MICROBIOLOGICAL ANALYSES OF EGGS AND EGG

PRODUCTS USING THE MICROFOSS SYSTEM

by

Amornrat Anprasertporn

M Sc. Food Science and Technology

(University of Western Sydney, Hawkesbury, Australia)

A thesis submitted for the degree of

Doctor of Philosophy

in

Food Science and Technology

School of Chemical Sciences and Engineering

The University of New South Wales

2009 PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES

Thesis/Dissertation Sheet

Surname or Family name: ANPRASERTPORN

First name: AMORNRAT Other name/s:

Abbreviation for degree as given in the University calendar: PhD

School: School of Chemical Science and Engineering Faculty: Faculty of Engineering

Title: Microbiological analyses of eggs and egg products using the MicroFoss system

Abstract 350 words maximum: (PLEASE TYPE) This thesis evaluates the MicroFoss system, also known as the BioSys or the Soleris® system, as a rapid method for microbiological analyses of eggs and egg products. Commercially assays for assessment of microbiological quality and sanitation conditions during processing, which are total viable counts (TVC), coliforms and Escherichia coli, are evaluated in comparison to the cultural methods. Concurrently, this thesis reports development of a new medium for the MicroFoss Salmonella detection, since this test is not available.

Good correlations between the cultural TVCs and the MicroFoss detection times (DTs) were obtained (r ≥-0.84) in all egg products. Preliminary results of a correlation between the cultural TVCs dominated by Pseudomonas fluorescens and the MicroFoss DTs suggested that the MicroFoss could be used for investigation of spoilage and shelf life prediction of eggs.

High agreements (r ≥-0.93) between the MicroFoss and cultural methods for coliforms and E. coli in eggs and egg products was achieved as well as the accuracy, sensitivity and specificity. The MicroFoss for estimations of TVC, coliforms and E. coli could be completed in less than 24h, compared to 48-72h required for the cultural methods. Hence, the MicroFoss could be used as a rapid screening for quality control of eggs and egg products.

While a new propylene glycol-based selective medium developed was effective for detection of Salmonella, possible false positive results caused by certain bacteria, such as Enterobacter cloacae, E. coli, Klebsiella pneumoniae and Proteus mirabilis, were encountered. When this selective medium was supplemented with 40 mg/L of novobiocin, or 1 or 2 mL/L of Niaproof 4, these non-salmonellae were greatly suppressed or eliminated, particularly Pr. mirabilis, whilst this did not affect growth of Salmonella. Various serotypes of Salmonella including strains of atypical biotype were able to acidify propylene glycol in these media.

Detection of Salmonella in naturally contaminated eggs by the MicroFoss using propylene glycol-based selective media showed equivalent sensitivity and specificity to the Australian Standard 5013.10-2004, according to numbers of egg samples tested in this study. Detection of low level of Salmonella (1-10 CFU/25g of eggs pre-enriched in buffered peptone water) was achieved in less than 24h.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the

University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

…………………………………………………………… ……………………………………..……………… ……….……………………...…….… Signature Witness Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

THIS SHEET IS TO BE GLUED TO THE INSIDE FRONT COVER OF THE THESIS ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………...... Date ……………………………………………...... ….

COPYRIGHT STATEMENT

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………...... Date ……………………………………………...... ….

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………...... Date ……………………………………………...... ….

i ACKNOWLEDGEMENTS

I would like to extend my sincere thanks firstly to my supervisor, A/Prof. Julian Cox. The guidance shown during the course of my study has been essential in achieving the results shown in this thesis.

Prof. Graham Fleet’s input has been invaluable over the last 9 months; helping steer and motivate me over the final stages of this work and for that I thank you.

This project could not have proceeded without the valuable material support provided by

Foss Pacific.

I also would like to thank the commercial egg supplier for providing products for testing.

For all the time spent looking after the laboratories, making sure everything ran smoothly I must thank Camillo Taraborrelli.

The statistical help provided by Prof. William Dunsmuir is also greatly appreciated.

All those long days and nights spent in the laboratory could not have allowed me to keep my sanity without the fun and supportive environment provided by all the students, especially Peter, Lidia, Hugh, Jum, Ai Lin, Sungsook and Maria. The friendship’s made during this time have been truly wonderful.

Lastly, thanks must be given to my wonderful parents, family, and friends for their kind support, and caring attitude; Sonia for reading, and my Samee for looking after me.

ii LISTS OF PRESENTATION AND PAPERS

Cox, J.M. and Anprasertporn, A., 2003. Estimation of coliforms in eggs and egg products by the MicroFoss system. 11th Australian Food Microbiology Conference, AIFST, 26th –

28th March 2003, Noosa.

The calibration curves were utilised by a commercial egg manufacturer. Results from ongoing testing of egg products by this commercial egg manufacturer corresponded well with those from this research. This confirmed the validity of the findings from this study.

Possible commercialisation of the medium for Salmonella detection developed in this research. Interest shown by a third party and New South Innovations.

iii TABLE OF CONTENTS

ORIGINALITY STATEMENT, COPY STATEMENT, AND i

AUTHENTICITY STATEMENT……………………………………………….

ACKNOWLEDGEMENTS……………………………………………………... ii

LISTS OF PRESENTATIONS AND PAPERS………………………………… iii

TABLE OF CONTENTS……………………………………………………….. iv

ABSTRACT…………………………………………………………………….. 1

CHAPTER 1: INTRODUCTION……………………………………………….. 3

CHAPTER 2: LITERATURE REVIEW………………………………………... 8

2.1 Introduction…………………………………………………………… 8

2.2 Eggs and egg products………………………………………………… 9

2.3 Microbiology of eggs and egg products………………………………. 9

2.3.1 Shell eggs………………………………………………………… 10

2.3.2 Pasteurized egg products………………………………………… 15

2.3.3 Dried egg products………………………………………………. 18

2.4 Structural properties of eggs to microbial resistance…………………. 20

2.4.1 Cuticle……………………………………………………………. 21

2.4.2 The true shell…………………………………………………….. 22

2.4.3 Shell membranes…………………………………………………. 23

2.4.4 Albumen…………………………………………………………. 24

2.4.5 Chalaziferous layer………………………………………………. 25

2.4.6 Vitelline membrane……………………………………………… 26

2.4.7 Yolk……………………………………………………………… 26

2.5 Microbiological examinations of eggs and egg products……….…….. 26

iv 2.5.1 Total viable count………………………………………………... 27

2.5.2 Spoilage index for shelf life assessment…………………………. 28

2.5.3 Enteric indicator organisms……………………………………… 28

2.5.3.1 Coliforms…………………………………………………… 29

2.5.3.2 Escherichia coli…………………………………………….. 30

2.5.4 Salmonellae……………………………………………………… 31

2.5.4.1 Salmonellosis……………………………………………….. 32

2.5.4.2 Evidence of Salmonella outbreaks from eggs and

egg-associated foods………………………………………... 33

2.5.4.3 Conventional detection of salmonellae in foods……………. 36

2.5.5 Microbiological testing and specifications of eggs and egg

products………………………………………………………….. 41

2.6 Implementation of rapid methods for microbiological testing………... 41

2.6.1 Performance indicator for evaluation of qualitative alternative

method…………………………………………………………… 42

2.6.2 Performance indicator for evaluation of quantitative alternative

method…………………………………………………………… 46

2.6.3 Test for significant difference of positive proportions for the

cultural and alternative assays…………………………………… 46

2.7 The MicroFoss method………………………………………………... 47

2.7.1 Principles………………………………………………………… 48

2.7.2 Operational procedure…………………………………………… 52

2.7.3 Commercial availability, applications and advantages…………... 53

2.8 Principles of detection of salmonellae………………………………… 56

2.8.1 Hydrogen sulfide (H2S) production as the differential system…... 56

v 2.8.2 Propylene glycol acidification as the differential system………... 56

2.8.3 Selective agents………………………………………………….. 57

CHAPTER 3: EVALUATION OF THE MICROFOSS SYSTEM FOR

DETERMINING TOTAL VIABLE COUNTS IN EGGS AND EGG

PRODUCTS…………………………………………………………………….. 64

3.1 Introduction…………………………………………………………… 64

3.2.1 Materials…………………………………………………………. 66

3.2.2 Sample handling prior to testing………………………………… 67

3.2.3 Inoculum preparation…………………………………………….. 68

3.2.4 Sample preparation for the TVC test…………………………….. 69

3.2.5 TVC cultural method…………………………………………….. 70

3.2.6 The MicroFoss TVC test………………………………………… 70

3.2.7 Statistical analyses……………………………………………….. 71

3.2.8 Identification of the isolates of predominant bacteria in raw

whole egg………………………………………………………… 72

3.3 Results………………………………………………………………… 74

3.3.1 Sample dilution for testing in the MicroFoss system……………. 74

3.3.2 Application of the MicroFoss system to estimate the TVC

in unpasteurized egg……………………………………………... 76

3.3.3 Application of the MicroFoss system to estimate the TVC

in pasteurized egg products……………………………………… 85

3.3.4 Application of the MicroFoss system to estimate populations of

Pseudomonas fluorescens in unpasteurized whole egg………….. 94

3.4 Discussion……………………………………………………………... 99

vi CHAPTER 4: EVALUATION OF THE MICROFOSS SYSTEM FOR 109

DETERMINING COLIFORMS IN EGGS AND EGG PRODUCTS…………..

4.1 Introduction…………………………………………………………… 109

4.2 Materials and methods………………………………………………… 111

4.2.1 Materials…………………………………………………………. 111

4.2.2 Sample handling prior to testing………………………………… 111

4.2.3 Inoculum preparation…………………………………………….. 111

4.2.4 Sample preparation………………………………………………. 112

4.2.5 Coliform cultural method………………………………………... 115

4.2.6 The MicroFoss coliform test……………………………………... 116

4.2.7 Statistical analyses……………………………………………….. 117

4.3 Results………………………………………………………………… 119

4.3.1 MicroFoss curves for coliform assays…………………………… 119

4.3.2 Variability of the MicroFoss assay for detection of coliforms

in eggs……………………………………………………………. 121

4.3.3 Application of the MicroFoss system to estimate coliforms

in unpasteurized egg……………………………………………... 122

4.3.4 Application of the MicroFoss system to estimate coliforms

in pasteurized egg products……………………………………… 128

4.3.5 Performance indicators…………………………………………... 132

4.3.6 Reproducibility of the MicroFoss assay and significant

differences between the cultural and MicroFoss coliform counts.. 135

4.3.7 Significant differences of proportions of confirmed coliforms

between the cultural and MicroFoss assays……………………… 137

4.4 Discussion……………………………………………………………... 138

vii CHAPTER 5: EVALUATION OF THE MICROFOSS SYSTEM FOR

DETERMINING ESCHERICHIA COLI IN EGGS AND EGG PRODUCTS…. 150

5.1 Introduction…………………………………………………………… 150

5.2 Materials and methods………………………………………………… 152

5.2.1 Materials…………………………………………………………. 152

5.2.2 Sample handling prior to testing………………………………… 152

5.2.3 Inoculum preparation…………………………………………….. 152

5.2.4 Sample preparation………………………………………………. 153

5.2.5 E. coli cultural method…………………………………………… 155

5.2.6 The MicroFoss E. coli test……………………………………….. 158

5.2.7 Statistical analyses……………………………………………….. 159

5.3 Results………………………………………………………………… 160

5.3.1 Application of the MicroFoss system to estimate E. coli

in unpasteurized egg……………………………………………... 160

5.3.2 Application of the MicroFoss system to estimate E. coli

in pasteurized egg products……………………………………… 167

5.3.3 Performance indicators…………………………………………... 172

5.3.4 Reproducibility of the MicroFoss assay and significant

differences between the cultural and MicroFoss E. coli counts…. 175

5.3.5 Significant differences of proportions of confirmed E. coli

between the cultural and MicroFoss assays……………………… 176

5.4 Discussion…………………………………………………………….. 176

CHAPTER 6: DEVELOPMENT OF A SALMONELLA MEDIUM FOR

DETECTION OF SALMONELLAE BY THE MICROFOSS SYSTEM……… 184

6.1 Introduction…………………………………………………………… 184

viii 6.2 Materials and methods………………………………………………… 185

6.2.1 Microbiological media, chemicals and reagents…………………. 185

6.2.2 Preparation of reference cultures………………………………… 188

6.2.3 Preparation of media……………………………………………... 192

6.2.4 Testing of media…………………………………………………. 193

6.3 Results………………………………………………………………… 194

6.3.1 Evaluation of xylose lysine deoxycholate (XLD) and lysine

mannitol glycerol (LMG) media as potential candidates for use

in the MicroFoss assay…………………………………………... 194

6.3.2 Development of propylene glycol-based selective medium for

potential use in the MicroFoss assay…………………………….. 199

6.3.3 Optimisation of propylene glycol-based selective medium……... 204

6.3.4 Modification of propylene glycol-based medium……………….. 222

6.3.4.1 Addition of novobiocin……………………………………... 222

6.3.4.2 Addition of Niaproof 4……………………………………... 226

6.3.5 Examination of the effectiveness of propylene glycol-based 230

medium…………………………………………………………...

6.3.5.1 Analyses with non-salmonellae…………………………….. 230

6.3.5.2 Analyses with salmonellae…………………………………. 234

6.3.6 Evaluation of propylene glycol acidification characteristic of

atypical biotypes of Salmonella………………………………….. 237

6.4 Discussion……………………………………………………………... 241

CHAPTER 7: EVALUATION OF PROPYLENE GLYCOL-BASED

MEDIUM FOR DETECTION OF SALMONELLA IN EGGS BY THE

MICROFOSS SYSTEM………………………………………………………… 252

ix 7.1 Introduction…………………………………………………………… 252

7.2 Materials and methods………………………………………………… 254

7.2.1 Materials…………………………………………………………. 254

7.2.2 Sample handling prior to testing…………………………………. 254

7.2.3 Microbiological media, chemicals and reagents and test kits..….. 254

7.2.4 Reference culture………………………………………………… 254

7.2.5 Preparation of media……………………………………………... 254

7.2.6 Procedure of Salmonella detection………………………………. 256

7.2.7 Conventional biochemical testing……………………………….. 258

7.2.8 Rapid biochemical and serological testing………………………. 261

7.2.9 Detection limit of the MicroFoss assay for Salmonella

detection………………………………………………………….. 262

7.3 Results………………………………………………………………… 264

7.3.1 Detection of Salmonella in naturally contaminated eggs………... 264

7.3.2 Sensitivity of the MicroFoss assay for Salmonella detection……. 281

7.3.3 Specificity of the MicroFoss assay for Salmonella detection……. 281

7.3.4 Detection limit of MicroFoss method for Salmonella detection… 281

7.4 Discussion……………………………………………………………... 285

CHAPTER 8: CONCLUSIONS………………………………………………… 294

CHAPTER 9: BIBLIOGRAPHY……………………………………………….. 299

APPENDIX 10.1………………………………………………………………... 324

APPENDIX 10.2………………………………………………………………... 325

APPENDIX 10.3………………………………………………………………... 325

x ABSTRACT

This thesis evaluates the MicroFoss system, also known as the BioSys or the Soleris® system, as a rapid method for microbiological analyses of eggs and egg products.

Commercial assays for assessment of microbiological quality and sanitation conditions during processing, which are total viable counts (TVC), coliforms and Escherichia coli, are evaluated in comparison to the cultural methods. Concurrently, this thesis reports development of a new medium for the MicroFoss Salmonella detection, since this test is not available.

Good correlations between the cultural TVCs and the MicroFoss detection times (DTs) were obtained (r ≥-0.84) in all egg products. Preliminary results of a correlation between the cultural TVCs dominated by Pseudomonas fluorescens and the MicroFoss DTs suggested that the MicroFoss could be used for investigation of spoilage and shelf life prediction of eggs.

High agreements (r ≥-0.93) between the MicroFoss and cultural methods for coliforms and

E. coli in eggs and egg products was achieved as well as the accuracy, sensitivity and specificity. The MicroFoss for estimations of TVC, coliforms and E. coli could be completed in less than 24h, compared to 48-72h required for the cultural methods. Hence, the MicroFoss could be used as a rapid screening for quality control of eggs and egg products.

While a new propylene glycol-based selective medium developed was effective for detection of Salmonella, possible false positive results caused by certain bacteria, such as

Enterobacter cloacae, E. coli, Klebsiella pneumoniae and Proteus mirabilis, were encountered. When this selective medium was supplemented with 40 mg/L of novobiocin, 1 or 1 or 2 mL/L of Niaproof 4, these non-salmonellae were greatly suppressed or eliminated, particularly Pr. mirabilis, whilst this did not affect growth of Salmonella. Various serotypes of Salmonella including strains of atypical biotype were able to acidify propylene glycol in these media.

Detection of Salmonella in naturally contaminated eggs by the MicroFoss using propylene glycol-based selective media showed equivalent sensitivity and specificity to the Australian

Standard 5013.10-2004, according to numbers of egg samples tested in this study.

Detection of low level of Salmonella (1-10 CFU/25g of eggs pre-enriched in buffered peptone water) was achieved in less than 24h.

2 CHAPTER 1

Introduction

Eggs are one of the most common foods and are used as ingredients in a variety of products. Eggs are mainly marketed and consumed as shell eggs, for instance, 70% of eggs reach the consumers as shell eggs, whilst egg products, such as pasteurised and dried egg products, are usually used in manufacturing and catering (ICMSF, 2005). Eggs from healthy, non-infected chickens are produced in a sterile reproductive area until they are laid, after which they are subject to contamination from the environment. Intact shell eggs, which are stored under appropriate conditions, are not prone to spoilage and pathogenic contamination. However, invasion of microorganisms present on the surface of the eggshell into the egg contents can occur, as eggs age and these microorganisms overcome the deteriorated defensive barriers of eggs (Cox, 2001). Raw, cooked eggs and egg-associated foods have been documented as a transmission vehicle for salmonellosis in human throughout the world (Adams and Moss, 2000; Cox, 2000; Hammack and Andrews, 2000;

Jay, 2000; Sparks, 2000; Cogan and Humphrey, 2003). Hence, it is essential to perform certain microbiological testings to ensure an overall quality of eggs and safety for consumption.

Eggs are analysed for several groups of microorganisms, such as total viable counts (TVC) or aerobic plate count (APC), coliforms, Escherichia coli, yeasts and moulds (particularly dried egg products), coagulase-positive staphylococci, and pathogens, such as Bacillus cereus, Campylobacter, Clostridium botulinum, Clostridium perfringens, salmonellae and

Vibrio parahaemolyticus (Andrews, 2000a; Ricke et al., 2001; AS 5013.7-2004). This 3 research applies certain microbiological testings to evaluate microbiological quality of eggs and egg products, such as TVC, coliforms, E. coli and salmonellae. Total viable counts indicate an overall microbiological quality, whilst coliforms and E. coli results indicate hygiene, sanitation, and efficiency of processing. Absence of salmonellae in eggs and egg products indicates safety for consumption.

Traditional cultural methods to enumerate total aerobic populations and indicator organisms involve plating method. Enumerations of TVC and coliforms are performed by pour plate method using Plate Count Agar (PCA) (AS 1766.1.3-1991; AS 1766.2.1-1991) and Violet

Red Bile Agar (VRBA) (AS 5013.4-2004), respectively. Enumeration of E. coli populations is performed by spread plate method using Eosin Methylene Blue (EMB)

(Levine) agar (AS 1766.1.4-1991). The traditional cultural method for the detection of

Salmonella is enrichment in non-selective and selective broths followed by plating onto solid media (AS 5013.10-2004). Although traditional cultural methods provide accurate and reliable results, the methods are laborious and lengthy, in which enumerations of TVC, coliforms and E. coli requires 24-48h (including confirmation for coliforms and E. coli) to obtain results. Detection of salmonellae by cultural method takes up to five to seven days to complete. Since eggs are a perishable food, it is beneficial to obtain timely microbiological results to facilitate an early release of the products to consumers. Demands for fast, automated, cost-effective and reliable methods to assess microbiological quality of foods have been increased due to the increasing consumer interest in food safety and modern approaches to food processing and quality assurance. Criteria to select a new rapid method or technology are accuracy and reliability of data; acceptability of data by regulatory authorities; speed of analysis and availability of results; degree of automation; convenience

4 and simplicity of the procedure; cost and economic benefit; requirements for staff training; availability of the technology and back-up service; requirements for space; and safety risks.

Examples of alternative rapid methods/techniques are membrane filter techniques; centrifugation; flow cytometry; impedance and conductance; optical growth measurement

(the MicroFoss system); bioluminescence; immunoassay; nucleic acid probes; amplification techniques; and typing techniques (Cox and Fleet, 2003).

This study evaluates the efficacy of the MicroFoss system to determine microbiological quality of eggs and egg products. Although certain microbiological tests by the MicroFoss system are commercially available, its application to eggs and egg products has not yet been established. The MicroFoss system is an automated instrument designed to detect the growth of microorganisms in food and industrial samples, which is represented by changes in colour of selective liquid medium as a result of microbial metabolites. The liquid medium is formulated with essential nutrients for microbial growth and/or specific substrates for detection of target organisms with pH dye indicators for optical detection.

Colour change of the inoculated medium results from chemical characteristics, such as pH, redox and enzymatic activity, of metabolic reactions. Detection occurs in a semi-solid medium with similar formulation to the liquid medium or an agar plug, which resides underneath the liquid medium in the MicroFoss vial. The agar plug mirrors colour change occurring in the inoculated liquid medium, in which changes are photometrically detected by light emitted diodes every six minutes. Periods detecting such changes are described as detection time (DT). Corresponding the MicroFoss DTs and the cultural microbial counts forms a calibration curve as a basis for rapid estimation of various microbiological populations in food and industrial samples by the MicroFoss system. It is noted that

5 detection times are inversely related to populations of microorganisms in samples. Previous studies have investigated the efficacy of the MicroFoss system to estimate populations of certain microbiological groups in food and environmental samples, such as TVC (Shelef et al., 1997; Russell, 2001; Firstenberg-Eden et al., 2002; Odumeru and Belvedere, 2002), coliforms (Russell, 2001; Firstenberg-Eden et al., 2002; Odumeru and Belvedere, 2002;

Firstenberg-Eden et al., 2004), E. coli (Russell, 2000; Russell, 2001; Odumeru and

Belvedere, 2002) and Enterobacteriaceae (Firstenberg-Eden et al., 2002), and detect salmonellae (Shelef and Tan, 1998; Shelef et al., 1998; Tan and Shelef, 1999; Peng and

Shelef, 2001) and listeriae (Firstenberg-Eden and Shelef, 2000; Peng and Shelef, 2000;

Peng and Shelef, 2001; Mbandi and Shelef, 2002; Odumeru, Belvedere & van

Donkersgoed, 2004) in various foods. All findings were observed to have good correlations between cultural and the MicroFoss methods to estimate TVC, coliforms and E. coli populations in various foods, with correlation coefficients (r) ranging from –0.90 to –0.99, as well as the relevant efficiency of both methods to detect salmonellae and listeriae in foods. Results of the estimation of TVC, coliforms, E. coli, and Enterobacteriaceae in all studies were obtained within two to 16h, whereas detection of salmonellae and listeriae completed within 24h including the pre-enrichment step.

6 This research is divided into two parts. The first part is the evaluation of commercially available assays for quantitative and qualitative analyses of total bacterial populations, indicator (coliforms), and index (E. coli) organisms in eggs and egg products. As the

MicroFoss assay for Salmonella detection is not commercially available, the second part of this research is the development of a new medium for detection of Salmonella in eggs and egg products by the MicroFoss system.

Research objectives

1. Evaluation of the MicroFoss system for estimation of TVC in eggs and egg products.

2. Evaluation of the MicroFoss system for estimation of coliform populations in eggs and egg products.

3. Evaluation of the MicroFoss system for estimation of E. coli populations in eggs and egg products.

4. Development and evaluation of a new Salmonella medium for detection of Salmonella in eggs and egg products by the MicroFoss system. The development of the new medium includes evaluation of its effectiveness in comparison to selective detection of Salmonella in culture media.

7 CHAPTER 2

Literature review

2.1 Introduction

This thesis is focused on evaluation of the MicroFoss system for microbiological analyses of eggs and egg products. The research is divided into two parts. The first part is the evaluation of commercially available assays for quantitative and qualitative analyses of total bacterial populations, indicator (coliforms), and index (Escherichia coli) organisms in eggs and egg products. The second part is aimed to develop a new medium for Salmonella detection by the MicroFoss assay, as this test is not commercially available. This literature review provides the occurrence of microorganisms in eggs and egg products and their role in terms of deterioration of egg qualities, which results in spoilage and pathogenic contamination of products. Certain microbiological analyses for routine screening to assure quality and safety for consumption are described. Since eggs are a perishable food and their shelf life and quality can deteriorate under improper handling and storage conditions prior to distribution to consumers, it is necessary to obtain rapid results to minimise these risks.

Applications of the MicroFoss system to rapidly enumerate and detect target organisms in foods and environmental samples are summarised. A final Section reviews potential biochemical reactions and selective agents that can be used for formulating a new medium for Salmonella detection by the MicroFoss.

8 2.2 Eggs and egg products

Eggs and egg products referred to in this thesis are products and derivatives of domestic chickens. While shell eggs are mostly marketed and consumed as is, other egg products such as dried, frozen and pasteurized egg are commonly used in manufacturing of various foods, such as bakery products, confectionery, drinks, special diets, infant foods, sauces, mayonnaise and noodles. For commercial use in a variety of foods, egg contents are broken from their shells, and either entirely mixed or separated into yolks and albumens (Ricke et al., 2001; ICMSF, 2005). Commercial liquid egg products are normally pasteurized, and subsequently chilled or frozen for storage (Delves-Broughton and Board, 2000; Cox, 2001;

Ricke et al., 2001; ICMSF, 2005). Although the addition of salt or sugar to yolk increases heat resistance of microflora, it promotes flow during heat treatment and therefore enhances the efficacy of pasteurization. Concurrently, the addition of salt or sugar to raw material may prohibit microbial growth prior to pasteurization. Albumens are mildly heat treated with or without the addition of other ingredients for specific purposes (Cox, 2001). On the other hand, liquid egg products can also be de-sugarised by enzymic or microbial action and dried for the purpose of a food ingredient (ICMSF, 2005).

2.3 Microbiology of eggs and egg products

Microbiology of eggs and egg products are separately described, as the impact of nature and processing of products results in differences in their microflora. The microflora of each product are considered as the initial microflora, spoilage organisms, and pathogens.

9 2.3.1 Shell eggs

2.3.1.1 Initial microflora

Eggs from healthy, non-infected chickens are produced in the sterile reproductive areas, the ovaries and oviduct (Cox, 2001). However, eggs can be contaminated by two principal ways, which are transovarian or trans-shell infection. Certain bacterial species, which were isolated from the laid eggs, suggested that those eggs were produced from the infected chicken through the oviduct (ICMSF, 2005). The ovary are colonised by a few microorganisms, such as Lactobacillus sp., Micrococcus sp. and Pasteurella haemolytica,

Spoilage of the eggs occurs if some of these microorganisms are transmitted through the ovaries and proliferate. Although bacteria may contaminate the yolk in the oviduct, disposal of these contaminants is effected by proteins in the albumen through the albumen-secreting segment of the oviduct (Kobayashi et al., 1997). Microorganisms, which reside in the ovarian tissue of chickens, usually present in low numbers do not lead to spoilage of eggs and egg-borne illnesses. However, Salmonella Enteritidis can be transmitted through the ovary of the infected hens, in which they inhabit in the yolk. Generally, the eggshell and contents of the laid eggs from non-infected chickens are considered hygienic until eggs are transported to the cloaca (anal sphincter) or the junction of the reproductive, urinary and alimentary tracts of the chicken. As a consequence, the eggshell is contaminated with different microorganisms, in which type of contaminants vary accordingly to the age of chickens. Microflora of the eggshells produced from young, non-infected chickens, whose cloaca muscle is toned, are predominantly Gram-positive bacteria such as Bacillus and skin microflora organisms, such as Micrococcus and Staphylococcus, due to their tolerance to the dry conditions (Board and Tranter, 1986; Kobayashi et al., 1997; Cox, 2001). On the

10 contrary, eggs produced from aged chickens, whose cloaca muscle depletes its tone, are more prone to faecal deposit on the eggshell (Cox, 2001). As a consequence, the eggshells are exposed to a variety of microorganisms including species of Gram-negative bacteria, such as Alcaligenes, Flavobacterium, Pseudomonas, Citrobacter, Enterobacter,

Escherichia, Hafnia, Proteus, and Serratia, and pathogenic bacteria, such as

Campylobacter and Salmonella (Board et al., 1964; Board, 1969; Harrigan and McCance,

1976; Board and Tranter, 1986; Kobayashi et al., 1997; Harrigan, 1998; Sparks, 2000; Cox,

2001; Ray, 2004). Other environmental matters, such as nesting materials, feeds, air, dust, soil, hatching equipment, trays, water, workers, insects, and animals are also responsible for the origin of microbial acquisition on the eggshell (Board and Tranter, 1986; Cox, 2001).

Once the microflora on the eggshell successfully invade into the egg content, Gram- negative bacteria are usually isolated from egg contents and play a major role in causing spoilage in eggs (Harrigan and McCance, 1976; Board and Tranter, 1986; Kobayashi et al.,

1997). Gram-positive bacteria are usually inhibited by antimicrobial proteins in albumen

(Board and Tranter, 1986; Kobayashi et al., 1997).

2.3.1.2 Spoilage organisms

Intact shell eggs from healthy, non-infected chickens tend to remain the original quality and not to be prone to spoilage when they are stored under proper storage conditions, such as consistent low temperature and moderate relative humidity. Indigenous defensive barriers of shell eggs naturally deteriorate through storage period. For instance, a reduction of thickness and viscosity of albumen results in decomposition and flattened setting of yolk.

Rate of deterioration accelerates upon the elevated storage temperature (Cox, 2001).

11 However, spoilage can still occur once microbial contaminants utilise nutritious egg content after penetrating through pore of the eggshell and overcoming natural defensive mechanisms. The microorganisms on the eggshell are not essentially the ones that commonly cause spoilage (Mayes and Takeballi, 1983). Spoilage organisms are predominantly bacteria, which are more frequently Gram-negative motile rods rather than

Gram-positive bacteria. This is a result of their: i) motility to allow their entry into egg contents and reach the nutritious yolk under favourable conditions; ii) ability to produce hydrolytic enzymes to break down physical barriers inside eggs, such as membranous and liquid components, to enhance the availability of smaller molecules of nutrients for microorganisms to utilise; iii) ability to capture indigenous iron in eggs for their growth; iv) resistance to lysozyme activity due to the advantageous components of the outer membrane of the cell; v) capability to utilise primary carbon and nitrogen compounds for synthesising most fundamental growth compounds (Cox, 2001; ICMSF, 2005). The common Gram- negative bacteria causing spoilage in eggs are Acinetobacter, Aeromonas, Alcaligenes,

Flavobacterium, Pseudomonas spp., and some members in the family Enterobacteriaceae, such as Citrobacter, Enterobacter, Escherichia, Proteus and Serratia (Ayres, 1960; Board et al., 1964; Board, 1969; Bhargava et al., 1975; Fields, 1979; Imai and Saito, 1985; Board and Tranter, 1986; Stadelman, 1994; Cousin, 2000; Jay, 2000; Cox, 2001; Ricke et al.,

2001; Ray, 2004; ICMSF, 2005). Spoilage of shell eggs is generally described as ‘rot’ or the visible effect of produced metabolites from responsible organisms characterising by colour and physical change of egg components (Cox, 2001). For instance, spoilage of albumen caused by Pseudomonas fluorescens is described as green rot, since it causes greening of albumen (Ray, 2004). Furthermore, fluorescent pseudomonads, such as

12 Pseudomonas fluorescens, commonly cause spoilage in shell eggs, as they have a potential to quickly invade into the egg contents and are not susceptible to natural antimicrobial substances in albumen (Jones et al., 2002; ICMSF, 2005). In contrast, spoilage in shell eggs caused by Alcaligenes does not exhibit any change in appearance or odour of the egg content (Imai and Saito, 1985).

To a lesser extent moulds can also cause spoilage in eggs when eggs are manufactured under the exposed conditions, such as free-range eggs, or stored at refrigerated temperature with high humidity for lengthy periods (Board and Tranter, 1986; Cox, 2001). The predominant flora to cause fungal spoilage under typical storage conditions are

Cladosporium, Penicillium and Sporotrichum (Fields, 1979). Growth from spores of filamentous fungi or mould usually initiates on the surface of eggs, namely whiskers. This is frequently caused by species of Cladosporium, particularly Cladosporium herbarum.

Thereafter, hyphae usually penetrate pores of the shell and shell membrane and spread throughout egg contents (Board and Tranter, 1986; Board et al., 1994; Sparks, 2000; Cox,

2001). The internal growth is more common than the external growth, which causes difficulty in grading eggs with this defect. More importantly, the external fungal growth prompts bacterial invasion into the egg content in regards to decomposition of cuticular material by fungi (Cox, 2001).

2.3.1.3 Pathogenic organisms

Pathogens derived from eggs are species of Aeromonas, Campylobacter, Listeria and

Salmonella. However, only salmonellae have been documented as a significant pathogen associating with outbreaks from eggs and egg-associated foods (Sparks, 2000). For

13 instance, in 1988 salmonellae were reported to be the cause of food poisoning in cooked eggs in Japan, which affected more than 10,000 people (Cox, 2000a). Egg contents can be contaminated through either the infection of the oviduct or penetration through the shell by contaminants from the intestinal tract or environmental matters. Several serotypes of salmonellae have been isolated from shells of eggs, such as Salmonella Anatum, S.

Bareilly, S. Enteritidis, S. Derby, S. Essen, S. Heidelberg, S. Montevideo, S. Oranienburg, S.

Thompson, S. Typhimurium and S. Worthington (Cantor and McFarlane, 1948; Humphrey,

1994; Sparks, 2000). However, apart from S. Enteritidis, most salmonellae normally contaminate egg content by penetration through the outside of the shell when the environmental conditions enhance egg deterioration, such as cuticle damage, temperature decline and low specific gravity of shell. Salmonellae on the outer shell are unable to grow and subsequently die when eggs are stored under 10°C. However, their survival could be increased under high relative humidity and low temperature (Jay et al., 2003). Humphrey et al. (1991) stated that the initial populations of salmonellae in contaminated shell eggs were as low as 10-20 CFU/egg, though Humphrey (1994) also reported the occurrence of higher levels of contamination.

Listeria spp., such as Listeria innocua and Listeria monocytogenes, were also isolated from commercially broken raw eggs (Leasor and Foegeding, 1989; Stadelman, 1994) and in-line filters from the processing of raw blended whole egg (Moore and Madden, 1993).

14 2.3.2 Pasteurized egg products

2.3.2.1 Initial microflora

Pasteurization of eggs and egg products is performed under mild conditions due to product sensitivity to heat, which affects their functional properties and texture. Pasteurization aims to destroy heat-sensitive bacteria such as salmonellae and Gram-negative bacteria including pseudomonads (Delves-Broughton and Board, 2000; Cox, 2001). Hence, the microflora of pasteurized egg products mainly consists of heat-resistant Gram-positive bacteria, predominantly species of Bacillus, Micrococcus and Staphylococcus and occasionally a few

Gram-negative rod survivors, such as Alcaligenes, Escherichia and Flavobacterium (Shafi et al., 1970; Stadelman, 1994; Delves-Broughton and Board, 2000; Cox, 2001; Ricke et al.,

2001; ICMSF, 2005). Afterwards, pasteurized products are preserved by chilling or freezing, which further reduces microbial populations in pasteurized products (Cox, 2001).

Freezing eliminates certain survivors in pasteurized egg products, such as Escherichia,

Flavobacterium and Gram-positive cocci, and prolong shelf life of finished products (Shafi et al., 1970). Pasteurization of raw albumen eliminates the flora residing in albumen, such as Acinetobacter, Enterobacter, and Pseudomonas (ICMSF, 2005). Pasteurization remarkably improves keeping quality of refrigerated egg products. Chilled pasteurized products must be used within a few days after pasteurization to avoid potential spoilage by psychrotrophs. Microbial populations of frozen products quickly declined instantly after freezing. Subsequently, they stabilised throughout freezing (Cox, 2001).

15 2.3.2.2 Spoilage organisms

Spoilage of refrigerated pasteurized liquid egg products is usually caused by psychrotrophic bacteria due to post-pasteurization contamination (Cox, 2001; Ray, 2004). Common spoilage microflora of liquid egg products are generally species of Gram-negative bacteria, mainly Pseudomonas, and infrequently members of the family Enterobacteriaceae. In some cases, psychrotrophic Bacillus spp., including Bacillus cereus, can cause spoilage in pasteurized liquid egg products, especially in defrosted pasteurized liquid egg products stored at the raised temperature for lengthy period. Spoiled liquid egg products are described as bitter or strong off-odours, or a bitter taste when the odour is not evident (Cox,

2001).

2.3.2.3 Pathogenic organisms

Pathogens recovered from liquid egg products are Bacillus cereus, Clostridium perfringens,

Salmonella and Staphylococcus aureus. However, salmonellae are the most significant human pathogen in liquid egg products. Pasteurization usually eliminates pathogens, such as salmonellae (Baker, 1974; Delves-Broughton and Board, 2000; Food Science Australia,

2008). Pasteurized egg products are recommended for the routine preparation of foods that are not subjected to be fully cooked (Food Science Australia, 2008). It was reported that pasteurization parameters established to destroy salmonellae were proved to be effective in controlling the epidemic of Salmonella Enteritidis phage type 4 without negatively altering the functional properties of eggs (Delves-Broughton and Board, 2000). However, pasteurized liquid eggs can be re-infected by salmonellae through post-pasteurization

16 contamination, most commonly from humans (Baker, 1974). It was noted that salmonellae had a rapid growth in yolks in either aerobic or anaerobic conditions (ICMSF, 2005).

Apart from salmonellae, psychrotrophic Listeria monocytogenes is also another important pathogen. Despite the heat resistance of Listeria monocytogenes being greater than that of salmonellae, the present pasteurization parameters used in the egg industry efficiently eliminate Listeria monocytogenes in liquid egg (Delves-Broughton and Board, 2000).

Nonetheless, pasteurized liquid egg products have yet to be identified as a transmission vehicle of listeriosis in human. Listeria monocytogenes was reported to be able to grow in raw and pasteurized liquid whole egg and yolk stored at 5°C to 30°C. However, it became dormant in unpasteurized liquid albumen at pH 7.0-8.9, due to lysis of bacterial cell walls by lysozyme in albumen (Stadelman, 1994; ICMSF, 2005).

The liquid whole egg may also support the growth of Staphylococcus aureus, when it is maintained at temperatures over 15.6°C. The occurrence of staphylococci in salted yolks has been increased due to the ability to grow at low water activity of 0.90 of salted yolks. A hazard from staphylococcal toxin could be encountered if staphylococci survive pasteurization or contaminate eggs or egg products through post-pasteurization (ICMSF,

2005). They must be stored under severe temperature abuse, for instance storage of the egg product at ambient temperature for a few days, to allow staphylococci to proliferate to the level of 105 CFU/mL before toxin forms (Ijichi et al., 1973). However, as salted yolks are pasteurized and filled hot into cans, the bacterial vegetative cells capable of growth in 10% salt are usually destroyed by the heat. This gives long shelf life to the product, even it is not rapidly cooled and is stored at room temperature (Cotterill et al., 1974). Hence, the incidence of potential hazard from consumption of salted yolk with staphylococci toxin is

17 less, as products are likely to be used in manufacturing or spoilage may occur and is noticed beforehand.

2.3.3 Dried egg products

2.3.3.1 Initial microflora

Dried egg products are manufactured using liquid egg, which has undergone pasteurization prior to drying. Dehydration of liquid egg products can be performed by three means, which are spray drying, drying on a heated surface (namely pan or drum drying) and freeze drying. Removal of glucose in albumen prior to drying is usually conducted to prevent non- enzymic browning (Maillard reactions) in dried albumen products, which could be mistakenly identified as spoilage (Cox, 2001; ICMSF, 2005). Insoluble brown compounds are produced as a result of a reaction between the aldehyde group of glucose and the amino groups of proteins. Furthermore, this reaction also results in the reduction of solubility of product and formation of off flavours. Glucose removal also enhances stability in a lesser degree in dried whole egg and dried yolk products (ICMSF, 2005). Regardless the method of drying, microflora of dried egg products is similar. Drying eliminates a number of bacteria primarily exist in liquid egg. Once the product is dried, the microbial populations become steady and the additional decline only gradually happens under prolonged storage, even at ambient temperature (Delves-Broughton and Board, 2000; ICMSF, 2005). Hence, the microflora of dried products consists of the resistant species of Gram-positive bacteria, including non-sporeformers, such as enterococci, and aerobic sporeforming bacilli (Delves-

Broughton and Board, 2000; Cox, 2001; ICMSF, 2005). Bacterial survival is more probable

18 in fermented albumen or temperature-abused whole egg, whose initial bacterial loads are high (ICMSF, 2005).

2.3.3.2 Spoilage organisms

Dry conditions usually inhibit spoilage of dried egg products, as microorganisms in dried egg products generally stay viable and dormant, but do not proliferate (Cox, 2001; ICMSF,

2005). However, most bacteria gradually die periodically due to various intrinsic and extrinsic factors, such as species, temperature, pH, water activity and atmosphere (ICMSF,

2005). Nonetheless, once the products are exposed to moisture and/or unstable storage temperatures, which allow condensation inside of the package, revival of the remaining bacteria due to the increased water activity can be enhanced. As a result, the product can become prone to bacterial and fungal spoilage, particularly by Penicillium species (Cox,

2001; ICMSF, 2005).

2.3.3.3 Pathogenic organisms

Historically prior to 1965, salmonellae were commonly implicated in the contamination of dried egg products. Low levels (0.1-0.01 CFU/g) of S. Enteritidis, S. Lille and S.

Typhimurium were isolated from whole egg powder manufactured in Yugoslavia (Matic et al., 1990). As a consequence, pasteurization, either wet or dry, has been introduced and this efficiently reduces the occurrence of salmonellae in dried egg products. Salmonellae should be eliminated completely by pasteurization and further reduced by 10000-fold during drying. Furthermore, salmonellae can be further eradicated by hot storage or hot room treatment once the product has been dried. However, contamination with salmonellae could

19 be increased if they grow due to the utilisation of glucose, or they could cross-contaminate to the finished product (Delves-Broughton and Board, 2000; ICMSF, 2005).

Listeria monocytogenes can remain unalterably in dried powdered whole egg or yolk for extended periods at refrigeration temperatures, but declines over time under storage at 20°C

(Brackett and Beuchat, 1991).

2.4 Structural properties of eggs to microbial resistance

Eggs have natural barriers for prevention of microbial contamination and growth, namely physical protectors, such as cuticle and shell and membranes, and chemical inhibitors, such as lysozyme and conalbumin (Cousin, 2000). However, the deterioration of natural protective barriers of the eggshell, such as the damage of cuticle, and the impact of processing could induce microbial migration from shell into the egg content. For instance,

Cotterill (1986) and The United States (US) Department of Agriculture (1999) recommended that wash water temperature should be at least 11.1°C warmer than the egg temperature, as well as that stating by the Australian Standard (AS) 2996-1987 that wash water temperature should be at least 12°C higher than the egg temperature in order to prevent microorganisms on the shell being drawn through the pores into the egg contents

(Srikaeo and Hourigan, 2002; ICMSF 2005).

The competency of the defensive barriers of eggs to prevent microbial contamination relies upon the soundness of the egg structure. A fracture of the shell partially damages the antimicrobial system, whilst the ability to penetrate through the yolk membranes entirely damages the antimicrobial system (Board and Tranter, 1986). The ability to overcome the defensive barriers through the influence of any exogenous and/or operational factors results

20 in contamination of eggs and egg products by spoilage and pathogenic organisms (Cox,

2001). The structure of an egg is shown in Figure 2.1.

Figure 2.1 Structure of an egg (Cox, 2001).

2.4.1 Cuticle

The outermost glycoprotein layer, approximately 10-30 μm thick, which typically covers most if not the entire outer surface of the shell, is the fundamental frontier of protection and the most significant physical barrier against microbial penetration from the environment or the surface of the shell into the egg content (Sparks, 2000; Cox, 2001). The structure of cuticle after oviposition is fragile, sponge-like and moist due to the open structure of cuticle and water retained in the cuticle and pore canals. Hence, this allows any bacteria on the surface of the shell to migrate from the shell into the shell membranes. However, the cuticle is usually dried up prior to bacterial exposure to the shell (Baker, 1974; Cox, 2001; ICMSF,

21 2005). Bacteria governing on the eggshell are not prone to dry and antagonistic conditions of the shell surface. Hence, the predominant species present on the shell of eggs are Gram- positive bacteria, such as Bacillus, Micrococcus and Staphylococcus, as well as a few genera of Gram-negative bacteria, such as Achromobacter, Aerobacter, Alcaligenes,

Arthrobacter, Cytophaga, Escherichia, Flavobacterium and Pseudomonas (Sparks, 2000).

It is noted that the active lifespan of the cuticle after oviposition is 96h. The cuticle provides the shell with a shield from water entry into the eggs as well as water loss, which subsequently protects the eggs from microbial penetration (Solomon et al., 1994; ICMSF,

2005).

2.4.2 The true shell

The shell composes of an interwoven mesh of glycoprotein fibres, whose pores of the mesh are pervaded and surrounded by calcium salt crystals. The shell is presumed to provide a great physical protection to microbial contamination into the eggs due to its hard texture and impenetrable appearance (Cox 2001). Furthermore, the shell of eggs does not usually sustain microbial proliferation unless eggs are stored under highly humid conditions (Board et al., 1964). However, the shell also contains thousands of pores, approximately 10-30 μm in diameter, which can promptly permit microbial entry into the eggs in the deficiency of cuticular material on the shell surface (Baker, 1974; Cox, 2001). Thus, the shell in conjunction with the cuticle provides the intact shell as one the major physical barriers.

Spoilage is more likely to occur to eggs, whose shells deposit observable dirt, than shells of eggs that appear clean (ICMSF, 2005). In contrast, eggs, whose shell surface is evidently

22 clean, can harbour a high number of microorganisms including pathogens (James et al.,

2002).

2.4.3 Shell membranes

Shell membranes are composed of the outer and inner membranes, which adjoin to the surface of the shell. They mainly consist of glycoprotein fibres (Cox, 2001). Puncture of the shell membranes escalates the likelihood of contamination of the egg contents (Board et al.,

1964).

• The outer shell membrane, approximately 50 μm thick, is composed of six layers of protein fibres, in which each layer aligns in different direction. However, the outer membrane is fairly porous and this therefore serves as a trivial barrier to microbial entry

(Cox, 2001; ICMSF, 2005).

• The inner shell membrane, approximately 20 μm thick, and consists of three layers of protein fibres. Despite its thickness being five times thinner than the outer membrane, the inner membrane acts as the second most critical barrier after the cuticle, in regards to defence against microbial incursion into the egg contents. This is due to the depth of pores and the fine structure of the inner membrane (Baker, 1974; Cox, 2001; ICMSF, 2005). The inner shell membrane also contains plenty of lysozyme, an anti-bacterial agent, in which

Gram-positive bacteria are more susceptible to lysozyme than Gram-negative bacteria

(Baker, 1974).

23 2.4.4 Albumen

The albumen offers a multifunctional barrier to chemical, biochemical and physical aspects, of microbial spoilage (Cox, 2001). The high viscosity of albumen, which is the outcome of the interaction between the proteins ovomucin and lysozyme at neutral pH, acts as a physical defence to microbial intrusion by hindering the motion of bacteria that penetrate through the shell membranes and obstruct their approach to the yolk (Sparks, 2000).

Concurrently, lysozyme, conalbumin, alkalinity (pH 9.1-9.6) and the unavailability of the indigenous protein to microbial growth of albumen serve as the key chemical defence to microbial spoilage (Brooks, 1960; ICMSF, 2005). However, the loss of carbon dioxide in albumen in a few days following oviposition, which results in an increase in pH of albumen, diminishes the interaction between ovomucin and lysozyme and consequently reduces viscosity of albumen. In spite of a decrease in viscosity of the albumen as the deterioration of a physical defence, the rise of pH of albumen itself serves as a chemical defence as well as intensifying the effectiveness of antimicrobial proteins in albumen

(Board, 1969; Sparks, 2000; Cox, 2001). Lysozyme, which is named after its ability to lyse bacterial cells, is a muramidase and splits murein layer in the peptidoglycan of bacterial cell walls (Cox, 2001; ICMSF, 2005). Gram-positive bacteria are more susceptible to lysis by lysozyme than Gram-negative bacteria, whose murein layer is sheltered by the outer cell membrane. The alkalinity of albumen enhances lysis of bacterial cells by lysozyme because it sensitised their cells. However, the efficacy of lysozyme is declined once the albumen is mixed with the yolk (ICMSF, 2005). The albumen is more lethal to Gram-negative bacteria at 39.5°C than at 30°C. However, the presence of iron in the albumen does not only deprive

Gram-negative bacteria of extermination by chemical defence in the albumen, but it also

24 promotes their growth, with the exceptions of Acinetobacter sp., Proteus vulgaris and

Pseudomonas fluorescens (Tranter and Board, 1984). Conalbumin has a significant role to chelate metal ions, especially iron, copper and zinc, markedly at the high pH of aged albumen. Hence, this makes these ions unavailable for utilisation by bacteria, which consequently confines their reproduction and growth (Board, 1969; Baker, 1974; Yadav and Vadehra, 1977; Fields, 1979; Board and Tranter, 1986; ICMSF, 2005). It is noted that many bacteria are inhibited by conalbumin, in which Gram-positive bacteria are more susceptible to conalbumin than Gram-negative bacteria (ICMSF, 2005). The unavailability of the indigenous proteins in the albumen; for instance, binding of the protein avidin with the B vitamin biotin, also restricts microbial growth (Yadav and Vadehra, 1977; Fields,

1979; Cox, 2001). Other proteins in the albumen, such as ovomucoid that inhibits trypsin; ovoinhibitor that inhibits trypsin, α-chymotrypsin, subtilisin and fungal proteinase and ovoflavoprotein that chelates riboflavin, also contribute to impede microbial spoilage of eggs.

2.4.5 Chalaziferous layer

This thick albumen layer displays as a membrane between the inner thin albumen and the yolk. The Chalaziferous layer twists to form cords to hold the yolk in the center of the egg, which makes the yolk inaccessible from the shell and membranes. Although microorganisms could overcome the exterior protective barriers of eggs, they still have to successfully travel over this thick layer to attain the nutritious yolk. Hence, chalaziferous layer provides another significant physical defence to the yolk (Baker, 1974; Cox, 2001).

25 2.4.6 Vitelline membrane

This membrane may provide slight resistance to microbial penetration into the yolk.

However, it is improbable to decrease or impede microbial spoilage, since microorganisms harbouring on the membranes still enable to approach the yolk nutrients. The significance of the vitelline membrane is to provide impermeability to water, hydrogen sulfide and other substances to entering into the yolk, whilst it assists in maintaining essentials, such as iron, in the yolk (Baker, 1974; Cox, 2001).

2.4.7 Yolk

Unlike the rest of the egg, the yolk is highly nutritious and therefore yields no resistance to microbial invasion. On the contrary, yolk promotes microbial growth if microorganisms can defeat its physicochemical barrier (Cox, 2001).

2.5 Microbiological examinations of eggs and egg products

Eggs are typically analysed for several groups of microorganisms, such as TVC or Aerobic

Plate Count (APC), coliforms, Escherichia coli, yeasts and moulds (particularly dried egg products), coagulase-positive staphylococci, and pathogens (such as Bacillus cereus,

Campylobacter, Clostridium botulinum, Clostridium perfringens, salmonellae and Vibrio parahaemolyticus) (Andrews, 2000; Ricke et al., 2001; AS 5013.7-2004).

The enumeration of quality indicator organisms (TVC, coliforms and E. coli) indicates the microbiological quality of the raw materials and the finished product, as well as ensuring in-process control (Betts, 2005).

26 2.5.1 Total viable count

The TVC or APC is one used to estimate numbers of total viable cells in food products, which is an indicator of the quality of foods, the adequacy and effectiveness of manufacturing hygiene and sanitation, and provides an estimation of shelf life. However, the results of TVC have no correlation to the safety of the product for human consumption, as TVC media is non-selective and favours growth of general mesophilic aerobes (ICMSF,

1978; Jay, 2000; Morton, 2001). Plate Count agar (PCA) is commonly used to enumerate aerobic mesophiles and does not contain any inhibitory agent, yet it is selective to fastidious organisms due to the absence of specific nutrients required for their growth

(Adams and Moss, 2000).

Aerobic Plate Count results of unpasteurized liquid egg from commercial egg breaking plants are usually in the range of 103 to 106 CFU/g. An aerobic plate count exceeding 107

CFU/g suggests that the shell eggs used had unsatisfactory microbiological quality, or the unpasteurized liquid egg underwent improper sanitation, handling and storage. On the other hand, APC of pasteurized egg products (liquid and frozen) and dried egg products should not exceed 2.5 x 104 CFU/g. Failure to meet this specification suggests the use of poor microbiological quality of unpasteurized liquid egg, improper pasteurization, or post- pasteurization contamination (Ricke et al., 2001).

Board et al. (1964) reported that the ranges of bacterial populations of the eggshell incubated at 20°C and 37°C were 1.3 x 104 to 8.0 x 106 CFU/shell and 3.5 x 104 to 1.6 x 106

CFU/shell of 130 farm eggs examined, respectively.

27 2.5.2 Spoilage index for shelf life assessment

Eggs are a good nutritious source of carbohydrate for microorganisms, which subsequently leads to microbial spoilage once microorganisms can invade into the egg contents and overcome their natural defensive barriers.

Shelf life prediction by the estimation of the spoilage time of foods can be performed by impedance. Impedance measures conductance changes in a medium caused by metabolites from microbial growth (de Boer and Beumer, 1999). Pseudomonads commonly causing spoilage in certain foods were selected as a representative organism to inoculate into several foods to predict their shelf life, as they are the primary invaders and are effortlessly detectable (Vadehra et al., 1970). Aerobic plate counts of foods inoculated with

Pseudomonas, such as ultra high temperature (UHT) low-acid foods (Coppola and

Firstenberg-Eden, 1988) and chilled fish and fish products (Koutsoumanis and Nychas,

2000), were performed simultaneously with the impedance to evaluate their shelf life.

Moreover, TVC estimation of pasteurized milk and other fluid dairy products by the impedance can be used to predict shelf life of these products (Kahn and Firstenberg-Eden,

1987).

2.5.3 Enteric indicator organisms

Enteric indicator organisms are comprised of several groups of microorganisms, such as coliforms and Escherichia coli, Enterobacteriaceae, and enterococci (Craven et al., 2003).

They have been extensively employed to a variety of foods to assess their safety for human consumption and the efficiency of processing and sanitation (Jay, 2000).

28 2.5.3.1 Coliforms

The enumeration of coliforms in food products is generally conducted in order to evaluate the adequacy of hygiene and sanitation of processing equipment and environment as well as the efficiency of heat treatment (Craven et al., 2003). The adequacy of pasteurization can be obtained through coliforms counts (Kornachi and Johnson, 2001).

Methods used for the enumeration of coliforms are based on the detection of their physiological characteristics, the ability to ferment lactose in the presence of selective agents, such as bile salts or surfactant such as sodium lauryl sulfate. Standard methods for coliform enumerations are performed by colony counts and Most Probable Number (MPN) technique (AS 5013.3-2009; AS 5013.4-2009). Violet Red Bile Agar (VRBA) is one of the most common media used to enumerate coliforms (Banwart, 1989; de Boer, 2000; Jay,

2000; Kornacki and Johnson, 2001; Craven et al., 2003). Colony counts from countable plates are calculated from the dilutions used and reported as presumptive coliforms

(Gorman and Ball, Jr., 1986; Herrera, 2001; Kornacki and Johnson, 2001; Leclercq et al.,

2002). These presumptive results are thereafter confirmed by an ability to produce gas in

Brilliant green lactose bile broth (AS 5013.4-2009). While VRBA is usually employed for pour plate method, LT broth is used to enrich and grow coliforms by MPN method. For

MPN method, presumptive coliforms are further confirmed by an observation of gas production in Brilliant green lactose bile broth (AS 5013.3-2009). Moreover, crystal violet in VRBL agar is also formulated to inhibit Gram-positive bacteria and some associated

Gram-negative bacteria (de Boer, 2000; Tortorello, 2003).

Alternative conventional methods include the use of Petrifilm series 2000 method which is comparable to cultural plate method using VRBA to enumerate coliforms in a range of

29 foods such as egg, frozen green beans, fresh sausages, a bakery product, raw minced meat and raw milk, with a correlation coefficient of 0.86 (Priego et al., 2000). Coliform counts by the Petrifilm 2000 method can be achieved in less than 12h.

2.5.3.2 Escherichia coli

Analysis of Escherichia coli is usually conducted in conjunction with coliform test. The presence of coliforms primarily indicates the inadequacy of processing, the improper sanitation, and possible post-processing contamination. Enumeration and detection of E. coli designates the exposure of foods to faeces and the possible presence of pathogenic strains of E. coli and other associated enteric pathogens (Craven et al., 2003).

The MPN procedure is widely used in standard methods (AS 5013.15-2006) to enumerate and detect E. coli in foods and environmental samples, due to the advantages of this technique, such as its relative simplicity, reproducibility of results between laboratories and the ability to enumerate and detect specific groups of organisms. However, the drawbacks of this technique have arisen, as the procedure is considered laborious, lengthy and imprecise at high populations. Therefore, the alternative methods can include direct plating and Petrifilm (de Boer, 2000; Jay, 2000). Direct plating onto eosin methylene blue (EMB) agar (Levine) for preliminary detection and differentiation of E. coli from other coliforms was recommended by Andrews (2000a) and Kornacki and Johnson (2001). Presumptive colonies of E. coli on EMB agar are subsequently confirmed by conducting the IMViC test.

These tests analyse i) indole production from tryptophan; ii) acidification of glucose in

MR-VP medium (Clarks and Lubs medium-buffered glucose broth, Oxoid) to reduce pH of the medium to <4.4, as indicated by methyl red; iii) production of acetylmethylcarbinol;

30 and iv) citrate utilisation. Typical Escherichia coli colonies possess the + + - - pattern of the

IMViC test (Adams and Moss, 2000; Jay, 2000; Pandey et al., 2000).

Niemi et al. (2003) indicated that the confirmation of E. coli should not solely rely on the indole production. Klebsiella pneumoniae consistently exhibited indole negative results, whilst 60% to 70% of Klebsiella oxytoca yielded the indole positive result. This subsequently led to a number of E. coli false positive results. Therefore, the simultaneous observation of gas production in liquid media, such as lauryl tryptose mannitole with tryptophan (LTM) broth, was recommended to assist the confirmation and decrease the indicence of false positive detection. E. coli normally survived and grew in liquid media incubated at 44°C or 44.5°C, as interpreted by gas production, whereas Klebsiella oxytoca was not adequately thermotolerant to grow in selective media at 44°C.

2.5.4 Salmonellae

Salmonella spp. are classified as a genus in the family Enterobacteriaceae, Gram-negative; straight rods; 0.3-1.0 x 1.0-6.0 μm; motile by peritrichous flagella or non-motile; able to grow in the presence and absence of oxygen; chemoorganotrophic; respiratory and fermentative metabolism; acid and often visible gas is produced during fermentation of D- glucose (Brenner, 1984).

Salmonella spp. are facultative anaerobe and usually motile with peritrichous flagella

(Brenner, 1984). They are rod-shaped bacteria in which the rod’s sizes are typically 0.7-1.5 x 2-5 μm, oxidase-negative and catalase-positive. Most strains ferment glucose with acid and gas production (Cox, 2000a).

31 Cells of Salmonella can survive under frozen and dried states for prolonged period of storage. More importantly, their cells in foods can proliferate without affecting the appearance of foods (Amaguana and Andrews, 1999). For instance, Board (1964) stated that salmonellae could multiply without creating any adverse effects than the vaguely turbid albumen.

2.5.4.1 Salmonellosis

Salmonellae are responsible for a number of different clinical syndromes, distinctly grouped as gastroenteritis and systemic disease. Systemic disease is usually associated with strains of serovars that inhabit a narrow range of hosts, such as Salmonella Dublin in cattle,

Salmonella Pullorum in poultry, and Salmonella Typhi, Paratyphi, and Sendai in humans.

Although these last three aforementioned serovars infect humans, they do not usually involve with food-borne illness from consumption of contaminated eggs or contaminated egg-containing foods (Cox, 2000a). On the other hand, the gastroenteric syndrome commonly results from most food-borne serovars occurring widely in humans and animals including those causing salmonellosis from contaminated eggs which are considered unadapted (no host preference) serovars (Jay, 2000). The symptoms can range in severity from asymptomatic carriage to severe diarrhea depending on the vulnerability of the person infected, the matrix where the strain is ingested, and the virulence of the strain (Cox, 2000a;

Sparks, 2000). Numbers of the cells on the order of 107–108 CFU/g are generally necessary for salmonellosis. Nonetheless, epidemiological evidence from a number of outbreaks has demonstrated that the infectious dose may be as little as a few cells (Cox, 2000a; Jay,

2000). Interestingly, the attribution of Salmonellosis to food groups is different between

32 countries. In particular, it has been observed that in Australia and New Zealand that higher numbers of outbreaks have been attributed to S. Typhimurium compares to Canada, the EU and US, and no outbreaks due to S. Enteritidis (Greig and Ravel, 2009).

2.5.4.2 Evidence of Salmonella outbreaks from eggs and egg-associated foods

It has been found since 1988 that eggs are one of the most frequent vehicles of Salmonella

Enteritidis inducing salmonellosis outbreaks in man through many parts of the world

(Adams and Moss, 2000; Greig and Ravel, 2009). The most common transmission vehicles of salmonellosis from food to human, approximately three-fourths, are eggs, poultry, meat, and meat products (Fagerberg and Avens, 1976; Jay, 2000). In cases of egg-borne illness caused by salmonellae, most outbreaks were associated with raw or undercooked eggs as ingredients in food products, such as homemade mayonnaise, salad dressing, ice cream and egg drink (Adams and Moss, 2000; Food Science Australia, 2008). Some cases even contributed with the consumption of cooked eggs as the case in Japan in 1988 (Cox,

2000a). Throughout the United States from 1973 through 1984, 44% of Salmonella

Enteritidis outbreaks were associated with egg-containing foods (Jay, 2000). This statement agreed with that of Andrews et al. (2001), in which illness caused by eggs contaminated with Salmonella Enteritidis were accounted for 25% of all salmonellosis reported in humans in the United States. Between January 1985 and May 1987, there were 65 food- borne outbreaks in the U.S. and interestingly 70% of the outbreaks were traced to grade A shell eggs or egg-containing foods. Moreover, in the U.S. from 1985-1991, there were

Salmonella Enteritidis outbreaks in hospitals or nursing homes where nine of 15 nursing home outbreaks caused from egg-containing dishes. It assumes that transovarian egg

33 contamination occurs in about one of 10,000 eggs produced in the U.S. according to the

USDA. Although this represents low level of contamination, this results in approximately

4.5 million final Salmonella Enteritidis-contaminated eggs in which a large number of people will greatly expose to this pathogen (Hammack and Andrews, 2000). Mayonnaise, particularly the homemade ones, is a significant transmission vehicle of Salmonella due to the survival beyond recipe ingredients and storage conditions (Radford and Board, 1993;

Adams and Moss, 2000). For example, the choice of acidulant used in the preparation such as acetic acid is more germicidal than citric acid (Radford and Board, 1993). In contrary, the recent study by Telo et al. (1999) showed that Salmonella were only isolated from the eggshell, but not from the egg contents. However, a study by Radkowski (2001) did not isolate Salmonella on the eggshell or in the egg contents of shell eggs sold in markets in

Olsztyn, Poland between June 1997 and December 1998.

There are many publications indicating food-borne illness and outbreaks from consumption of contaminated eggs and egg products throughout the world. The involved pathogens are

Aeromonas, Campylobacter, Clostridium perfringens, Listeria, Salmonella, Staphylococcus aureus and Shigella (Adams and Moss, 2000; Sparks, 2000). However, it has been documented that Salmonella spp. is the most significant species in terms of reported outbreaks of food poisoning attributed to eggs (Table 2.1). Notably, Salmonella

Typhimurium was traditionally the predominant serovar of salmonellae implicated in egg- associated outbreaks in the mid-1980s. Subsequently, Salmonella Enteritidis became the major egg-associated source of food-borne illness in human (ICMSF, 2005).

34 Table 2.1 Major egg-associated outbreaks of human salmonellosis

Year Country Vehicle Salmonella Cases serovar 2002a Spain Vanilla cream Enteritidis 1,435 pastry 1985-1995b USA Raw shell eggs Enteritidis 582 outbreaks, 34,058 cases, 70 deaths 1993c France Mayonnaise Enteritidis 751 1988c Japan Cooked eggs Salmonella 10,476 spp. 1987c China Egg drink Typhimurium 1,113 1977c Sweden Mustard Enteritidis PT4 2,865 dressing 1976c Spain Egg salad Typhimurium 702

References: a Camps et. al. (2005); b Jay (2000); c Cox (2000a)

Recent Australian foodborne illness associated with eggs was reported by Food Science

Australia (2008). Of the twenty-seven Australian foodborne disease outbreaks reported between October and December 2007, thirteen (48%) were caused by Salmonella

(Department of Health and Ageing, 2007). Over one-third (5 out of 13) of these Salmonella outbreaks were associated with the use of raw or undercooked eggs in combination with poor food handling techniques. Foods implicated included uncooked cheese/cream cake containing raw egg (prepared in a bakery) and raw egg aioli dressing (prepared in two different restaurants).

In January 2008, seventy-eight cases of salmonellosis were linked to consumption of aioli dressing containing raw egg yolks served at several catered functions in Tasmania

(Department of Health and Ageing, 2008a). As a result of this and several previous outbreaks (Department of Health and Ageing, 2008b), the Tasmanian Government has

35 implemented new legally enforceable requirements for businesses manufacturing raw egg products, designed to prevent further outbreaks of salmonellosis associated with these products (Tasmanian Department of Health & Human Services, 2008). In late 2007 the

NSW Food Authority and NSW Health investigated three outbreaks of salmonellosis in

NSW, affecting up to forty-nine people. The foods linked to these outbreaks all contained raw egg and were prepared in private residences, such as caesar salad dressing, chocolate mousse, eggnog and homemade fried ice cream. In response to these outbreaks the

Authority issued a media release to remind the public about safe use of eggs (NSW Food

Authority, 2008).

2.5.4.3 Conventional detection of salmonellae in foods

Identification of salmonellae is based on certain criteria, such as diagnosis of the family

Enterobacteriaceae; diagnosis of the genus Salmonella; diagnosis of the serotype; determination of the antibiotic susceptibility pattern, and further study of the biovar and phagovar, if indicated. Selective procedures are essential for the isolation of Salmonella from samples containing mixed bacterial flora (such as faecal sample, foods, and environmental samples). The key to the isolation of Salmonella is the enrichment step, which increases the ratio of Salmonella cells compared to other bacterial cells during incubation. This can be achieved by using liquid nutrient media containing selective agents to inhibit or suppress growth of non-Salmonella competitors. The use of enrichment is crucial when low number of salmonellae is present in the sample (Brenner, 1984).

The conventional cultural technique (Figure 2.2) comprises of a five-stage procedure, as described in the following Sections (Amaguana and Andrews, 2000).

36 ••• Resuscitation

This first step is performed with a non-selective broth medium, which provides for the non- inhibited growth of indigenous bacterial flora and the resuscitation and proliferation of stressed or injured Salmonella spp. cells from exposure to food processing to detectable levels. Lactose broth and buffered peptone water are among the most widely used pre- enrichment media for foods including eggs and egg products. However, lactose broth has a disadvantage according to specificity because it does not provide specific enrichment only for Salmonella spp., but also a self-limiting environment for non-Salmonella spp. organisms. Other alternative pre-enrichments include distilled water with 1% brilliant green for instant non-fat dry milk, nutrient broth for frosting mixes and trypticase soy broth for spices. However, Jay et al. (2003) stated that lactose broth would favour the growth of lactose-fermenting competitors to Salmonella. In addition, the carbohydrate in pre- enrichment media does not induce the effectiveness to multiply Salmonella numbers.

Therefore, this suggests that nutrient broth and buffered peptone water are suitable for pre- enriching most foods. Bailey and Cox (1992) described that Universal pre-enrichment broth can be simultaneously used to recover Salmonella and Listeria. High recovery (10 CFU) of heat-injured Salmonella and Listeria is achieved in both pure and mixed flora and several different foods.

37 Figure 2.2 Flow diagram of detection of Salmonella. Source: Jay et al. (2003).

38 • Selective enrichment

Selective enrichment broths are used to increase the population of Salmonella spp. organisms, while restricting the growth of other organisms in the food samples. Three of the more commonly used selective enrichments are Rappaport-Vassiliadis (RV) medium, selenite cystine (SC) broth and tetrathionate (TT) broth. As none of these selective enrichments is capable of recovering all of the 2,400 recognised serovars, the various cultural methods typically recommend the use of mixed two selective enrichment broths

(AS 5013.10-2006). The incubation temperature and period are critical to achieve optimal performance of the respective media. For foods with low microbial loads, SC broth is normally incubated at 35°C, TT broth at 35°C, and RV medium at 42°C, while elevated incubation temperature for high microbial loaded foods in order to enhance the recovery of

Salmonella spp. and increase the inhibition potential towards competitive micro flora, thus

RV medium at 42°C, and TT broth at 43°C. Nonetheless, the amount of SC broth use is restricted due to its toxicity and it is classified as a hazardous waste by the US

Environmental Protection Agency regarding to its critical toxic level of selenium containing.

• Selective plating

This step begins with streaking the selective enrichment broths on to selective solid media to isolate and differentiate Salmonella spp. in which the selective plating agars normally contain ingredients to suppress the growth of competitive non-Salmonella spp. Commonly used agars are brilliant green agar (BGA), xylose lysine Tergitol-4 agar (XLT4), bismuth sulphite agar (BS), Hekton enteric agar (HE), xylose lysine desoxycholate agar (XLD) and

39 Chromogenic agar. It is also recommended that two or more combined agars should be used to ensure all atypical strains are detected.

• Biochemical testing

Presumptive positive Salmonella spp. colonies from selective plating media are then tested for typical biochemical characteristics. Suspect colonies from the selective agar plates are transferred to tubes of differential agar for the preliminary biochemical characterisation of the isolates. Biochemical tests such as urease, lysine decarboxylase, dulcitol broth, KCN

(potassium cyanide) broth, malonate broth and indole reactions are usually sufficient to obtain presumptive identification of Salmonella spp. The results of these tests provide a relatively reliable indication of Salmonella spp. contamination in the food.

• Serological confirmation

To confirm the presence of Salmonella spp. the isolates are characterised serologically by determining their antigenic composition. The antigens are classified as somatic (O) and flagellar (H). The somatic (O) antigens are determined by performing a slide agglutination test with somatic (O) antisera and growth from an agar culture of the food. The flagellar

(H) antigens are determined by performing an agglutination test with the flagellar (H) antisera and a formalised saline infusion broth culture in a test tube. This test can also be performed using the Spicer-Edwards flagellar (H) antisera test. The test is positive if a specific somatic group (O) and Spicer-Edwards reactions are obtained in one or more TSI cultures.

40 2.5.5 Microbiological testing and specifications of eggs and egg products

Microbiological testing and specifications of eggs and egg products are shown in Table

2.2.

Table 2.2 Microbiological testing and specifications of eggs and egg products

Products Tests Specifications References (CFU/g)

Interior contents APC <10 Ricke et al. (2001) of shell eggs

Unpasteurized liquid egg APC 103 - 106 Ricke et al. (2001) (Egg contents separated Coliforms 102 - 105 Ricke et al. (2001) from shell eggs) Salmonellae <1 Ricke et al. (2001) Yeast and moulds <10 Ricke et al. (2001)

Pasteurized egg products APC <2.5 x 104 Ricke et al. (2001) Coliforms <10 Ricke et al. (2001) Salmonellae Nil Ricke et al. (2001); Food Standards Australia New Zealand (2002a) Yeast and moulds <10 Ricke et al. (2001)

2.6 Implementation of rapid methods for microbiological testing

Conventional microbiological methods for the detection of food-borne pathogenic organisms suffer from several disadvantages, such as, labour intensity, poor timeliness, and tediousness (Steegmans, 1994/1995), and insufficient sensitivity and specificity (Bennett et al., 1998; Manzano et al., 1998). As Hazard Analysis of Critical Control Points (HACCP) is being implemented increasingly in food manufacturing, manufacturers demand faster microbiological results to decrease the release-time of perishable food products as well as 41 reducing costs of holding these food products. As a result, automated rapid methods can facilitate screening of large numbers of samples (Steegmans, 1994/1995). Slow and cumbersome conventional methods neither allow rapid evaluation of raw material on delivery nor on-line control measures during processing. End product testing by conventional methods often permits only a retrospective assessment of the food’s microbiological condition, since many foods are perishable. Therefore, much effort has been made to develop methods, which enable a more rapid and automated estimation of the microbiological quality of foods (Watkins, 1999; Upmann and Bonaparte, 2000). The benefits of automated pathogen testing was described by Maucci (2005). Various techniques including rapid assays for detection of salmonellae in foods (Feng, 1992;

Blackburn, 1993; Giese, 1995; Giese, 1998; Andrews et al., 2001; Jay et al., 2003) and poultry feeds (Ricke et al., 1998) have been reviewed.

Fung (1995), Fung (1997), AS/NZS 4659.3-1999, de Boer and Beumer (1999), Feldsine et al. (2002), and Cox and Fleet (2003) described the criteria of developing and accomplishing the ideal rapid and/or automated food-borne microbiological assay system.

Details are shown in the following Sections.

2.6.1 Performance indicator for evaluation of qualitative alternative method

Performance indicators required for evaluation of qualitative alternative methods are sensitivity rate, specificity rate, false negative rate and false positive rate (Table 2.3) and details are described in the following Sections (AS/NZS 4659.3-1999; de Boer and

Beumer, 1999; Feldsine et al., 2002).

42 Sensitivity rate

Sensitivity is the probability that the alternative method will correctly detect the target organism present in food samples, given that these samples contain this organism. The cultural reference and the alternative assays are simultaneously performed and their results are compared.

Sensitivity rate (%) = N11 x 100

N11 + N12

Concurrently, sensitivity is also defined as the limit of detection or the lowest concentration of target organism that can be reliably determined by the alternative method (Cox and

Fleet, 2003).

Table 2.3 Calculation of performance indicators after generalised categorisation of test samples

b Test results Status of test samplesa Positive Negative Total

Positive N11 N12 N1• Negative N21 N22 N2• Total N•1 N•2 N = N1• + N2• or N•1 + N•2 a = Status defined by reference method. b = Test result is defined as result of the alternative method. For sensitivity calculation, results after confirmation are used.

43 N = Number of results in any particular cell. First subscript represents the row, whilst second subscript represents the column. Test results of each cell are defined in the subsequent Section.

• N11 (row 1, column 1) = An agreement of positive results from reference and alternative methods.

• N12 (row 1, column 2) = Positive results given by reference method, whilst negative results given by alternative method, or false negative results given by the alternative method.

• N21 (row 2, column 1) = Negative results given by reference method, whilst positive results given by alternative method, or false positive results given by the alternative method.

• N22 (row 2, column 2) = An agreement of negative results from reference and alternative methods.

Adapted from Feldsine et al. (2002).

Specificity rate

Specificity of the alternative assay is defined as the ability of the alternative method to distinguish between target organism and the other organisms in food samples. The cultural reference and the alternative assays are simultaneously performed and their results are compared.

Specificity rate (%) = N22 x 100

N22 + N21

44 False negative rate

The false negative rate of the alternative assay is the probability that food samples tested contain target organism, when they have been classified as negative by the alternative assay.

False negative rate (%) = N12 x 100

N11 + N12

Or;

100 – sensitivity rate

False positive rate of the MicroFoss assay for coliforms

The false positive rate of the alternative assay is the probability that food samples tested do not contain target organism; however, they have been classified as positive by the alternative method.

False positive rate (%) = N21 x 100

N22 + N21

Or;

100 – specificity rate

Hammack and Andrews (2000) stated that rapid methods that exhibited high sensitivity and specificity in comparison to the cultural method, particularly when the cultural and rapid

45 methods share the same pre-enrichment strategy, can be used as an alternative screening tool.

2.6.2 Performance indicator for evaluation of quantitative alternative method

Performance indicators required for evaluation of quantitative alternative methods are repeatability and standard deviations. The repeatability value is the value below which the absolute difference between 2 single test results obtained under repeatability conditions may be expected to lie within 95% probability (Feldsine et al., 2002).

The coefficient of repeatability (CR) can be calculated as 1.96 times the standard deviations of the differences between the two repeated measurements using one single method (Bland and Altman, 1999).

Coefficient of repeatability = 1.96 x Standard deviation

If the values of the test results are between mean + CR, this indicates that the method is repeatable at 95% confidence interval.

2.6.3 Test for significant difference of positive proportions for the cultural and

alternative assays

McNemar’s test [a Chi square (χ2) test] is used to compare the proportions confirmed positive for target organism between reference and alternative methods (Feldsine et al.,

2002; Silbernagel and Lindberg, 2002). Together with performance indicators, these tests provide a total assessment of the alternative method (Feldsine et al., 2002).

46 2 2 χ value = (DN12 – N21D - 1) ; degrees of freedom = 1

N12 + N21

Interpretations of χ2 value obtained are described in the subsequent Sections.

• A χ2 value <3.84 indicates that the proportions positive for the alternative and cultural methods are not statistically different at the 5% level of significance (P>0.05).

• A χ2 value >3.84 indicates that the proportions positive for the alternative and cultural methods differ significantly at the P≤0.05 (Feldsine et al., 2002).

2.7 The MicroFoss method

The MicroFoss (currently ‘the SolerisTM’) is a computerised instrument designed to rapidly detect microbial contamination in industrial samples. It can be applied to foods, beverages, cosmetics, toiletries, and pharmaceuticals.

The system has applications in the detection of the presence of various groups of organisms

(such as TVC, coliforms, Escherichia coli, yeast and moulds and lactic acid bacteria).

Other tests that can be performed by the system are the detection of spoilage organisms, shelf-life assessment, microbial limits and preservative challenge test, environmental sponges and swabs for the presence of certain organisms such as Listeria spp.

The system is based upon the detection of the metabolic processes of the organisms, by combining dye technologies and optical sensors. The MicroFoss technology is protected by

US Patent no. 5366873.

47 2.7.1 Principles

2.7.1.1 Basic principles

The principle of the MicroFoss system is similar to that of the reflectance colorimeter, the

Omnispec bioactivity monitor system (Wescor, Inc., Logan, UT), which monitors changes in dye pigmentation of the medium formulated with appropriate dyes indicating growth and metabolism of microorganisms presenting in a sample. Colour changes of the medium detected by dye indicators are a result of changes in pH or redox potential of the medium, or the presence of compounds with free amino groups. Samples are placed in microtitre wells or other types of containers and are scanned by an automated light source with computer interface during the growth stages (0-24h). The change of colour can be monitored similar to impedance curve and conductance curve. This technology offers its applications for estimation of microbial numbers, coliforms, and yeast and mould counts for assessment of quality and shelf life prediction of milk and dairy products, evaluation of starter activity, detection of bacteriophage, and assessment of milk abnormality (Vasavada,

1993; Fung 1997).

The key principle of the impedance method to estimate populations of viable microorganisms based on the relationship between microbial metabolic activity and electrochemical changes in a growth medium. Microbial growth results in metabolising large, comparatively uncharged molecules, such as sugars, proteins and fats, to smaller, highly charged molecules, such as lactic acid, amino acids and fatty acids, respectively. As a result of production of these highly charged molecules, a decline in the electrical impedance of the growth medium occurs, which can be detected and measured by the system. This consumes less time than that required for visible growth of microorganisms on

48 the agar plates. Hence, the impedance method offers an alternative rapid assay to estimate total viable microorganisms in foods (Mendonca and Juneja, 2000).

Calibration curves are generated from a relationship between the impedance detection time

(IDT) and a parameter of a comparison method, such as standard plate count (SPC). The results of approximately 80 to 100 food samples, whose microbiological populations range over various log10 cycles (log CFU/g), are used to generate a valid calibration curve. Linear or quadratic regression analyses are usually used to evaluate correlation of the impedance and comparison methods. However, care must be taken when analysing food samples with high (107 CFU/mL) and low (<10 CFU/mL) populations, since the detection threshold is rapidly accomplished for high populations and more sampling errors and encountered for low populations. As a result of more sampling errors, this escalates scattering of data points. Once the accurate, reliable calibration curve is established, it can be applied to estimate TVC of foods, evaluate generation times of contaminating microorganisms and grade food samples accordingly to specifications (Mendonca and Juneja, 2000).

2.7.1.2 Principles of the MicroFoss method

The technology is based on monitoring changes in the chemical characteristics, such as pH, redox, and enzymatic activity, of metabolic activities in a microbial liquid growth medium, where target microorganisms grow and are detected by optically sensitive reagents. Various dyes, which are indicators of changes, can be used in the system. The reagents change their spectral patterns, as the metabolic process occurs. These changes are photometrically detected by an optical instrument, the MicroFoss system, and are monitored every six minutes by the system. Details of detection of changes are as follows:

49 • pH change: As a result of microbial growth and subsequent fermentation of sugars to acid, pH of medium changes. Many commonly used media utilise pH indicators to monitor growth. Nutrient broth with the addition of bromocresol purple can be utilised for total counts. Purple broth can be utilised in conjunction with many carbon sources to explore their utilisation. Various inhibitors can be added to the system to study their effect on microbial kinetics. A variety of acidified mould media can be utilised in conjunction with pH indicator, bromocresol green. The coliform medium (CM) developed for impedance, whose bromocresol purple acts as the pH indicator, is used for the detection of coliforms.

• Redox reactions: These reactions depend upon the Eh (oxidation-reduction potential) of the medium. The method relies on the ability of microorganisms to transfer electrons to oxygen and to reduce the dyes. As the dye is reduced, resazurin sequentially passes through a spectrum of colours. It firstly converts from blue through violet and to pink, as resazurin is irreversibly reduced to resorfurin. Thereafter, the colour changes from pink to colourless, as resorfurin is reversibly reduced to dehydroresorfurin. Both resazurin and methylene blue are used in the dairy and meat industries to assess the quality of raw products. Both dyes can be used in certain media to assess the total viable microbial populations. Media containing dyes have been accepted by microbiologists and have withstood the test of time.

The key to the technology is the monitoring of these changes in a semifluid agar zone, which is separated from the liquid medium. As a consequence, this eliminates the interference of the optical pathway food particles or microbial turbidity. Light from light emitting diodes passes through the agar and a photo diode on the other side of the vial and reads colour change as microbial growth occurs.

50 Changes in colour, expressed as optical units, are sensed by the optical sensor and recorded in the computer, where the period of detecting such changes is described as detection time

(DT). It is noted that detection times are inversely related to the number of organisms in the sample.

A significant advantage of the technology is that many of the currently available media can be used with simple addition of a dye indicator.

The MicroFoss 128 system, which is comprised of four incubators that can be set at different temperatures ranging from 15°C to 55°C with temperature stabiliser system, and the MicroFoss vial are shown in Figures 2.3 and 2.4.

Figure 2.3 The MicroFoss 128 system (Neogen Corporation, Lansing, USA).

51

Figure 2.4 The MicroFoss vial and the location of detection.

2.7.2 Operational procedure

• Samples at the appropriate dilutions, no dilution for liquid samples and 1:10 dilution for viscous or solid samples, are simply added to the MicroFoss vials containing media designed for the individual testing.

• Testing configurations (Threshold, Skip, Shuteye, Cutoff-time, Caution-time, incubation temperature, duration) are individually set for each test. These configurations are either previously developed and provided by the company, or developed in this study by a number of trial combinations of three significant configurations, which are Threshold, Skip and

Shuteye, to obtain the corresponding simulated growth curve and DT.

• Judgment of test configurations involved trials and adjustments of system parameters, namely, detection threshold, skip factor and shut-eye period. Optimum combinations of these three parameters for detection were justified by corresponding DTs to the positions of detection. Detection should occur at the beginning of the accelerating phase of the

MicroFoss curve.

52 • The system regularly monitors any changes in the liquid medium and recorded every six minutes. Equation of the linear regression and its correlation coefficient (r) is generated by the MicroFoss system. The detection time obtained is subsequently converted to microbial populations of the tested food samples.

2.7.3 Commercial availability, applications and advantages

The current commercially available tests by the Soleris system are TVC, coliforms, E. coli, yeast and/or moulds, lactic acid bacteria, Enterobacteriaceae and sterility testing. These tests are available for a wide range of food commodities, such as meat (carcasses, ground meats, processed meats), egg, dairy products (milk, cheese, yoghurt, ice cream), seafood

(fresh fish, shrimp, oysters), beverages, salad dressings, chocolate and candies, frozen foods, fruits and vegetables and nutraceuticals (vitamins & minerals, supplements, nutritional bars, health drinks) (Russell, 2000; Firstenberg-Eden et al., 2004; Neogen

Corporation, 2008). Advantages of the Soleris system in comparison to traditional cultural methods are shown in Table 2.4.

Table 2.4 Advantages of the Soleris system in comparison to traditional cultural methods

Desired Soleris Time for Tests Specification Time to negative traditional Early alert time (CFU/g) result methods TVC <104 6-8h 14g 48h Coliforms <102 6-10h 14h 24h E. coli Not Detected 6-10h 20h 24h Yeast & moulds <100 30-48h 72h 5 days Lactic acid <100 30-35h 48h 3-5 days bacteria

Source: Neogen Corporation (2008).

53 Although the MicroFoss technology has focussed on the spoilage bacteria, methods and applications to pathogenic bacteria have also been considered. Pathogens such as Listeria and Salmonella spp. have been evaluated by the MicroFoss technology for food and environmental samples (Table 2.5).

The advantages of using the MicroFoss technology for the detection of pathogens are similar to those previously stated for the spoilage microorganisms with faster turn-around time to results.

54 Table 2.5 Application of MicroFoss technology to the detection of pathogenic bacteria

Year Event Reference 1996 Application of MicroFoss technology to detection of Shelef and pure cultures in naturally contaminated foods. Firstenberg-Eden, 1996 1997 MicroFoss technology used to study growth of Shelef and pathogenic and spoilage organisms and to detect the Firstenberg-Eden, presence of target microorganism by one or more 1997 biochemical reactions.

1998 Amino acid decarboxylation coupling with detection of Shelef et al., 1998 other phenotypic reaction used to detect and differentiate salmonellae from other Enterobacteriaceae.

Hydrogen sulfide used in the detection of different Shelef and Tan, 1998 strains of salmonellae.

1999 MicroFoss assay in conjunction with immunological Tan and Shelef, 1999 technology was used to detect salmonellae in liquid eggs, shell eggs, dried eggs, skim milk and chicken.

2000 MicroFoss system used to detect Listeria spp. from Firstenberg-Eden environmental swabs and sponges. and Shelef, 2000

MicroFoss system used to detect Listeria spp. in Peng and Shelef, pasteurised milk, shell eggs, fresh ground meat, hot 2000 dogs, sausage and deli meats.

2001 Detection of salmonellae and Listeria spp. in milk, shell Peng and Shelef eggs, fresh and ready-to-eat meats, and raw chicken. 2001

2004 Evaluation of the MicroFoss system for the detection of Odumeru et al., 2004 Listeria species in environmental samples

2007 High sensitivity, specificity and overall accuracy for Yan et al., 2007 recovery of Listeria from environmental samples using a novel one-ply composite tissue method of the Soleris (formerly BioSys) optical analysis system

55 2.8 Principles of detection and differentiation of salmonellae from non-salmonellae

The principles of developing a medium to detect and differentiate salmonellae from non- salmonellae, particularly other related genera in the family Enterobacteriaceae, are based on selecting the distinctive biochemical reactions of salmonellae as the differential system in the presence of potential selective agents. Selective media generally contain basic nutrients with the addition of dyes, bile salts, antibiotics and/or other chemicals to inhibit growth of non-Salmonella competitors (Jay et al., 2003).

2.8.1 Hydrogen sulfide (H2S) production as the differential system

Salmonella detection based on hydrogen sulfide production as a differential property has been employed in a number of culture media, such as XLD (Taylor, 1965), HE (King and

Metzger, 1968), lysine mannitol glycerol (LMG) agar (Cox, 1993), and Miller-Mallinson

(MM) agars (Miller and Mallinson, 2000). Development of a medium to detect Salmonella in this thesis will combine features of XLD and LMG media. XLD is widely used as one of the selective differential media for the isolation and detection of Salmonella (Fagerberg and

Avens, 1976; Jay et al., 2003; AS 5013.10-2004). LMG agar facilitates the detection of strains of Salmonella that have atypical fermentation pattern and Salmonella Typhi as well as differentiates Salmonella and Citrobacter spp. (Cox, 1993).

2.8.2 Propylene glycol acidification as the differential system

Propylene glycol-based medium as Rambach agar has been proved to provide the effective detection of non-Typhi Salmonella from other enteric bacteria. Rambach agar (RA) incorporates phenotypic characteristic of Salmonella to ferment propylene glycol, chromogenic indicator of β-galactosidase to detect lactose-utilising organisms, 56 desoxycholate to inhibit Gram-positive bacteria and neutral red as a pH indicator. Among

100 strains of non-Typhi Salmonella tested, 97 isolates can ferment propylene glycol in RA exhibiting bright red colonies within 16-24h. All strains of Salmonella Typhi tested are negative for this reaction and appears colourless on RA. Colonies of microorganisms that can utilise lactose appears blue, such as Enterobacter, Escherichia and Klebsiella. Colonies of microorganisms possessing both characteristics exhibits in violet, such as Citrobacter.

Proteus and Shigella colonies appear colourless on RA (Rambach, 1990; Freydiere and

Gille, 1991; Gruenewald et al., 1991; Dusch and Altwegg, 1993; Pignato et al., 1995a;

Pignato et al., 1995b; Cooke et al., 1999).

2.8.3 Selective agents

••• Bile salts

Bovine bile and other bile salts are one of the most commonly used inhibitors in selective media to detect Salmonella, such as Salmonella-Shigella agar and Hektoen Enteric (HE) agar (Arroyo and Arroyo, 1995). Small amount of bile salts (1.5 g/L) and crystal violet

(0.001 g/L) formulated in MacConkey agar yielded an inhibitory effect on Gram-positive bacteria (Fagerberg and Avens, 1976).

Busse (1995) described that bile salts have been formulated in commercially available media, such as deoxycholate citrate agar originally proposed by Leifson (1935) (cited in

Busse, 1995), SS agar, HE agar and XLD agar. Plain deoxycholate and XLD agars contained bile salts and citrate at the concentration of 1-2 g/L, whilst concentration of bile salts in the more selective deoxycholate citrate agar was increased to 5-10 g/L. The main

57 selective agent in deoxycholate citrate agar (Leifson) was citrate, whereas bile salts were the principal selective agent in HE agar.

••• Brilliant green

Brilliant green was found to be toxic to most bacteria, particularly has been used as a growth inhibitor to highly competitive non-salmonellae, such as Proteus spp. (Arroyo and

Arroyo, 1995). BGA is one of the commonly used media to isolate salmonellae (Jones et al.

1984). The amount of brilliant green used in commercial modified BGA complying with

ISO is 4.7 mg/L, whereas concentration of brilliant green in BGA (Kauffmann medium) was almost three times higher (Busse, 1995). Although 10 mg/L of brilliant green used in

BGA did not totally inhibit potential non-Salmonella competitors, an increase of concentration of brilliant green in BGA to 15-20 mg/L completely inhibited non- salmonellae, except for Citrobacter sp. However, concentration of brilliant green of 15-20 mg/L also significantly inhibited salmonellae (Moats and Kinner, 1974). Therefore, the use of brilliant green concentration in combination with other inhibitors shall be decreased.

••• Magnesium chloride

Busse (1995) stated that the original Rappaport medium (Rappaport et al., 1956, cited in

Busse, 1995), which contained high amounts of malachite green and magnesium chloride

(MgCl2), was developed for the enrichment of Salmonella Paratyphi and other serotypes that were comparatively hostile to brilliant green. Selectivity of Rappaport medium was based on malachite green, a relatively high ionic strength and a low pH. MgCl2 was found to neutralise the toxic effect of brilliant green for salmonellae. Vassiliadis et al. (1976)

(cited in Busse, 1995) modified Rappaport medium by decreasing the concentration of brilliant green to one third of the original. The Rappaport-Vassiliadis (RV) medium was

58 comprised of 41 μM of malachite green and 141-177 mM of MgCl2. The high concentration of MgCl2 provided hypertonic conditions or high osmotic pressure.

According to RV medium, tolerance of Salmonella to high osmotic pressure by

MgCl2.6H2O and minimal nutritional requirement of salmonellae compared to non- salmonellae provided the selectivity to the developed medium. It was recommended the effective and optimum amount of MgCl2.6H2O used should be approximately 28.6 g/L

(Peterz et al., 1989) or 29 g/L (van Schothorst et al., 1987). It was crucial to specify the correct term and amount of MgCl2 used in formulating the medium, either 29g of the hexahydrate of MgCl2 or 13.58g of the anhydrous salt per litre (Busse, 1995). Incubation temperature at 42.0 + 0.1°C in water bath was recommended for the effective use and selectivity increase of RV medium for selective enrichment of salmonellae (Peterz et al.,

1989).

••• Niaproof 4 (Surfactant)

XLT4 agar incorporated Niaproof type 4 (formerly Tergitol 4, 7-ethyl-2-methyl-4- undecanol-hydrogen sulfate, sodium tetradecyl sulfate) to enhance selectivity of the medium and minimise or eliminate false-positive results. Niaproof-4 has been found to largely inhibit or suppress the competing flora, markedly Proteus, Providencia, and

Pseudomonas, and other non-salmonellae. Although Citrobacter could grow on XLT4 agar, they could not produce H2S in 24h. Thus, colonies of lactose-positive strain of

Citrobacter remained yellow with sticky texture on XLT4 agar. Colonies of typical H2S- positive strain of Salmonella exhibited as a black or black-centered colony in smooth and creamy texture, whose yellow (acid) periphery morphs into pink (alkaline), whereas colonies of H2S-negative strain of Salmonella expressed as pinkish-yellow colonies on

59 XLT4 agar. Hence, XLT4 agar served as a good differentiation medium to distinguish between Salmonella and Citrobacter. Concentration of 4.6 mL/L of Niaproof 4 added to xylose lysine agar provided superior selectivity to the medium (Miller et al., 1991; Sherrod et al., 1995). Other competing non-salmonellae, which usually masked Salmonella colonies by crowding or overgrowth, were generally declined or destroyed. Hence, this precluded false-negative results. In spite of the ability of Citrobacter spp. to grow on XLT4 agar, their colony size was reduced and they could hardly produce black colonies within 24h (Miller et al., 1995).

Dusch and Altwegg (1995) indicated the outstanding specificity of XLT4 agar for the isolation and recovery of nontyphoid salmonellae from stool samples.

Miller and Mallinson (2000) developed a strategically balanced medium, Miller-Mallinson

(MM) agar, to improve detection of non-typhoid and typhoid salmonellae. The medium was composed of balanced sugars (cellobiose, lactose, mannitol and trehalose), proteins

(beef extract and polypeptone peptone), inhibitory agent (Niaproof 4) and chromogenic substrate (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside or X-gal). Use of nutritious peptone as polypeptone peptone with beef extract favoured detection of typhoid carriers in contaminated food. Hence, MM agar was formulated to promote detection of H2S production by salmonellae that traditional agar media might fail to detect, particularly for non-typhoid Salmonella with weak H2S production and typhoid Salmonella. This medium enhanced desirable H2S production by true salmonellae over the false H2S-positive production of non-salmonellae. The incorporation of Niaproof 4 in the medium considerably suppressed the overgrowths of Proteus and Providencia. α-lactose was formulated to prevent H2S production by non-salmonellae. Mannitol and trehalose

60 stimulated the initial growth of Salmonella including Salmonella serotype Typhi. The quantity of four sugars used balanced and supported Salmonella growth, whilst they did not interfere the alkaline condition to enhance H2S production. Interestingly, this medium did not depend on pH change from sugar fermentation, which might not clearly differentiate between Salmonella and non-Salmonella. Therefore, 5-bromo-4-chloro-3-indolyl-β-D- galactopyranoside (XGAL) was incorporated into the medium to detect lactose-fermenting organisms by producing bright bluish green colour. Tris buffer was added to simulate and maintain the alkaline condition to enhance H2S production. As a result, MM agar did not only have similar specificity to XLT4 agar, but it also had greater sensitivity than XLT4 agar.

••• Novobiocin

Moat (1978) found that using tryptic soy agar as a base greatly improved the growth of salmonellae, whilst it did not interfere the desirable differential properties of selective agents. The most recoveries and the fewest false-positive results were obtained by the use of tryptic soy-xylose-lysine (TSXL) agar with or without the addition of novobiocin.

Novobiocin was added into a number of media to detect Salmonella in order to inhibit the growth of non-Salmonella competitors. Restaino et al. (1977) indicated the incorporation of novobiocin at low level as 4 μg/mL into XLD agar, which contained 1.0 g/L of sodium desoxycholate as a selective agent, completely inhibited the growth of Proteus mirabilis.

Proteus vulgaris was completely inhibited by 5 g/L of sodium desoxycholate (Blivet et al.,

1998). The amount of novobiocin was increased to 10 μg/mL to completely inhibit the growth of Proteus mirabilis in HE agar (Restaino et al., 1977). In contrast, Citrobacter freundii was found to tolerate up to 250 μg/mL of novobiocin in both XLD and HE agars.

61 However, Citrobacter freundii was reported to maximally resist when 80 μg/mL of novobiocin was incorporated into HE agar (Hoben, Ashton & Peterson, 1973). However, the concentration of novobiocin used was recommended at 40 μg/mL to assure the efficacy not to suppress Salmonella growth (Blivet et al., 1998).

Restaino, Komatsu & Syracuse (1982) reported that 40 and 7 μg/mL of novobiocin supplemented to HE and XLD agars, suppressed the growth of some strains of Proteus mirabilis, whereas the survivors lost the ability to produce hydrogen sulphide. Meanwhile, these concentrations gave the greatest recovery of salmonellae without altering their colony morphology. Moreover, Restaino et al. (1977) and Komatsu and Restaino (1981) modified

HE and XLD agars by adding various amounts of novobiocin. Incorporating 80 and 5

μg/mL of novobiocin into HE and XLD agars, respectively, completely inhibited Proteus mirabilis, whilst it did not affect the growth and colony morphology of salmonellae.

••• Sodium mandelate

Jones et al. (1984) investigated the effectiveness of supplementation of BGA with sodium sulphacetamide (1000 mg/L) and sodium mandelate (250 mg/L) to Salmonella growth in comparison to commercially available media, such as Salmonella-Shigella (SS), deoxycholate citrate, deoxycholate citrate-Hynes modification, and bismuth sulphite agars.

Watson and Walker (1978) reported that supplementation of BGA with sulphacetamide and mandelic acid diminished growth of non-Salmonella contaminants. However, findings of

Jones et al. (1984) found that several isolates of Salmonella Dublin and Salmonella

Paratyphi B could not grow on BGA supplemented with sodium sulphacetamide and sodium mandelate and the incubation period had to prolong beyond 24h. Improvement of

Salmonella detection by this medium could be done by reducing the concentration of

62 sulphacetamide rather than mandelic acid, since selectivity of the medium relied on the efficacy of mandelic acid to inhibit most contaminants.

Bullock and Frodsham (1989) described lysine-iron-cystine-neutral red (LICNR) broth, which was primarily formulated for rapid sensitive detection of Salmonella by the impedance method. This broth well differentiated Salmonella and non-Salmonella competitors when tested with pure cultures. However, incorporation of selective agents in the medium are essential when testing with mixed flora, particularly in the presence of

Citrobacter fruendii and Enterobacter cloacae. The addition of 0.15 μg novobiocin and

0.009 μg mandelic acid was found to inhibit Enterobacter cloacae. With the exception of

Citrobacter fruendii that gave false-positive result, LICNR broth could be used to screen

Salmonella and non-Salmonella. Since lactose was excluded from the formula and fermentation did not rely on lactose, it allowed detection of lactose-fermenting strains of

Salmonella.

63 CHAPTER 3

Evaluation of the MicroFoss system for estimating total viable counts in eggs and egg products

3.1 INTRODUCTION

The shell and contents of eggs that are produced in healthy hens are indigenously sterile.

Thereafter, microflora from the environment may subsequently contaminate the egg shell and egg contents. Hence, the eggshells are subjected to contamination from faecal matter, nesting materials, air, equipment, soil and feeds. The microflora of eggshells from healthy young hens primarily consists of Gram-positive bacteria, such as species of Bacillus,

Micrococcus and Staphylococcus, whereas faecal matter is more frequently accumulated on the shell of eggs from aged hens. Therefore, the eggshells of eggs from aged hens are exposed not only to Gram-positive bacteria, but also Gram-negative bacteria, including environmental species of Alcaligenes and Pseudomonas, enteric species of Citrobacter,

Enterobacter, Escherichia, Hafnia, Proteus and Serratia, and pathogens such as

Campylobacter and Salmonella (Board et al., 1964; Board and Tranter, 1986; Banwart,

1989; Sparks, 2000; Cox, 2001; Ray, 2004). As eggs age, their natural defensive barriers deteriorate. If the shell of eggs is impaired and eggs are stored under relatively humid conditions, this will allow microorganisms deposited on the eggshells to penetrate into the egg contents. This results in microbial spoilage and growth of pathogens. Eggshell breaking results in contamination of egg contents with a variety of microorganisms. Consequently, egg products are pasteurized in order to avoid these microbiological risks (Delves-

Broughton and Board, 2000). 64 The microflora of pasteurized egg products is dominated initially by species of Gram- positive bacteria rather than Gram-negative bacteria due to the greater heat resistance of the former species (Shafi et al., 1970; Delves-Broughton and Board, 2000; Cox, 2001; Ricke et al., 2001).

The total viable count (TVC) or aerobic plate count (APC) is one of the most common tests conducted to assess the general microbiological quality of foods. It indicates the adequacy and effectiveness of manufacturing sanitation, and provides an estimation of shelf life, though not necessarily serving as a correlate of safety for human consumption (ICMSF,

1978; Morton, 2001).

Pseudomonas spp. have been documented as major spoilage organisms of egg products

(Ayres, 1960; Board, 1969; Bhargava et al., 1975; Fields, 1979; Board and Tranter, 1986;

Jay, 2000; Cox, 2001; Ray, 2004; ICMSF, 2005). Pseudomonas spp. generally cause spoilage in shell eggs, unpasteurized liquid egg and refrigerated pasteurized egg products, and produce defects, such as green, fluorescent green and pink rots (Ayres, 1960; Board and Tranter, 1986; Banwart, 1989; Jay, 2000; Cox, 2001). As a consequence, a shelf life estimate of eggs can be evaluated by a Pseudomonas test.

Since egg products are produced on a large, continuous scale, and are highly perishable, a rapid semi-automated to automated method is needed to assess their general microbiological quality. The MicroFoss system provides a rapid, semi- automated method for estimating the total bacterial load in a product (Shelef et al., 1997; Russell, 2001;

Firstenberg-Eden et al., 2002; Odumeru and Belvedere, 2002), but its application to egg products requires evaluation.

The aim of this Chapter was to evaluate the MicroFoss system for estimating the total bacterial load in unpasteurized liquid eggs and pasteurized egg products. Such microbial 65 load can then be used to assess their general microbial quality and the efficiency of processing. Unpasteurized liquid eggs and pasteurized egg products were simultaneously analysed for total viable bacterial counts (TVC) using the standard cultural, plate count assay and detection times (DTs) using the MicroFoss assay. The TVC and DTs were correlated to form calibration curves as a basis for the enumeration of total bacterial load in eggs and egg products and their correlations were evaluated. The ability of the MicroFoss assay to estimate populations of Pseudomonas fluorescens, the main spoilage organism in egg products, was also examined.

3.2 MATERIALS AND METHODS

3.2.1 Materials

Egg samples and pasteurized egg products used in this research are described in the following Sections.

3.2.1.1 Shell eggs

Shell eggs were purchased from retail grocery stores where eggs were stored on the shelf at room temperature. All individual shell eggs used had a minimum weight of 59g each and a minimum total weight of 708g for a dozen pack size. The recommended shelf life on the package was approximately 6 weeks from the purchase date.

3.2.1.2 Commercial egg samples

Egg samples, all supplied frozen, were provided by a commercial supplier. They are divided into two types, which are described in the following Sections.

Raw egg samples

These were unpasteurized samples of: whole egg; egg yolk; and egg albumen.

66 Pasteurized egg products

Pasteurized egg products provided were

• whole egg, 10% salted whole egg, scrambled egg mix;

• 10% salted yolk, 12% salted yolk, 10% sugared yolk, low solids yolk (yolk mixed with albumen to reduce the total solids to 40%), and;

• plain albumen and albumen mixed with vegetable gum (single and double strength).

Pasteurized egg products, as supplied frozen, were estimated to have a 24-month shelf life when stored at -20°C.

3.2.2 Sample handling prior to testing

Shell eggs and commercial egg samples were handled and treated according to AS

1766.3.8-1991. Procedures are described in the following Sections.

3.2.2.1 Shell eggs

All samples were refrigerated (5°C) and analysed within three days. Prior to cracking the shells, eggs were immersed in 70% (v/v) ethanol for 10 min. Eggs were then removed and allowed to dry. After cracking the eggs, the whole egg contents were placed in a Stomacher bag (Interpath Services Pty Ltd, Sydney, Australia) and mixed using a Stomacher

(Colworth Stomacher 400, Seward Limited, West Sussex, England) to obtain an homogenate. Mixing was performed for 2 min to ensure homogeneity without formation of excessive foam. Care was taken to prevent dropping eggshell pieces into the egg contents.

Concurrently, shell eggs were broken, after which yolk and albumen contents were separated by using a sterile commercial metal separator. Care was taken to avoid inclusion of eggshell pieces into the contents and avoid mixing of yolk and albumen. The separated

67 yolks and albumens were then individually and thoroughly mixed using a Stomacher until they appeared homogenous. Lengthy mixing was avoided to prevent superfluous foam.

3.2.2.2 Commercial egg samples

Commercial raw egg samples and pasteurized egg products supplied frozen, were delivered in that condition and stored in a freezer at -20°C until tested. All frozen samples were thawed at 5°C for less than 24h.

3.2.3 Inoculum preparation

The reference strain of Pseudomonas fluorescens UNSW 036800 or ATCC 13525 was used to evaluate the MicroFoss assay for detection of this major spoilage organism in shell eggs.

This strain was received from the culture collection of the School of Biotechnology and

Biomolecular Sciences, UNSW. The stock culture of the strain was maintained on slants of tryptone soya agar (TSA) (Oxoid) at 5°C after culture at 25°C for 24-36h, and sub-cultured monthly. Before use in experiments, the culture was tested for purity by streaking onto fresh plates of TSA.

The culture of Ps. fluorescens was prepared for inoculation into whole egg homogenate by transferring a single colony of biomass from a plate of TSA into 10 mL of tryptone soya broth (TSB) (Oxoid) and incubating at 25°C for 24-36h. The broth culture was serially diluted 1:10 using 9 mL of 0.1% peptone water (bacteriological peptone, Oxoid) to give suspensions of different populations for inoculation.

68 3.2.4 Sample preparation for the TVC test

3.2.4.1 Generation of calibration curves as a basis for enumeration of TVC in eggs

and egg products by the MicroFoss assay

Commercial egg samples were prepared according to AS 1766.3.8-1991. Portions (10g) of commercial raw egg samples and pasteurized egg products were diluted 1:10 with 90 mL of

0.1% peptone water. The diluted contents were mixed thoroughly using a Stomacher until homogenous. An aliquot (1 mL) of samples at 10-1 and further decimal dilutions were used for the analyses. The cultural TVC and the MicroFoss analyses were performed without delay after sample preparation.

To create samples with various populations of bacteria, raw and pasteurized samples of each product were temperature abused by storage for up to 3 days. Development of populations of the indigenous microflora in the products was promoted by prolonged storage at 5°C or time-temperature abuse at 25°C. Samples were stored at 5°C for 1-3 days, with the exception of pasteurized albumen, which was stored for up to 5 days.

Concurrently, samples were temperature abused at 25°C by storage for 1 day, except pasteurized albumen, which was held for up to 3 days.

3.2.4.2 Detection of Pseudomonas fluorescens by the MicroFoss for TVC

Whole egg homogenates for inoculation with different populations of Ps. fluorescens were prepared according to AS 1766.3.8-1991. Homogenates were aseptically prepared from shell eggs purchased from the supermarket as described in Section 3.2.2.1, and assumed to be sterile or minimally contaminated. To obtain a wide range of TVC levels to establish a regression line, portions (10g) of whole egg homogenate in Stomacher bags were inoculated with 0.1 mL of Ps. fluorescens cell suspension containing 102 – 107 CFU/ ml.

69 The inoculated samples were diluted with 90 mL of 0.1% peptone water to obtain samples at 10-1 dilution and thoroughly blended using a Stomacher to yield an homogenate.

Aliquots (1 mL) of these homogenates were used for the analyses. Portions (10g) of uninoculated whole egg homogenates were similarly diluted and analysed as uninoculated controls. Homogenates were immediately analysed for bacteria by the cultural plating method and DTs by the MicroFoss method.

3.2.5 TVC cultural method

The conventional TVC was performed according to AS 1766.1.3-1991 and AS 1766.2.1-

1991. Volumes of 1 mL of appropriate sample dilutions were pipetted into sterile Petri dishes in duplicate. Molten plate count agar (PCA) (Oxoid), 12 to 15 mL, at approximately

45°C was poured into each Petri dish. The contents of each dish were mixed by gentle swirling and allowed to solidify. The plates were then inverted and placed in an incubator at 30°C + 1°C for 72 + 2h. Plates with 25 to 250 colonies were selected for counting.

3.2.6 The MicroFoss TVC test

An aliquot (2 mL) of each egg sample at the 10-1 dilution was added to a TVC vial (Foss

Pacific, North Ryde, Australia). The vials were mixed by gently inversion 10 times, then inserted into the MicroFoss instrument and monitored during their incubation, according to the manufacturer’s instructions.

Judgment of testing configurations involved trials and adjustments of system parameters, namely, detection threshold, skip factor and shut-eye period. Optimum combinations of these three parameters for detection were justified by corresponding DTs to the positions of

70 detection. Detection should occur at the beginning of the accelerating phase of the

MicroFoss curve. Configurations for analyses of yolk are different from those for testing of whole egg and albumen. Details of test configurations are shown in Table 3.1.

Table 3.1 Test configurations of the MicroFoss TVC assay for eggs and egg products

Tests Threshold Skip Shuteye Cutoff Caution Temperature Duration (°C) (h) TVC (whole egg & 8 1 25 8 8 30 36 albumen) TVC-yolk 14 1 35 10 10 30 36

Source: Derived from the MicroFoss manual.

3.2.7 Statistical analyses

Correlations of the calibration curves, which were a linear regression between the cultural

TVC counts (log CFU/g) and the MicroFoss DTs, were determined by correlation coefficients (r value) processed by the MicroFoss system. Linear equations were generated by the system.

Quadratic regression was alternatively applied to represent and possibly improve correlations of the relationships between the cultural TVC and the MicroFoss DTs, which was determined by the r value. This was processed by SPSS 14.0 software for Windows

Evaluation Version.

71 3.2.8 Identification of the isolates of predominant bacteria in raw whole egg

Isolates of predominant bacterial colonies on PCA plates used for TVC analyses of commercial raw whole egg samples were broadly identified to indicate the possible presence of indigenous Ps. fluorescens as the main spoilage species in eggs. Bacterial isolates were streaked onto PCA plates and incubated at 30°C for 24-36h to prepare for use in identification tests. The basic identification tests are described in the following Sections.

• Gram stains: Gram stains were performed on fresh cultures using the method of the

Australian Standard AS 5013.14-2004: Microbiology of food and animal feeding stuffs-

General rules for microbiological examinations (Standards Australia committee FT-004,

Food Microbiology, 2004). This staining of bacterial cells allows description of the morphology of the bacteria and classification of them into two groups as a function of whether or not they are capable of retaining the violet stain of Crystal violet under the test conditions. The division results mainly from differences in the structure of the cell walls of the two groups and it is correlated with other major differences between the two groups.

Reagents used in order are Crystal violet solution, iodine solution, 95% ethanol, and safranine solution. Bacterial cells which appear blue or violet are termed Gram-positive, whilst those which are coloured dark pink to red are termed Gram-negative.

• Morphology: Morphology of the Gram stained isolates was examined by microscopy

(Olympus CH30). Procedure and maintenance followed the method of AS 5013.14-2004.

• Catalase test: Catalase tests were performed on fresh cultures using the method of AS

5013.14-2004. Cover the culture with 1 mL to 2 mL of a 10 volume hydrogen peroxide solution. Observe immediately and after 5 min for formed oxygen bubbles, which results from decomposition of catalase enzyme into water and oxygen.

72 • Oxidase test: Oxidase tests were performed on fresh cultures using the method of AS

5013.14-2004. The detection of oxidase is carried out by the change in colouring of a compound at the time of oxidation under the action of this enzyme. In the case of the presence of oxidase, a violet to purple colour appears within a period of between 5s and

10s. If the colour has not changed after 10s, the test is considered as being negative.

• Oxidation/Fermentation of glucose test: Two tubes of Hugh & Leifson medium (1953) were used. Fresh cultures were stabbed into both tubes, in which one tube was covered with sterile mineral oil, and both tubes were subsequently incubated at 30°C for up to 48h in an aerobic atmosphere. An uninoculated control (green medium) was also incubated at 30°C for 48h. Glucose metabolism by oxidation or fermentation was checked at 24-h intervals.

Oxidative organisms only produced an acid reaction in the open tube, which was indicated by the change of medium colour from green to yellow, whereas no growth or no acid formation was observed in the covered tube. Blue colour of the medium may be initially observed at the surface of the open tube for weak or slow oxidative reaction due to the alkaline reaction from utilisation of peptone in the medium, which would eventually change to acid (yellow). Fermentative organisms produced an acid reaction in both types of tubes, which resulted in colour change of the medium from green to yellow. Organisms that possessed neither oxidative nor fermentative reaction produced slight alkalinity in the open tube (blue-green colour), whilst the covered tube remained unchanged (green).

• Motility test: Fresh cultures were stabbed into motility medium and incubated at 30°C for up to 48h. Motility was checked at 24-h intervals. Motility medium is comprised of nutrient broth (Oxoid), 0.2-0.4% (w/v) agar (Davis Gelatine (Australia) Co., Botany, Australia) and

73 0.005% triphenyl tetrazolium chloride (Aldrich Chemical Company, Inc., Milwaukee,

USA).

• Aerobic/anaerobic growth: Fresh cultures were simultaneously streaked onto two separate fresh plates of nutrient agar (NA) (Oxoid). Plates were separately incubated under aerobic and anaerobic conditions at 30°C for up to 48h.

3.3 RESULTS

3.3.1 Sample dilution for testing in the MicroFoss system

Egg samples were evaluated as to whether they required dilution prior to addition into the

MicroFoss vials as previous studies have shown that the food matrix can affect microbial growth in the vials as well as have an independent effect on the chemistry of the indicator dye (Shelef and Eden, 1996). For this purpose, samples of commercial unpasteurized whole egg, undiluted or diluted 1 in 10, were inoculated into the MicroFoss TVC vials for analysis as well as confirmation of counts by the cultural method. The MicroFoss detection curves of the undiluted and 10-1 diluted portions of each sample were then compared and are shown in Figure 3.1.

Figure 3.1 shows the MicroFoss curves for detection of 105 CFU/g of total bacteria in the undiluted and 10-1 diluted portions of commercial unpasteurized whole egg yielding detection times of 2.3h and 7.9h, respectively.

74 Figure 3.1 The MicroFoss curves for detection of 105 CFU/g of total bacteria in the undiluted (DT 2.3h) and 10-1 diluted (7.9h) portions of commercial unpasteurized whole egg.

Although the undiluted and 10-1 diluted samples of commercial unpasteurized whole egg yielded similar metabolic growth profiles and curve amplitudes ranging from 200 to 600 optical units, these portions gave different detection times of 2.3h and 7.9h, respectively.

Detection time should occur at the beginning of the accelerating phase of the MicroFoss curve or log phase of the normal bacterial growth curve, as the organisms presented in samples proliferated to reach the detection threshold of 106-107 CFU/mL. This showed that the egg matrix in the undiluted sample caused an early increase in the optical unit that did not correlate with bacterial growth. Hence, these findings indicated that egg samples for analyses in the MicroFoss system should be at 10-1 dilution.

75 3.3.2 Application of the MicroFoss system to estimate the TVC in unpasteurized

egg

3.3.2.1 Unpasteurized whole egg

Commercial raw whole egg samples (Section 3.2.1.2) were thawed and then used for estimations of TVC by plating onto PCA and analyses of DTs by the MicroFoss system.

The two measurements were used to form a calibration curve as a basis for subsequent enumeration of TVC in unpasteurized whole egg samples. After thawing, raw whole egg samples generally gave TVCs of 104-105 CFU/g. In order to obtain product with higher populations, some samples (20 samples) were stored at 5°C for 24-72h to give populations of 104-106 CFU/g, whilst the others (5 samples) were deliberately temperature abused by storage at 25°C for 24h to reach populations of 107 CFU/g (Section 3.2.4.1).

Figure 3.2 shows representative MicroFoss detection curves for raw whole egg samples containing 104 CFU/g, 105 CFU/g, 106 CFU/g and 107 CFU/g, in which these samples gave detection times of 8.3h, 7.5h, 4.4h and 2.3h, respectively. As expected, samples with higher populations of bacteria gave curves with shorter detection times.

76 TVC 104 CFU/g 8.3h

TVC 105 CFU/g 7.5h

TVC 106 CFU/g 4.4h

TVC 107 CFU/g 2.3h

Figure 3.2 The MicroFoss detection curves for total bacteria in raw whole egg samples exhibiting TVCs of 104 CFU/g (8.3h); 105 CFU/g (7.5h); 106 CFU/g (4.4h) and 107 CFU/g

(2.3h), respectively.

Figure 3.3 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 50 samples of commercial raw whole egg covering the range of 4 log CFU/g to 7 log CFU/g, yielding a correlation coefficient of –0.90. Detection times were inversely proportional to the initial populations in the samples, and ranged from DT of 2h for 107 CFU/g to a DT of 9h for a sample with

104 CFU/g.

77 Figure 3.3 Linear regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial raw whole egg. The regression is described by the equation log CFU/g = 8.155 – 0.426 (DT), (r = -0.90, n = 50).

Salvat and Blivet (2000) stated that a declining exponential function was more appropriate to describe the relationship between TVCs and impedance measurements than a linear regression. An exponential function as a quadratic regression was therefore alternatively applied to correlate and possibly improve the relationship between the cultural TVC results and the MicroFoss DTs. Figure 3.4 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 50 samples of raw whole egg, yielding a correlation coefficient of –0.91.

78     !  "  #

  $ % 

     

      

Figure 3.4 Quadratic regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for commercial raw whole egg. The regression is described by the equation log CFU/g = 8.7888 - 0.7162 (DT) + 0.0279 (DT)2, (r = -0.91, n

= 50).

Quadratic regression did not markedly improve correlation of the cultural TVC results and the MicroFoss DTs compared to linear regression analysis, as determined by their correlation coefficients, since the trend of the entire paired results appeared linear. Hence, linear regression was suitable to represent the relationship between the cultural TVC and

DTs used to form the calibration curve as a basis for enumeration of TVC in raw whole egg by the MicroFoss assay.

3.3.2.2 Unpasteurized yolk

Commercial raw yolk samples (Section 3.2.1.2) were thawed and then used for estimations of TVC by plating onto PCA and analyses of DTs by the MicroFoss system. Raw yolk samples generally gave TVCs of 103-104 CFU/g. In order to obtain product with higher

79 populations, some samples were stored at 5°C, whilst the others were deliberately temperature abused by storage at 25°C (Section 3.2.4.1).

Figure 3.5 shows representative MicroFoss detection curves for raw yolk samples containing 103 CFU/g, 104 CFU/g, 105 CFU/g, 106 CFU/g and 107 CFU/g, in which these samples gave detection times of 8.7h, 6.8h, 5.5h, 4.5h and 3.8h, respectively.

Figure 3.5 The MicroFoss detection curves for total bacteria in raw yolk samples exhibiting TVCs of 103 CFU/g (8.7h); 104 CFU/g (6.8h); 105 CFU/g (5.5h); 106 CFU/g

(4.5h) and 107 CFU/g (3.8h), respectively.

Figure 3.6 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 87 samples of commercial raw yolk, with DTs ranging from 3h for a sample containing 107 CFU/g to 13h for a sample with 102 CFU/g, yielding a correlation coefficient of –0.89.

80 Figure 3.6 Linear regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial raw yolk. The regression is described by the equation log CFU/g = 8.190 – 0.472 (DT), (r = -0.89, n = 87).

Figure 3.7 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 87 samples of raw yolk yielding a correlation coefficient of –

0.91.

 & '  ! 

"  #

    $ % 

     

      

Figure 3.7 Quadratic regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for commercial raw yolk. The regression is described by the equation log CFU/g = 9.8538 -1.0813 (DT) + 0.0448 (DT)2, (r = -0.91, n =

87). 81 Quadratic regression did not substantially improve correlation of the cultural TVC results and the MicroFoss DTs compared to linear regression analysis, as determined by their correlation coefficients, since the trend of the entire paired results appeared linear. Hence, linear regression was suitable to represent the relationship between the cultural TVC and

DTs used to form the calibration curve as a basis for enumeration of TVC in raw yolk by the MicroFoss assay.

3.3.2.3 Unpasteurized albumen

Commercial raw albumen samples (Section 3.2.1.2) were thawed and then used for estimations of TVC by plating onto PCA and analyses of DTs by the MicroFoss system.

Raw albumen samples generally gave TVCs of 102-104 CFU/g. In order to obtain product with higher populations, some samples were stored at 5°C, whilst the others were deliberately temperature abused by storage at 25°C (Section 3.2.4.1).

Figure 3.8 shows representative MicroFoss detection curves for raw albumen samples containing 102 CFU/g, 103 CFU/g, 104 CFU/g, 105 CFU/g, 106 CFU/g and 107 CFU/g, in which these samples gave detection times of 11.8h, 10.7h, 8.6h, 7.3h, 6.4h and 4.7h, respectively.

82 Figure 3.8 The MicroFoss detection curves for total bacteria in raw albumen samples exhibiting TVCs of 102 CFU/g (11.8h); 103 CFU/g (10.7h); 104 CFU/g (8.6h); 105 CFU/g

(7.3h); 106 CFU/g (6.4h) and 107 CFU/g (4.7h), respectively.

Figure 3.9 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 72 samples of commercial raw albumen, with DTs ranging from 4h for a sample containing 107 CFU/g to 15h for a sample with 102 CFU/g, yielding a correlation coefficient of –0.84.

83 Figure 3.9 Linear regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial raw albumen. The regression is described by the equation log CFU/g = 7.710 – 0.334 (DT), (r = -0.84, n = 72).

Figure 3.10 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 72 samples of raw albumen yielding a correlation coefficient of

–0.86.

  (   ! 

"  #

  $

%       

      

Figure 3.10 Quadratic regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial raw albumen. The regression is described by the equation log CFU/g = 9.8821 - 0.8427 (DT) + 0.0275 (DT)2, (r = -0.86, n = 72).

84 Quadratic regression did not considerably improve correlation of the cultural TVC results and the MicroFoss DTs, compared to linear regression analysis, as determined by their correlation coefficients, since the trend of the entire paired results appeared relatively linear. Hence, linear regression was suitable to represent the relationship between the cultural TVC and DTs used to form the calibration curve as a basis for enumeration of TVC in raw albumen by the MicroFoss assay.

3.3.3 Application of the MicroFoss system to estimate the TVC in pasteurized egg

products

3.3.3.1 Pasteurized whole egg products

Commercial pasteurized whole egg products (Section 3.2.1.2) were thawed and then used for estimations of TVC by plating onto PCA and analyses of DTs by the MicroFoss system.

Pasteurized whole egg products generally gave TVCs of 102-104 CFU/g. In order to obtain product with higher populations, some samples were stored at 5°C, whilst the others were deliberately temperature abused by storage at 25°C (Section 3.2.4.1).

Figure 3.11 shows representative MicroFoss detection curves for pasteurized whole egg products containing 103 CFU/g, 104 CFU/g, 105 CFU/g, 106 CFU/g and 107 CFU/g, in which these samples gave detection times of 7.0h, 5.6h, 2.4h, 2.3h and 2.3h, respectively.

As expected, samples with higher populations gave curves with shorter detection times.

85 Figure 3.11 The MicroFoss detection curves for total bacteria in pasteurized whole egg products exhibiting TVCs of 103 CFU/g (7.0h); 104 CFU/g (5.6h); 105 CFU/g (2.4h); 106

CFU/g (2.3h) and 107 CFU/g (2.3h), respectively.

Figure 3.12 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 120 samples of commercial pasteurized whole egg products, with DTs ranging from 2h for a sample containing 107

CFU/g to 13h for a sample with 102 CFU/g, yielding a correlation coefficient of –0.89.

86 Figure 3.12 Linear regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for commercial pasteurized whole egg products.

The regression is described by the equation log CFU/g = 7.622 – 0.444 (DT), (r = -0.89, n =

120).

Figure 3.13 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 120 samples of pasteurized whole egg products yielding a correlation coefficient of –0.94.

)  *    !

 "  #

  $ %

      

      

Figure 3.13 Quadratic regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial pasteurized whole egg products. The regression is described by the equation log CFU/g = 9.5660 - 1.3234 (DT) + 0.0696 (DT)2, (r = -0.94, n = 120).

87 Correlation of the cultural TVC results and the MicroFoss DTs as represented by quadratic regression was noticeably improved, with a correlation coefficient of –0.94, compared to a coefficient of –0.89 by linear regression. This was due to numerous samples yielding a range of different DTs; however, comparable TVCs, 102-104 CFU/g, resulted in a cluster of these paired measurements as a flat base of the curve. As a result, quadratic regression was more appropriate to represent the relationship between the cultural TVC and DTs of pasteurized whole egg products.

3.3.3.2 Pasteurized yolk products

Commercial pasteurized yolk products (Section 3.2.1.2) were thawed and then used for estimations of TVC by plating onto PCA and analyses of DTs by the MicroFoss system.

Pasteurized yolk products generally gave TVCs of 10-102 CFU/g. In order to obtain product with higher populations, some samples were stored at 5°C, whilst the others were deliberately temperature abused by storage at 25°C (Section 3.2.4.1).

Figure 3.14 shows representative MicroFoss detection curves for pasteurized yolk products containing 10 CFU/g, 102 CFU/g, 103 CFU/g, 104 CFU/g, 105 CFU/g, 106 CFU/g and 107

CFU/g, in which these samples gave detection times of 10.8h, 8.6h, 7.5h, 6.4h, 4.9h, 3.9h and 3.3h, respectively.

88 Figure 3.14 The MicroFoss detection curves for total bacteria in pasteurized yolk products exhibiting TVCs of 10 CFU/g (10.8h); 102 CFU/g (8.6h); 103 CFU/g (7.5h); 104 CFU/g

(6.4h); 105 CFU/g (4.9h); 106 CFU/g (3.9h) and 107 CFU/g (3.3h), respectively.

Figure 3.15 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 46 samples of commercial pasteurized yolk products, with DTs ranging from 2h for a sample containing 107 CFU/g to

12h for a sample with 10 CFU/g, yielding a correlation coefficient of –0.90.

89 Figure 3.15 Linear regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for commercial pasteurized yolk products. The regression is described by the equation log CFU/g = 8.570 – 0.580 (DT), (r = -0.90, n =

46).

Figure 3.16 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 46 samples of pasteurized yolk products yielding a correlation coefficient of –0.95.

)  * & '  !

 "

 #

  $

% 

     

      

Figure 3.16 Quadratic regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial pasteurized yolk products. The regression is described by the equation log CFU/g = 11.3981 -1.6331 (DT) + 0.0738 (DT)2, (r = -0.95, n = 46).

90 Correlation of the cultural TVC results and the MicroFoss DTs as represented by quadratic regression was markedly improved, with a correlation coefficient of –0.95, compared to a coefficient of –0.90 by linear regression. This was due to a number of samples yielding different DTs; however, similar TVCs, 10-103 CFU/g, resulted in a cluster of these paired measurements as a flat base of the curve. As a result, quadratic regression was more appropriate to represent the relationship between the cultural TVC and DTs of pasteurized yolk products.

3.3.3.3 Pasteurized albumen products

Commercial pasteurized albumen products (Section 3.2.1.2) were thawed and then used for estimations of TVC by plating onto PCA and analyses of DTs by the MicroFoss system.

Pasteurized albumen products generally gave TVCs of non-detectable to 102 CFU/g.

Among 110 samples tested, no bacteria were detected in 67 samples by either the cultural

TVC method or the MicroFoss assay. In order to obtain product with higher populations, some samples were stored at 5°C, whilst the others were deliberately temperature abused by storage at 25°C (Section 3.2.4.1).

Figure 3.17 shows representative MicroFoss detection curves for pasteurized albumen products containing <10 CFU/g, 10 CFU/g, 102 CFU/g and 104 CFU/g, in which these samples gave detection times of 14.0h, 12.3h, 10.4h and 8.8h, respectively.

91 Figure 3.17 The MicroFoss detection curves for total bacteria in pasteurized albumen products exhibiting TVCs of <10 CFU/g (14.9h); 10 CFU/g (12.3h); 102 CFU/g (10.4h) and 103 CFU/g (8.8h), respectively.

Figure 3.18 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 43 samples of commercial pasteurized albumen products, with DTs ranging from 7h for a sample containing 103

CFU/g to 15h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.89.

92 Figure 3.18 Linear regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for commercial pasteurized albumen products. The regression is described by the equation log CFU/g = 6.903 – 0.424 (DT), (r = -0.89, n =

43).

Figure 3.19 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 43 samples of pasteurized albumen products yielding a correlation coefficient of –0.89.

)  *  (   !  "  #

    $ %        $

      

Figure 3.19 Quadratic regression relationship between the log-transformed populations of TVC and the MicroFoss detection times for commercial pasteurized albumen products. The regression is described by the equation log CFU/g = 7.3624 - 0.5122 (DT) + 0.0041 (DT)2, (r = -0.89, n = 43).

93 Quadratic regression did not improve correlation of the cultural TVC results and the

MicroFoss DTs compared to linear regression analysis and gave the same correlation coefficient of –0.89, as the trend of the entire paired results appeared linear. Moreover, since the lowest TVC populations of pasteurized albumen products were <10 CFU/g, this did not allow a decreased exponential regression to be effective and observable. These findings indicated that linear regression was suitable to represent the relationship between the cultural TVC and DTs of commercial pasteurized albumen products.

Linear regression analysis was generally suitable to represent the relationship between the cultural TVC results and the MicroFoss DTs. Quadratic regression was more suitable to represent the relationship, where the paired results of the cultural TVC and the MicroFoss

DTs were not evenly distributed.

3.3.4 Application of the MicroFoss system to estimate populations of Pseudomonas

fluorescens in unpasteurized whole egg

Pseudomonas fluorescens is the most prevalent spoilage species in refrigerated eggs and egg products (Ayres, 1960; Imai, 1976; Board and Tranter 1986). The MicroFoss assay, therefore, should reliably estimate this species. Homogenates of whole egg samples, inoculated with different populations of Ps. fluorescens, were simultaneously examined by the cultural TVC and MicroFoss assays (Section 3.2.4.2).

Figure 3.20 shows representative MicroFoss detection curves for homogenates containing

Ps. fluorescens at <10 CFU/g, 10 CFU/g, 102 CFU/g, 103 CFU/g, 104 CFU/g, 105 CFU/g,

106 CFU/g and 107 CFU/g, in which these samples gave detection times of 13.2h, 10.7h,

9.3h, 8.4h, 7.5h, 6.2h, 5.5h and 3.8h, respectively. 94 Figure 3.20 The MicroFoss detection curves for whole egg homogenates inoculated with

Pseudomonas fluorescens, yielding samples with populations of <10 CFU/g (13.2h); 10

CFU/g (10.7h); 102 CFU/g (9.3h); 103 CFU/g (8.4h), 104 CFU/g (7.5h); 105 CFU/g (6.2h)

106 CFU/g (5.5h) and 107 CFU/g (3.8h), respectively.

Figure 3.21 shows the linear correlation between TVC by the plate method (log CFU/g) and the detection times estimated by the MicroFoss system for 334 samples of whole egg homogenates inoculated with different populations of Ps. fluorescens, with DTs ranging from 2h for a sample containing 107 CFU/g to 17h for a sample with <10 CFU/g yielding a correlation coefficient of –0.76.

95 Figure 3.21 Linear regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for whole egg homogenates inoculated with

Pseudomonas fluorescens. The regression is described by the equation log CFU/g = 7.630 -

0.524 (DT), (r = -0.76, n = 334).

Figure 3.22 shows the quadratic regression relationship between the cultural TVC results and the MicroFoss DTs for 334 samples of whole egg homogenates inoculated with different populations of Ps. fluorescens yielding a correlation coefficient of –0.79.

96 10

9

8

7

6

5 +   4

3

2

1

0 0 6 12 18 24 Detection times (hours)

Figure 3.22 Quadratic regression relationship between the log-transformed populations of

TVC and the MicroFoss detection times for whole egg homogenates inoculated with

Pseudomonas fluorescens. The regression is described by the equation log CFU/g = 9.834 –

1.103 (DT) + 0.035 (DT)2, (r = -0.79, n = 334).

Quadratic regression did not markedly improve correlation of the cultural TVC results and the MicroFoss DTs, compared to linear regression analysis, as determined by their correlation coefficients, since the trend of the entire paired results appeared relatively linear. Hence, linear regression was suitable to represent the relationship between the cultural TVC and DTs of whole egg homogenates inoculated with different populations of

Ps. fluorescens.

97 During the course of the investigation, many samples of commercial raw whole egg were examined by culture on PCA. Generally, four to five distinct and different colony types were observed in most of these plates. One particular colony type exhibited morphology characteristic of Ps. fluorescens as judged by comparison with growth of the reference culture of Ps. fluorescens. Clearly, this was the most prevalent colony on the plates, and represented approximately 30-40% of total populations. Representatives (30 strains) of this colony were selected and purified by re-streaking onto PCA. Pure cultures were then examined for a series of basic identification tests, as shown in Table 3.2.

Table 3.2 Presumptive identification tests for Pseudomonas fluorescens

Test Results Commercial sample Reference culture of Pseudomonas fluorescens

Gram stain Gram-negative Gram-negative (Cousin, 2000)

Colony morphology Small, single, double Straight to curved rods & cluster rods (Cousin, 2000)

Motility Motile Motile (Imai, 1976; Cousin, 2000)

Aerobic/anaerobic growth Aerobic Aerobic (Cousin, 2000)

Catalase Positive Positive (Imai, 1976; Cousin, 2000)

Oxidase Positive Positive (Imai, 1976; Cousin, 2000)

Oxidative/fermentative Oxidative-positive Oxidative-positive Glucose metabolism Fermentative-negative Fermentative-negative (Hugh & Leifson medium) (Imai, 1976)

98 Results of the identification of presumptive Ps. fluorescens found in commercial raw whole egg samples corresponded to those of the reference culture of Ps. fluorescens.

This indicated the occurrence and predominance of indigenous Ps. fluorescens in unpasteurized whole egg. Thus, this finding justified the use of a Ps. fluorescens as an inoculum culture to evaluate the MicroFoss system.

3.4 DISCUSSION

The MicroFoss system has now been applied as an alternative, rapid method to estimate the TVC in a range of foods, including ground beef (Shelef et al., 1997; Odumeru and

Belvedere, 2002), ground pork (Russell, 2001) and dairy products (Firstenberg-Eden et al., 2002). Linear regression relationships between the MicroFoss DTs and the cultural

TVC (log CFU/g) results have been established in these commodities, with correlation coefficients in excess of -0.90. These calibration curves then become the basis for subsequent routine use of the MicroFoss for estimating the microbial load and microbial quality of the product, in quality assurance programs. The MicroFoss assays could be accomplished in less than 24h, whereas approximately 48-72h are needed to obtain results by the cultural method. Moreover, the MicroFoss enabled the convenience of semi-automated management of large numbers of samples (Russell, 2001; Firstenberg-

Eden et al., 2002; Odumeru and Belvedere, 2002).

Judgment of test configurations involved trials and adjustments of system algorithm, which were detection threshold, skip factor and shut-eye period. Optimum combinations of these three parameters were justified by corresponding DTs to the positions of detection. Detection should occur at the beginning of the accelerating phase of the

MicroFoss curve or log phase of the normal microbial growth curve. The strong colour of egg yolk sensitized the detection threshold, which resulted in earlier detection than

99 when the actual detection should take place. Hence, detection threshold for the

MicroFoss analyses of yolk was desensitized by setting at 14, compared to detection threshold at 8 for analyses of whole egg and albumen. As a consequence, whole egg and albumen samples were analysed in the MicroFoss system under the same configurations, whereas yolk samples were examined under different configurations, as shown in Table 3.1.

Egg samples were evaluated whether they required dilution prior to analyses by the

MicroFoss system. Direct addition of the undiluted egg samples to the MicroFoss TVC vials resulted in non-corresponding DTs to the location where the actual detection should take place. On the other hand, DTs for 10-1 diluted samples were recorded at the beginning of the accelerating phase of the MicroFoss curves, even under several trials of different sets of test configurations. The opacity of eggs and egg matrix presumably overwhelmed colour of the inoculated MicroFoss medium, which masked and consequently disguised exhibition of the actual detection. Therefore, all egg samples required 1:10 dilution prior to analyses with the MicroFoss system. These findings agreed with results of a comparison of the MicroFoss curves for analyses of TVC in pasteurized whole egg products using undiluted and 10-1 diluted samples conducted by

MicroSys, Inc. (Ann Arbor, Michigan, USA). Dilution (1:10) of these samples was therefore instructed in the company’s manual.

Because different foods have different microflora and different matrix compositions, it is necessary to establish the MicroFoss detection time and plate count correlations as calibration curves for each product (Firstenberg-Eden et al., 2002). Calibration curves for unpasteurized liquid eggs and pasteurized egg products were separately established due to differences in their indigenous microflora. Concurrently, calibration curves to estimate TVC in whole egg, yolk and albumen were individually generated due to

100 differences in their matrix compositions and antimicrobial properties, which affected growth and detection of total bacteria in the cultural and MicroFoss media. Estimation of TVC in pasteurized albumen products in this study were problematic, considered due to the natural antimicrobial substances in albumen. Several proteins in albumen inhibit microbial multiplication. For example, lysozyme lyses cells of Gram-positive bacteria by degrading peptidoglycan in their cell walls, whilst avidin and ovotransferrin bind biotin and iron to make them bio-unavailable to microorganisms to utilize. Moreover, the alkalinity in albumen impedes microbial growth (Yadav and Vadehra, 1977; Cox,

2001; ICMSF, 2005). Pasteurization of albumen usually destroys 99% of the initial bacterial populations (Stadelman, 1994). Thus, no bacteria were detected in most pasteurized albumen samples after thawing by either the cultural or MicroFoss methods.

TVCs in pasteurized albumen products were generally lower than those in pasteurized whole egg and pasteurized yolk products that were treated under the same manners. For instance, TVCs of pasteurized albumen products that were stored at 5°C for 24-72h were <10-102 CFU/g, compared to 102-104 CFU/g and 10-103 CFU/g for pasteurized whole egg and pasteurized yolk products, respectively. Highest populations of total bacteria in pasteurized albumen products after storage at 25°C was 103 CFU/g, compared to 107 CFU/g for pasteurized whole egg and pasteurized yolk products. As a result, longer DTs were generally required for detection of microorganisms presenting in albumen than those required by whole egg and yolk at the same microbial populations.

All calibration curves in this Chapter were generated from 50 to 100 valid paired data of the cultural TVC (log CFU/g) and DTs due to the availability of commercial raw egg samples (as unpasteurized eggs) and pasteurized egg products supplied.

101 Linear regression was generally applied to correlate microbial counts and measurements from the alternative methods in several foods, such as the colorimetric method

(Manninen and Fung, 1992), and the impedance method (Lach et al., 1978; Gibson,

1988; Johansen et al., 1995; Koutsoumanis and Nychas, 2000), yielding correlation coefficients in excess of –0.80. Findings in this Chapter shows that linear regression was generally suitable to represent the relationship between the cultural TVC and the

MicroFoss DTs in unpasteurized liquid eggs and pasteurized egg products, with correlation coefficients in the range of –0.89 to –0.90, with an exception of –0.84 for unpasteurized albumen. Estimations of TVC in raw egg samples and pasteurized egg products by the MicroFoss TVC assay were completed within 2-16h, as opposed to 48-

72h required to complete the cultural TVC method. As expected, higher populations gave earlier detection times. Findings of this study suggested the MicroFoss assay as an alternative rapid method to estimate TVC in unpasteurized liquid eggs and pasteurized egg products. Previous studies correlating the cultural TVC results and the MicroFoss

DTs showed similar and higher correlation coefficients and completion of the test in less than 24h suggesting that the MicroFoss system could be used as an alternative rapid method to estimate TVC populations in other food products, such as ground beef

(Shelef et al., 1997; Odumeru and Belvedere, 2002), ground pork (Russell, 2001) and dairy products (Firstenberg-Eden et al., 2002).

However, correlation coefficients between the cultural TVC and the MicroFoss DTs of unpasteurized liquid eggs and pasteurized egg products obtained were lower than those reported in other studies (Shelef et al., 1997; Russell, 2001; Firstenberg-Eden et al.,

2002; Odumeru and Belvedere, 2002). Correlation coefficients less than –0.90 suggest detection of a diversity of flora. The impedance method, which measures conductance changes in a medium causing by metabolites from microbial growth, was previously

102 developed to use as an alternative rapid method to assess microbial quality of foods (de

Boer and Beumer, 1999; Mendonca and Juneja, 2000). Since the principles of the development of the MicroFoss method to be potentially used as an alternative rapid method are similar to those of the impedance method, certain rules of impedance method could be applied to the MicroFoss assay. Curda (2000) stated that a correlation coefficient of a linear relationship between the plate count and the impedance detection time for an analysis of a single strain could be as high as >0.97. However, a correlation coefficient of the analysis of samples containing various strains of one species decreased to approximately 0.90, whilst a correlation coefficient of 0.80 was expected from the analysis of samples containing a mixture of different species.

Hence, r values of linear relationships between the cultural TVC results and the

MicroFoss DTs of unpasteurized liquid eggs, which ranged from –0.84 to –0.90, were not greater than those of pasteurized egg products, which ranged from –0.89 to –0.90.

Lower r values obtained from raw egg samples were a result of a more diverse flora presenting in these samples, compared to higher r values yielding from the more consistent microflora of pasteurized egg products. They were heat-resistant Gram- positive bacteria, predominantly species of Bacillus, Micrococcus and Staphylococcus and occasionally a few Gram-negative rod survivors, such as Alcaligenes, Escherichia and Flavobacterium (Shafi et al., 1970; Stadelman, 1994; Delves-Broughton and Board,

2000; Cox, 2001; Ricke et al., 2001; ICMSF, 2005). On the other hand, r values of linear relationships between the cultural TVC results and the MicroFoss DTs of raw milk and pasteurized milk, -0.93 and –0.92, respectively, were not different

(Firstenberg-Eden et al., 2002).

Optionally, correlation of linear regression could be increased by excluding the TVC results of >107 CFU/g or <102 CFU/g (Salvat and Blivet, 2000). This was due to the

103 excessive sensitivity of detection threshold at high populations and the increased errors from both of the cultural and impedance methods that commonly occurred at low populations, causing scatter of data (Mendonca and Juneja, 2000). However, generation of the calibration curves in this Chapter deliberately included TVC results of <102

CFU/g to enable the MicroFoss assay to accurately detect low populations of total bacteria in pasteurized egg products and minimise extrapolating of TVC estimation.

This option was not applicable for all types of commercial raw egg samples, as their lowest TVCs were ≥102 CFU/g. Concurrently, since highest TVCs in all types of raw egg samples and pasteurized egg products were 107 CFU/g, the option of excluding

TVC results of >107 CFU/g to improve correlation of the cultural TVC and DTs was irrelevant.

However, Salvat and Blivet (2000) stated that a declining exponential function was more appropriate to represent the calibration curves of microbial counts and the impedance detection times than a linear regression. A declining exponential function as a quadratic regression was employed by Kahn and Firstenberg-Eden (1987) to represent the relationship between plate counts and impedance detection times to predict shelf life in pasteurized milk and other fluid dairy products, with correlation coefficients in the range of –0.80 to –0.90. Thus, a quadratic regression was alternatively applied to correlate the cultural TVCs and the MicroFoss DTs in raw egg samples and pasteurized egg products, in which the analysis was performed by SPSS 14.0 software for Windows

Evaluation Version (Section 3.2.7).

Quadratic regression did not noticeably improve correlations of the cultural TVC and the MicroFoss DTs of most commercial egg samples, except for pasteurized whole egg and pasteurized yolk products, compared to linear regression analyses. The reason that quadratic regression more suitably represented the relationship between the cultural

104 TVCs and the MicroFoss DTs for pasteurized whole egg and pasteurized yolk products was because the paired results were linear at high populations, 105-107 CFU/g, for both egg products but they were scattering to form a flat area at the bottom of the curves at lower populations, 102-104 CFU/g in pasteurized whole egg products (Figure 3.13) and

10-104 CFU/g in pasteurized yolk products (Figure 3.16). Moreover, these two matrices were reported to have similar microflora and TVCs (Shafi et al., 1970) and this may contribute to the former findings. In general, linear regression was sufficiently efficient to represent the calibration curves of TVC estimations by the MicroFoss assay in all types of eggs and egg products because the correlation coefficients obtained for all linear regressions were relatively high. Furthermore, linear regression was more practical, as only linear regression analysis was available in the MicroFoss system.

In Australia, TVC of pasteurized egg products should not exceed 2.50 x 104 CFU/g.

This specification was similar to that stated by Ricke et al., (2001) for liquid, frozen and dried pasteurized products. TVC of all pasteurized egg products tested freshly after thawing in this study complied to the above specifications, as they were in the range of

102-104 CFU/g, 10-102 CFU/g and non-detectable to 102 CFU/g for pasteurized whole egg, pasteurized yolk and pasteurized albumen products, respectively. Any pasteurized whole egg, pasteurized yolk and pasteurized albumen products that gave DT values earlier than 7.25h, 7.19h and 5.92h, respectively, or approximately 7.5h for pasteurized whole egg and pasteurized yolk products and 6h for pasteurized albumen products would be rejected due to unsatisfactory microbial quality, suggesting improper pasteurization. These results would be notified by the cutoff time of the MicroFoss system. Hence, estimations of TVC in raw egg samples and pasteurized egg products by the MicroFoss system would not only facilitate rapid microbiological testing of the products, but it would also assist decisions on compliance with specifications. Russell

105 (2001) and Firstenberg-Eden et al. (2002) also reported the usefulness of the MicroFoss system to assess whether TVC in ground pork and raw milk, respectively, exceeded the specification limits by setting up a cutoff time to the software based on the calibration curve line equations obtained in those studies.

Concurrently, Delves-Broughton and Board (2000) noted that pasteurization usually decreased microbial populations of liquid eggs by 100- to 1000- fold, leaving residual populations of about 102 CFU/g. TVC in some pasteurized whole egg products tested in this Chapter exceeded 102 CFU/g, perhaps because the initial bacterial loads in the egg pools prior to pasteurization of these samples could be as high as 106-107 CFU/g.

Reduction in total counts could not be directly compared between raw egg samples and pasteurized egg products in this Chapter, since these samples were not from the same egg pools.

The MicroFoss TVC assay was assessed whether it could be used to estimate populations of Ps. fluorescens in eggs. Pseudomonas fluorescens was inoculated into shell egg homogenates to create populations of spoilage species with the minimal impact of natural microflora. The MicroFoss DTs were inversely proportional to populations of Ps. fluorescens inoculated into the homogenates. Correlations of the cultural TVCs dominated by Ps. fluorescens and the MicroFoss DTs expressed as linear and quadratic regressions did not markedly differ, with relatively low correlation coefficients of –0.76 and –0.79, respectively. This marginal correlation could be because the acidic pH of the MicroFoss TVC medium after D-glucose was utilised may not be suitable for Pseudomonas growth, as Pseudomonas are not very acid tolerant and can hardly grow below pH 5.0-6.0 (Cousin, 2000). Furthermore, the incubation temperature at 30°C used in the cultural and MicroFoss methods in this study, as stated in Section 3.2.5 and Table 3.1, respectively, may not be best suitable for growth of

106 Pseudomonas. Although Pseudomonas can grow readily at 5-30°C, but the optimum growth temperature is at 25°C (Ayres, 1960; Burger et al., 2000). Pseudomonas species require proteases, peptidases, and associating enzymes when growing in eggs due to very few residual or utilizable carbohydrates available in egg (Cousin, 2000). The unavailability of the indigenous proteins in albumen; for instance, binding of the protein avidin with the B vitamin biotin could be responsible for restriction of Ps. fluorescens growth inoculated into whole egg homogenates in this study (Yadav and Vadehra, 1977;

Fields, 1979; Cox, 2001). This could contribute to obtaining the varied cultural TVC results with the same or closed DTs of whole egg homogenates dominated with Ps. fluorescens or non-correlating paired results. As a consequence, this resulted in relatively low correlation coefficient of –0.76 of this linear relationship. Findings of this study suggested that the MicroFoss assay could be used for further investigation of spoilage and shelf life prediction of eggs. According to findings in this Chapter, eggs with high TVC, presumably 106-107 CFU/g, were more likely to spoil, as this population level causes spoilage in pasteurized milk and other fluid dairy products

(Kahn and Firstenberg-Eden, 1987) and beef liver (Hernandez-Herrero et al., 1999).

Further examinations, such as threshold level of Pseudomonas populations to cause spoilage in eggs, were necessary to be conducted.

Some studies indicated that spoilage and shelf life prediction of foods could be conducted by TVC estimations (cottage cheese and cream dressing) inoculated with Ps. fluorescens (Bishop and White, 1985), or a relationship between pH and pseudomonad counts of time-temperature treated sliced beef liver (Hernandez-Herrero et al., 1999).

Commercial raw whole egg samples were checked for presence of Ps. fluorescens, as the most prevalent spoilage species in refrigerated eggs and egg products (Ayres, 1960;

Imai, 1976; Board and Tranter 1986). Colony morphology of isolates from raw whole

107 egg samples on PCA corresponded with that of the reference culture of Ps. fluorescens.

These isolates were classified as Gram-negative rods, catalase-positive, oxidase- positive, motile, aerobic and possessed oxidative metabolism of glucose, as shown in

Table 3.2. Although it could not be concluded from these results that these isolates were

Ps. fluorescens, they exhibited general characteristics of spoilage organisms, including

Pseudomonas spp. and indicated possible presence of presumptive Ps. fluorescens in raw egg. It is necessary to perform further specific examinations to confirm the identity of Ps. fluorescens isolates from the egg samples.

This study suggested marginal to good correlations between the cultural TVCs and the

MicroFoss DTs to be used as a basis for rapid enumeration of total bacteria in unpasteurized liquid eggs and pasteurized egg products. Analyses can be completed in less than 24h, compared to 48-72h required to obtain the cultural TVC results.

Furthermore, the MicroFoss system could also serve as an assessor of adequate and efficient pasteurization and possibly a shelf life predictor. Rapid estimation of TVC in perishable foods, such as eggs, would benefit the egg manufacturers for a fast release of products to markets and consumers and cost-saving storage.

108 CHAPTER 4

Evaluation of the MicroFoss system for estimating coliforms in eggs and egg products

4.1 INTRODUCTION

The coliform group is comprised of several genera in the family Enterobacteriaceae, including Citrobacter, Enterobacter, Escherichia and Klebsiella (Jay, 2000; Ray, 2001;

Craven et al., 2003). Since Escherichia coli is present in the intestines of humans and animals, its occurrence in foods suggests possible faecal contamination. However, many coliforms also inhabit environments, which are not associated with faeces. Hence, their presence in foods also indicates contamination from environmental sources, such as water, soil and plants. Therefore, the presence of coliforms, as indicator organisms, in processed foods suggests one or more lapses in quality control, including inadequate processing, lack of adequate hygiene and sanitation, or improper handling and storage.

In summary, coliform populations can reflect the overall microbial quality of raw materials and finished products, the presumptive presence of pathogens, the efficacy of food processing, such as pasteurization, and possible post-processing contamination

(ICMSF, 1978; Adams and Moss, 2000; Jay, 2000; Pandey et al., 2000; Kornacki and

Johnson, 2001; Craven et al., 2003).

The microflora on the shell of freshly laid eggs consists of a variety of Gram-positive and Gram-negative bacteria, including coliforms (Board et al., 1964; Board and Tranter,

1986; Banwart, 1989; Ray, 1996; Sparks, 2000; Cox, 2001). High coliform counts (>105

CFU/g) in unpasteurized liquid eggs indicate heavy contamination of coliforms from the shell to the egg contents as well as temperature abuse of the product that allows growth

109 of coliforms. Coliform counts of pasteurized egg products that exceed 10 CFU/g suggest poor microbiological quality of unpasteurized liquid eggs and/or improper pasteurization that allowed post-pasteurization contamination (Ricke et al., 2001).

Although the standard cultural plate method using violet red bile agar (VRBA) (Oxoid) has been widely used to enumerate viable coliforms in foods including eggs and egg products, it suffers several limitations including its tediousness, speed, sensitivity and detection limit (Craven et al., 2003). In recent years, the needs for reliable, rapid analytical methods have increased due to increased demands in quality assurance of foods and food safety. Since eggs are a perishable food, it is beneficial to obtain timely microbiological results to facilitate the early release of products to consumers. Various techniques including rapid assays for enumeration and detection of coliforms in foods

(Craven et al.2003) and drinking water (Rompre et al., 2002) have been reviewed.

These methods include a MicroFoss assay for coliforms (Russell, 2001; Firstenberg-

Eden et al., 2002; Odumeru and Belvedere, 2002; Cox and Fleet, 2003; Firstenberg-

Eden et al., 2004).

Although the MicroFoss coliform assay is commercially available, its application to unpasteurized liquid eggs and pasteurized egg products was not yet established. The aim of this Chapter was to evaluate the MicroFoss assay for estimating coliform populations in unpasteurized liquid eggs and pasteurized egg products. Unpasteurized liquid eggs and pasteurized egg products were analysed for coliform counts simultaneously by the standard cultural, plate count approach and the MicroFoss assay, after which the results from the two methods were correlated to generate calibration curves. Calibration curves for unpasteurized liquid eggs were generated from egg homogenates derived from shell eggs inoculated with different populations of coliforms. Thereafter, the effectiveness of these calibration curves was examined by applying their linear equations to estimate

110 coliform populations in commercial raw egg samples. The MicroFoss assay for coliforms was evaluated for accuracy, sensitivity, specificity, reproducibility, and speed of detection to determine its use as an alternative to the cultural method. The Chi square test and t-test were used to compare the populations of confirmed coliforms obtained using the MicroFoss assay and significant differences in coliform counts obtained from the cultural and MicroFoss assays, respectively.

4.2 MATERIALS AND METHODS

4.2.1 Materials

Shell eggs and commercial raw egg samples and pasteurized egg products used in this research are as described in Section 3.2.1.

4.2.2 Sample handling prior to testing

Handling of shell eggs and frozen pasteurized egg products are as described in Section

3.2.2.

4.2.3 Inoculum preparation

Four reference cultures, Citrobacter freundii, Enterobacter aerogenes, Escherichia coli and Klebsiella pneumoniae, were used to evaluate the MicroFoss assay to estimate and detect coliforms in eggs and egg products. These cultures, except for Escherichia coli, were received from the culture collection of the School of Biotechnology and

Biomolecular Sciences, UNSW. Escherichia coli UNSW 048200 or ATCC 11775 was a reference culture maintained by the Food Science and Technology group, School of

Chemical Sciences and Engineering, UNSW. Stock cultures were maintained on slants

111 of TSA at 5°C and sub-cultured monthly. Before use in experiments, all cultures were tested for purity by streaking onto fresh plates of TSA.

Inocula were prepared by transferring a single colony of biomass from a plate of TSA into 10 mL of TSB and incubating at 37°C for 24h. The broth culture was serially diluted 1:10 using 9 mL of 0.1% peptone water to give suspensions of different populations for inoculation into egg homogenates derived from shell eggs. For generation of calibration curves, these homogenates were inoculated with a cocktail of four coliform species. Cultures of individual coliforms were prepared as just described and mixed to give the cell suspension for inoculation.

4.2.4 Sample preparation

4.2.4.1 Inoculation of samples with defined populations of coliforms

Egg homogenates derived from shell eggs for inoculation with different levels of coliform populations were prepared according to AS 1766.3.8-1991, as described in

Section 3.2.2.1. Purposes of inoculation of these samples are described in the subsequent Sections.

••• Applicability and variability of the MicroFoss assay for coliforms

Aliquots of either each species of coliforms (0.1 mL each of Citrobacter freundii,

Enterobacter aerogenes, Escherichia coli and Klebsiella pneumoniae) or a cocktail of the four coliform species (0.025 mL of each species) at selected populations were individually inoculated into 10g of whole egg homogenates derived from shell eggs.

Homogenates were aseptically prepared from shell eggs purchased from the supermarket as described in Section 3.2.1.1, and assumed to be sterile or minimally contaminated. The inoculated samples were then diluted with 90 mL of 0.1% peptone

112 water to obtain a 10-1 dilution. The mixtures were thoroughly blended using a

Stomacher for 2 min to yield a homogenate.

Further decimal dilutions were prepared by adding 1 mL to 9 mL of 0.1% peptone water. After preparation, these samples were simultaneously analysed using the cultural and MicroFoss methods without delay. The counts from culture and the results of the

MicroFoss analysis (DTs, curve shapes and curve amplitudes) for samples inoculated with either individual species or the cocktail of the four coliforms were compared.

MicroFoss data for the inoculated samples were compared with those of non-inoculated

MicroFoss coliform media and MicroFoss media inoculated with sterile whole egg homogenate.

Variability of detection of populations of coliforms in whole egg homogenates inoculated with various batches of the cocktail was subsequently assessed using the same preparation and testing procedures.

••• Generation of calibration curves as a basis for enumeration of coliforms in unpasteurized liquid eggs by the MicroFoss assay

Portions (10g) of homogenates of shell egg derivatives (whole egg, yolk and albumen) were inoculated with aliquots of a cocktail of the four coliform species (0.1 mL composite) to obtain samples containing coliform populations in the range of <10 to 107

CFU/g. These samples were prepared and analysed using both assays as above. The cultural coliform counts (log CFU/g) and the MicroFoss DTs were correlated to form linear calibration curves generated by the MicroFoss system as a basis for the enumeration of coliforms in unpasteurized liquid eggs by the MicroFoss assay.

113 4.2.4.2 Analysis of commercial egg samples for coliforms

Commercial egg samples (Section 3.2.1.2) were prepared according to AS 1766.3.8-

1991. Portions (10g) of commercial raw egg samples and pasteurized egg products were serially diluted and analysed using the two assays as described above.

To create samples with various populations of coliforms, raw and pasteurized samples of each product were abused by storage for up to 3 days. Development of populations of indigenous coliforms in the products was promoted by extended storage at 5°C, or at

25°C, to simulate potential abuse storage conditions. Samples at 5°C were stored for 1-3 days, with the exception of pasteurized albumen, which was stored for up to 5 days.

Samples were abused at 25°C by storage for 1 day, except pasteurized albumen, which was held for up to 3 days.

••• Determination of the effectiveness of the MicroFoss assay to estimate coliforms in unpasteurized liquid eggs by testing with commercial raw egg samples

Differences between the cultural coliform counts and the counts obtained from calculations of DTs based on linear equations describing the calibration curves in commercial raw egg samples were compared to determine the effectiveness of the

MicroFoss assay for coliforms in unpasteurized liquid eggs.

••• Generation of calibration curves for the MicroFoss assay as a basis to estimate coliforms in pasteurized egg products

The cultural coliform counts and DTs of commercial pasteurized egg products were correlated to form linear calibration curves by the MicroFoss system as a basis for the enumeration of coliforms in pasteurized egg products by the MicroFoss assay.

114 4.2.5 Coliform cultural method

The cultural coliform method was carried out according to AS 5013.4-2004. Aliquots (1 mL) of sample dilutions were pipetted into sterile Petri dishes to which approximately

15 mL of molten VRBA at 45°C were added, and the contents were carefully mixed.

After the agar had solidified, a further volume (approximately 4 mL) of molten VRBA was added as an overlay and allowed to solidify. The plates were inverted and incubated at 30 + 1°C for 24 + 2h. Typical coliforms appearing on VRBA plates were characterised by purplish red or fuchsia colonies, 1 to 2 mm in diameter, occasionally as small as 0.5 mm. These colonies were generally surrounded by a reddish fuchsia zone of precipitated bile salts, which indicated acid produced from lactose fermentation.

Plates apparently containing between 15 and 150 presumptive coliform colonies were counted. Populations were calculated from the dilutions used and recorded as presumptive coliforms. Each dilution of sample homogenate was analysed using duplicate plates.

The method for enumeration of coliforms in pasteurized egg products, in which the populations of these organisms were expected to be reduced or eliminated, was modified slightly. A 10mL sample suspension at the 10-1 dilution was used. Each 2.5 mL of this suspension was pipetted into one of four plates. A total colony counts (CFU) from the four plates were related to the coliform population per gram of egg product.

Further dilutions were plated using the method described above.

Confirmation of presumptive coliforms: A number of colonies were selected for confirmation based on an approximation of the square root of the number of typical colonies on VRBA plates, to a maximum of 10. Each colony was selected randomly and inoculated into 10 mL of single-strength LT broth containing an inverted Durham tube.

The inoculated LT broths were incubated at 30 + 1°C for 48h. Colonies producing gas

115 in Durham tubes were recorded as confirmed coliforms (Bloch et al., 1996; AS

1766.2.3-1992).

4.2.6 The MicroFoss coliform test

An aliquot (5 mL) of each egg sample at the 10-1 dilution was added to a coliform vial

(Foss Pacific, North Ryde, Australia). The vials were mixed by gentle inversion 10 times, then inserted into the MicroFoss instrument and monitored during their incubation, according to the manufacturer’s instructions. Details of test configurations are shown in Table 4.1.

Table 4.1 Test configurations of the MicroFoss coliform assay for eggs and egg products

Tests Threshold Skip Shuteye Cutoff Caution Temperature Duration (°C) (h) Coliforms 10 1 25 10 10 35 24

Source: Derived from the MicroFoss manual.

The vials giving a detection time, indicating the presence of coliforms, were further confirmed by checking gas production in LT broth. Approximately five drops of the culture from a vial giving positive a result were pipetted into 10 mL of a single-strength

LT broth containing a Durham tube, which was incubated at 30°C for up to 48h. The cultures from positive vials that produced gas in LT broth were recorded as confirmed coliforms, whereas those from positive vials that did not produce gas in the LT broth were recorded as false positive results.

116 4.2.7 Statistical analyses

4.2.7.1 Effectiveness of the MicroFoss assay for coliforms

••• Correlations of the data set to produce a calibration curve

Correlations of the calibration curves, which were a linear regression between the cultural coliform counts (log CFU/g) and the MicroFoss DTs, were determined by correlation coefficients (r value) processed by the MicroFoss system. Linear equations were generated by the system.

••• Effectiveness of coliform estimations using the calibration curves

Effectiveness of the calibration curves to estimate coliform counts was tested by applying them to test for coliforms in raw egg samples from a commercial producer of egg products (Section 3.2.1.2). Raw egg samples were thawed and then simultaneously used for estimation of coliform populations using the cultural assay with VRBA and the MicroFoss system. The MicroFoss DTs were transformed to coliform counts using the linear equation derived from each calibration curve. Standardised residuals, at 95% prediction intervals, were calculated to determine the effectiveness of the MicroFoss assay to estimate coliform populations in unpasteurized liquid eggs. Standard errors of the estimate were calculated from the whole series of the paired data (log CFU/g and DTs) of homogenates derived from shell eggs, which were used for generation of the calibration curves. This was calculated using SPSS 14.0 software for Windows Evaluation Version.

Standardised residuals = (log cultural coliform counts – log MicroFoss coliform counts) Standard error of the estimate

Within normal distribution, approximately 95% of standardised residuals should fall in the interval of +2 (Montgomery et al., 2004). Hence, the range between –2 and 2 of standardised residuals at 95% prediction intervals indicates comparable cultural and

MicroFoss coliform counts.

117 4.2.7.2 Performance indicators

The MicroFoss assay detects and estimates coliforms in eggs and egg products. Hence, criteria for evaluation of qualitative (sensitivity rate, specificity rate, false negative rate and false positive rate) and quantitative (standard deviations and repeatability) alternative methods were therefore evaluated.

4.2.7.3 Repeatability of the MicroFoss assay for coliforms

Repeatability of the MicroFoss assay for coliforms was assessed among ≥5 replicates of one identical commercial egg sample conducted by the same operator using the same method and apparatus within the same laboratory. The repeatability value is the value below which the absolute difference between 2 single test results obtained under repeatability conditions may be expected to lie within 95% probability (Feldsine et al.,

2002).

The coefficient of repeatability (CR) can be calculated as 1.96 times the standard deviations of the differences between the two repeated measurements using one single method (Bland and Altman, 1999).

Coefficient of repeatability = 1.96 x Standard deviation

If the values of the test results are between mean + CR, this indicates that the method is repeatable at 95% confidence interval.

118 4.2.7.4 Test for significant difference of positive proportions for the cultural and

MicroFoss assays a Chi square (χ2) value was calculated to assess whether proportions confirmed positive for coliforms between the cultural and MicroFoss methods were significantly different.

4.2.7.5 Comparison of coliform counts from the cultural and MicroFoss assays

Student’s t-test for two samples, assuming equal variances, is used to determine if the mean of the cultural and MicroFoss coliform counts are not statistically different at the

5% level of significance (AS/NZS 4659.2-1999; Feldsine et al., 2002; Silbernagel and

Lindberg, 2002). This was performed using Microsoft Excel 2000.

4.3 RESULTS

4.3.1 MicroFoss curves for coliform assays

Figure 4.1 shows the MicroFoss curves for the detection of different coliforms inoculated into whole egg homogenates, compared to the MicroFoss curves for detection of non-inoculated coliform medium and medium inoculated with coliform- free whole egg homogenate (controls). These controls yielded no detection, whereas samples inoculated with Citrobacter freundii, Enterobacter aerogenes, Escherichia coli and Klebsiella pneumoniae gave detection times of 5.1h, 5.1h, 4.7h and 5.5h, respectively, whilst those inoculated with a cocktail of the four coliforms gave similar detection times of 5.3h and 5.0h (done in duplicate).

119 Non-inoculated medium N.DT.

Medium inoculated with coliform-free sample N.DT.

Citrobacter fruendii 5.1h

Enterobacter aerogenes 5.1h

Escherichia coli 4.7h

Klebsiella pneumoniae 5.5h

A cocktail of mixed coliforms I 5.3h

A cocktail of mixed coliforms II 5.0h

Figure 4.1 The MicroFoss curves for detection of a non-inoculated coliform medium and a medium inoculated with coliform-free whole egg homogenate (controls) giving no detection, and detection of coliforms in whole egg homogenates inoculated to approximately 105 CFU/g with individual strains of coliforms, Citrobacter freundii (DT 5.1h); Enterobacter aerogenes (5.1h); Escherichia coli (4.7h); and Klebsiella pneumoniae (5.5h), together with a mixture of the four coliforms (done in duplicate) yielding DTs of 5.3h and 5.0h.

Regardless of the individual strain or different batches of a cocktail of the four coliforms, all inoculated whole egg homogenates containing approximately 105 CFU/g of coliforms gave similar DTs, ranging from 4.7h to 5.5h, comparable curve amplitudes ranging from 200 to 300 optical units and similar MicroFoss detection curves. The slightly elevated straight curve from 130 to 160 optical units of a non-inoculated

MicroFoss coliform medium and a medium inoculated with coliform-free whole egg homogenate (controls) was a result of a difference in colour of the medium, due to the addition of the egg itself and some dilution of the colour in the agar plug throughout the incubation period.

As acid produced from metabolism of coliforms to utilise fermentable substrates in the coliform medium, a change in the optical units from 130 to 400 was detected by the

MicroFoss system. The shape of the MicroFoss curves for coliform detection was comprised of the pre-accelerating, accelerating and declining phases, which was similar 120 to the lag, log and decline phases of the normal bacterial growth curve. These findings indicated applicability of the MicroFoss assay to the detection of either each strain of coliforms or a cocktail of the four coliforms in egg products.

4.3.2 Variability of the MicroFoss assay for detection of coliforms in eggs

To evaluate the variability of the MicroFoss assay, many individual samples of egg homogenates were prepared. Each sample was inoculated with a mixture of the four coliforms to give a final (approximate) population of either 10 or 102 CFU/g coliforms, as egg products may have very low counts of coliforms. These samples were then concurrently assayed for coliforms by the cultural and MicroFoss methods.

Similar results were obtained for an inoculum to yield 102 and 10 CFU/g of coliforms in egg homogenates, in which the mean populations were 2.88 x 102 CFU/g for a range of

1.00 x 102 to 9.12 x 102 CFU/g, and 3.63 x 10 CFU/g for a range of 1.00 x 10 to 8.91 x

10 CFU/g, respectively. The mean DTs for samples containing 102 and 10 CFU/g of coliforms were 8.0g for a range of 7.0h to 9.5h with the variation coefficient of 0.05, and 8.8h for a range of 7.5h to 10.9h with the variation coefficient of 0.09, respectively.

It was presumed that coliform strains that dominated in each inoculation gave varied metabolic activity rate, resulting in slightly different DTs at each population level. This may have reflected naturally uneven presence of types and numbers of each species of coliforms in contaminated unpasteurized eggs. Variability of the MicroFoss assay allowed detection of any species of coliforms.

121 4.3.3 Application of the MicroFoss system to estimate coliforms in

unpasteurized egg

4.3.3.1 Unpasteurized whole egg

Whole egg homogenates derived from shell eggs inoculated with different populations of coliforms were simultaneously analysed for coliform counts by the cultural plating method and DTs by the MicroFoss system.

Figure 4.2 shows the linear correlation between coliform counts by the cultural method

(log CFU/g) and DTs as estimated by the MicroFoss system for 234 samples of whole egg homogenates inoculated with different coliform populations yielding a correlation coefficient of –0.93. Detection times ranged from 3h for 107 CFU/g to 13h for a sample with <10 CFU/g.

Figure 4.2 Linear regression relationship between the log-transformed populations of coliforms and MicroFoss detection times for whole egg homogenate samples inoculated with different populations of coliforms. The regression is described by the equation log CFU/g = 8.997 – 0.786 (DT), (r = -0.93, n = 234).

The cultural coliform counts in commercial raw whole egg samples were compared to coliform counts obtained from mathematically transforming DTs, using the linear equation describing the calibration curve, log CFU/g = 8.997 – 0.786 (DT), as shown in

122 Figure 4.2, to calculate standardised residuals at 95% prediction intervals, as shown in

Table 4.2. The standard error of the estimate of the calibration curve was 0.60851.

After thawing, commercial raw whole egg samples generally gave coliform counts of

10-104 CFU/g, as determined by the cultural plating method. Populations of indigenous coliforms in the products stored at 5°C (15 samples), or at 25°C (8 samples) were developed to 104 CFU/g and 105-107 CFU/g, respectively.

Table 4.2 A comparison of the cultural and MicroFoss coliform counts of commercial raw whole egg samples and their standardised residuals at 95% prediction intervals n DT VRBA counts MF countsa ZRESIDb n DT VRBA counts MF countsa ZRESIDb (h) (log CFU/g) (log CFU/g) 95% P.I. (h) (log CFU/g) (log CFU/g) 95% P.I. 1 5.6 3.48 4.59 -1.82 24 2.9 6.41 6.72 -0.51 2 7.3 1.00 3.26 -3.71 25 3.6 6.09 6.17 -0.13 3 6.1 4.27 4.20 0.12 26 3.5 6.25 6.25 0 4 6.0 4.20 4.28 -0.13 27 3.1 6.41 6.56 -0.25 5 5.2 4.32 4.91 -0.97 28 3.5 6.24 6.25 -0.02 6 6.0 4.44 4.28 0.26 29 2.9 5.39 6.72 -2.19 7 6.1 4.53 4.20 0.54 30 7.4 4.38 3.18 1.97 8 7.4 3.92 3.18 1.22 31 6.9 3.88 3.57 0.51 9 7.4 3.86 3.18 1.12 32 6.5 5.13 3.89 2.04 10 7.4 3.95 3.18 1.27 33 6.4 4.69 3.97 1.18 11 7.5 4.15 3.10 1.73 34 7.2 4.44 3.34 1.81 12 7.4 4.18 3.18 1.64 35 7.2 4.16 3.34 1.35 13 7.5 4.48 3.10 2.27 36 6.9 4.24 3.57 1.10 14 7.2 4.52 3.34 1.94 37 6.8 4.32 3.65 1.10 15 7.3 4.59 3.26 2.19 38 7.1 4.31 3.42 1.46 16 7.3 4.41 3.26 1.89 39 6.8 4.37 3.65 1.18 17 7.3 4.42 3.26 1.91 40 6.3 4.03 4.04 -0.02 18 8.8 4.47 2.08 3.93 41 6.2 4.08 4.12 -0.07 19 9.1 3.80 1.84 3.22 42 6.1 4.34 4.20 0.23 20 9.3 3.82 1.69 3.50 43 6.0 4.20 4.28 -0.13 21 6.1 4.20 4.20 0 44 6.4 4.66 3.97 1.13 22 6.5 4.11 3.89 0.36 45 6.3 4.83 4.04 1.30 23 2.6 7.18 6.95 0.38 a Coliform counts of commercial raw whole egg samples determined by the MicroFoss method was obtained from mathematic transformation of DTs to coliform counts using linear equation describing the calibration curve, log CFU/g = 8.997 – 0.786 (DT). b ZRESID 95% P.I. = Standardised residuals at 95% prediction intervals. Black and red fonts represented commercial raw whole egg samples, whose standardised residuals at 95% prediction intervals fell within and exceeded the range of + 2, respectively.

Among 45 commercial raw whole egg samples tested, coliform counts were comparable for the cultural and MicroFoss assays in 37 samples (82.22%) and different in eight

123 samples (17.78%), as indicated by their standardised residuals at 95% prediction intervals, highlighted in black and red in Table 4.5.

4.3.3.2 Unpasteurized yolk

Yolk homogenates, derived from shell eggs, and inoculated with different populations of coliforms were simultaneously analysed for coliform counts by the cultural plating method and DTs by the MicroFoss system.

Figure 4.3 shows the linear correlation between coliform counts by the cultural method

(log CFU/g) and DTs as estimated by the MicroFoss system for 123 samples of yolk homogenates inoculated with different coliform populations. DTs ranged from 2.5h for a sample containing 107 CFU/g to 13h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.95.

Figure 4.3 Linear regression relationship between the log-transformed populations of coliforms and MicroFoss detection times for yolk homogenate samples inoculated with different populations of coliforms. The regression is described by the equation log CFU/g = 9.183 – 0.820 (DT), (r = -0.95, n = 123).

The cultural coliform counts in commercial raw yolk samples were compared to coliform counts obtained from mathematically transforming DTs using the linear

124 equation describing the calibration curve, log CFU/g = 9.183 – 0.820 (DT), as shown in

Figure 4.3, to calculate standardised residuals at 95% prediction intervals, as shown in

Table 4.3. The standard error of the estimate of the calibration curve was 0.60111.

After thawing, commercial raw yolk samples generally gave coliform counts of 10-103

CFU/g, as determined by the cultural plating method.

Table 4.3 A comparison of the cultural and MicroFoss coliform counts of commercial raw yolk samples and their standardised residuals at 95% prediction intervals n DT (h) VRBA counts MF countsa ZRESIDb (log CFU/g) (log CFU/g) 95% P.I. 1 7.4 1.98 3.11 -1.88 2 7.3 2.48 3.19 -1.18 3 8.1 2.80 2.54 0.43 4 8.4 3.48 2.29 1.98 5 8.4 3.45 2.29 1.93 6 8.7 3.18 2.05 1.88 7 2.3 7.74 7.30 0.73 8 2.3 7.63 7.30 0.55 9 2.3 7.94 7.30 1.06 10 3.6 7.25 6.23 1.70 11 2.5 7.11 7.13 -0.03 12 8.6 2.70 2.13 0.95 13 8.4 2.39 2.29 0.17 14 11.3 1.00 -0.09 1.81 15 10.0 1.45 0.98 0.78 16 11.8 1.00 -0.50 2.50 17 10.3 1.18 0.73 0.75 18 7.8 2.92 2.78 0.23 19 7.7 3.43 2.87 0.93 20 7.8 2.90 2.78 0.20 21 7.9 3.41 2.70 1.18 22 8.4 2.30 2.29 0.02 23 8.3 2.78 2.37 0.68 24 8.5 2.00 2.21 -0.35 25 15.4c No detection Not applicable Not applicable a Coliform counts of commercial raw yolk samples determined by the MicroFoss method was obtained from mathematic transformation of DTs to coliform counts using linear equation describing the calibration curve, log CFU/g = 9.183 – 0.820 (DT). b ZRESID 95% P.I. = Standardised residuals at 95% prediction intervals. Black and red fonts represented commercial raw yolk samples, whose standardised residuals at 95% prediction intervals fell within and exceeded the range of + 2, respectively. c A false positive result obtained from the MicroFoss assay, with DT of 15.4h, the MicroFoss coliform count and standardised residuals at 95% prediction intervals of this sample were therefore not applicable.

The MicroFoss assay gave a false positive detection of coliforms in one sample of commercial raw yolk, which was then excluded from calculating proportions of similar coliform counts between the cultural and MicroFoss assays. Hence, among 24

125 commercial raw yolk samples tested, coliform counts for the cultural and MicroFoss assays were comparable in 23 samples (95.83%) but different in 1 sample (4.17%), as indicated by their standardised residuals at 95% prediction intervals, highlighted in black and red fonts in Table 4.3.

4.3.3.3 Unpasteurized albumen

Albumen homogenates, derived from shell eggs, and inoculated with different populations of coliforms were simultaneously analysed for coliform counts by the cultural plating method and DTs by the MicroFoss system.

Figure 4.4 shows the linear correlation between coliform counts by the cultural method

(log CFU/g) and DTs as estimated by the MicroFoss system for 125 samples of albumen homogenates inoculated with different coliform populations. DTs ranged from

2.5h for a sample containing 107 CFU/g to 14h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.95.

Figure 4.4 Linear regression relationship between the log-transformed populations of coliforms and MicroFoss detection times for albumen homogenate samples inoculated with different populations of coliforms. The regression is described by the equation log CFU/g = 8.683 – 0.721 (DT), (r = -0.95, n = 125).

126 The cultural coliform counts in commercial raw albumen samples were compared to coliform counts obtained from mathematically transforming DTs using linear equation describing the calibration curve, log CFU/g = 8.683 – 0.721 (DT), as shown in Figure

4.4, to calculate standardised residuals at 95% prediction intervals, as shown in Table

4.4. The standard error of the estimate of the calibration curve was 0.50329.

After thawing, commercial raw albumen samples generally gave coliform counts of

<10-102 CFU/g, as determined by the cultural plating method.

Table 4.4 A comparison of the cultural and MicroFoss coliform counts of commercial raw albumen samples and their standardised residuals at 95% prediction intervals

n DT VRBA counts MF countsa ZRESIDb n DT VRBA counts MF countsa ZRESIDb (h) (log CFU/g) (log CFU/g) 95% P.I. (h) (log CFU/g) (log CFU/g) 95% P.I. 1 10.4 0.90 1.19 -0.58 22 8.4 3.00 2.63 0.74 2 10.0 1.16 1.48 -0.64 23 9.2 3.40 2.05 2.68 3 9.3 1.95 1.98 -0.06 24 10.6 1.70 1.05 1.29 4 9.8 1.92 1.62 0.60 25 10.2 2.30 1.33 1.93 5 8.9 2.39 2.27 0.24 26 9.4 4.28 1.91 4.71 6 9.1 2.59 2.13 0.91 27 10.0 2.48 1.48 1.99 7 12.3 1.43 -0.18 3.20 28 10.2 1.70 1.33 0.74 8 11.5 1.63 0.40 2.44 29 6.0 4.78 4.36 0.83 9 11.1 1.60 0.69 1.81 30 6.1 4.77 4.29 0.95 10 10.4 1.16 1.19 -0.06 31 5.9 4.61 4.43 0.36 11 10.2 1.06 1.33 -0.54 32 6.4 4.49 4.07 0.83 12 10.7 1.36 0.97 0.77 13 10.7 1.90 0.97 1.85 14 10.4 2.13 1.19 1.87 15 9.0 2.30 2.20 0.20 16 11.8 1.29 0.18 2.21 17 11.7 1.16 0.25 1.81 18 8.5 2.52 2.56 -0.08 19 9.6 2.40 1.77 1.25 20 9.6 3.29 1.77 3.02 21 8.9 3.58 2.27 2.60 a Coliform counts of commercial raw albumen samples determined by the MicroFoss method was obtained from mathematic transformation of DTs to coliform counts using linear equation describing the calibration curve, log CFU/g = 8.683 – 0.721 (DT). b ZRESID 95% P.I. = Standardised residuals at 95% prediction intervals. Black and red fonts represented commercial raw albumen samples, whose standardised residuals at 95% prediction intervals fell within and exceeded the range of + 2, respectively.

127 Among 32 commercial raw albumen samples tested, coliform counts for the cultural and

MicroFoss assays were comparable in 25 samples (78.13%), but different in 7 samples

(21.87%), as indicated by their standardised residuals at 95% prediction intervals, highlighted in black and red fonts in Table 4.7.

In conclusion, coliform estimations by the cultural and MicroFoss assays yielded comparable counts in excess of 78% of samples tested.

4.3.4 Application of the MicroFoss system to estimate coliforms in pasteurized

egg products

4.3.4.1 Pasteurized whole egg products

Commercial pasteurized whole egg products were simultaneously analysed for coliform counts by the cultural method and the MicroFoss system. Among 112 samples examined, no coliforms were detected in 91 samples by either the cultural method or the

MicroFoss method. Positive results in nine samples were obtained by the MicroFoss method, but the cultural method failed to detect coliforms in these samples. However, positive results from the MicroFoss method in seven out of nine samples were confirmed as coliforms by the ability to produce gas in the LT broth. As a result, the

MicroFoss assay gave false positive results for two samples. Hence, results of 12 samples, for which typical colonies on VRBA plates and vials giving detection times were confirmed as coliforms, were used to form the calibration curve.

After thawing, commercial pasteurized whole egg products generally gave coliform counts of 10-102 CFU/g, as determined by the cultural plating method.

Figure 4.5 shows the linear correlation between coliform counts by the cultural method

(log CFU/g) and DTs as estimated by the MicroFoss system for 12 samples of

128 commercial pasteurized whole egg products. DTs ranged from 4h for a sample containing 106 CFU/g to 15h for a sample with 10 CFU/g, yielding a correlation coefficient of –0.90.

Figure 4.5 Linear regression relationship between the log-transformed populations of coliforms and MicroFoss detection times for pasteurized whole egg products. The regression is described by the equation log CFU/g = 7.600 – 0.446 (DT), (r = -0.90, n =

12).

Analyses of seven pasteurized whole egg products giving DTs of 13.1h, 14.3h, 14.8h,

14.9h, 15.8h, 16.7h and 16.8h, which were subsequently confirmed as coliforms in the

LT broths, were equivalent to approximate coliform populations of 5.70 x 10, 1.70 x 10,

1.00 x 10, 9, 4, 2 and 2 CFU/g, respectively based on the linear equation, log CFU/g =

7.600 – 0.446 (DT), as shown in Figure 4.5.

4.3.4.2 Pasteurized yolk products

Commercial pasteurized yolk products were simultaneously analysed for coliform counts by the cultural method and the MicroFoss system. Among 56 samples examined, no coliforms were detected in 19 samples by either the cultural method or the 129 MicroFoss method. Positive results in two samples were obtained by the MicroFoss method, but the cultural method failed to detect coliforms in these samples. However, positive results from the MicroFoss method in one out of two samples was confirmed as coliforms by the ability to produce gas in the LT broth. As a result, the MicroFoss assay gave false positive result for one sample. Hence, results of 35 samples, for which typical colonies on VRBA plates and vials giving detection times were confirmed as coliforms, were used to form the calibration curve.

After thawing, commercial pasteurized yolk products generally gave coliform counts of

<10-10 CFU/g, as determined by the cultural plating method.

Figure 4.6 shows the linear correlation between coliform counts by the cultural method

(log CFU/g) and DTs as estimated by the MicroFoss system for 35 samples of commercial pasteurized yolk products. DTs ranged from 2h for a sample containing 107

CFU/g to 14h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.96.

Figure 4.6 Linear regression relationship between the log-transformed populations of coliforms and the MicroFoss detection times for pasteurized yolk products. The regression is described by the equation log CFU/g = 8.454 – 0.532 (DT), (r = -0.96, n =

35).

130 Analysis of one pasteurized yolk product giving a DT of 14.5h, which was subsequently confirmed as coliforms in the LT broth, was equivalent to approximate coliform populations of 6 CFU/g based on the linear equation, log CFU/g = 8.454 – 0.532 (DT), as shown in Figure 4.6.

4.3.4.3 Pasteurized albumen products

Commercial pasteurized albumen products were simultaneously analysed for coliform counts by the cultural method and the MicroFoss system. Among 55 samples examined, no coliforms were detected in 42 samples by either the cultural method or the

MicroFoss method. Positive results in two samples were obtained by the MicroFoss method, but the cultural method failed to detect coliforms in these samples. However, positive results from the MicroFoss method in one out of two samples was confirmed as coliforms by the ability to produce gas in the LT broth. As a result, the MicroFoss assay gave false positive result for one sample. Hence, results of 11 samples, for which typical colonies on VRBA plates and vials giving detection times were confirmed as coliforms, were used to form the calibration curve.

After thawing, commercial pasteurized albumen products generally gave coliform counts of <10 CFU/g, as determined by the cultural plating method.

Figure 4.7 shows the linear correlation between coliform counts by the cultural method

(log CFU/g) and DTs as estimated by the MicroFoss system for 11 samples of commercial pasteurized albumen products. DTs ranged from 12.5h for a sample containing 7.50 x 10 CFU/g to 23h for a sample with 1 CFU/g, yielding a correlation coefficient of –0.95.

131 Figure 4.7 Linear regression relationship between the log-transformed populations of coliforms and the MicroFoss detection times for pasteurized albumen products. The regression is described by the equation log CFU/g = 3.809 – 0.167 (DT), (r = -0.95, n = 11).

Analysis of one pasteurized albumen product giving a DT of 15.2h, which was subsequently confirmed as coliforms in the LT broth, was equivalent to approximate coliform populations of 1.90 x 10 CFU/g based on the linear equation, log CFU/g =

3.809 – 0.167 (DT), as shown in Figure 4.7.

4.3.5 Performance indicators

Sensitivity rate, specificity rate, false negative rate and false positive rate of the

MicroFoss assay for coliforms were evaluated by analyses of commercial raw egg samples and pasteurized egg products. Results of coliform detection in these samples by the cultural and MicroFoss assays used for calculation of performance indicators are shown in Table 4.5.

132 Table 4.5 Results of coliform detection in commercial raw egg samples and pasteurized egg products by the cultural and MicroFoss assays

Sample n MicroFoss resultsa True positive False negative False positive True negative (N11) (N12) (N21) (N22) Raw whole egg 45 45 0 0 0 Raw yolk 25 24 0 1 0 Raw albumen 32 32 0 0 0 Pasteurized whole egg 112 12 0 2 91 Products Pasteurized yolk products 56 35 0 1 19 Pasteurized albumen 55 11 0 1 42 Products a Results of coliform detection by the MicroFoss method were compared with results from the cultural method.

4.3.5.1 Sensitivity and false negative rates

There was no occasion that coliforms were not detected by the MicroFoss method and were present by the cultural method for analyses of 45, 25, 32, 112, 56 and 55 samples of raw whole egg, raw yolk, raw albumen, pasteurized whole egg products, pasteurized yolk products and pasteurized albumen products, respectively, across population ranges of <1 log CFU/g to 7 log CFU/g, except ranges of <1 log CFU/g to <2 log CFU/g for pasteurized albumen products. Hence, there was no false negative result for coliform detection in eggs and egg products by the MicroFoss assay in this Chapter, for which its sensitivity rate ([N11/(N11 + N12)] x 100) and false negative rate ([N12/(N11 + N12)] x 100 were 100% and 0%, respectively. This indicated equivalent sensitivity and great accuracy of the MicroFoss assay to the cultural method for coliform detection in egg and egg products.

Concurrently, sensitivity as lowest concentration of coliforms in commercial raw egg samples and pasteurized egg products that can be reliably detected by the MicroFoss

133 method was determined. The MicroFoss assay could detect ≤10 CFU/g in eggs and eggs products tested in this Chapter.

4.3.5.2 Specificity and false positive rates

There was no occasion that coliforms were not detected by either the cultural or the

MicroFoss method for analyses of 45, 25 and 32 samples of raw whole egg, raw yolk and raw albumen, respectively. Hence, specificity rate ([N22/(N22 + N21)] x 100) and false positive rate ([N21/(N22 + N21)] x 100) of these samples were not applicable, since the former was 0%.

On the other hand, results of coliform detections of all types of commercial pasteurized egg products in this Chapter, as summarised in Table 4.5, allowed specificity and false positive rates of the MicroFoss assay to be calculated, as shown in Table 4.6.

Table 4.6 Specificity and false positive rates of the MicroFoss assay to detect coliforms in commercial pasteurized egg products Sample True negative False positive Specificity False positive (N22) (N21) rate (%) rate (%) Pasteurized whole 91 2 97.85 2.15 egg products Pasteurized yolk 19 1 95.00 5.00 products Pasteurized albumen 42 1 97.67 2.33 products

There were a few occasions where coliforms were not detected by the cultural method, but positive results were obtained from the MicroFoss method and were subsequently confirmed as non-coliforms for analyses of 112, 56 and 55 samples of pasteurized whole egg products, pasteurized yolk products and pasteurized albumen products,

134 respectively. Nonetheless, high specificity and low false positive rates of 95-98% and 2-

5%, respectively of the MicroFoss assay for coliforms were accomplished.

On the other hand, there were occasions among 112, 56 and 55 samples of pasteurized whole egg products, pasteurized yolk products and pasteurized albumen products, where confirmed coliforms in 7 (6.25%), 1 (1.79%) and 1 (1.82%) samples, respectively, were detected by the MicroFoss, but they were not detected by the cultural method. Hence, the MicroFoss was likely to give more accurate, reliable result than the cultural method.

4.3.6 Repeatability of the MicroFoss assay and significant differences between

the cultural and MicroFoss coliform counts

Repeatability of the MicroFoss assay for coliforms was determined among coliform counts of replicates (≥5) from the same samples of commercial raw egg analysed on the same day. Concurrently, significant differences (P<0.05) between means of coliform counts in these replicates from the cultural and MicroFoss assays were evaluated by t- test. Results of repeatability of the MicroFoss assay and significant differences between means of coliform counts from both methods are shown in Table 4.7.

135 Table 4.7 Repeatability of the MicroFoss assay and significant differences between the cultural and MicroFoss coliform counts in commercial raw egg samples

Sample VRBA counts MF counts Mean + CR t P (log CFU/g) (log CFU/g)b (MF method) Raw whole egg 4.03 4.05 4.11 + 0.22 1.76 0.109 4.08 4.12 = 3.89 to 4.33 4.34 4.20 4.20 4.28 4.66 3.97 4.83 4.04 4.36 + 0.32a 4.11 + 0.11a 4.44 3.34 3.50 + 0.29 11.282 <0.0001* 4.16 3.34 = 3.21 to 3.79 4.24 3.57 4.32 3.65 4.31 3.42 4.37 3.65 4.31 + 0.10a 3.50 + 0.15a 6.09 6.17 6.39 + 0.47 -1.51 0.17 6.25 6.25 = 5.92 to 6.86 6.41 6.56 6.24 6.25 5.39 6.72 6.08 + 0.40a 6.39 + 0.24a Raw yolk 2.92 2.78 2.57 + 0.53 1.105 0.291 3.43 2.87 = 2.04 to 3.10 2.90 2.78 3.41 2.70 2.30 2.29 2.78 2.37 2.00 2.21 2.82 + 0.53a 2.57 + 0.27a Raw albumen 1.60 0.69 1.03 + 0.47 2.105 0.068 1.16 1.19 = 0.56 to 1.50 1.06 1.33 1.36 0.97 1.90 0.97 1.42 + 0.34a 1.03 + 0.24a 1.16 1.48 1.90 + 0.67 0.367 0.723 1.95 1.98 = 1.23 to 2.57 1.92 1.62 2.39 2.27 2.59 2.13 2.00 + 0.55a 1.90 + 0.34a a Coliform counts (log CFU/g) obtained from the cultural and MicroFoss assays are presented as mean + SD. b The MicroFoss coliform counts were mathematically transformed using DTs, which were based on linear equations describing the calibration curves, log CFU/g = 8.997 – 0.786 (DT), log CFU/g = 9.183 – 0.820 (DT) and log CFU/g = 8.683 – 0.721 (DT) of commercial raw whole egg, raw yolk and raw albumen samples, as shown in Figures 4.2, 4.3 and 4.4, respectively. * The cultural and MicroFoss coliform counts in commercial raw egg samples were significantly different (P<0.05).

In most cases, SD values of coliform counts determined by the MicroFoss assay were smaller than those of coliform counts estimated by the cultural method, particularly at

≤102 CFU/g.

136 Estimation of coliform populations by the MicroFoss assay was repeatable at 95% confidence interval, as results of all replicates of each individual sample in all types of raw egg samples were within mean + correlation of repeatability, as shown in Table 4.7.

Coliform counts in replicates of commercial raw egg samples estimated by either the cultural or the MicroFoss assay were generally not significant different (P≥0.05). The significant difference in counts was observed among replicates of one raw whole egg sample, in which the cultural counts in all replicates were 10-fold higher than the

MicroFoss counts.

4.3.7 Significant differences of proportions of confirmed coliforms between the

cultural and MicroFoss assays

2 2 A chi square value (χ = [(DN12 – N21D - 1) ] / (N12 + N21)) could not be calculated for commercial raw whole egg and raw albumen samples in this Chapter, as false negative and false positive results of coliform detection in these samples by the MicroFoss assay were not obtained, as shown in Table 4.5.

On the other hand, results of coliform detection for commercial raw yolk samples and all types of pasteurized egg products by both methods allowed chi square values to be calculated, as shown in Table 4.8.

137 Table 4.8 Test for significant differences of proportions of confirmed coliforms between the cultural and MicroFoss assays

Sample False negatives False positives χ2 (N12) (N21) Raw yolk samples 0 1 0 Pasteurized whole egg products 0 2 0.5 Pasteurized yolk products 0 1 0 Pasteurized albumen products 0 1 0

Since chi square values of commercial raw yolk samples and all types of pasteurized egg products were <3.84, this indicated that the proportions of confirmed coliforms detected by the cultural and MicroFoss assays were not statistically different (P>0.05).

4.4 DISCUSSION

The MicroFoss system has now been applied as an alternative, rapid method to estimate the coliform populations in a range of foods, including ground pork (Russell, 2001), dairy products (Firstenberg-Eden et al., 2002; Firstenberg-Eden et al., 2004), ground beef (Odumeru and Belvedere 2002; Firstenberg-Eden et al., 2004), pork sausage raw chicken raw eggs and chocolate (Firstenberg-Eden et al., 2004). Linear regression relationships between the MicroFoss DTs and the cultural coliform counts (log CFU/g) have been established in these commodities, with correlation coefficients in excess of -

0.90. These calibration curves then become the basis for subsequent routine use of the

MicroFoss for estimating the overall microbial quality of the product, the presumptive presence of pathogens, the efficacy of food processing, such as pasteurization, and possible post-processing contamination. The MicroFoss assays could be accomplished in less than 24h, whereas approximately 48-72h were needed to obtain results by the cultural method.

138 The shape of the MicroFoss curves for detection of the individual coliforms,

Citrobacter freundii, Enterobacter aerogenes, Escherichia coli and Klebsiella pneumoniae, and a cocktail of these four coliforms inoculated into whole egg homogenates derived from shell eggs followed the pre-accelerating, accelerating and declining phases, which was similar to lag, log and decline phases of the normal bacterial growth curve, as shown in Figure 4.1. The liquid medium used for detection of coliforms, namely, coliform medium (CM), was specifically formulated by Firstenberg-

Eden and Klein (1983) to enhance the rapid impedimetric procedure for selective detection and estimation of coliforms. Russell, Fletcher and Cox (1995) stated that CM was an effective medium for selective detection and estimation of coliforms in broiler carcasses, particularly in temperature abused samples, using impedance microbiological techniques. Moreover, Firstenberg-Eden et al. (1984) indicated the advantages and superior selectivity of CM to other conventional coliform media when applied to the impedance method to detect and estimate coliforms in dairy products. Consequently,

CM has been used as a basal medium for detection of coliforms by the MicroFoss system (bioMerieux Vitek, Inc., Durham, NC, USA) (Shelef and Firstenberg-Eden,

1997; Russell, 2001). For preparation of the vials for detection, CM was supplemented with 2% dextrose (CMD) (BioSys, Inc., Ann Arbor, MI, USA) to allow bromcresol purple to completely migrate into the agar detection zone of the assay vial (Russell,

2001). The CMD medium contained proteose peptone No. 3 (Difco, Detroit, MI, USA) as a main nitrogen source, lactose and glucose as a main carbon source (Oxoid,

Ogdensburg, NY, USA) and bromocresol purple (Sigma, St Louis, MO, USA) as pH indicator. The colour of the non-inoculated CMD is reddish purple that gives an initial optical unit of 130 in the MicroFoss detection system. Acid production from lactose and glucose metabolism by coliform growth in the inoculated CMD shifted the pH from

139 neutral to acid, which changed the medium colour from reddish purple to yellow, as detected by bromocresol purple. The pre-accelerating and accelerating phases or the flat and sharp rise sections of the detection curve, respectively, represented time required for coliform growth to reach the detection threshold level of approximately 106-107

CFU/mL, as recorded as detection time (Firstenberg-Eden and Klein, 1983; Shelef et al., 1998; Russell, 2001). As a result of coliform growth to the detection threshold level of 106-107 CFU/mL, acid produced gives an increase in transmittance or the sharp rise of the optical unit from about 200 to 400-500. Thereafter, degradation of proteose peptone or the bovine β-casein comprising of sequences of amino acids (Andrews,

1978) by coliforms inoculated into the CMD medium gives various amino acids.

Subsequently, coliforms decarboxylate these amino acids (Brenner, 1984) to amines, causing the pH of the medium to increase, and thereby change the colour of the indicator dye from yellow back to reddish purple. This results in a decrease in transmittance and a decline of optical units as seen in the curves in Figure 4.1.

The range of optical unit shift, DTs and metabolic growth profiles of the MicroFoss curves for detection of individual coliforms and a cocktail mixture of the four coliforms inoculated to whole egg homogenates were comparable. Although different coliform species possess different taxonomy and biochemical activities, the CMD was specifically developed for broad detection of coliforms. Firstenberg-Eden et al. (2002) demonstrated similar metabolic growth profiles for Enterobacter aerogenes and E. coli in the MicroFoss Enterobacteriaceae medium and the MicroFoss coliform medium

(CMD), respectively. Although detections of Enterobacter aerogenes and E. coli were compared between different MicroFoss media, both media contained glucose that was metabolised by these two strains of coliforms, in which similar growth profiles were

140 observed. Hence, the MicroFoss assay for coliforms was applicable for detection of either individual strains of coliforms or a cocktail of the four coliforms (Figure 4.1).

As different coliforms metabolise matrix in slightly different ways, variability of detection times and counts of different cocktails of coliforms inoculated into homogenate samples of whole egg were observed, with the variation coefficients of 0.05 and 0.09 for 102 and 10 CFU/mL of coliforms, respectively.

This Chapter evaluated the MicroFoss assay for coliforms in eggs and egg products, as the MicroFoss assay for these products had not yet been established. Calibration curves to estimate coliform populations in unpasteurized liquid eggs and pasteurized egg products were separately developed due to differences in the nature of each product and likely lower populations that need to be detected in the pasteurized products. Results of this study indicated that the linear relationship of the cultural coliform counts and the

MicroFoss DTs of all types of egg products (n >100) inoculated with different populations of coliforms were well correlative, with correlation coefficients in excess of

–0.93 to –0.95. Firstenberg-Eden et al. (2004) indicated high correlation between coliform counts and DTs of analyses of 49 artificially contaminated raw egg samples, with a correlation coefficient of –0.98. Subsequently, the effectiveness of these calibration curves to estimate the MicroFoss coliform populations in unpasteurized liquid eggs was tested with commercial raw egg samples.

Coliforms counts in excess of 78% of 45, 24 and 32 samples of commercial raw whole egg, raw yolk and raw albumen tested were comparable. In most cases, the cultural coliform counts, which were corrected coliform counts after confirmation in LT broth, for all types of commercial raw egg samples were higher than coliform counts estimated by the MicroFoss method. Since an approximation of the square root of the number of typical colonies on VRBA to a maximum of 10 colonies were selected for confirmation

141 in LT broth (Section 4.2.5), the cultural coliform counts obtained were, in effect, only estimates numbers. Concurrently, the generation time of each coliform detected should be similar in order to obtain corresponding DTs with the initial coliform concentration in these commercial raw egg samples. Firstenberg-Eden and Klein (1983) indicated that in order for the impedance detection times to correlate with the initial bacterial concentration of a sample the generation time of each member of the populations being tested must be similar under the conditions of the test. This could contribute to the differences in coliform counts obtained from the cultural and MicroFoss methods, as shown in Tables 4.2, 4.3 and 4.4 for raw whole egg, raw yolk, and raw albumen, respectively.

Findings in this study showed that a number of typical colonies on VRBA from detection of coliforms in commercial raw egg samples were confirmed as non- coliforms, as they did not produce gas in LT broth. Firstenberg-Eden and Klein (1983) stated that non-coliforms isolated from beef and ground beef samples that produced typical coliform colonies on VRBA plates were Proteus, Serratia, Aeromonas, and

Pseudomonas. These organisms were also reported by Hussong et al. (1981) to interfere with the coliform test using MPN. Firstenberg-Eden et al. (2004) reported that the organisms producing typical colonies on VRBA without producing gas in brilliant green lactose broth obtained from chicken rinse were identified as Aeromonas hydrophila,

Pantoea spp., Pseudomonas spp., Serratia liquefaciens, and Vibrio fluvialis by the API

20E kit (bioMerieux, Hazelwood, Mo.). On the other hand, all coliform-positive vials, except for one false positive result from raw yolk (Table 4.5), were confirmed as coliforms by gas production in LT broth. Hence, the MicroFoss could possibly be used to estimate coliform populations in unpasteurized liquid eggs.

142 The false positive result obtained from the MicroFoss method when coliforms was not detected on VRBA in raw yolk, as shown in Table 4.3, could cause from interference in detection by the MicroFoss system under the extended incubation period.

Pasteurization usually reduces or eliminates the population of initial organisms including coliforms often present in unpasteurized liquid eggs (Delves-Broughton and

Board, 2000). However, coliforms were detected in all types of pasteurized egg products tested in this study, as shown in Sections 4.3.4.1, 4.3.4.2, and 4.3.4.3. The actual cause of the presence of coliforms in commercial pasteurized egg products used in this study was unknown. Possible causes could be the inadequacy of pasteurization or post-pasteurization contamination. Coliforms that survive pasteurization could resuscitate once the optimum conditions are introduced. For example, storage temperature fluctuation during transportation and along the distribution chain of these pasteurized egg products. If post-pasteurization in chilled liquid pasteurized whole egg products, which is mainly caused by coliforms, can be avoided, these products can be safely stored for a minimum of 6 days without significant increases in bacterial counts

(Forsythe and Hayes, 2000).

The cultural method failed to detect coliforms in certain numbers of samples of pasteurized whole egg products, pasteurized yolk products, and pasteurized albumen products, whereas coliforms were detected by the MicroFoss method, as shown in

Sections 4.3.4.1, 4.3.4.2, and 4.3.4.3, respectively. This could be due to limit of detection ability of the cultural plate method in products with not only expectedly low numbers of coliforms, but also in the injured stage like those in pasteurized egg products that may survive pasteurization and were undergone further processing steps as freezing. A study of Ray and Speck (1978) indicated that detection of injured coliforms present in cottage cheese, which was made from pasteurized skim milk, using the

143 cultural plate method with VRBA was enhanced when 5 mL of TSA was used to pour plate and left at room temperature for 1h to allow repair of the injured cells prior to being overlaid by 10 to 12 mL of VRBA. This showed the limit of coliform detection in heat-treated food products by the cultural plate method using VRBA. Similar case was reported by Firstenberg-Eden et al. (2004), in which one of the five VRBA plates failed to detect coliforms inoculated to five replicate samples of pasteurized milk, whereas the

MicroFoss assay detected the presence of coliforms in all five replicate samples.

According Table 4.6, false positive results of coliform detection in all types of pasteurized egg products by the MicroFoss assay were observed. This could be a result of non-coliforms presenting in these pasteurized egg products that could utilize lactose and dextrose in the CMD medium and change medium colour from purple to yellow sufficiently for detection to occur. Serratia was reported to yield an early detection in

CD medium used in the impedimetric method and could therefore be mistaken for coliform detection (Firstenberg-Eden and Klein, 1983).

Although many samples of the pasteurized egg products commercially available showed no detectable coliforms by either the cultural or MicroFoss assay, certain samples yielded confirmed cultural coliform results and positive-vials confirmed as coliforms could be used to generate calibration curves for each type of pasteurized egg products.

In this study, numbers of coliforms presenting in these commercial pasteurized egg products could be promoted by temperature abuse at 25°C for 1-3 days. Generation of a calibration curve using naturally contaminated pasteurized egg products in this study should give an accurate estimation of coliform populations in pasteurized egg products.

Since only a limited number of samples were available for validation analyses, thereby diminishing the strength of the conclusions. Nevertheless, the results of this study indicated good correlation between coliform counts obtained by the cultural and the

144 MicroFoss methods for all types of commercial pasteurized egg products with correlation coefficients in excess of –0.90.

For each type of pasteurized egg products, there were samples for which positive results were detected by the MicroFoss method, whereas the cultural method failed to detect coliforms in these samples. These positive results from the MicroFoss method were confirmed as coliforms by an observation of gas production in the LT broths. This difference could be explained by the greater sample size (5 mL) used in the MicroFoss method, compared to 2.5 mL used in the cultural plating method, with several plates per sample being used (Section 4.2.5). However, some of the data from the MicroFoss assay were confirmed as false positive results. Hence, specificity of the MicroFoss assay for coliforms had to be evaluated and this is discussed in the subsequent section.

According to Ricke et al. (2001), raw liquid eggs harbour coliform counts from 102 to

105 CFU/g. Microbiological analyses for grading raw whole egg, of unsatisfactory quality with respect to coliforms, could be achieved effectively in faster time by the

MicroFoss system. For example, raw whole egg, whose coliforms populations exceed

105 CFU/g would be detected within 5h, and withdrawn from the manufacturing line.

Normally, raw materials would have to be held for about 48-72h until completion of the conventional cultural methods. High coliform counts in unpasteurized liquid eggs indicate poor hygiene and sanitation as well as improper handling, and might influence coliform and microbial quality of pasteurized egg products. In Australia, coliforms in pasteurized egg products are not expected to exceed 10 CFU/g. Hence, pasteurized whole egg, pasteurized yolk and pasteurized albumen products that gave DT values earlier than 14-15h would be rejected due to unsatisfactory coliform quality, suggesting improper pasteurization.

145 The above findings indicated high correlations between the cultural method and the

MicroFoss assay in eggs and egg products and suggested that the MicroFoss assay could be effective for estimation of coliforms in eggs and egg products. Further evaluation of the MicroFoss in regards to performance indicators and repeatability are discussed in the following Sections.

Accuracy

Accuracy of the MicroFoss method was determined by its false negative and false positive rates, which must be minimal or preferably zero (AS/NZS 4659.3-1999; de

Boer and Beumer, 1999). False negative rate should be less than 0.5%, whilst false positive rate should not exceed the range of 1 to 5% (Cox and Fleet, 2003). The false negative rate of the MicroFoss assay for coliforms was 0%, based on 45, 24 and 32 samples of commercial raw whole egg, raw yolk and raw albumen, respectively, and

112, 56 and 55 samples of pasteurized whole egg, pasteurized yolk and pasteurized albumen products, respectively tested. The false positive rates of the MicroFoss assay for coliforms were in the range of 2%-5%, based on 112, 56 and 55 samples of pasteurized whole egg, pasteurized yolk and pasteurized albumen products, respectively tested. Such low false negative and false positive rates indicated the high accuracy of the MicroFoss assay for coliform detection.

Sensitivity

Sensitivity rate also reflects the accuracy of the MicroFoss assay (de Boer and Beumer,

1999). Sensitivity rate of the MicroFoss assay for coliforms obtained was 100%, based on numbers of commercial raw egg samples and pasteurized egg products tested. This finding indicated high accuracy of the MicroFoss assay for coliform detection.

146 Concurrently, the limit of detection also exhibits sensitivity of the MicroFoss assay for coliforms (Cox and Fleet, 2003). The MicroFoss assay detected <10 CFU/g of confirmed coliforms in commercial raw egg samples and pasteurized egg products tested. This finding indicated high sensitivity and accuracy of the MicroFoss assay for coliforms.

Specificity

Specificity rate also indicates accuracy of the MicroFoss assay (de Boer and Beumer,

1999). Specificity rates of the MicroFoss assay for coliforms were in the range of 95%-

98%, based on numbers of commercial pasteurized egg products tested.

SD values of the MicroFoss coliform counts were less than the cultural coliform counts, particularly at low populations (≤10 CFU/g) in commercial raw egg samples and this implied that the MicroFoss assay could produce repeatable results.

Proportions of confirmed coliforms detected by the cultural and MicroFoss assays were not statistically different (P>0.05) (McNemar’s test, Section 4.3.7). Results of

McNemar’s test in conjunction with results of performance indicators (false negative, false positive, sensitivity and specificity rates) provided a total assessment of the

MicroFoss assay as an accurate alternative method to detect coliforms in eggs and egg products.

Repeatability

In most cases, smaller SD values of the MicroFoss coliform counts than the cultural coliform counts in commercial raw egg samples were obtained, particularly at moderately low populations (≤102 CFU/g) in raw albumen samples. This finding was

147 useful since coliforms are naturally present in low numbers. This would benefit coliform estimations in eggs and other food products using the MicroFoss method, as coliforms are naturally present in low numbers. Findings in this study indicated that the

MicroFoss coliform assay could be likely to give repeatable results on an ongoing, routine basis.

Speed

An acceptable rapid assay should provide an accurate result within hours or generally within 24h (de Boer and Beumer, 1999). Coliform estimations in unpasteurized liquid eggs and pasteurized egg products were generally completed in 2h for samples containing 107 CFU/g and in 15h for samples with <10 CFU/g. This time frame is much shorter than the 24h to obtain presumptive results on VRBA plates and an additional 24-

48h for confirmation in LT broth. Consequently, the speed of the MicroFoss assay offers a significant time advantage if quality assurance programs require coliform analyses. Some pasteurized albumen samples may require longer times in order to obtain a MicroFoss DT because they may contain substances that retard or delay coliform growth, such as lysozyme. Lysozyme is shown not to be inactivated by heat treatment (62°C) employed in pasteurization (Henderson and Robinson, 1969).

Simplicity

Sample preparation and operation procedure of detection of target organisms by the

MicroFoss assay were simple and easy to follow. Liquid samples can be directly added into the ready-to-use MicroFoss vials, whereas solid or semi-solid samples only require

10-1 dilution prior to inoculating to the vials.

148 This study suggested high correlations between the cultural coliform results and the

MicroFoss DTs of analyses of unpasteurized liquid eggs and pasteurized egg products, with correlations of the two measurements similar to or higher than those reported in other studies (Russell, 2001; Firstenberg-Eden et al., 2002; Odumeru and Belvedere,

2002; Firstenberg-Eden et al., 2004). Coliform counts obtained from the MicroFoss assay were generally comparable to those determined by the cultural method. Moreover, the MicroFoss method was shown to have relevant or greater accuracy, sensitivity, specificity, and repeatability to the cultural method; and provided a significant advantage of time require to obtain results. From this Chapter, it can be concluded that the MicroFoss system could be used as an alternative method to rapidly detect and estimate coliform populations in unpasteurized liquid eggs and pasteurized egg products. The ability to obtain rapid results for coliform testing will assist and benefit egg manufacturers by providing for more effective control of the microbial quality of eggs and egg products. Corrective action could be more effectively implemented in the case of any eggs or egg products failing to meet the specifications. As a consequence, the rapid determination of coliforms in eggs will facilitate the early distribution of eggs to markets and consumers, as eggs are a perishable food.

149 CHAPTER 5

Evaluation of the MicroFoss system for estimating

Escherichia coli in eggs and egg products

5.1 INTRODUCTION

Escherichia coli (E. coli) is a member of the family Enterobacteriaceae as well as being a member of the coliform group. It is a Gram-negative, non-sporeforming rod which is facultative anaerobic and oxidase-negative (Brenner, 1984; Adams and Moss, 2000;

Batt, 2000). E. coli is a principal component of the so-called faecal coliform group in which E. coli type I and other ‘faecal coliforms’ (better termed thermotolerant coliforms) can ferment lactose and produce gas in brilliant green lactose bile (BGLB) broth at 44.5°C. The group is used generally as an indicator of sanitation, whereas E. coli is more specific and appropriate as an indication of faecal contamination, suitability and safety of foods for human consumption. Since E. coli and some enteric pathogens are present in faeces and have the same survival rates, E. coli has therefore been used as an index organism, its presence suggesting the presence of faeces-borne pathogens.

Typically, E. coli is present in larger numbers and is more readily detectable than the associated pathogens. However, the occurrence of E. coli in foods does not necessarily imply food is hazardous; this needs to be confirmed by specific tests (Banwart, 1989;

Adams and Moss, 2000; Batt, 2000; Craven et al., 2003). Also, its absence does not always assure that the food is safe and free of pathogens (Adams and Moss, 2000;

Craven et al., 2003).

The presence of E. coli in unpasteurized liquid eggs and pasteurized egg products significantly indicates unsanitary conditions of handling and processing of eggs (Ricke

150 et al., 2001). Although the most probable number (MPN) technique has been widely used to estimate the number of E. coli in foods including eggs and egg products, it encounters several limitations related to its precision (Jay, 2000; Williams and Busta,

2000; Craven et al., 2003), tediousness (Presland, 2005), laboratory intensive procedure

(Jay, 2000; Presland, 2005), the lack of opportunity to observe colony morphology of organisms present (Jay, 2000) and speed (Craven et al., 2003). In recent years, the needs for reliable, rapid analytical methods have increased due to increased demands in quality assurance of foods and food safety. Since eggs are a perishable food, it is beneficial to obtain timely microbiological results to facilitate an early release of the products to consumers. Various techniques including rapid assays for enumeration and detection of E. coli in foods (Blackburn and McCarthy, 2000; Craven et al., 2003) have been reviewed. These methods include a MicroFoss assay for E. coli (Russell, 2000;

Russell, 2001; Odumeru and Belvedere, 2002; Cox and Fleet, 2003).

Although the MicroFoss E. coli assay is commercially available, its application to unpasteurized liquid eggs and pasteurized egg products was not yet been established.

The aim of this Chapter was to evaluate the MicroFoss assay for estimating E. coli populations in unpasteurized liquid eggs and pasteurized egg products. Unpasteurized liquid eggs and pasteurized egg products were simultaneously analysed for E. coli counts by the cultural spread plate method using eosin methylene blue (EMB) agar

(Oxoid) and DTs by the MicroFoss assay, after which the results from the two methods were correlated to generate calibration curves. Since estimation of E. coli using the

MPN technique gives the statistical estimate of the mean number of E. coli populations present in food samples within 95% confidence intervals (Williams and Busta, 2000;

Presland, 2005), the MPN technique should not be used as a method to enumerate E. coli present in food sample (Craven et al., 2003). Calibration curves for unpasteurized

151 liquid eggs were generated from egg homogenates derived from shell eggs inoculated with different populations of E. coli. Thereafter, effectiveness of these calibration curves was examined by applying their linear equations to estimate E. coli populations in commercial raw egg samples. The MicroFoss assay for E. coli was evaluated for accuracy, sensitivity, specificity, reproducibility, and speed of detection to determine its use as an alternative method to the cultural method. The Chi square test and t-test were used to compare the populations of confirmed E. coli obtained using the MicroFoss assay and significant differences in E. coli counts obtained from the cultural and

MicroFoss assays, respectively.

5.2 MATERIALS AND METHODS

5.2.1 Materials

Shell eggs and commercial raw egg samples and pasteurized egg products used in this research are as described in Section 3.2.1.

5.2.2 Sample handling prior to testing

Handling of shell eggs and frozen pasteurized egg products are as described in Section

3.2.2.

5.2.3 Inoculum preparation

A reference culture of E. coli UNSW 048200 or ATCC 11775 was used to evaluate the

MicroFoss assay to estimate and detect E. coli in eggs and egg products. This culture, which was originally received from the culture collection of the School of

Biotechnology and Biomolecular Sciences, UNSW, was maintained by the Food

Science and Technology group, School of Chemical Sciences and Engineering, UNSW.

152 The stock culture was maintained on slant of TSA at 5°C and sub-cultured monthly.

Before use in experiments, it was tested for purity by streaking onto fresh plates of

TSA.

Inocula were prepared by transferring a single colony of biomass from a plate of TSA into 10 mL of TSB and incubating at 37°C for 24h. The broth culture was serially diluted 1:10 using 9 mL of 0.1% peptone water to give suspensions of different populations for inoculation into egg homogenates derived from shell eggs.

5.2.4 Sample preparation

5.2.4.1 Inoculation of samples with defined populations of E. coli

Egg homogenates derived from shell eggs for inoculation with different levels of E. coli populations were prepared according to AS 1766.3.8-1991. Volume (0.1 mL) of E. coli suspension was individually inoculated into 10g of homogenates of shell egg derivatives

(whole egg, yolk and albumen) to give samples with final populations of <10 to 107

CFU/g. Homogenates were aseptically prepared from shell eggs purchased from the supermarket as described in Section 3.2.1.1, and assumed to be sterile or minimally contaminated. The inoculated samples were then diluted with 90 mL of 0.1% peptone water to obtain samples at 10-1 dilution. The mixtures were thoroughly blended using a

Stomacher for 2 min to yield an homogenate. An aliquot (1 mL) of samples at 10-1 and further dilutions was serially diluted with 9 mL of 0.1% peptone water to prepare for the analyses. These samples were simultaneously analysed for E. coli populations by the cultural spread plate assay using EMB agar, and DTs by the MicroFoss assay, in which these two measurements were correlated to form linear calibration curves generated by the MicroFoss system as a basis for the enumeration of E. coli in unpasteurized liquid eggs by the MicroFoss assay.

153 5.2.4.2 Analysis of commercial egg samples for E. coli

Portions (10g) of commercial raw egg samples and pasteurized egg products (Section

3.2.1.2) were diluted 1:10 with 90 mL of 0.1% peptone water and thoroughly mixed using a Stomacher until homogenous. Preparations of samples for simultaneous analyses by the cultural and the MicroFoss methods were as described in the above

Section.

To create samples with various populations of E. coli, raw and pasteurized samples of each product were abused by storage for up to 3 days. Development of populations of indigenous E. coli in the products was promoted by extended storage at 5°C, or at 25°C, to simulate potential abuse storage conditions. Samples at 5°C were stored for 1-3 days, with the exception of pasteurized albumen, which was stored for up to 5 days. Samples were abused at 25°C by storage for 1 day, except pasteurized albumen, which was held for up to 3 days.

••• Determination of the effectiveness of the MicroFoss assay to estimate E. coli in unpasteurized liquid eggs by testing with commercial raw egg samples

Differences between the cultural E. coli counts and the counts derived from calculations of DTs based on linear equations describing the calibration curves in commercial raw egg samples were compared to determine the effectiveness of the MicroFoss assay for

E. coli in unpasteurized liquid eggs.

154 ••• Generation of calibration curves as a basis for enumeration of E. coli in pasteurized egg products by the MicroFoss assay

The cultural E. coli counts and DTs of commercial pasteurized egg products were correlated to form linear calibration curves by the MicroFoss system as a basis for the enumeration of E. coli in pasteurized egg products by the MicroFoss assay.

5.2.5 E. coli cultural method

Estimations of E. coli populations using the surface spread method were performed according to AS 1766.1.4-1991 and Swanson et al. (2001). EMB agar (Levine) was used in this study for preliminary detection and differentiation of E. coli from coliforms, as recommended by Andrews (2000a) and Kornacki and Johnson (2001). EMB plates were prepared according to the Oxoid instructions 24 to 48h prior to testing. Aliqouts

(0.1 mL) of appropriate sample dilutions were pipetted onto pre-poured EMB plates in duplicate. The inoculated suspensions were spread over the surface of the EMB plates, allowed to dry for approximately 30 min, and the plates incubated at 37°C for 24 + 2h.

Typical E. coli colonies appearing on EMB agar were characterised as 2-3 mm in diameter, with a green metallic sheen by reflected light and a dark purple/blue-black centre by transmitted light. All colonies appearing on plates were counted, including those with high numbers that appeared without confluence or diminution in size as a result of overcrowding. Colony counts were calculated from results of the mean colony counts multiplied by dilution factor and subsequently divided by the volume of inoculum (0.1 mL). These results were recorded as presumptive E. coli (AS 1766.1.4-

1991; Messer et al., 2000).

Presumptive E. coli results were confirmed by performing an IMViC test (Banwart,

1989; AS 4276.6-1995; Andrews, 2000a; Jay, 2000; Kornacki and Johnson, 2001;

155 Tortorello, 2003). Typical Escherichia coli colonies possess the + + - - pattern of the

IMViC test (Jay, 2000; Pandey et al., 2000). Presumptive colonies of E. coli were selected and prepared for use in the IMViC test by streaking onto NA plates and incubated at 37°C + 1°C for 24h. The IMViC test is described in the following Sections.

• Indole production: A single colony of biomass from the NA plate was inoculated into

10 mL of tryptone water (tryptone water, Oxoid), which was previously heated to 45°C.

After incubation in the water bath (HTB-150-D, Thermoline Scientific Equipment Pty.

Ltd., Smithfield, Australia) set at 44°C + 1°C for 48h + 2h, 0.5 mL of the indole reagent

(Kovac’s reagent) was added to the cultured broths and the contents were mixed well by gently shaking. Results were read after 1 min. Indole production is exhibited by a cherry red/pinkish red ring on the surface of the broth culture (AS 5013.15-2004).

• Methyl red reaction: A single colony of biomass from the NA plate was inoculated into 10 mL of MRVP medium (Clarks and Lubs medium-buffered glucose broth,

Oxoid). After incubation at 35°C for 48h + 2h, an aliquot (5 mL) of the cultured MRVP broths was transferred into a sterile test tube, to which 5-6 drops of methyl red indicator solution was added. Methyl red indicator solution was prepared by dissolving 0.1g of methyl red (BDH Chemicals Ltd., Poole, England) in 300 mL of absolute ethanol and diluting to 500 mL with sterile distilled water. Results were read immediately. A positive result is indicated by red colour throughout the tube, whilst yellow colour represents a negative reaction (Andrews, 2000b).

• Voges-Proskauer (VP) reaction: VP reagent is a composite of two reagents, which are

5% α-naphthol solution and 40% potassium hydroxide (KOH) solution. 5% α-naphthol solution was prepared by dissolving 5.0 g of α-naphthol (BDH Chemicals Ltd., Poole,

England) in 100 mL of absolute ethanol. The KOH solution was prepared by dissolving

40 g of KOH (May & Baker Australia Pty Ltd., West Footscray, Australia) in sterile

156 distilled water and diluting to 100 mL. The VP test was conducted by transferring 1 mL of the cultured MRVP broths into a sterile test tube, to which 0.6 mL of the 5% α- naphthol solution was added and the broths were mixed thoroughly. An aliquot (0.2 mL) of 40% KOH solution was subsequently added to the cultured broths, which were then well shaken. Reaction was intensified and speeded up by the addition of a small amount of creatine crystals (Ajax Chemicals Ltd., Sydney, Australia) to the broths.

Results were read after 4h. A positive reaction is indicated by a development of eosin pink to ruby colour throughout the broths as a result of acetylmethylcarbinol production, whilst colour of the broths exhibiting a negative result remains unchanged (Andrews,

2000b).

• Citrate utilisation: Slants of Simmons Citrate agar (Oxoid) were prepared 24-48h prior to testing. A single colony of biomass from the NA plate was streaked onto the slant and stabbed into the butt of Simmons Citrate agar slant and incubated at 35°C for 96 + 2h. A positive reaction is indicated by growth with colour change of medium from green to intense blue on the slant and throughout the medium, whereas a negative reaction is exhibited by no growth to trace growth with no colour change of the medium (medium remains dark green) (Andrews, 2000b; Kornacki and Johnson, 2001).

157 5.2.6 The MicroFoss E. coli test

An aliquot (5 mL) of each egg sample at the 10-1 dilution was added to an E. coli vials

(Foss Pacific, North Ryde, Australia). The vials were mixed by gently inversion 10 times, then inserted into the MicroFoss instrument and monitored during their incubation, according to the manufacturer’s instructions. Details of test configurations are shown in Table 5.1.

Table 5.1 Test configurations of the MicroFoss E. coli assay for eggs and egg products

Threshold Skip Shuteye Cutoff Caution Temperature Duration (°C) (h) 12 1 25 10 10 42 24

Source: Derived from the MicroFoss manual.

The vials giving a detection time, indicating the presence of E. coli, were further confirmed by the IMViC test. Positive vials for E. coli were prepared for use in the

IMViC test by streaking a loopful of culture from the positive vials on NA plates and incubating at 37°C for 24h. Procedures of the IMViC test are as described in Section

5.2.5. Vials showing positive results and exhibiting typical reactions of E. coli for the

IMViC test were recorded as confirmed E. coli, whereas those that did not exhibit typical reaction of E. coli for the IMViC test were recorded as false positive results.

158 5.2.7 Statistical analyses

5.2.7.1 Effectiveness of the MicroFoss assay for E. coli

••• Correlations of the data set to produce a calibration curve

Correlations of the calibration curves, which were a linear regression between the cultural E. coli counts (log CFU/g) and the MicroFoss DTs, were determined by correlation coefficients (r value) processed by the MicroFoss system. Linear equations were generated by the system.

••• Effectiveness of E. coli estimations using the calibration curves

Effectiveness of the calibration curves to estimate E. coli populations in unpasteurized liquid eggs was tested with commercial raw egg samples (Section 3.2.1.2), as determined by standardised residuals at 95% prediction intervals.

5.2.7.2 Performance indicators

Performance indicators, which were sensitivity rate, specificity rate, false negative rate, false positive rate, repeatability, and standard deviations, were calculated for evaluations of the MicroFoss assay for E. coli.

5.2.7.3 Test for significant difference of positive proportions for the cultural and

MicroFoss assays

McNemar’s test or a Chi square (χ2) value was assessed to compare the proportions confirmed positive for E. coli between the cultural and MicroFoss methods.

5.2.7.4 Comparison of E. coli counts from the cultural and MicroFoss assays

Student’s t-test for two samples, assuming equal variances, for determining statistical difference at the 5% level of significance of the mean of E. coli counts from the cultural and MicroFoss assays were calculated.

159 5.3 RESULTS

5.3.1 Application of the MicroFoss system to estimate E. coli in unpasteurized egg

5.3.1.1 Unpasteurized whole egg

Whole egg homogenates derived from shell eggs inoculated with different populations of E. coli were simultaneously analysed for E. coli counts by the cultural spread plate method and DTs by the MicroFoss system.

Figure 5.1 shows the linear correlation between E. coli counts by the cultural spread plate method (log CFU/g) and DTs as estimated by the MicroFoss system for 119 samples of whole egg homogenates inoculated with different E. coli populations.

Detection times ranged from 2h for a sample containing 107 CFU/g to 9.5h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.97.

Figure 5.1 Linear regression relationship between the log-transformed populations of E. coli and the MicroFoss detection times for whole egg homogenate samples inoculated with different populations of E. coli. The regression is described by the equation log

CFU/g = 9.671 – 1.002 (DT), (r = -0.97, n = 119).

160 The cultural E. coli counts in commercial raw whole egg samples were compared to E. coli counts obtained from mathematically transforming DTs using linear equation describing the calibration curve, log CFU/g = 9.671 – 1.002 (DT), as shown in Figure

5.1 to calculate standardised residuals at 95% prediction intervals, as shown in Table

5.2. Standard error of the estimate of the calibration curve was 0.45838.

After thawing, commercial raw whole egg samples generally gave E. coli counts of 103-

104 CFU/g, as determined by the cultural spread plate method.

Table 5.2 A comparison of the cultural and MicroFoss E. coli counts of commercial raw whole egg samples and their standardised residuals at 95% prediction intervals

n DT EMB counts MF countsa ZRESIDb n DT EMB counts MF countsa ZRESIDb (h) (log CFU/g) (log CFU/g) 95% P.I. (h) (log CFU/g) (log CFU/g) 95% P.I. 1 6.0 4.41 3.66 1.64 24 6.1 4.30 3.56 1.61 2 6.0 4.46 3.66 1.75 25 6.2 4.27 3.46 1.77 3 7.1 3.37 2.56 1.77 26 6.1 4.27 3.56 1.55 4 7.0 3.70 2.66 2.27 27 6.1 4.33 3.56 1.68 5 6.8 4.45 2.86 3.47 28 6.0 4.26 3.66 1.31 6 6.9 2.95 2.76 0.41 29 5.8 4.28 3.86 0.92 7 7.2 4.54 2.46 4.54 30 6.1 4.25 3.56 1.51 8 7.3 3.10 2.36 1.61 31 6.4 4.22 3.26 2.09 9 6.7 3.70 2.96 1.61 32 6.2 3.84 3.46 0.83 10 6.4 3.70 3.26 0.96 33 6.0 3.83 3.66 0.37 11 3.5 7.06 6.16 1.96 12 4.3 5.90 5.36 1.18 13 6.2 3.60 3.46 0.31 14 6.2 3.57 3.46 0.24 15 6.1 4.54 3.56 2.14 16 6.1 3.42 3.56 -0.31 17 6.1 3.71 3.56 0.33 18 6.6 3.77 3.06 1.55 19 6.4 3.77 3.26 1.11 20 6.1 3.83 3.56 0.59 21 6.2 3.92 3.46 1.00 22 6.2 4.27 3.46 1.77 23 6.0 3.98 3.66 0.70 a E. coli counts of commercial raw whole egg samples determined by the MicroFoss method was obtained from mathematic transformation of DTs to E. coli counts using linear equation describing the calibration curve, log CFU/g = 9.671 – 1.002 (DT). b ZRESID 95% P.I. = Standardised residuals at 95% prediction intervals. Black and red fonts represented commercial raw whole egg samples, whose standardised residuals at 95% prediction intervals fell within and exceeded the range of + 2, respectively.

161 Among 33 commercial raw whole egg samples tested, E. coli counts were comparable for the cultural and MicroFoss assays in 28 samples (84.85%) and different in 5 samples

(15.15%), as indicated by their standardised residuals at 95% prediction intervals, highlighted in black and red fonts in Table 5.2.

5.3.1.2 Unpasteurized yolk

Yolk homogenates derived from shell eggs and inoculated with different populations of

E. coli were simultaneously analysed for E. coli counts by the cultural spread plate method and DTs by the MicroFoss system.

Figure 5.2 shows the linear correlation between E. coli counts by the cultural plating method (log CFU/g) and DTs as estimated by the MicroFoss system for 86 samples of yolk homogenates inoculated with different E. coli populations. DTs ranged from 2h for a sample containing 107 CFU/g to 10h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.98.

Figure 5.2 Linear regression relationship between the log-transformed populations of E. coli and the MicroFoss detection times for yolk homogenate samples inoculated with different populations of E. coli. The regression is described by the equation log CFU/g = 9.541 – 0.968 (DT), (r = -0.98, n = 86).

162 The cultural E. coli counts in commercial raw yolk samples were compared to E. coli counts obtained from mathematically transforming DTs using linear equation describing the calibration curve, log CFU/g = 9.541 – 0.968 (DT), as shown in Figure 5.2 to calculate standardised residuals at 95% prediction intervals, as shown in Table 5.3. The standard error of the estimate of the calibration curve was 0.40467.

After thawing, commercial raw yolk samples generally gave E. coli counts of 10-103

CFU/g, as determined by the cultural spread plate method.

Table 5.3 A comparison of the cultural and MicroFoss E. coli counts of commercial raw yolk samples and their standardised residuals at 95% prediction intervals n DT EMB counts MF countsa ZRESIDb n DT EMB counts MF countsa ZRESIDb (h) (log CFU/g) (log CFU/g) (h) (log CFU/g) (log CFU/g) 1 8.6 2.00 1.22 1.94 25 2.3 7.74 7.31 1.05 2 7.7 2.88 2.09 1.96 26 9.3c No detection 0.54 N/A 3 8.9 1.70 0.93 1.91 27 4.0 5.49 5.67 -0.44 4 8.7 1.70 1.12 1.43 28 9.8 1.70 0.05 4.07 5 8.0 2.24 1.80 1.09 29 10.1 2.00 -0.24 5.52 6 8.0 2.29 1.80 1.22 30 10.0 1.70 -0.14 4.54 7 2.3 7.69 7.31 0.93 31 10.4d No detection N/A N/A 8 2.3 7.19 7.31 -0.31 32 2.3 7.19 7.31 -0.31 9 2.3 7.24 7.31 -0.18 33 2.3 7.25 7.31 -0.16 10 3.0 7.35 6.64 1.76 34 9.9 1.70 -0.04 4.31 11 3.1 7.28 6.54 1.83 35 9.9 1.70 -0.04 4.31 12 2.7 6.99 6.93 0.15 36 10.3 2.00 -0.43 6.00 13 2.7 7.13 6.93 0.50 37 9.7c No detection 0.15 N/A 14 10.4 1.70 -0.53 5.50 38 2.3 7.11 7.31 -0.51 15 2.3 7.67 7.31 0.88 39 10.7d No detection N/A N/A 16 3.5 6.23 6.15 0.19 40 10.0d No detection N/A N/A 17 2.5 7.31 7.12 0.47 41 2.3 6.81 7.31 -1.25 18 2.9 7.41 6.73 1.67 42 8.2 2.40 1.60 1.97 19 3.1 6.29 6.54 -0.62 43 9.4c No detection 0.44 N/A 20 4.4 5.22 5.28 -0.15 44 8.3 2.18 1.51 1.66 21 4.1 5.30 5.57 -0.67 45 6.8 3.08 2.96 0.30 22 9.4c No detection 0.44 N/A 46 6.7 3.20 3.06 0.36 23 4.2 5.39 5.48 -0.21 47 6.6 3.06 3.15 -0.23 24 10.2 1.70 -0.33 5.02 48 6.3 3.02 3.44 -1.04 a E. coli counts of commercial raw yolk samples determined by the MicroFoss method was obtained from mathematic transformation of DTs to E. coli counts using linear equation describing the calibration curve, log CFU/g = 9.541 – 0.968 (DT). b ZRESID 95% P.I. = Standardised residuals at 95% prediction intervals. Black and red fonts represented commercial raw yolk samples, whose standardised residuals at 95% prediction intervals fell within and exceeded the range of + 2, respectively. c Detection of confirmed E. coli only by the MicroFoss assay. d Detection of false positive E. coli by the MicroFoss assay. N/A = Not applicable.

163 Among 48 samples examined, positive results in seven samples were obtained by the

MicroFoss method, but the cultural method failed to detect coliforms in these samples.

However, positive results from the MicroFoss method in four out of seven samples were confirmed as E. coli by the IMViC test. As a result, the MicroFoss assay gave false positive results for three samples. Hence, results of 41 samples, for which typical colonies on EMB plates and vials giving detection times were confirmed as E. coli, were used to calculate standardised residuals.

Hence, among 41 commercial raw yolk samples tested, E. coli counts for the cultural and MicroFoss assays were comparable in 33 samples (80.49%) but different in 8 samples (19.51%), as indicated by their standardised residuals at 95% prediction intervals, highlighted in black and red fonts in Table 5.3.

Samples of raw yolk yielding DTs of 9.3h, 9.4h, 9.4h and 9.7h were equivalent to approximate E. coli populations of 4, 3, 3 and 2 CFU/g, respectively based on linear equation describing the calibration curve, log CFU/g = 9.541 – 0.968 (DT), as shown in

Figure 5.2.

5.3.1.3 Unpasteurized albumen

Albumen homogenates derived from shell eggs and inoculated with different populations of E. coli were simultaneously analysed for E. coli counts by the cultural spread plate method and DTs by the MicroFoss system.

Figure 5.3 shows the linear correlation between E. coli counts by the cultural plating method (log CFU/g) and DTs as estimated by the MicroFoss system for 96 samples of albumen homogenates inoculated with different E. coli populations. DTs ranged from

2.5h for a sample containing 107 CFU/g to 11h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.94.

164 Figure 5.3 Linear regression relationship between the log-transformed populations of E. coli and the MicroFoss detection times for albumen homogenate samples inoculated with different populations of E. coli. The regression is described by the equation log

CFU/g = 9.155 – 0.829 (DT), (r = -0.94, n = 96).

The cultural E. coli counts in commercial raw albumen samples were compared to E. coli counts obtained from mathematically transforming DTs using linear equation describing the calibration curve, log CFU/g = 9.155 – 0.829 (DT), as shown in Figure

5.3 to calculate standardised residuals at 95% prediction intervals, as shown in Table

5.4. Standard error of the estimate of the calibration curve was 0.63872.

After thawing, commercial raw albumen samples generally gave E. coli counts of <10 to <102 CFU/g, as determined by the cultural spread plate method.

165 Table 5.4 A comparison of the cultural and MicroFoss E. coli counts of commercial raw albumen samples and their standardised residuals at 95% prediction intervals

n DT EMB counts MF countsa ZRESIDb n DT EMB counts MF countsa ZRESIDb (h) (log CFU/g) (log CFU/g) (h) (log CFU/g) (log CFU/g) 1 12.5 0.70 -1.21 2.99 35 10.9 1.18 0.12 1.66 2 10.6 1.00 0.37 0.99 36 11.0 1.00 0.04 1.51 3 10.5 1.18 0.45 1.14 37 12.0d No detection N/A N/A 4 10.7 1.48 0.28 1.87 38 10.8 1.00 0.20 1.25 5 9.2 2.26 1.53 1.15 39 11.4 0.70 -0.30 1.56 6 9.6 2.27 1.20 1.68 40 10.3 1.00 0.62 0.60 7 8.5 2.36 2.11 0.39 41 11.6 0.70 -0.46 1.82 8 11.0 0.70 0.04 1.04 42 10.1 0.70 0.78 -0.13 9 12.3 0.70 -1.04 2.73 43 10.8 1.18 0.20 1.53 10 10.7 1.48 0.28 1.87 44 9.2 2.54 1.53 1.58 11 10.3 1.40 0.62 1.23 45 10.0 1.65 0.87 1.23 12 10.2 1.85 0.70 1.80 46 9.6 1.65 1.20 0.71 13 11.7 0.70 -0.54 1.95 47 11.7 0.70 -0.54 1.95 14 11.3 1.00 -0.21 1.90 48 10.6 1.40 0.37 1.62 15 10.7 1.40 0.28 1.75 49 11.0 1.18 0.04 1.79 16 9.1 2.08 1.61 0.73 50 10.4 1.78 0.53 1.95 17 9.0 2.48 1.69 1.23 51 10.4 1.00 0.53 0.73 18 9.4 2.40 1.36 1.62 52 9.8 1.00 1.03 -0.05 19 11.1 1.00 -0.05 1.64 53 6.0 5.08 4.18 1.41 20 11.2 1.00 -0.13 1.77 54 5.7 5.19 4.43 1.19 21 12.7 0.70 -1.37 3.25 55 5.6 4.94 4.51 0.67 22 11.4 1.00 -0.30 2.03 56 5.9 5.14 4.26 1.37 23 10.8 1.00 0.20 1.25 24 11.0 1.00 0.04 1.51 25 11.3 1.00 -0.21 1.90 26 11.3 0.70 -0.21 1.43 27 9.8 1.18 1.03 0.23 28 8.7 1.90 1.94 -0.07 29 12.9 0.70 -1.54 3.51 30 10.1c No detection 0.78 N/A 31 12.4 0.70 -1.12 2.86 32 12.9 0.70 -1.54 3.51 33 9.6 1.18 1.20 -0.03 34 10.4 1.00 0.53 0.73 a E. coli counts of commercial raw albumen samples determined by the MicroFoss method was obtained from mathematic transformation of DTs to E. coli counts using linear equation describing the calibration curve, log CFU/g = 9.155 – 0.829 (DT). b ZRESID 95% P.I. = Standardised residuals at 95% prediction intervals. Black and red fonts represented commercial raw albumen samples, whose standardised residuals at 95% prediction intervals fell within and exceeded the range of + 2, respectively. c Detection of confirmed E. coli only by the MicroFoss assay. d Detection of false positive E. coli by the MicroFoss assay. N/A = Not applicable.

166 Among 86 samples examined, no E. coli was detected in 30 samples by either the cultural method or the MicroFoss method. Positive results in two samples were obtained by the MicroFoss method, but the cultural method failed to detect E. coli in these samples. However, positive results from the MicroFoss method in one out of two samples were confirmed as E. coli by the IMViC test. As a result, the MicroFoss assay gave false positive result for one sample. Hence, results of 54 samples, for which typical colonies on EMB plates and vials giving detection times were confirmed as E. coli, were used to calculate standardised residuals.

Hence, among 54 commercial raw albumen samples tested, E. coli counts for the cultural and MicroFoss assays were comparable in 47 samples (87.04%) but different in

7 samples (12.96%), as indicated by their standardised residuals at 95% prediction intervals, highlighted in black and red fonts in Table 5.4.

Sample of raw albumen yielding a DT of 10.1h was equivalent to approximate E. coli populations of 6 CFU/g based on linear equation describing the calibration curve, log

CFU/g = 9.155 – 0.829 (DT), as shown in Figure 5.3.

5.3.2 Application of the MicroFoss system to estimate E. coli in pasteurized egg

products

5.3.2.1 Pasteurized whole egg products

Commercial pasteurized whole egg products were simultaneously analysed for E. coli counts by the cultural plating method and DTs by the MicroFoss system. Among 116 samples examined, no E. coli was detected in 105 samples by either the cultural method or the MicroFoss method. Positive results in three samples were obtained by the

MicroFoss method, but the cultural method failed to detect E. coli in these samples.

However, positive results from the MicroFoss method in one out of three samples were

167 confirmed as E. coli by the IMViC test. As a result, the MicroFoss assay gave false positive results for two samples. Hence, results of eight samples, for which typical colonies on EMB plates and vials giving detection times were confirmed as E. coli, were used to form the calibration curve.

E. coli was generally not detected in most pasteurized whole egg products tested after thawing. For detectable results (2 samples), the counts were generally 010 CFU/g, as determined by the cultural plating method.

Figure 5.4 shows the linear correlation between E. coli counts by the cultural plating method (log CFU/g) and DTs as estimated by the MicroFoss system for 8 samples of commercial pasteurized whole egg products. DTs ranged from 4.5h for a sample containing 106 CFU/g to 10h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.98.

Figure 5.4 Linear regression relationship between the log-transformed populations of E. coli and the MicroFoss detection times for pasteurized whole egg products. The regression is described by the equation log CFU/g = 10.779 – 1.030 (DT), (r = -0.98, n

= 8).

168 Analysis of one pasteurized whole egg product giving DT of 8.9h, which was subsequently confirmed as E. coli by the IMViC test, was equivalent to approximate E. coli populations of 4.10 x 10 CFU/g based on the linear equation, log CFU/g = 10.779 –

1.030 (DT), as shown in Figure 5.4.

5.3.2.2 Pasteurized yolk products

Commercial pasteurized yolk products were simultaneously analysed for E. coli counts by the cultural plating method and DTs by the MicroFoss system. Among 82 samples examined, no E. coli was detected in 37 samples by either the cultural method or the

MicroFoss method. Positive results in six samples were obtained by the MicroFoss method, but the cultural method failed to detect E. coli in these samples. However, positive results from the MicroFoss method in five out of six samples were confirmed as E. coli by the IMViC test. As a result, the MicroFoss assay gave false positive results for one sample. Hence, results of 39 samples, for which typical colonies on EMB plates and vials giving detection times were confirmed as E. coli, were used to form the calibration curve.

After thawing, commercial pasteurized yolk products generally gave E. coli counts of

<10 to <102 CFU/g, as determined by the cultural plating method.

Figure 5.5 shows the linear correlation between E. coli counts by the cultural plating method (log CFU/g) and DTs as estimated by the MicroFoss system for 39 samples of commercial pasteurized yolk products. DTs ranged from 2h for a sample containing 107

CFU/g to 13h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.94.

169 Figure 5.5 Linear regression relationship between the log-transformed populations of E. coli and the MicroFoss detection times for pasteurized yolk products. The regression is described by the equation log CFU/g = 8.263 – 0.572 (DT), (r = -0.94, n = 39).

Analyses of five pasteurized yolk products giving DTs of 11.4h, 12.0h, 12.1h, 12.2h and

12.6h, which were subsequently confirmed as E. coli by the IMViC test, were equivalent to approximate E. coli populations of 5.52 x 10, 2.51 x 10, 2.20 x 10, 1.93 x

10 and 1.14 x 10 CFU/g, respectively based on the linear equation, log CFU/g = 8.263 –

0.572 (DT), as shown in Figure 5.5.

5.3.2.3 Pasteurized albumen products

Commercial pasteurized albumen products were simultaneously analysed for E. coli counts by the cultural plating method and DTs by the MicroFoss system. Among 79 samples examined, no E. coli was detected in 71 samples by either the cultural method or the MicroFoss method. Positive result in one sample was obtained by the MicroFoss method, but the cultural method failed to detect E. coli in this samples. This positive result was subsequently confirmed as E. coli by the IMViC test. Hence, results of seven

170 samples, for which typical colonies on EMB plates and vials giving detection times were confirmed as E. coli, were used to form the calibration curve.

E. coli was generally not detected in most pasteurized albumen products tested after thawing. For detectable results (1 sample), the counts were generally <10 CFU/g, as determined by the cultural plating method.

Figure 5.6 shows the linear correlation between E. coli counts by the cultural plating method (log CFU/g) and DTs as estimated by the MicroFoss system for 7 samples of commercial pasteurized albumen products. DTs ranged from 13h for a sample containing 103 CFU/g to 20h for a sample with <10 CFU/g, yielding a correlation coefficient of –0.94.

Figure 5.6 Linear regression relationship between the log-transformed populations of E. coli and the MicroFoss detection times for pasteurized albumen products. The regression is described by the equation log CFU/g = 9.964 – 0.482 (DT), (r = -0.94, n =

7).

171 One sample of pasteurized albumen product yielding DT of 19.3h, in which confirmed

E. coli was only detected by the MicroFoss assay, was equivalent to approximate E. coli populations of 5 CFU/g based on linear equation of the calibration curve, log CFU/g =

9.964 – 0.482 (DT), as shown in Figure 5.9.

Analysis of one pasteurized albumen product yielding a DT of 19.3h, which was subsequently confirmed as E. coli by the IMViC test, was equivalent to approximate E. coli populations of 5 CFU/g based on the linear equation, log CFU/g = 9.964 – 0.482

(DT), as shown in Figure 5.6.

5.3.3 Performance indicators

Sensitivity rate, specificity rate, false negative rate and false positive rate of the

MicroFoss assay for E. coli were evaluated by analyses of commercial raw egg samples and pasteurized egg products. Results of E. coli detection in these samples by the cultural and MicroFoss assays used for calculation of performance indicators are shown in Table 5.5.

Table 5.5 Results of E. coli detection in commercial raw egg samples and pasteurized egg products by the cultural and MicroFoss assays Sample n MicroFoss resultsa True positive False negative False positive True negative (N11) (N12) (N21) (N22) Raw whole egg 33 33 0 0 0 Raw yolk 48 41 0 3 0 Raw albumen 86 54 0 1 30 Pasteurized whole egg 116 8 0 2 105 Products Pasteurized yolk products 82 39 0 1 37 Pasteurized albumen 79 7 0 0 71 Products a Results of coliform detection by the MicroFoss method were compared with results from the cultural method.

172 5.3.3.1 Sensitivity and false negative rates

There was no occasion that E. coli was not detected by the MicroFoss method and were present by the cultural method for analyses of 33, 48, 86, 116, 82 and 79 samples of raw whole egg, raw yolk, raw albumen, pasteurized whole egg products, pasteurized yolk products and pasteurized albumen products, respectively, across population ranges of

<1 log CFU/g to 7 log CFU/g, except ranges of <1 log CFU/g to <4 log CFU/g for pasteurized albumen products. Hence, there was no false negative result for E. coli detection in eggs and egg products by the MicroFoss assay in this Chapter, for which its sensitivity rate and false negative rate were 100% and 0%, respectively. This indicated equivalent sensitivity and great accuracy of the MicroFoss assay to the cultural method for E. coli detection in egg and egg products.

5.3.3.2 Specificity and false positive rates

There was no occasion that E. coli was not detected by either the cultural or the

MicroFoss method for analyses of 33 and 48 samples of raw whole egg and raw yolk, respectively. Hence, specificity rate and false positive rate of these samples were not applicable, since the former was 0%.

On the other hand, results of E. coli detections of commercial raw albumen samples and all types of pasteurized egg products in this Chapter, as summarised in Table 5.5, allowed specificity and false positive rates of the MicroFoss assay to be calculated, as shown in Table 5.6.

173 Table 5.6 Specificity and false positive rates of the MicroFoss assay to detect E. coli in commercial raw albumen and all types of pasteurized egg products

Sample True negatives False positives Specificity False positive by both methods by the MicroFoss rate (%) rate (%) (N22) (N21) Raw albumen samples 30 1 96.77 3.23 Pasteurized whole 105 2 98.13 1.87 egg products Pasteurized yolk 37 1 97.37 2.63 products Pasteurized albumen 71 0 100.00 0 products

There were a few occasions where E. coli was not detected by the cultural method, but positive results were obtained from the MicroFoss method and were subsequently confirmed as non-E. coli for analyses of 30, 105 and 37 samples of raw albumen, pasteurized whole egg products and pasteurized yolk products, respectively.

Nonetheless, high specificity and low false positive rates of 96-100% and <5%, respectively of the MicroFoss assay for E. coli were accomplished.

On the other hand, there were occasions among 48, 105, 37 and 71 samples of raw yolk, pasteurized whole egg products, pasteurized yolk products and pasteurized albumen products, where confirmed E. coli in 4 (8.33%), 1 (0.95%), 5 (13.51%) and 1 (1.41%) samples, respectively, were detected by the MicroFoss, but they were not detected by the cultural method. Hence, the MicroFoss was likely to give more accurate, reliable result than the cultural method.

174 5.3.4 Repeatability of the MicroFoss assay and significant differences between

the cultural and MicroFoss E. coli counts

Repeatability of the MicroFoss assay for E. coli and significant differences (P<0.05) between means of E. coli counts from both methods are shown in Table 5.7.

Table 5.7 Repeatability of the MicroFoss assays and significant differences between the cultural and MicroFoss E. coli counts in commercial raw egg samples

Sample n EMB counts MF counts Mean + CR t P (log CFU/g + SD)a (log CFU/g + SD)a, b (MF method) Raw whole egg 6 3.79 + 0.15 3.48 + 0.13 3.23 to 3.73 3.913 0.001c 6 4.34 + 0.11 3.63 + 0.14 3.36 to 3.90 9.981 <0.0001c Raw yolk 6 2.33 + 0.30 1.67 + 0.30 1.08 to 2.26 3.849 0.002c 5 3.09 + 0.08 3.15 + 0.21 2.74 to 3.56 -0.566 0.296 7 7.09 + 0.61 6.94 + 0.49 5.98 to 7.90 0.509 0.31 Raw albumen 5 1.99 + 0.51 1.56 + 0.27 1.03 to 2.09 1.669 0.067 a E. coli counts (log CFU/g) obtained from the cultural and MicroFoss assays are presented as mean + SD values. b The MicroFoss E. coli counts were mathematically transformed using DTs, which were based on linear equations describing the calibration curves, log CFU/g = 9.671 – 1.002 (DT), log CFU/g = 9.541 – 0.968 (DT) and log CFU/g = 9.155 – 0.829 (DT) for commercial raw whole egg, raw yolk and raw albumen samples, as shown in Figures 5.1, 5.2 and 5.3, respectively. c The cultural and MicroFoss E. coli counts in commercial raw egg samples were significantly different (P<0.05).

SD values of E. coli counts in commercial raw egg samples estimated by the MicroFoss assay were generally equal or smaller than those of E. coli counts obtained from the cultural method, particularly at ≤102 CFU/g in raw albumen.

Estimation of E. coli populations by the MicroFoss assay was repeatable at 95% confidence interval, as results of all replicates of each individual sample in all types of raw egg samples were within mean + correlation of repeatability, as shown in Table 5.7.

Significant differences (P <0.05) between the cultural and MicroFoss E. coli counts were observed in samples of raw whole egg and raw yolk examined in this Chapter.

However, E. coli counts obtained from both methods at low populations (<102 CFU/g) in sample of raw albumen was not significantly different.

175 5.3.5 Significant differences of proportions of confirmed E. coli between the

cultural and MicroFoss assays

Since false negative and false positive results of E. coli detection by the MicroFoss assay in commercial raw whole egg samples and pasteurized albumen products were not obtained, calculations of a chi square value (χ2) of these samples were not applicable.

On the other hand, results of E. coli detection of samples of raw yolk, raw albumen, pasteurized whole egg and pasteurized yolk products by both methods in this Chapter allowed chi square values to be calculated, as shown in Table 5.8.

Table 5.8 Test for significant differences of proportions of confirmed E. coli between the cultural and MicroFoss assays

Sample False negatives False positives χ2 (a) by the MicroFoss by the MicroFoss (N12) (N21) Raw yolk samples 0 3 1.33 Raw albumen samples 0 1 0 Pasteurized whole egg products 0 2 0.5 Pasteurized yolk products 0 1 0

As chi square values of all commercial egg samples analysed were <3.84, this indicated that the proportions of confirmed E. coli detected by the cultural and MicroFoss assays were not statistically different (P>0.05).

5.4 DISCUSSION

The MicroFoss system has now been applied as an alternative, rapid method to estimate

E. coli populations in a range of foods, including chicken carcasses (Russell, 2000), ground beef (Russell, 2000; Odumeru and Belvedere 2002), ground pork (Russell,

2001). Linear regression relationships between the MicroFoss DTs and the MPN E. coli

176 counts (log CFU/g) results have been established in these commodities, with correlation coefficients from and in excess of -0.90. These calibration curves then become the basis for subsequent routine use of the MicroFoss for assessing the sanitation and efficacy of processing as well as indicating faecal contamination, possible presence of pathogens and safety of foods for human consumption. The MicroFoss assays could be accomplished in less than 24h, whereas approximately four to five days were needed to obtain results by the MPN procedure.

Calibration curves to estimate E. coli populations in unpasteurized liquid eggs and pasteurized egg products were separately established due to difference of the nature of each product. Egg homogenate derived from shell eggs were used as an egg matrix for detection of E. coli in unpasteurized liquid eggs. Since E. coli were not detected or were naturally present in low numbers in contents of shell eggs, shell eggs were therefore inoculated with various populations of E. coli to achieve a range of populations of <10 to 107 CFU/g. The cultural spread plate method using the EMB plates was selected to detect and estimate E. coli populations in this study, since the MPN technique was unsuitable to be used to enumerate organisms present in food sample (Craven et al.,

2003). Results of this study indicated that the linear relationship of the cultural E. coli counts and the MicroFoss DTs of all types of egg homogenates inoculated with different populations of E. coli were well correlative, with correlation coefficients in excess of –

0.90. However, the effectiveness of these calibration curves to estimate the MicroFoss

E. coli populations in unpasteurized liquid eggs was tested with commercial raw egg samples.

E. coli counts in excess of 80% of 33, 48 and 86 samples of commercial raw whole egg, raw yolk and raw albumen tested were comparable. In all cases, the cultural E. coli counts of all types of commercial raw egg samples were higher than E. coli counts

177 estimated by the MicroFoss method. Most E. coli-positive vials, except for three and one false positive results from raw yolk (Table 5.3) and raw albumen (Table 5.4), respectively, were confirmed as E. coli by the IMViC test. On the other hand, a number of typical colonies on EMB plates were confirmed as non E. coli by the IMViC test. All colonies on EMB plates were counted (Section 5.2.5) and typical colonies were selected for further confirmation. Corrected cultural counts were then used to compare with E. coli counts estimated from the MicroFoss method. The cultural E. coli counts obtained were, in effect, only estimates numbers, which could contribute to the difference in E. coli counts between the cultural and MicroFoss methods. Hence, the MicroFoss could possibly be used to estimate E. coli populations in unpasteurized liquid eggs.

E. coli populations initially presented in unpasteurized liquid eggs are usually declined or eliminated by pasteurization and completely destroyed by subsequent freezing (Shafi et al., 1970). Many samples of the pasteurized egg products showed no detectable E. coli by either the cultural or MicroFoss assay.

However, E. coli was detected in all types of pasteurized egg products tested in this study, as shown in Sections 5.3.2.1, 5.3.2.2, and 5.3.2.3. The actual cause of the presence of E. coli in commercial pasteurized egg products used in this study was unknown. Possible causes could be the inadequacy of pasteurization or post- pasteurization contamination. E. coli that survive pasteurization could resuscitate once the optimum conditions are introduced. For example, storage temperature fluctuation during transportation and along the distribution chain of these pasteurized egg products.

Although many samples of the pasteurized egg products commercially available showed no detectable E. coli by either the cultural or MicroFoss assay, certain samples yielded confirmed cultural E. coli results and positive-vials confirmed as coliforms could be used to generate calibration curves for each type of pasteurized egg products. In this

178 study, numbers of E. coli presenting in these commercial pasteurized egg products could be promoted by temperature abuse at 25°C for 1-3 days. Generation of a calibration curve using naturally contaminated pasteurized egg products in this study should give an accurate estimation of E. coli populations in pasteurized egg products.

Since only a limited number of samples were available for validation analyses, thereby diminishing the strength of the conclusions. Nevertheless, the results of this study indicated good correlation between E. coli counts obtained by the cultural and the

MicroFoss methods for all types of commercial pasteurized egg products with correlation coefficients in excess of –0.90.

For most commercial raw egg samples and pasteurized egg products tested, except for raw whole egg, there were samples for which confirmed E. coli were only detected by the MicroFoss method and these counts were generally low, ranging from <10 to <102

CFU/g. This superior detection could be because the greater sample size (5 mL) used in the MicroFoss assay compared to 0.1 mL used in the cultural spread plate method.

These findings showed that the MicroFoss assay could be used for detection of E. coli, as E. coli in foods are naturally present in low numbers.

However, some of these positive results detected solely by the MicroFoss assay were confirmed as false positive results. Edmiston and Russell (2000) evaluated the specificity of the E. coli conductance method in comparison between detection of E. coli and non-E. coli organisms that are closely related to E. coli in poultry products using coliform medium (CM) supplemented with 2% dextrose (CMD), which was the same as the MicroFoss coliform medium. This was to assess any interference with conductance readings from these non-E. coli organisms by multiplying and producing similar conductance readings to that of E. coli. Citrobacter freundii did not proliferate to a considerable degree in the CMD medium at 44°C to yield similar conductance

179 reading to that of E. coli. As Shigella sonnei grew at a more rapid rate than E. coli in

CMD, high levels of Shigella sonnei that could reach the detection threshold (106

CFU/mL) may interfere with estimation of E. coli populations using conductance method. Although Salmonella Enteritidis grew similarly to E. coli, an initial level of 106

CFU/mL would be essential for Salmonella to interfere with E. coli estimation by conductance method. As the MicroFoss E. coli medium used in this study was double- strength CMD (Russell, 2000), similar to that used in Edmiston and Russell (2000), false positive results obtained from the MicroFoss method could be from false detection of non-E. coli organisms that are closely related to E. coli, such as Shigella (Brenner,

1984). Specificity of the MicroFoss assay for E. coli had to be evaluated and this will be discussed in the subsequent Section.

In Australia, E. coli should be absent in properly pasteurized egg products. Hence, pasteurized whole egg, pasteurized yolk and pasteurized albumen products that gave DT values earlier than approximately 10.5, 14.5 and 20.7h, respectively were would be rejected due to unsatisfactory E. coli quality, suggesting improper pasteurization.

The above findings indicated high correlations between the cultural method and the

MicroFoss assay in eggs and egg products and suggested that the MicroFoss assay could be effective for estimation of E. coli in eggs and egg products. Further evaluation of the

MicroFoss in regards to performance indicators and repeatability are discussed in the following Sections.

Accuracy

False negative rate of the MicroFoss assay for E. coli obtained was 0%, based on 33, 48 and 86 samples of commercial raw whole egg, raw yolk and raw albumen, respectively, and 116, 82 and 79 samples of pasteurized whole egg, pasteurized yolk and pasteurized

180 albumen products, respectively tested. False positive rates of the MicroFoss assay for E. coli obtained were in the range of 0%-3.50%, based on 86 samples of commercial raw albumen samples, and 116, 82 and 79 samples of pasteurized whole egg, pasteurized yolk and pasteurized albumen products, respectively tested in this Chapter. Such low false negative and false positive rates indicated the high accuracy of the MicroFoss assay for E. coli detection.

Sensitivity

Sensitivity rate also reflects accuracy of the MicroFoss assay (de Boer and Beumer,

1999). Sensitivity rate of the MicroFoss assay for E. coli obtained was 100%, based on numbers of commercial raw egg samples and pasteurized egg products tested in this

Chapter. This finding indicated high accuracy of the MicroFoss assay for E. coli.

Concurrently, limit of detection also exhibits sensitivity of the MicroFoss assay (Cox and Fleet, 2003). The MicroFoss assay detected as low as <10 CFU/g of confirmed E. coli in commercial raw albumen samples and all types of pasteurized egg products tested. This finding indicated high sensitivity and accuracy of the MicroFoss assay for

E. coli.

Specificity

Specificity rate also indicated the accuracy of the MicroFoss assay (de Boer and

Beumer, 1999). Specificity rates of the MicroFoss assay for E. coli obtained were in the range of 96%-100%, based on numbers of commercial raw albumen samples and all types of pasteurized egg products tested in this Chapter. Results of high specificity rates indicated high accuracy of the MicroFoss assay for E. coli.

181 Repeatability

SD values of E. coli counts in commercial raw egg samples estimated by the MicroFoss assay were generally equal or smaller than those of E. coli counts obtained from the cultural method, particularly at ≤102 CFU/g in raw albumen. This finding was useful since E. coli is naturally present in low numbers. This would benefit E. coli estimations in eggs and other food products using the MicroFoss method, as E. coli are naturally present in low numbers. Findings in this study indicated that the MicroFoss E. coli analysis could be likely to be repeatable on an ongoing, routine basis.

Speed

E. coli estimations in unpasteurized liquid eggs and pasteurized egg products can be completed in 2h for samples containing 107 CFU/g and in 13h for samples with <10

CFU/g. This showed significant time advantage for E. coli detection by the MicroFoss, compared to 24h to obtain presumptive results on EMB plates and an additional 24-

120h for confirmation with the IMViC test. This suggested that the use of the

MicroFoss assay as an alternative rapid method for detection and enumeration of E. coli in unpasteurized liquid eggs and pasteurized egg products.

Simplicity

Sample preparation and operation procedure of detection of target organisms by the

MicroFoss assay were simple and easy to follow. Liquid samples can be directly added into the ready-to-use MicroFoss vials, whereas solid or semi-solid samples only require

10-1 dilution prior to inoculating to the vials.

182 Although proportions of confirmed E. coli detected by both of the cultural and

MicroFoss assays were not statistically different (Table 5.8), significant differences

(P<0.05) between the cultural and MicroFoss E. coli counts in samples of raw whole egg and raw yolk were observed (Table 5.7). However, E. coli counts at low levels

(<102 CFU/g) in raw albumen samples obtained from both methods were not significantly different. This finding was important, since E. coli in foods are naturally present in low numbers.

This study suggested high correlations between the cultural E. coli results and the

MicroFoss DTs of analyses of unpasteurized liquid eggs and pasteurized egg products, with correlations of the two measurements similar to or higher than those reported in previous studies (Russell, 2000; Russell, 2001; Odumeru and Belvedere 2002).

Moreover, the MicroFoss method was shown to have relevant or greater accuracy, sensitivity, specificity, and repeatability to the cultural method; and provided a significant advantage of time require to obtain results. Therefore, the overall results in this Chapter suggested that the MicroFoss system could be used as an alternative method to rapidly detect and estimate E. coli populations in unpasteurized liquid eggs and pasteurized egg products. The ability to detect E. coli results in a timely fashion can assist egg manufacturers in more effective control of the microbial quality of eggs.

183 CHAPTER 6

Development of a Salmonella medium for detection of salmonellae by the MicroFoss system

6.1 INTRODUCTION

Salmonella is a major human pathogen that is responsible for food poisoning outbreaks from eggs and egg-associated foods throughout the world (Humphrey, 1994; Kobayashi et al., 1997; Cox, 2000; Sparks, 2000; ICMSF, 2005). The most common food vehicles of human salmonellosis are eggs, poultry, meat and their products (Brenner, 1992; Jay,

2000). Ingestion of Salmonella cells in the order of 106–108 CFU is generally necessary to cause salmonellosis. However, epidemiological evidence from a number of outbreaks has demonstrated that the infectious dose may be as little as a few cells (Cox, 2000).

Although standard cultural methods are commonly used to detect salmonellae in a variety of foods, these methods are tedious (Flowers, 1985; Adams and Moss, 2000;

Andrews et al., 2001; Jay et al., 2003) and labour-intensive (Andrews et al., 2001).

Development of rapid detection methods has, therefore, gained interest as they offer advantages in speed, timely corrective action for occurrence of contamination, cost effectiveness, and early release of Salmonella free perishable foods such as egg products (Flowers, 1985; Jay et al., 2003). Various techniques including rapid assays for detection of salmonellae in foods (Feng, 1992; Blackburn, 1993; Andrews et al.,

2001; Jay et al., 2003) and poultry feeds (Ricke et al., 1998) have been reviewed. The

MicroFoss assay could also be used for rapid detection of salmonellae in foods (Tan and

Shelef, 1999; Peng and Shelef, 2001).

184 Since detection of salmonellae by the MicroFoss assay has not been commercially available, this study aimed to develop and optimise a selective medium that could be used in the MicroFoss system to detect salmonellae. The creation of this MicroFoss

Salmonella medium was based on selecting distinctive biochemical characteristics of salmonellae, which would differentiate salmonellae from the other related genera in the family Enterobacteriaceae, such as Citrobacter freundii, Escherichia spp., Hafnia alvei and Proteus mirabilis (Brenner, 1984; le Minor, 1992). Systematic investigation was done to formulate the medium that gave the best detection and differentiation of salmonellae from the other potential non-salmonellae. Different medium formulations were tested with a range of Salmonella serovars, representative species of the other related genera in the family Enterobacteriaceae and the other prevalent organisms in eggs and egg products.

6.2 MATERIALS AND METHODS

6.2.1 Microbiological media, chemicals and reagents

All microbiological media, chemicals, and reagents used in this Chapter are listed in

Table 6.1. All chemicals used are of general laboratory grade or greater quality.

185 Table 6.1 Microbiological media, chemicals and reagents used in this Chapter

Media, Chemicals & Reagents Manufacturers Ammonium iron (III) citrate Sigma Chemical CO., St. Louis, USA Amyl alcohol Sigma-Aldrich, Australia Bile bovine Sigma-Aldrich Inc., St. Louis, USA Bismuth sulphite agar (modified) Oxoid , Adelaide, Australia Brilliant green BDH Chemicals Ltd, Poole, UK Bromocresol purple Ajax Chemicals, Sydney, Australia Bromophenol red, Riedel-de Haen, Sigma-Aldrich Laborchemikalien GmbH, Seelze, Germany Bromothymol blue sodium salt, ACS agent, Sigma Chemical CO. Bromoxylenol blue, Riedel-de Haen, Sigma-Aldrich Chemie GmbH, Steinheim, Germany Buffered peptone water Oxoid Chlorophenol red, indicator grade Sigma-Aldrich Pty Ltd, Castle Hill, Australia CLED medium Oxoid Creatine Ajax Chemicals Ltd., Sydney, Australia Davis powdered agar, grade J3 Davis Gelatine (Australia) Co., 4-(Dimethylamino)benzaldehyde, ACS reagent 99%, Sigma-Aldrich, Inc. di-Potassium hydrogen phosphate, May & Baker Ltd., Dagenham, UK (Potassium phosphate dibasic) D(-)Sorbitol Ajax Chemicals, Auburn, Australia Glucose Jaegar Chemicals, Brookvale, Australia Glycerol (Glycerine) Chem-Supply Pty Ltd, Gillman, Australia Hydrochloric acid 36% Ajax Finechem, Seven Hills, Australia Iodine BDH Chemicals Australia Pty Ltd, Australia Lab'Lemco' powder Oxoid L-Lysine (L-2, 6-Diaminohexanoic acid) Sigma Chemical CO. monohydrochloride, minimum 98% (TLC) Magnesium chloride Ajax Finechem Malachite green oxalate Merck KgaA, Darmstadt, Germany Mannitol Ajax Chemicals Ltd. Manntiol selenite cystine broth base Amyl Media Pty Ltd, Dandenong, Australia Methyl red BDH Chemicals Ltd

186 Table 6.1 (continued) Microbiological media, chemicals and reagents used in this

Chapter

Media, Chemicals & Reagents Manufacturers MRVP medium (Clarks and Lubs medium) Oxoid Muller-Kauffmann-tetrathionate (MKTTn) broth base Oxoid 1-Naphthol (-Naphthol) BDH Chemicals Ltd Niaproof 4, type 4, 27% Sigma-Aldrich, Inc. Novobiocin sodium salt (minimum 90%-HPLC) Sigma-Aldrich Chemie Gmbh, Steinheim, Germany Peptone, bacteriological Oxoid Phenol red Merck, Darmstadt, Germany Potassium dihydrogen phosohate (Potassium biphosphate), May & Baker Ltd.

KH2PO4 Potassium hydroxide pellets May & Baker Australia Pty Ltd, West Footscray, Australia Potassium iodide (KI) Ajax Chemicals Pty Ltd, Auburn, Australia Propylene glocol (propane-1, 2-diol) Ajax Chemicals Proteose peptone Amyl Media Pty Ltd Rappaport-Vassiliadis (RV) enrichment broth Oxoid Salicin (2-[Hydroxymethyl] phenyl -D-glucopyranoside) Sigma Chemical CO. Simmons citrate agar Oxoid Sodium biselenite (sodium hydrogen selenite) Oxoid Sodium chloride Chem-Supply Pty Ltd Sodium thiosulphate Ajax Chemicals Sulphamandelate supplement (vials) Oxoid 2, 3, 5-Triphenyl-2H-tetrazolium chloride, 98% Aldrich Chemical Company, Inc., Milwaukee, USA Triple sugar iron agar Oxoid Tryptone Oxoid Urea Fisons Scientific Equipment, Loughborough, UK Urea agar base Oxoid Xylose (D(+)-Xylose, Sigma grade, 99-100% Sigma Chemical CO. Xylose Lysine Deoxycholate (XLD) medium Oxoid Yeast extract Oxoid Zinc granule Sigma Chemical CO.

187 6.2.2 Preparation of reference cultures

Bacterial reference cultures of non-salmonellae and salmonellae used in these evaluation studies and their sources are shown in Tables 6.2 and 6.3, respectively. Stock cultures were maintained on slants of TSA at 5°C and sub-cultured monthly. Before use in experiments, all cultures were tested for purity by streaking onto fresh plates of TSA.

For use in experiments, reference cultures were prepared freshly by transferring a single colony of biomass from a plate of TSA into 10 mL of TSB. Most of the TSB cultures were incubated at 37°C for 24h. The exceptional cultures that were incubated at 30°C for 24-48h are Acinetobacter baumannii subsp. anitratus, Acinetobacter calcoaceticus/baumannii, Acinetobacter calcoaceticus/haemolyticus, Aeromonas hydrophila, Flavobacterium sp., Listeria monocytogenes L4, Micrococcus luteus,

Proteus vulgaris, Pseudomonas fluorescens, Pseudomonas putida and

Stenotrophomonas maltophilia, whilst Serratia marcescens was incubated at 25°C for

48h. The broth cultures were serially diluted 1:10 using 9 mL of 0.1% peptone water to give suspensions of different populations for inoculation to testing media. In some cases, cultures of individual bacteria were prepared as just described and mixed to give the cell suspension for inoculation.

188 Table 6.2 Reference cultures of non-salmonellae used in this Chapter

Reference cultures Source Acinetobacter baumannii subsp. anitratus School of Biotechnology and Biomolecular Sciences, UNSWa (UNSW 050200, NCTCb 7844, ATCCc 15308) Acinetobacter calcoaceticus/baumannii School of Chemical Sciences and Engineering, UNSW Acinetobacter calcoaceticus/haemolyticus School of Chemical Sciences and Engineering, UNSW Aeromonas hydrophila School of Chemical Sciences and Engineering, UNSW Bacillus cereus TECRAd (BC TICC 1996) Bacillus subtilis TECRA (BS TICC 11) School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Citrobacter freundii 047800, NCTC 9750, ATCC 8090) Citrobacter freundii IMVS culture collection (IMVS 1263) School of Biotechnology and Biomolecular Sciences, UNSW Enterobacter aerogenes (UNSW 045800, NCTC 10006, ATCC 13048) School of Biotechnology and Biomolecular Sciences, UNSW Enterobacter cloacae (UNSW 038000, ATCC 13047) School of Chemical Sciences and Engineering, UNSW Escherichia coli (UNSW 048200, ATCC 11775) Flavobacterium sp. School of Chemical Sciences and Engineering, UNSW School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Klebsiella pneumoniae 004600, ATCC 12657) Listeria monocytogenes L4 1771 serotype 4c, TECRA, Sydney (ATCC 19116) School of Biotechnology and Biomolecular Sciences, UNSW Micrococcus luteus (UNSW 011000, NCTC 2665, ATCC 4698) School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Proteus mirabilis 059300) School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Proteus vulgaris 046100, NCTC 4175, ATCC 13315) School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Providencia stuartii 058600) School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Pseudomonas aeruginosa 029100, NCTC 6749, ATCC 19582) Pseudomonas fluorescens School of Biotechnology and Biomolecular Sciences, UNSW (NCTC 10038, ATCC 13525 NCIMBe B9406, ACMf 441) Pseudomonas putida School of Chemical Sciences and Engineering, UNSW School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Serratia marcescens 052001, NCTC1377, ATCC 274, AGALg 165) School of Biotechnology and Biomolecular Sciences, UNSW (UNSW Shigella sonnei 004900) Staphylococcus aureus School of Chemical Sciences and Engineering, UNSW Stenotrophomonas maltophilia School of Chemical Sciences and Engineering, UNSW Yersinia enterocolitica Biotype 1a, O non-typable IMVSh (IMVS2861) a UNSW The University of New South Wales, Sydney, Australia. b NCTC National Collection of Type Cultures, 61 Colindale Rd., London, NW9 5HT, U.K. c ATCC American Type Culture Collection, 10801 University Boulevard, Manassas, VA 20110-2209. d TECRA Tecra International, Frenchs Forest, Sydney, Australia. e NCIMB National Collections of Industrial Food and Marine Bacteria, 23 St. Machar Drive, Aberdeen, AB24 3RY, Scotland, U.K. f ACM Australia Collection of Microorganisms, Department of Microbiology, University of Queensland, Brisbane, Queensland, 4072, Australia. g AGAL Australian Government Analytical Laboratories. h IMVS Institute of Medical and Veterinary Science, PO Box 14, Rundle Mall, South Australia 5000, Australia.

189 Table 6.3 Reference cultures of salmonellae used in this Chapter

Reference cultures Source Salmonella Agona 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Amsterdam School of Chemical Sciences and Engineering, UNSW Salmonella Anatum 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Blockley School of Chemical Sciences and Engineering, UNSW Salmonella Bredeney School of Chemical Sciences and Engineering, UNSW Salmonella Derby 2 School of Chemical Sciences and Engineering, UNSW Salmonella Dublin School of Chemical Sciences and Engineering, UNSW School of Biotechnology and Biomolecular Sciences, Salmonella Enteritidis (SE) UNSW (UNSW 031900) SE FS1 School of Chemical Sciences and Engineering, UNSW SE 1, 2, 4, 14, 15, 16, 17, 20, 25, 31, 33, 53, 59, 67, 69, 70, 71, 73, C6B, G7/All, School of Chemical Sciences and Engineering, UNSW RPTs(I), rSE6 Salmonella Hadar 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Infantis 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Mbandaka School of Chemical Sciences and Engineering, UNSW Salmonella Montevideo 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Muenster School of Chemical Sciences and Engineering, UNSW Salmonella Newport 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Oranienburg School of Chemical Sciences and Engineering, UNSW Salmonella Saint Paul 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Schwarzengrund 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Senftenberg 1 & 2 School of Chemical Sciences and Engineering, UNSW Salmonella Sofia 1, 2, 76, 89, sswt School of Chemical Sciences and Engineering, UNSW Salmonella Typhimirium (ST) School of Chemical Sciences and Engineering, UNSW ST 1 School of Chemical Sciences and Engineering, UNSW Salmonella Choleraesuis IMVS (IMVS1259) Salmonella Choleraesuis var Kunzendorf IMVS (IMVS1269) Salmonella Choleraesuis var Australia IMVS (IMVS1594) Salmonella subsp 3b ser 61:i:z53 IMVS (IMVS1679) Salmonella Derby IMVS (IMVS0878) Salmonella subsp 2 ser 1,9,12:g,z62:- IMVS SRL04-803419

190 Details of biochemical characteristics of atypical biotypes of Salmonella supplied by the

IMVS, which were used to determine the effectiveness of their detection, are shown in

Table 6.4.

Table 6.4 Biochemical characteristics of atypical biotypes of Salmonella supplied by the IMVS

Atypical biotypes IMVS references Serotyping Biochemical characteristics of Salmonella Salmonella Derby IMVS 0878 4,12:f,g:- Lactose +ve, ONPG +ve, L’tartrate +ve, Dulcitol +ve, Mucate +ve Salmonella Choleraesuis IMVS 1259/92 6,7:c:1,5 H2S –ve, Citrate –ve, Malonate –ve, Xylose –ve, L’tartrate +ve, Dulcitol –ve, Mucate -ve Salmonella Choleraesuis IMVS 1269 6,7:c:1,5 H2S +ve, L’tartrate +ve, var Kunzendorf Dulcitol –ve, Mucate -ve Salmonella Choleraesuis IMVS 1594/84 6,7:c:1,5 H2S weak +ve by conventional method var Australia & -ve by Gram Negative Bacilli 24E-Oxidase Negative test kit, L’tartrate +ve, Dulcitol –ve, Mucate +ve Salmonella subsp 3b ser IMVS 1679 61:i:z53 ONPG +ve, Malonate +ve, 61:i:z53 Lactose –ve, H2S -ve Salmonella subsp 2 ser SRL 02-803419 1,9,12:g,z62:- ONPG +ve, Indole +ve, 1,9,12:g,z62:- Citrate +ve, Malonate –ve, Dulcitol +ve, Mucate +ve

+ve = Positive reaction. -ve = Negative reaction.

H2S = Hydrogen sulphide. ONPG = o-Nitrophenyl β-galactopyranoside.

191 6.2.3 Preparation of media

6.2.3.1 Media evaluated in test tubes or screw-cap test tube experiments

Media created in this Chapter were sterilised by either autoclaving at 121°C for 15 min, or by boiling for those containing chemicals that were sensitive to autoclave temperatures. Media that were sterilised by autoclaving were dissolved in normal distilled water, whereas sterilised distilled water was used to dissolve media that were boiled. Media were thoroughly mixed to ensure that all ingredients were completely dissolved and pH was adjusted prior to sterilisation with either 0.1 N of NaOH or 0.1 N of HCl. pH was calibrated and measured accordingly to the actual ambient temperature at the time of preparation, approximately 25°C.

For media that were sterilised by autoclaving, an aliquot (10 mL) of medium was dispensed to test tubes or screw-cap test tubes prior to autoclaving. For media that were sterilised by boiling, aliquots (10 mL) were cooled to room temperature and aseptically dispensed to sterilised test tubes or screw-cap test tubes. Glassware and equipment associated with media that were sterilised by boiling were pre-sterilised by autoclaving at 121°C for 15 min.

6.2.3.2 Media tested in the MicroFoss system

The MicroFoss vials consist of two compartments. The lower section of the vial contains medium as a semi-solid agar plug. The section above this plug contains liquid medium. Medium for the agar plug was prepared by mixing liquid media (Section

6.2.3.1) with 1% agar (w/v) and boiling to completely dissolve and sterilise. Molten agar media (approximately 0.3 mL) was then aseptically added into sterile vials

(Neogen Corporation, Lansing, USA) to precisely fill up the agar plug zone without allowing any bubbles to form on the surface and in the agar plug. The agar plug of medium was allowed to solidify prior to adding 5 mL of liquid medium to the vial.

192 6.2.4 Testing of media

Various medium formulations were initially evaluated in test tubes and screw-cap test tubes. Thereafter, analyses of selected formulae were performed in the MicroFoss system.

6.2.4.1 Test tube evaluations

Suspensions (0.1ml) of bacterial cells of individual species or mixtures of species

(Section 6.2.2) were inoculated into 10 mL of liquid media in tubes (Section 6.2.3.1).

The inoculated media in test tubes were incubated in an incubator, whilst those in screw-cap test tubes were incubated in the higher temperature water baths (HTB-150-D,

Thermoline Scientific Equipment Pty. Ltd., Smithfield, Australia) set to obtain the required incubation temperatures. Reactions were recorded at specified intervals during and/or after completion of the incubation period.

6.2.4.2 Evaluations in MicroFoss vials

Suspensions (0.05mL) of bacterial cells of individual species or mixtures of species were added to 5 mL of liquid media in the MicroFoss vials. The vials were mixed by gently inversion 10 times, then inserted into the MicroFoss instrument and monitored during their incubation.

Various system algorithms, which are detection threshold, skip factor, and shut-eye period, were trialled and adjusted to obtain the optimum combinations of these three parameters for the best-fit MicroFoss curves. The trialled values for the detection threshold were 8, 10, and 12, whilst the shuteye values trialled were 25 and 35. The skip factor value was fixed at 1. Cutoff and caution values of 8 were constant. The inoculated and control vials were incubated at 42°C in the MicroFoss instrument for

48h.

193 6.3 RESULTS

The principles for creation of a novel Salmonella medium involved selecting and coupling biochemical characteristics of Salmonella that were distinctive from those of other related genera in the family, Enterobacteriaceae. Trial experiments based on different principles were performed to achieve the appropriate medium formulation.

6.3.1 Evaluation of xylose lysine deoxycholate (XLD) and lysine mannitol glycerol (LMG) media as potential candidates for use in the MicroFoss assay

The principles of formulating selective media in this Section were based on L-lysine decarboxylation and hydrogen sulfide (H2S) production of typical Salmonella using combined features of XLD and lysine-mannitol-glycerol (LMG) media. XLD is widely used as a selective differential medium for the isolation and detection of Salmonella

(Fagerberg and Avens, 1976; Jay et al., 2003; AS 5013.10-2004). It consists of yeast extract as a nutrient base; xylose, sucrose and lactose as carbon sources; sodium desoxycholate as a selective agent; and L-lysine hydrochloride as a differential agent.

On this medium, typical Salmonella are able to utilise only xylose and subsequently decarboxylate lysine to produce an amine that causes a reversion to alkaline pH or red colour of the medium, which is indicated by phenol red indicator. As an additional differential test, the medium contains H2S substrates, sodium thiosulphate and ammonium ferric citrate, that Salmonella can utilise for H2S production, which results in blackening of the medium. Lysine-positive coliforms are unable to exhibit lysine decarboxylation due to excessive acid produced from fermentation of all three sugars in this medium (Taylor, 1965; Fagerberg and Avens, 1976). Cox (1993) introduced a new medium, LMG agar, which featured XLD and mannitol-lysine-crystal violet-brilliant green (MLCB) media, with an addition of glycerol for differentiation of Salmonella and

194 Citrobacter spp. Salmonella fermented mannitol in this medium and subsequently decarboxylated lysine to provide an alkaline condition for H2S production. pH of the medium inoculated with Citrobacter freundii remained acid, as a result utilisation of both of mannitol and glycerol, and consequently no blackening was observed after prolonged incubation (up to seven days). LMG medium proved to be a superior medium to differentiate salmonellae from non-salmonellae, particularly H2S-producing non- salmonellae as Citrobacter spp. and Proteus spp. This medium also promoted detection of atypical biotypes of Salmonella, such as lactose- or sucrose-positive strains, and

Salmonella Typhi.

Evaluation of lysine decarboxylation and hydrogen sulfide production

A medium formulated in this Experiment consisted of 3 g/L of proteose peptone and 5 g/L of yeast extract as nutrient bases; 5 g/L of mannitol and 5 g/L of glycerol as carbon sources; 1 g/L of sodium desoxycholate as a selective agent; and 5 g/L of L-lysine, 4 g/L of sodium thiosulphate and 1 g/L of ammonium ferric citrate as differential agents.

Although xylose is a more selective substrate for fermentation by most salmonellae, mannitol was selected to allow fermentation by those strains that do not ferment xylose

(Brenner, 1984). pH dye indicator was not used in this occasion, as H2S production was exhibited by blackening of the colourless uninoculated medium, which would be indicated by a decline of transmittance monitored in the MicroFoss system. Test tubes containing 10 mL of this medium were inoculated with a selection of non-salmonellae of the family Enterobacteriaceae and representative serotypes of Salmonella, S.

Enteritidis and S. Typhimurium, as listed in Tables 6.2 and 6.3, respectively and incubated at 37°C for 48h.

195 Results showed that Citrobacter freundii and Proteus mirabilis produced H2S within

24h, whereas both serotypes of Salmonella did not produce H2S within 48h. Slow lactose-fermenting Citrobacter strains and Proteus mirabilis was able to produce alkaline H2S-positive colonies on medium with thiosulfate substrate (Hoben et al.,

1973; Restaino et al., 1977; Komatsu and Restaino, 1981; Moats, 1981). This indicated that this medium was not sufficiently selective in liquid form. As a result, selective agents were added to this medium to enhance selectivity of detection. Key selective agents used followed those of Rappaport-Vassiliadis (RV) broth, which included

MgCl2.6H2O and malachite green oxalate. Rappaport et al. (1956) introduced an enrichment broth containing 4% of MgCl2 and 0.012% of malachite green to inhibit growth of coliforms, yet allowed growth of salmonellae. Rappaport and Konforti (1959) reported that incorporation of MgCl2.6H2O (28.6 g/L) in RV broth neutralised the toxic effect of malachite green oxalate to allow growth of salmonellae. Since malachite green oxalate was also a dye, colour of the uninoculated medium was then intense blue.

Hydrogen sulfide production was indicated by blackening of the medium.

Test tubes containing 10 mL of this medium with MgCl2.6H2O and malachite green oxalate were analysed with the same set of non-salmonellae and representative serotypes of Salmonella, as analysed with the medium without MgCl2.6H2O and malachite green oxalate, and incubated at 37°C for 48h. Results showed that

Citrobacter freundii was inhibited in this medium, whereas H2S produced by Proteus mirabilis was observed within 24h. Both S. Enteritidis and S. Typhimurium were inhibited by the selectivity of this medium.

Proteus mirabilis did not only persist in giving a false positive result in this medium containing additional selective agents, but Salmonella was also inhibited by its increased selectivity. Hence, detection of Salmonella in a medium based on these two

196 principles was discontinued. The following Experiment evaluated detection and differentiation of Salmonella using a medium based on lysine decarboxylation.

Evaluation of lysine decarboxylation

A medium formulated in this Experiment consisted of 3 g/L of proteose peptone and 5 g/L of yeast extract as nutrient bases; 5 g/L each of mannitol and glycerol as carbon sources with additional 7.5 g/L of lactose and/or 7.5 g/L of sucrose to enhance excessive fermentation by organisms that are positive for lysine decarboxylation; 10 g/L of L-lysine as a differential agent; and 0.08 g/L of phenol red as a pH indicator. Acid produced from fermentation of mannitol by Salmonella would result in a rise in transmittance detected by the MicroFoss system, which would subsequently decline due to pH reversal to alkaline from lysine decarboxylation. However, pH reversal to alkaline could also result from utilisation of proteinaceous substrates (proteose peptone and yeast extract) in this medium by both of Salmonella and non-salmonellae. Thus, selectivity of detection could be enhanced by the addition of combined selective agents that were reported the effectiveness to suppress non-salmonellae without any adverse effect on Salmonella growth by several authors, which were MgCl2.6H2O (Rappaport et al., 1956; Busse, 1995), brilliant green (Jones et al., 1984; Arroyo and Arroyo, 1995), bovine bile (Fagerberg and Avens, 1976; Arroyo and Arroyo, 1995), and in one instance with sulphamandelate (Watson and Walker, 1978; Moats, 1981; McGibbon et al., 1984;

Stallard and Cox, 1994; Arroyo and Arroyo, 1995). Brilliant green did not only act as a selective agent, but it also served as a pH indicator in conjunction with phenol red, which was similar to those in modified brilliant green agar (Edel-Kampelmacher medium). Malachite green oxalate was not used in this medium to avoid complication of colour changes from fermentation of carbon sources and lysine decarboxylation.

197 Test tubes containing 10 mL of this medium with MgCl2.6H2O, brilliant green and bovine bile as selective agents were inoculated with non-salmonellae and selected serotypes of Salmonella and incubated at 37°C for 48h. Potential non-salmonellae

(Enterobacter aerogenes, Citrobacter freundii UNSW 047800, Escherichia coli,

Klebsiella pneumoniae and Proteus mirabilis) well fermented carbon sources in this medium to produce excessive acid, as indicated by high degree of colour change of medium from brown to yellow, whereas fermentation by Salmonella was in lesser extent, as indicated by colour change of medium from brown to green. Other non- salmonellae, Proteus vulgaris, Shigella sonnei, Bacillus cereus, B. subtilis and

Staphylococcus aureus, were completely inhibited in this medium. Nonetheless, neither non-salmonellae nor Salmonella exhibited lysine decarboxylation in this medium.

It was noted that the incorporation of sulphamandelate to the medium containing three selective agents (MgCl2.6H2O, brilliant green and bovine bile) improved medium selectivity by completely suppressing coliforms and Proteus mirabilis. Although sulphamandelate did not inhibit Salmonella, lysine decarboxylation by Salmonella was not observed.

It was concluded that media formulated based on the combined selective and differential principles of XLD and LMG media were not sufficiently selective and effective for use as a liquid medium in the MicroFoss assay. Attempts to improve their selectivity by the addition of selective agents as previously described gave conditions that either inhibited

Salmonella growth, or interfered with their differential properties. As a result, a new concept of detection and differentiation would be investigated in the following Section.

198 6.3.2 Development of propylene glycol-based selective medium for potential use in the MicroFoss assay

Rambach agar is another differential selective agar medium used for the detection of

Salmonella. It incorporates propylene glycol as the sole substrate for formation of acid from metabolism of Salmonella, sodium desoxycholate as a selective agent to inhibit

Gram-positive bacteria, and the chromogenic β-galactosidase substrate, X-Gal (5- bromo-4-chloro-3-indolyl β-D-galactopyranoside) which, when hydrolysed by the β-D- galactosidase activity of lactose-utilising members of the Enterobacteriaceae, yields a blue colour (Rambach, 1990; Freydiere and Gille, 1991; Gruenewald et al., 1991; Feng,

1992; Pignato et al., 1995a; Manafi, 1996; de Boer, 1998; Manafi, 2000; Jay et al.,

2003). Rambach agar has been used widely for the examination of Salmonella in foods and environmental samples and found to give reliable results (Garrick and Smith, 1994;

Joosten et al., 1994; Monfort et al., 1994; Pignato et al., 1995b). The principle of its formulation, fermentation of propylene glycol, was used to develop a medium for use in the MicroFoss. The chromogenic reaction based on β-D-galactosidase activity to X-Gal would not be adaptable to the MicroFoss assay as a differential property, so phenol red was incorporated into the formulation to serve as an indicator dye for acid production from propylene glycol. Consequently, a liquid medium with the following formulation was examined.

The formulation of propylene glycol-based liquid medium (per litre) contained 3.0g of proteose peptone, 5.0g of yeast extract, 5.0g of sodium chloride, 10.0g of propylene glycol, 29.0g of MgCl2.6H2O, 4.7mg of brilliant green, 0.08g of phenol red and 1 L of sterilised distilled water. Calculations of volumes of 40% (w/v) working MgCl2.6H2O and 0.1% working brilliant green solutions required are shown in Appendix 10.1.

199 Propylene glycol is used as the sole substrate for metabolism of Salmonella, and the selective agents are MgCl2.6H2O and brilliant green.

6.3.2.1 Evaluation of the propylene glycol-based medium

Test tubes containing 10 mL of this medium were inoculated to yield 104-106 CFU/mL of a selection of non-Salmonella species and strains of two serotypes of Salmonella, S.

Enteritidis and S. Typhimurium, as shown in Tables 6.2 and 6.3, respectively, and incubated at 42°C for 48h. The higher temperature of incubation added to the selectivity of this test, as media containing MgCl2.6H2O should be incubated at 42.0 + 0.1°C in a water bath (Peterz et al., 1989). As shown in the Table 6.5, both serotypes of

Salmonella grew in this medium, producing acid from propylene glycol to change colour of the medium from dark red to yellowish green or yellow, as indicated by phenol red. All non-salmonellae gave a negative reaction, as none changed the colour of the medium.

200 Table 6.5 Acidification of propylene glycol in propylene glycol-based medium after inoculation with selected non-Salmonella and Salmonella cultures and incubation at 42°C for 48h Culture Propylene glycol acidification Bacillus cereus (BC TICC 1996) -ve Bacillus subtilis (BC TICC 11) -ve Citrobacter freundii UNSW 047800 -ve Enterobacter aerogenes -ve Escherichia coli -ve Klebsiella pneumoniae -ve Micrococcus luteus -ve Proteus mirabilis -ve Proteus vulgaris -ve Pseudomonas aeruginosa -ve Staphylococcus aureus -ve Salmonella Enteritidis (SE) 1 +ve SE 2 ++ve SE 25 ++ve SE 64 ++ve SE RPTs (I) +ve SE G7/All ++ve SE UNSW 031900 +ve Salmonella Typhimurium (ST) +ve ST 1 ++ve

-ve reaction +ve reaction Uninoculated control

SE 1 SE 2 SE 25 ST ST 1 (+) (++) (++) (+) (++)

Figure 6.1 Colour changes in the medium as a consequence of propylene glycol acidification.

201 It may be concluded from these observations that this medium formulation had sufficient selectivity to inhibit competing non-Salmonella species, whilst it did not inhibit the growth of Salmonella. Moreover, the Salmonella growth was readily seen as a change in colour of the medium as a consequence of utilisation of propylene glycol and acid production. This differential detection mechanism was most suitable for the

MicroFoss system.

6.3.2.2 Evaluation of propylene glycol-based medium in the MicroFoss assay

• The optimum combination of the MicroFoss test parameters

According to Section 6.2.4.2, various combinations of the MicroFoss test parameters were trialled to assess the optimum combinations that provided the best-fit MicroFoss curves in terms of the correspondence between the DTs and the positions of detection.

The liquid medium and the semi-solid agar plug was prepared and dispensed in

MicroFoss vials, as shown in Section 6.2.3.2. These vials were inoculated to give 106 and 103 CFU/mL of Salmonella Enteritidis 1 for each set of the trialled combined parameters. The optimum combination of these three parameters yielding the best-fit

MicroFoss curves was the set of the detection threshold at 10, the skip factor at 1, and the shuteye period at 25 (Figure not shown).

• The MicroFoss curves of detection of Salmonella and non-salmonellae in propylene glycol-based medium

The propylene glycol medium was subsequently evaluated for its performance in the

MicroFoss assay. The liquid medium and the semi-solid agar plug was prepared and dispensed in MicroFoss vials, as shown in Section 6.2.3.2. These vials were inoculated to give 106 CFU/mL of: (1) Enterobacter aerogenes; (2) Citrobacter freundii UNSW

047800; (3) Salmonella Enteritidis 1; and (4) 10 CFU/mL of SE 1 with 103-104

202 CFU/mL of C. freundii UNSW 047800, Escherichia coli and Proteus mirabilis, and incubated at 42°C in the MicroFoss system. As shown in Figure 6.2, none of the non-

Salmonella species gave a detection time, whereas vials inoculated with the individual culture of SE 1, and a low level of SE 1 in the presence of high levels of non- salmonellae, gave optical transmission curves with amplitudes of 300-550 optical units, and detection times of 6.1h and 15.1h, respectively.

Enterobacter aerogenes No detection

Citrobacter freundii No detection UNSW 047800

Salmonella Enteritidis 1 6.1h

SE 1 (10 CFU) with 15.1h high levels of non-salmonellae

Figure 6.2 The MicroFoss detection curves for the propylene glycol-based medium inoculated to give 106 CFU/mL of: (1) Enterobacter aerogenes (No detection); (2)

Citrobacter freundii UNSW 047800 (No detection); (3) SE 1 (DT 6.1h); and (4) 10

CFU/mL of SE 1 with high levels of selected non-salmonellae (15.1h), and incubated at

42°C for 48h.

These findings indicated that the MicroFoss assay was likely to provide an effective detection system for Salmonella. Investigations to optimise the performance of this new medium are presented in the following Sections.

203 6.3.3 Optimisation of propylene glycol-based selective medium

The following Section optimised incubation temperature and choice and concentration of key ingredients of the propylene glycol-based selective medium for effective detection of Salmonella.

6.3.3.1 Incubation temperature

A slight colour change was occasionally observed in this selective medium after inoculation with Klebsiella pneumoniae and incubation at 42°C for 24-36h. This suggested that its selectivity could be improved. It has been recommended that RV broth, which contains MgCl2.6H2O as a selective agent, should be incubated at 43°C to achieve maximum selectivity (Vassiliadis, 1976, cited in Busse, 1995; Vassiliadis,

1983). Consequently, trials were conducted to determine if increasing the temperature of incubation from 42°C to 43°C would improve the selectivity of this new propylene glycol-based medium.

Test tubes of the medium (10 mL) were inoculated to give 104-106 CFU/mL of selected non-salmonellae, 12 strains of Salmonella Enteritidis and mixed cultures of low level

(10 CFU/mL) of SE 1 and high levels of competing bacteria, and incubated at 43°C for

48h.

All species of non-salmonellae yielded a negative reaction throughout the incubation period, whilst all individual strains of Salmonella Enteritidis and 10 CFU/mL of SE 1 in the presence of high levels of non-salmonellae produced acid from propylene glycol.

These findings indicated that selectivity of this propylene glycol-based medium was enhanced by incubation at 43°C, whilst incubation at this temperature did not affect propylene glycol acidification and growth of Salmonella.

204 As the incubation temperature at 43°C of the experiment in test tubes was proven to improve selectivity of this propylene glycol-based medium, the incubation temperature used in the MicroFoss system should also be at 43°C. Hence, the final test configurations of the MicroFoss Salmonella assay assembling results from the first part of Section 6.3.2.2 and this Section 6.3.3.1 are shown in Table 6.6.

Table 6.6 Test configurations of the MicroFoss Salmonella assay

Threshold Skip Shuteye Cutoff Caution Temperature Duration (°C) (h) 10 1 25 8 8 43 48

6.3.3.2 pH dye indicator

This Section assessed the appropriate pH dye indicator and its concentration to measure propylene glycol acidification by Salmonella.

••• Evaluation of the appropriate pH indicator in test tubes

The dye indicator for monitoring optical changes from metabolism of Salmonella in the

MicroFoss system was justified by sensitivity, specificity and speed of detection, as well as curve amplitudes. Selected pH dye indicators tested were bromocresol purple, bromophenol red, bromothymol blue, bromoxylenol blue, chlorophenol red and phenol red, whose pH and colour ranges are 5.2-6.8 (yellow→purple), 5.2-6.8

(orange→purple), 6.0-7.6 (yellow→blue), 5.7-7.4 (yellow→blue), 4.8-6.4

(yellow→purple) and 6.4-8.2 (yellow→red), respectively (Merck, 2002). These pH dye indicators were separately incorporated into the same basal formulation, as described in

Section 6.3.2. Concentrations of bromocresol purple, bromophenol red, bromothymol blue, bromoxylenol blue, chlorophenol red and phenol red used were 0.016, 0.020,

205 0.090, 0.090, 0.080 and 0.080 g/L, respectively. pH of the final medium in each case was in the range of 6.80-7.00 (+ 0.20).

Media in test tubes were inoculated to give 102-107 CFU/mL of selected non- salmonellae, C. freundii UNSW 047800, Ent. aerogenes, E. coli, K. pneumoniae and Pr. mirabilis, and representative serotypes of Salmonella, SE 64, ST 1, S. Newport 1 and S.

Schwarzengrund 2. Media were also inoculated to yield 102 CFU/mL each of

Salmonella, SE 64, ST 1 and S. Newport 1, in the presence of high concentrations (104-

105 CFU/mL) of non-salmonellae, which were C. freundii UNSW 047800, E. coli, K. pneumoniae and Pr. mirabilis. The inoculated media were incubated at 43°C for 48h.

Performance of pH indicators to exhibit reactions of Salmonella and non-salmonellae in propylene glycol-based selective media are shown in Table 6.7 and discussed in the following Sections.

Detection of non-salmonellae

Enterobacter aerogenes, C. freundii UNSW 047800 and K. pneumoniae were completely inhibited by the selectivity system of this medium, as neither of pH indicators detected any colour change of media inoculated with these cultures throughout the incubation period, as shown in Table 6.7.

Bromothymol blue exhibited superior specificity of detection to the other pH dye indicators, as it did not detect colour change of media inoculated with any species of non-salmonellae, as shown in Table 6.7.

Among slight colour changes of media inoculated with Escherichia coli, colour change of the medium containing phenol red was the most noticeable and quite likely to yield a

206 false positive result, as shown in Figure 6.4c, indicating that it would not be the optimum dye to use in the MicroFoss assay.

The degree of colour changes of media inoculated with Proteus mirabilis containing bromoxylenol blue and chlorophenol red was similar to that of media as a consequence of propylene glycol acidification by Salmonella, as shown in Figures 6.4a and 6.4b, respectively. This indicated that these indicators were not the optimum dye to use in the

MicroFoss assay.

Detection of Salmonella

Media formulated with any of the dyes gave a clear and distinct colour change after inoculation with high levels (107 CFU/mL) of SE 64 and incubation for 21h. Of all the dyes, the extent of colour change was less with bromothymol blue, as shown in Figure

6.3c, and this would be likely to produce an insufficient curve amplitude when analysed in the MicroFoss system.

Media inoculated with 102 CFU/mL of Salmonella did not show any colour change after

21h, but some changes were observed after 42h only in media containing bromophenol red, bromoxylenol blue and chlorophenol red, as shown in Table 6.7. This showed that these indicators could be sufficiently sensitive to detect colour changes of the medium as a consequence of propylene glycol acidification by low levels of Salmonella.

Media formulated with all dyes showed definite colour changes when inoculated with

102 CFU/mL of Salmonella and 104-105 CFU/mL of non-salmonellae and incubated for

21h, in which such changes were more readily detected by bromophenol red, bromoxylenol blue, chlorophenol red and phenol red rather than bromocresol purple and bromothymol blue, as shown in Table 6.7.

207 Table 6.7 Performance of different pH dye indicators for exhibition of acidification of propylene glycol in the medium after inoculation with selected non-Salmonella and

Salmonella cultures and incubation at 43°C for 21h and 42h

Acidification of propylene glycol in the medium Culture Bromocresol Bromophenol Bromothymol Bromoxylenol Chlorophenol Phenol red purple red blue blue red 21h 42h 21h 42h 21h 42h 21h 42h 21h 42h 21h 42h Citrobacter freundii ------UNSW 047800 Enterobacter ------aerogenes Escherichia coli ------very very wk + wk + wk + wk + Klebsiella pneumoniae ------

Proteus mirabilis ------+ - wk + - -

SE 64 (107 CFU/mL) ++ ++ ++ ++ + + ++ ++ + ++ ++ ++

SE 64 (102 CFU/mL) ------

ST 1 (102 CFU/mL) ------+ - -

S. Newport 1 - - - + - - - + - + - - (102 CFU/mL) S. Schwarzengrund 2 - - - ++ - - - ++ - + - - (102 CFU/mL) SE 64 (102 CFU/mL) wk + ++ ++ ++ - ++ ++ ++ + ++ + ++ & non-salmonellae ST 1 (102 CFU/mL) wk + ++ ++ ++ wk + ++ ++ ++ + ++ + ++ & non-salmonellae S. Newport 1 + ++ ++ ++ wk + ++ ++ ++ + ++ + ++ (102 CFU/mL) & non-salmonellae

Colour changes of media indicated by different dyes as a consequence of acidification of propylene glycol are shown in the following Section, bromocresol purple (Figure

6.3a), bromophenol red (Figure 6.3b), bromothymol blue (Figure 6.3c), bromoxylenol blue (Figure 6.4a), chlorophenol red (Figure 6.4b), and phenol red (Figure 6.4c).

208 (a) Bromocresol purple

-ve reaction +ve reaction Uninoculated control

No reaction Slight change caused by E. coli (+) (++) (+++)

(b) Bromophenol red

-ve reaction +ve reaction Uninoculated control

No reaction Slight change caused by E. coli (+) (++)

(c) Bromothymol blue

-ve reaction +ve reaction Uninoculated control

(+) (++)

Figure 6.3 Colour changes of propylene glycol-based media containing (a) bromocresol purple; (b) bromophenol red; and (c) bromothymol blue for exhibition of acidification of propylene glycol after inoculation with selected non-salmonellae and Salmonella and incubation at 43°C for 48h.

209

(a) Bromoxylenol blue

-ve reaction +ve reaction Uninoculated control

Pr. mirabilis S. Newport 1 S. Schwarzengrund 2 (+) (102 CFU) (+) (102 CFU) (++)

(b) Chlorophenol red

-ve reaction +ve reaction Uninoculated control

E. coli Pr. mirabilis Salmonella Salmonella (very wk +) (wk +) (102 CFU) (+) (107 CFU) (++)

(c) Phenol red

-ve reaction +ve reaction Uninoculated control

E. coli Salmonella Salmonella (wk +) (+) (++)

Figure 6.4 Colour changes of propylene glycol-based media containing (a) bromoxylenol blue; (b) chlorophenol red; and (c) phenol red for exhibition of acidification of propylene glycol after inoculation with selected non-salmonellae and Salmonella and incubation at 43°C for 48h. 210 Findings of the test tube evaluation showed that bromophenol red was the optimum pH indicator to exhibit acidification of propylene glycol in the medium in regards to sensitivity, specificity and speed of detection.

••• Evaluation of the appropriate pH indicator in the MicroFoss assay

Propylene glycol-based medium with different pH indicators that was inoculated with

107 CFU/mL of Salmonella Enteritidis 64 was analysed in the MicroFoss assay to evaluate the appropriate indicator that produced a sufficient curve amplitude and gave a good response to exhibit acidification of propylene glycol in the medium.

Figure 6.5 shows variation of the MicroFoss detection of SE 64 inoculated into the propylene glycol-based medium as a consequence of formulation with different indicators yielding detection times ranging from 2.3h to 7.3h.

Phenol red 2.3h

Bromocresol purple 3.8h

Bromothymol blue 2.5h

Chlorophenol red 5.1h

Bromoxylenol blue 2.8h

Bromophenol red 7.3h

Figure 6.5 The MicroFoss curves of detection of 107 CFU/mL of SE 64 inoculated into propylene glycol-based medium formulated with different dye indicators, phenol red

(DT 2.3h); bromocresol purple (3.8h); bromothymol blue (2.5h); chlorophenol red

(5.1h); bromoxylenol blue (2.8h); and bromophenol red (7.3h), which were incubated at

43°C for 48h.

211 Results of the MicroFoss detection of propylene glycol acidification of SE 64 using media formulated with different pH indicators are discussed in the following Section.

Curve amplitude

Different curve amplitudes resulted from variation in degree of colour changes of the medium depending on choice of pH indicator. Curve amplitudes of detection of SE 64 inoculated into media formulated with chlorophenol red, bromothymol blue and phenol red were shallow to extremely shallow. The optical units started at 150, 300 and 400 in respect for the uninoculated media containing chlorophenol red, bromothymol blue and phenol red and increased to 400, 420 and 600, respectively when acid was produced from metabolism of SE 64, as shown in Figure 6.5. These indicators were not the appropriate dye for the MicroFoss detection of Salmonella, as they did not give sufficiently optimum curve amplitude and could be interpreted as a false positive result.

For instance, the MicroFoss detection of 105 CFU/mL each of Proteus mirabilis and S.

Enteritidis 31 in the propylene glycol-based medium containing bromothymol blue yielded DTs of 3.4h and 5.3h and shallow curve amplitudes, with the optical units in the range of 260 to 480 and 260 to 500, respectively, as shown in Figure 6.6. Furthermore, the MicroFoss detection of 106 CFU/mL each of Klebsiella pneumoniae and Pr. mirabilis, S. Enteritidis 25 and S. Typhimurium 1 in propylene glycol-based media containing phenol red yielded DTs of 21.9h, 30.6h, 3.5h and 3.1h and shallow curve amplitudes, with the optical units in the range of 400 to 520 and 400 to 550, 400 to 520 and 400 to 580, respectively, as shown in Figure 6.7.

The MicroFoss detection of SE 64 using the propylene glycol-based medium formulated with bromophenol red, bromocresol purple and bromoxylenol blue produced curves with distinct curve amplitudes, in which their optical units were in the range of 180 to

212 600, as shown in Figure 6.5. This indicated that these indicators could be optimum for use in the MicroFoss assay.

Detection time

Among all pH indicators, detection of high levels of SE 64 (107 CFU/mL) using the medium containing phenol red, bromothymol blue, bromoxylenol blue and bromocresol purple gave good response, yielding early detection with DTs of 2.3h, 2.5h, 2.8h and

3.8h, respectively, as shown in Figure 6.5. However, detection of the same level of SE

64 using the medium containing chlorophenol red and bromophenol red was prolonged to 5.1h and 7.3h, respectively, as shown in Figure 6.5. This showed that chlorophenol red and bromophenol red were not the optimum dyes to use in the MicroFoss assay.

Evaluation of the appropriate pH indicator for use in the MicroFoss assay, as determined by curve amplitude and detection time, indicated that bromocresol purple and bromoxylenol blue were the optimum dyes to exhibit colour change of the medium as a result of propylene glycol acidification by Salmonella. However, it was concluded from the evaluation in test tube that bromoxylenol blue was not the optimum dye for use in the MicroFoss assay, as this dye also detected a false positive result caused by

Proteus mirabilis.

Although results of the evaluation in test tubes showed the lack of sensitivity of detection of 102 CFU/mL of Salmonella using the medium with bromocresol purple, results of other Experiments demonstrated good sensitivity of this dye to detect low level of Salmonella. For example, the MicroFoss detection of <10 CFU/mL of S.

Typhimurium 1 and 10 CFU/mL of S. Enteritidis 59 using media containing bromocresol purple yielded DTs of 19.6h and 16.5h with the optimum curve amplitudes in the range of 180 to 600 optical units, as shown in Figures 6.8 and 6.9, respectively.

213 Hence, the overall results indicated that bromocresol purple was the most suitable pH indicator to exhibit acidification of propylene glycol by Salmonella.

Citrobacter freundii No detection UNSW 047800

Proteus mirabilis 3.4h

SE 31 (105 CFU/mL) 5.3h

SE 31 (<10 CFU/mL) 17.9h

SE 31 (<10 CFU/mL) 8.0h with high levels of non-salmonellae

Figure 6.6 The MicroFoss detection of selected non-salmonellae and SE 31 inoculated into propylene glycol-based media containing bromothymol blue: (1) 105 CFU/mL of Citrobacter freundii UNSW 047800 (No detection); (2) 105 CFU/mL of Proteus mirabilis (DT 3.4h); (3) 105 CFU/mL of SE 1 (5.3h); (4) <10 CFU/mL of SE 31 (17.9h); and (5) <10 CFU/mL of SE 31 with high levels of selected non-salmonellae (8.0h), and incubated at 43°C for 48h.

Klebsiella pneumoniae 21.9h

Proteus mirabilis 30.6h

SE 25 3.5h

ST 1 3.1h

Figure 6.7 The MicroFoss detection of 106 CFU/mL each of selected non-salmonellae and Salmonella in propylene glycol-based media containing phenol red: (1) Klebsiella pneumoniae (DT 21.9h); (2) Proteus mirabilis (30.6h); (3) SE 25 (3.5h); and (4) ST 1 (3.1h), and incubated at 43°C for 48h.

214 ST 1 (106 CFU/mL) 2.5h

ST 1 (<10 CFU/mL) 19.6h

ST 1 (<10 CFU/mL) 14.4h with high levels of non-salmonellae

Figure 6.8 The MicroFoss detection of ST 1 inoculated into propylene glycol-based media containing bromocresol purple: (1) 106 CFU/mL of ST 1 (2.5h); (2) <10 CFU/mL of ST 1 (19.6h); and (3) <10 CFU/mL of ST 1 with high levels of selected non- salmonellae (14.4h), and incubated at 43°C for 48h.

SE 59 (105 CFU/mL) 8.1h

SE 59 (10 CFU/mL) 16.5h

SE 59 (10 CFU/mL) 15.9h with high levels of non-salmonellae

Figure 6.9 The MicroFoss detection of SE 59 inoculated into propylene glycol-based media containing bromocresol purple: (1) 105 CFU/mL of SE 59 (8.1h); (2) 10 CFU/mL of SE 59 (16.5h); and (3) 10 CFU/mL of SE 59 with high levels of selected non- salmonellae (15.9h), and incubated at 43°C for 48h.

215 ••• Evaluation of the appropriate concentration of bromocresol purple

The appropriate concentration of bromocresol purple to exhibit acidification of propylene glycol was determined by detection of 106 CFU/mL of S. Enteritidis 17 using the propylene glycol-based medium containing various concentrations of bromocresol purple, 16, 32, 48, 64 and 80 mg/L, in the MicroFoss assay at 43°C for 48h.

As shown in Figure 6.10, the MicroFoss detection of 106 CFU/mL of SE 17 in the medium containing 16 mg/L of bromocresol purple gave the best detection, yielding a

DT of 3.8h with the optimum curve amplitude in the range of 180 to 550. Although detection of SE 17 using the medium containing 32 mg/L of bromocresol purple also gave a curve with the optimum curve amplitude (130 to 500 optical units), the detection was prolonged to 5.2h. Detection of SE 17 in the medium containing 48, 64 and 80 mg/L of bromocresol purple was retarded, yielding DTs of 8.8h, 14.9h and 16.0h with declined curve amplitudes ranging from the optical units of 130 to 450, 130 to 400 and

130 to 350, respectively. These findings indicated that 16 mg/L of bromocresol purple was optimum to exhibit acidification of propylene glycol in the MicroFoss assay.

BCP 16 mg/L 3.8h

BCP 32 mg/L 5.2h

BCP 48 mg/L 8.8h

BCP 64 mg/L 14.9h

BCP 80 mg/L 16.0h

Figure 6.10 The MicroFoss curves for detection of 106 CFU/mL of SE 17 in propylene glycol-based media containing 16 mg/L of BCP (DT 3.8h); 32 mg/L of BCP (5.2h); 48 mg/L of BCP (8.8h); 64 mg/L of BCP (14.9h); and 80 mg/L of BCP (16.0h).

216 6.3.3.3 Proteinaceous substrate

This Section evaluated choices and concentrations of proteinaceous substrates, yeast extract and proteose peptone, of the propylene glycol-based medium for the effective detection of Salmonella. Miller et al. (1995) showed the benefit of incorporation of small amount (1.2 g/L) of proteose peptone No. 3 in xylose lysine Tergitol-4 (XLT4) agar to enhance greater H2S production and facilitate detection of weaker H2S- producing strains of Salmonella, whilst it maintained great inhibition on other competing bacteria. However, concentration of proteose peptone No. 3 could be varied in the range of 1.2-1.8 g/L when examining different food samples and Salmonella serotypes. Thus, trials were conducted to determine if different choice and decreased concentration of proteinaceous substrates, (i) only 5 g/L of yeast extract; (ii) 5 g/L of yeast extract and 1.5 g/L of bacteriological peptone; and (iii) 5 g/L of yeast extract and

1.5 g/L of proteose peptone would improve detection of Salmonella in this propylene glycol-based medium.

Media were inoculated to give 102-107 CFU/mL of selected non-salmonellae, C. freundii UNSW 047800, Ent. aerogenes, E. coli, K. pneumoniae and Pr. mirabilis, and representative serotypes of Salmonella, SE 17, SE 64, S. Agona 1, S. Dublin, S. Hadar

2, S. Sofia sswt, S. Montevideo 1 and S. Newport 1. Media were also inoculated to yield

102 CFU/mL each of Salmonella, SE 64, S. Newport 1 and S. Agona 1, in the presence of high concentrations (104-106 CFU/mL) of non-salmonellae, which were E. coli, K. pneumoniae and Pr. mirabilis. The inoculated media were incubated at 43°C for 48h.

Results in Table 6.8 clearly showed that the medium containing greater nutritious proteinaceous substrates promoted more growth of Salmonella, as indicated by greater extents of propylene glycol acidification by Salmonella. All non-salmonellae gave negative reaction in all formulations. Therefore, the optimum choices and

217 concentrations of proteinaceous substrates in the propylene glycol-based medium were

5 g/L of yeast extract and 3 g/L of proteose peptone.

Table 6.8 Effectiveness of different choices and concentrations of proteinaceous substrates for exhibition of acidification of propylene glycol in the medium after inoculation with selected non-Salmonella and Salmonella cultures and incubation at 43°C for 24h and 48h

Acidification of propylene glycol in the medium Culture 5 g/L of yeast extract 5 g/L of yeast extract 5 g/L of yeast extract 5 g/L of yeast extract yeast extract & 1.5 g/L of & 1.5 g/L of & 3 g/L of bacteriological peptone proteose peptone proteose peptone 24h 48h 24h 48h 24h 48h 24h 48h Citrobacter freundii ------UNSW 047800 (105 CFU/mL) Enterobacter ------aerogenes (106 CFU/mL) Escherichia coli ------(106 CFU/mL) Klebsiella pneumoniae ------(106 CFU/mL) Proteus mirabilis ------(106 CFU/mL) SE 64 (107 CFU/mL) - +++ + +++ +++ +++ +++ +++

SE 17 (106 CFU/mL) + ++ ++ ++ ++ ++ ++ +++

S. Montevideo 1 - +++ + +++ +++ +++ +++ +++ (107 CFU/mL) S. Newport 1 + +++ + +++ +++ +++ +++ +++ (107 CFU/mL) S. Dublin - + - ++ +++ +++ +++ +++ (107 CFU/mL) S. Agona 1 - - - ++ +++ +++ +++ +++ (106 CFU/mL) S. Sofia sswt - - - - + +++ ++ +++ (106 CFU/mL) S. Hadar 2 - ++ - ++ +++ +++ +++ +++ (107 CFU/mL) SE 64 (102 CFU/mL) - - - - - +++ +++ +++

S. Newport 1 - - - - - +++ +++ +++ (102 CFU/mL) S. Agona 1 ------(102 CFU/mL) SE 64 (102 CFU/mL) - - - ++ +++ +++ +++ +++ & non-salmonellae S. Newport 1 - ++ + +++ ++ +++ +++ +++ (102 CFU/mL) & non-salmonellae S. Agona 1 - - - ++ - +++ - +++ (102 CFU/mL) & non-salmonellae

+, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow. 218 6.3.3.4 Selective agent

This Section evaluated if the increase of concentration of MgCl2.6H2O as a selective agent from 29 g/L to 32 g/L would improve the selectivity of propylene glycol-based medium.

••• Evaluation of selectivity of propylene glycol-based medium in test tubes

Media were inoculated to give 10-107 CFU/mL of selected non-salmonellae, C. freundii

UNSW 047800, Ent. aerogenes, E. coli, K. pneumoniae, Pr. mirabilis and

Staphylococcus aureus, and representative serotypes of Salmonella, SE 2, ST 1, S.

Muenster and S. Senftenberg 2. The medium was also inoculated to yield 102 CFU/mL of SE 2 in the presence of high concentrations (103-105 CFU/mL) of non-salmonellae, which were C. freundii UNSW 047800, E. coli and Pr. mirabilis. The inoculated media were incubated at 43°C for 48h.

As shown in Table 6.9, it clearly showed that low levels (10-102 CFU/mL) of

Salmonella were completely inhibited in the medium containing 32 g/L of MgCl2.6H2O, or their ability to acidify propylene glycol in the medium was retarded in a greater extent than those inoculated into the medium containing 29 g/L of MgCl2.6H2O. This indicated that 32 g/L of MgCl2.6H2O formulated in the medium yielded an inhibitory effect on Salmonella. Among non-salmonellae tested, C. freundii UNSW 047800, Ent. aerogenes, K. pneumoniae and Staph. aureus were completely inhibited in both of media containing 29 and 32 g/L of MgCl2.6H2O, whereas E. coli and Pr mirabilis were not inhibited and caused slight colour changes to these media. Nonetheless, the degree of colour changes was small and this would be insufficient to be detected by the

MicroFoss assay. These findings indicated that propylene glycol-based medium containing 29 g/L of MgCl2.6H2O had sufficient selectivity of detection and allowed growth of Salmonella.

219 Table 6.9 Evaluation of selectivity of propylene glycol-based medium containing different concentrations of MgCl2.6H2O after inoculation with selected non-Salmonella and Salmonella cultures and incubation at 43°C for 48h

Culture Acidification of propylene glycol in the medium after incubation at 43°C for 48h

29 g/L of MgCl2.6H2O 32 g/L of MgCl2.6H2O Citrobacter freundii - - UNSW 047800 (105 CFU/mL) Enterobacter aerogenes - - (107 CFU/mL) Escherichia coli (107 CFU/mL) - -

Klebsiella pneumoniae - - (107 CFU/mL) Proteus mirabilis (107 CFU/mL) - -

Staphylococcus aureus - - (107 CFU/mL) SE 2 (107 CFU/mL) +++ +++

SE 2 (102 CFU/mL) - -

SE 2 (10 CFU/mL) - -

ST 1 (107 CFU/mL) +++ +++

ST 1 (102 CFU/mL) +++ -

ST 1 (10 CFU/mL) +++ -

S. Muenster 2 (107 CFU/mL) +++ +++

S. Muenster 2 (102 CFU/mL) +++ -

S. Muenster 2 (10 CFU/mL) - -

S. Senftenberg 2 (107 CFU/mL) +++ +++

S. Senftenberg 2 (102 CFU/mL) - -

S. Senftenberg 2 (10 CFU/mL) - -

SE 2 (102 CFU/mL) +++ +++ & non-salmonellae

+, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow.

220 ••• Evaluation of selectivity of propylene glycol-based medium in the MicroFoss assay

Selectivity of propylene glycol-based medium containing 29 g/L of MgCl2.6H2O was assessed with selected non-salmonellae and Salmonella Enteritidis 2 and incubated at

43°C for 48h. As shown in Figure 6.11, media inoculated to yield 107 CFU/mL each of all non-salmonellae, E. coli, K. pneumoniae, Pr. mirabilis and Staph. aureus, were not detected by the MicroFoss system. The MicroFoss detection of 107 CFU/mL of SE 2 in the medium yielded a DT of 5.9h as a consequence of acidification of propylene glycol, producing a curve with the optimum curve amplitude of 150 to 530 optical units.

However, low level (10 CFU/mL) of SE 2 was inhibited in the medium and therefore not detected by the MicroFoss.

Escherichia coli No detection

Klebsiella pneumoniae No detection

Proteus mirabilis No detection Staphylococcus aureus No detection

SE 2 (107 CFU/mL) 5.9h

SE 2 (10 CFU/mL) No detection

Figure 6.11 The MicroFoss detection of 107 CFU/mL of selected non-salmonellae and

SE 2 in propylene glycol-based media containing 29 g/L of MgCl2.6H2O: (1)

Escherichia coli (No detection); (2) Klebsiella pneumoniae (No detection); (3) Proteus mirabilis (No detection); (4) Staphylococcus aureus (No detection); (5) Salmonella

Enteritidis 2 (DT 5.9h); and (6) 10 CFU/mL of SE 2 (No detection), and incubated at

43°C for 48h.

221 It was concluded propylene glycol-based medium containing 29 g/L of MgCl2.6H2O yielded the optimum selectivity of detection for use in the MicroFoss assay.

6.3.4 Modification of propylene glycol-based medium

Although the propylene glycol-based medium provided a good detection system for

Salmonella in the MicroFoss assay, slight colour changes were observed inf the medium, caused by certain non-salmonellae such as E. coli, K. pneumoniae and Pr. mirabilis. In some cases, the changes were insignificant, though in others, the changes approached those produced by Salmonella. Hence, the following Section modified this propylene glycol-based medium to improve the selectivity by addition of an antibiotic, novobiocin, or a surfactant, Niaproof 4. From this point onwards, the formulation of propylene glycol-based medium originally developed will be termed ‘the original formula’, whilst the formulation, for which novobiocin or Niaproof 4 was added, will be termed ‘the modified formula’.

6.3.4.1 Addition of novobiocin

Jeffries (1959) and Hargrove et al. (1971) demonstrated the effectiveness of novobiocin added to enrichment broths to increase the selectivity of these broths to facilitate the isolation of Salmonella.Restaino et al. (1977) and Komatsu and Restaino (1981) indicated that the incorporation of 80 and 5 mg/L of novobiocin into Hektoen enteric

(HE) and XLD agars, respectively did not affect growth and colony morphology of salmonellae, whilst these concentrations completely inhibited growth of Pr. mirabilis and facilitated differentiation of colonies of C. freundii and E. coli from Salmonella.

Restaino et al. (1982) reported that 40 and 7 mg/L of novobiocin supplemented to HE and XLD agars, respectively yielded the greatest percentage of recovery of salmonellae from fresh food products without altering their colony morphology. These

222 concentrations on both media suppressed growth of some strains of Pr. mirabilis, while those that could grow lost the ability to produce H2S. The colony morphology of C. freundii on XLD agar supplemented with 7 mg/L of novobiocin was altered and this was likely to reduce the probability of false positive results. Hydrogen sulfide production by C. freundii and E. coli was reduced when HE agar was supplemented with 40 mg/L of novobiocin.

Novobiocin at concentrations of 5, 10, 20, 40 and 80 mg/L were added to the formulation of the original formula, which were equivalent to the volumes of 0.28, 0.55,

1.11, 2.21 and 4.42 mL, respectively of working novobiocin solution (20 mg/mL).

Details of preparation and calculations are shown in Appendix 10.2.

Media of all formulations were inoculated to give 105-107 CFU/mL of selected non- salmonellae, C. freundii UNSW 047800, Ent. aerogenes, E. coli, K. pneumoniae and Pr. mirabilis, and 103-104 CFU/mL of representative serotypes of Salmonella, SE 1, SE 16,

ST 1, S. Muenster and S. Senftenberg 2. Media were also inoculated to yield 103

CFU/mL of SE 1 in the presence of high concentrations (104-106 CFU/mL) of all selected non-salmonellae, as just described. The inoculated media were incubated at

43°C for 48h.

Each medium, supplemented novobiocin, was inoculated with a pool of Salmonella and non-Salmonella cultures. After incubation at 43°C for 20h, the inoculated media were streaked onto XLD plates. XLD plates were incubated at 37°C for 24h.

Results of the improved selectivity of the modified formula with novobiocin are shown in Table 6.10 and discussed in the following Sections.

223 While Proteus mirabilis caused colour change in the original formula of propylene glycol-based medium, it was completely inhibited in the presence of ≥10 mg/L of novobiocin.

Although Escherichia coli was not completely inhibited at all concentrations of novobiocin, the very slight colour changes caused by E. coli in these media were insufficient to give a detection in the MicroFoss system.

Klebsiella pneumoniae was not inhibited at all concentrations of novobiocin and produced colour changes similar to those produced by Salmonella. However, the degree of colour change was decreased noticeably in media supplemented with ≥20 mg/L of novobiocin. Review of detection curves would differentiate those produced by

Klebsiella compared to those produced by Salmonella.

Recovery of 103 CFU/mL SE 1 in the presence of 104-106 CFU/mL of non-salmonellae was significantly increased when 40 and 80 mg/L of novobiocin were added to the original formula, as shown in Figures 6.12e and 6.12f, respectively.

All individual strains of Salmonella were able to utilise propylene glycol in media supplemented with ≤40 mg/L of novobiocin after incubation at 43°C for 18h. It was noted that the medium supplemented with 80 mg/L of novobiocin gave a partially inhibitory effect on propylene glycol acidification of some strains of Salmonella, such as SE 16 and S. Senftenberg 2, as shown in Table 6.10. As a result, acidification of propylene glycol by these strains of Salmonella was prolonged and completion was observed within 48h.

Hence, the optimum concentration of novobiocin supplemented to the original formula for improvement of the selectivity without any adverse effect on the growth of

Salmonella was 40 mg/L.

224 Table 6.10 Evaluation of the selectivity of propylene glycol-based medium supplemented with various concentrations of novobiocin after incubation at 43°C for 18h and 48h

Acidification of propylene glycol in the medium supplemented with various concentrations of novobiocin Culture 0 mg/L 5 mg/L 10 mg/L 20 mg/L 40 mg/L 80 mg/L 18h 48h 18h 48h 18h 48h 18h 48h 18h 48h 18h 48h Citrobacter freundii ------UNSW 047800 Enterobacter ------aerogenes Escherichia coli + + + + + + + + - - - - Klebsiella pneumoniae +++ +++ +++ +++ +++ +++ ++ ++ ++ ++ ++ ++ Proteus mirabilis - +++ - + ------SE 1 (104 CFU/mL) +++ +++ +++ +++ +++ +++ ++ ++ ++ ++ ++ ++ SE 16 (103 CFU/mL) +++ +++ +++ +++ +++ +++ +++ +++ ++ ++ - ++ ST 1 (104 CFU/mL) +++ +++ +++ +++ +++ +++ +++ +++ ++ ++ ++ ++ S. Muenster +++ +++ +++ +++ +++ +++ +++ +++ ++ ++ ++ ++ S. Senftenberg 2 +++ +++ +++ +++ +++ +++ +++ +++ ++ ++ - ++ SE 1 (103 CFU/mL) & +++ +++ +++ +++ +++ +++ +++ +++ ++ ++ ++ ++ non-salmonellae (104-106 CFU/mL) +, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow.

a b c

d e f

Figure 6.12 The increased recovery of SE 1 in the presence of high levels of non- salmonellae with (a) 0 mg/L; (b) 5 mg/L; (c) 10 mg/L; (d) 20 mg/L; (e) 40 mg/L; and (f) 80 mg/L of novobiocin supplemented to the original formulation of propylene glycol- based medium, as indicated by the ratio of Salmonella to non-salmonellae on XLD plates.

225 6.3.4.2 Addition of Niaproof 4

Niaproof type 4 (7-ethyl-2-methyl-4-undecanol-hydrogen sulfate, sodium tetradecyl sulphate, formerly Tergitol 4) can be added to commercial culture media for enhancing greater selectivity of Salmonella detection. Niaproof 4 (4.6 mL/L) formulated in xylose- lysine-tergitol 4 (XLT4) agar largely or completely inhibited a range of non- salmonellae, notably Proteus, Providencia, and Pseudomonas, and while C. freundii was able to grow on the medium, it could not produce H2S. This medium therefore provided for effective detection and differentiation of Salmonella (Miller et al., 1991;

Miller et al., 1995; Sherrod et al., 1995).

Niaproof 4 at concentrations of 1, 2 and 3 mL/L was added to the original formulation of the propylene glycol-based medium, as shown in Appendix 10.3.

Media of all formulations were inoculated to give 105-107 CFU/mL of selected non- salmonellae, C. freundii UNSW 047800, C. freundii IMVS 1263, Ent. aerogenes, E. coli, K. pneumoniae, Pr. mirabilis, Pr. vulgaris, Pseudomonas aeruginosa, Shigella sonnei and Yersinia enterocolitica IMVS 2861, and 102-106 CFU/mL of representative strains of Salmonella, SE 1 and ST 1. Media were also inoculated to yield 103 CFU/mL of SE 1 in the presence of high concentrations (104-106 CFU/mL) of selected non- salmonellae, C. freundii UNSW 047800, Ent. aerogenes, E. coli, K. pneumoniae and Pr. mirabilis. The inoculated media were incubated at 43°C for 48h, and then streaked onto

XLD plates, which were incubated at 37°C for 24h.

Results of selectivity of detection using the original formula and the modified formula with Niaproof 4 are shown in Table 6.11 and discussed in the following sections.

226 While all strains of Salmonella (102-106 CFU/mL) produced acid from propylene glycol in media supplemented with 1 and 2 mL/L of Niaproof 4 within 18h, they were suppressed or completely inhibited in the medium supplemented with 3 mL/L of

Niaproof 4. For those strains suppressed, acid production from propylene glycol was prolonged and completion was observed within 48h (Table 6.11).

Recovery of 103 CFU/mL SE 1 in the presence of 104-106 CFU/mL of non-salmonellae was significantly increased when 1, 2 and 3 mL/L of Niaproof 4 was added to the original formula (Figures 6.13b, 6.13c and 6.13d, respectively).

All species of non-salmonellae were completely inhibited at all concentrations of

Niaproof 4.

Hence, the optimum concentrations of Niaproof 4 supplemented to the original formula for improvement of selectivity without any adverse effect on the growth of Salmonella were 1 and 2 mL/L.

227 Table 6.11 Evaluation of the selectivity of propylene glycol-based medium supplemented with various concentrations of Niaproof 4 after incubation at 43°C for 18h and 48h

Acidification of propylene glycol in the medium supplemented with various concentrations of Niaproof 4 Culture 0 mL/L 1 mL/L 2 mL/L 3 mL/L 18h 48h 18h 48h 18h 48h 18h 48h Citrobacter freundii ------UNSW 047800 Citrobacter freundii ------IMVS 1263 Enterobacter aerogenes ------

Escherichia coli + + ------

Klebsiella pneumoniae +++ +++ ------

Proteus mirabilis + +++ ------

Proteus vulgaris ------

Pseudomonas aeruginosa ------

Shigella sonnei ------

Yersinia enterocolitica ------IMVS 2861 SE 1 (106 CFU/mL) ++ +++ ++ ++ ++ ++ - -

SE 1 (105 CFU/mL) ++ +++ ++ ++ ++ ++ - -

SE 1 (104 CFU/mL) - +++ ++ ++ ++ ++ - -

SE 1 (103 CFU/mL) - +++ ++ ++ ++ ++ - -

SE 1 (102 CFU/mL) - ++ ++ ++ ++ ++ - -

ST 1 (106 CFU/mL) +++ +++ ++ +++ ++ ++ ++ ++

ST 1 (105 CFU/mL) ++ +++ ++ ++ ++ ++ + ++

ST 1 (104 CFU/mL) + +++ ++ ++ ++ ++ - ++

ST 1 (103 CFU/mL) - +++ ++ ++ ++ ++ - ++

ST 1 (102 CFU/mL) - +++ ++ ++ + ++ - ++

SE 1 (103 CFU/mL) +++ +++ ++ ++ + ++ + ++ & non-salmonellae (104-106 CFU/mL) +, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow. 228 a b

c d

Figure 6.13 The increased recovery of SE 1 in the presence of high levels of non- salmonellae with (a) 0 mL/L; (b) 1 mL/L; (c) 2 mL/L; and (d) 3 mL/L of Niaproof 4 supplemented to the original formulation of propylene glycol-based medium, as indicated by the ratio of Salmonella to non-salmonellae on XLD plates.

229 6.3.5 Examinations of the effectiveness of propylene glycol-based media

The following Sections will evaluate the selectivity and effectiveness of Salmonella detection using all formulation of propylene glycol-based media, the original formula and the modified formulae with supplementations of 40 mg/L of novobiocin, and 1 and

2 mL/L of Niaproof 4, with a range of non-Salmonella species and various serotypes of

Salmonella.

6.3.5.1 Analyses with non-salmonellae

Media of all formulations were inoculated to yield 105-107 CFU/mL of non-Salmonella cultures, which were a number of genera in the family Enterobacteriaceae, miscellaneous Gram-negative bacteria and Gram-positive bacteria (Table 6.2). Media were also inoculated to yield 104-106 CFU/mL of mixed cultures, as described in the following Sections.

• The four coliforms: C. freundii UNSW 047800, Ent. aerogenes, E. coli and K. pneumoniae;

• The five coliforms: The aforementioned four coliforms and Ent. cloacae;

• The four coliforms and Pr. mirabilis;

• The five coliforms and Pr. mirabilis.

The inoculated media were incubated at 43°C for 48h.

230 Results of the evaluation of selectivity of propylene glycol-based media with non- salmonellae are shown in Table 6.12 and discussed in the following Sections.

The family Enterobacteriaceae

Most species of the family Enterobacteriaceae were substantially or completely inhibited in all formulations of propylene glycol-based medium, as shown in Table 6.12.

Although E. coli was not totally suppressed in all formulations of the medium, the degree of slight to very slight colour changes caused by E. coli in these media was insufficient to give a detection in the MicroFoss system.

Proteus mirabilis was completely inhibited in the presence of novobiocin and Niaproof

4 at all concentrations.

Enterobacter cloacae was completely inhibited in the presence of 2 mL/L of Niaproof

4, whereas it produced colour changes in all of the other formulations. The degree of colour changes in the original formula and the modified formula with 40 mg/L of novobiocin was similar to those produced by Salmonella, whilst the very slight colour change in the modified formula with 1 mL/L of Niaproof 4 was insufficient to give a detection in the MicroFoss system.

Klebsiella pneumoniae was the most resistant species to the selectivity of the medium compared to the other species tested, except for the medium supplemented with 2 mL/L of Niaproof 4. Colour changes of the other formulations of media caused by K. pneumoniae resembled those produced by Salmonella.

Colour changes of the original formula inoculated with all pools of mixed cultures were similar to those produced by Salmonella. Supplementations of the original formula with both novobiocin and Niaproof 4 effectively suppressed these mixed cultures to a great extent that reduced the probability of these cultures to cause a false positive result.

231 Miscellaneous Gram-negative and Gram-positive bacteria

All strains of miscellaneous Gram-negative and Gram-positive bacteria were completely inhibited in all formulations of propylene glycol-based medium.

It was concluded that the original formula of propylene glycol-based medium generally provided good selectivity for Salmonella detection. Supplementations of the original formula with novobiocin or Niaproof 4 improved selectivity of detection by decreasing or eliminating false positive results caused by potential non-salmonellae, such as Ent. cloacae, K. pneumoniae and Pr. mirabilis. Greatest selectivity of detection was achieved in the presence of 2 mL/L of Niaproof 4.

232 Table 6.12 Evaluation of the selectivity of the original formula and the modified

formulae with novobiocin or Niaproof 4 of propylene glycol-based medium after

inoculation with a range of non-Salmonella cultures and incubation at 43°C for 48h

Acidification of propylene glycol in the medium with and without modification Culture The original The modified formula The modified formula The modified formula formula with 40 mg/L of NV* with 1 mL/L of NP4** with 2 mL/L of NP4**

Enterobacteriaceae

Citrobacter freundii - - - - UNSW 047800 Citrobacter freundii - - - - IMVS 1263 Enterobacter aerogenes - - - - Enterobacter cloacae +++ ++ - - Escherichia coli - - - - Klebsiella pneumoniae +++ ++ + - Proteus mirabilis ++ - - - Proteus vulgaris - - - - Providencia stuartii - - - - Serratia marcescens - - - - Shigella sonnei - - - - Yersinia enterocolitica - - - - IMVS 2861

Miscellaneous Gram-negative bacteria

Acinetobacter baumannii - - - - subsp. anitratus UNSW 050200 Acinetobacter - - - - calcoaceticus/baumannii Acinetobacter - - - - calcoaceticus/haemolyticus Aeromonas hydrophila - - - - Flavobacterium spp. - - - - Pseudomonas aeruginosa - - - - Pseudomonas fluorescens - - - - Pseudomonas putida - - - - Stenotrophomonas - - - - maltophilia

Gram-positive bacteria

Bacillus cereus - - - - BC TICC 1996 Bacillus subtilis - - - - BS TICC 11 Listeria monocytogenes L4 - - - - Micrococcus luteus - - - - Staphylococcus aureus - - - - +, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow. * NV = Novobiocin.

233 6.3.5.2 Analyses with salmonellae

Media of all formulations were inoculated to yield 103-104 CFU/mL of various strains of S. Enteritidis; a range of other serotypes; and six strains of atypical biotype of

Salmonella. The inoculated media were incubated at 43°C for 48h.

Results of the effectiveness of detection of a range of Salmonella serotypes and atypical biotype of Salmonella using propylene glycol-based media, as shown in Tables 6.13 and

6.14, respectively, are discussed in the following Sections.

Salmonella Enteritidis

Analyses of propylene glycol acidification of various strains of S. Enteritidis indicated that 23 out of 26 strains of S. Enteritidis (88.46%) were able to acidify propylene glycol in all formulations of the medium. Strains of S. Enteritidis that gave negative reactions in all formulations were SE 13, SE C6B and rSE 6.

Other serotypes of Salmonella

Among 27 strains across the other 18 serotypes tested, 25 strains (92.59%) were able to acidify propylene glycol in all formulations of the medium. It was noted that the presence of 40 mg/L of novobiocin did not have an inhibitory effect on propylene glycol acidification of these serotypes of Salmonella. Regardless of the presence of novobiocin or Niaproof 4, S. Sofia 89 gave a negative reaction in this propylene glycol- based medium. Salmonella Typhimurium was inhibited at both concentrations of

Niaproof 4.

Atypical biotype of Salmonella

Among six strains of atypical biotype of Salmonella, Salmonella subsp 2 ser

1,9,12:g,z62:- SRL 04-803419 and S. Derby IMVS 0878 were able to acidify propylene glycol in all formulations of the medium.

234 As the degree of colour changes of all formulations of propylene glycol-based media caused by S. Choleraesuis var Australia IMVS 1594/84 was very small, this strain will be investigated whether it characterised propylene glycol acidification, or it was inhibited by selectivity of the medium in the following Section.

The strains that gave negative reactions in all formulations of the medium were S.

Choleraesuis IMVS 1259/92, S. Choleraesuis var Kunzendorf IMVS 1269 and S. subsp

3b ser 61:i:z53 IMVS 1679.

It was concluded that detection of Salmonella using all formulations of propylene glycol-based media was effective and this could provide an efficient detection of

Salmonella by the MicroFoss assay.

Since a majority of strains of atypical biotype of Salmonella yielded either a weak positive or negative reactions in the original formula of propylene glycol-based medium, all strains of atypical biotype of Salmonella will be evaluated whether they were able to acidify propylene glycol, or they were inhibited by selectivity of the medium in the following Section.

235 Table 6.13 Evaluation of the effectiveness of the original formula and the modified formulae with novobiocin or Niaproof 4 of propylene glycol-based medium after inoculation with various serotypes of Salmonella and incubation at 43°C for 18h and 48h

Acidification of propylene glycol in the medium with and without modification Culture The original The modified formula The modified formula The modified formula formula with 40 mg/L of NV* with 1 mL/L of NP4** with 2 mL/L of NP4** 18h 48h 18h 48h 18h 48h 18h 48h

S. Enteritidis

SE 1 ++ ++ ++ +++ ++ +++ ++ ++ SE 2 ++ +++ ++ +++ ++ +++ ++ ++ SE 4 +++ +++ ++ +++ ++ +++ ++ ++ SE 13 ------SE 14 + +++ + +++ ++ +++ ++ ++ SE 15 +++ +++ ++ +++ ++ ++ ++ ++ SE 16 + +++ + +++ + +++ ++ ++ SE 17 - ++ - ++ - ++ - ++ SE 20 +++ +++ +++ +++ +++ +++ ++ ++ SE 25 + ++ + ++ + ++ + ++ SE 31 + +++ +++ +++ +++ +++ ++ ++ SE 33 +++ +++ ++ +++ ++ +++ ++ ++ SE 53 ++ +++ ++ +++ +++ +++ ++ ++ SE 59 ++ +++ ++ +++ ++ +++ + ++ SE 64 +++ +++ +++ +++ +++ +++ ++ ++ SE 67 + +++ + +++ + +++ + ++ SE 69 ++ ++ ++ +++ ++ +++ ++ ++ SE 70 + +++ + +++ +++ +++ ++ ++ SE 71 +++ +++ +++ +++ +++ +++ +++ +++ SE 73 +++ +++ +++ +++ +++ +++ ++ ++ rSE 6 ------SE C6B ------SE RPTs (I) + ++ ++ ++ ++ +++ ++ +++ SE G7/All + +++ + +++ + +++ + ++ SE FST - + - + + + + + SE UNSW 031900 - + - + + + + +

Other serotypes

S. Agona 1 ++ +++ ++ +++ ++ +++ ++ ++ S. Amsterdam ++ +++ ++ +++ ++ +++ ++ ++ S. Anatum 1 ++ +++ ++ +++ ++ +++ ++ ++ S. Anatum 2 ++ ++ ++ ++ ++ ++ ++ ++ S. Blockey ++ ++ +++ +++ +++ +++ ++ ++ S. Bredeney +++ +++ ++ +++ ++ +++ ++ ++ S. Dublin +++ +++ +++ +++ +++ +++ ++ ++ S. Hadar 2 +++ +++ +++ +++ ++ +++ ++ ++ S. Infantis 1 ++ ++ ++ +++ ++ ++ ++ ++ S. Infantis 2 ++ ++ ++ +++ ++ ++ ++ ++ S. Mbandaka ++ ++ ++ ++ ++ ++ ++ ++ S. Montevideo 1 ++ ++ ++ +++ ++ ++ ++ ++ S. Montevideo 2 ++ ++ ++ +++ ++ ++ ++ ++ S. Muenster ++ +++ +++ +++ +++ +++ ++ ++ S. Newport 1 +++ +++ +++ +++ +++ +++ ++ ++ S. Oranienburg ++ ++ ++ +++ +++ +++ ++ ++ S. Saint Paul 1 + +++ ++ +++ ++ +++ ++ +++ S. Saint Paul 2 + ++ ++ +++ ++ +++ ++ +++ S. Schwarzengrund 1 +++ +++ +++ +++ +++ +++ ++ ++ S. Schwarzengrund 2 +++ +++ +++ +++ +++ +++ ++ ++ S. Senftenberg 1 + +++ + +++ ++ +++ ++ +++ S. Senftenberg 2 ++ +++ +++ +++ +++ +++ ++ ++ S. Sofia 76 ++ ++ ++ +++ +++ +++ ++ ++ S. Sofia 89 ------S. Sofia sswt - +++ - +++ ++ +++ ++ ++ S. Typhimurium + +++ + ++ - - - - S. Typhimurium 1

+, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow. 236 * NV = Novobiocin. ** NP4 = Niaproof 4. Table 6.14 Evaluation of the selectivity of the original formula and the modified formulae with novobiocin or Niaproof 4 of propylene glycol-based medium after inoculation with strains of atypical biotype of Salmonella and incubation at 43°C for

18h and 48h

Acidification of propylene glycol in the medium with and without modification Culture The original The modified formula The modified formula The modified formula formula with 40 mg/L of NV* with 1 mL/L of NP4** with 2 mL/L of NP4** 18h 48h 18h 48h 18h 48h 18h 48h Salmonella subsp 2 - ++ - +++ +++ +++ ++ ++ ser 1,9,12:g,z62:- SRL 02-803419 Salmonella Derby + +++ + +++ +++ +++ ++ ++ IMVS 0878 Salmonella Choleraesuis ------IMVS 1259/92 Salmonella Choleraesuis ------var Kunzendorf IMVS 1269 Salmonella Choleraesuis wk + wk + wk + wk + wk + wk + wk + wk + var Australia IMVS 1594/84 Salmonella subsp 3b ------ser 61:i:z53 IMVS 1679

Wk +, +, ++ and +++ represent different degree of colour changes of propylene glycol-based media: uninoculated control = purplish blue; wk + = from purplish blue to bluish green; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow.

6.3.6 Evaluation of propylene glycol acidification characteristic of atypical

biotypes of Salmonella

The following Sections will assess if these six strains of atypical biotype of Salmonella could acidify propylene glycol in the medium. Hence, media with and without selective agents, MgCl2.6H2O and brilliant green, were inoculated with these strains and incubated at 43°C and the decreased incubation temperature, 37°C. Propylene glycol-

237 based medium with and without selective agents would be termed ‘selective medium’ and ‘non-selective medium’, respectively.

Both selective and non-selective propylene glycol-based media were inoculated to yield

103-107 CFU/mL of six strains of atypical biotype of Salmonella and incubated at 37°C and 43°C for 48h.

Results of the evaluation of propylene glycol acidification characteristic of each strain of atypical biotype of Salmonella are shown in Table 6.15 and discussed in the following Sections.

Salmonella subsp 2 ser 1,9,12:g,z62:- SRL 02-803419 and S. Derby IMVS 0878

Regardless of the presence of selective agents and incubation temperatures, S. subsp 2 ser 1,9,12:g,z62:- SRL 04-803419 and S. Derby IMVS 0878 at all populations were able to acidify propylene glycol.

Salmonella Choleraesuis IMVS 1259/92

Regardless of the presence of selective agents and incubation temperatures, high level

(106 CFU/mL) of S. Choleraesuis IMVS 1259/92 was able to acidify propylene glycol.

However, the decrease of incubation temperature enhanced acidification of propylene glycol by lower populations (103-105 CFU/mL) of this strain.

Salmonella Choleraesuis var Kunzendorf IMVS 1269

It was concluded that S. Choleraesuis var Kunzendorf IMVS 1269 was unable to acidify propylene glycol, as colour change of the non-selective medium inoculated with this strain was not observed at both incubation temperatures.

238 Salmonella Choleraesuis var Australia IMVS 1594/84

Salmonella Choleraesuis var Australia IMVS 1594/84 exhibited a weak positive reaction to propylene glycol acidification, as indicated by slight colour changes of non- selective media incubated at both temperatures. As a consequence, this strain was greatly inhibited by selective agents formulated in propylene glycol-based medium.

Salmonella subsp 3b ser 61:i:z53 IMVS 1679

Salmonella subsp 3b ser 61:i:z53 IMVS 1679 was able to acidify propylene glycol, as the high degree of colour changes of selective media incubated at both temperatures was observed. The decrease of incubation temperature slightly enhanced acidification of propylene glycol of this strain inoculated into the selective medium.

239 Table 6.15 Evaluation of propylene glycol acidification characteristic of atypical biotype of Salmonella using selective and non-selective formulations of propylene glycol-based medium and incubating at 37°C and 43°C for 24h and 48h

Acidification of propylene glycol in selective and non-selective propylene glycol-based media Culture Selective medium* Selective medium* Non-selective medium** Non-selective medium** 43°C 37°C 43°C 37°C 24h 48h 24h 48h 24h 48h 24h 48h S. subsp 2 ser 1,9,12:g,z62:- SRL 02-803419

107 CFU/mL +++ +++ +++ +++ + + + + 106 CFU/mL +++ +++ +++ +++ + + + + 105 CFU/mL ++ ++ ++ ++ + + + + 104 CFU/mL ++ ++ ++ ++ + + + +

S. Derby IMVS 0878

106 CFU/mL +++ +++ +++ +++ + + + + 105 CFU/mL +++ +++ +++ +++ + + + + 104 CFU/mL ++ ++ ++ ++ + + + + 103 CFU/mL ++ ++ ++ ++ + + + +

S. Choleraesuis IMVS 1259/92

106 CFU/mL ++ ++ +++ +++ + + + + 105 CFU/mL - - + +++ wk + wk + wk + wk + 104 CFU/mL - - + ++ wk + wk + wk + wk + 103 CFU/mL - - + ++ wk + wk + wk + wk +

S. Choleraesuis var Kunzendorf IMVS 1269

107 CFU/mL ------106 CFU/mL ------105 CFU/mL ------104 CFU/mL ------

S. Choleraesuis var Australia IMVS 1594/84

107 CFU/mL - - - - wk + wk + wk + wk + 106 CFU/mL - - - - wk + wk + wk + wk + 105 CFU/mL - - - - wk + wk + wk + wk + 104 CFU/mL - - - - wk + wk + wk + wk +

S. subsp 3b ser 61:i:z53 IMVS 1679

106 CFU/mL ++ +++ ++ +++ wk + wk + wk + wk + 105 CFU/mL + +++ ++ +++ wk + wk + wk + wk + 104 CFU/mL + +++ ++ +++ wk + wk + wk + wk + 103 CFU/mL + +++ ++ +++ - - wk + wk +

* +, ++ and +++ represent different degree of colour changes of selective propylene glycol- based media: uninoculated control = purplish blue; + = from purplish blue to dark green or green; ++ = from purplish blue to yellowish green; and +++ = from purplish blue to yellow. ** Wk + and + represent different degree of colour changes of non-selective propylene glycol-based media: uninoculated control = purple; wk + = light brown; and + = yellow. 240 It was concluded S. subsp 2 ser 1,9,12:g,z62:- SRL 02-803419, S. Derby IMVS 0878, S.

Choleraesuis IMVS 1259/92 and S. subsp 3b ser 61:i:z53 IMVS 1679 were able to acidify propylene glycol, whilst S. Choleraesuis var Australia IMVS 1594/84 characterised a weak positive reaction to propylene glycol acidification. Salmonella

Choleraesuis var Kunzendorf IMVS 1269 was the only strain that was unable to acidify propylene glycol. This indicated that propylene glycol-based selective medium was effective for detection of atypical biotype of Salmonella and the decrease of the incubation temperature from 43°C to 37°C facilitated propylene glycol acidification by these strains of Salmonella.

6.4 DISCUSSION

Salmonella are one of the most common pathogens that cause food poisoning throughout the world. Since natural habitats of Salmonella are fresh and perishable foods, it will be beneficial to obtain a rapid result of Salmonella detection to assist decisions on compliance to specifications and release of food products to the markets.

The standard method for Salmonella detection require four to seven days to obtain results (AS 5013.10-2004). Hence, the aim of this Experiment was to develop a new medium for use in the MicroFoss assay to rapidly detect Salmonella in foods and environmental samples.

Previous studies indicated good detection and differentiation of Salmonella from non- salmonellae by the MicroFoss system, based on L-lysine decarboxylation (Shelef et al.,

1998) and H2S production (Shelef and Tan, 1998; Tan and Shelef, 1999; Peng and

Shelef, 2001) characteristics of typical Salmonella. These characteristics of Salmonella have been used as the differential agents for formulating commercial plating media, such as XLD (Taylor, 1965), HE (King and Metzger, 1968) and LMG (Cox, 1993)

241 media. Thus, trials conducted were based on combined features of differential properties of XLD and LMG media in the presence of selective agents for inhibiting growth of interfering bacteria, whilst their concentrations allowed growth of Salmonella without any adverse effect (Moats, 1981). The efficiency of selective agents used in the trial

Experiments, such as malachite green oxalate (Rappaport et al., 1956); bovine bile salt

(Miller and Banwart, 1965; Moats and Kinner, 1974; Busse, 1995); brilliant green

(Miller and Banwart, 1965; Moats and Kinner, 1974; Busse, 1995); MgCl2.6H2O

(Rappaport et al., 1956; Busse, 1995); and sulphamandelate (Moats and Kinner, 1974 &

1976; Watson and Walker, 1978; Jones et al., 1984; McGibbon et al., 1984; Stallard and

Cox, 1994), were previously evaluated and reviewed. Several carbon sources were formulated to enhance excessive acidification of non-salmonellae to consequently prevent lysine decarboxylation by organisms that exhibited this characteristic. Several commercial media for Salmonella detection are based on excessive acidification of carbon sources by non-salmonellae as a differential property (Taylor, 1965; King and

Metzger, 1968; Moats and Kinner, 1976; Cox, 1993; Miller and Mallinson, 2000).

Reactions produced by Salmonella and non-Salmonella cultures were indicated by phenol red.

The 1:100 ratio of volumes of testing portion to the medium (0.1 mL and 0.05 mL of an aliquot inoculated into 10 mL of medium in test tubes and 5 mL of medium in vials, respectively) in these Experiments was followed that inoculated into the RV broth (AS

5013.10-2004). The use of 1:100 or 1:200 ratio of volumes of the inoculum to RV media was suggested by several authors (Vassiliadis, 1983; Vassiliadis et al., 1985;

Fricker, 1984 & 1985).

242 While the representative strain of Salmonella, S. Enteritidis UNSW 031900, did not produce hydrogen sulfide in the formulated liquid medium based on combined features of XLD and LMG agars in the presence of sodium desoxycholate as a selective agent,

C. freundii UNSW 047800 and Pr. mirabilis were able to produce H2S in this medium.

The additional selective agents, MgCl2.6H2O and malachite green oxalate, incorporated into this medium did not only inhibit C. fruendii, but this also inhibited SE UNSW

031900. However, Pr. mirabilis continued producing H2S in this medium. Citrobacter spp. and/or Proteus spp. often caused false positive results in commercially available media based on H2S production, such as HE agar (Taylor and Schelhart, 1971; Hoben et al., 1973b; Restaino et al., 1977; Moats, 1978; Komatsu and Restaino, 1981; Perez et al., 2003), XLD agar (Taylor and Schelhart, 1971; Restaino et al., 1977; Moats, 1978;

Komatsu and Restaino, 1981).

While Salmonella and non-Salmonella cultures were primarily differentiated by different degree of colour changes as a consequence of acid produced from fermentation of carbon sources, lysine decarboxylation of Salmonella was not observed.

It was concluded that liquid media formulated based on the combined selective and differential principles of XLD and LMG media in the presence of selective agents were not effective for Salmonella detection, as selectivity of these media either inhibited

Salmonella growth, or interfered with their differential properties. As a result, a new concept of detection was investigated in the following Section.

A new liquid medium formulated for an effective detection of Salmonella was based on a novel phenotypic characteristic of Salmonella to form acid from utilisation of propylene glycol in the presence of selective agents, MgCl2.6H2O and brilliant green.

Significance of incorporating precise concentration of MgCl2 in media, such as RV

243 broth, to achieve optimum selectivity and allow growth of Salmonella was described by van Schothorst et al. (1987), Peterz et al. (1989) and Maijala et al. (1992). This could be achieved by incorporating these selective agents into propylene glycol-based medium as working solutions, as shown in Appendix 10.1. Selectivity of propylene glycol-based medium was also enhanced by incubation at 42°C. Incubation of MgCl2.6H2O- containing medium, such as RV medium, was recommended at 42°C (Truscott and

Lammerding, 1987; Peterz et al., 1989; AS 5013.10-2004). D’Aoust et al. (1992) stated that greater selectivity of RV broth could be achieved when incubated this medium at an elevated temperatures (42°C-43°C) than at lower temperatures (35°C-37°C).

Results from evaluation of propylene glycol-based medium in test tubes showed distinct colour changes of this medium produced by Salmonella, and the degree of colour changes produced the MicroFoss detection curves with the optimum curve amplitude.

On the other hand, all potential non-Salmonella cultures tested gave negative reaction to this medium. This propylene glycol-based medium could be effective for detection of

Salmonella. The effectiveness of detection of non-Typhi Salmonella using Rambach agar containing propylene glycol was shown in several studies (Rambach, 1990;

Freydiere and Gille, 1991; Gruenewald et al., 1991; Dusch and Altwegg, 1993;

Monnery et al., 1994; Pignato et al., 1995b).

Key components of propylene glycol-based medium, which were incubation temperature, selective agent (MgCl2.6H2O), pH indicator and proteinaceous substrates, were optimised for an effective detection of Salmonella.

As a slight colour change of propylene glycol-based medium that was inoculated with

K. pneumoniae and incubated at 42°C for 24-36h was occasionally observed, incubation of this medium at 43°C was assessed if this could improve selectivity of this medium.

Results showed that all species of non-salmonellae yielded a negative reaction

244 throughout the incubation period at 43°C, whilst all strains of Salmonella Enteritidis tested were able to acidify propylene glycol in this medium. Harvey and Thomson

(1953, cited in Fagerberg and Avens, 1976) reported superior efficiency of selenite broth incubated at 43°C to 37°C for the isolation of salmonellae from human faecal samples. Further investigation among the high incubation temperatures, 42°C, 43°C and

44°C, indicated that 43°C was the optimum temperature for the isolation of salmonellae.

Other studies also reported greater efficiency of incubation at 43°C to 37°C for the isolation of Salmonella, which markedly suppressed competing Gram-negative bacteria, such as coliforms and Proteus, and allowed growth of Salmonella without any adverse effect (Spino, 1966; Carlson et al., 1967; Smyser and Snoeyenbos, 1969; Morris and

Dunn, 1970). Watson and Walker (1978) indicated the improved selectivity without affecting growth of Salmonella using antibiotic-enriched brilliant green agar that was incubated at 43°C, compared to the incubation at 35°C.

Evaluation of the appropriate pH indicator and its concentration showed that bromocresol purple at the concentration of 16 mg/L was the most suitable to exhibit acidification of propylene glycol by Salmonella. Bromocresol purple at this concentration was also used in lysine decarboxylase broth (Taylor modification)

(Oxoid) and peptone glucose yeast broth (Falkow’s medium) as a basal medium for detection of amino acid decarboxylation by salmonellae and the other

Enterobacteriaceae in the MicroFoss system (Shelef et al., 1998).

Evaluation of the appropriate proteinaceous substrates for propylene glycol-based medium indicated that 3 g/L of proteose peptone and 5 g/L of yeast extract were the optimum substrates to enhance acidification of propylene glycol and growth of

Salmonella.

245 A comparison of detection of Salmonella and non-salmonellae using propylene glycol- based media containing 29 g/L and 32 g/L of MgCl2.6H2O indicated that the optimum concentration of MgCl2.6H2O to suppress non-salmonellae and allow growth of

Salmonella without an inhibitory effect was 29 g/L. Growth of low levels of Salmonella

(10-102 CFU/mL) was suppressed in propylene glycol-based medium containing 32 g/L of MgCl2.6H2O, as shown in Table 6.9. van Schothorst et al. (1987) compared growth of Salmonella in the RV medium with various concentrations of MgCl2.6H2O, 29, 32,

34 and 36 g/L. Higher concentrations of MgCl2 in the RV medium inhibited growth of

Salmonella after incubation at 42 + 1°C for 24h, as indicated by the decrease of

Salmonella numbers. For example, the mean log number of Salmonella + standard deviation was 7.5 + 0.72 when enriched in RV medium with 29 g/L of MgCl2, whereas

Salmonella populations were declined to 6.0 + 1.12 in RV medium with 32 g/L of

MgCl2. These findings agreed with those of Peterz et al. (1989) that an increase of concentration of MgCl2.6H2O by 1 g/L resulted in a decrease of log numbers of

Salmonella by almost a half log unit.

In some instances, colour changes of propylene glycol-based medium caused by certain non-salmonellae, such as E. coli, K. pneumoniae and Pr. mirabilis, were similar to those produced by Salmonella. As a consequence, the medium was supplemented with novobiocin or Niaproof 4 to enhance greater selectivity.

Supplementation with novobiocin

Improved selectivity of Salmonella detection and greater recovery of Salmonella using culture media supplemented with novobiocin, in which the optimum concentrations of novobiocin varied with different media, were evaluated in other studies (Boothroyd and

246 Baird-Parker, 1973; Hoben et al., 1973a; Hoben et al., 1973b; Reamer et al., 1974;

Restaino et al., 1977; Moats, 1978; Swaminathan et al., 1978; Rappold and Bolderdijk,

1979; Komatsu and Restaino, 1981; Devenish et al., 1986; Bullock and Frodsham,

1989; Poisson, 1992; Morinigo et al., 1993).

Proteus mirabilis was completely inhibited when the original formula of propylene glycol-based medium was supplemented with ≥10 mg/L of novobiocin. Hoben et al.

(1973b) and Moats and Kinner (1976) reported that the incorporation of 10 mg/L of novobiocin into HE agar completely inhibited the growth of Proteus. This was also supported by Devenish et al. (1986) and Poisson (1992) that the incorporation of 10 mg/L of novobiocin into novobiocin-brilliant green-glucose agar and novobiocin- brilliant green-glycerol-lactose agar, respectively restricted growth of Proteus spp.

Although Escherichia coli was not completely inhibited in media supplemented with 40 and 80 mg/L of novobiocin, as extremely slight colour changes were still observed in these media, the probability of false positive results caused by E. coli was markedly declined at these concentrations. These findings were similar to those of Restaino et al.

(1982) and Restaino et al. (1977) that the incidence of detecting H2S-producing strains of E. coli was decreased when HE agar was supplemented with 40 and 80 mg/L of novobiocin, respectively.

Klebsiella pneumoniae was not inhibited at all concentrations of novobiocin and continued giving a false positive result. McKeon et al. (1995) showed the high frequency of the isolation of antibiotic and multiple-antibiotic (10μg of ampicillin and

30μg of novobiocin) resistant strains of K. pneumoniae from rural, untreated groundwater supplies.

Significant recovery of Salmonella in the presence of high levels of non-salmonellae was achieved when the original formula was supplemented with 40 and 80 mg/L of

247 novobiocin, as shown in Figures 6.12e and 6.12f, respectively. However, some individual strains of Salmonella (103-104 CFU/mL) were partially inhibited in the medium supplemented with 80 mg/L of novobiocin and acidification of propylene glycol was consequently prolonged to be completed within 48h. Hence, the optimum concentration of novobiocin supplemented to the original formula to improve the selectivity of detection without any adverse effect on Salmonella growth was 40 mg/L.

This findings agreed with those of Restaino et al. (1977) and Komatsu and Restaino

(1981) that while the incorporation of 80 mg/L of novobiocin into HE agar did not affect growth and alter colony morphology of selected strains of salmonellae in these studies, the greatest recovery of Salmonella on HE agar was achieved when supplemented with 40 mg/L of novobiocin (Restaino et al., 1982).

Supplementation with Niaproof 4

The optimum concentrations of Niaproof 4 supplemented to the original formula of propylene glycol-based medium were 1 and 2 mL/L. The individual strains of

Salmonella were able to acidify propylene glycol in media with both concentrations of

Niaproof 4. Furthermore, significant recovery of Salmonella in the presence of high levels of non-salmonellae was also achieved at these concentrations. All potential non-

Salmonella cultures were completely inhibited by both concentrations. Niaproof 4 formulated in XLT4 medium was found effective to inhibit Proteus, Providencia, and

Pseudomonas, and provided good differentiation between Salmonella and Citrobacter

(Miller et al., 1991; Miller et al., 1995; Sherrod et al., 1995). XLT4 medium was proven to be effective for the isolation of non-typhoid Salmonella from chicken caeca

(Carli et al., 2001) and stool samples (Dusch and Altwegg, 1995).

248 Examinations of the selectivity of propylene glycol-based media at all formulations were conducted with 12 genera of non-salmonellae in the family Enterobacteriaceae, nine strains of miscellaneous Gram-negative bacteria and five strains of Gram-positive bacteria. It was concluded that the original formula of propylene glycol-based medium generally provided good selectivity. High selectivity using propylene glycol containing medium as Rambach agar was achieved in several studies (Freydiere and Gille, 1991;

Gruenewald et al., 1991). Supplementations of the original formula with novobiocin or

Niaproof 4 improved selectivity of detection by decreasing or eliminating false positive results caused by potential non-salmonellae, such as Ent. cloacae, K. pneumoniae and

Pr. mirabilis. Greatest selectivity of detection was achieved in the presence of 2 mL/L of Niaproof 4.

Examinations of the effectiveness of propylene glycol-based media at all formulations for Salmonella detection were conducted with 26 strains of Salmonella Enteritidis, 27 strains across the other 18 serotypes and six strains of atypical biotype of Salmonella. A majority of strains of S. Enteritidis (88.46%) and other serotypes (92.59%) were able to acidify propylene glycol in all formulations. Among six strains of atypical biotype of

Salmonella, only two strains of these Salmonella, Salmonella subsp 2 ser 1,9,12:g,z62:-

SRL 04-803419 and S. Derby IMVS 0878, were able to acidify propylene glycol in all formulations of the medium. Hence, propylene glycol acidification characteristic of atypical biotype of Salmonella was subsequently evaluated.

Evaluations of the effectiveness of propylene glycol containing medium as Rambach agar for the detection of Salmonella were conducted in several studies (Freydiere and

Gille, 1991; Gruenewald et al., 1991; Monnery et al., 1994; Cooke et al., 1999).

Various serotypes of Salmonella in excess of 82% produced typical Salmonella colonies

249 (red) on this medium, whereas non-salmonellae in excess of 86% produced colonies with different colour (blue, green, beige, or colourless).

Further investigation of propylene glycol acidification characteristic of these six strains of atypical biotype of Salmonella concluded that S. subsp 2 ser 1,9,12:g,z62:- SRL 02-

803419, S. Derby IMVS 0878, S. Choleraesuis IMVS 1259/92 and S. subsp 3b ser

61:i:z53 IMVS 1679 were able to acidify propylene glycol in propylene glycol-based medium containing selective agents. Although S. Choleraesuis IMVS 1259/92 was not inhibited by selective agents in propylene glycol-based medium, this strain was susceptible to high incubation temperature at 43°C. Thus, propylene glycol acidification of this strain was enhanced at 37°C. Salmonella Choleraesuis var Australia IMVS

1594/84 exhibited a weak positive reaction to propylene glycol acidification. As a result, this strain was completely inhibited by selective agents (MgCl2.6H2O and brilliant green) in the original formula of propylene glycol-based medium. Salmonella

Choleraesuis var Kunzendorf IMVS 1269 was the only strain that did not exhibit this characteristic.

Several studies were conducted to examine propylene glycol acidification of Salmonella subspecies I to VI using Rambach agar. Most strains of Salmonella subspecies I, II, IV and VI were able to acidify propylene glycol and produced typical Salmonella colonies on Rambach agar, except for subspecies IIIa, IIIb and V that were positive for β-D- galactosidase and produced blue, blue-green or blue-violet colonies (Kühn et al., 1994;

Pignato et al., 1995a; Poppe et al., 1996).

250 In conclusion, propylene glycol-based medium originally developed in this Chapter could be used for an effective detection of Salmonella, including strains of atypical biotype, by the MicroFoss system. False positive results caused by competing bacteria, such as Ent. cloacae, E. coli, K. pneumoniae and Pr. mirabilis could be decreased or eliminated when the original formulation was supplemented with novobiocin or

Niaproof 4. Selected concentrations of these additional selective agents did not affect growth of the individual strains of Salmonella and enhanced great recovery of

Salmonella in the presence of high levels of non-salmonellae. The effectiveness of detection of Salmonella using all formulations of propylene glycol-based medium will be assessed with naturally contaminated egg samples in Chapter 7.

251 CHAPTER 7

Evaluation of propylene glycol-based medium for detection of

Salmonella in eggs by the MicroFoss system

7.1 INTRODUCTION

Egg contents can be contaminated with Salmonella by either transovarian transmission, which is the infection via the ovary and oviduct of hens during the process of egg formation (Board et al., 1994; Bruce and Drysdale, 1994; Kobayashi et al., 1997), or trans-shell transmission, which is penetration of Salmonella on the shell into the egg contents once they overcome the natural defensive barriers (cuticle and shell membranes) of eggs under improper handling and storage conditions (Bruce and

Drysdale, 1994; Kobayashi et al., 1997; Jay et al., 2003). Contamination of the egg contents is one cause of human salmonellosis in Europe and North and South America, mainly from consumption of raw or undercooked eggs (Schultz et al., 1987; Hopper and

Mawer, 1988; Humphrey, 1990; Rodriguez et al., 1990; Steinert et al., 1990; Duguid and North; 1991; Caffer and Eiguer, 1994; Morse et al., 1994; CDC, 2000; Reporter et al., 2000; Drociuk et al., 2002; Jay et al., 2003). Salmonella Enteritidis is a prevalent serotype that is responsible for human salmonellosis, because of its unusual ability to colonise the ovarian tissue of hens and to be present in the contents of intact shell eggs

(Cogan and Humphrey, 2003). Humphrey (1994) and Food Standards Australia New

Zealand (2006) stated that contamination of egg contents with S. Enteritidis phage type

4 is the result of infection of the reproductive tissue rather than invasion through the shell. According to the study by Humphrey et al. (1991), the initial number of S.

Enteritidis in naturally contaminated eggs was low (10-20 CFU/egg) and there is no connection between shell contamination and the presence of S. Enteritidis in egg 252 contents. Cox (2000) and Jay (2000) reported that numbers of Salmonella cells of the order of 107–108 CFU/g are generally necessary for salmonellosis. However, epidemiological evidence from a number of outbreaks has demonstrated that the infectious dose may be as little as a few cells. Salmonellae should be absent in 25g of properly pasteurized egg products (Food Standards Australia New Zealand, 2006 &

2009). Hence, it is essential for the detection system to achieve as low level as <10

CFU. The ability of several rapid methods to detect <10 CFU of Salmonella in eggs has been reported (Wang et al., 1995; Ueda et al., 2000; Seo et al., 2003; Seo et al., 2004).

This Chapter evaluated the effectiveness of propylene glycol-based selective media (the original formula and three modified formulae with 40 mg/L of novobiocin, or 1 or 2 mL/L of Niaproof 4) developed in Chapter 6 for detection of Salmonella in naturally contaminated egg samples by the MicroFoss method in comparison to the standard cultural method. Sensitivity and specificity of the MicroFoss method will be subsequently assessed.

This Chapter also evaluated detection limit of the MicroFoss method for Salmonella by the MicroFoss assay.

253 7.2 MATERIALS AND METHODS

7.2.1 Materials

Commercial raw egg samples and pasteurized egg products used in this research were described in Section 3.2.1.2.

7.2.2 Sample handling prior to testing

Handling of commercial raw egg samples and pasteurized egg products are as described in Section 3.2.2.2.

7.2.3 Microbiological media, chemicals, reagents and test kits

All microbiological media, chemicals, and reagents used in this Chapter are as shown in

Table 6.1.

7.2.4 Reference culture

The reference culture of Salmonella Enteritidis 1 (SE1) (School of Chemical Sciences and Engineering, UNSW) was used to evaluate the detection limit of the MicroFoss assay. Preparation of SE 1 for analyses and maintenance of the stock culture are described in Section 6.2.2.

7.2.5 Preparation of media

7.2.5.1 Media for the cultural method

Details of preparation of media and reagents used for the cultural method are shown in the following Sections.

• Buffered peptone water (BPW) (Oxoid): BPW was prepared according to Oxoid’s instruction.

254 • Rappaport-Vassiliadis (RV) broth (Oxoid): RV was prepared according to Oxoid’s instruction.

• Manntiol selenite cystine (MSC) broth: MSC broth was prepared according to instructions of Oxoid and Amyl Media Pty Ltd. An amount of 19g of MSC broth base

(Amyl Media Pty Ltd) was added to one litre of distilled water containing 4g of sodium biselenite (Oxoid). Sodium biselenite was separately added to water to minimise risk of teratogenicity. The mixture was thoroughly dissolved and dispensed into test tubes and sterilised by heating in free flowing steam in an autoclave for 10 min.

• Muller-Kauffmann-tetrathionate with novobiocin (MKTTn) broth (Oxoid): MKTTn broth was prepared according to Oxoid’s instruction and AS 5013.10-2004.

• XLD agar (Oxoid): XLD agar was prepared according to Oxoid’s instruction.

• Bismuth sulphite (BS) agar (modified) (Oxoid): BS agar was prepared according to

Oxoid’s instruction.

• NA agar: NA agar was prepared according to Oxoid’s instruction.

7.2.5.2 Media for the MicroFoss assay

The original and three modified formulae of selective media created in Chapter 6 are shown in Table 7.1

255 Table 7.1 Formulae of selective media for detection of Salmonella

Formula (g/L) Composition Original Modified with Modified with Modified with 40 mg/L of 1 mL/L of 2 mL/L of novobiocin Niaproof 4 Niaproof 4 Proteose peptone 3.00 3.00 3.00 3.00 Yeast extract 5.00 5.00 5.00 5.00 Sodium chloride 5.00 5.00 5.00 5.00 a MgCl2.6H2O (40% solution) 100.70 mL 100.70 mL 100.70 mL 100.70 mL Brilliant greenb (0.1% solution) 5.20 mL 5.20 mL 5.20 mL 5.20 mL Bromocresol purple 0.016 0.032 0.048 0.064 Propylene glycol 10.00 10.00 10.00 10.00 Novobiocinc (20 mg/mL solution) - 2.21 mL - - Niaproof 4d - - 1.11 mL 2.21 mL H2O 1000 mL 1000 mL 1000 mL 1000 mL Adjusted pH (6.80 + 0.20) 6.78 6.80 6.78 6.79

a Final concentration of MgCl2.6H2O in all media was 2.90% (w/v). b Final concentration of brilliant green in all media was 0.00047%.

Media were thoroughly mixed and sterilised by boiling for dispensing in the MicroFoss vials. The agar plug was prepared by mixing liquid media with 1% agar (w/v) and boiling to completely dissolve and sterilise. Molten agar media (approximately 0.3 mL) was then aseptically added into sterile vials to precisely fill up the agar plug zone without allowing any bubbles in the contents and on the surface. The agar plug was allowed to dry prior to filling the vial with 9 mL of sterilised liquid media.

7.2.6 Procedure of Salmonella detection

The procedures for simultaneous Salmonella detection in commercial raw egg samples by the cultural and MicroFoss methods are shown in Figure 7.1.

256 Pre-enrichment of 25g of commercial raw egg samples in 225 mL of buffered peptone water

Incubation of samples at 37°C + 1°C for 18h + 2h

Inoculation of pre-enriched samples into selective enrichment broths Inoculation of 0.1 mL of pre-enriched samples 0.1 mL 1 mL 1 mL into 9 mL of media in vials

10 mL of RV broth 10 mL of MSC broth 10 mL of MKTTn broth Testing in the MicroFoss system* Incubation at 41.5°C + 1°C Incubation at 37°C + 1°C Incubation at 37°C + 1°C Incubation at 43°C and monitoring for 48h for 24h + 3h for 24h + 3h for 24h + 3h

XLD** BS** XLD** BS** XLD** BS** Vials giving DTs*** Vials yielding no DT Incubation at 37°C + 1°C Incubation at 37°C + 1°C Incubation at 37°C + 1°C for 24h + 3h for 24h + 3h for 24h + 3h XLD** BS** TSA**

+ve -ve

Selection of presumptive colonies for further biochemical and serological confirmatory tests

Preparation for confirmatory tests by streaking the isolate of presumptive colonies onto NA plates

Incubation at 37°C + 1°C for 24h + 3h

Serological confirmation Biochemical confirmation (Latex agglutination-SerobactTM) Conventional Microbact****

TSI Urea split LDC ONPG Methyl red VP Indole Citrate

Figure 7.1 Procedure of simultaneous Salmonella detection in commercial raw egg samples by the cultural and MicroFoss methods. Adapted from Amaguana and Andrews (2000); Andrews et al. (2001); Jay et al. (2003); AS 5014.10-2004. TSI = Triple sugar iron. LDC = L-Lysine decarboxylation. ONPG = ο-Nitrophenyl β-D-galactopyranoside. VP = Voges-Proskauer. * Test configurations of the MicroFoss assay for Salmonella are as described in Table 6.4. ** Selective plating onto XLD, BS and TSA plates was done in duplicate. *** A loopful of vials giving detection times prior to and after 24h was streaked onto NA plates after incubation for 24h and 48h, respectively. **** Gram stains were performed prior to conducting biochemical testing by the Microbact system, whilst motility and oxidase tests were simultaneously performed with the Microbact.

257 7.2.7 Conventional biochemical testing

A single colony of biomass of selected presumptive colonies from the NA plate was inoculated into 1 mL of 0.1% peptone water to prepare cell suspension for a series of conventional biochemical tests. Preparations of media and procedure for conventional biochemical tests are shown in the following Sections.

• TSI reaction: An aliquot of cell suspension was streaked onto slants and stabbed into butts of TSI medium (Oxoid). TSI slants were incubated at 37°C + 1°C for 24h + 3h

(AS 5013.10-2004).

• Urea split reaction: A loopful of cell suspension was streaked onto surface of urea agar (Christensen) slant (Oxoid). Urea agar slants were incubated at 37°C + 1°C for 24h

+ 3h (AS 5013.10-2004).

• Lysine decarboxylation: A loopful of cell suspension was inoculated into LDC broth and incubated at 37°C + 1°C for 24h + 3h (AS 5013.10-2004).

•β-galactosidase reaction: Testing procedure of β-galactosidase reaction followed

Oxoid’s instruction. An ONPG disc (Oxoid) was placed into sterile test tubes, for which

0.1 mL of sterile 0.88% saline solution was subsequently added. A loopful of cell suspension was inoculated into these test tubes and incubated at 37°C + 1°C for 24h +

3h. Results were checked at an hour interval for up to six hours.

• Methyl red reaction: A loopful of cell suspension was inoculated into MR-VP medium

(Clarks and Lubs medium-buffered glucose broth, Oxoid) and incubated at 37°C + 1°C for 24h + 3h. Volume (5 mL) of the MR-VP broth culture was aseptically transferred to sterile test tubes, to which 5 to 6 drops of methyl red indicator solution were subsequently added. Methyl red indicator solution was prepared by dissolving 0.1g of methyl red (BDH Chemicals Ltd.) in 300 mL of absolute ethanol and diluting to 500

258 mL with sterile distilled water. Methyl red reaction was read immediately (Andrews,

2000b).

• VP reaction: Volume (3 mL) of the MR-VP broth culture that was incubated at 37°C +

1°C for 24h + 3h was aseptically transferred to sterile test tubes, for which two drops of creatine solution, 0.6 mL of α-naphthol solution and 0.2 mL of 40% potassium hydroxide solution (w/v), respectively were added and well shaken after the addition of each reagent. VP reaction was checked within 15 mins (AS 5013.10-2004).

• Indole reaction: A loopful of cell suspension was inoculated into 5 mL of tryptone/tryptophan medium and incubated at 37°C + 1°C for 24h + 3h. After incubation, 1 mL of Kovacs reagent was added to the broth culture (AS 5013.10-2004).

• Citrate utilisation: An aliquot of cell suspension was streaked onto the slant and stabbed into the butt of Simmons Citrate agar slant (Oxoid) and incubated at 35°C for

96 + 2h (Andrews, 2000b; Kornacki and Johnson, 2001).

• Motility test: Motility semi-solid medium was prepared according to Ball and Sellers

(1966). Motility medium was composed of 25g of heart infusion broth, 2 g of K2HPO4,

2g of KNO3, 30g of gelatin and 1g of agar. Contents were thoroughly mixed and dissolved in 1 litre of distilled water and autoclaved at 121°C for 15 min. Volume (10 mL) of 1% filter-sterilised, aqueous solution of 2, 3, 5-Triphenyl-2H-tetrazolium chloride (Aldrich Chemical Company, Inc.) was aseptically added to molten sterilised motility medium and swirled to obtain uniform distribution. Volume (3 mL) was aseptically transferred to sterile test tubes. An aliquot of cell suspension was stabbed into centre of motility medium further down half of the depth of medium. The inoculated medium was incubated at 35°C + 2°C for 24-48h.

• Oxidase test: Oxidase tests were performed on fresh cultures using the method of AS

5013.14-2004.

259 Interpretation of biochemical reactions and reaction of typical Salmonella are shown in

Table 7.2.

Table 7.2 Interpretation of biochemical reactions and reaction of typical Salmonella

Biochemical test Positive reaction Negative reaction Reactions of Salmonella TSI reaction-glucose utilisation Yellow butt Red or unchanged butt +ve TSI reaction-lactose and/or Yellow slant Red or unchanged slant -vea sucrose utilisation b TSI reaction-H2S & Blackening of medium No blackening of medium +ve bubble production with gas formation or cracks in medium Urea split-urease reaction Bright pink or Bright yellow medium -ve pinkish red medium LDC reaction Turbid and purple broth Yellow broth +vec β-galactosidase reaction Yellow medium Clear, unchanged medium -ved Methyl red reaction Diffuse red medium Diffuse yellow medium +ve VP reaction Pink to Unchanged medium -ve pinkish red medium Indole formation Cherry red ring on surface Yellow to yellowish brown -ve ring on surface Citrate utilisation Growth and No growth to trace growth +vee blue medium on the slant and no colour change and throughout the medium of medium

+ve = Positive reaction, which represents 90-100% positive biochemical reaction. -ve = Negative reaction, which represents 0-10% positive biochemical reaction. a Salmonellae usually do not utilise lactose and/or sucrose, except for Salmonella subsp. III, Salmonella Arizona, which yield weak positive reaction for lactose utilisation. b Most salmonellae produce H2S, except for Salmonella choleraesuis and Salmonella Paratyphi A, which yield weak positive and negative reactions, respectively. Concurrently, most salmonellae produce gas from glucose utilisation, except for Salmonella Gallinarum and Salmonella Typhi. c Most salmonellae decarboxylate L-lysine, except Salmonella Paratyphi A. d Salmonellae usually do not possess β-galactosidase reaction, except for Salmonella subsp. II and III, which yield weak positive and positive reactions, respectively. e = Salmonella subsp. I to IV utilise citrate, whereas Salmonella Choleraesuis yield a very weak positive reaction. Salmonella Gallinarum, Salmonella Paratyphi A, Salmonella Pullorum and Salmonella Typhi yield a negative reaction to citrate utilisation.

Adapted from Brenner (1984); Andrews; (2000b); Jay et al. (2003); AS 5013.10-2004.

260 7.2.8 Rapid biochemical and serological testing

The Microbact system (Medvet Diagnostics) is a standardised micro-substrate system designed to simulate conventional biochemical substrates used for the identification of

Enterobacteriaceae and common miscellaneous Gram-negative bacilli (MGNB). The

Microbact test kit consists of two separate strips, 12A and 12B. Each strip is comprised of 12 different biochemical testings. The 12A strip alone could be used for identification of oxidase negative, nitrate positive glucose fermenters, which represent

15 genera in the family Enterobacteriaceae. Concurrently, the 12B strip in conjunction with the 12A strip could be used for identification of oxidase positive, nitrate negative glucose non-fermenters or MGNB, as well as the Enterobacteriaceae. Three consecutive biochemical reactions are grouped, in which positive results of the first, second and third reactions in each group are equivalent to the values of 4, 2 and 1, respectively. On the other hand, negative result yields the value of 0. A summary of values of each group forms a single digit. Hence, results of a series of 24 biochemical reactions of the 12A and 12B strips (24E) produce eight digits or an octal code. Interpretation of an octal code for the identification of the isolate is performed by the Microbact computer aided identification software provided with the Microbact product. Results are shown as the probability of microorganisms (%). Details of testing are as described in the manufacturer’s booklet (Medvet Diasnostics, Thebarton, Australia).

Serological confirmation of presumptive results was performed by a commercial latex agglutination kit, SerobactTM (Oxoid). Details of principles, testing instruction and interpretation of results are described in product leaflet (Oxoid). Latex particles are coated with polyvalent antisera against an extensive range of Salmonella flagellar antigens. A suspension of suspicious colonies is made by thorough mixing a single colony in 0.85% isotonic saline. When a suspension of suspicious colonies possessing

261 these antigens are blended with polyvalent antisera, the latex particles rapidly agglutinate to form visible white clumps. Negative result remains unchanged. Details of testing are as described in Oxoid’s instruction.

7.2.9 Detection limit of the MicroFoss assay for Salmonella detection

Portions (25g) of pasteurized whole egg products from three batches were separately inoculated with low levels of SE 1 (1 and 10 CFU), as the individuals and in a culture pool with 103 CFU each of Citrobacter freundii, Enterobacter aerogenes, Enterobacter cloacae, Escherichia coli and Klebsiella pneumoniae, for evaluation of detection limit of the MicroFoss assay using the original formula of selective medium. Procedure of the evaluation of detection limit of the MicroFoss assay is shown in Figure 7.2.

262 Inoculation of 25g of pasteurized whole egg products* to yield low levels of SE 1

Uninoculated SE 1 SE 1 SE 1 SE 1 control 1 CFU 1 CFU + 103 of five coliforms 10 CFU 10 CFU + 103 of five coliforms

Pre-enrichment of the inoculated pasteurized whole egg products in 225 mL of buffered peptone water.

Incubation at 37°C + 1°C for 18h + 2h.

Inoculation of 0.1 mL of pre-enriched samples into 9 mL of selective medium in vials.

Testing in the MicroFoss system at 43°C for 48h.

Figure 7.2 Procedure for the evaluation of detection limit of the MicroFoss assay for

Salmonella.

* Detection limit of the MicroFoss assay for Salmonella were conducted using three batches of pasteurized whole egg products.

263 7.3 RESULTS

7.3.1 Detection of Salmonella in naturally contaminated eggs

Detection of Salmonella in naturally contaminated eggs using the original formula and three modified formulae of selective media by the MicroFoss system were simultaneously performed with the cultural method. Naturally contaminated egg samples used were one sample each of commercial raw whole egg and raw albumen.

Results for detection of each sample by both methods are shown in the following

Sections.

7.3.1.1 Commercial raw whole egg

• Results of the cultural method

Detection of Salmonella in commercial raw whole egg by the cultural method gave colonies exhibiting typical morphology of Salmonella on XLD and BS plates from all selective enrichment broths. It was noted that proportions of colonies characterising typical morphology of Salmonella to non-salmonellae on XLD plates were sequentially increased from samples selectively enriched in MSC, MKTTn and RV broths (data not shown). These presumptive colonies were selected for further biochemical

(conventional and the Microbact) and serological tests, which were termed as codes representing media used in each step of detection. Confirmation and identification of these presumptive results are shown in Table 7.3.

264 Table 7.3 Confirmation and identification of presumptive colonies obtained from detection of Salmonella in commercial raw whole egg by the traditional method

Presumptive result Results of a series of Identification by the Microbact Serological conventional confirmation* biochemical tests indicating presence of Salmonella* BPW-RV-XLD I + 96.79%-Salmonella subsp. 1 + BPW-RV-XLD II + 96.79%-Salmonella subsp. 1 + BPW-RV-BS I + 61.27%-Salmonella Pullorum + 37.14%-Salmonella subsp. 1 BPW-RV-BS II + 61.27%-Salmonella Pullorum + 37.14%-Salmonella subsp. 1 BPW-MSC-XLD I + 96.79%-Salmonella subsp. 1 + BPW-MSC-XLD II + 96.79%-Salmonella subsp. 1 + BPW-MSC-BS I - 85.89%-Citrobacter braakii - 7.50%-Citrobacter youngae 4.21%-Salmonella subsp. 3B 1.76%-Escherichia coli BPW-MSC-BS II - 85.89%-Citrobacter braakii - 7.50%-Citrobacter youngae 4.21%-Salmonella subsp. 3B 1.76%-Escherichia coli BPW-MKTTn-XLD I + 75.28%-Salmonella subsp. 3B + 15.34%-Salmonella Arizonae subsp. 3A 6.44%-Salmonella subsp. 1 2.20%-Salmonella subsp. 5 BPW-MKTTn-XLD II + 75.28%-Salmonella subsp. 3B + 15.34%-Salmonella Arizonae subsp. 3A 6.44%-Salmonella subsp. 1 2.20%-Salmonella subsp. 5 BPW-MKTTn-BS I - 55.73%-Citrobacter freundii - 32.58%-Salmonella subsp. 3B 5.46%-Citrobacter youngae 4.95%-Citrobacter braakii BPW-MKTTn-BS II - 76.67%-Enterobacter cloacae - 13.76%-Enterobacter biogp 1 6.30%-Enterobacter aerogenes 3.13%-Enterobacter intermedium

* Results of a series of conventional biochemical tests and serological test indicate presence of Salmonella, + = Positive reaction for typical Salmonella; - = Negative reaction for typical Salmonella. BPW = Buffered peptone water; RV = Rappaport-Vassiliadis medium; MSC = Mannitol selenite cystine broth; MKTTn = Muller-Kauffmann-tetrathionate with novobiocin broth; XLD = xylose lysine desoxycholate agar; BS = Bismuth sulphite agar.

265 Results of biochemical (conventional and the Microbact) and serological confirmatory tests of Salmonella detection in commercial raw whole egg by the cultural method corresponded, in which eight out of 12 presumptive results were confirmed Salmonella

(Table 7.3). Selective enrichment in RV broth yielded greater selectivity than results obtained from MSC and MKTTn broths, as presumptive results of sample selectively enriched in RV broth were entirely confirmed as Salmonella (Table 7.3). False positive results of sample selectively enriched in MSC and MKTTn broths were observed when selective plating on BS plates, which were identified as Citrobacter braakii, C. freundii and Enterobacter cloacae. These findings indicated that Salmonella was present in commercial raw whole egg, as determined by the cultural method.

••• Results of the MicroFoss method

An aliquot (0.1 mL) of raw whole egg at 10-1 dilution (25g in 225 mL of BPW) prior to pre-enrichment and after pre-enrichment at 37°C + 1°C for 18h + 2h was separately inoculated into 9 mL of selective media from all formulae (the original and three modified formulae) and analysed in the MicroFoss system at 43°C for 48h. The

MicroFoss detection of samples with and without pre-enrichment were compared.

Figures 7.3, 7.4, 7.5 and 7.6 show the MicroFoss curves of Salmonella detection in samples of raw whole egg, with and without pre-enrichment in BPW at 37°C + 1°C for

18h + 2h, using the original formula, the modified formula with 40 mg/L of novobiocin, and the modified formulae with 1 and 2 mL/L of Niaproof 4, respectively.

266 The original formula: Sample (in duplicate) containing presumptive Salmonella without pre-enrichment gave DTs of 41.4h and 41.5h with the optical units in the range of 200-480, whilst pre-enriched sample (in duplicate) containing presumptive

Salmonella yielded DTs of 3.3h and 2.3h with the optical units in the range of 200-650

Figure 7.3).

The modified formula with 40 mg/L of novobiocin: Sample (in duplicate) containing presumptive Salmonella without pre-enrichment gave DTs of 16.7h and 16.5h, whilst pre-enriched sample (in duplicate) containing presumptive Salmonella yielded DTs of

3.3h and 3.5h with the optical units of both curves in the range of 200-500 (Figure 7.4).

The modified formula with 1 mL/L of Niaproof 4: Sample (in duplicate) containing presumptive Salmonella without pre-enrichment gave DTs of 15.3h and 15.5h, whilst pre-enriched sample (in duplicate) containing presumptive Salmonella yielded DTs of

3.6h and 3.2h with the optical units of both curves in the range of 200-550 (Figure 7.5).

The modified formula with 2 mL/L of Niaproof 4: Sample (in duplicate) containing presumptive Salmonella without pre-enrichment gave DTs of 24.3h and 20.9h, whilst pre-enriched sample (in duplicate) containing presumptive Salmonella yielded DTs of

5.5h and 5.6h with the optical units of both curves in the range of 200-500 (Figure 7.6).

267 Figure 7.3 The MicroFoss curves of detection of Salmonella in commercial raw whole egg using the original formula, uninoculated control (No detection); duplicate of sample analysed prior to pre-enrichment (DT 41.4h & 41.5h); and duplicate of sample analysed after pre-enrichment (3.3h and 2.3h).

Figure 7.4 The MicroFoss curves of detection of Salmonella in commercial raw whole egg using the modified formula with 40 mg/L of novobiocin, uninoculated control (No detection); duplicate of sample analysed prior to pre-enrichment (DT 16.7h & 16.5h); and duplicate of sample analysed after pre-enrichment (3.3h and 3.5h).

268 Figure 7.5 The MicroFoss curves of detection of Salmonella in commercial raw whole egg using the modified formula with 1 mL/L of Niaproof 4, uninoculated control (No detection); duplicate of sample analysed prior to pre-enrichment (DT 15.3h & 15.5h); and duplicate of sample analysed after pre-enrichment (3.6h and 3.2h).

Figure 7.6 The MicroFoss curves of detection of Salmonella in commercial raw whole egg using the modified formula with 2 mL/L of Niaproof 4, uninoculated control (No detection); duplicate of sample analysed prior to pre-enrichment (DT 24.3h & 20.9h); and duplicate of sample analysed after pre-enrichment (5.5h and 5.6h).

269 Late detections of presumptive Salmonella in non-pre-enriched sample by the

MicroFoss assay were observed, markedly in sample tested in the original formula, as shown in Figure 7.3. It clearly showed that pre-enrichment was essential to resuscitate injured Salmonella cells in frozen commercial raw egg samples and proliferate these injured cells to detectable levels for detection in selective media, as the injury increases cell susceptibility to selective agents in these media (Amaguana and Andrews, 2000;

Jay, 2003). As a result, detection of presumptive Salmonella in samples pre-enriched in

BPW yielded DTs within 2-6h. Speed of detection of Salmonella in pre-enriched sample using either the original formula or modified formulae was not different, except for sample tested in the modified formula with 2 mL/L of Niaproof 4. Curve amplitudes were in the optimum ranges, which were similar to those of detection of the reference cultures of Salmonella in these media, as shown in Chapter 6.

All vials giving detection times that were streaked onto XLD and BS plates showed colonies exhibiting typical morphology of Salmonella on these plates. These presumptive results were selected and prepared for biochemical (conventional and the

Microbact) and serological tests for confirmation and identification, as shown in Table

7.4.

270 Table 7.4 Confirmation and identification of presumptive results obtained from detection of Salmonella in commercial raw whole egg by the MicroFoss assay

Medium DT Presumptive result Results of a series Identification by Serological (h) of conventional the Microbact confirmation* biochemical tests indicating presence of Salmonella* The original 41.4 Non-PE-Org-XLD I + 49.44%-S. subsp. 1 + formula 38.52%-S. subsp. 4 41.5 Non-PE-Org-XLD II + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 3.3 BPW-Org-XLD I + 61.27%-S. Pullorum + 37.14%-S. subsp. 1 BPW-Org-BS I + 96.79%-S. subsp. 1 + 2.3 BPW-Org-XLD II + 96.79%-S. subsp. 1 + The modified 16.7 Non-PE-NV 40-XLD I + 78.26%-S. subsp. 3B + formula with Non-PE-NV 40-BS I + 84.51%-S. subsp. 1 + 40 mg/L of 16.5 Non-PE-NV 40-XLD II + 96.79%-S. subsp. 1 + novobiocin Non-PE-NV 40-BS II + 96.79%-S. subsp. 1 + 3.3 BPW-NV 40-XLD I + 61.27%-S. Pullorum + 37.14%-S. subsp. 1 3.5 BPW-NV 40-XLD II + 96.79%-S. subsp. 1 + BPW-NV 40-BS II + 96.79%-S. subsp. 1 + The modified 15.3 Non-PE-NP4 1-XLD I + 61.27%-S. Pullorum + formula with 37.14%-S. subsp. 1 1 mL/L of Non-PE-NP4 1-BS Ia + 61.27%-S. Pullorum + Niaproof 4 37.14%-S. subsp. 1 Non-PE-NP4 1-BS Ib + 96.79%-S. subsp. 1 + 15.5 Non-PE-NP4 1-XLD II + 96.79%-S. subsp. 1 + Non-PE-NP4 1-BS IIa + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 Non-PE-NP4 1-BS IIb + 96.79%-S. subsp. 1 + 3.6 BPW-NP4 1-XLD I + 93.25%-S. subsp. 3B + BPW-NP4 1-BS Ia + 73.54%-S. subsp. 1 + 14.59%-S. subsp. 5 BPW-NP4 1-BS Ib + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 3.2 BPW-NP4 1-XLD II + 96.79%-S. subsp. 1 + BPW-NP4 1-BS IIa + 96.79%-S. subsp. 1 + BPW-NP4 1-BS IIb + 96.79%-S. subsp. 1 + The modified 24.3 Non-PE-NP4 2-XLD I + 96.79%-S. subsp. 1 + formula with Non-PE-NP4 2-BS Ia + 96.79%-S. subsp. 1 + 2 mL/L of Non-PE-NP4 2-BS Ib + 96.79%-S. subsp. 1 + Niaproof 4 20.9 Non-PE-NP4 2-XLD II + 96.79%-S. subsp. 1 + Non-PE-NP4 2-BS IIa + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 Non-PE-NP4 2-BS IIb + 96.79%-S. subsp. 1 + 5.5 BPW-NP4 2-XLD I + 96.79%-S. subsp. 1 + BPW-NP4 2-BS Ia + 99.36%-S. subsp. 1 + BPW-NP4 2-BS Ib + 96.79%-S. subsp. 1 + 5.6 BPW-NP4 2-XLD II + 96.79%-S. subsp. 1 + BPW-NP4 2-BS IIa + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 BPW-NP4 2-BS IIb + 96.79%-S. subsp. 1 +

* Results of a series of conventional biochemical tests and serological test indicate presence of Salmonella, + = Positive reaction for typical Salmonella; - = Negative reaction for typical Salmonella. BPW = Buffered peptone water; XLD = xylose lysine desoxycholate agar; BS = Bismuth sulphite agar. Non-PE = Non-pre-enriched sample; Org = The original formula; NV 4 = The modified formula with 40 mg/L of novobiocin; NP4 1 = The modified formula with 1 mL/L of Niaproof 4; and NP4 2 = The modified formula with 2 mL/L of Niaproof 4.

271 Results of biochemical (conventional and the Microbact) and serological tests of presumptive results of Salmonella well corresponded for both of samples with and without pre-enrichement, in which all presumptive results were identified as

Salmonella. This indicated that Salmonella was present in raw whole egg sample, as determined by the MicroFoss assay. Although Salmonella in samples without pre- enrichment were remarkably detected by the MicroFoss assay, pre-enrichment greatly enhanced an effective detection of Salmonella, as previously described. These findings therefore showed good sensitivity and specificity of Salmonella detection by the

MicroFoss assay.

Results of detection of Salmonella in raw whole egg by the MicroFoss assay agreed with those obtained from the cultural method and indicated presence of Salmonella in this sample.

272 7.3.1.2 Commercial raw albumen

••• Result of the cultural method

Detection of Salmonella in commercial raw albumen by the cultural method gave colonies exhibiting typical morphology of Salmonella on XLD and BS plates from all selective enrichment broths. A majority of presumptive results (14 out of 16) were biochemically and serologically confirmed as Salmonella (Table 7.5).

Table 7.5 Confirmation and identification of presumptive colonies obtained from detection of Salmonella in commercial raw albumen by the traditional method

Presumptive result Results of a series of Identification by the Microbact Serological conventional confirmation* biochemical tests indicating presence of Salmonella* BPW-RV-XLD I + 98.83%-Salmonella subsp. 1 + BPW-RV-BS Ia + 99.84%-Salmonella subsp. 1 + BPW-RV-BS Ib - 98.58%-Enterobacter aerogenes - BPW-RV-BS Ic - 77.74%-Enterobacter cloacae - 22.10%-Enterobacter aerogenes BPW-RV-XLD II + 49.44%-Salmonella subsp. 1 + 38.52%-Salmonella subsp. 4 BPW-RV-BS IIa + 96.79%-Salmonella subsp. 1 + BPW-RV-BS IIb + 96.79%-Salmonella subsp. 1 + BPW-RV-BS IIc + 49.44%-Salmonella subsp. 1 + 38.52%-Salmonella subsp. 4 BPW-MSC-XLD I + 99.99%-Salmonella subsp. 1 + BPW-MSC-XLD II + 98.83%-Salmonella subsp. 1 + BPW-MKTTn-XLD I + 99.94%-Salmonella subsp. 1 + BPW-MKTTn-BS Ia + 98.98%-Salmonella subsp. 1 + BPW-MKTTn-BS Ib + 52.22%-Salmonella subsp. 1 + 35.05%-Salmonella subsp. 3B BPW-MKTTn-XLD II + 99.81%-Salmonella subsp. 1 + BPW-MKTTn-BS IIa + 99.94%-Salmonella subsp. 1 + BPW-MKTTn-BS IIb + 98.83%-Salmonella subsp. 1 +

* Results of a series of conventional biochemical tests and serological test indicate presence of Salmonella, + = Positive reaction for typical Salmonella; - = Negative reaction for typical Salmonella. BPW = Buffered peptone water; RV = Rappaport-Vassiliadis medium; MSC = Mannitol selenite cystine broth; MKTTn = Muller-Kauffmann-tetrathionate with novobiocin broth; XLD = xylose lysine desoxycholate agar; BS = Bismuth sulphite agar.

273 Results of biochemical (conventional and the Microbact) and serological confirmatory tests of Salmonella detection in commercial raw albumen by the cultural method corresponded, in which 14 out of 16 presumptive results were confirmed Salmonella

(Table 7.5). In this Experiment, selective enrichment in RV broth yielded lesser selectivity than that in MSC and MKTTn broths, as some presumptive results of sample selectively enriched in RV broth were confirmed as false positive results, which were identified as Enterobacter aerogenes and Ent. cloacae. These findings indicated that

Salmonella was present in commercial raw albumen, as determined by the cultural method.

••• Results of the MicroFoss method

An aliquot (0.1 mL) of raw albumen at 10-1 dilution (25g in 225 mL of BPW) prior to pre-enrichment and after pre-enrichment at 37°C + 1°C for 18h + 2h was separately inoculated into 9 mL of selective media from all formulae and analysed in the

MicroFoss system at 43°C for 48h. The MicroFoss detection of samples with and without pre-enrichment was compared.

Figures 7.7, 7.8, 7.9 and 7.10 show the MicroFoss curves of Salmonella detection in samples of raw albumen, with and without pre-enrichment in BPW at 37°C + 1°C for

18h + 2h, using the original formula, the modified formula with 40 mg/L of novobiocin, and the modified formulae with 1 and 2 mL/L of Niaproof 4, respectively.

274 The original formula: Salmonella was not detected in sample without pre-enrichment prior to analyses in the MicroFoss assay, whilst the pre-enriched sample (in duplicate) containing presumptive Salmonella yielded DTs of 9.2h and 2.3h with the optical units in the range of 200-650 (Figure 7.7).

The modified formula with 40 mg/L of novobiocin: Salmonella was not detected in sample without pre-enrichment prior to analyses in the MicroFoss assay, whilst the pre- enriched sample (in duplicate) containing presumptive Salmonella yielded DTs of 7.2h and 8.0h with the optical units in the range of 180-450 (Figure 7.8).

The modified formula with 1 mL/L of Niaproof 4: Sample (in duplicate) containing presumptive Salmonella without pre-enrichment gave DTs of 26.2h and 32.8h with the optical units in the range of 200-340, whilst the pre-enriched sample (in duplicate) containing presumptive Salmonella yielded DTs of 2.3h and 2.4h with the optical units in the range of 200-450 (Figure 7.9).

The modified formula with 2 mL/L of Niaproof 4: The first duplicate of sample containing presumptive Salmonella without pre-enrichment gave a DT of 27.2h with the optical units in the range of 200-270, whilst Salmonella was not detected in the second duplicate. Sample pre-enriched in BPW (in duplicate) containing presumptive

Salmonella yielded DTs of 4.9h and 5.0h with the optical units in the range of 200-500

(Figure 7.10).

275 Figure 7.7 The MicroFoss curves of detection of Salmonella in commercial raw albumen using the original formula, uninoculated control (No detection); duplicate of sample analysed prior to pre-enrichment (No detection); and duplicate of sample analysed after pre-enrichment (9.2h and 2.3h).

Figure 7.8 The MicroFoss curves of detection of Salmonella in commercial raw albumen using the modified formula with 40 mg/L of novobiocin, uninoculated control

(No detection); duplicate of sample analysed prior to pre-enrichment (No detection); and duplicate of sample analysed after pre-enrichment (7.2h and 8.0h).

276 Figure 7.9 The MicroFoss curves of detection of Salmonella in commercial raw albumen using the modified formula with 1 mL/L of Niaproof 4, uninoculated control

(No detection); duplicate of sample analysed prior to pre-enrichment (DT 26.2h &

32.8h); and duplicate of sample analysed after pre-enrichment (2.3h and 2.4h).

Figure 7.10 The MicroFoss curves of detection of Salmonella in commercial raw albumen using the modified formula with 2 mL/L of Niaproof 4, uninoculated control

(No detection); duplicate of sample analysed prior to pre-enrichment (No detection &

DT 27.4h); and duplicate of sample analysed after pre-enrichment (4.9h and 5.0h).

277 Sample without pre-enrichment that was analysed in the original formula and the modified formula with 40 mg/L of novobiocin yielded no detection by the MicroFoss system. In order to evaluate if this is the true negative result, a loopful of these vials was streaked onto TSA plates as preparation for further biochemical and serological tests.

However, no colony was observed on these TSA plates. On the other hand, late detections of presumptive Salmonella were obtained from analyses of non-pre-enriched sample in the modified formulae with 1 and 2 mL/L of Niaproof 4. It was noted that curve amplitudes of these detection curves were relatively shallow. These presumptive results were subsequently confirmed and identified.

Results of detection of presumptive Salmonella in sample pre-enriched in BPW could be achieved within 2-9h. Curve amplitudes were in the optimum ranges, which were similar to those of detection of confirmed Salmonella in commercial raw whole egg. All vials giving detection times that were streaked onto XLD and BS plates showed colonies exhibiting typical morphology of Salmonella on these plates. These presumptive results were selected and prepared for biochemical (conventional and the

Microbact) and serological tests for confirmation and identification, as shown in Table

7.6.

278 Table 7.6 Confirmation and identification of presumptive results obtained from detection of Salmonella in commercial raw albumen by the MicroFoss assay

Medium DT Presumptive result Results of a series Identification by the Serological (h) of conventional Microbact confirmation* biochemical tests indicating presence of Salmonella* The original 9.2 BPW-Org-XLD I + 99.87%-S. subsp. 1 + formula BPW-Org-BS Ia + 99.87%-S. subsp. 1 + BPW-Org-BS Ib + 82.72%-S. subsp. 3B + 2.3 BPW-Org-XLD II + 98.83%-S. subsp. 1 + BPW-Org-BS IIa + 99.87%-S. subsp. 1 + BPW-Org-BS IIb + 99.94%-S. subsp. 1 + The modified 7.2 BPW-NV 40-XLD I + 98.86%-S. subsp. 1 + formula with BPW-NV 40-BS Ia + 99.81%-S. subsp. 1 + 40 mg/L of BPW-NV 40-BS Ib + 72.57%-S. subsp. 1 + novobiocin 18.03%-S. subsp. 3B 8.0 BPW-NV 40-XLD II + 98.83%-S. subsp. 1 + BPW-NV 40-BS II + 98.86%-S. subsp. 1 + The modified 26.2 Non-PE-NP4 1-BS Ia - 73.99%-E. coli - formula with 25.39%-Inactive E. coli 1 mL/L of Non-PE-NP4 1-BS Ib - 97.26%-E. coli - Niaproof 4 2.72%-Inactive E. coli 32.8 Non-PE-NP4 1-BS IIa - 73.99%-E. coli - 25.39%-Inactive E. coli Non-PE-NP4 1-BS IIb - 97.26%-E. coli - 2.72%-Inactive E. coli 2.3 BPW-NP4 1-XLD I + 98.86%-S. subsp. 1 + BPW-NP4 1-BS Ia + 99.95%-S. subsp. 1 + BPW-NP4 1-BS Ib + 98.72%-S. subsp. 1 + 2.4 BPW-NP4 1-XLD II + 99.96%-S. subsp. 1 + BPW-NP4 1-BS IIa + 99.60%-S. subsp. 1 + BPW-NP4 1-BS IIb + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 The modified 24.3 Non-PE-NP4 2-BS II - 71.72%-Inactive E. coli - formula with 28.13%-E. coli 2 mL/L of 4.9 BPW-NP4 2-XLD I + 98.83%-S. subsp. 1 + Niaproof 4 BPW-NP4 2-BS Ia + 99.66%-S. subsp. 1 + BPW-NP4 2-BS Ib + 98.83%-S. subsp. 1 + BPW-NP4 2-BS Ic + 98.83%-S. subsp. 1 + 5.0 BPW-NP4 2-XLD II + 99.96%-S. subsp. 1 + BPW-NP4 2-BS IIa + 49.44%-S. subsp. 1 + 38.52%-S. subsp. 4 BPW-NP4 2-BS IIb + 96.79%-S. subsp. 1 +

* Results of a series of conventional biochemical tests and serological test indicate presence of Salmonella, + = Positive reaction for typical Salmonella; - = Negative reaction for typical Salmonella. BPW = Buffered peptone water; XLD = xylose lysine desoxycholate agar; BS = Bismuth sulphite agar. Non-PE = Non-pre-enriched sample; Org = The original formula; NV 4 = The modified formula with 40 mg/L of novobiocin; NP4 1 = The modified formula with 1 mL/L of Niaproof 4; and NP4 2 = The modified formula with 2 mL/L of Niaproof 4.

279 Presumptive results obtained from analyses of non-pre-enriched sample in the modified formulae with 1 and 2 mL/L of Niaproof 4 were confirmed as a false positive result and were identified as E. coli, which corresponded with relatively shallow curve amplitudes shown in Figures 7.9 and 7.10, respectively. Shallow curve amplitude exhibited fairly small degree of colour change in the medium, which was generally observed in Chapter

6 for false positive results caused by certain competing bacteria.

On the other hand, presumptive results of samples pre-enriched in BPW prior to analyses in the MicroFoss assay were entirely confirmed as Salmonella. It was assumed that the number of Salmonella present in raw albumen sample after thawing did not reach the detectable level for the MicroFoss system, whilst the number of E. coli presented was sufficient to be detected and caused a false positive result. Pre- enrichment substantially promoted the number of Salmonella in this sample to enhance an accurate and effective detection by the MicroFoss assay. Findings of this Experiment could be concluded that pre-enrichment was essential to enhance specificity of

Salmonella detection by the MicroFoss assay.

Results of detection of Salmonella in raw albumen by the MicroFoss assay agreed with those obtained from the cultural method and indicated presence of Salmonella in this sample.

Findings of Salmonella detection in raw whole egg and raw albumen samples indicated the relevant efficiency of the MicroFoss assay to the cultural method.

280 7.3.2 Sensitivity of the MicroFoss assay for Salmonella detection

Findings in previous Experiments showed an agreement of detection of confirmed

Salmonella in raw whole egg and raw albumen samples by the cultural and MicroFoss methods. Among these two samples, there was no occasion that confirmed Salmonella was not detected by the MicroFoss assay, but it was present in samples and was detected by the cultural method. Hence, sensitivity of the MicroFoss assay for

Salmonella detection was 100%, based on numbers of samples tested and results in these Experiments.

7.3.3 Specificity of the MicroFoss assay for Salmonella detection

There was no occasion that Salmonella was absent in the medium and was agreeably not detected by the cultural and MicroFoss methods. Specificity therefore could not be calculated in this instance. Further evaluation of specificity of the MicroFoss method with more samples is recommended.

7.3.4 Detection limit of MicroFoss method for Salmonella detection

Commercial pasteurized whole egg products (25g) from three batches (A, B and C), assumed to be free of Salmonella, were inoculated with 1 and 10 CFU of Salmonella

Enteritidis 1, as the individuals and in a culture pool with 103 CFU each of the five coliforms, Citrobacter freundii, Enterobacter aerogenes, Enterobacter cloacae,

Escherichia coli and Klebsiella pneumoniae. The inoculated samples were pre-enriched in BPW by incubation at 37°C for 16h, and subsequently analysed in the MicroFoss system at 43°C for 48h using the original formula in duplicate.

Salmonella was not detected in all three uninoculated pasteurized whole egg products throughout the incubation period of 48h. Detection of 1 CFU and 10 CFU of SE 1

281 inoculated into 25g of all three samples, following pre-enrichment in BPW for 37°C for

16h, was accomplished within 2.5 to 4.0 h (Figure 7.11) and 2.0h to 4.0h (Figure 7.12), respectively. However, detection of 1 CFU and 10 CFU of SE 1 in the presence of 103

CFU each of the five coliforms inoculated into 25g of these samples, following pre- enrichment for 16h, required longer time, 7.0-8.5h (Figure 7.11) and 6.5-8.0h (Figure

7.12), respectively, to completion. Curve shapes of all samples were similar with the optical units in the range of 200-600. The MicroFoss detection of 1 and 10 CFU of SE 1 in pasteurized whole egg products are shown in Table 7.7.

Table 7.7 The MicroFoss detection of 1 and 10 CFU of SE 1 in pasteurized whole egg products

Pasteurized Uninoculated SE 1 1 CFU SE 1 1 CFU & SE 1 10 CFU SE 1 10 CFU & whole egg control the five coliforms the five coliforms products DT (h) DT (h) DT (h) DT (h) DT (h) I II I II I II I II A No detection 2.9 3.1 7.3 7.8 3.3 3.6 7.5 6.6 B No detection 3.2 3.1 7.3 8.5 2.3 2.6 7.7 7.9 C No detection 3.5 3.6 7.2 7.0 3.3 3.1 7.5 7.6

282 a

b

c

Figure 7.11 The MicroFoss curves for detection of 1 CFU of SE 1, as individuals and in the presence of 103 CFU each of the five coliforms, in pasteurized whole egg products

(a) sample A; (b) sample B; and (c) sample C, respectively.

283 a

b

c

Figure 7.12 The MicroFoss curves for detection of 10 CFU of SE 1, as individuals and in the presence of 103 CFU each of the five coliforms, in pasteurized whole egg products (a) sample A; (b) sample B; and (c) sample C, respectively.

284 Since Salmonella is naturally present in eggs with other background microorganisms, it was concluded that the total time required for the MicroFoss assay to detect 1-10 CFU of Salmonella in the presence of other organisms in 25g of pasteurized whole egg products, including the cultural pre-enrichment for 16h, was 22.5-24.5h, compared to four to seven days to obtain results from the cultural method.

7.4 DISCUSSION

This Chapter evaluated the effectiveness of MicroFoss assay in comparison to the cultural method to simultaneously detect Salmonella in naturally contaminated eggs, which were commercial raw whole egg and raw albumen, using the original formula and three modified formulae of selective media created in Chapter 6. Salmonella detection in selective liquid media by the MicroFoss assay was equivalent to selective enrichment and selective plating steps of the cultural method.

Salmonella detection by the cultural method

Confirmed Salmonella in raw whole egg and raw albumen samples was detected by the cultural method. Greatest recovery of Salmonella in raw whole egg sample on XLD plates was obtained when this sample was selectively enriched in RV broth, followed by

MKTTn and MSC broths, respectively, whereas recovery of Salmonella in raw albumen on XLD plates by these three selective enrichment broths was not different. Superiority of RV broth to tetrathionate brilliant green and selenite cystine broths to promote growth of Salmonella was indicated in previous studies (Allen et al., 1991; Maijala et al., 1992; June et al., 1996; Rall et al., 2005). Schönenbrücher et al. (2008) reported that

RV broth with soypeptone (RVS) yielded highest recovery of Salmonella spp. (97.4%) compared to 94.9% and 38.5% for MKTTn and SC broths, respectively for Salmonella

285 detection in 286 naturally contaminated raw porcine and bovine minced meat samples.

Fries and Steinhof (1997) reported the ability to detect very low numbers of Salmonella

Enteritidis in the presence of high numbers of competing bacteria by selective enrichment in RV broth. Vassiliadis et al. (1991) indicated that RV medium was more sensitive and selective than tetrathionate brilliant green broth for detection of

Salmonella Typhimurium inoculated into pasteurized fluid whole milk samples.

Al-Hindawi and Rished (1979) reported that tetrathionate brilliant green bile (TBGB) broth was superior to SC broth for isolation of some serotypes of Salmonella in local foods in Iraq and recommended the use of more than one selective enrichment medium for Salmonella detection. D’Aoust et al. (1992) reported greater sensitivities of commercial MKTTn broth to recover Salmonella in foods than selenite cystine (SC) broth.

Greatest selectivity of Salmonella detection in raw whole egg sample was achieved when this sample was selectively enriched in RV broth, as presumptive colonies appeared on XLD and BS plates were entirely confirmed as Salmonella. Cherrington and Huis in’t Veld (1993) stated that false positive results of Salmonella detection in pig faeces were obtained when samples pre-enriched in BPW were selectively enriched in

MKTTn and SC broths, except for RV broth. It was noted that false positive results in raw whole egg sample were solely obtained from presumptive colonies on BS plates, after selective enrichment in MSC and MKTTn broths. This finding disagreed with data by D’Aoust et al. (1992), in which BS plates were found highly selective and provided an effective detection of typical and atypical biotypes of Salmonella spp. False positive results in raw whole egg sample were identified as Citrobacter braakii, C. freundii and Enterobacter cloacae. Citrobacter freundii were indicated to commonly

286 cause false positive result to Salmonella detection in foods including eggs (Ogden and

Cann, 1987; D’Aoust and Sewell, 1988; Blackburn et al., 1994; Coleman et al., 1995).

In contrast, lesser selectivity of Salmonella detection in raw albumen sample enriched in

RV broth was obtained, as some presumptive colonies on BS plates of sample enriched in RV broth were identified as false positive results. These false positive results were identified as Enterobacter aerogenes and Ent. cloacae. Enterobacter cloacae caused false positive result for detection of Salmonella in food samples using an improved

Impedance-Splitting method (Pless et al., 1994).

Hara-Kudo et al. (2001) evaluated detection of Salmonella Enteritidis in naturally contaminated liquid eggs and artificially inoculated shell eggs by the cultural method comparing between various enrichment and plating media and immunoconcentration.

These findings indicated that the most effective procedure for detection of S. Enteritidis in shell eggs and liquid eggs was pre-enrichment in BPW supplemented with cysteine at

35°C-37°C for 20-24h, followed by selective enrichment in tetrathionate broth at 42°C for 20h, and selective plating onto any of the selected agar plates in this study, such as

XLD, XLT4, modified brilliant green, SMID and Rambach agars.

Salmonella detection by the MicroFoss assay

Buffered peptone water was used for pre-enrichment in this study as it was recommended for pre-enrichment of all foods, except for cocoa and cocoa products, by

AS 5013.10-2004. However, different pre-enrichment media for Salmonella detection in eggs, egg products and egg-associated foods were suggested, such as buffered nutrient broth/milk medium for dried eggnog and egg noodles (Goepfert et al., 1970); nutrient broth for irradiated whole egg magma (Licciardello et al., 1970); 0.5% lactose broth for dried egg albumin (Price et al., 1972); mannitol broth for eggs (Al-Hindawi and Rished,

287 1979); trypticase (tryptic) soy broth for shell eggs, liquid whole egg homogenates and hard-boiled eggs (Amaguana and Andrews, 2000).

Salmonella detection by the MicroFoss assay without pre-enrichment was also conducted to assess whether this could shorten the detection process. Although the

MicroFoss assay exhibited remarkably good sensitivity by the ability to detect injured cells of confirmed Salmonella in raw whole egg sample without pre-enrichment in buffered peptone water, detection of Salmonella in this sample was delayed, particularly until 41.5h for samples analysed in the original formula, as shown in Figure 7.1.

Detection of presumptive Salmonella in non-pre-enriched raw albumen sample using the modified formulae with 1 and 2 mL/L of Niaproof 4 did not only prolong (Figures

7.7 and 7.8), but these presumptive results were also identified as false positive results, which was Escherichia coli. On the other hand, Salmonella detection in raw whole egg and raw albumen samples pre-enriched in BPW, followed by detection in the MicroFoss system using all formulae of selective media, could be accomplished within 2-6h and 2-

9.5h, respectively. Presumptive results obtained were confirmed as Salmonella. Hence, these findings indicated that pre-enrichment of foods, particularly processed foods as frozen egg samples used in these Experiments, rather than direct selective enrichment was crucial, as pre-enrichment: (i) resuscitates and proliferates injured cells of

Salmonella to detectable levels; (ii) enhances greater recovery of Salmonella; (iii) dilutes toxic or inhibitory substances in foods; and (iv) provides an accurate and effective detection of Salmonella (Gabis and Silliker, 1974; Amaguana and Andrews,

2000; Jay, 2003; Liao and Fett, 2003). Results from this study were supported by Duffy and Catarame (2005) that rapid methods still lack sufficient sensitivity and specificity for direct testing. Hence, all current rapid methods should be undergone at least one

288 enrichment step, involving incubation of at least six and often up to 48h to increase numbers of target pathogen to a detectable level.

Bailey and Cox (1992) suggested pre-enrichment in universal pre-enrichent (UP) broth for 24h to allow simultaneous recovery of as few as 10 cells each of heat-injured

Salmonella and Listeria, as pure culture and in culture pool as well as in several foods.

The beneficial use of pre-enrichment in UP broth at 42°C for 6h to facilitate detection of low levels of Salmonella inoculated into shell eggs and egg products by the MicroFoss system was demonstrated by Tan and Shelef (1999) and Peng and Shelef (2001).

Preenrichment of the egg samples in UP broth prior to detection in the MicroFoss assay using propylene glycol-based media could provide an effective and rapid detection of

Salmonella in these samples. Detection of Salmonella in raw whole egg and raw albumen samples after pre-enrichment in buffered peptone water using propylene glycol-based media was achieved within 6h and 9h, as shown in Figures 7.3-7.6 and

7.7-7.10, respectively.

According to the number of raw egg samples analysed in these Experiments, it could be concluded that detection of Salmonella by the MicroFoss assay had equivalent sensitivity to the cultural method. High selectivity of Salmonella detection using propylene glycol-based selective media developed in Chapter 6 in the MicroFoss system was achieved, as 100% of presumptive results of pre-enriched samples were confirmed as Salmonella. However, specificity of the MicroFoss method could not be calculated, as there was no occasion that Salmonella was not detected by both methods, and

Salmonella was in fact not present in samples. Further evaluation of the efficiency of the MicroFoss method for Salmonella detection with more egg samples is recommended. High specificity (≥97%) of detection of Salmonella in food samples

289 (poultry, meat and eggs) (Pignato et al., 1995b) and stool specimens (Dusch and

Altwegg, 1993; Dusch and Altwegg, 1995; Ruiz et al., 1996) using propylene glycol- based selective medium as Rambach agar after pre-enrichment and selective enrichment was achieved. However, variations in sensitivity of detection of Salmonella using

Rambach agar were reported, with sensitivity ranging from 56.8% (Ruiz et al., 1996),

69% (Dusch and Altwegg, 1993), 88% (Manafi and Willinger, 1994), 91% (Manafi and

Sommer, 1992), 97.9% (Pignato et al., 1995b), to 100% (Monnery et al., 1994).

Detection of low levels of SE 1 (1-10 CFU) inoculated into 25g of pasteurized whole egg products and pre-enriched in BPW at 37°C for 16h was achieved by the MicroFoss assay within 2-4h. Longer time (6.5-8.5h) was required for detection of 1-10 CFU of SE

1 in the presence of competing bacteria (103 CFU each of the five coliforms) in 25g of pre-enriched samples of pasteurized whole egg product. Hence, a total time required for detection of 1-10 CFU of Salmonella in 25g of eggs by the MicroFoss system, including the cultural pre-enrichment, was approximately 24h, compared to five to seven days required to obtain results of the cultural method. This finding agreed with those of previous studies that detection of low levels of Salmonella, 2-25 CFU (Tan and Shelef,

1999); and 10-50 CFU (Peng and Shelef, 2001), in 25g of eggs by the MicroFoss system could be achieved in 24h. Hence, detection of low levels of Salmonella using propylene glycol-based media in the MicroFoss assay could provide an effective and rapid detection system for Salmonella.

The findings reported in this Chapter demonstrate the accuracy and sensitivity of the

MicroFoss method in comparison to the cultural method for detecting Salmonella in naturally contaminated eggs using the selective media created in Chapter 6. High selectivity of Salmonella detection using these media was enhanced when samples were pre-enriched prior to the MicroFoss analyses, as all presumptive results of pre-enriched

290 samples were confirmed as Salmonella. Detection of low levels (1-10 CFU) of

Salmonella in 25g of eggs by the MicroFoss system, including the cultural pre- enrichment, could be completed within 24h, as opposed to five to seven days required to obtain results by the cultural method. These findings suggested the MicroFoss method can be used as a rapid alternative method for detection of Salmonella in eggs and other foods. Future research for evaluation of the effectiveness of Salmonella detection using propylene glycol-based media in the MicroFoss assay should be conducted, as findings in this study were based on limited numbers of naturally contaminated samples analysed and one representative strain of Salmonella Enteritidis. While the other commercially available assays may have the equivalent sensitivity and specificity to the MicroFoss, disadvantages of these assays, such as polymerase chain reaction (PCR)-based method, were encountered, as they are labour-intensive, relative time-consuming, and expensive

(Bassam et al., 1996). The MicroFoss assay provides a simple, rapid automated system and facilitates simultaneous detection of up to 128 samples. As a consequence, the

MicroFoss is proven to provide an effective detection system for Salmonella.

Other commercially available assays for Salmonella detection in raw eggs and egg products also offer relevant sensitivity, simplicity, and speed. Seo et al. (2003) reported the efficiency of the antibody-based lateral flow assay (Neogen, Lansing, MI) as a presumptive qualitative test for detection of Salmonella group D1 serovars, such as SE, in raw eggs samples in a collaborative study with Neogen, Lansing, MI. Pre-enrichment of raw egg pools inoculated with 10 SE cells per mL of egg in BPW or tetrathionate brilliant green broth at 37°C for 24h, followed by Neogen’s lateral flow assay, yielded

100% detection of SE in raw egg pools inoculated to yield 10 SE cells per mL of egg.

Results could be observed 20 mins after adding a portion of samples (100 μL) into the round sample port of the test device. Salmonella detection using the lateral flow assay

291 was equivalent to the MicroFoss Salmonella detection assay, which combined selective enrichment and selective plating steps of the Australian Standard method (AS 5013.10-

2004) for Salmonella detection, after pre-enrichment in BPW at 37°C + 1°C for 18h +

2h performed in this study. Detection limit of this test was increased approximately tenfold up to 105-106 cells/mL in whole egg contents when the novel antigen extraction method developed for use with this test panel kit was used. The developed lateral flow test kit could provide a simple, rapid and inexpensive method for detection of

Salmonella group D1 serovars, such as SE, in raw egg samples. This finding indicated that the Neogen’s lateral flow device could detect lower levels of SE in raw egg samples, 1 cells/mL of egg, compared to 1-10 CFU/25g of eggs by the MicroFoss assay in comparable time.

McMahon et al. (2004) reported results of a collaborative study among 25 laboratories to compare the VIDAS Salmonella (SLM) with RV method for detection of Salmonella in certain foods including dried egg to the current standard method presented in the U.S.

Food and Drug Administration’s Bacteriological Analytical Manual (BAM) and the culture method presented in AOAC’s Official Methods of Analysis. Sensitivity rates for the high (10-50 cells/25g) and low (1-5 cells/25g) inoculum levels were 98% and 100%, respectively, for the VIDAS method, and 100% for both levels for the culture method.

False negative rates for the VIDAS method were 1.5% and 0% for thr high and low inoculum levels, respectively. False negative rates for the culture method were 0% for both inoculum levels. Agreement between the two methods was 99% and 100% for the high and low inoculum levels, respectively. The number of positives obtained from the

VIDAS and culture methods was not significantly different (p > 0.05) between the two methods. Sensitivity of the MicroFoss Salmonella assay obtained in this study based on

292 number of commercial raw egg samples tested indicated similar or greater sensitivity of the MicroFoss compared to the VIDAS Salmonella method.

293 CHAPTER 8

Conclusions

The objective of this research was to evaluate the efficacy of the MicroFoss system as an alternative rapid assay to determine microbiological quality of eggs and egg products. Although more microbiological tests (TVC; coliforms; E. coli; yeast and moulds; and lactic acid bacteria) by the MicroFoss system for eggs are currently available, only certain tests, which were TVC; coliforms; and lactic acid bacteria, were available and only for pasteurized egg products at the time of commencing this research.

Detection of Salmonella by the MicroFoss system has not yet been available. Hence, this research evaluated the effectiveness of the MicroFoss system for use in routine quality assessment of eggs and egg products in regards to the overall microbiological quality (TVC); hygiene and adequacy of processing (coliforms and E. coli); and safety for human consumption (Salmonella).

Chapter 3 evaluated the MicroFoss system for enumeration of TVC in eggs and egg products. The MicroFoss method was shown to have good correlations with the cultural

TVC method in all types of eggs and egg products, with correlation coefficients of -0.84 to –0.90 for linear regression analysis by the MicroFoss system and -0.86 to –0.95 for quadratic regression analysis by the SPSS software. However, only slight improvement in correlations of two methods was achieved by quadratic regression. Thus, linear regression analysis was suitable to represent a relationship of the cultural TVC results and DTs. Enumerations of TVC in eggs and egg products by the MicroFoss system completed within two to 14h, with the delayed detection of four to 15h and seven to 16h of raw albumen and pasteurized albumen product, respectively, compared to 48-72h required for completion of cultural method. This facilitated the egg manufacturers to

294 obtain timely TVC results to determine whether the products comply with the specifications and benefited them for an early release of the products to consumers.

Corrective action could be therefore more effectively implemented in the case of any eggs or egg products failing to meet the specifications. Furthermore, preliminary results of the enumeration of TVC of homogenates of whole egg inoculated with Pseudomonas fluorescens by the MicroFoss system suggested that the MicroFoss method could be useful to predict spoilage of eggs. Further study to evaluate threshold level of

Pseudomonas populations to cause spoilage was essential to be conducted.

Chapters 4 and 5 evaluated the MicroFoss system for detection and enumeration of coliforms and E. coli in eggs and egg products. Good agreements between the cultural and the MicroFoss methods of the enumeration of coliforms and E. coli in all types of eggs and egg products were achieved, with correlation coefficients in excess of –0.93 to

–0.95 and –0.94 to –0.98, respectively. The enumerations of coliforms and E. coli in eggs and egg products completed within two to 14h and two to 17h, respectively, with the delayed detection of 12-23h and 12.5-20h for raw albumen and pasteurized albumen products, respectively. The effectiveness of the MicroFoss method to estimate populations of coliforms and E. coli was evaluated with commercial raw egg samples.

Results showed that coliforms and E. coli counts in a majority of egg samples, in excess of 78% and 80%, respectively, obtained from the cultural and MicroFoss methods were relatively comparable. This showed that the MicroFoss system could be effective for estimation of coliforms and E. coli in eggs and egg products. Further evaluations indicated the high or equivalent accuracy, sensitivity, specificity, and repeatability of the MicroFoss method in comparison to the cultural methods. Furthermore, the

MicroFoss was also able to detect confirmed coliforms and E. coli, particularly in pasteurized egg products, when they were not detected by the cultural method. This

295 indicated the efficiency of the MicroFoss method for use as an alternative rapid method for enumeration of coliforms and E. coli in eggs and egg products.

Results of Chapters 3-5 suggested the potential use of the MicroFoss system as an alternative rapid assay to enumerate TVC, coliforms and E. coli in eggs and egg products for assessing their microbiological qualities. The calibration curves obtained in these studies were utilized by a commercial egg manufacturer. Results from ongoing testings of eggs and egg products corresponded well with those from these studies. This confirmed the validity of the findings from these studies and demonstrated the routine and ongoing use in a commercial setting of the MicroFoss method to rapidly estimate

TVC, coliform and E. coli populations in eggs and egg products.

As the MicroFoss assay for Salmonella detection has not yet been available. Hence,

Chapter 6 describes the development of a selective medium for use in the MicroFoss assay. The propylene glycol-based medium containing brilliant green and MgCl2.6H2O as selective agents and incubating at high incubation temperature (43°C) was found effective for detection of Salmonella, whilst most non-Salmonella cultures tested gave negative reactions to this medium. Although the original formulation of this propylene glycol-based medium was effective for detection of Salmonella, colour changes of this medium caused by non-salmonellae, such as E. coli, K. pneumoniae and Pr. mirabilis, was observed occasionally. Hence, the original formulation of this medium was modified by supplementation with 40 mg/L of novobiocin or 1 or 2 mL/L of Niaproof 4.

Supplementations of the original formula with these concentrations of novobiocin or

Niaproof 4 substantially inhibited E. coli, whilst false positive result caused by Pr. mirabilis was eliminated. Enterobacter cloacae was greatly inhibited at all concentrations of Niaproof 4, whereas it was not inhibited in the presence of 40 mg/L of novobiocin and caused a colour change in the medium similar to those caused by

296 Salmonella. Regardless of the presence of these additional selective agents, Klebsiella pneumoniae continued giving a false positive result. Supplementation of the original formula with 2 mL/L of Niaproof 4 yielded the greatest selectivity.

Examination of the selectivity and effectiveness of propylene glycol-based media at all formulations with a range of non-salmonellae (genera in the family Enterobacteriaceae, miscellaneous Gram-negative bacteria and Gram-positive bacteria) and various serotypes of Salmonella indicated that propylene glycol-based medium can be used for an effective detection of Salmonella including atypical biotype strains.

Chapter 7 then Evaluation of the effectiveness of Salmonella detection in eggs using propylene glycol-based media by the MicroFoss system is conducted in Chapter 7. The findings demonstrate the accuracy and sensitivity of the MicroFoss method in comparison to the cultural method for detecting Salmonella in naturally contaminated eggs using the propylene glycol-based media at all formulations developed in Chapter 6.

High selectivity of Salmonella detection using these media was enhanced when samples were pre-enriched prior to the MicroFoss analyses, as all presumptive results of pre- enriched samples were confirmed as Salmonella. Detection of low levels (1-10 CFU) of

Salmonella in 25g of eggs by the MicroFoss system, including the cultural pre- enrichment, could be completed within 24h, as opposed to five to seven days required to obtain results by the cultural method. These findings suggested the MicroFoss method could be used as a rapid alternative method for detection of Salmonella in eggs and other foods. Further studies of the effectiveness of Salmonella detection using propylene glycol-based media by the MicroFoss system could be conducted on i) the ability of the MicroFoss assay to detect low levels of serotypes of Salmonella other than

S. Enteritidis used in this study ; ii) the ability of the MicroFoss assay to detect low levels of serotypes of Salmonella including S. Enteritidis used in this study in the

297 modified formula of propylene glycol-based media; and iii) the assessment of the application of the MicroFoss assay to detect Salmonella in a variety of food products.

298 CHAPTER 9

Bibliography

Adams, M.R., Moss, M.O., 2000. Food Microbiology, 2nd ed, The Royal Society of Chemistry, Cambridge.

Al-Hindawi, N., Rished, R., 1979. Presence and distribution of Salmonella species in some local foods from Baghdad City, Iraq. Journal of Food Protection 42, 877- 880.

Allen, G., Bruce, V.R., Stephenson, P., Satchell, F.B., Andrews, W.H., 1991. Recovery of Salmonella from high-moisture foods by abbreviated selective enrichment. Journal of Food Protection 54, 492-495.

Amaguana, R.M., Andrews, W.H., 2000. Salmonella-Detection by classical cultural techniques. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, London, pp. 1948-1952.

Andrews, A.T., 1978. The composition, structure and origin of proteose-peptone component 5 of bovine milk. European Journal of Biochemistry 90, 59-65.

Andrews, W.H., 2000a. Microbiological methods-Subchapter 1 Eggs and egg products. In: Horwitz, W. (Ed.), Official Methods of Analysis of AOAC International, 17th ed, AOAC International, Gaithersburg, pp. 1-3.

Andrews, W.H., 2000b. Microbiological methods-Subchapter 9 Salmonella. In: Horwitz, W. (Ed.), Official Methods of Analysis of AOAC International, Vol. 1, 17th ed, AOAC International, Gaithersburg, pp 78-137.

Andrews, W.H., Flowers, R.S., Silliker, J., Bailey, J.S., 2001. Salmonella. Downes, F.P., Ito, K. (Eds.), Compendium of Methods for the Microbiological Examination of Foods, 4th ed, American Public Health Association, Washington, D.C., pp. 357-380.

Arnaut-Rollier, I., Vauterin, L., De Vos, P., Massart, D.L., Devriese, L.A., De Zutter, L., Van Hoof, J., 1999. A numerical taxonomic study of the Pseudomonas flora isolated from poultry meat. Journal of Applied Microbiology 87, 15-28.

Arroyo, G., Arroyo, J.A., 1995. Selective action of inhibitors used in different culture media on the competitive microflora of Salmonella. Journal of Applied Bacteriology 78, 281-289.

299 AS 1766.1.3-1991, Standards Australia committee FT/4, Food Microbiology, 1991. Method 1.3: General procedures and techniques-Colony count-Pour plate method, Standards Australia, Australia.

AS 1766.1.4-1991, Standards Australia Committee FT/4, Food Microbiology, 1991. Method 1.4: General procedures and techniques-Colony count-Surface spread method, Standards Australia, Australia.

AS 1766.2.1-1991, Standards Australia committee FT/4, Food Microbiology, 1991. Method 2.1: Examination for specific organisms-Standard plate count, Standards Australia, Australia.

AS 1766.3.8-1991, Standards Australia committee FT/4, Food Microbiology, 1991. Method 3.8: Examination of specific products-Eggs and egg products, Standards Australia, Australia.

AS 1766.2.3-1992, Standards Australia committee FT/4, Food Microbiology, 1992. Method 2.3: Examination for specific organisms-Coliforms and Escherichia coli, Standards Australia, Australia.

AS 4276.6-1995, the Joint Standards Australia/Standards New Zealand Committee on Water Microbiology, FT/20, 1995. Method 6: Thermotolerant coliforms and Escherichia coli-Estimation of most probable number (MPN), Standards Australia, Australia.

AS 5013.4-2004, the Joint Standards Australia/Standards New Zealand Committee FT- 004, 2004. Method 4: Microbiology-General guidance for the enumeration of coliforms-Colony count technique, Standards Australia International Ltd, Sydney.

AS 5013.7-2004, the Joint Standards Australia/Standards New Zealand Committee FT- 004, Food Microbiology, 2004. Method 7: Examination of specific products- Eggs and egg products, Standards Australia, Sydney.

AS 5013.10-2004, Standards Australia committee FT-004, Food Microbiology, 2004. Method 10: Microbiology of food and animal feeding stuffs-Horizontal method for the detection of Salmonella spp., Standards Australia Limited, Sydney.

AS 5013.14-2004, Standards Australia committee FT-004, Food Microbiology, 2004. Method 14: Microbiology of food and animal feeding stuffs-General rules for microbiological examinations, Standards Australia International Ltd, Sydney.

AS 5013.15-2004, the Joint Standards Australia/Standards New Zealand Committee FT- 004, 2004. Method 15: Microbiology-General guidance for enumeration of presumptive Escherichia coli-Most probable number technique, Standards Australia International Ltd, Sydney.

300 AS 5013.15-2006, the Standards Australia Committee FT-024, Food Products and Constituted Subcommittee FT-024-01, Food Microbiology, 2006. Method 15: Microbiology of food and animal feeding stuffs-Horizontal method for the detection and enumeration of presumptive Escherichia coli- Most probable number technique, Standards Australia, Sydney.

AS 5013.3-2009, the Standards Australia Committee FT-024, Food Products and Constituted Subcommittee FT-024-01, Food Microbiology, 2009. Method 3: Microbiology of food and animal feeding stuffs-Horizontal method for the detection and enumeration of coliforms-Most probable number technique, Standards Australia, Sydney.

AS 5013.4-2009, the Standards Australia Committee FT-024, Food Products and Constituted Subcommittee FT-024-01, Food Microbiology, 2009. Method 4: Microbiology of food and animal feeding stuffs-Horizontal method for the detection and enumeration of coliforms-Colony-count technique, Standards Australia, Sydney.

AS/NZS 4659.2-1999, Joint Technical Committee FT/4, Food Microbiology, 1999. Guide to determining the equivalence of food microbiology test methods-Part 2: Quantitative tests, Standards Australia International Ltd, Australia.

AS/NZS 4659.3-1999, Joint Technical Committee FT/4, Food Microbiology, 1999. Guide to determining the equivalence of food microbiology test methods-Part 3: Confirmation tests, Standards Australia International Ltd, Australia.

Ayres, J.C., 1960. The relationship of organisms of the genus Pseudomonas to the spoilage of meat, poultry and eggs. Journal of Applied Bacteriology 23, 471-486.

Bailey, J.S., Cox, N.A., 1992. Universal pre-enrichment broth for the simultaneous detection of Salmonella and Listeria in foods. Journal of Food Protection 55, 256-259.

Baker, R.C., 1974. Microbiology of eggs. Journal of Milk and Food technology 37, 265- 268.

Baker, R.C., Bruce, C., 1994. Effects of processing on the microbiology of eggs. In: Board, R.G., Fuller, R. (Eds.), Microbiology of the Avian Egg, Chapman & Hall, London, pp. 153-173.

Banwart, G.J., 1989. Basic Food Microbiology, 2nd ed, Van Nostrand Reinhold, New York.

Bassam, B.J., Allen, T., Flood, S., Stevens, J., Wyatt, P., Livak, K.J., 1996. Nucleic acid sequence detection systems: Revolutionary automation for monitoring and reporting PCR products. Australasian Biotechnology 6, 285-294.

301 Batt, C.A., 2000. Escherichia coli-Escherichia coli. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 633-640.

Bennett, A.R., Greenwood, D., Tennant, C., Banks, J.G., Betts, R.P., 1998. Rapid and definitive detection of Salmonella in foods by PCR. Letters in Applied Microbiology 26, 437-441.

Bhargava, D.N., Mandokhot, U., Panghal, R.S., Naryan, K.G., Singh, I.P., 1975. Quantitative and qualitative microflora on the poultry egg shell surface. Indian Journal of Poultry Science 10, 19-23.

Bishop, J.R., White, C.H., 1985. Estimation of potential shelf life of cottage cheese utilising bacterial numbers and metabolites. Journal of Food Protection 48, 1054-1057.

Blackburn, C. de W., 1993. Rapid and alternative methods for the detection of salmonellas in foods. Journal of Applied Bacteriology 75, 199-214.

Blackburn, C. de W., Curtis, L.M., Humpheson, L., Petitt, S.B., 1994. Evaluation of Vitek Immunodiagnostic Assay System (VIDAS) for the detection of Salmonella in foods. Letters in Applied Microbiology 19, 32-36.

Blackburn, C. de W., McCarthy, J.D., 2000. Modifications to methods for the enumeration and detection of injured Escherichia coli O157:H7 in foods. International Journal of Food Microbiology 55, 285-290.

Bland, J.M., Altman, D.G., 1999. Measuring agreement in method comparison studies. Statistical Methods in Medical Research 8, 135-160.

Blivet, D., Salvat, G., Humbert, F., Colin, P., 1998. Development of a new culture medium for the rapid detection of Salmonella by indirect conductance measurements. Journal of Applied Microbiology 84, 399-403.

Bloch, N., Sidjabat-Tambunan, H., Tratt, T., Bensink, J., Lea, K., Frost, A.J., 1996. The enumeration of coliforms and E. coli on naturally contaminated beef: a comparison of the PetrifilmA method with the Australian standard. Meat Science 43, 187-193.

Board, R.G., 1964. The growth of Gram-negative bacteria in the hen’s egg. Journal of Applied Bacteriology 27, 350-364.

Board, R.G., Ayres, J.C., Kraft, A.A., Forsythe, R.H., 1964. The microbiological contamination of egg shells and egg packing materials. Poultry Science 43, 584- 595.

Board, R.G., Clay, C., Lock, J., Dolman, J., 1994. The egg: A compartmentalized aseptically packaged food. In: Board, R.G., Fuller, R., (Eds.), Microbiology of the Avian Egg, Chapman & Hall, London, pp. 43-61. 302 Board, R.G., Loseby, S., Miles, V.R., 1979. A note on microbial growth on hen egg-shells. British Poultry Science 20, 413-420.

Board, R.G., Tranter, H.S., 1986. The microbiology of eggs. In: Stadelman, W.J., Cotterill, O.J. (Eds.), Egg Science and Technology, 3rd ed, MacMillan Publishers, Westpost, pp. 75-95.

Boothroyd, M., Baird-Parker, A.C., 1973. The use of enrichment serology for Salmonella detection in human foods and animal feeds. Journal of Applied Bacteriology 36, 165-172.

Brackett, R.E., Beuchat, L.R., 1991. Survival of Listeria monocytogenes in whole egg and egg yolk powders and in liquid whole eggs. Food Microbiology 8, 331-337.

Brenner, D.J., 1984. Facultatively anaerobic Gram-negative rods-Family I Enterobacteriaceae. In: Krieg, N.R. (Ed.), Bergey’s Manual of Systematic Journal of Applied Bacteriology 23, 499-509.

Brenner, D.J., 1992. Introduction to the family Enterobacteriaceae. In: Balows, A., Truper, H,G, Dworkin, M., Harder, W., Schleifer, K.-H. (Eds.), The Prokaryotes-A Handbook on the Biology of Bacteria: Ecophysiology, Isolation, Identification, Applications Vol. 3, 2nd ed, Springer-Verlag, New York, pp. 2673-2695.

Brooks, J., 1960. Mechanism of the multiplication of Pseudomonas in the hen’s egg. Journal of Applied Bacteriology 23, 499-509.

Bruce, J., Drysdale, E.M., 1994. Trans-shell transmission. In: Board, R.G., Fuller, R. (Eds.), Microbiology of the Avian Egg, Chapman & Hall, London, pp. 63-91.

Bullock, R.D., Frodsham, D., 1989. Rapid impedance detection of salmonellas in confectionery using modified LICNR broth. Journal of Applied Bacteriology 66, 385-391.

Burger, M., Woods, R,G, McCarthy, C., Beacham, I.R., 2000. Temperature regulation of protease in Pseudomonas fluorescens LS107d2 by an ECF sigma factor and a transmembrane activator. Microbiology 146, 3149-3155.

Busse, M., 1995. Media for Salmonella. International Journal of Food Microbiology 26, 117-131.

Caffer, M.I., Eiguer, T., 1994. Salmonella Enteritidis in Argentina. International Journal of Food Microbiology 21, 15-19.

Camps, N., Dominguez, A., Company, M., Perez, M., Pardos, J., Llobet, T., Usera, M.A., Salleras, L. 2005. A foodborne outbreak of Salmonella infection due to overproduction of egg-containing foods for a festival. Epidemiology and Infection 303 133, 817-822

Cantor, A., McFarlane, V.H., 1948. Salmonella organisms on and in chicken eggs. Poultry Science 27, 350-355.

Carli, K.T., Eyigor, A., Caner, V., 2001. Prevalence of Salmonella serovars in chickens in Turkey. Journal of Food Protection 64, 1832-1835.

Carlson, V.L., Snoeyenbos, G.H., McKie, B.A., Smyser, C.F., 1967. A comparison of incubation time and temperature for the isolation of Salmomella. Avian Diseases 11, 217-225.

CDC (Centersfor Disease Control), 2000. Outbreaks of Salmonella serotype Enteritidis infection associated with eating raw or undercooked shell eggs-United States, 1996-1998. Morbidity and Mortality Weekly Report 49, 73-79.

Cherrington, C.A., Huis in’t Veld, J.H.J., 1993. Comparison of classical isolation protocols with a 24h screen to detect viable salmonellas in faeces. Journal of Applied Bacteriology 75, 65-68.

Clark, M.A., Barrett, E.L., 1987. Catabolite repression of thiosulfate reduction by Salmonella Typhimurium. Current Microbiology 16, 27-31.

Cogan, T.A., Humphrey, T.J., 2003. The rise and fall of Salmonella Enteritidis in the UK. Journal of Applied Microbiology 94, 114S-119S (supplement).

Coleman, D.J., Nye, K.J., Chick, K.E., Gagg, C.M., 1995. A comparison of immunomagnetic separation plus enrichment with conventional Salmonella culture in the examination of raw sausages. Letters in Applied Microbiology 21, 249-251.

Cooke, V.M., Miles, R.J., Price, R.G., Richardson, A.C., 1999. A novel chromogenic ester agar medium for detection of salmonellae. Applied and Environmental Microbiology 65, 807-812.

Coppola, K., Firstenberg-Eden, R., 1988. Impedance based rapid method for detection of spoilage organisms in UHT low-acid foods. Journal of Food Science 53, 1521- 1524, 1527.

Cotterill, O.J., 1986. Egg breaking. In: Stadelman, W.J., Cotterill, O.J. (Eds.), Egg Science and Technology, 3rd ed, MacMillan Publishers, Westpost, Connecticut, pp. 193-215.

Cotterill, O.J., Glauert, J., Steinhoff, S.E., Baldwin, R.E., 1974. Hot-pack pasteurization of salted egg products. Poultry Science 53, 636-645.

Cousin, M.A., 2000. Introduction-Pseudomonas. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, San Diego pp. 304 1864-1867.

Cox, J.M., 1993. Lysine-mannitol-glycerol agar, a medium for the isolation of Salmonella spp., including S. typhi and atypical strains. Applied and Environmental Microbiology 59, 2602-2606.

Cox, J.M., 2000a. Salmonella-Introduction. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, London, pp. 1928-1937.

Cox, J.M., 2000b. Salmonella-Salmonella Typhi. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, London, pp. 1943-1947.

Cox, J.M., 2001. Eggs and egg products. In: Moir, C.J., Kabilafkas, C.A., Arnold, G., Cox, B.M., Hocking, A.D., Jenson, I. (Eds.), Spoilage of Processed Foods: Causes and Diagnosis, Australian Institute of Food Science and Technology (AIFST) Incorporated, NSW Branch, Food Microbiology Group, Waterloo DC, pp. 167-175.

Cox, J.M., Fleet, G.H., 2003. New directions in the microbiological analysis of foods. In: Hocking, A.D., (Ed.), Foodborne Microorganisms of Public Health Significance, AIFST Incorporated, NSW Branch, Food Microbiology Group, Waterloo DC, pp. 103-161.

Craven, H.M., Eyles, M.J., Davey, J.A., 2003. Enteric indicator organisms in foods. In: Hocking, A.D. (Ed.), Foodborne Microorganisms of Public Health Significance, 6th ed, AIFST Incorporated, NSW Branch, Food Microbiology Group, Waterloo DC, pp. 163-193.

Curda, L., 2000. Electrical techniques-lactics and other bacteria. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 580-588.

D’Aoust, J.Y., Sewell, A.M., 1988. Detection of Salmonella with the BioEnzabead enzyme immunoassay technique. Journal of Food Protection 51, 538-541.

D’Aoust, J.-Y., Sewell, A.M., Jean, A., 1992. Efficacy of prolonged (48h) selective enrichment for the detection of foodborne Salmonella. International Journal of Food Microbiology 15, 121-130.

D’Aoust, J.Y., Sewell, A.M., Warburton, D.W., 1992. A comparison of standard ultural methods for the detection of foodborne Salmonella. International Journal of Food Microbiology 16, 41-50.

Dawson, R.C., Cox, J.M., Almond, A., Moses, A., 2001. Food Safety Risk Management in Different Egg Production Systems-A report for the Rural Industries Research and Development Corporation (RIRDC), RIRDC, Australia. 305 de Boer, E., 1998. Update on media for isolation of Enterobacteriaceae from foods. International Journal of Food Microbiology 45, 43-53. de Boer, E., 2000. Enterobacteriaceae, coliforms and E. coli-Classical and modern methods for detection/enumeration. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 610- 617. de Boer, E., Beumer, R.R., 1999. Methodology for detection and typing of foodborne microorganisms. International Journal of Food Microbiology 50, 119-130.

Delves-Broughton, J., Board, R.G., 2000. Microbiology of egg products. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 569-573.

Devenish, J.A., Ciebin, B.W., Brodsky, M.H., 1986. Novobiocin-brilliant green-glucose agar: New medium for isolation of salmonellae. Applied and Environmental Microbiology 52, 539-545.

Drociuk, D., Carnesale, S., Elliot, G., Bell, L.J., Gibson, J.J., Wolf, L., Briggs, D., Jenkins, B., Maillard, J.M., Huddle, M., Virgin, F., Braden, C., Srikantiah, P., Stoica, A., Chiller, T., 2002. Outbreaks of Salmonella serotype Enteritidis infection associated with eating shell eggs-United States, 1999-2001. Morbidity and Mortality Weekly Report 51, 1149-1152.

Duffy, G., Catarame, T., 2005. Tools for safe food. New Food 8, 35-38.

Duguid, J.P., North, R.A.E., 1991. Egg and Salmonella food-poisoning: An evaluation. Journal of Medical Microbiology 34, 65-72.

Dusch, H., Altwegg, M., 1993. Comparison of Rambach agar, SM-ID medium and Hektoen Enteric agar for primary isolation of non-Typhi salmonellae from stool samples. Journal of Clinical Microbiology 31, 410-412.

Dusch, H., Altwegg, M., 1995. Evaluation of five new plating media for isolation of Salmonella species. Journal of Clinical Microbiology 33, 802-804.

Edmiston, A.L., Russell, S.M., 2000. Specificity of a conductance assay for enumeration of Escherichia coli from broiler carcass rinse samples containing genetically similar species. Journal of Food Protection 63, 264-267.

Fagerberg, D.J., Avens, J.S., 1976. Enrichment and plating methodology for Salmonella detection in foods. A review. Journal of Milk and Food Technology 39, 628-646.

Feldsine, P., Abeyta, C., Andrews, W.H., 2002. AOAC international methods committee guidelines for validation of qualitative and quantitative food microbiological official methods of analysis. Journal of AOAC International 85, 1187-1200. 306 Feng, P., 1992. Commercial assay systems for detecting foodborne Salmonella: A review. Journal of Food Protection 55, 927-934.

Fields, M.L., 1979. Fundamentals of Food Microbiology, The AVI Publishing Company, Inc., Westport.

Firstenberg-Eden, R., Foti, D.L., McDougal, S.T., Baker, J., 2002. Optical instrument for the rapid detection of microorganisms in dairy products. International Dairy Journal 12, 225-232.

Firstenberg-Eden, R., Foti, D., McDougal, S., Beck, S., 2004. Performance comparison of the BioSys optical assay and the Violet Red Bile agar method for detecting coliforms in food products. Journal of Food Protection 67, 2760-2766.

Firstenberg-Eden, R., Klein, C.S., 1983. Evaluation of a rapid impedimetric procedure for the quantitative estimation of coliforms. Journal of Food Science 48, 1307-1311.

Firstenberg-Eden, R., Shelef, L.A., 2000. A new rapid automated method for detection of Listeria from environmental swabs and sponges. International Journal of Food Microbiology 56, 231-237. Firstenberg-Eden, R., van Sise, M.L., Zindulis, J., Kahn, P., 1984. Impedimetric estimation of coliforms in dairy products. Journal of Food Science 49, 1449-1452.

Flowers, R.S., 1985. Comparison of rapid Salmonella screening methods and the conventional culture method. Food Technology 39, 103-108.

Forsythe, S.J., Hayes, P.R., 2000. Food Hygiene, Microbiology and HACCP. 3rd ed, Aspen Publishers, Inc., Gaithersburg.

Freydiere, A.-M., Gille, Y., 1991. Detection of salmonellae by using Rambach agar and by a C8 esterase spot test. Journal of Clinical Microbiology 29, 2357-2359.

Fricker, C.R., 1984. A comparison of isolation procedures for salmonellas from polluted water using two forms of Rappaport’s medium. Journal of Applied Bacteriology 56, 305-309.

Fricker, C.R., Quail, E., McGibbon, L., Girdwood, R.W.A., 1985. An evaluation of commercially available dehydrated Rappaport-Vassiliadis medium for the isolation of salmonellae from poultry. Journal of Hygiene 95, 337-344.

Fries, R., Steinhof, U., 1997. Growth kinetics of Salmonella in mixed cultures incubated in Rappaport-Vassiliadis medium. Food Microbiology 14, 505-513.

Fung, D.Y.C., 1995. What’s needed in rapid detection of foodborne pathogens. Food Technology 49, 64-67.

Fung, D.Y.C., 1997. Overview of rapid methods of microbiological analysis. In: Tortorello, 307 M.L., Gendel, S.M. (Eds.), Food Microbiological Analysis New Technologies, Marcel Dekker, Inc., New York, pp. 1-25.

Gabis, D.A., Silliker, J.H., 1974. ICMSF methods studies. II. Comparison of analytical schemes for detection of Salmonella in high-moisture foods. Canadian Journal of Microbiology 20, 663-669.

Garrick, R.C., Smith, A.D., 1994. Evaluation of Rambach agar for the differentiation of Salmonella species from other Enterobacteriaceae. Letters in Applied Microbiology 18, 187-189.

Gibson, A.M., 1988. Use of impedance measurements to estimate numbers of antibiotic resistant Salmonella strains. Letters in Applied Microbiology 6, 89-92.

Giese, J., 1995. Rapid microbiological testing kits and instruments. Food Technology 49, 64-71.

Goepfert, J.M., Mann, M.E., Hicks, R., 1970. One-day fluorescent-antibody procedure for detecting salmonellae in frozen and dried foods. Applied Microbiology 20, 977-983.

Gorman, J.M., Ball, Jr., H.R., 1986. Quality control and product specifications. In: Stadelman, W.J., Cotterill, O.J. (Eds.), Egg Science and Technology, 3rd ed, The AVI Publishing Company, Inc., Westport, pp. 325-344.

Greig, J.D., Ravel, A. 2009. Analysis of foodborne outbreak data reported internationally for source attribution. International Journal of Food Microbiology 130, 77-87

Gruenewald, R., Henderson, R.W., Yappow, S., 1991. Use of Rambach propylene glycol containing agar for identification of Salmonella spp. Journal of Clinical Microbiology 29, 2354-2356.

Hammack, T.S., Andrews, W.H., 2000. Salmonella-Salmonella Enteritidis. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, London, pp. 1937-1943.

Hara-Kudo, Y., Kumagai, S., Masuda, T., Goto, K., Ohtsuka, K., Masaki, H., Tanaka, H., Tanno, K., Miyahara, M., Konuma, H., 2001. Detection of Salmonella Enteritidis in shell and liquid eggs using enrichment and plating. International Journal of Food Microbiology 64, 395-399.

Hargrove, R.E., McDonough, F.E., Reamer, R.J., 1971. A selective medium and presumptive procedure for detecting Salmonella in dairy products. Journal of Milk and Food Technology 34, 6-11.

Harrigan, W.F., McCance, M.E., 1976. Laboratory Methods in Food and Dairy Microbiology, Academic Press, London. 308 Henderson, A.E., Robinson, D.S., 1969. Effect of heat pasteurization on some egg white enzymes. Journal of the Science of Food & Agriculture 20, 755-760.

Hernandez-Herrero, M.M., Roig-Sagues, A.X., Lopez-Sabater, E.I., Rodriguez-Jerez, J.J., Mora-Ventura, M.T., 1999. Influence of storage temperature on the quality of beef liver; pH as a reliable indicator of beer liver spoilage. Journal of the Science of Food and Agriculture 79, 2035-2039.

Herrera, A.G., 2001. Coliforms. In: Spencer, J.F.T., Ragout de Spencer, A.L. (Eds.), Food Microbiology Protocols Methods in Biotechnology, Vol. 14, Humana Press Inc., Totowa, pp. 29-36. Hoben, D.A., Ashton, D.H., Peterson, A.C., 1973a. A rapid, presumptive procedure for the detection of Salmonella in foods and food ingredients. Applied Microbiology 25, 123-129.

Hoben, D.A., Ashton, D.H., Peterson, A.C., 1973b. Some observations on the incorporation of novobiocin into Hektoen Enteric agar for improved Salmonella isolation. Applied Microbiology 26, 126-127.

Hopper, S.A., Mawer, S.L., 1988. Salmonella Enteritidis in a commercial layer flock. The Veterinary Record 123, 351.

Horwitz, W. (Ed.), 2000. Official Methods of Analysis of AOAC International Vol. I, 17th ed, AOAC International, Gaithersburg.

Hugh, R., Leifson, E., 1953. The taxonomic significance of fermentative versus oxidative metabolism of carbohydrates by various Gram-negative bacteria. Journal of Bacteriology 66, 24-26.

Humphrey, T.J., 1990. Public health implications of the infection of egg-laying hens with Salmonella Enteritidis phage type 4. World’s Poultry Science 46, 5-13.

Humphrey, T.J., 1994. Contamination of eggs with potential human pathogens. In: Board, R.G., Fuller, R. (Eds.), Microbiology of the Avian Egg, Chapman & Hall, London, pp. 93-116.

Humphrey, T.J., Whitehead, A., Gawler, A.H.L., Henley, A., Rowe, B., 1991. Numbers of Salmonella Enteritidis in the contents of naturally contaminated hen eggs. Epidemilogy Infection 106, 489-496.

Hussong, D., Damare, J.M., Weiner, R.M., Colwell, R.R., 1981. Bacteria associated with false-positive most-probable-number coliform test results from shellfish and estuaries. Applied and Environmental Microbiology 41, 35.

Ijichi, K., Garibaldi, J.A., Kaufman, V.F., Hudson, C.A., Lineweaver, H., 1973. Microbiology of a modified procedure for cooling pasteurized salt yolk. Journal of

309 Food Science 38, 1241-1243.

Imai, C., 1976. Some characteristics of psychrophilic bacteria isolated from green rotten eggs. Poultry Science 55, 606-610.

Imai, C., Saito, J., 1985. Detection of spoiled eggs using a new-type black light egg inspection unit. Poultry Science 64, 1891-1899.

ICMSF, 1978. Microorganisms in Foods 1-Their Significance and Methods of Enumeration, 2nd ed, University of Toronto Press, Toronto.

ICMSF, 2005. Microorganisms in Foods 6-Microbial Ecology of Food Commodities, 2nd ed, Kluwer Academic/Plenum Publishers, New York.

James, C., Lechevalier, V., Ketteringham, L., 2002. Surface pasteurisation of shell eggs. Journal of Food Engineering 53, 193-197.

Jay, J.M., 2000. Modern Food Microbiology, 6th ed, Aspen Publishers, Inc., Gaithersburg.

Jay, S., Davos, D., Dundas, M., Frankish, E., Lightfoot, D., 2003. Salmonella. In: Hocking, A.D. (Ed.), Foodborne Microorganisms of Public Health Significance, 6th ed, AIFST Incorporated, NSW Branch, Food Microbiology Group, Waterloo DC, pp. 207-266.

Jeffries, L., 1959. Novobiocin-tetrathionate broth: a medium of improved selectivity for the isolation of salmonellae in faeces. Journal of Clinical Pathology 12, 568- 571.

Johansen, C., Gill, T., Gram, L., 1995. Antibacterial effect of protamine assayed by impedimetry. Journal of Applied Bacteriology 78, 297-303.

Jones, D.R., Anderson, K.E., Curtis, P.A., Jones, F.T., 2002. Microbial contamination in inoculated shell eggs: I. Effects of layer strain and hen age. Poultry Science 81, 715- 720.

Jones, P.W., Collins, P., Hayle, A.J., 1984. The effect of sodium sulphacetamide and sodium mandelate in brilliant green agar on the growth of salmonellas. Journal of Applied Bacteriology 57, 423-428.

Joosten, H.M.L.J., van Dijck, W.G.F.M., van der Velde, F., 1994. Evaluation of motility enrichment on modified semisolid Rappaport-Vassiliadis (MSRV) and automated conductance in combination with Rambach agar for Salmonella detection in environmental samples of a milk powder factory. International Journal of Food Microbiology 22, 201-206.

June, G.A., Sherrod, P.S., Hammack, T.S., Amaguana, R.M., Andrews, W.H., 1996. Relative effectiveness of selenite cystine broth, tetrathionate broth, and 310 Rappaport-Vassiliadis medium for recovery of Salmonella spp. from raw flesh, highly contaminated foods, and poultry feed: collaborative study. Journal of AOAC International 79, 1307-1323.

Kahn, P., Firstenberg-Eden, R., 1987. Prediction of shelf-life of pasteurized milk and other fluid dairy products in 48 hours. Journal of Dairy Science 70, 1544-1550.

King, S., Metzger, W.I., 1968. A new plating medium for the isolation of enteric pathogens. I. Hektoen enteric agar. Applied Microbiology 16, 57-578.

Kobayashi, M., Gutierrez, M.A., Hatta, H., 1997. Microbiology of eggs. In: Hen Eggs: Their Basic and Applied Science, Yamamoto, T., Juneja, L.R., Hatta, H., Kim, M. (Eds.), CRC Press, Boca Raton, Florida, pp. 179-191.

Komatsu, K., Restaino, L., 1981. Determination of the effectiveness of novobiocin added to two agar plating media for the isolation of Salmonella from fresh meat products. Journal of Food Safety 3, 183-192.

Kornacki, J.L., Johnson, J.L., 2001. Enterobacteriaceae, coliforms, and Escherichia coli as quality and safety indicators. In: Downes, F.P., Ito, K. (Eds.), Compendium of Methods for the Microbiological Examination of Foods, 4th ed, American Public Health Association, Washington, D.C., pp. 69-82.

Koutsoumanis, K., Nychas, G.-J.E., 2000. Application of a systematic experimental procedure to develop a microbial model for a rapid fish shelf life predictions. International Journal of Food Microbiology 60, 171-184.

Kühne, H., Wonde, B., Rabsch, W., Reissbrodt, R., 1994. Evaluation og Rambach agar for detection of Salmonella subspecies I to VI. Applied and Environmental Microbiology 60, 749-751.

Lach, V.H., Wood, J.M., Jarvis, B., 1978. The rapid estimation of colony count levels in foods by impedance measurements-abstract of Society for Applied Bacteriology meetings. Journal of Applied Bacteriology 45, xxvi-xxvii.

Laycock, R.A., Regier, L.W., 1970. Pseudomonads and Achromobacters in the spoilage of irradiated haddock of different preirradiation quality. Applied Microbiology 20, 333-341.

Leclercq, A., Wanegue, C., Baylac, P., 2002. Comparison of fecal coliforms agar and violet red bile lactose agar for fecal coliform enumeration in foods. Applied and Environmental Microbiology 68, 1631-1638. le Minor, L., 1992. The genus Salmonella. In: Balows, A., Truper, H.G., Dworkin, M., Harder, W., Schleifer, K.-H. (Eds.), The Prokaryotes-a Handbook on the Biology of Bacteria: Ecophysiology, Isolation, Identification, Applications Vol. 3, Springer- Verlag, New York, pp. 2760-2774. 311 Liao, C.-H., Fett, W.F., 2003. Isolation of Salmonella from alfafa seed and demonstration of impaired growth of heat-injured cells in seed homogenates. International Journal of Food Microbiology 82, 245-253.

Licciardello, J.J., Nickerson, J.T.R., Goldblith, S.A., 1970. Recovery of salmonellae from irradiated foods. Journal of Food Science 35, 620-624.

Maijala, R., Johansson, T., Hiern, J., 1992. Growth of Salmonella and competing flora in five commercial Rappaport-Vassiliadis (RV) media. International Journal of Food Microbiology 17, 1-8.

Manafi, M., 1996. Fluorogenic and chromogenic enzyme substrates in culture media and identification tests. International Journal of Food Microbiology 31, 45-48.

Manafi, M., 2000. New developments in chromogenic and fluorogenic culture media. International Journal of Food Microbiology 60, 205-218.

Manafi, M., Sommer, R., 1992. Comparison of three rapid screening methods for Salmonella spp.: ‘MUCAP test, MicroScreenReg. Latex and Rambach agar’. Letters in Applied Microbiology 14, 163-166.

Manafi, M., Willinger, B., 1994. Comparison of three rapid methods for identification of Salmonella spp. Letters in Applied Microbiology 19, 328-331.

Manzano, M., Cocolin, L., Astori, G., Pipan, C., Botta, G.A., Cantoni, C., Comi, G., 1998. Development of a PCR microplate-capture hybridization method for simple, fast and sensitive detection of Salmonella serovars in food. Molecular and Cellular Probes 12, 227-234.

Maucci, E., 2005. The benefits of automated pathogen testing. New Food 8, 33-36.

Mbandi, E., Shelef, L.A., 2002. Automated measurements of antilisterial activities of lactate and diacetate in ready-to-eat meat. Journal of Microbiological Methods 49, 307-314.

McGibbon, L., Quail, E., Fricker, C.R., 1984. Isolation of salmonellae using two forms of Rappaport-Vassiliadis medium and brilliant green agar. International Journal of Food Microbiology 1, 171-177.

McKeon, D.M., Calabrese, J.P., Bissonnette, G.K., 1995. Antibiotic resistant Gram- negative bacteria in rural groundwater supplies. Water Research 29, 1902-1908.

McMahon, W.A., Schultz, A.M., Johnson, R.L., 2004. Evaluation of VIDAS® Salmonella (SLM) Immunoassay method with Rappaport-Vassiliadis (RV) medium for detection of Salmonella in foods: Collaborative study. Journal of AOAC International 87, 867-883.

312 Mellado, V., Yanez, M.A., Catalan, V., 2006. Evaluation of the MicroFoss system for the analysis of Escherichia coli in water. Microbiological Research 161, 20-24.

Mendonca, A.F., Juneja, V.K., 2000. Total viable counts-Metabolic activity tests. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, San Diego, pp. 2168-2176.

Merck, 2002. Acid-base indicators. In: Baden, W. (Ed.), Chemicals Reagents 2002, Merck KGaA, Darmstadt.

Messer, J.W., Rice, E.W., Johnson, C.H., 2000. Total viable count-Spread plate technique. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, San Diego, pp. 2159-2160.

Miller, V.R., Banwart, G.J., 1965. Effect of various concentrations of brilliant green and bile salts on salmonellae and other microorganisms. Applied Microbiology 13, 77-80.

Miller, R.G., Mallinson, E.T., 2000. Improved detection of nontyphoid and typhoid salmonellae with balanced agar formulations. Journal of Food Protection 63, 1443- 1446.

Miller, R.G., Tate, C.R., Mallison, E.T., Scherrer, J.A., 1991. Xylose-lysine-tergitol 4: an improved selective agar medium for the isolation of Salmonella. Poultry Science 70, 2429-2432. Miller, R.G., Tate, C.R., Mallinson, E.T., 1995. Improved XLT4 agar: small addition of peptone to promote stronger production of hydrogen-sulfide by salmonellae. Journal of Food Protection 58, 115-119.

Moats, W.A., 1978. Comparison of four agar plating media with and without added novobiocin for isolation of salmonellae from beef and deboned poultry meat. Applied and Environmental Microbiology 36, 747-751.

Moats, W.A., 1981. Update on Salmonella in foods: Selective plating media and other diagnostic media. Journal of Food Protection 44, 375-380.

Moats, W.A., Kinner, J.A., 1974. Factors affecting selectivity of brilliant green-phenol red agar for salmonellae. Applied Microbiology 27, 118-123.

Moats, W.A., Kinner, J.A., 1976. Observations on brilliant green agar with an H2S indicator. Applied and Environmental Microbiology 31, 380-384.

Monfort, P., le Gal, D., Saux, J.C. le, Piclet, G., Raguenes, P., Boulben, S., Plusquellec, A., 1994. Improved rapid method for isolation and enumeration of Salmonella from bivalves using Rambach agar. Journal of Microbiological Methods 19, 67- 79.

313 Monnery, I., Freydiere, A.M., Baron, C., Rousset, A.M., Tigaud, S., Boude-Chevalier, M., de Montclos, H., Gille, Y., 1994. Evaluation of two new chromogenic media for detection of Salmonella in stools. European Journal of Clinical Microbiology and Infectious Diseases 13, 257-261.

Montgomery, D.C., Runger, G.C., Hubele, N.F., 2004. Engineering Statistics, 3rd ed, John Wiley & Sons, Inc., New York.

Montville, R., Schaffner, D.W., 2004. Statistical distributions describing microbial quality of surfaces and foods in food service operations. Journal of Food Protection 67, 162-167.

Moore, J., Madden, R.H., 1993. Detection and incidence of Listeria species in blended raw egg. Journal of Food Protection 56, 652-654, 660.

Morinigo, M.A., Martinez-Manzanares, E., Munoz, A., Cornax, R., Romero, P., Borrego, J.J., 1989. Evaluation of different plating media used in the isolation of salmonellas from environmental samples. Journal of Applied Bacteriology 66, 353-360.

Morinigo, M.A., Munoz, M.A., Martinez-Manzanares, E., Sanchez, J.M., Borrego, J.J., 1993. Laboratory study of several enrichment broths for the detection of Salmonella spp. particularly in relation to water samples. Journal of Applied Bacteriology 74, 330-335.

Morris, G.K., Dunn, C.G., 1970. Influence of incubation temperature and sodium heptadecyl sulfate (Tergitol No. 7) on the isolation of salmonellae from pork sausage. Applied Microbiology 20, 192-195.

Morse, D.L., Birkhead, G.S., Guardino, J., Kondracki, S.F., Guzewich, J.J., 1994. Outbreak and sporadic egg-associated cases of Salmonella Enteritidis: New York’s experience. American Journal of Public Health 84, 859-860.

Morton, R.D., 2001. Aerobic plate count. In: Downes, F.P., Ito, K. (Eds.), Compendium of Methods for the Microbiological Examination of Foods, 4th ed, American Public Health Association, Washington, D.C., pp. 63-67.

National Advisory Committee on Microbiological Criteria for Foods, 1998. Hazard Analysis and Critical Control Point principles and application guidelines. Journal of Food Protection 61, 1246-1259.

Niemi, R.M., Mentu, J., Siitonen, A., Niemela, S.I., 2003. Confirmation of Escherichia coli and its distinction from Klebsiella species by gas and indole formation at 44 and 44.5°C. Journal of Applied Microbiology 95, 1242-1249.

Odumeru, J.A., Belvedere, J., 2002. Evaluation of the MicroFoss system for enumeration of total viable count, Escherichia coli and coliforms in ground beef. Journal of 314 Microbiological Methods 50, 33-38.

Odumeru, J.A., Belvedere, J., van Donkersgoed, C., 2004. Evaluation of the MicroFoss system for the detection of Listeria species in environmental samples. Journal of Microbiological Methods 57, 1-7.

Ogden, I.D., Cann, D.C., 1987. A modified conductance medium for the detection of Salmonella spp. Journal of Applied Bacteriology 63, 459-464.

Pandey, A., Joshi, V.K., Nigam, P., Soccol, C.R., 2000. Enterobacteriaceae, coliforms and E. coli-Introduction. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 604- 610.

Peng, H., Shelef, L.A., 2000. Rapid detection of low levels of Listeria in foods and next- day confirmation of L. monocytogenes. Journal of Microbiological Methods 41, 113-120.

Peng, H., Shelef, L.A., 2001. Automated simultaneous detection of low levels of listeriae and salmonellae in foods. International Journal of Food Microbiology 63, 225-233.

Perez, J.M., Cavalli, P., Roure, C., Renac, R., Gille, Y., Freydiere, A.M., 2003. Comparison of four chromogenic media and Hektoen agar for detection and presumptive identification of Salmonella strains in human stools. Journal of Clinical Microbiology 41, 1130-1134.

Peterz, M., Wiberg, C., Norberg, P., 1989. The effect of incubation temperature and magnesium chloride concentration on growth of Salmonella in home-made and in commercially available dehydrated Rappaport-Vassiliadis broths. Journal of Applied Bacteriology 66, 523-528.

Pettipher, G.L., Watts, Y.B., 1989. Evaluation of lysine-iron-cystine-neutral red for the detection of salmonellas in the bactometer. Letters in Applied Microbiology 9, 243- 244.

Pignato, S., Giammanco, G., Giammanco, G., 1995a. Rambach agar and SM-ID medium sensitivity for presumptive identification of Salmonella subspecies I-VI. Journal of Medical Microbiology 43, 68-71.

Pignato, S., Marino, A.M., Emanuele, M.C., Iannotta, V., Caracappa, S., Giammanco, G., 1995b. Evaluation of new culture media for rapid detection and isolation of salmonellae in foods. Applied and Environmental Microbiology 61, 1996-1999.

Pless, P., Futschik, K., Schopf, E., 1994. Rapid detection of salmonellae by means of a new Impedance-Splitting method. Journal of Food Protection 57, 369-376.

315 Poisson, D.M., 1992. Novobiocin, brilliant green, glycerol, lactose agar: a new medium for the isolation of Salmonella strains. Research in Microbiology 143, 211-216.

Poisson, D.M., Nugier, J.P., Rousseau, P., 1993. Study of Rambach and NBGL agar on 4037 stools of human origin and 584 veterinary samples submitted for isolation of salmonellae. Pathologie Biologie 41, 543-546.

Poisson, D.M., Nugier, J.P., Rousseau, P., 1993. Study of Rambach and NBGL agar on 4037 stools of human origin and 584 veterinary samples submitted for isolation of salmonellae. Pathologie Biologie 41, 543-546.

Poppe, C., Elliot, L.A., Duncan, C.L., 1996. Evaluation of immunomagnetic separation in combination with modified semi-solid Rappaport Vassiliadis medium and Rambach agar for the isolation of Salmonella. Journal of Microbiological Methods 25, 237-244.

Powrie, W.D., Nakai, S., 1986. The chemistry of eggs and egg products. In: Stadelman, W.J., Cotterill, O.J. (Eds.), Egg Science and Technology, 3rd ed, MacMillan Publishers, Westport, pp. 97-139.

Presland, F., 2005. Microbial update-faecal indicators. International Food Hygiene 16, 20- 21.

Price, W.R., Olsen, R.A., Hunter, J.E., 1972. Salmonella testing of pooled pre- enrichment broth cultures for screening multiple food sample. Applied Microbiology 23, 679-682.

Priego, R., Medina, L.M., Jordano, R., 2000. Evaluation of Petrifilm series 2000 as a possible rapid method to count coliforms in foods. Journal of Food Protection 63, 1137-1140.

Radford, S.A., Board, R.G., 1993. Review: Fate of pathogens in home-made mayonnaise and related products. Food Microbiology 10, 269-278.

Radkowski, M., 2001. Occurrence of Salmonella spp. in consumption eggs in Poland. International Journal of Food Microbiology 64, 189-191.

Rall, V.L.M., Rall, R., Aragon, L.C., Silva, M.G. da, 2005. Evaluation of three enrichment broths and five plating media for Salmonella detection in poultry. Brazilian Journal of Microbiology 36, 147-150.

Rambach, A., 1990. New plate medium for facilitated differentiation of Salmonella spp. from Proteus spp. and other enteric bacteria. Applied and Environmental Microbiology 56, 301-303.

Rappaport, F., Konforti, N., Navon, B., 1956. A new enrichment medium for certain salmonellae. Journal of Clinical Pathology 9, 261-266. 316 Rappaport, F., Konforti, N., 1959. Selective enrichment medium for paratyphoid bacteria. Inhibitory and growth promoting factors. Applied Microbiology 7, 63- 66.

Rappold, H., Bolderdijk, R.F., 1979. Modified lysine iron agar for isolation of Salmonella from food. Applied and Environmental Microbiology 38, 162-163.

Ray, B., 2004. Fundamental Food Microbiology, 3rd ed, CRC Press LLC, Boca Raton.

Ray, B., Speck, M.L., 1978. Plating procedure for the enumeration of coliforms from dairy products. Applied and Environmental Microbiology 35, 820-822.

Reamer, R.H., Hargrove, R.E., McDonough, F.E., 1974. A selective plating agar for direct enumeration of Salmonella in artificially contaminated dairy products. Journal of Milk and Food Technology 37, 441-444.

Reporter, R., Mascola, L., Kilman, L., Medina, A., Mohle-Boetani, J., Farrar, J., Vugia, D., Fletcher, M., Levy, M., Ravenholt, O., Empey, L., Maxson, D., Klouse, P., Bryant, A., Todd, R., Williams, M., Cage, G., Bland, L., 2000. Outbreaks of Salmonella serotype Enteritidis infection associated with eating raw or undercooked shell eggs-United States, 1996-1998. Journal of the American Medical Association 283, 1132-1134.

Restaino, L., Grauman, G.S., McCall, W.A., Hill, W.M., 1977. Effects of varying concentrations of novobiocin incorporated into two Salmonella plating media on the recovery of four Enterobacteriaceae. Applied and Environmental Microbiology 33, 585-589.

Restaino, L., Komatsu, K.K., Syracuse, M.J., 1982. A note on novobiocin in XLD and HE agars: the optimum levels required in two commercial sources of media to improve isolation of salmonellas. Journal of Applied Bacteriology 53, 285-288.

Ricke, S.C., Birkhold, S.G., Gast, R.K., 2001. Eggs and egg products. In: Downes, F.P., Ito, K. (Eds.), Compendium of Methods for the Microbiological Examination of Foods, 4th ed, American Public Health Association, Washington, D.C., pp. 473-481.

Ricke, S.C., Pillai, S.D., Norton, R.A., Maciorowski, K.G., 1998. Applicability of rapid methods for detection of Salmonella spp. in poultry feeds: A review. Journal of Rapid Methods and Automation in Microbiology 6, 239-258.

Rodriguez, D.C., Tauxe, R.V., Rowe, B., 1990. International increase in Salmonella Enteritidis: A new pandemic. Epidemiology and Infection 105, 21-27.

Ruiz, J., Nunez, M.L., Lorente, I., Perez, J., Simarro, E., Gomez, J., 1996. Performance of six culture media for isolation of Salmonella species from stool samples. European Journal of Clinical Microbiology and Infectious Diseases 15, 922- 926. 317 Russell, S.M., 2000. Comparison of the traditional three-tube most probable number method with the Petrifilm, SimPlate, BioSys optical and Bactometer conductance methods for enumerating Escherichia coli from chicken carcasses and ground beef. Journal of Food Protection 63, 1179-1183.

Russell, S.M., 2001. Evaluation of an optical microbiological method for rapidly estimating populations of aerobic bacteria, coliforms, and Escherichia coli from ground pork. Journal of Food Protection 64, 669-673.

Russell, S.M., Fletcher, D.L., Cox, N.A., 1995. Comparison of media for determining temperature abuse of fresh broiler carcasses using impedance microbiology. Journal of Food Protection 58, 1124-1128.

Salvat, G., Blivet, D., 2000. Electrical techniques-Food spoilage flora and total viable count (TVC). In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 578-580.

Sauter, E.A., Petersen, C.F., 1969. The effect of egg shell quality on penetration by Pseudomonas fluorescens. Poultry Science 48, 1525-1528.

Schönenbrücher, V., Mallinson, E.T., Bülte, M., 2008. A comparison of standard cultural methods for the detection of foodborne Salmonella species including three new chromogenic plating media. International Journal of Food Microbiology 123, 61-66.

Schultz, S., Morse, D., Parkin, W., Grady, G.F., Witte, E.J., Hadler, J.L. Vogt, R.L., Schwartz, E., Gensheimer, K.F., Silverman, P.R., Israel, E., 1987. Update: Salmonella Enteritidis infections in the Northeastern United Staes. Morbidity and Mortality Weekly Report 36, 204-205.

Seo, K.H., Holt, P.S., Stone, H.D., Gast, R.K., 2003. Simple and rapid methods for detecting Salmonella Enteritidis in raw eggs. International Journal of Food Microbiology 87, 139-144.

Seo, K.H.,, Valentin-Bon, I.E., Brackett, R.E., Holt, P.S., 2004. Rapid, specific detection of Salmonella Enteritidis in pooled egg by real-time PCR. Journal of Food Protection 67, 864-869.

Shafi, R., Cotterill, O.J., Nichols, M.L., 1970. Microbial flora of commercially pasteurised egg products. Poultry Science 49, 578-585.

Shaw, B.G., Latty, J.B., 1982. A numerical taxonomic study of Pseudomonas strains from spoiled meat. Journal of Applied Bacteriology 52, 219-228.

Shelef, L.A., Eden, G., 1996. Optical instrument rapidly detects and enumerates microorganisms in food. Food Technology 50, 82-85.

318 Shelef, L.A., Firstenberg-Eden, R., 1997. Novel selective and non-selective optical detection of microorganisms. Letters in Applied Microbiology 25, 202-206.

Shelef, L.A., Mohammed, S., Tan, W., Webber, M.L., 1997. Rapid optical measurements of microbial contamination in raw ground beef and effects of citrate and lactate. Journal of Food Protection 60, 673-676.

Shelef, L.A., Surtani, A., Kanagapandian, K., Tan, W., 1998. Automated detection of amino acid decarboxylation in salmonellae and other Enterobacteriaceae. Food Microbiology 15, 199-205. Shelef, L.A., Tan, W., 1998. Automated detection of hydrogen sulfide release from thiosulfate by Salmonella spp. Journal of Food Protection 61, 620-622.

Sherrod, P.S., Amaguana, R.M., Andrews, W.H., June, G.A., Hammack, T.S., 1995. Relative effectiveness of selective plating agars for recovery of Salmonella species from selected high-moisture foods. Journal of AOAC International 78, 679-690.

Silbernagel, K.M., Lindberg, K.G., 2002. Evaluation of the 3M Petrifilm Enterobacteriaceae count plate method for the enumeration of Enterobacteriaceae in foods. Journal of Food Protection 65, 1452-1456.

Silk, T.M., Ryser, E.T., Donnelly, C.W., 1997. Comparison of methods for determining coliforms and Escherichia coli levels in apple cider. Journal of Food Protection 60, 1302-1305. Smyser, C.F., Snoeyenbos, G.H., 1969. Evaluation of several methods of isolating salmonellae from poultry litter and animal feedingstuffs. Avian Diseases 13, 134- 141.

Solomon, S.E., Bain, M.M., Cranstoun, S., Nascimento, V., 1994. Hen’s egg shell structure and function. In: Board, R.G., Fuller, R. (Eds.), Microbiology of the Avian Egg, Chapman & Hall, London, pp. 1-24.

Sparks, N.H.C., 2000. Microbiology of fresh eggs. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 1, Academic Press, San Diego, pp. 563-569.

Spino, D.F., 1966. Elavated temperature technique for the isolation of Salmonella from streams. Applied and Environmental Microbiology 14, 591-596.

Spino, D.F., Geldreich, E.E., 1981. Acinetobacter spp.: distinct morphology on Eosin Methylene Blue agar as an aid to identification in drinking water. Applied and Environmental Microbiology 41, 1063-1064.

Srikaeo, K., Hourigan, J.A., 2002. The use of statistical process control (SPC) to enhance the validation of critical control points (CCPs) in shell egg washing. Food Control 13, 263-273.

319 Stadelman, W.J., 1994. Contaminants of liquid egg products. In: Board, R.G., Fuller, R. (Eds.), Microbiolgy of the Avian Egg, Chapman & Hall, London, pp. 139-151.

Stallard, K.R., Cox, J.M., 1994. Lysine mannitol glycerol agar (LMG) and LMG with sulphamandelate for isolation of Salmonella spp. from clinical specimens. Letters in Applied Microbiology 19, 83-87.

Steegmans, W.J.L., 1994/1995. Rapid detection of foodborne pathogens. Food Tech Europe 1, 132-134.

Steinert, L., Virgil, D., Bellemore, E., Williamson, B., Dinda, E., Harris, D., Scheider, D., Fanella, L., Bogacki, V., Liska, F., Birkhead, G.S., Guzewich, J.J., Fudala, J.K., Kondracki, S.F., Shayegani, M., Morse, D.L., Dennis, D.T., Healey, B., Tavris, D.R., Duffy, M., Drinnen, K., Hutcheson, R.H., 1990. Update: Salmonella Enteritidis infections and grade A shell eggs-United States, 1989. Morbidity and Mortality Weekly Report 38, 877-880.

Swaminathan, B., Denner, J.M., Ayres, J.C., 1978. Rapid detection of salmonellae in foods by membrane filter-disc immunoimmobilization technique. Journal of Food Science 43, 1444-1447.

Swanson, K.M.J., Petran, R.L., Hanlin, J.H., 2001. Culture methods for enumeration of microorganisms. In: Downes, F.P., Ito, K. (Eds.), Compendium of Methods for the Microbiological Examination of Foods, 4th ed, American Public Health Association, Washington, D.C., pp. 53-62.

Tan, W., Shelef, L.A., 1999. Automated detection of Salmonella spp. in foods. Journal of Microbiological Methods 37, 87-91.

Taylor, W.F., 1965. Isolation of shigellae. I. Xylose-lysine agars: new media for the isolation of enteric pathogens. American Journal of Clinical Pathology 44, 471-475.

Taylor, W.I., Schelhart, D., 1971. Isolation of shigellae. VIII. Comparison of xylose lysine desoxycholate agar, Hektoen enteric agar, Salmonella-Shigella agar, and eosin methylene blue agar with stool specimens. Applied Microbiology 21, 32- 37.

Telo, A., Bijo, B., Sulaj, K., Beli, E., 1999. Occurrence of Salmonella spp. in imported eggs into Albania. International Journal of Food Microbiology 49, 169-171.

Tortorello, M.L., 2003. Indicator organisms for safety and quality-uses and methods for detection: minireview. Journal of AOAC International 86, 1208-1217.

Tranter, H.S., Board, R.G., 1984. The influence of incubation temperature and pH on the antimicrobial properties of hen egg albumen. Journal of Applied Bacteriology 56, 53-61.

320 Truscott, R.B., Lammerding, A.M., 1987. Millipore filtration and use of RV medium for isolation of Salmonella from preenrichment broths. Journal of Food Protection 50, 815-819.

Ueda, S., Umesako, S., Mineno, J., Kuwabara, Y., 2000. The magnetic immuno polymerase chain reaction assay for detection of Salmonella from food and fecal samples. Biocontrol Science 5, 25-32.

Upmann, M., Bonaparte, C., 2000. Rapid methods for food hygiene inspection. In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, London, pp. 1887-1895.

Vadehra, D.V., Baker, R.C., 1973. Effect of egg shell sweating on microbial spoilage of chicken eggs. Journal of Milk and Food Technology 36, 321-322.

Vadehra, D.V., Baker, R.C., Naylor, H.B., 1970. Infection routes of bacteria into chicken eggs. Journal of Food Science 35, 61-62.

Vaidya, V.M., Paturkar, A.M., Waskar, V.S., Zende, R.J., Rawool, D.B., 2005. Detection of indicator organisms on poultry carcass sites in an organised slaughterhouse. Journal of Muscle Foods 16, 289-297. van Schothorst, M., Renaud, A., van Beek, C., 1987. Salmonella isolation using RVS broth and MLCB agar. Food Microbiology 4, 11-18.

Vasavada, P.C., 1993. Rapid methods and automation in dairy microbiology. Journal of Dairy Science 76, 3101-3113.

Vassiliadis, P., 1983. The Rappaport-Vassiliadis enrichment media for the isolation of salmonellas:An overview. Journal of Applied Bacteriology 54, 69-76.

Vassiliadis, P., Kalapothaki, V., Trichopoulos, D., 1991. Isolation of Salmonella from fluid milk with the use of Rappaport-Vassiliadis medium. Journal of Food Protection 54, 421-423.

Vassiliadis, P., Mavrommati, C.H., Kalopathaki, V., Chronas, G., Efstratiou, M., 1985. Salmonella isolation with Rappaport-Vassiliadis enrichment medium seeded with different sized inocula of pre-enrichment cultures of meat products and sewage polluted water. Journal of Hygiene 95, 139-147.

Wang, H., Blais, B.W., Yamazaki, H., 1995. Rapid and economical detection of Salmonella Enteritidis in eggs by the polymyxin-cloth enzyme immunoassay. International Journal of Food Microbiology 24, 397-406.

Watkins, P., 1999. Keeping pathogens under control. Food Processing UK 68, 32-33.

Watson, D.C., Walker, A.P., 1978. A modification of brilliant green agar for improved 321 isolation of Salmonella. Journal of Applied Bacteriology 45, 195-204.

White, C.H., Wilson, J., Schilling, M.W., 2006. An investigation of the use of the MicroFoss as an indicator of the shelf life of pasteurized fluid milk. Journal of Dairy Science 89, 2459-2464.

Williams, M.G., Busta, F.F., 2000. Total viable counts-Most probably number (MPN). In: Robinson, R.K., Batt, C.A., Patel, P.D. (Eds.), Encyclopedia of Food Microbiology, Vol. 3, Academic Press, San Diego, pp. 2166-2168.

Yadav, N.K., Vadehra, D.V., 1977. Mechanism of egg white resistance to bacterial growth. Journal of Food Science 42, 97-99.

Yan, Z., Vorst, K.L., Zhang, L., Ryser, E.T., 2007. Use of one-ply composite tissues in an automated optical assay for recovery of Listeria from food contact surfaces and poultry-processing environments. Journal of Food Protection 70, 1263-1266.

Department of Health and Ageing, 2007. OzFoodNet Quarterly report 1 October to 31

December2007. Available: http://www.health.gov.au/internet/main/Publishing.nsf/Content/cda-cdi3201o.htm. [19-Aug-2008].

Department of Health and Ageing, 2008a. OzFoodNet Quarterly report 1 January to 31

March 2008. Available: http://www.health.gov.au/internet/main/publishing.nsf/Content/cda-cdi3202-pdf- cnt.htm/$FILE/cdi3202k.pdf. [19-Aug-2008].

Department of Health and Ageing, 2008b. Large outbreaks of Salmonella Typhimurium

phage type 135 infections associated with the consumption of products containing raw egg in Tasmania. Available: http://www.health.gov.au/internet/main/Publishing.nsf/Content/cda- cdi3101h.htm#top. [19-Aug-2008].

Food Science Australia, 2008. Food safety issue update: Salmonella in food containing raw or undercooked eggs-information for the food service industry August 2008. Available:

Food Standards Australia New Zealand, 2002b. Food Standard Code-Standard 1.6.2 Processing requirements. Available: http://www.foodstandards.gov.au/foodstandardscode/standard16/standard162.cfm. [15-Apr-2002].

Food Standards Australia New Zealand, 2006. Initial assessment report-Proposal P301 Primary production and processing standard for eggs and egg products. 322 Available: http://www.foodstandards.gov.au/_srcfiles/IAR_P301_Eggs_PPPS.pdf. [20-Feb-2009].

Food Standards Australia New Zealand, 2009. Standard 1.6.1 Microbiological limits for food. Available: http://www.foodstandards.gov.au/_srcfiles/Standard_1_6_1_Micro_v103.pdf. [01- Mar-2009].

Neogen Corporation, 2008. General microbiology-The SolerisTM system. Available: http://www.neogen.com. [10-Apr-2008].

NSW Food Authority, 2008. Improper use of raw eggs linked to food poisoning. Available:

http://www.foodauthority.nsw.gov.au/aboutus/media-releases/mr-26-dec-improper- use-of-raw-eggs-food-poisoning/ [19-Aug-2008].

Tasmanian Department of Health & Human Services, 2008. Salmonella and eggs –

important information for restaurants, takeaways and delicatessens. Available: http://www.dhhs.tas.gov.au/__data/assets/pdf_file/0017/31184/salmonella_info_for _restaurants_takeaways_and_delis_May_08.pdf. [19-Aug-2008].

US Department of Agriculture, 1999. The USDA regulations governing the voluntary grading of shell eggs. Available: http://www.ams.usda.gov/poultry/regulations/56CFR10-25-99.html. [03-Aug-2000].

323 APPENDIX 10.1

MgCl2.6H2O is hygroscopic and its density is 1.57 g/mL (20°C).

Working MgCl2.6H2O solution (40% w/v) was prepared by mixing 40g of MgCl2.6H2O, which is equivalent to 25.48 mL, with 100 mL of distilled water, giving a total volume of

125.48 mL.

A total volume of 125.48 mL of working MgCl2.6H2O solution contains 40g of

MgCl2.6H2O. Therefore, the volume of 100 mL of working MgCl2.6H2O solution contains

31.88g of MgCl2.6H2O.

Concurrently, 0.1% working brilliant green solution was prepared by mixing 0.1g of brilliant green with 100 mL of distilled water. Hence, 5 mL of 0.1% working brilliant green solution contained 0.0050g of brilliant green.

In each Experiment, volumes of 40% working MgCl2.6H2O and 0.1% working brilliant green solutions were correspondingly calculated to achieve their expected concentrations.

In order to obtain 2.90% (w/v) of MgCl2.6H2O and 0.00047% of brilliant green in the medium, it required 100.70 mL of 40% working MgCl2.6H2O solution and 5.20 mL of

0.1% working brilliant green solutions, which was equivalent to 32.10g of MgCl2.6H2O and

0.0052g of brilliant green, respectively.

Hence, a total volume of 1105.90 mL of the medium (1000 mL of sterile distilled water,

100.70 mL of 40% working MgCl2.6H2O solution and 5.20 mL of 0.1% working brilliant green solution), containing 32.10g of MgCl2.6H2O and 0.0052g of brilliant green are equivalent to 2.90% (w/v) or 29 g/L of MgCl2.6H2O and 0.00047% or 4.7mg/L of brilliant green, respectively.

324 APPENDIX 10.2

Working novobiocin solution was prepared by aseptically dissolving 20mg of novobiocin powder in 1 mL of sterile distilled water. Subsequently, working novobiocin solution was sterilised by filtration through a 0.22-μm membrane filter.

Note: Sterilised novobiocin solution must be added to the sterile medium only after the medium was cooled down to room temperature.

As it was aimed to add 5, 10, 20, 40 and 80 mg of novobiocin to 1000 mL of the medium, a total volume of 1105.90 mL of the original formulation therefore requires approximately

5.53, 11.06, 22.12, 44.24 and 88.47mg of novobiocin, respectively.

Every 1 mL of working novobiocin solution contains 20mg of novobiocin. Hence, it requires approximately 0.28, 0.55, 1.11, 2.21 and 4.42 mL of working novobiocin solution for supplementation to the original formula to achieve 5.53, 11.06, 22.12, 44.24 and

88.47mg of novobiocin in 1105.90 mL of the medium, in equivalence to 5, 10, 20, 40 and

80mg of novobiocin, respectively in 1000 mL of the medium.

APPENDIX 10.3

It was aimed to add 1, 2 and 3 mL of Niaproof 4 to 1000 mL of the medium, a total volume of 1105.90 mL of the original formulation therefore requires approximately 1.11, 2.21 and

3.32 mL of Niaproof 4, respectively.

Niaproof 4 (sodium tetradecylsulfate solution) was added to the medium as is and it must be added after the other ingredients were completely dissolved in distilled water prior to sterilisation by boiling.

325