APPLICATION AND EVALUATION OF BACTERIAL WRUSES IN RAPID
METHODOLOGIES FOR THE DETECTION OF FOOD-BORNE PATHOGENS
A Thesis
Presented to
nie Faculty of Graduate Studies
of
The University of Guelph
In partial fulfilment of requirements
for the degree of
Doctor of Philosophy
July, 1998
O Lynn McIntyre, 1998 National Library Bibliothèque nationale 1*1 ofCamda du Canada Acquisitions and Acquisitions et Bibliographie Services services bibliographiques 395 Weilingîon Street 395. rue Wellington OitawaON K1AON4 Ottawa ON K1A ON4 canada Canada
The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant a la National Lîbrary of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sell reproduire, prêter, distribuer ou copies of this thesis in microfom, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/nlm, de reproduction sur papier ou sur format électronique.
The author retaios ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. ABSTRACT
APPLICATION AND EVALUATION OF BACTERIAL VIRUSES IN RAPID METHODOLOGIES FOR THE DETECTION OF FOOD-BORNE PATHOGENS
Lynn McIntyre Advisor: University of Guelph, 1998 Dr. M. W. Griffrths
Bacteriophages (or phages) are viruses that replicate only by identieing and idecting specific host bacteria. This property has facilitated their application in microbial typing schemes, and more recently in various bacterial detection methods. However, lack of specificity and sensitivity, dong with time-consuming and costly procedures have hindered their application in species-specific pathogen detection. This thesis will address these issues with regard to development of phage-based methods for the detection of Listeria rnonocytogenes, Salmonella spp., and Escherichia di.
A number of commercial and environmentally-isolated phages were first evaluated, based on host specificity ranges. Four phages were selected for Merinvestigation, two demonstrating broad-range specificity for Lisieria and Salmonella spp., and two exhibiting specificity for S. Enteritidis and non-VTEC (verotoxigenic E. di)isolates. Transmission electron microscopy was used to classiQ each virus into a specific group based on tail length and head size, and their mode of infection was elucidated.
Pathogen detection methods were chosen, based on phage properties. First, phage replication requires an actively metabolising host, making metabolic methods, including impedance. turbidirnetry and colorimetry, ideal candidates for evaluation of bactenophage behaviour. Additionally, phage lytic properties make them potentiaily specific biological extractânts for ATP bioluminescence, whic h ty picaily uses a non-speci fic chemicai extraction,
Phage-based ATP bioluminescence was successful in specificaily identiQing pathogens, but poor sensitivity was a problem. Filtration, increased phage exposure Meand an ATP amplification method were dl evaluated as means of improving sensitivity, but were limited in their application. Phage-mediated impedimeûic and colorimetric methods were developed for the sensitive and specific detection and confirmation of bovine non-VTEC isolates in raw milk (40 CFU/mL) within 12 to 16 hours. Colorimetric analysis of artificially contarninated ground beef in combination with phage AT20 was capable of confirming 4000 CFUIg of E. coli G2-2. Phage-based irnpedimetric assays were aiso developed for differentiai detection of SalmonelZu spp. in skim milk powder, and L. rnonocytogenes in raw milk. Phage-based turbidimetric pathogen assays were also demonstrated, but diluting samples to reduce initiai turbidity detracted from the sensitivity and usefùlness of the assay. This dissertation is dedicated to the mernories of
Isobel McMillan Calder
Jean Greenhalgh
John Phillips
Craig Thornley
al1 of whom died before tlieir prime, and before witnessing the successfu1 completion of my doctorate.
1 miss you al1 very much. Hrnmm ... where to begin?
First of ail, 1 would like to extend my doepest thanks and admiration to my supervisor, Dr.
Mansel W. Griffiths. Not only did he encourage me to pursue an interest in scientific
research. but he ailowed me the opportunity to broaden my horizons by moving to Canada to undertake graduate studies with him at the University of Guelph. I have learned a great deal about science fiom him, but more than that, he has unfailingly demonstrated how to be a fair and decent individual. Thank you Mansel for your guidance, support, patience and humour (the latter being of immense important when one is a Welsh rugby fan).
Many thanks to Dr. Heidi Schraft and Dr. Joseph Odumeru for their constructive cnticisms of this dissertation, their attention to detail, and their arnazing endurance in reading this entire piece of work fiom cover to cover under a tight deadline. Thanks also to Dr.
Massimo Marcone, Dr. Yukio Kakuda, and my extemal examiner Dr. Michael Brodsky for making my defence experience such a rewarding ordeal.
1 can't Say enough about the many wondemil people 1 have had the opportunity to work with over the course of my Ph.D. There are of course too many to narne but honourable mentions go to Ann Toner, Veena Kau., Yolanda HWi, Doug McPhee, Dr. Luba Brovko and Stacy Favrin. Much appreciation is extended to Dr. Sabah Jassim, Dr. JinChen,
Wendy Cladman, Riske Meewisse and Andrew Moore for technical assistance. A special
1 thank you to Master Bill Lachowsky for being my partner in crime during the long and ktrating weeks of writing -you kept me sane, on track, and 1 do indeed owe you a bottle of whisky! Gratitude is also extended to the faculty, staffand students of Food Science and
Laboratory Services Division for support and assistance during my research.
I have been fortunate to make many good &ends during my studies, most of them as a result of working for several years in the Graduate Student Lounge. Thank you Bonnie for keeping me financially solvent during some lean tirnes. Erin, thank you for being so thoughtful of others despite your own difficulties. Cheers to my bartender colleagues and the Lounge patrons, past and present, for providing such a fun environment to work in.
To rny close friends Brian, David, Donna, Elizabeth, Jarnie, Jan, Laurie, Lisa D., Lisa P.,
Louis, and Moma thank you for being so caring and supportive - 1 love you all. To my fnends in Scotland - particularly Nicola, Magnus, Stuart, and David - thanks for coming to visit. and for continuing to participate in my life even though I've been thousands of miles away. To my housernates Nancy, Mark, Kristen, and Don, thanks for being such laid- back people - 3 1 Fountain Street won? be the same without you.
Finally, this would not be complete without acknowledging the love and support of my family both in Scotland and in Michigan, USA. Leaving Scotiand kvas one of the hardest things I've ever done (apart from hting this dissertation!), but 17mglad I did because it has made me appreciate you even more than before. Thank you di.
1.2.4.2. Turbidimetry ...... 45 1.2.4.3. Colonmetty ...... 46 1.2.5. MISCELLANEOUS PHAGE PROBE APPLICATIONS ...... 47 1 .2.5 .1 . High-performance Iiquid chromatography ...... 47 1.2.5 .2 . Immobilised-phage capture assay ...... 49 1.2.5.3. Phage-linked immunosorbant assays (PHALISA) ...... 51 1 .2.5.4. Phage amplification assays ...... 51
CHAPTER TWO: ISOLATION AND CHARACTEEUSATION OF BACTERIOPHAGES ACTIVE AGAINST THE GENERA LISTERIA. SALMONELLA. AND ESCHERICHIA 55
2.2. MATERIALS AND METHODS ...... 56 2.2.1. Bacteriai cultures ...... 56 2.2.1.1. Listeria strains ...... 57 2.2.1.2. Salmonella strains ...... 59 2.2.1 .3 . Escherichia coli strains ...... 60 2.2 .2 . Bacteriophages and propagating strains ...... 60 2.2.2.1. Listerin phage ...... 60 2.2.2.2. Salmonella phages ...... 61 2.2.2.3. E. coliphage ...... 61 2.2.3. Propagation and enurneration of phages ...... 61 2.2.3.1. Phage lysis of host bacteria ...... 62 2.2.3 .2 . Cleaning of phage lysates ...... 63 2.2.3 .3. Enurneration of phage preparations ...... 64 2.2.4. Phage typing of bacteria ...... 65 2.2.5. Transmission electron microscopy of phages ...... 65
2.3. RESULTS AND DISCUSSION ...... 65 2.3.1. Phage enurneration and plaque morphology ...... 65 2-32Phage typing of Listeria spp ...... 66 2.3 .3 . Microscopic charactensation of Listeria phage ATCC 23074-B 1 ...... 72 2.3 .4 . Phage typing of Salmonella spp...... 74 2-35Microscopie characterisation of SaZrnonelia phages Felix 0-1 and SJ2 ...... 77 2.3.6. Phage typing of VTEC and non-VTEC isolates ...... 77 2.3 .7. Microscopic charactensation of E. coli phage AT20 ...... 83 2.3 .8. Applicability of phages in detection methods ...... 85
CHAPTER THREE: SPECIFIC BACTERIAL DETECTION BY PHAGE-MEDIATED ATP BIOLUMINESCENCE ...... 87 3.2. MATERIALS AND METHODS ...... 88 3.2.1. The phage lysis assay ...... 88 3.2.2. Establishing the detection sensitivity of the assay ...... 91 3 .2.3. Detection of target bacteria in mixed cultures ...... 91 3 .2.4. Increasing assay time ...... 93 3.2.5. Target ce11 concentration by filtration ...... 93 3.2.6. ATP recycling ...... 94
3.3. RESULTS AND DISCUSSION ...... 95 3.3.1. Practical considerations ofthe phage lysis assay ...... 95 3 -3.2 . Phage-mediated biolurninescent detection of L. monocytogenes ...... 96 3 -3.3 . Detection limits of the Listerio phage assay ...... 98 3.3.4. Phage-mediated bioluminescent detection of E coli and S. Enteritidis ...... 100 3 -3.5 . Detection limits for E. coli and S. Enteritidis assays ...... 105 3 -3.6 . Effect of contaminants on target ce11 detection ...... 105 3.3.7. Improving asçay sensitivity ...... 113 3.3 .8. Effect of extended assay tirne on detection sensitivity ...... 114 3.3.9. Filtration as a means of concentrating target cells ...... 116 3 31O . Improving assay sensiûvity by ATP amplification ...... 119
CHAPTER FOUR: THE USE OF BACTERIOPHAGES TO IMPROVE THE SPECIFICITY OF TURBIDIMETRIC DETECTTON METHODS ...... 126
4.2. MATERIALS AND METHODS ...... 128 1.2.1. Instrument set-up ...... 128 4.2.2. Establishing the detection limits of the assay in pure culture ...... 129 4.2.3. Effect of phage on bacterial growth ...... 130 42.4. Effect of contarninants on phage-based detection ...... 131 4.2.5. Effect of medium selectivity on detection of L . monocytogenes ...... 131 4.2.6. Phage-based turbidimetric detection of L . rnonocyfogenes in milk ...... 132
4.3. RESULTS AND DISCUSSION ...... 132 4.3.1. Turbidimetric detection ofL . monocytogenes: assay sensitivity ...... 133 4.3 .2. Comparison of andysis parameters ...... 136 4.3 .2.1. Initial and maximum optical density ...... 137 4.3.2.2. Time to reach maximum optical density ...... 139 4.3.2.3. Slope of log phage ...... 140 4.3.2.4. Detection time ...... 141 4.3.2.5. Tuniing point optical density and time ...... 141 4.3.3. Effect of phage on bacterial detection ...... 143 4.3.4. Cornparison of phages ...... 147 4.3.5.Effect of contaminants on phage-based bacterial detection ...... 153 4.3 .6 . E ffect of increased selectivity on phage-based bacterial detection ...... 157 4.3 .7. Phage-based detection of L . monocytogenes in pasteurised milk ...... 159
CHAPTER FTVE: ALTERNATIVE PHAGE-BASED BACTERIAL DETECTION METHODS : IMPEDANCE MICROBIOLOGY AND REFLECTANCE COLONMETRY ...... 166
INTRODUCTION
5.2. MATERIALS AND METHODS ...... 168 5.2.1. Establishing the detection lirnits of the impedimetric assay ...... 168 5 2.2. Effect of phage on impedimetric detection ...... 169 5.2.3. Evaluation of media selectîvity on impedimetric detection ...... 170 5.2.4. Detection of L . monocyfogenes in raw milk ...... 170 5.2.5. Detection of Salmonella spp. in skim milk powder ...... 171 5 .2.6. Cornparison of colorimetric and impedirnetric detection of E. coli ...... 172 5.2.6.1. Colorimetric instrument set-up ...... 172 5.2.6.2. Colorimetric and impedirnetric detection of E . coli G2-2 in raw rnilk ...... 173 5.2.7. Colorimetric detection of E-coli G2-2 in raw minced beef ...... 174
5.3. RESULTS AND DISCUSSION ...... 175 5.3.1. Interpretation of impedimetric detection times ...... 175 5.3 .2 . Detection limits of the impedance detection assay ...... 177 5.3.3. Effect of phage on impedimetric detection ...... 180 5.3.4. Effect of media selectivity on impedimetric detection ...... 183 5 -3.5 . Detection of L . monocytogenes in raw milk ...... 187 5 -3.6 . Detection of Salmonella spp. in skim rnilk powder ...... 190 5.3.7. Cornparison of colorirnetric and impedimetric detection of E. coli ...... 193 5.3 .7.1. Interpretation of endpoint analysis data ...... 193 5.3 .7.2. Interpretation of phage-based endpoint information ...... 195 5.3.7.3. Colorimetric and impedimetric detection of E. coli in raw milk ...... 198 5.3.8. Phage-based colorimetric detection of E. coli in raw minced beef ...... 202
CHAPTER SIX: CONCLUSIONS ...... 209
CHAPTER SEVEN: REFERENCES ...... 225 LIST OF TABLES
Table 2.1. Biochemical profiles used for differentiation of Listeria species ...... 58
Table 2.2. Typing of Listeria monocytogenes and innocua isolates by phage ATCC 23074-BI ...... ,...... ~...... 67
Table 2.3. Phage typing of Listeria species ivanovii, seeligeri, and welshimeri using phage ATCC 23074-B 1 ...... 69
Table 2.4. Typing of selected Group B, C, and D Salmonella isolates by phages Felix 0-1andSJ2 ...... 75
Table 2.5. Typing of ver0 cytotoxigenic E. coli isolates by phage AT20 ...... 8 1
Table 2.6. Typing of non-vero cytotoxigenic E. coli isolates by phage AT20 ...... 82
Table 3.1. Combinations of bacteria and phages wdto examine the effect of non- target cells on phage-mediated detection ...... 92
Table 3 2. Effect of altering amplification components on bioluminescent detection of 10' CFU/ml Listeria monocytogenes ATCC 23074 ...... 123
Table 4.1. Validity of Bioscreen analysis parameters for the differentiation of initial concentrations of L. monocytogenes ATCC 23074 ...... 138
Table 4.2. Evaluation of phage-based turbidimetric detection of pure cultures of E. coli, S. Typhimurium, and L. monocytogenes grown in Trypticase Soy Broth 149
Table 4.3. Effect of contarninants on phage-based turbidimetric detection of E. coli, S. Typhimurium, and L. monocytogenes grown in Trypticase Soy Broth 1 54
Table 4.4. Phage-based turbidimetric detection of L. monocytogenes in artificially contarninated 2% milk using non-selective (TSB) and selective (SEL) media . . 162
Table 5.1. Effect of increased media selectivity on phage-mediated impedimetric detection of L. monocytogenes ATCC 23074 in pure culture ...... 184
Table 5 2. Effect of media selectivity on phage-mediated impedimetric detection in raw rnilk artificially contaminated with L. monocytogenes ATCC 23074 . . . 188
Table 5.3. Influence of phage specificity on irnpedimetric detection of Salmonella serovars Enteritidis and Braenderup in skun milk powder ...... 192
vii Table 5.4. Examples of interpretation of colorimetric endpoint data in the absence and presence of bactenophage ...... 196
Table 5.5. Colorimetric endpoint detection of E. cofi G2-2 in raw rniik using Coliform Medium (CM) and phage AT20 ...... 199
Table 5.6. Cornparison of impedimetric and colorimetric data for detection of E. coli G2-2 in raw milk using Coliform Medium (CM) and phage AT20 ..... 203
Table 5.7. Colorimetric endpoint detection of E. coli G2-2 in raw minced beef using Coliform Medium (CM) and phage AT20 ...... 204
viii LIST OF FIGURES
Figure 2.1. Transmission electron micrograph of Listeria phage ATCC 23074-8 1 ... 73
Figure 2.2. Transmission electron micrograph of Salmonella phage Felix 0-1 ...... 78
Figure 2.3. Transmission electron micrograph of Salmoneh phage SJ2 ...... 79
Figure 2.4. Transmission electron micrograph of E. coli phage AT20 ...... 84
Figure 3.1. The phage-mediated ATP Bioluminescence Assay ...... 89
Figure 3.2. Release of ATP f?om L . monocytogenes ATCC 23074 by phage B 1- rnediated lysis ...... 97
Figure 3 .3 . Detection limits of phage B 1mediated L. monocytogenes bioluminescence assay ...... 99
Figure 3.4. Release of ATP fiom E . coli G2-2 by phage AT20 ...... 101
Figure 3 .5 . Release of ATP £Yom S. Enteritidis by phage SJ2 ...... 102
Figure 3.6. Detection limits of phage AT2O-mediated E. coli bioluminescence assay . 106
Figure 3.7. Detection limits of phage SJ2-mediated S. Enteritidis bioluminescence assay ...... 107
Figure 3.8. Effect of contaminants on phage B 1-mediated bioluminescent detection of L . monoqtogenes ATCC 23074 ...... 108
Figure 3.9. Effect of contaminants on phage AT20-rnediated bioluminescent detection of E. coli G2-2 ...... 110
Figure 3.10. Effect of contaminants on phage SJ2-mediated bioluminescent detection of S. Enteritidis S 1O ...... Ill
Figure 3.12. Effect of extended incubation time on release of ATP from E. coli G2-2byphageAT20 ...... 115
Figure 3.13. Effect of filtration on phage B 1mediated biolurninescent detection of L.monocytogenes ...... 117 Figure 3.14. ATP amplification of the phage B I -mediated L. rnonocyiogenes assay . . 12 1
Figure 4.1. Optical density profiles for IO-fold serial dilutions of L. monocytogenes ATCC 23074 in Trypticase Soy broth ...... 134
Figure 4.2. Reproducibility of detection times for 10-fold senal dilutions of L. rnonocytogenes ATCC 23074 in Trypticase Soy broth ...... 144
Figure 4.3. Turbidimetric detection of L. monocytogenes ATCC 23074 in Trypticase Soy broth: Influence of bacterial and viral levels on detection times ...... 145
Figure 4.4. Turbidimetric detection of L. monocytogenes ATCC 23074: Interaction of phage B 1 and increased media selecîivity ...... 158
Figure 5.1. Capacitance detection curve for L. rnonocytogenes ATCC 23074 growing in Trypticase Soy broth ...... 176
Figure 5.2. Relationship between bacteriai count and Irnpedirnetric detection the (IDT) for L. monocytogenes ATCC 23074 growing in Trypticase Soy broth . . 178
Figure 5.3. Effect of phage B 1 on the capacitance curve for L. monocytogenes ATCC 23074 growing in Trypticase Soy broth ...... 18 1
Figure 5.4. Effect of phage B 1 on IDT7sfor L. monocytogenes ATCC 23074 ...... 182
Figure 5.5. Cornparison of IDT's for E. coli G2-2, L. rnonocytogenes ATCC 23074 and S. Typhimurium ATCC 14028 ...... 186
Figure 5.6. Influence of endpoint selection values on colorimetric detection of E coli G2-2 ...... 194
Figure 5.7. Influence of phage AT20 and endpoint selection values on colorimetric detection of E. coli G2-2 ...... 197 ClAAPTER ONE
LITERATURE REVIEW
1.1. AN INTRODUCTION TO BACTERIAL VIRUSES
1.1.1. WISTORY OF DISCOVERY
1.1.1.1. The "Twort-d9HereiiePhenornenon"
The existence of bacterial vinises known as bacteriophages - eaters of bacteria - was first reported by the British scientist F. W.Twoa in 19 15, and somewhat controversiaily in
19 17 by the Canadian bacteriologist F. d'Herelle. Twort's publication in the Lancet documented the phenornenon of "glassy transformation" observed in colonies of micrococci grown on agar in the presence of small pox vaccine. He postulated that the
'lransparent dissolving material" might be caused by: (i) an enzyme; (ii) could be part of the life cycle of the bactenurn; or (iii) was caused by a vina, a term first used vaguely by
Jenner in 1796 and used more specifically in 1898 by the Dutch botanist Martinus
Beijerinck in his study of Tobacco Mosaic Virus (Radetsky, 1994).
d'Herelle, however, was much more certain of the viral nature of this bactenolytic agent, subsequently published a book in 1921 entitied Le bactériophage: son rde dans immunité. However, he mistalcenly assurned that a single bacteriophage existed (Fenner,
1994)' and that it could act against a number of different species of bacteria due to adaptive variation (Maniloff et al., 1994). This is, in hindsight, not a surprising assumption when one considers that virus taxonomy and classification was still many years away from the generally accepted system in use today (Fauquet, 1994). Australian scientiçt Bumett, between 1924 and 1934, disproved d7Herelle'stheory of a single bacteriophage, and went on to identifi a myriad of vinises, which while exhibiting different physical and biological properties, were able to maintain these characteristics despite repeated subculturing (Maniloff et al., 1994).
1.1.13. Origins of Molecular Biology
Significant scientific advances were to be made in the shidy of bacteriophages during the late 1930's to 50's. in particular, the formation of the Phage Group - a group of scientists to include Max Delbruck, Alfied D. Hershey, Salvador E. Luria, and James D.
Watson - played an enormous role in both the beginnings of molecular biology and the elucidation of the structure of DNA. During this penod, researchers showed that vimes were composed of either DNA or RNA with a protein component, and in 1946, Hershey demonstrated that bacterial viruses could undergo genetic recombination. In 1952, Zinder and Lederberg first demonstrated host transduction (the transfer of genetic information via a virus particle) by bacteriophage P22, while Brenner, Jacob and Meselson discovered messenger RNA in 196 1 (Fermer, 1994). The presence within the Phage Group of the electron microscopist Anderson enabled the identificatioii of viral architecture, a feature now used by the International Committee on Taxonomy of Viruses (ICTV) to aid in identification and classification of viruses dong with nucleic acid idormation (Fauqet, 1994). The attachent of bacteriophages to bacterid cells was also demonstrated by eiectron microscopy. Ultimately this led to the elucidation of replication of these parasites within the host cell.
1.1.2. PHAGE REPLICATION
Phages, in cornmon with other viruses, do not possess the biochemical machinery required to replicate themselves, and thus rely on the assistance of an actively metabolising bacterial ce11 to do the job for them. As phages have no means of actively seeking out replicative hosts, corning into contact with one is a completely random event, relying on their interaction during movement through a medium (Prescott et al., 1993).
Replication will only follow if the ce11 contains specific receptor sites for the phage in question, such as ce11 wall lipopolysaccharides, proteins, teichoic acids, pili or flagella.
Variations within receptor groups Merincrease the phage's specificity for a particular host, with the result that bacteriophages are typically specific in their host requirements to at least the genus level, and usually beyond (Duckworth, 1987). The replicative cycle can differ depending on the phage, but there are suffcient commonalities to use T4 (Mosig,
1994) and Lambda (Campbell, 1994) phages as general examples of the lytic and lysogenic replicative cycles respectively.
1.1,2.1, Attachment and infection Attachent, as descnbed above, occurs when a phage encounters a bacterial ce11 possessing a particular receptor specific for that particular phage. in the case of the extensively studied DNA bactenophage T4, fibres associated with its contractile tail
mediate attachrnent? and as more fibres make contact with the receptor, the tail plate of the phage settles down ont0 the surface of the cell. Thus attachment is generally a two- step event involving first a reversible step, followed by irreversible binding. For each phage-host combination, various salt concentrations (in particular calcium and magnesium ions) govem the rate of attachment, innuencing the formation of electrostatic bonds between tail fibres and ce11 recepton. Either reduction or removal of these ions, or receptor mutations will effectively prevent adsorption and subsequent phage multiplication. Afier attachment the next step of phage replication follows: infection of the host.
First, conformational changes occur to the adsorbed tail base plate and sheath, and the tail sheath contracts. The central tube of the phage tail then entee through the bacterial wall and DNA is pushed out of the phage head, similar to the action of a sy~ge,and into the host cell. This constitutes infection of the cell, although it cm also occur by other means, such as attachment to pili and transfer of DNA either through or dong the pilus. Shortly der introduction of phage DNA into the bactenum, synthesis of host DNA and protein is suppressed. Two events govem the subsequent synthesis of phage DNA: first, the degradation of host DNA into its nucleotide components occurs, providing building blocks for viral DNA synthesis and preventing merhost gene expression. Second, modification of bacterial RNA polymerase occurs, with the result that it now recognises viral promoters, thus promoting the transcription of viral genes rather than host genes.
Viral DNA synthesis commences within 5 minutes of infection, followed by synthesis of proteins responsible for phage structure, assembly, and ce11 lysis and phage release. It takes approximately 15 minutes for the first complete T4 particles to appear following infection, and another 7 minutes for ce11 lysis (Prescott et ai., 1993) -an incredibly efficient process!
1.1.2.2. The lytic versus lysogenic cycle
Two modes of replication have been observed for bactenophages: (i) the lytic cycle, where replicated DNA is packaged into phage heads and subsequently released fiom the bacterial host by ce11 lysis; such phages (e-g. T4) are described as virulent, and (ii) the lysogenic cycle, where phage DNA is incorporated into and replicated within the bacterial genome; in this case phages are termed temperate, the best charactensed being the double-stranded DNA phage lambda. Virulent phages will typically lyse and destroy susceptible hosts (although some filamentous phages are secreted without ce11 lysis), while temperate phages cm adopt either mode. The choice between the lytic and lysogenic cycle depends on the relative expression rates of phage repressor encoded by the cil gene (promoting lysogeny) and the cro protein, capable of tuming off repressor gene expression and starting the lytic pathway (Campbell, 1994). Following infection of susceptible cells with lambda phage, a very small proportion will be lysed but the majority will survive to become lysogens. These infected cells continue to replicate as normal, creating a clone of cells containhg phage DNA. The latent fom of the phage genome remaining in the host is temed the prophage, and its presence prevents superinfection of the host with the same virus. Temperate phages can be induced, in particular by agents causing DNA damage, such as W light or chernical mutagens, and once induced will enter the lytic cycle, resulting in production of phage at the expense of the bacterial host. Two classes of genes are involved in lysogeny and induction; those establishing the lysogenic state, and those maintainhg the prophage in a repressed state within the bacterial genome. Mutations in either group of genes will result in loss of lysogenic function, and depending on the class involved mutants can either resort to lytic activity, lose the prophage completely, or be unable to release viable phage particles
(Gottesman & Oppenheim. 1994).
1.1.2.3. Host Implications
The lysogenic relationship between host and phage can be mutually beneficial. The incorporation of phage DNA into the bacterial genome is considered a survival tactic of the virus, and typically occurs under adverse conditions such as host nutrient deprivation or a high multiplicity of infection where the host cells are in danger of being completely destroyed, exposing phage to environmental hazards. One of the most significant properties conferred to host bacteria following lysogenisation by a temperate phage is that of toxin production. Toxins including Shiga-like toxins in Escherichia cd, botulinum toxins in Clostridium botulinum, enterotoxuis in Staphylococcus meus, diphtheria toxin in Corynebacterium diphtheriae, and pyrogenic toxins in Streptococcus pyogenes have al1 been identified as proteins produced as a result of "phage conversion7' of the bacterial host (Holmes & Schmitt, 1994). Another benefit of lysogenic infection has been reported in the literature as a type of "suicide bombing" where a small proportion of Bacih subtilis cells infected with a phage release virus, thus killing off any susceptible strains present and giving the remaining bactena (those which can safely harbour the phage) a cornpetitive edge (Anon., 1995).
Although it is mentioned above that phage infection can be an advantage to the host, a proportion of the bactenal population is going to be lyset as the virus takes over the replicative machinery of the ce11 and puts it to the task of maiiufacturing more viral particles. This has been problematic in fermentation processes, where starter culture failures can occur as a result of bactenophage infection, and control is difficult particularly when a large number of variable phages are continually introduced into the processing environment via raw materials such as raw rniik (Bruttin et al., 1997), and even from the starter cultures themselves (Sanders, 1994).
1.1.3. MICROBIOLOGICAL APPLICATIONS OF BACTERIOPHAGES
1.1.3.1. Bacterial typing The use of characterised bacteriophages to type bactena beyond the species and serotype
level is well known, and perhaps the best recognised system is the International Typing
System for Listeria monocytogenes defined in 1985 by Rocourt et al. developed
following the successful application of phage typing to other bactena such as
Staphylococcus aureuî (Ryser, 199 1). Identification of new bacteriophages has resuited
in an optirnised typing system (Marquet-Van der Mee & Audurier, 1995), as well as the
development of altemate bactenophage typing sets (Loessner et al., 1994) and a reversed
phage typing procedure (Loessner, 1991 ; Estela & Sofos, 1993). Bacteriophage typing
has also proven usehl as an epiderniological tool in the study of food poisoning
outbreaks associated with L. monocytogenes, BacilZus cerem (Ahrned et al., 1999,
SuZrnonella Enteritidis (Boonmar et al., 1998; Ward et al., l987), and E. coli 0 1 57:H7
(Samadpour et al., 1993; Barrett et al., 1994; Grimm et al., 1995). This allows the epidemiologist to confimi the reiationship between outbreak strains isolated fkom infected
patients and those cultured fiom suspected food samples. In addition, environmental sources of contamination cm be identified, their movement within food processing plants rnonitored, and persistent problems distinguished fiom sporadic contaminants (Loessner
& Busse, 1990).
1.1.3.2. Biological indicators
The most fiequently used phages as biological indicators are those of the coliphage group.
Their use as indicatoa of water quality was proposed based on their occurrence in waters dong with their bacterial hosts, with numbers of phage theoretically reflecting the numbers of host bacteria present. Several studies have utilised these phages instead of the host strains as an indication of faecal pollution (Borrego et al., 1987; 07Keefe&
Green, 1989), and evidence suggests that they are as good as, or indeed better indicators than bacterial systems currently in use. A coliphage-based water assay is described in the
Standard Methods for the Examination of Waters and Wastewater (Amencan Public
Health Association, 1985), and using E. coli main C as the host bacterium, five or more phaged100mL of water can be detected within 4-6 hours. Problems have been encountered with naturally occumng coliphages in environmental waters, which affect the correlation between phages and faecal pollution. However, vhestend to survive better than coliforms in water, and have been show to exhibit a higher resistance to chlorine. Their application as indicators for colifomis in foods has seen Iimited study
(Kennedy et al., 1994; 1996). Overall this method may be useful as an alternative to E. coli or coliform determinations, or as an indicator of enterovins survival (Jay, 1992).
1.1.3.3. Cloning vectors
Extensive genetic characterisation of certain bacteriophages has seen their extensive application as cloning vectors for amplification of DNA in E. coli. Indeed, phage
Lambda was the prototype vector long before recombinant molecules could be made in vitro. Lambda phage offers a number of attractive features for cloning: recombinant genomes can be eficiently recovered in packaged phage heads, plaques provide a large amount of DNA for hybridisation studies, and cloned genes cm be easily anaiysed.
Phage M 13 has been particularly useful in cases where single-stranded DNA is required,
such as in site-directed mutagenesis, while phage P 1 offen a large cloning capacity of up
to 100 kilobases. P 1 has dso been demonstrated to have a wide host range, which couId
facilitate the transfer of DNA to bacteria other than E. coli (Murray, 1994).
1.1.3.4. Phage therapy
A prophetic aspect of the discovery of bacteriophages was the suggestion by d'tierelle
that bacteriophage could be used as a form of therapy for the treatments of 'intestinal
disturbances', Bacillary dy sentery , Staphy lococcal infections, and Bubonic plague to
name but a few (d'Herelle, 1926). While this idea was viewed, certainly by d'Herelle himself, as perhaps the universal cure for al1 bacterial infections, it was never explored to any great extent, and indeed with the advent of antibiotics was ail but discarded.
However, with the increasing incidence of bacteriai resistance to certain antibiotics, bacteriophage therapy has been revisited in the 1990's (Levin & Bull, 1996; Meml et al.,
1996; Radetsky, 1 996; Soothill, 1992).
While scientists such as Lederberg (1996) have been reluctant to completely mle out the usefulness of phage therapy, a number of pertinent questions remain unanswered in the wake of work by Meml et al. (1 996) to isolate and utilise long-circulahg mutants of
Escherichia coli phage L and Salmonella Typhimuriurn phage P22. An identified alteration to the phage head protein was found to be responsible for the mutants? success
in avoidïng elimination by the reüculoendothelial system of test mice, one of the more
important problems to be addressed in phage therapy. Despite this work, scepticism remains for many reasons such as the potential for bacteria to develop mistance to phages (a phenomenon which may be slowed by the use of phage combinations) and a lack of hard data despite claims that phage therapy has been successfully used in parts of
Europe for many years (Radetsky, 1996).
1.1.3.5. Bacterial detection
As a result of the usefulness of phages in the typing of bacteria, and as cloning vectors, their ability to identie microorganisms to at least the genus level has seen their increased application in techniques for the detection of important food-associated bacteria.
Ultimately. their specificity dictates the success of any phage-based system, and thus identification of useful phages for species- or snain-specific detection of bactena is of the utmost importance. The next section of this chapter will focus on the use of phages in a variety of molecular, biochemical and physical methods applicable to the microbiological analysis of foods, some of which are in use, and their imminent application in more recently developed technologies.
1.2. BACTERIOPHAGE-BASED DETECTION METHODS
1.2.1. PHAGE-BASED REPOR'IER GENES 1.2.1.1. lux-based bioluminescence
The use of bacterial bioluminescence, in particular the application of lm genes, and
aspects of its molecular organisation and regulaîion have ken extersively reviewed
(Baker et al., 1992; Hi11 et al., 1993; Hill & Stewart, 1994; Meighen, 199 1, 1993;
Schauer, 1988; Stewart et al., 1989; Stewart & Williams, 1992, 1993; Stewart, 1997), and
readers are referred to these publications for a comprehensive introduction. Briefly,
bacterial bioluminescence is a light-emitting reaction involving the oxidation of reduced
flavin mononucleotide (FMNH,) and a long chah aliphatic aldehyde, such as tetradecanal
(considered the natural substrate of the reaction) or dodecanal, by rnolecular oxygen. The
reaction is catalysed by intracellular luciferase according to the following equation:
FMNH, + RCHO + O2 -+ FMN + RCOOH + H20+ LIGHT
where light emission is in the form of a glow with an estimated quantum yield (photons of light produced per molecule of substrate) of 0.1 (Bronstein et al., 1994) Given that the production of FMNHz fiorn flavin mononucleotide is essential to the reaction, only living cells capable of electron transport will be able to produce this blue-green light (490nm emission), and bioluminescence is thus an excellent means of establishg the metabolic status of an organism. Light can be measured by instrumentation including luminometen, scintillation counters and sensitive photon counting imagers such as the
BIQ Bioview Image Quantifier (Cambridge Imaging, Cambridge, U.K.) which is capable of detecting single bioluminescing cells (Hooper et al., 1990). Naturally bioluminescent
bacteria have been isoIated fiom marine, fieshwater, and terrestrial environments, and the
majority belong to one of four genera: Vibrio, Alteromonm, Photobacteriurn, or
Xenorhabdur (Baker et al., 1992). In symbiotic relationships, bacterial bioluminescence
is thought to serve as a means of communication for the host, but its purpose is less clear
for fiee-living bacteria (Meighen, 1993).
A heterodimeric enzyme, bacterial luciferase is composed of two subunits, a (40kDa)
and p (36kDa), respectively encoded by the IuxA and ZiaB genes. These genes are tlanked
by lux genes C. D,and E, respectively coding for reductase, tramferase, and synthetase
proteins responsible for the production of aldehydes fkom fatty acids. These five genes
compose the lux operon, transcnbed in the order luxCDABE, and these genes have been cloned and sequenced for some luminescent Vibrio, Photobacteriurn, and Xenorhabdus
spp. (Meighen, 1993). Other lux genes have since been identifieci in Y.fischeri, including a second operon consisting of regulatory genes lu1 and IuxR located immediately upstream from the lux operon. These genes code for an autoinducer and receptor involved in activation and expression of luminescence, in conjunction with another gene, luxG, encoding a reductase involved in the production of FMNHz (Meighen, 1993).
Transfer of lux genes has been approached fiom a number of angles, depending on both the organisms to be transformed, and the application of the biolurninescent phenotype.
Plasmids, containing either the complete 1uxCDABE operon, ldBluciferase genes, or hised llaAB genes, have been constructed and employed to transfer a bioluminescent
phenotype to a desired bacterial recipient (Hill et al.,1993). The complete operon
encodes substrate in addition to luciferase, thus providing a continuously luminescent
organism. Innate substrate production provides a non-destmctive means of real-tirne
monitoring of luminescent bacteria in rnixed culture and within eukaryotic cells, but is
only useN for Gram-negative organisms capable of recognising transcription and
translation sipals O rig inating fiom biolurninescent bacteria Additionally , continuous expression of substrate places energy conshaints on the cell, and is therefore not suitable
for physiological studies.
Altematively, only the genes for luciferase are required if substrate is added exogenously, and this does not significantly influence ce11 metabolism. Bactena expressing lwcAB are therefore non-luminescent before substrate is added, and this less complicated procedure is favoured for use with Gram-positive bacteria where a promoter also has to be engineered for expression. ldBfusions are another option when working with Gram- positive bacteria, and the resulting monomenc luciferase has been reported to increase quantum yield about 100 times as compared to the dimeric form (Hill et al., 1993).
Transposon vectors have also been engineered to contain lux genes without promoters, such that the genes are only expressed when the transposon inserts into the bacterial chromosome beside a promoter region (Hill et al., 1993). In addition, the source of luciferase genes influences the temperature stability of the resulting enzyme, and should be carefully considered based on the growth temperature of the microorganism of interest and the application of the bioluminescent phenotype. It is worth noting that fusion genes produce products with temperature optima 7-1 O°C Iower than that of the wild type (Hill et al., 1993).
Such a variety of alternatives has resulted in the transformation of a nurnber of both
Gram-negative and Gram-positive bacteria, and these now bioluminescent microorganisms have most commody been wdas reporters of, for example, bacteriai inj ury and repair, biocide efficacy (WaIker er al., 1992), regulation of gene expression
(Eaton et al., 1993; Park et al., 1992; Schauer, 1988) and survival of pathogens such as
SalmoneZIa Enteritidis (Chen et al., 1996) and Escherichia coli 0 157:H7 in food processing (Hudson et al., 1997; Tomicka et al., 1997).
Alternatively, bioluminescence-based detection methods for bactena have been developed by the use of a third gene vector: bactenophage engineered to carry Z2a genes within their genome. Depending on the host-specificity range of these phage vecton, generic or specific detection of microorganisms cmbe achieved by bacterial expression of bioluminescence within 30-50 minutes of infection and transduction by lux' phage unable to express the bioluminescent phenotype themselves (Kodikara et al., 1991). This approach to bacterial detection, first described by Ulitzer and Kuhn in 1987, has led to a nurnber of microbiological applications, in particuiar the detection of various food-borne and clinically important pathogens. A near on-line method for detecting enteric bacteria using recombinant lia bacteriophage
was subsequently proposed by Kodikara et al. (1991). Using a slaughterhouse processing
pig carcasses as a test environment, the authoe reported the ability of recombinant lm'
Enterobactenaceae-specific phages to convert enterics sampled fiom carcasses, meat-
contact fac tory surfaces, and various abattoir areas into biolurninescent bacteria within
one hour of sampling. The limit of detection of this assay was around l(r target cells per
g or cm', but was substantially improved by the incorporation of a 4 hour enrichment in
Luria broth at 37"C, reducing the detection limit to less than 10 enterics per g or cm' with
a required total assay theof 5 hours. Despite these prornising results and the
applicability of such a rapid detection system in a Hazard Analysis Critical Control Points
(HACCP) system. industrial application of such on-line analysis has not been adopted
(Stewart, 1997). Regardless, this research prompted significant research activities in the area of bacterial detection, and alternative applications have been forthcoming.
In 1993. Turpin et al. (1 993) reported the developrnent of a phage-based luminescent rnost-probable-number (MPN) method for the detection of Salmonella Typhimwiurn, employing a commercially available lux' P22 Salmonella phage available from
Arnersham International. A 15-tube MPN test protocol was established, with each of 0.1.
1, and 10 rnL volumes of samples (including soils, sewage sludge and water) artificially inoculated with S. Typhimuriurn pipetted into five tubes of stenle buffered peptone water.
Afier overnight enrichment, samples were tested by both the MPN method (inoculated into Rappaport-Vassiliadis broth) and the In-baçed protocol where samples were inoculated into Luria broth containhg 6.9~10'~plaque forming units (PFU) of lux-P22 phage. MPN samples were incubated at 43OC ovemight, streaked ont0 xylose lysine desoxycholate and brilliant green agas after 24 and 48 hour incubations, and presumptive colonies were biochemically confirmed using the API system of identification. lux samples were incubated for 90 minutes at 30°C and aldehyde substrate was then added.
Subsequent luminescence was rneasured uskg a luminometer, with positive tubes producing relative light outputs 100 to 1O00 times greater than negative samples. When compared with the MPN method, the lux-based method had an esciency of 100% with no false positive or negative results. A S. Typhimurhm detection limit of 10 colony forming units (CFU)per gram of soi1 or per 100 mL of lake water was dernonstrated in the presence of large numbers of competing bacteria, thereby reducing the requirements for selective enrichment and decreasing the duration of the protocol to less than 24 hours.
Three recombinant lux phages, containing either ldBor the entire lm operon, were later constructed by Chen & GriEths (1996) to expand the detection range to SalmoneUa isolates belonging to different serogroups. One hundred percent of the Salmonella strains belonging to groups B and D, and selected group C isolates tested were positively identified using a recombinant phage plaque assay, while dl of the 22 non-Salmonella isolates tested negative. The optimum adsorption theand temperature for subsequent phage infection and luminescence was detemined to be 30 minutes at 42OC, producing detectable luminescence in less than one hour for Salmonella concentrations of 1OS
CFUlmL. Employing a 6 hour pre-incubation step, as few as 10 CFU/mL of Salmonella could be reliably detected, which was reported to agree well with previous estimates
using a similar detection system. The method was then evaluated in a food system by
directly inoculating S. Enteritidis into the white of eggs followed by recombinant phage
addition. Mer a 16 hour incubation penod, luminescence was detected in eggs
inoculated at an initial level of about 1o3 CFU, but lower concentrations of 1O2 and 10'
CFU of S. Enteritidis per egg went undetected. However, increasing the incubation
penod to 24 hours resulted in luminescence detection at al1 levels, with the opporhmity to observe the intemal location of luminescent organisms by CCD imaghg of the whole egg. Additional protocols using both (i) E. coli petrifilm and X-ray autoradiography, and
(ii) liquid media and lurninometry were evaluated and demonstrated to be useful alternatives to the original methodology, offering a cheaper means of light detection and a more portable assay system.
Similarly, Loessner et al. (1W6a) reported the construction and use of a lzd reporter bacteriophage, based on the specificity of Listeria phage A5 1 1 for about 95% of Listerio rnonocyfogenes isolates belonging to serogroups 1/2 and 4. A K hanteyii /dBgene fusion was inserted just downstrem of the strong promoter for the major capsid protein gene cps. Using the recombinant phage A51 l::luxAB,as few as 5 x 10' CFUImL of tested isolates could be detected after phage infection and a two hour incubation at 20°C.
To increase the sensitivity even Mer,Listeria selective enrichment broth was utilised to e~chartificially contarninated lettuce samples over a 16 hour penod at 30°C before phage-based testing. The sensitivity of detection was subsequently increased to less than one ce11 of L. monocytogenes Scott A per gram of salad, with a total time investment of
22-24 hours. While the bacteriophage used was reported to be highly specific for certain
L. monocytogenes serogroups. it has a very broad host range with demonstrable activity
against L. ivanovii and L. seeligeri isolates. Therefore, this particular phage will only
gain application as a generic detection method for Listeria spp., assuming that it is also capable of infecting other species within the genus such as L. innocua, which has been
proposed as a usefbi indicator for the presence of L. monocytogenes.
nie detection of L. rnonocytogenes Scott A in various artificially contaminated foods by the above method was merexarnined by Loessner's research group (1997a), and a
MPN-based method was also evaluated. The method was shown to reliably detect L. rnonocytogenes at a level of 100 CFUIg or less in al1 tested foods following a 20 hour selective incubation, with even lower detection lirnits for liverwurst, shrùnp, pasteurised milk, cottage cheese, ricotta cheese, chocolate pudding and cabbage. Detection of lower levels of target cells was improved for rninced meat, Iivenvurst, shredded Swiss cheese, and cabbage samples following an increase in enrichment time to 44 hours, although detection sensitivity never improved beyond 10 CFU/g for minced meat and Camembert, likely due to interference by a large microbid background flora. Naturally biolurninescent bactena were also responsible for non-specific light emission by a shrimp sample, but this was easily identified due to light emission by a negative sample not containing luphage. Naturally contaminated food samples were also tested, with 57 determined to be Listeria-positive by selective plating, and 55 positive by the lux phage method. Differences were observed in the nurnber of positive results each method identified. with the phage-based method identiQing more positive dairy products and dairy plant environmental isolates, while the plating method identified 20% more positive pouitry samples. Failure to isolate Listeria ceIls by plating could be related to sensitivity of some strains to selective agents used in Oxford agar, while failure to detect poultry isolates by phage-mediated lumuiescence was due to the fact that phage AS 1 1::ldB was deterrnined not to infect those isolates.
A modified lux-MPN method was developed and evaluated, using Listeria selective broth as the growth medium in a 3 x 3 tube method, with artificially contaminated samples of minced meat, shrimp, ricona cheese, and cabbage (Loessner et al., 1997a). Samples were removed from positive tubes der20 and 44 hours of enrichment at 30°C, and tested using the luvphage method. Numben calculated fiom the luminescent MPN method agreed well with the levels of L. monocytogenes Scon A initially inoculated into food samples following both 20 and 44 hour enrichment, with only minced meat samples showing slightly higher counts after 44 hours, demonstrating the rapidity and reliability of such an assay. Furthemore, it was demonstrated to be easily modifiable to a microtitre plate format, enabling a large number of samples to be processed simultaneously.
1.2.1.2. Luc-based bioluminescence
The natural bioluminescence of the North American firefly Photinus pyrillis has been
20 similarly exploited in bacterial systems, following the successful cloning and expression of the cDNA of the single polypeptide lue gene, encoding a 62 kDa luciferase, in E. coli
(de Wet et al., 1985). The production of light by this rneans relies on the intedon of
ATP, present in dl living cells as an energy transfer molecule, with substrate, heterocyclic carboxylic acid luciferin, in the presence of molecular oxygen, and is catalysed by the firefly luciferase enzyme in the presence of magnesiurn according to the following equations:
Luciferase + luciferin + ATP - luciferase-luciferyl-AMP + PPi (Eq. 1)
Luciferase-luciferyl-AMP + 0 - luciferase + oxyluciferin + AMP + CO+ light (Eq.2)
This reaction results in the highly efficient production of light, in the form of a flash, with an estimated quantum yield of 0.88 - substantially more efficient than that generated by bactenal luciferase (Bronstein et al.. 1994). The luciferase of P. pyrafis emits yellow- green light with peak emission at 56011x11,but other f~eflyand insect species (including click beetles) have been shown to emit different colours of light, ranging fiom green
(547nm)to red (604nm) (Ohmiya et al., 1996). This reporter system is particularly attractive for use with bacteria because they are not capable of natural luc-encoded luminescence, and therefore will not interfere with detection of luc-transformed bacteria.
Any natural bioluminescence will be of lux-origin, and cm be differentiated fkom luc by their diffenng wavelengths of light emission. Given that only one gene is involved in encoding insect luciferases, they also offer the advantage over lux genes of being somewhat simpler to work with. luc-encoded bioluminescence can be measured by the same instrumentation used for lm' detection, or a customised apparatus such as the system reported by Gailey et al. (1997), who developed an imaging system based on a single-photon counting photomultiplier tube. This system is capable of real-time monitoring of luciferase gene expression in cancer cells but at a substantially reduced cos.
There have, however, been problems associated with the use of firefly luciferase relating to its pH and temperature stabilities, sensitivity to proteolysis, and the rapid decay of light output during the bioluminescence reaction. Given that different species of insects can produce different light emissions while sharing a comrnon substrate, differences have been attributed to enzyme structure (Ohmiya et al., 1996). This, dong with the recent publication of the structure of firefly luciferase, has been the impetus for genetic manipulation studies of the structure-function relationship in an atternpt to creaie enzyme denvatives with improved light eniission, and more applicable pH and thermal stabilities.
For example, substitution of the arnino acid serine at position 286 in the firefly luciferase sequence by either leucine, valine, glutamine, tyrosine, lysine or phenylalanine results in a shift in the luciferase emission specmim producing red-orange light as opposed to the wild-type yellow-green (Mamaev et al., 1996; Arslan et al., 1997). Additionally, amino acid substitutions cm decrease or improve thermal stability, as demonstrated by White et al. ( 1996) who generated a series of mutant enzymes by replacing glutamate at position 354 with al1 other possible amino acid residues. With the exception of glycine, proline and aspartate, any amino acid replacing the glutamate residue produced an enzyme with enhanced thermostability as compared to the wild-type fuefly luciferase. The flash kinetics of the reaction, caused by the interaction of luciferase and oxyluciferin, have been addressed to some degree by the use of chemicais such as coenzyme A, reported to prolong the half-life of light to ten minutes. Another system of reagents, known as
LucLiteTM,bas been developed and it has been estimated to improve the half-life of light output to over 5 hours, with the production of a glow-type light signal (Roelant et al.,
1996). In addition, Thompson et al. (1997) reported that a single serine to threonine substitution at position 198 resulted in an enzyme with a 150-fold increase in half-tirne for light emission decay, as well as a higher pH optimum.
The permutations are nurnerous, and offer an approach which will be able to produce enzymes with more desirable properties for a particular application, dong with the simultaneous differentiation of bacteria based on the emission spectra of the luc genes used. The use of dual bio- and chemi-luminescence reporter genes for luciferase and P- galactosidase has also been reported (Martin et al., 1996), and current advances in characterisation of luciferases From other species of insects (Viviani & Bechara, 1994; Ye et al., 1997) will no doubt advance the application of luc-based bioluminescence even fiirther.
As with the lm genes, the firefly luc gene has been engineered into both plasmid and
23 bacteriophage delivery vectors, facilitating gene cloning of many eukaryotic and
prokaryotic cells, including E. coli 0 157:H7 (Pagotto el al., 1996; Fratamïco et al., 1997)
and StaphyZococcur aureus (Steidler et al., 1996). Much of the applied research using
phage-based luc gene technology has focused on the detection of the problematic human
pathogen Mycobacterium tuberculosis, causative agent of tuberculosis, the leading cause
of death from a single infectious disease with fatalities estimated at over 2 million per
year. With a trial-proven vaccine against tuberculosis estunated to be another 30-40 years
away, a delay which could resdt in up to 80 million more deaths (Day, 1998), anti-
tuberculosis drugs are still the only means of combating the disease. However, the emergence of multidrug- resistant isolates has Mercomplicated matters. Due to the slow 20-24 hou doubling rate of M. tuberculosis, dmg susceptibility tests take an unfeasibly long time before resistance profiles are available to act upon (Jacobs, Jr. et al.,
1993), but phage-based luc technology has offered an alternative approach.
Arnerican and British researchers (Carrière et al., 1997; Jacobs, Jr. et ai., 1993; Pearson et al., 1996; Sarkis et al., 1995) have engineered firefly luciferase reporter myco- bactenophages designed to both detect and assess the sensitivities of M. tubercuZosis and other mycobacteria to various drugs including rifampicin, isoniazid, and streptomycin.
Various strains of mycobacteria infected by luc' phage were exposed to antibiotics commonly used in TB treatment: those that are dmg-resistant continued to biolurninesce, while those sensitive to drug treatments did not. Irnprovements to the sensitivity of the assay have been achieved by the replacement of lytic phages, in which light output was lost following ce11 lysis, with lysogenic phages capable of producing lysogens with
continuous light output. This has hproved the sensitivity from 500-5000 cells domto
as few as 13 mycobacteria, with dmg susceptibility determinable within 24 to 48 hours
(Carrière et al., 1997). The use of phages with broader host ranges has expanded the
applicability of the assay to other mycobacteria of medical importance, while the
selectivity of the assay can be tailored with the inclusion of p-Nitro-a-Acetylamino-P-
Hydroxy Propiophenone (NAP), inhibitory to rnembes of the tuberculosis cornplex and
thus enable differentiation between these and other non-tuberculosis mycobacteria (Riska
et al.. 1997).
1.2.1.3. Ice nucleation activity
Ice nucleation, Le., formation of ice nuciei in pure water supercooled below its Freezing
point. by bactena was fist reported in 1973 by Schnell and Vali, and this frost damage
phenornenon was subsequently demonstrated to be caused by the plant pathogen
Pseudomonas syringae. Since then, a number of strains of P. syringae as well as strains of Xanîhomonas campestris, Envinia ananas, E. herbicola, P. fluorescens and P. viridiflrra have been identified as ice nucleation active (INA) bacteria (Margaritis &
Bassi, 1 99 1). Bacterial ice nucleation has been shown to occur at threshold temperatures ranging from -2 to -1 O°C, higher than those associated with the ice nucleation activity of lichens, fieeze-tolerant amphibians and insects (Warren & Wolber, 1991) and these values have been used to classifi MA bacteria into three groups (Schmid et al., 1997). Although it is not known what advantage this phenotype conveys to MA Gram-negative bacteria, a single chromosomally-encoded gene has been determined to be responsible for ice nucleation activity, and a variety have already been sequenced, including inmi (E. ananas), iceE (E. herbicola), inail (E. uredovora), inaW (P.fluorescens), inaX (X campestris pathovar tranrlucem) and inaZ (P. syringae), ranging in size fkom 3.4kb to
7.5kb (Li & Lee, 1995). Recently, a new ice nucleation gene, inaV, of about 3.5kb in size was isolated from P. syringue (Schmid et al., 1997). This gene encodes a 1200 arnino acid residue protein, InaV, and shares 92.4% homology with inaZ. Based on 77% sequence conservation of their protein products (Schmid et al., 1997), it is thought that these individual ina genes hmdifferent bacteria evolved fiom a common gene. Also, their action seems to be host independent in nature, since antibodies raised against the inaW protein of P. fluorescens will recognise and react with the inaZ protein of P. syringae expressed in E. coli (Margaritis & Bassi, 1991). Ina proteins are located in the outer membrane of bacterial cells, and have a functional requirement for Iipids (Gurian-
Sherman & Lindow, 1993). It has been proposed that ina proteins act as templates for ice formation, and that differing aggregation of these proteins dictates variations in threshold temperatures (Schmid et al., 1997).
Ice nucleation genes cloned into Escherichia coli have been show to convey ice nucleating properties to their host (Oner et al., 1985), and this has led to the application of bacterial ice nucleation in a variety of ways, ranging fkom fieezing processes in the food industry (Watanabe & Arai, 1994), artificial snow production (Gurian-Sherman & Lindow, 1993)- a reporter for gene expression in moderately hdophilic bactena
(Arvanitis et al., 1995) and promoter analysis in Zyrnomonas mobilis (Drainas et ai.,
1999, to a reporter for the detection of bactena such as Salmonella spp. in food (Wolber
& Green, 1990). It has been estimated that the use of ina gene fusions for the detection of gene expression is 100- to 1000-fold more sensitive than the lux detection system
(Guian-Sherman & Lindow, 1993). With a low number of proteins (300 or less) required for ice nucleation to occur, such activity could be a useful proposition for the detection of injured cells still capable of minimal cellular activity.
By cloning an ice nucleation gene into a delivery vector such as a bacteriophage, which cannot express the MA phenotype on its own, this phage can be used to tramduce susceptible cells. An MA phenotype cm thus be passed on to infiected cells, with ice nuclei synthesised within 15 minutes of ha gene transcription. This principle was employed to develop a sensitive and npid method for the detection of Salmonella spp., known as the BMDTM(Bacterial Ice Nucleation Diagnostic) assay, marketed by IDEXX
Laboratories, Inc., located in Westbrook Maine, USA. This test employs the specificity of the SuIrnonella phage P22 engineered to contain an ina gene within its DNA to detect ice nucleation in susceptible Salmonella cells. Phage and sarnples being tested for the presence of Sdmonella are mixed in the presence of a freezing-indicator dye, carboxyfluorescein, and incubated at 37OC to allow phage attachent and transduction of susceptible bacteria. Samples are then incubated at 23°C to optimise expression of the ice nucleation protein and formation of ice nuclei. Ice nucleation activity is subsequently detected by chilling treated samples and controls to a temperature of -9.j°C. Freezing of cells is indicated by an orange, non-fluorescent appearance, while negative samples appear fluorescent green. The level of bacterial contamination can be cdculated based on the concentration of ice nuclei determined by the droplet fieezing assay (Wolber &
Green, 1990).
The performance of the BMDTM prototype assay in foods such as miik, meat, gelatin and eggs was most encouraging, with littie or no interference fiom the food samples themselves or fkom background ice nuclei. Using ina-transduced phage constnicted fkom
P22 (with a similar host specificity to the parent phage), a sensitivity of 10 cells per mL or less was obtained in test samples containing either S. Enteritidis, S. Typhimurium, S.
Dublin S. Paratyphi B, or S. Gallinariurn. A sirnilar detection limit was achieved for S.
Dublin in the presence of high nurnbers of non-Salmonella bacteria. The test is also capable of detecting sublethally injured cells which are still able to produce ice nuclei.
Any e~chmentstep can thus be relatively short, enhancing the rapidity of the method, which. at around hvo hours for a positive result is a considerable time saver as compared to other detection methods for Salmonella spp. (Wolber & Green, 1990).
Information regarding the BINDTM assay and the agreement of assay-positive isolates with identification by the API 20E biochemical system was presented by Wolber in
Washington at the 9S" general meeting of the American Society for Microbiology
(Wolber & Green, 1995). Ten food matrices were spiked at low (20 to 50 cells/25g) and high ( 10' to 1o3 cellslg) levels with 20 of the commonest Salmonella serotypes isolated in the U.S., and dera non-selective pre-enrichment using buffered peptone water, samples were tested for ice nucleation ability. Positive isolates were then identified by the API
ZOE system, and agreed 100% with the results obtained by the BMDTM assay.
More recent uiformation koom IDEXX regarding the BINDTMassay claims an assay sensitivity of one viable Salmonella ce11 per 25g of sample, following a single ovemight enrichment step. HandIing time for the assay is estimated at less than one minute per sample, and including enrichment, results are obtained within 22 hours. The test can now detect 500 different Salmonella spp. without producing false-positive results (Anon.,
1997).
1.2.1.4. Green Fluorescent Protein
Another reporter system currently receiving a great deal of attention as a means of tracking gene expression and protein localisation in a variety of organisms and cells is the green fluorescent protein (GFP), naturally occurring in the bioluminescent jellyfish
Aequorea victoria. GFP is activated in A. victoria by the absorption of blue cherniluminescence generated fiom the calcium binding activation of photoprotein aequorin, and produces a stable- strong green fluorescence (Chalfie et al., 1994). It has been proposed that GFP evolved as a means of arnplifying the relatively low levels of cherniluminescence produced by photoproteins into higher intensity fluorescence with no extra energy expenditure on the part of the ce11 (Cubitt et al., 1995). The 27-kDa GFP,
encoded by the gfp gene, is 238 amino acid residues in size and absorbs light at 395nm
(with a smaller peak at 475n.m). The formation of GFP is favoured by lower temperatures
(-22°C) and takes about four hours in E. coii (Heim et al-, 1994). Once GFP is formed, it
is stable at temperatures up to 65OC, at aikaline pH, and is resistant to proteases. Its
crystd structure has been determined to consist of 1 1 beta-sheets forming a barrel-like
arrangement around a central alpha-helix containing the fluorescence centre (Misteli &
Spector. 1997). Its activity requires the presence of oxygen, but does not need additionai
genes, CO-factorsor substrates for fluorescence to occur. This allows for the detection of
sub-lethaily injured cells, making it an attractive alternative to the use of substrate-
dependent marker systems such as lux and luc.
Research interest in GFP soared following the demonstration by Chalfie et al. (1994) that
the & gene engineered into a plasmid vector could produce fluorescence in both E. coli and eukaryotic cells when the transformed cells were excited by near-U.V. or blue light.
Additionally, the g& gene cmbe fused to genes of interest in living cells without interfenng with ce11 function and growth, although it has been suggested that the fùnctionality of some fusion proteins may be affected by the size of the GFP tag (Misteli
& Spector, 1997). This makes it a potentially useful marker system in the study of, for example, bacterial biofilm formation. Such breakthroughs, together with the availability of fluorescence detection methods such as flow cytornetry, epifluorescent microscopy, spectrofluonmetry, and confocal laser microscopy, have led to the application of GFP technology in a number of research areas.
in a special issue of the journal "Geney7dedicated to GFP, some bacterial applications were reported. For exampie, host-pathogen interactions following infection of live marnmaiian cells with either Salmonella Typhimuriurn, Yersiniapseudotuberculosis, or
Mycobacterium marinum. were monitored by bacterial expression of the GFP phenotype
(Valdivia et al., 1996). Bacterial movement through environmental materials such as quartz sand could be tracked by fluorescence (Burlage et al., 1 996) and plasmid tram fer was studied using GFP as a marker of genetic transfer between bacterial cells
(Christensen et al., 1996). Recent publications have focused on GFP as a successful marker system for evaluating movement (Tombolini et al., i 997) and survival of genetically engineered rnicroorganisms in the environment (Leff & Leff, 1996), the survival of E. coli O 157:H7 in orange juice and apple cider (Fratamico et al., 1 997), and detection of individual P. ueruginosa cells in a mixed biofilm population (Bloemberg et al., 1997).
Despite the obvious attractiveness of this marker system, it has yet to be reported as a bacteriophage-based detection method. Promising resuits have been published reporting the use of virus-based dpvectors such as potato virus X (Boevink et al.. 1996), tobacco rnosaic virus (Casper & Holt, 1996), and retrovimes (Klein et al., 1997), and it is anticipated that bactenophage-mediated delivery of GFP will be available in the near future. Inevitably, problems associated with this new technology have arisen, which have in part already been addressed by researchers. One of the most crucial aspects is the
fluorescence intensity of the protein, and efforts have been made to improve fluorescence
by single point mutations targeting primarily the serine65-tyrosine66-glycine67 chomo p hore. Replacement of Se65 with alanine, valine, leucine, cy steine, and threonine residues resulted in loss of the 395m excitation emission peak, and moved the more photostable 475rm peak slightly towards a longer wavelength, producing spectra similar to that of the sea pansy Renilla GFP. When Ser65 was replaced by threonine, a six-fold increase in brightness of GFP was observed as compared to wild-type GFP, with decreased photobleaching Wei.et al., 1995). Cormack et al. (1996) reported the genention of GFP variants, by random amino acid substitution, which when expressed in
E. coli, showed a 100-fold increase in fluorescence intensity, attributed to shifts in excitation maxima and more efficient protein folding than that of wild-type GFP.
The colour emission of GFP has also been studied, and it was demonstrated that replacement of Tyr66 and Tyr 145 with histidine and phenylalanine, respectively, produced isoforms emitting blue instead of green light (Heim et al., 1994; Mistell &
Spector, 1997). A red-shifted variant, RSGFP4, has also been created by mutagenesis
(Yang et al.. 1996), and similarly, photoactivation of GFP in low oxygen environments produced red instead of green fluorescence (Elowitz et al., 1997). This provides exciting advantages to the use of GFP and its variants in simultaneous detection of cells expressing different fluorescence emissions as demonstrated by Yang et al. (1996), who used dual colour microscopy to visualise cells expressing red and green fluorescence. While the use of GFP has been praised due to the absence of interfering background fluorescence, care should be taken in the application of its variants, as reported by Lewis
& Emngton ( 1996). These researchers found that a blue gfp variant was not useful in the study of sporuiation of Bacillu subtilis, due to the innate fluorescence of the spore- former in the blue portion of the spectrum.
Little information is currently available regarding the detection limits of GFP. While it has been estimated as being 1OS bacterial cells using a derivative Tn5GFP 1 transposon vector (Burlage et al., 1996) and as low as 1 CFUImL using a commercial cDNA vector
(Fratamico et al., 1997), the w of a strong promoter and rnodified & genes (Suarez et al.. 1997) should improve sensitivity greatly.
1.2.2. PHAGE-BASED ATP BIOLUMINESCENCE
The use of ATP (adenosine-5'-triphosphate) bioluminescence as a rapid method for the detection of bacterial biomass and its applications in food microbiology, hygiene monitoring and HACCP programmes have been extensively reviewed (Ugarova, 1993;
Griffiths. 1993, 1996; Hawronskyj & Holah, 1997). The ATP bioluminescence reaction has already been discussed with regard to lm-encoded bioluminescence and readers are directed to this section for details of the reaction. Protocols for the analysis of food samples typically include primary chernical treatrnents, filtration and washing steps to remove background andor non-microbial ATP. Bacterial cells are concentrated on the filter and are then chernically lysed to release microbiai ATP which can be quantified by
the addition of the luciferin-luciferase cornplex. This assay can be perfonned either in
liquid form using a luminometer, or by direct application of ATP extractants and
luminescent reagents ont0 filters, followed by image processing (Tanaka et d,1997).
Light output, quantified in terms of relative light units, can thus be used to estirnate
biomass (Grifiths et al., 199 1) based on the generaily accepted fact that the ATP content
of each bacterial ce11 is reasonably constant. Alternatively, in hygiene monitoring, both
microbial ATP and ATP present in food debns are of relevance and thus a total ATP
estimate is desirable to evaluate hygiene levels relating to microbial contamination and
overall cleanliness (Seeger & Griffiths, 1994).
While ATP bioluminescence has been widely accepted as a rapid and convenient method
for the anaiysis of overall bacterial load, the lack of specificity of the procedure is a major drawback in iis application to many areas of food microbiology. Chernical extraction of microbial ATP is inherently non-specific, and while measurement of total ATP levels has been shown to be sufEcient for the purposes of quality assessrnent of food products and factory hygiene (Bautista et al., l992), the detection of specific groups of microorganisms in mixed cultures, for example, food-borne pathogens, starter cultures, and spoilage bacteria. has been impossible to achieve. Two approaches have been studied in an attempt to improve the specificity of the ATP bioluminescence reaction. both of which exploit the specificity of lytic bacteriophages. These breakthroughs could make ATP bioluminescence a more complete tool for the microbiological analysis of foods. 1.2.2.1. Lytic phages
The fmt approach has focused on the replacement of chernical extractants with bactenophages capable of specifically lysing groups of microorganisms. Very simply, the lysis of bacterial cells by phage will result in release of intracellular ATP which can be detected by bioluminescence. A sample containing added bacteriophage and bacterial cells susceptible to phage infection will show an increased amount of ATP in the system. as compared to a control sample containing no phage. This approach was first published in 1995 by Sanders, who reported the use of a bacteriophage for the specific detection of
Lisleria monocytogenes even in the presence of the closely related species Listeria
Nlnocua. In a broth system, the presence of L. monocytogenes ATCC 23074 was confirmed within 80 minutes of addition of bactenophage ATCC 2307443 1, when the
Iight output was at its maximum. This method was also capable of detecting 100 cells/g of L. monocytogenes in artificially contaminated lemice within 15 hours, incorporating a selective e~chmentstep.
Despite the specificity of the assay, the lack of sensitivity has become an ovemding concem. The bioluminescence assay itself has a detection limit of around 10)-1O4 bacterial cells, and is unreliable for ce11 numbers below this level. For the phage assay, even in pure culture, at least 1OS to 106 cells are required to produce a detectable increase in ATP as compared with a control in which no phage lysis kas occurred. For optimum results an even higher ce11 count is required. As an alternative to pre-enrichment techniques, suggestions to improve sensitivity have focused on (i) the use of adenylate kinase (AK) and excess amounts of ADP as a means of driWig the equilibrium of AMP,
ADP and ATP towards the anabolism of ATP (Squirrell & Murphy, 1995), and (ii) recycling of ATP by the use of specific substrate-enzyme mixtures to effectively increase the net production of ATP (Hawronskyj et al., 1995). The AK assay, employing lytic phage to release intracellular AK, currently bas a detection limit of less than 1x 10' cells/mL for both E. coli and S. Newport. Results were achievable in less than one hour for E. coli while the Salmonella assay took up to two hours (Blasco et al., 1998). The
ATP recycling method was demonstrated to improve the sensitivity of the standard ATP assay from 550nM to 250pM ATP (Hawronskyj et al., 1995). However, this was achieved using pure ATP standards and not by ATP extraction from microorganisms, as the publication title suggests.
1.2.2.2. Bacteriolysins
A second approach to improving the ATP bioluminescence assay has also exploited the activity of lytic bactenophages, more specifically the endolysins they produced as late gene products during phage replication. Endolysins have been demonstrated to facilitate the release of progeny phage from the bacterial cell; this is known as "lysis fiom within"
(Tarahovsky et al., 1994). These lysins fa11 into one of at lest four categories of enzymes: lysozymes (as is the case with phage T4), transglycosylases, endopeptidases, and amidases. The enzymes gain access to the peptidoglycan layer by the formation of
non-specific membrane lesions or holes, produced by a group of membrane proteins
known as holins (Young & Blasi, 1995).
The genes responsible for endolysin production have been isolated and cloned for a
number of phages (Young, 1992), and subsequent information gained fiom
charactensation of these lytic enzymes has led the way towards more applied uses for
phage endolysins. More recently, Loessner et al. (1995a) reportec! the use of a
recombinant phage L-aianoyl-D-glutamate peptidase endolysin ply 1 18 of L.
monocytogenes, overexpressed in E. coli to very high levels, as an efficient means of
gently releasing DNA, RNA and proteins fiom Listeria cells. This procedure was
subsequently modified to incorporate a 12 amino acid purification tag without adversely
affecting the lytic activity of the protein, and enabled the production of lysins purified to
more than 90% homogeneity (Loessner et al., 1996b). This endolysin was show to be specific for Listeria cells, but did not lyse other Gram-positive and Gram-negative bacteria. It thus provides an efficient and rapid means of lysing specific cells. Analysis of gene sequences upstrearn of endolysin genes ph118 andply500 revealed highly conserved holin (ho0 genes, which the endolysins require for transportation across the ce11 membrane (Loessner et al., 1995b).
The continued characterisation of phage lytic enzymes (Hertwig et al., 1997; Loessner et al.. 1997b; Sheehan et al. , 1997) will greatly enhance our knowledge of the rnolecular action of such enzymes, but as well, will increase their applicability in other areas of rnicrobiology. For exarnple, such recombinant lysins are currently being evaluated as
ATP extractants for use in bioluminescence assay s (Loessner et al., 1996b), in which intact target cells codd be selectiveiy lysed and rapidly detected by firefly iuciferase. In addition, they offer the potential of being used as antimicrobial agents to eliminate pathogens in the food industry and in medical applications.
1.2.3. PHAGE-BASED FLUORESCENCE
1.2.3.1. Labeling of intracellular phage nucleic acid
Fluorescent stains have been reported as a useful accessory to the use of phage probes, as a rneans of labeling phage nucleic acid within idected cells. Sanders et al. (1 99 1) reported the use of two fluorescent dyes, chromomycin A3 (CA3) and Hoechst 33258
(H33258). to bind and label respectively G-C- and A-T-nch regions of phage T4 DNA present inside infected E. coli cells. Following infection of E. coli ATCC 1 1303 with phage T4, sarnples were placed on ice to intermpt viral replication, washed and ethanol- fixed in preparation for fluorescence staining. Cells were then analysed by dual-laser flow cytometry, and within 35 minutes of phage infection, T4-infected celis could be differentiated from uninfected cells based on fluorescent detection of phage DNA. While this technique was developed to study viral DNA within the infected bacterial host, it has also provided an excellent means of fluorescent labeling of bacterial cells infected by phage. Thus it could be used in conjunction with phages of known host specificity as a detection method for particular groups of rnicroorganisms.
1.2.3.2. Labeling of virus-packaged nucleic acid
Fluorescent labeling of nucleic acid packaged within the virus head has also been used to
directly count viruses in sarnples fiom keshwater and marine environments using
epifluorescence microscopy (Hennes & Suttie, 1995). This labeling technique was
employed to create differential phage probes using fluorescent stains with different
excitation wavelengths for the identification and enurneration of specific groups of
marine bacteria (Hennes et al., 1995). More recently, labeling of phage DNA has been
adapted to enable fluorescent detection of E. di01 57:H7 in ground beef, in conjunction with immunomagnetic separation and flow cpometry (Goodridge, 1997).
Prior to sampling, phage DNA was fluorescently iabeled with the benzoxazoliurn-4- quinoliniurn fluorescent dye YOYO. Ground beef and milk sarnples were artificially
inoculated with low levels of E. coli 0 1 57:H7 and pre-enriched in non-selective media
for 6 hours. Target bacteria were subsequently removed fiom the food matrices by irnrnunomagnetic separation with E. di 0157:H7 Dynabeads, and the fluorescent phage preparation was then used to label the recovered cells. Using both confocal laser scanning and epifluorescent microscopy techniques, in addition to a more automated flow cytometric method, target E. coli 0 157:H7 cells were counted based on their
Iluorescence. The detection limits achieved were approximately 2.2 CFU/g in ground beef and 1 x 1O' CFU/mL in milk. The potential application of this method for detection of other food-borne pathogens is enormous, given the large nurnber of fluorescent stains
cornmercialiy available for phage labeling and the adaptability of magnetic separation
techniques for the detection of a wide variety of bacteria (Olsvik et al.. , 1994).
f .2.4. METABOLIC INHIBITION ASSAYS
A number of methodologies, including electlical impedance, turbidimetry and colorimetry, that rely on metabolic activity and microbial growth as a means of investigating microbial populations are available. Similar to culturai methods, total counts of microorganisrns can be determined by using non-selective media, while specific detection of microorganisms has involved the development and use of suitable differential or selective culture media. By incorporation pH indicatoe, fermentable carbohydrates, organic acids and alcohols, or other inhibitors such as antibiotics, preferential growth conditions for the microorganism of interest are provided at the expense of the background microbial flora. Such technologies have found many applications in food microbiology, particularly with advancement in the automation of such systems (Vasavada, 1993). They are particularly promising for the study and application of bacteriophages. where metabolically active bacterial cells are required for viral replication (Wiggins & Alexander, 1985).
l.t.4.l. Electrical Impedance Of the metabolic methods to be considered in this section, electrical impedance, though
more complicated in theory, has been the most widely studied and developed in ternis of
its microbiological applications. Detection by impedance is based on changes to the growth medium associated with the increase in biomass of metabolically active microorganisms. The components of growth media, such as proteins, carbohydrates and
lipids. are uncharged or weakly charged substrates, but as microorganisms metabolise these compounds, they are converted into more highly charged molecules such as amino acids, lactate, and acetate. These metabolic products increase the conductivity of the medium. However, this increase is small, therefore requiring a microbial level of at least
1O6 cells/mL to be reached before significant changes can be detected. The period between initiation of the test and detection is known as the detection time (D.T.), the value of which is inversely proportional to the number of cells initially present in the test sample.
A large number of reports regarding the electncal detection of specific microorganisms has been published, particularly since the introduction of automated systems such as the
Vitek BactometedB Microbial Monitoring System and the Malthus Microbial Growth
Analyser. There are a number of differences between the two systems, but the most pertinent is their electrical signal measurement. The Malthus analyser measures oniy conductance, while the Vitek system offers more flexibility by measuring either conductance, capacitance, or total impedance. Applications utilising these systems have focused primarily on microbial enmeration, detection of contarninants (Irvine et al., 1989; Kyriakides & Thurston, 1989) and pathogens, particularly Enterobacteriaceae (Bird
et al., 1989; Fryer and Forde, 1989; Petitt, 1989; Prentice et al., 1989; Colquhoun et al.,
1995) and Listeria spp. (Phillips and Grifiths, 1989; Hancock et al., 1993; Rodrigues et
al., 1993, shelf-life studies (Bishop et al., 1984; Phillips & Griffiths, 1985), stenlity
testing of UHT products (Suhren & Heeschen, 1987), and detection of bacteriophages
implicated in cheese starter culture failws (Waes and Bossuyt, 1984; Canninati and
Neviani, 1 99 1).
What these methods have in common (regardless of their mode of detection), and what
has primarily lirnited their application in paihogen detection in foods, is their requirement
for differential or selective media. The use of such media cmincrease testing time
considerably, and does not always allow for the detection of stressed and injured bacterial
cells. Generally they are not capable of differentiating to the species level, and cm aiso
be affected by the growth of contaminating microorganisms associated with the
background microflora of foods, resulting in false-positive detections. The specificity of
bacteriophages and their effect on the metabolic activity of their hosts can offer an
alternative means of detecting specific microorganisms in the presence of the background
Bora of foods. Based on the target population's inability to replicate (in the presence of specific phage) to similar ce11 concentrations achieved by a negative control not exposed to phage, positive detection of pathogens can be confirmed. If target cells are present, they will be infected and lysed by the phage, resulting in delayed detection in samples with added phage. Control samples without phage will have nomal metabolism and growth, resulting in a shorter detection tirne.
This approach was evaluated by Pugh & Arnott (1989) in their development of impedimeûic media for the detection of Salmonella spp. in confectionery products. A combination of two irnpedimetric media, selenite-cystine-trimethylamine oxide-dulcitol medium (SCKID) and modified lysine decarboxylare broth (MLD), was used in an attempt to take into account those Salmonella shains not capable of fermenting dulcitol. but able to decarboxylate lysine. In addition, phage Felix 0-1, estimated to be lytic for
96.1 % of tested Salmonella strains, and phage G47, included to compensate for the tendency of Felix 0-1 not to infect O groups El to E4, were employed to conhthe positive genus identification made by impedance microbiology. Of 8 1 Salmonella strains tested in the Bactometer system by both media and the phage test, ail strains except one strain of S. Houten, were positively identified by at least two of the three protocols, while false-positive reactions by Citrobacterfieundii, E. di,Proteus vulgarir, and
Pseudomonas aeruginosa were obtained in either SC/T/D or MLD, but never both broths.
The non-Salmonella strains were al1 negative according to the phage test. A pre- enrichment step using brilliant green-milk broth was incorporated into the assay when artificially inoculated samples were analysed. In al1 cases but one, where Salmonella isolates were identified by conventional selective enrichment and plating, positive identifications were made by at least two of the three test protocols. The one exception was an isolate of S. Typhimurium which was positively identified by the phage test but neither by the conventional nor the Bactometer methods. Despite the apparent usefulness of this phage-based impedimetric method, confirmation of presurnptive positives made
the method more time consurning, and no merapplication of bactenophages to
impedance has been reported.
Despite its many uses, the direct impedance technique has disadvantages, the most crucial
being that it does not operate under hi& salt concentrations found in some growth media,
and this has prompted the development of alternative rnethods. An indirect approach,
where metabolic production of CO2is considered as an indicator of microbial growth
rather than biomass, prompted the development of the RABIT (rapid autornated bacterial
impedance technique) by Don Whitley Scientific in the W. Detection is based on the
absorption of COz by potassium hydroxide, and its conversion into carbonate at pH levels
above 1 1, which effectively produces changes in rnolar conductivity. These changes are
independent of media composition, and have thus provided fewer limitations to selective
culture media development for detection of bactena such as Pseudornonas (Salvat et al.,
1997), Sulmonella spp. (Blivet et al., 1998), and other microorganisms in foods (Bolton,
1 990), beverages (Dniggan et al., 1993) and potable water (Pndmore & Silley, 1997).
An additional systern, known as the automated BacTracTM4 100 system (SY-Lab, Vienna,
Austria), employs an impedance-splitting (1s) method differing from other impedance
measunng systems by simultaneously detecting and differentiating impedance changes in both the growth medium (termed the M-value) and around the electrodes (E-value). Pless er al. (1 994) reported the use of this system for the detection of Salmonella spp., employing an irnpedance-splitting Salmonella (ISS) medium. Despite a high salt concentration in the medium, changes in E-value up to 100% were achieved in the presence of Salmonellae, while M-values changed by only a few percent. On this basis, a positive result was deemed to be achieved when E-value changes exceeded 15%, and the
ISS medium was demonstrated to be a more efficient Salmonella detection medium than four conventional selective media,
1.2.4.2. Turbidimetry
Turbidimetry is a relatively simple means of estimating bacterial generation times. The use of turbidity measurements takes advantage of the increase in optical density of sterile microbiological growth media associated with growth of inoculated bacteria. Although this method is used to estirnate the rate of growth of microorganisms, generated optical density data can also be used to estimate ceil populations when compared to total plate counts. The automation of optical density readen has been particularly usehl in increasing the number of samples capable of being analysed at one time, and cornputer interfacing has provided an easier means of graphing and analysing generated growth cuves. One particular automated system, the Bioscreen C microbiology plate reader
(Labsystems. Helsinki, Finiand), has provided food rnicrobiologists with a useful tool for the evaluation of turbidity as a means of estimating bacterial levels in food samples.
Capable of incubating up to 200 samples at various ternperatures, the Bioscreen C can be programmed to read the optical density of each sarnple well a particular number of times over a defined incubation penod. In the case of food samples, dilutions are often required
to reduce the background turbidity to a minimum, and the resulting growth cuves can be
analysed using a number of parameten appropriate to the application. These include the
time taken for optical density to increase by a defmed amount, maximum optical density,
slope of the logarithmic growth phase, or tuming point time (the middle of the
logarithmic phase).
The Bioscreen C has been used by researchers to, for example, predict colony fomiing
units in raw mik (Mattila and Alivehmas, 1987) and other foods (Mattila, 1987). and assess the effects of antimicrobials such as bacteriocins (Smand Mattila-Sandholm,
199 1) on bacterial growth. In addition it has proved very useful in the development of data for predictive modeling (McClure et al., 1993), and most recently has been used to
investigate recovery of single heat-injured Salmonella cells by rneasuring the lag phase of cells pre-e~chedin non-selective media (Stephens et al., 1997). To date bacteriophage- based turbidimetric methods for the detection of food-borne pathogens have not been reported.
1.2.4.3. Colorimetry
The suitability of colour changes to indicate the statu of biochemical reactions is evident by their extensive application in numerous methodologies used in microbiology, such as differentiai and selective culture media, biochemical tests for identification of isolates, and enzyme-linked immunosorbent assays (ELISA). This approach has also been used in the development of media for both electncal impedance, where the incorporation of a colour change is typically used in conjunction with detection times and cuve characteristics to confm a particular microorganisrn, and colorimetric assays where the colour of growth media containing pH indicators is monitored over the for changes corresponding to, for example, fermentation of carbohydrates. This has been applied particularly in dairy rnicrobiology where colorimetric tests have been successfully developed to evaluate the presence of antibiotics, coliforms, pathogens or psychrotrophs, enzyme activity, stale flavour development. lactic culture activity, and shelf life of dairy products. Instrumentation uicludes the the OrnnispecTM4000 bioactivity monitor
(Richardson et al., 1994) and the versatile LabSMART, colour scanner and software package (Richardson, 1997). Both instruments operate on the bais of reflectance colour readings, which eliminates any problems associated with background nirbidity of food sarnples (Richardson et al., 1988). Phage-based colorirnetric assays have not been reported to date.
1.2.5. MISCELLANEOUS PHAGE PROBE APPLICATIONS
1.2.5.1. High-performance liquid chrornatography
A High-Performance Liquid Chromatography (HPLC)method for the detection of
Salmonella spp. utilising the Felix 0-1 phage was reported in 1983 (Hirsh & Martin, l983a). Using an E-1000 u-Bondagel gel permeation column, it was dernonstrated that HPLC could be used to distinguish and detect Salmonella spp. based on the detectable
increase in bacteriophage titres associated with phage replication and ce11 lysis.
Susceptible host cells and Felix 0-1 phage were mixed and allowed to interact for two hours during which replication of phages would occur. The aqueous phase was removed, treated with chloroform, and centrifuged to remove ce11 debris. This bacteriophage supernatant was then injected into the HPLC column and the resulting chromatograph was analysed for a specific peak corresponding to the retention of bacteriophage previously detennined to occur 3.2 minutes after injection. It was detemiuied from the results obtained that a minimum bacteriophage concentration of approxirnately 1O9
PFU/mL was required to give a reliable detection (response of at least lOmm over background), corresponding to around 3x1O6 bacteria per mL required in the reaction mixture. This peak occurred oniy when SaIrnonella spp. were present in the reaction mixture, and a positive detection was demonstrated even when non-Salmonella spp. were aiso present. The high levels of bactena required for detection was a defite shortcoming of this method, and was subsequently addressed by the use of filtration and enrichment in follow-up research outlined below.
These modifications were incorporated in order to detect Salmonella spp. in milk using this HPLC method (Hirsh & Martin, 1983b). Large pore electropositive filters were ernployed to remove bactena fiom artificially-inoculated milk samples, and subsequently treated with alkaline broth media to elute the trapped microorganisms from the filter.
This method was estimated to recover about 40% of the bacteria onginally added to the mi&, and subsequent ovemight e~chrnentin the presence of brilliant green dye was
adequate to ensure that 75- 100% of the total bacterial population consisted of
salmonellae. Following enrichment, the cells were centrifbged and washed, resuspended
in CaC1,-supplernented nutrient broth, and bacteriophage was added to defuied ce11
densities. Following preparation of bactenophage supernatants (as previously descnbed),
25S volumes were injected into the HPLC coiumn, and the presence of a 3.2 minute
bactenophage peak was indicative of the presence of Salmonella in the original sample.
Less than five salmonellae per mL of mik were detectable by this method, resulting in a
positive detection result within 24 hr of sarnpling. However, as previously reported
(Hirsh & Martin, 1983a), problems arose when certain SuZrnonella serotypes were not
infected by this phage. The false-negative results were explained by the potential
presence of either prophages, known to limit or inhibit replication of Felix 0-1, or bactenophage-restricting plasmids. Despite the improved detection limit achieved in this research, the use of this assay has not been more recently reported.
1.2.5.2. Immobilised-phage capture assay
More recently, the 'Sapphire' lytic phage fiom Amersharn International has been evaluated in a novel immobilised capture method for the separation and concentration of
Salmonella strains by Bennett er al. (1997). Using both a microtitre plate and dipstick method, phage particles were allowed to adhere to the solid phase, washed several times to remove unbound virus, and treated with bovine semm albumin to minimise non- specific binding of cells. The phage-coated solid phase was then exposed to several
Salmonella serotypes uicluding S. Typhimurium and S. Enteritidis, and non-salmonellae
including E. coli, Proteus rnirubilis, and Hafnia alvei in Iiquid suspensions. Using the polymerase chah reaction (PCR), the method was evaluated for (i) its ability to remove cells From the suspensions and (ii) specifically distinguish and remove Salmonella cells
From mixed cultures of Enterobacteriaceae. A comrnercially available Salmonella- specific BAX PCR system was chosen, but used in a procedure slightly different from that suggested by the manufacturer, previously demonstrated to significantly increase sensitivity of the system. Data obtained showed that capture of Salmonella spp. was possible using both solid phases mentioned above, but, based on the detection limits of the PCR assay, high nurnbers of cells (around 10' CFU/mL) were required to subsequently obtain a positive end result. Specific capture of S. Typhimuriurn in the presence of equal nurnbers of competing Enterobacteriaceae using the microtitre plate test format was also a viable option, but again, high numben of cells were required for a positive result. While these results were encouraging, a number of problems were encountered in this assay. Most irnportantly, S. Ealing and S. Arizonae were not detected by this capture method, and the 'Sapphire' phage was subsequently show not to infect these particular organisms. Second, phage capture was very inefficient and it is likely that some phage particles were binding to the solid phase via the tail rather than the head, thus rendering them useless in the ce11 capture process. The small surface area of the microtitre plates may also have contributed to the poor capture rate. In addition, unusually high ce11 counts obtained using control (uncoated) dipsticks demonstrated that phage or ce11 clurnping may have aisen as a result of uneven surface chernistry. A
solution to the surface area issue is the use of magnetic spheres as a solid phase, but
further work in this area was not reported by these or other authors.
1.2.53. Phage-linked immunosorbant assays (PHALISA)
The use of phages as reporter molecules has also been extended to the developrnent of
phage-linked immunosorbant assays (PHALISA), similar to the ELISA method, where
the amount of phage bound by irnrnobilised antigen-antibody complexes cm be related
back to the nurnber of target cells present by its subsequent infection of susceptible
bacterial cells. Block et al. (1989) employed bacteriophage Ml 3, engineered to contain the E. coli lac2 gene, as a "bio-amplifiable tag" for the detection of herpes simplex virus
(HSV). HSV-infected cells were incubated with HSV-specific monoclonal antibodies raised in mice, followed by rabbit anti-mouse serum and mouse anti-Ml3 senim. The resulting complexes were incubated with phage M 13. The bound phage was eluted and then quantified by plaque assay and acquired beta-galactosidase activity in phage-infected
E. coli cells. This PHALISA assay was more sensitive than an ELISA method under similar test conditions and demonstrated the potential for detecting bacteriai and viral antigens in foods.
1.2.5.4. Phage amplification assays A similar role for phage has been exploited in the development of methods to detect
Pseudomonas aenrginosa and Mycobocterium t uberculosis (Stewart, 1997). Kno wn as the 'phage amplification assay', it exploits the lytic cycle of bactenophage to indicate the presence of target bacterial cells. In the case of M. iuberculosis, the British Company
Biotec Diagnostic has developed a test called BioPhab (Graham, 1W6), relying on infection of mycobacteria present in clinical samples by addition of a specific bacteriophage. Following infection of cells, samples are first treated with a specid virucidal reagent to destroy extemal virus particles not already involved in the infection process, and then the reagent is neutralised to prevent later destruction of phage progeny.
If susceptible target cells are not present in the sample, then al1 bacteriophage will be destroyed. Otherwise, the bacteriophage replicates normally within the host ce11 and eventually lyses the ce11 wall, releasing its progeny into the external medium. Resulting virus particles can then be amplified Merusing a non-pathogenic host and detected by plaque assay. This method has been reported to have a detection bitof about 100 mycobactena per mL of sample, detectable within 10 hours, and offers the advantages of destroying virulent M. tuberculosis cells during testing if they are indeed present in the clinical sample, simplicity of use with no expensive instrumentation required, and a less expensive application of phages as opposed to more costly clonuig techniques mentioned previously. As with other phage-based detection methods, resistance to a particular bacteriophage is the most likely problem associated with this approach. However the use of phage cocktails could be a means of overcoming this shortcoming. 1.3. CONCLUSIONS
Bacteriophages are versatile biological particles, capable of identifying host bacteria to levels well beyond the capabilities of many conventional microbiological methods. Their demonstrated suitability for bacterial typing and cloning has resulted in their application in various methodologies, particularly the detection of pathogenic bacteria. However, many of these methods have employed molecular cloning techniques, which. although sensitive in nature, are rather time-consuming and costly endeavourj.
Phages are known to have a universal requirement for a metabolically active bacterial host in order to replicate, but few researchers to date have taken advantage of this fact to explore their potential in phage-based metabolic detection assays. A number of such technologies are available, including impedance, turbidimetry, and coionmetry. These offer both a rapid and sensitive means of detecting viable microbid cells, and an ideal oppominity to evaluate the behaviour and suitability of phages in more specific detection applications. Another phage property is that of host ce11 lysis which can be exploited in the development of specific extractants for the ATP bioluminescence assay. Phages specific for various bacteria could thus offer a means of selectively analysing samples for the presence of pathogens, or other bacteria of choice, as an alternative to the non-specific chernical extraction of ATP curredy available.
The lirniting factor in the advancement and application of phage-based methodologies will always be the bacteriophages themselves. For example, the generic detection of
Salmonella spp. using the broad-spectrum phage Felix 0-1 has been extensively documented, but it is not a viable option for more specific detection to the species or serovar level. Continuing phage isolation and characterisation still rernains the best way of identiQing potentially usehl bacterial viruses, and this avenue, in conjunction with the development of phage-based metabolic and bioluminescent detection rnethods for food- borne pathogens will be explored in this thesis. CHAPTER WO
ISOLATION AND CHARACTERTSATION OF BACTERIOPHAGES ACTIVE
AGALNST TAE GENERA LTSTERL4, mOlVELLA,AND ESCHERICHZ4
2.1 INTRODUCTION
A plethora of bacteriophages are commercially available for bacterial typing and microbiological research applications. While this provides a convenient and consistent source of virus, little information is available regarding their origins and host specificities, a notable exception being the extensively studied Snlmonella phage Felix O-
1. The broad-spectnim idectivity of the Felix phage has facilitated its application in a nurnber of Salmonella-specific detection methods (descnbed in Chapter One), but none to date appear to be capable of a single species-specific Ievel of detection. This aspect is of particular interest when considering detection methods for genera such as Listeria where only one species, rnonocyfogenes, has been reported as a human pathogen, and
Escherichia, where serovar differentiation is usefûl for detection of potentially pathogenic strains. To date, al1 isolated Listeria phages have been identified as genus-specific
(Rocourt, 1986; Loessner, 1994), while E. coli phages demonstrate variability in their host ranges (Samadpour et al, 1993). The continued isolation and characterisation of new phages is thus particularly important as a means of increasing the numbers and host range information pertaining to their suitability in rnicrobiological applications. Given that bacteriophages require bacterial hosts in order to propagate, bacteria and their specific bacteriophages cm be easily located within the same environmental niche. This offers an advantage to the microbiologist, where appropriate microbial ecosystems can be exploited to provide phages with the host specificity required. In addition, the hypothetical isolation of a bacterial strain unidentifiable by alternate means can be readily addressed by probing the environment for phages capable of infecthg such a strain. This offen great flexibility in their application, where detection methods can be simply upgraded by the addition of newly isolated phages, and "cocktails" of phages can be used
(similar to starter culture combinations and rotations) to minimise and control the development of phage resistance.
This chapter serves as an introduction to selected commercial and environmentdly isolated phages active against the genera Listeria, SalmonelIa, and Escherichia.
Characterisation information acquired by bacterial typing and virai taxonornic properties is presented, and their suitability in bacterial detection applications is discussed.
2.2. MATERIALS AND METHODS
2.2.1. Bacterial cultures
Pure cultures of bacterial strains were stored at 4OC on slants prepared from Trypticase
Soy Agar (TSA; Difco Laboratories. Detroit, MI) and also at -80°C in Nutrient Broth (Difco) supplemented with 15% giycerol. Slants were restreaked every month to maintain ce11 viability.
Phage host L. monocytogenes ATCC 23074 (originating fiom the Department of Medical
Microbiology , Stanford Universify ) was purchased fiom the Amencan Type Culture
Collection (Rockville, MD). Source and serovar idormation was unavailable upon request. In addition, thuty-two Listeria isolates of both environmental and cluiical ongins (donated by Dr. P. Slade, University of Guelph) were employed for phage charactensation (see Tables 2.2. and 2.3.). As far as possible, where the source was identified, food isolates were selected for testing but other environmentai and human isolates were also used. Strains were also selected based on avaiIabIe serovar information, with an effort made to include as many different ones as possibie.
The identity of al1 species of Listeria to be screened was first confirmed by Gram staining and a positive catalase test, followed by carbohydrate fermentation profiles including hydrolysis of aesculin on Oxford agar plates (Difco), and beta-haemolysis according to the biochemical profile in Table 2.1. (adapted from Jay, 1992).
Cultures were screened easily using the Cathra RepliScan plater system (AutoMed, Inc.,
St Paul. MN). Eighteen hour TSB cultures were pipetted into individual sterile wells of a Table 2.1. Biochemical profiles used for differentiation of Liîteriu species (Adapted from Jay, 1992)
Biochemical profiles
Listeria species Xylose Rharnnose Mannitol Aesculin P-haemolysis L. rnonocytogenes No" Yes No Yes Yes L. innocuu No Yes No Yes No L. seeligeri yesb No No Yes Weakd L. welshimeri Yes Variablec No Yes No L. ivanovii Yes No No No Yes " Negative carbohydrate fermentation Positive carbohydrate fermentation ' Carbohydrate fementatiori strain-variabIe Strains weakly haemolytic stainless steel mould, and the 37-pin plater head was used to transfer smdl amounts of each culture fiom the mould ont0 agar plates placed undemeath. Carbohydrate plates were prepared using Bacto purple agar base (Difco), supplemented with sterile carbohydrate solutions to give a fmd concentration of 1%. Haemolysis was examined on
TSA agar plates supplemented with 5% sheep blood (Becton Dickinson Microbiology
Systems. Cockeysville, MD). Oxford agar plates (Difco) were made according to manufacturer specifications. Quality control cultures obtained frorn C. Larkin
(Laboratory Services Division, University of Guelph) were used throughout to ensure appropriate interpretation of results. Plates were incubated at 37°C for 18-24 hours and then examined for colour changes &om purple to yellow associated with carbohydrate fermentation, black bacterial growth resulting from hydrolysis of aesculin in the presence of femc ions, and zones of clearing consistent with haemolytic activity.
2.2.1.2. Salmonella strains
Twenty Salmonella strains representing 20 different serovars and belonging to groups
B,C, and D (Table 2.4.) were obtained fkom Dr. J. Chen (Chen & Griffiths, 2996).
Quality control cultures S. nphimurium ATCC 14028 and S. Enteritidis ATCC 13016 were obtained from C. Larkin (Laboratory Services Division, University of Guelph). Al1 cultures were Gram stained and drops of broth culture were dropped ont0 Modified
Semisolid Rappaport-Vassiliadis (MSRV) medium (Difco Laboratones, Detoit, MI).
After incubation at 42OC for up to 24 hours, Salmonella isolates were re-confïrmed by the appearance of opaque halos.
2.2.1.3. Escherichia coli strains
Eighteen ver0 cytotoxigenic E. coli (VTEC)stralns (Table 2.5.) were obtained from
Heaith of Animais Laboratory, Guelph, ON. Twelve non-VTEC strains (Table 2.6.) tested at Health of Animais Laboratory were obtained fiom Dr. S. Chen (Laboratory
Services Division, University of Guelph, ON). Al1 mains were both Gram stained and streaked on MacConkey Agar, incubated at 37T,to resodirm identification.
2.2.2. Bacteriophages and propagating strains
Bacteriophages suspended in ~arnbda-Ca" baer with added chloroforrn were stored in steriie 15mL tubes at 4OC, while 1mL aliquots (without chlorofom added) were placed in the -80°C freezer for long-term storage.
2.2.2.1. Listeria phage
Lisleria phage ATCC 2307443 1 (originating fiom the Department of Medical
Microbiology. Stanfbrd University) was purchased fiom the American Type Culture
Collection (Rockville, MD). No further information regarding source was available upon request. The phage was propagated using L. monocytogenes ATCC 23074. 2.2.2.2. SaIrnoneNa phages
Salmonella phage Felix 0-1 was obtained fiom S. Favrïn (University of Guelph), and propagated using S. Typhimurium ATCC 14028 (C. Larkin). Fhage SJ2 was isolated korn a chicken egg (Dr. S. Jassiin, University of Guelph, ON), but Merdetails regarding isolation techniques were not made available. The propagating strain used was
S. Enteritidis S 1O (Chen & GriEths, 1 996).
2.2.2.3. E. coli phage
E. coli phage was isolated according to the protocol of Ahmed et al (1999, with the following modifications: 1m. of faeces was mixed with lOrnL TSB, plus 1 OOpL of an overnight culture of E. coli G2-2 (A.Toner, University of Guelph), and incubated overnight at 37OC. The suspension was then centrifuged, and filtered through a 0.45pm sterile disposable filter (Mandel Scientific, Guelph, ON). Instead of using the drop technique to screen isolated phages, the agar overlay method used for enumeration in section 2.2.3.3. was employed to better view any plaques formed. A number of E-coli isolates and phages were initially available, and these were first screened in order to select a representative phage for subsequent analysis. Phage AT20 and its host G2-2 were chosen, based on strong lytic activity and broad infection panems.
2.2.3. Propagation and enumeration of phages To obtain sufficient volumes and titres of al1 four bactenophages andysed, the following
propagation and enurneration methods were employed (Jassim, unpublished). The
primary advantage of this method was that only small volumes of bacteriophage were required for propagation, and thus most of the lysate could be used for experimentation purposes.
2.2.3.1. Phage Iysis of host bacteria
Lambda buffer was prepared according to the formulation: 2 Sg/L MgS0,. 7H,O; 0.05glL gelatin; 6mLL 1M tris buffer (adjusted to pH 7.2). Following autoclaving at 12 1OC for
15 minutes, filter-stenlised CaC11.2H20 was added to lambda baer to give a 1-Ca" bufTer with a final ~a"concentration of 5m.M The use of this buffer ensured that suitable electrolytes were supplied in the phage-bacterial host reaction mixture to facilitate phage attachent. Serial dilutions of bactenophage (100pL phage in 900pL A-
C'a2+ buffer) were prepared in sterile eppendorf tubes, and placed on ice. When al1 dilutions had been prepared, IOOpL of each was added to 1OOpL of an 18 hour old
Trypticase Soy Broth (TSB) culture of the appropriate host bacteria in a sterile plastic
5mL capped tube. Afier gentle mixing of tube contents, the phage-bacteria mixtures were incubated at 37°C for 10 minutes to allow attachent of bactenophages to commence.
At the end of the incubation period, 3.2rnL of rnolten 0.4% TSA (cooled to approximately
50-55OC before use) was added to each tube. Contents were gently mixed, and quickly overlayed on the surface of pre-poured 1.5% TSA plates. Plates were left at room temperature to solidifi, and then placed in a 37OC incubator for up to 18 hours.
Following incubation, plates were examuied for zones of clearing, termed "plaques". Al1
plates showing complete lysis and the first successive plate to demonstrate individual
plaques were chosen for harvesting phage. Four rnL of A-ca"' buffer was added to the
first plate and the top layer agar was scraped off using an alcohol-fiamed glas hockey stick. This agar was then added into the next plate and the top layer agar of that plate scraped off and mixed with the additional agar. This procedure was continued until ail the selected plates had been scraped and ail the top layer agar was contained in the final scraped plate. The top layer agar was then transferred into a stenle centrifuge tube using a 25mL pipette. Any remaining agar (and phage) was swirled off by adding another 4mL of A-Ca2*buffer to each scraped plate, which in tum was pipetted and added into the centrifuge tube. The tube was then placed on ice while other phage lysis plates were prepared for the cleaning step outlined below. Care was taken at al1 time to ensure phage preparations remained fkee of contaminating phage.
A control preparation was prepared for each of the four phages by the same method outlined above, but in the absence of phage. This resulted in baer solutions treated in exactly the same way as phage lysates that could be used as "minus phage" controls in further experimentation.
2.2.3.2. Cleaning of phage lysates Chloroform (CHCI,) was added to phage preparations in centrifuge tubes (approximately
3mL CHCl, per 5mL phage lysate). The tubes were weighed and balanced with additional chloroform, mixed gently and placed on ice for 10 minutes. Tubes were then centrifuged at 7000 rpm (rotor J-2 1) for 20 minutes at 4OC in a Beckman mode1 J2-M 1 refngerated centrifuge (Beckman instnunents (Canada) inc., Mississauga, ON). Three distinct phases were observed, where a layer of agar and cells separated the phage in buffer From the chloroform added. This phagelbuffer supernatant was carefully pipetted off into a sterile 15mL polystyrene tube, and then filter-sterilised using a 0.22pm stenle disposable syringe filter unit (Mandel Scientific, Guelph, ON). Two 1mL aliquots of each phage preparation were removed to sterile eppendorfs: one tube was placed in -80°C freezer storage. and the second tube was stored at 4"C, the contents intended for use in subsequent phage preparations.
2.2.3.3. Enurneration of phage preparations
Enurneration of prepared phage lysates was conducted according to the procedure used for phage propagation of serial dilutions above. After incubation, plates were observed for lysis plaques, each one assumed to originate fiom infection by one bacteriophage.
Duplicate plates falling within the 30- 100 plaque range were counted. The "titre" of the phage preparation was calculated by multiplying the average plaque count by the inverse of the dilution factor and the volume of phage used in the assay. 2.2.4. Phage typing of bacteria
The plaque assay was employed to examine the specificity of each phage when exposed to strains of the proposed host genus (see results tables for more information). A positive result was indicated by the formation of plaques following a 24 hou penod of incubation of TSA plates overlaid with 0.4% TSA containing 100pL each of bacterial test strain and phage. Incubation was increased to 48 hours in the case of Lisleria strains and phage, according to the fmdings of Loessner & Busse (1 990).
2.2.5. Transmission electron microscopy of phages
Drops of phage suspension (approx. 10" PFUImL) were placed on individual formvar- coated copper grids and allowed to sit for two minutes. Sarnples were wicked off with filter paper and a drop of 2% phosphotungstic acid was added for 30 seconds. Following removal of excess acid with filter paper, grids were air-dried and examined ushg an
Hitachi H-7 100 Transmission Electron Microscope. Micrographs were taken for al1 four studied phages.
2.3. RESULTS AND DISCUSSION
2.3.1. Phage enumeration and plaque morphology Bacterid phage suspensions were propagated with appropriate bacterial hosts until titres of approximately 10' ' PFUIrnL were obtahed for each phage. This required several propagations in the case of environmentally isolated phages where the initiai titre was substantidly lower at only 10'-1 O6 PFU/mL. It is well documented that plaque size and morphology are dependent on the virus, the host, and the conditions of culture, and cm be used as important criteria for the differentiation of bacterial species and saaùis (DuBow,
1994). Inspection of plaque characteristics for each of the host-phage combinations re- confïied this phenornenon. Listeria phage ATCC 23 074-B 1 revealed small, turbid plaques, with a diarneter of 4 mm. Their turbid nature suggests a lysogenic origin, and according to Rocourt (1986), most Listeria phages are indeed isolated from lysogenic strains. The other phages were identified as vident: Salmonella phages Felix 0-1 and
SJ2 produced larger (-2rnm) clear plaques, while the largest plaques were produced by E. coli phages (3-hm).resuiting in completely clear overlays for the fust seven dilutions.
2.3.2. Phage typing of Listeria spp.
In order to establish the host range of this phage, a number of cultures were obtained of various species, strains, serovars, and sources in order to establish whether the tested phage demonstrated specific infection spectra based on any of these properties. The results for Listeria phage ATCC 2307443 1 and Listeria species monocytogenes and innocua are shown in Table 2.2. Table 2.2. Typing of Listeria monocytogenes and innocua isolates by phage ATCC
II Listeria SDD. 1 Isolate Source Serovar Phage sensitivity II L. monocytogenes 1 ATCC N.K. N.A. Yes
II L. monocytogenes 1 ~487 Cheese 1 112b 1 No II L. monocytogenes 1 L 19 Faeces 1 3a 1 No 1 L- monocytogenes 1 y.8 Meat 1 N.A. 1 Yes L. monocytogenes II L. monocytogenes 1 Mg Meat 1 N.A. 1 No II L. monocytogenes 1 714 Meat 1 1/2a 1 No II L. monocytogenes 1 2 II L. rnonocytogenes 1 409 Yes L. monocytogenes L 18 Faeces 1 3b 1 No L. innocua 309 Raw rnilk 1 N.A. 1 Yes II L. innocua 1 LA1 N.K. 1 N.A. 1 No L. innocua M7 Meat 1 NA. 1 No L. innocua TRUSB N.K. I N.A. I Yes L. innocua 3 76 Rawrnilk 1 N.A. 1 Yes L. innocua H7 Raddish 1 4ab 1 Yes L. innocua Hl6 Yes II L. innocua Mushroom 1 4a 1 Yes L. innocua Faeces 1 6b 1 Yes L. innocua N.K. 1 N.A. 1 Yes N.K.: Not known N.A.: Not available ( ): Information incomplete/inconsistent Interestingly, only two other isolates of L. monocytogenes apart fkorn the propagating
strain (Le. 30% of tested strains) demonsûated sensitivity to the Listeria phage, one of
unknown serology and the other belonging to serovar 4b. Other serovars including 1/2a,
1/2b, 3% 3b, and another 4b isolate of human ongin gave negative results. However,
80% of L. innocua isolates of serovars 4% 4ab, and 6b, demonstrated sensitivity to the
Listeria phage. Sensitivity to phage was low for both L. ivanovii (33%) and L. seeligeri
(25%), while 100% of L welshimeri isolates were sensitive (Table 2.3.). Overall, the
majority (83%) of known serovar 4 isoiates were sensitive to phage ATCC 23074-B 1.
Eighteen isolates were of undetermined serology and of these, 1 1 isolates were sensitive
to phage infection. Based on the Iack of serological information, and the low nurnbers of
isolates of each serovar tested, no other trends could be determined. In terms of the source of isolates, 7/16 (43.8%) food isolates were phage-sensitive (mostly serovar 4), while 24(50%) fàecai samples (3b and 6b) and the only water sarnple (also 4b) tested positive. OF 10 isolates of unknown source, 50% demonstrated sensitivity. Overail typability of Listeria species, disregarding source and serovar, was 57.6%.
Regardless of the rather limited information available about source and serovar of the
Listeria isolates tested, some sirnilarities were observed between this typing study and another conducted by Loessner & Busse (1 990). In cornmon with their fmdings, phage susceptibility was highest (83.3%) among isolates of serovar 4. In addition, they identified serovar 3 of L. monocytogenes as behg phage-resistant, which was also observed in this study (although only 2 isolates were tested). However, none of the four Table 23. Phage typing of Listeria species ivanovii, seeiigeri, and wehhimeri using phage ATCC 23074-BI
Lisieria spp. Isolate Source Serovar Phage sensitivity L. ivanovii Lab N.K. N.A. Yes L. ivanovii KC1714 N.K. N.A. No L. ivanovii ATCC 19 1 19 N.K. N.A. No L. saeligeri 472595 PA N.K. N.A. Nc L. seeligeri F4080 N.K. N.A. Yes L. seeligeri HI 1 Cabbage la No L. seeligeri LL4 12 N.K. N.A. No L. welshimeri L18 Faeces 3b Yes L. welshirneri M6 Meat N.A. Yes L. welshirneri Ml 1 Meat N.A. Yes L. welshirneri Ml4 Meat N.A. Yes L. welshimeri 502 -Meat 4a Yes N.K.: Not known N.A.: Not avaiiable known serovar 1 isolates demonstrated sensitivity to this phage, which conflicts with the
90.6%typing rate achieved by Loessner & Busse. In their study, these strains were lysed by phages propagated using serovar 1 hosts, it is therefore possible that phage ATCC
23074-8 1 was not propagated by such a strain in this study. Unfominately this remains uncertain as the serovar information for the ATCC host used here was not available upon request and was not initially tested. What should also be taken into account is that these data presented are the result of typing using only one phage, while 27 phages were included in Loessner & Busse's typing scheme. In addition, induction techniques to remove potentially problematic prophages were not used before testing strains. It is possible that prophage-rnediated immunity may have played a role in the apparent
"resistance" of isolates to phage ATCC 23074-8 1. However, in the present study, the percentage of Listeria isolates typable using only this one phage (57.6%) was within reported ranges of 34-80% (Rocourt, 1986) and 45-9 1% (Loessner et al, 1994), with variations dependent on the type and nurnber of phages employed.
Isolates were confirmed to the species level before testing by a battery of biochernical tests, including haemolytic activity. It is worth noting however that chromosomally- encoded haemolytic activity cm be lost spontaneously, or as a result of gene blockading or loss of promoter kctionality (Hof & Rocourt, 1992). As haemolysis is the only property able to discriminate between monocytogenes and innocua strains in routine biochemical testing, it is possible that non-haemolytic monocytogenes strains were incorrectly identified as innocua. This could explain, at least in part, the unexpectedly hi& susceptibility of innocua strains to the Listeria phage as compared to only 30% of
monocytogenes strains. Additionally, serotyping was not repeated . It is thus possible
that initial typing results were inaccurate, or that switching of cultures may have occurred
during the loan of cultures &or repeated subculturing onto fiesh aga.slants. According
to information acquired by Loessner and Busse (1990), it is not unusud for serotyping at
a later date to deviate fiom initial results, particulariy when cultures have been stored for
extended periods of time as was the case here. It is thus likely that, although typing
strains before use would have been a better course of action, conflicting results would
have merely confused the issue Mer.
What is apparent fiom this research is that phage ATCC 23074-B 1 does not exhibit
specificity to only one species or serovar, previously demonstrated by both Rocourt and
Loessner's phage sets (Loessner & Busse, 1990; Rocourt, 1986). While a more complete serovar and source testing scheme wouid be better able to address this and other observations more conclusively, published findings suggest that this is unnecessary as most listerial phages have been reported to demonstrate a broad lytic spectrum (Jay,
1992). This would imply the involvement of a fairiy cornmon receptor site for phage attachment. The only accessible ceil wall polymers of Listeria spp. are the O antigenic deteminants: poly (nbito 1phosphate)-type teichoic acids. The former genus members grayi and rnurrayi differ from the five true Listeria species by containing modified lipoteichoic acids, and it is thought that these modifications are the reason for their resistance to phages that lyse the other species of Listeria (Jay, 1992; Loessner & Busse, 1990). However, as phage sensitivity does not appear to be serovar-specific, based on
this and other research, the phage receptor sites are likely to be different &om the O
antigenic determiriants (Jay, 1992).
2.33. Microscopie characterisation of L&eria phage ATCC 23074-B1
Previously identified Listeria phages have been shown to contain double-straoded DNA and belong to one of two virus families: the Myoviridae, with contractile tails, and the
Siphoviridae, with non-contractile tails. Myoviridue include the well-known vident T2,
T4 and T6 coliphages, while Siphoviridae are charactensed by the À. phage group, and are
lysogenic (Maniloff et al, 1994). Both the observation of plaque turbidity and the transmission electron micrograph obtained for phage ATCC 2307443 1 (Figure 2.1) indicate that this phage is a member of the Siphoviridae phage family. This family includes phages active against Bacillus, Lactobacillus, Lactococcus, Leuconostoc,
.Micrococctis, and Srreptococcus (Maniloff ef al, 1 994), dl close1y related to L isferia.
In Recourt's previously cited phage study (1986), only two L. hnocua phages of a total
55 Listeria phages were identified as belonging to the Myoviridae farnily. The remaining
53 phages of L. rnonocytogenes, L. ivanovii, and L. innocua were identified as siphoviruses, and could be subdivided into three groups based on tail length. Similarly,
Loessner et al (1994) investigated the taxonomy of 20 isolated Listeria phages, and 19 of them were identified as siphoviruses, with isometric (icosahedral) capsids and long, Figure, 2.1. Transmission electron micrograph of Listeria phage ATCC 23074-B1 flexible. non-contractile heads. Interestingly, defective Lisferiaphages lacking a head region. termed monocins, lose their genus specificity and uifect closely related genera such as Stuphylococcllr and i3aciZlus (Jay, 1992; Rocout, 1986).
2.3.4. Phage typing of Salmonella spp.
A more complete set of Salmonella isolates aiready determined to belong to groups B, C,, and D, (Chen & Griffiths, 1996) were made available to evaluate the infection spectra of
Salmonella phages Felix 0-1 and S12. Sensitivity results are shown in Table 2.4.
As compared to the results obtained for Listeria phage ATCC 23074-B 1, sensitivity to
SufmonelZa phages was much more clear-cut. Of 22 Salmonella isolates tested, al1 but one, S. Tennessee, demonstrated sensitivity to phage Felix 0- 1, with each of two tested S.
Typhimurium and S. Enteritidis isolates also positive. This overall sensitivity of 95.5% is consistent with reported estimates of 96.1-99.5% obtained for over 5,000 screened isolates (Gunnarssson et al, 1977; Kallings, 1967; Welkos et al, 1974). Some researchers have reported resistance of isolates of S. Enteritidis and S. Agona to Felix 0-1 infection
(Hirsh & Martin, 1983). However, this was not observed in this phage screening, although only two isolates of S. Ententidis were tested. In those cases, resistance was attributed to the possible presence of (i) a prophage, making these isolates immune to infection, or (ii) the presence of plasrnids capable of restricting bacteriophages. This Table 2.4. Typing of selected Group B, C, and D Saimonella isolates by phages Felh 0-1 and SJ2
Salmonella isolate Group Number of Sensitivity to: isolates tested Feiix 0-1 SJ2 S. Typhimurium B 2 Yes* Yes* S. Heidelberg B 1 Yes No S. Saintpaui B 1 Yes No S. Bredeney B 1 Yes No S. Schwmenground B 1 Yes No S. Agona B 1 Yes No
- -- - S. Reading B 1 Yes No S. Indiana B 1 Yes No S Brandenburg B 1 Yes No
S. Infantis CI 1 Yes No S. Thompson CI 1 Yes No S. Mbandaka CI 1 Yes No S. Braenderup CI 1 Yes No S. Ohio CI 1 Yes No S. Montevideo CI 1 Yes No S. Oranienburg CI 1 Yes No S. Tenessee C, 1 No No S. Ententidis DI 2 Yes* Yes* S. Berta Di 1 Yes No S. Panama D, 1 Yes No * Both isolates positive may also explain the resistance of S. Tennessee to Felix 0-1. According to published
information, S. Tennessee is typically sensitive to Felix 0-1 phage (Kallings, 1967; Pugh
& Arnott, 1989). Given that a retest of the isolate gave another negative resdt, testhg of
other S. Tennessee isolates or irradiation to induce prophage followed by retesting would
be the mon appropriate course of action.
In contrast to Felix 0-1, phage SJ2 demonstrated a much narrower host specificity by
infecting only two species: S. Typhimurium and S. Enteritidis, of groups B and D,
respectively. Group C, isolates remained unaffected. The reason for this phage's specificity is probably related to more specific receptor site(s) utilised by this phage.
S. Typhimuriurn and S. Enteritidis belong to different serogroups, but they have some antigenic determinants in comrnon, which are not shared by group C Salmonella serovars.
In particular, O antigens 1 and 12 are cornmon to groups A, B, and D, but are not found in groups C, and C2. However, this still does not explain why other isolates of groups B and
D were not sensitive to phage SJ2.
It could be speculated that phage SJ2 uses a sirnilar receptor arrangement to that known for other SalmoneIla bactetiophages. For example, N-acetylglucosamine found in the core polysaccharide of lipopolysaccharide (LPS) is an essential part of the receptor for adsorption of phage Felix 0-1 (Lindberg & Holme, 1969). These researchen also demonstrated that the structure and composition of O side-chains (the outer polysaccharide component of LPS) deterrnined phage adsorption. The temperate phage
P22, which infects S. ïyphimurium, does so by the enzymatic action of the P22 baseplate
protein. It binds and hydrolyses the O side-chah, which effectively serves as an initial
phage receptor. A secondary receptor is also implicated in virion attachent, resulting in
transfer of phage DNA Uito the cytoplasm of the host ce11 (Poteete, 1994). It is therefore a reasonable proposition that more than one LPS component dictates the specific attachent of Salmonella phages, althou& Merresearch is required to ascertain the nature of such receptor sites.
2.3.5. Microscopic characterisation of Saimonella phages Felix 0-1 and SJ2
Transmission electron micrographs of phages Felix 0-1 and SJ2 (Figures 2.2 & 2.3) placed hem both into the Myoviridoe family characterised by phage T4, based on relative head sizes and taii lengths (Maniloff et al, 1994). This is consistent with the observation of clear plaques formed as a result of infection, indicating lytic activity.
2.3.6. Phage typing of VTEC and non-VTEC isolates
E. coli isolates of various serotypes and sources were tested for sensitivity to environmentally isolated phage AT20. Eighteen isolates were previously detemiined to produce ver0 cytotoxin (VT), while the remaining 12 were confirmed as non-toxigenic by both conventional and PCR-based methods (Personal communication, S. Chen). Figure 2.2. Transmission electron micrograph of Snltnonelln phage Felix 0-1
75 Figure 2.3. Transmission electrun microgrnph of Salmonella phage SJ2
79 Results are shown in Tables 2.5 and 2.6.
Al1 of the VTEC isolates were identifieci as serovar 0157:H7 and al1 of hem, regardless of source, demonstrated resistance to phage AT20. In contrast, al1 non-VTEC strains of various serovars, including bovine isolates of 0157:H7, were sensitive to phage infection.
These conflicting resuits can probably be best explained by addressing toxin production as opposed to considering isolate serovars or source.
The ability of E. coli to produce ver0 cytotoxin (also known as shiga-like toxin; SLT) is known to be determined by temperate bacteriophages (Holmes & Schmitt, 1994). As previously described, harbouring lysogenic phages gives host bacteria a cornpetitive edge, including immunity to superinfection, by repression of RNA polymerase activity. The apparent resistance of a11 toxin-producing isolates to phage AT20 may be caused by such prophages, rather than being attributed to a physiological feature of the phage itself.
Indeed, it is likely, based on the lack of serovar specificity, that AT20 is a broad-spectrum coliphage capable of infecting large cross-sections of E. coli populations. The use of induction, to expel prophages, and subsequent infection following exposure ro phage
AT20. would confirm both: (i) the non-specific nature of the phage, and (ii) the role of prophage-mediated irnrnunity to infection, as opposed to resistant. Loss of the toxin phenotype would Merimplicate the role of toxin-encoding phages. However, if subsequent phage infection was not observed, then other possible mechanisms would have to be considered such as inhibition of replication by plasmids (Duckworth, t 987). Table 2.5. Typing of vero cytotoxigenic E. coli isolates by phage AT20
E. coli isolate Source Serovar Sensitivity to phage AT20 EC 920005 Bovine 0157:H7 No 11 EC 920026 1 Bovine 1 0157:H7 1 No II EC 920027 1 Bovine 1 0157:H7 1 No II EC 920037 1 Bovine 1 0157:H7 1 No EC 92008 1 Bovine 0157:H7 No EC 920 1 92 Bovine 0157:H7 No EC 920321 N.K. 0157:H7 No 1 - r EC 950050 Bovine 0157:H7 No EC 960275 Bovine 0157:H7 No EC 920267 Bovine 0157:H7 No
-- - - II EC 940468 1 Human 1 0157:~7 1 No Table 2.6. Typhg of non-vero cytotoxigenic E. coli isolates by phage AT20
E. di 1 Source 1 Serovar 1 Sensiùvity to phage AT2C isolate I I EC 960264 1 Bovine 1 01 57:H7 1 Yes
EC 920048 Bovine O 1 56:H7 Yes EC 950334 Lamb 0128:H2 Yes EC 960001 Lamb 091 :Hl0 Yes EC 95021 1 Feline 04:HS Yes EC 960266 Bovine 01 57:H7 Yes EC 940287 1 Bovine 1 01 12:H2 1 Yes EC 920232 1 Bovine 1 02:H5 1 Yes - - EC 960282 Bovine 0157:H25 Yes EC 960265 Bovine 01 57:H7 Yes EC 920229 Bovine 02:H6 Yes 2.3.7. Microscopic characterisation of E. coli phage AT20
Transmission electron microscopy of phage AT20 (Figure 2.4) reveaied a viral
architecture similar to that of the lytic myovirus T4, with lytic activity confinned by the
observation of large clear plaques in viral assay. An elongated head can be clearilly
distinguished, as well as the baseplate and tail spikes. Tai1 fibres are not visible, and
were probably sheared off during the negative staining procedure. This phage differs
substantially from isolated prototype phages, capable of converthg bacterial hosts to a
toxin-producing phenotype (Holmes & Schmitt, 1994). Toxin structural variants, VT-1
and VT-II, were found to be encoded by genes located on different bacteriophages. VT-1
determinants were identified on phages H 19A and H 19B, isolated from a 026:Hll E. coli
strain. These temperate phages are characterised by a hexagonal head and long, non-
contractile tail. Genes for VT-II production were found in phage 933W isolated fiom a
0 157:H7 strain of E. coli, morphologically distinct fiom the Hl 9 phages by a very short,
non-contractiie tail (Hohes & Schmitt, 1994).
If host irnrnunity is indeed responsible for the insensitivity of VTEC isolates to lytic
phage infection, then this research indicates that acquired immunity can prevent
superinfection by heterologous phages, as descnbed previously by Duckworth. Induction of prophage and examination of phage properties would be required to confirm this phenornenon, described as mutual exclusion between unrelated phages (Duckworth,
1987). Figure 2.4. Transmission eiecbon micrograph of E. coli phage AT20
84 2.3.8. Applicability of phages in detection methods
The potentid usefulness of phages in methods for the identificationldetection of bacteria
depends on both the level of specificity required of the test, and the availability of suitable
phages. There may be situations where only a generic assay is required for screening
purposes, while a more specific assay wodd focus on, for exarnple, the pathogenic
members of the genus.
Based on the information obtained fiom characterisation studies, a number of
observations can be made regarding the applicability of each of the tested phages. Both
the Listeria ATCC 2307443 1 and Salmonella Felix 0-1 phages demonstrated broad-
spectnim activity against many different species of their respective genera. Indeed, Felix
0- 1 has already been applied to various Salmonella-specific assays with promising results
(Chapter One). Although it is not capable of identieing 100% of Salmonella isolates,
improvements have been demonstrated if used in combination with other Salmonella
phages (Pugh & Arnott, 1989). Altematively, the species-specific nature of the
environmental SJ2 phage could facilitate its use either alone or in tandem with Felix 0- 1
to confirm the presencelabsence of S. Typhimuriurn or S. Ententidis.
Like Felix 0-1, ATCC 23074-8 1 would be an appropriate phage to use in combination
with other phages for detection of Listeria spp. Furthemore, with its broadest specificity demonstrated against L. innocua and L. welshimeri, it could be used as a rneans of detecting indicator species such as L. irznocuo instead of oniy testing for L. rnonocytogenes strains. Further screening is also required to establish whether non- defective phage B 1 can infect isolates closely related to the Listeria genus such as
Bacillus.
Finaily. phage AT20 was not able to detect ver0 cytotoxin-producing strains of E coli
0 157:H7. However, based on this research it is proposed that immunity to infection, conveyed by lysogenic phages encoding toxin genes, could be a usehl means of screening strains for VT production. Further screening of E. coli strains is necessary to establish whether phage AT20 possesses the ability to infect a broader range of isolates. SPECIFIC BACTERIAL DETECTION BY PHAGE-MEDIATED ATP
BIOLUMINESCENCE
First described in the late 19403, the nahirally occuning ATP bioluminescence reaction of the North American firefly Photinus pyraks has since been developed into a sensitive and rapid means of accurately determining levels of rnicrobial and somatic ATP. While such measurements are useN for estimating microbial biomass in foods and overall cleanliness in food processing facilities, the non-selective nature of ATP extraction means that they cannot offer specific information regarding the sources of ATP being evaluated.
Currently, with the commercial ATP bioluminescence assay unable to diflerentiate bacteria its use as a detection method for specific pathogens is impossible.
This gap in the application of ATP bioluminescence has been addressed recently by the use of bacteriophages (Sanders, 1995; Blasco et al., 1998). Exploiting the specificity of these bacterial viruses, physical and molecular techniques have been employed in the development of detection methods for important food-borne pathogens such as Listeria monocytogenes (Sanders, 1995; Loessner et al., 1995). Both methods rely on the lytic nature of selected bacteriophages as a means of extracting ATP fkom target bacterial cells. The first method uses lytic phages themselves as biological extractants (Figure 3.1 ), while the second has relied on the identification and cloning of genes encodhg endolysins, enzymes responsible for bacteriai ce11 lysis and release of progeny phage.
While both approachrs have been show to be potentially successful ways of detecting specific bacteria, their wide-spread application will depend on both the availability of suitable bactenophages and endolysins for the detection of desired target bactena.
Although the use of lytic phages is the least sensitive of the two proposed methods, it is a much less expensive, simple and time-efficient proposition to the expensive and labour- intensive endeavour of molecular gene cloning.
The use of bacteriophages, characterised in Chapter 2, as biological ATP extractants was therefore investigated. This chapter will focus on their application in the development of specific ATP bioluminescence assays for the detection of Listeria monocytogenes,
Escherichia coli and Salmonella Ententidis.
3.2. MATEMALS AND METHODS
3.2.1. The phage lysis assay
The lytic activity of bacteriophages charactensed in Chapter 2 was studied to determine Figure 3.1. The phage-mediated ATP Bioluminescence Assay
89 their usefulness as specific bacterial ATP extractants in the bioluminescence reaction descnbed above. Pure cultures of bacteria were grown in Trypticase Soy broth (TSB)at
37°C for 18 hours. Nine hundred pL volumes of ce11 suspensions were then pipetted into paired sterile capped tubes, each pair of tubes corresponding to progressive 30 minute
incubation periods, up to two hours total assay tirne. Tubes were prepared in triplicate for each of two studies. At tirne zero, test and control tubes were set up by addition of either
100 pL of the appropriate undiluted bacteriophage or 100pL of propagation buffer
(prepared in accordance with section 2.1.1. but without addition of phage) and assayed using the protocol outlined below. Following assay of time zero tubes, the remaining tubes were set up in the presencektbsence of phage as before, and placed into a 37OC incubator for timed incubation periods.
Tube contents were assayed for ATP by vortexing, followed by removd of 100 pL into a stede ATP- free cuvette. One hundred PL of prepared luc i fenn-luci ferase reagent was added to each tube. Following bnef gentle shaking, tubes were then placed into the charnber of a bench-top luminometer pre-programmed to perform readings integrated over ten seconds. The mean relative light unit (RLU) output for the six replicates was recorded, and covariance values for each tnplicate reading were available fiom the luminometer print-out of the data..
The ATP monitoring reagent MR 5000, included in the High Sensitivity Surface
Monitoring Kit (Catalogue number 1320-0301; GEM Biomedical, Inc., Hamden, CT), was employed for bioluminescent assay using the Optocomp I luminometer (both manufactured by MGM Instruments, Inc., Hamden, CT).
3.2.2. Establishing the detection sensitivity of the assay
To establish the limit of sensitivity of this assay, seriai dilutions of bacterial suspensions were prepared in stenle TSB, and assayed according to the procedure described in section
3 -2.1 . above. Duplicate 100 pL volumes of each dilution were surface plated onto TSA plates and incubated for 18-24 hours at 37°C to establish ce11 concentrations. The mean
RLU reading obtained for each dilution &er 60 minutes was then compared with ce11 concentration to detennine assay sensitivity.
3.2.3. Detection of target bacteria in mixed cultures
Based on the detection limits established in section 3.2.2., combinations of target and non-target bacteria were set up to compare the selectivity of phages, and the potential influence of non-target bacteria on the lytic assay (see Table 3.1 .). Target ce11 concentrations of 10' CFU/mL were mixed 1: 1 with 104 CFU/mL of non-target bactena.
Phages for both target and non-target cells were then added individually as before and bioluminescence was measured in triplicate at time zero and after 60 minutes of incubation at 37°C. Table 3.1. Combinations of bacteria and phages used to examine the effect of non- target ceUs on phage-rnediated detection
Target organism Contaminant P hage/Control L. monocytogenes S. Ententidis B1 L. monocytogenes S. Ententidis SJ2 ------L. monocytogenes S. Enteritidis ControI II L. monocytogenes 1 E. coli I BI
L. monocytogenes E. coli Control E. coli L. monocytogenes AT20 E. coli L. monocytogenes B1 E. coli L. monocytogenes Control II E. coli 1 S. Enteritidis 1 AT20 II E. coli 1 S. Enteritidis 1 SJ2 II E. coli 1 S. Enteritidis 1 Control II S. Enteritidis 1 L. monocytogenes 1 SJ2 II S. Enteritidis 1 L. rnonocytogenes 1 B 1 II S. Enteritidis 1 L. monocytogenes 1 Control II S. Enteritidis 1 E. coli 1 SJ2 II S. Enteritidis 1 E. coli 1 AT20 S. Enteritidis 1 E. coii Control 3.2.4. Increasing assay time
Increasing the assay tirne for detection of E. coli G2-2 was investigated. An overnight culture was diluted to a level of 10' CFUIrnL (well below the detection limit) and the lytic phage assay was set up employing phage AT20. Triplicate measurements of ATP bioluminescence for controls and tests was recorded over an extended period of 3.5 hours, and bioluminescence was plotted versus phage exposure time.
3.2.5. Target ce11 concentration by filtration
The use of filtration to improve the sensitivity of the phage lysis assay for Listeria monocytogenes was evaluated using 13mm filters and holders included in the Biotrace
Raw Milk Quality Kit (see section 3 -2.1. for kit details). Nine hundred uL volumes of culture dilutions containing 10' and 1O6 CFUIrnL were filtered and washed according to manufacturer's instructions. Filters were then aseptically removed fiom filter units and placed in stenle tubes. One hundred pL of phage suspension was added directly to the filter, and gently vortexed to aid the movernent of the phage within the filter matrix. At time zero 900 pL of TSB was added to filters, followed by vigorous vortexing, and bioluminescence was compared to 900 pL of unfiltered ce11 suspension. To incubated filters, 100 pL of TSB was added to keep the filter moist, and prior to time 60 minute readings, 800 pL of TSB was added. Bioluminescence for filtered and unfiltered samples in the absence and presence of phage was measured (in triplicate) for the Listeria phage ATCC 23074-B 1 and L. monocyiogenes ATCC 23074 host combination.
3.2.6. ATP recycling
ATP recycling was perfomed according to the method of Hawronskyj et al. (1994) based on the following reactions:
ATP + luciferin + O2+ AMP + oxyluciferin + PPi + CO+ light (catalysed by luciferase)
ATP + AMP - 2ADP (catalysed by myokinase) 7ADP + 2PEP - 2ATP + pyruvate (catalysed by pyruvate kinase)
Similar concentrations of enzymes and substrates were employed (Sigma-Aldrich Canada
Ltd., Missisauga, ON). These included: 0.1 1 mM phosphoenolpyruvate (PEP). 0.1 1 mM adenosine monophosphate (AMP), 0.45 unitdml pymvate kinase (PK), and 0.75 units/mL myokinase (MK). It should be noted that the initial lower MK concentration was later lowered to 0.0075 units/mL to reduce bioluminescence output to a more suitable level. The reaction mixture contained 100 pL each of PEP, AMP, MK, and Lysed bacterial suspension. Five hundred pL of luciferin-luciferase was added to the tube with
100 pL of PK added last to start the reaction. Al1 bioluminescence readings were measured in duplicate, using either the kinetics program of the GEM Optocomp 1 luminometer, or single measurements. Bacterial ATP was extracted using both the phage lysis method previously described and the ATP releasing agent from the High Sensitivity Surface Monitoring Kit (as per instructions) before addition of recycling reagents.
3.3. RESULTS AND DISCUSSION
3.3.1. Practical considerations of the phage lysis assay
The use of lytic bacteriophages for selective release of bacterial ATP, as proposed by
Sanders (1 995) was investigated initially as a potential detection system for Listeria monocytogenes, and later applied to strains of E. coli and S. Enteritidis. Using the bacteria-phage combination originally employed (L. monocytogenes ATCC 23074 and phage 23074-B 1). phage-rnediated ATP release was determined by bioluminescent measurement, with the mean relative light unit (RLU) output plotted against time of exposure to phage. General purpose Trypticase Soy broth was used as a phage lysis medium as opposed to buffer, as intracellular replication of bacteriophages requires an actively growing host (Wiggins & Alexander, 1985).
Initial bacteriophage titres for L. monocyîogenes ATCC 23074-B 1 were approximately
1O9 PFU/mL. but were improved to -10" PFU/mL following propagation outlined in
Chapter Two. Al1 phages, including AT20 and SJ2, were used at levels consistently around 10" PFU/mL. 3.3.2. Phage-mediated bioluminescent detection of L. monocytogenes
The interaction of phage ATCC 23074-BI with L. monocytogenes ATCC 23074 was subsequently investigated and the mean bioluminescence profile is demonsûated in
Figure 3.2. The mean bioluminescence is the resuit of six readings, with calculated coefficient of variation (CV) values ranging fiom 0.16-3.8%. Analysis of CV values for al1 luminometer data demonstrated that values were always below 5%, and in most cases were better than 3%. Variation between triplkate readings, and duplication of experiments was low. The mean bioluminescence of six data points was thus used for every sample in this study.
Comparing this resuit with published data using the same phage-host combination. the
ATP peak observed in this experiment occurred approximately 50 minutes before the peak observed by Sanders. However, because a number of samples were being manually tested in this case (as opposed to the automated testing protocol of Sanders), a longer interval between tests was necessary and thus an 80 minute sampling time was not evaluated. Similarly, the exact onset of ATP release fiom cells could not be ascertained from Figure 3.2, as incubation times between O and 30 minutes were not tested, but it did occur at least 30 minutes before the onset of ATP release reported by Sanders. It is likely that ATP levels did not begin to increase until at least 20 minutes of phage exposure had passed, which is the minimum length of tirne required for the phage lytic cycle to occur, and is dependent on the phage-host combination used. It is also feasible that the amount 1 2 Time (hours)
Figure 3.2. Release of ATP from L. monocytogener ATCC 23074 by phage BI- mediated lysis of ATP released by bacteriophage did indeed reach a maximum level later than Figure
3 -2. indicates. Regardless of these issues, what was important here was the ability to differentiate samples infected with bactenophage fkom those not, and this was demonstrated to be achievable within 30 minutes using this assay.
3-33. Deteetion limits of the Lûteria phage assay
To establish the detection limits of this phage-based bioluminescence assay, initial bacterial concentrations fiom 1 x 10' to 1 x 10' CFU/rnL were tested- ATP levels of controis and tests assayed after 60 minutes of exposure to phage were then compmed at each ce11 concentration (Figure 3.3). There was little difference between controls and tests at levels from 1 x 10" to 1 x 106, while a slight increase in measured ATP was observed at a level of 106 CFU/mL. However, si~nilarincreases were demonstrated by both uninfected and infected samples, therefore indicating either a fluctuation in the testing procedure or non-phage related release of ATP. A significant increase in bioluminescence was finally observed in the presence of phage at a level of 10' CFU/mL of L. monocytogenes, similar to the increased bioluminescence illustrated previously in
Figure 3.2. The detection limit for this particular phage-bacteria combination was therefore estimated to be greater than 1 x 1O6 CFU/mL within 60 minutes of exposure of bactena to phage.
A steady increase in ATP levels for uninfected ce11 suspensions was also observed over
98 1 -plus phage
Bacterial concentration (CFU/m L)
Figure 3.3. Detection limits of phage BI-mediated L. monocytogenes bioluminescence assay time. This could be due to autolysis of cells, or leakage of ATP during the replicative
cycle. It does however indicate that total ATP levels observed for infected cells are not
solely due to phage lysis, and that increases in ATP levels should always be compared to
controls.
33.4. Phage-mediated bioluminescent detection of E. coli and S. Enteritidis
Having established the phage lysis protocol for detection of L. monocytogenes, similar
conditions were then applied to evaluate the use of E. coli AT20 and Salmonella SJ2
phages for the detection of similar levels of E. coli (32-2 (Figure 3.4.) and S. Enteritidis
S 1 O respectively (Figure 3.5.).
Mean ATP release profiles were different in both cases, and differed again from that
observed for L. monocytogenes. The lytic activity of phage AT20 was not apparent until at leasr 30 minutes of exposure time had elapsed. An almost exponential increase in
measured ATP followed, continuing until the maximum assay time of 120 minutes was
reached. Although a peak level was not obtained in this case, after 60 minutes an
increase of about 40,000 relative light units was observed. with an ATP level of over
100.000 RLU reached by completion of the assay. Contrastingly, phage SJ2 had already produced a peak increase in mean bioluminescence of 150,000 RLU within 30 minutes of exposure to S. Ententidis cells. However, this level unlike the ATP peak of L. monocytogenes remained high even after over three hours of assay. Figure 3.4. Release of ATP from E. coii G2-2 by phage AT20
101 - minus phage SJ2 - plus phage SJ2
O 1 2 3 4 Time (hoursl
Figure 3.5. Release of ATP from S. Enteritidis by phage SJ2
1O2 In both cases above a ~~cientlylarge increase in bioluminescence was observed after 60 minutes, and this cut-off time was therefore retained for Merevaiuation of detection
Iimits.
These results indicated that the ATP release profiles obtained were host-phage dependent, and this has been Mersubstantiated by the work of Blasco and colleagues (Blasco et ol. 1998) who identified a similar trend when exarnining the use of lytic bacteriophages to spccifically release cellular adenylate kinase (AK). Interestingly, they observed that the E. coli phage employed was capable of releasing AK (and presumably ATP) &ce as quickly as compared to the Salmonella Felix or Newport phages. Given that different bacterial strains and phages were used in their study as opposed to this one, this adds further credence to the claim that different host-phage combinations behave differently, even within the sarne bacterial host genus.
Modes of action of these phages can be speculated at this point, taking into consideration what is already known about them fkom Chapter Two. In the case of Listeria phage B 1, it was identified as a lysogenic phage, capable by definition of infecting host cells by incorporating its DNA into the host chromosome rather than adopting the lytic cycle.
This confers imrnunity to the host, and Merinfection is not seen. This could explain why, in Figure 3.2., a single peak ATP reading was obtained which was then observed to decrease below the contra1 readings over the. It is possible that the Listeria phage initially lysed a large number of cells, followed by subsequent lysogenic behaviour promoted by the decrease in the bacterial population.
In contrast, phage lysis of E- coli G2-2 did not reach a maximum ATP peak but instead conûnued to increase the amount of measurable ATP over tirne. Phage AT20 is a lytic phage (based on information fiom Chapter Two) and it would appear kom its infection profile that it undergoes a senes of lytic cycles. This in tum produces increasing arnounts of ATP in the system as the number of phages capable of infection is increasing by lytk replication in the E. coli host. Eventually, it is assumed that this balance between bacterial and viral replication would reach a point where bacterial numbers were too low for phage replication, and ATP levels would then start to decrease. Such a phenornenon was not studied in this case, but such information would be helpful in attempting to explain this very complicated virus-host relationship.
Finally , the Salmonella phage SJ2 demonstrated yet another variation of the ATP bioluminescence profile. In this case, a peak ATP level was achieved within 30 minutes of phage exposure, and this level remained reasonably constant over a three hour penod.
This indicates initially high lytic activity by the phage to reach the ATP peak so quickly.
If infection was not occurring then ATP levels wouid be expected to decrease due to enzymatic degradation (similar to the scenario with Listeria). As this is not the case, it is proposed that a more balanced infection occurred where ATP Ievels were maintained over the course of the incubation period. 3.3.5. Detection limits for E. coli and S. Enteritidis assays
Detection limits were subsequently examined for phages AT20 and SJ2. Figures 3.6. and
3 -7. demonstrate the effect of addition of these phages on bioluminescence levels for various serial dilutions of. respectively, E. coli G2-2 and S. Enteritidis. In both cases a constant level was observed for concentrations of 10' and 1O6 CFUImL as compared to the control, but bioluminescence began to increase at a ce11 level of 1 x lo7 CFU/mL.
Figures 3.3., 3.6., and 3.7. demonstrated that the detection limit of the bioluminescence assay is the same for dl three phages, but that the relative arnounts of released ATP di ffered.
3.3.6. Effect of contaminants on target ce11 detection
The influence of non-target rnicroorganisms was evaluated by combining different levels of target and non-target bacteria, and assessing the ability of each phage to detect its bactenal host (Table 3.1 .). A target:non-target ce11 ratio of approximately 2: 1 was chosen where tne level of target cells (1 0' CFU/mL) was above the detection limit but non-target levels (1 0'CFUlmL) were below. Each combination of cells was tested with the two corresponding phages, and bioluminescence, as before, was measured in triplicate at the zero and after 60 minutes of exposure to phage. Figure 3.8. illustrates the ability of phage
B 1 to detect L. monocyfogenes in the presence of either S. Enteritidis or E. coli after 60 minutes of incubation. However, phages SJ2 and AT20 demonstrated ATP profiles -ninus phage AT20 -plus phage AT20
Bacterial concentration (CFU/m L)
Figure 3.6. Detection limits of phage AT20-mediated E. coli bioluminescence assay bacterial concentration (hours)
Figure 3.7. Detection limits of phage SJ2-mediated S. Enteritidis bioluminescence assay Figure 3.8. Effect of contaminants on phage BI-mediated bioluminescent detection of L. rnonocyfogenes dmost identical to uninfected controls and therefore were unable to both detect their
hosts (being below the detection limit of the assay) and infect non-target Listeria cells.
The bioluminescence level achieved after 60 minutes was slightly lower than that
previously dernonstrated for the Listeria detection assay (Figure 3.2.). This is a result to
be expected when a large proportion of the bacterial population includes non-target cells,
effectively hindering the ability of the phage to encounter target cells. Such an effect was
recently reported by Blasco (1 998), who dso observed slightly lower bioluminescence
signals when assaying S. Newport in the presence of approximately equal concentrations
of non-target Citrobacter freundii.
Figures 3 -9. and 3.1 0. both demonstrated the same trend as Figure 3.8.. where target cells
were detectcd by their specific phage, but non-target cells were not. Again, phages did
not cross-react with bacteria of different genera and bioluminescent signais were slightly
lower than seen in the absence of non-target cells. To put into perspective the relative abilities of the three phages to detect their host cells, results for each phase and bacteriai combination were graphically compared as shown in Figure 3.1 1. The highest bioluminescence output afler 60 minutes was achieved by the Listeria phage ATCC
23074-B 1, followed by phages AT20 and SJ2. Apart fiom the already discussed reduction in bioluminescence associated with the presence of contarninants. none of them played a role in delaying or preventing the detection of one target pathogen over another, and Blasco's group demonstrated that a similar target ce11 phage assay was relatively unaf3ected by the inclusion of 10' non-target cells/mL. 80 Erposure to phage (minutes)
Figure 3.9. Effect of contaminants on phage ATZO-mediated bioluminescent detection of E. coli G2-2 Figure 3.10. Effect of contaminants on phage SJ2-mediated bioluminescent detection of S. Enteritidis SI0 Figure 3.1 1. Relative release of ATP by phages BI, AT20 and SJ2 3.3.7. Improving assay sensitivity
This phage assay has been demonstrated to be a simple, quick and inexpensive means of making the ATP bioluminescence assay more specific. However, the detection limit for various pathogens was determined to be greater than 1 x 1O6 CFU/mL, about 100-fold higher than the lower sensitivity limit of ATP bioluminescence itself. To address this issue a few alternative approaches were investigated, including (i) increased assay tirne,
(ii) the incorporation of filtration as reported by Phillips & Grïffiths (1989), Griffith et al. (199 l), and more recently by Tanaka's research group (1997), and (iii) an investigation of the ATP recycling reaction proposed by Hawronskyj and colleagues
(1 995)- The adenylate kinase assay has been reported to improve the sensitivity of the lytic phage bioluminescence assay by a factor of 10-100. Based on the results published by Blasco et al. (1998) it does appear to be a promising method applicable to any bactena-phage combination, and will likely receive more attention in the near future.
However, this methcd was not explored in this Chapter for two main reasons: First, reagents for the AK assay need to be exuemely pure, and it was found that ADP was a particular problem. resulting in an overload RLU luminometer reading. Purification is in itself an involved process, and was felt to be beyond the scope of this research. Second,
Blasco and colleagues (1998) also employed a proprietary extractant, to release vesicle- bound AK. By not incorporating this step, the sensitivity of the assay would be immediately decreased, negating the usehilness of the assay in the first place. 3.3.8. Effect of extended assay time on detection sensitivity
The idea of extending the phage assay tirne was prompted by both the inability of phage
AT20 to produce a peak ATP value (Figure 3.4.) and the published work of Wiggins and
Alexander ( 1985). They reported that a minimum bacterid density was required for bacteriophage replication. This threshold was estimated to be between 7 x 10' and 3 x
1 CFU/mL for three different host-phage combinations, including E. coli.
Stuphylococcus aureus, and Bacillus subtiiis. These threshold values are already
imposing on the limitations of the ATP bioluminescence assay, and given that the modified phage assay is only an hour in duration, and it is unlikely that bactenophage replication could occur at lower bacterial concentrations during this period suficien? for lysis and subsequent detection by ATP bioluminescence.
The time of incubation was therefore extended to 3.5 hours, and the concentration of E. coli (32-2 was reduced to I x 10' CFU/mL, below the detection limit previously established. Bioluminescence was evaluated for infected and non-infected ce11 suspensions every 30 minutes aqd results are show in Figure 3.12. Two trends are apparent fiom the ATP profile obtained: first, the levels of detectable ATP were about
100 times lower than those determined in Figure 3.4., in keeping with the 100-fold decrease in bacterial population employed in this experirnent. Second, as previously demonstrated, in the presence of phage the amount of detected ATP increased progressively over time as compared to the uninfected ce11 suspension. This would -- O 30 60 90 1ZQ 150 180 21O Exporure to phage AM0 (minutes)
Figure 3.12. Effect of extended incubation time on release of ATP from E. coli G2- by phage AT20 indicate that the lyçis of target cells continues over an extended period of time. likely correlating with a nurnber of lytic cycles subsequentiy increasing the amount of phage each time. It thus appears that extended assay time could be a useful means of increasing sensitivity, but this would have to be assessed for different phage-host combinations. For example, it codG be postdated that phages exhibiting extremely efficient target ce11 lysis would not be useful in such a case. It is probable that they would initially deplete their host population to a level where further viral replication would subsequently require significant replication of host cells to exceed the minimum density requirement for propagation (Wiggins & Alexander, 1985).
3.3.9. Filtration as a means of concentrating target cells
The concentration of bacterid cells by means of filtration has been a useful addition to the ATP bioluminescence asay and has been effectively incorporated into both liquid and solid media systems (Griffiths et ai., 1991; Tanaka et al., 1997). Its use in the phage lysis assay as a way of concentrating host cells, followed by addition of bacteriophage was therefore evaluated. Filtered and unfiltered ce11 suspensions of L. monocytogenes
ATCC 23074 at concentrations of 1 x 10' and 1 x 106 CFU/rnL were tested in the presence and absence of phage ATCC 23074-B 1.
Bioluminescence readings were taken at zero and 60 minutes exposure to phage and the resulting differences are shown in Figure 3.13. At an initial concentration of 1 x log Figure 3.13. Effect of filtration on phage BI-mediated bioluminescent detectioo of L. monocytogenes CFUIrnL, bioluminescence results for samples not exposed to phage were sirnilar regardless of the incorporation of a filtration step. This is to be expected when an extraction step is not being used and therefore bacterial concentration is not a contributing factor to the bioluminescence output. It was thought however that the included rinsing step would produce a lower bioluminescence result in the filtered samples, due to the rernoval of background ATP, and this was indeed the case at a levei of 1 x 10' CFU/rnL where unflltered samples had an initial ATP count about 100,000
RLU higher than filtered samples (data not shown). However. the dilution of cultures effèctively reduced ATP present and so no difierences were observed between filtered and unfiltered samples in the absence of phage at a ce11 concentration of 1 x 106CFUhL.
Regardless, levels of background ATP were negated by subtraction of time zero values. but the usehlness of nnsing was apparent.
Filtration did dernonstrate an effect when bactenophage was included as an ATP extractant. After 60 minutes of exposure to bactenophage ATCC 23074-B 1, an increase of approximately 270,000 RLU was observed in unfiltered samples at the higher bacterial concentration as compared to control readings. Filtration reduced this bioluminescence by almost 70% to an ATP extraction level of approximately 82,000 RLU. Similady, at the lower bactenal concentration examined. filtration lowered the subsequent bioluminescsnce of infected ce11 suspensions by approximately 30% to a level comparable with the filtered control. This was the reverse of the result hoped for, and can be explained by the inability of the bacteriophage to gain access to bacterial attachment sites and subsequently infect the cells trapped in the filter matrix. This is not an issue with chemical extractants where lysis is rapid, and allows more thorough penetration of the extractant. DifTerences in the bioluminescence reduction rates observed after prior filtration are likely due to the entrapment of a much larger number of cells at the higher bacterial concentration, which could not be accessed by the phage suspension. At lower concentrations, lower ce11 congestion and thus better phage accessibility was evident, resulting in a lesser reduction in bioluminescence as compared to unfiltered ce11 suspensions. Options to improve the filtration assay could be increased exposure theto phage (dependent on the phage-host combination). alternative filter matrices and an increased filter surface area. Ultimately however it would appear that concentration of bacteria by filtration for phage-mediated lysis is not an effective means of increasing the sensitivity of this particular assay. The lirniting factor is rather the lytic eaciency related prirnarily to the particular host-phage combination.
3.3.10. Improving assay sensitivity by ATP amplification
The use of ATP amplification, or "recycling", to improve the sensitivity of the ATP bioluminescence assay was first reported by HawronsLyj and colleagues ( 1995). and was later proposed as a useful means of detecting low levels of bacterial ATP (Hawronskyj &
Holah. 1997). The amplification system is composed of the following enzymes and substrates: luciferin, luciferase, phosphoenolpynwate (PEP), adenosine monophosphate
(AMP), myokinase (MK) and pyruvate kinase (PK). It has been reported to increase the sensitivity of ATP detection fiom 550nM down to 250pM (Hawronskyj et al., 1995).
This method was therefore examined as a potentid means of increasing sensitivity of the phage lytic assay.
The amplification reagents employed in this research were fmt evaluated to ensure that a linear relationship existed between ATP concentration and the time interval required to reach half peak height (t%). Analysis of kinetic data revealed a sirnilar result to that observed by Hawronskyj (1995), where increasing ATP concentrations produced a corresponding decrease in t%. while peak height remained the same (data not shown).
This confirmed the suitability of the recycling reaction concentrations for ATP amplification.
These reagents were then used to ampli@ ATP released fiom different concentrations of
L. monocytogenes ATCC 33074 following a 30 minute exposure to phage B 1.
Bioluminescence was monitored using the kinetic mode of the luminometer and the resulting data were plotted versus time (Figure 3.14). Bioluminescence outputs were extremely high for al1 samples, at over 20 million RLU, and al1 bioluminescence profiles e,xhibited sirnilar peak heights afier about five minutes of analysis. If we compare these levels with those achieved by the original phage lysis assay (Figure 3.2.) a 100-fold increase in light output was apparent. However, even at high target ce11 concentrations Figure 3.14. ATP amplification of the phage B1-mediated L. rnunocytogenes assay
121 previously detected by the original phage lysis assay, difierences in t% were not observed between samples exposed to phage and those not. This implied that the amount of ATP being released by phage lysis was not suficient to influence the amplification of background ATP. Indeed, significant amplification of background ATP, present as reagent impurities, was demonstrated by Hawronskyj (1995) in a system where no additional ATP had been added to theoretically start the reaction. To investigate this mer,phage lysis was subsequently replaced by chemical lysis using ATP releasing agent and concentrations of enzymes were adjusted in an attempt to lower the light output to a more suitable level.
Reducing the pyruvate kinase concentration had previously been reported by Hawronskyj
( 1994) to effectively reduce the maximum light output of the bioluminescence reaction.
Pyruvate kinase converts ADP into ATP, thus less enzyme would obviously convert less
ADP and bioluminescence would be lower. Similady, myokinase catalyses the reversible catabolism of ADP to ATP and AMP, and lowenng the concentration of the enzyme should theoreticaily resul t in a decrease in ATP available for the bioluminescence reaction.
The effect of varying concentrations of enzymes and substrates was evaluated by first extracting ATP fiom a 1 O8 CFU/mL suspension of L. rnonocyrogenes using chemical lysis as opposed to lytic phage. Table 3.2. shows the maximum bioluminescence achieved using both original and reduced enzyme and substrate concentrations. Using original Table 3.2. Effect of altering amplineation cornponents on bioluminescent detection of 10' CFUIml Lisleria monocyiogenes ATCC 23074
11 Components 1 Alterations 1 Max. Bioluminescence
1 PEP. AMP,MK. PK 1/10 pK, pK] 2.7 r~illion 1 PEP, AMP, MK, pK l/iOO Fm, pK] 2.8 million
II AMP 1 NoMK, PK,PEP 1 4 -4 thousand 1 AMP, 111 O IpEP] No MK, PK 3.8 mitlion AMP, 1/100 PEP] No MK, PK 56 thousand PEP: Phosphoenolpyruvate AMP: Adenosine monophosphate MK: Myokinase PK: Pyruvate kinase concentrations of recycling components, a peak bioluminescence output of 2.7 million
RLU was obtained. However, serially diluting the concentrations of MK and PK did not change this peak and indeed, complete removal of both enzymes had no effect.
Subsequent removal of AMP also had no effect, but complete removal of PEP resulted in a 1000-fold decrease in maximum bioluminescence. Subsequent addition of 1O-' diluted
PEP returned the bioluminescence peak back at 3.8 million RLU, while a further serial dilution lowered it to 56,000. It wouid thus appear fiom these results that the only criticai component of the original amplification reaction mixture is phosphoenolpyruvate, involved in the conversion of ADP to ATP by pyruvate kinase. Indeed, a similar phenornenon was observed for AMP, where at lower concentrations peak heigh was significantly affected (Hawronskyj et al., 1994), demonstrating the requirement of AMP for the myokinase reaction. Removing it from this reaction mixture had no eRect on peak height, probably because enough AMP was released fiom bacterial cells to drive the myokinase reaction. The fact that additional enzymes required for ATP production did not have any effect on the reaction would aiso imply that these were aiready present in the system as a result of bactena ce11 lysis. It is probable that, with only the addition of PEP, these enzymes were able to ampli@ ATP to give a maximum level 1000-fold higher than that achieved by the normal bioluminescence reaction.
Based on these results, it would appear that while this system is usehl for amplifiing pure standards of ATP, it has limited application in combination for improving the phage- based lytic assay. The use of PEP alone however appeared quite promising, and further research in this area would be usehl to establish its role in amplification further.
In conclusion, this research has demonstrated the use of bacterial phages to develop a
more specific ATP bioluminescence assay, and it has been possible, based on
observations in this research, to speculate Merregarding the potential infection
mechanisms of different phage-host combinations. However, sensitivity has been lost as
a result of increasing the specificity of this assay, and a number of options should
therefore be considered in future research. If this application is to be transferred to a food
system, a pre-enrichment step will have to be incorporated. This would enable target cells to reach detectable levels, but also allow for transfer of the food into the broth system required for the phage assay. A coarse filtration step or centrifugation would eliminate food debris, and a further filtration step could be developed to clean up the sample further. The infective nature of potential phages should be investigated further as they ultirnately dictate the efficiency of the lytic assay. The adenylate kinase assay appears to be the only mechanism proven to increase the sensitivity of such phage-based assays, but it is currently limited in its application by the need for pure reagents and proprietary extractants. The use of the ATP recycling method was demonstrated, as a protocol. to be ineffective as there appeared to be interference fiom bacterial eniymes present as a result of chernical lysis. However, the addition of PEP was demonstrated to increase bioluminescence readings alone, and may be usefid in future work. CHAPTER FOUR
THE USE OF BACTERIOPHAGES TO WROVE THE SPECIFICITY OF
TURBIDIMETRIC DETECTION METHODS
4.1. INTRODUCTION
The introduction of automated technologies for the measurement of turbidity has resulted
in their increased application as an alternative means of evaluating the microbiological quality of foods. However, as with ATP bioluminescence. the technique has been used as an estirnate of total bacterial Ioad in foods rather than for detection of specific microorganisms. With bacteriai detection relying on metabolic activity, development of pathogen-based assays using turbidimetry would have to focus on the use of differentid and seicciivc iiiedin as a means of improving their specificity. This in itself suffers from a nurnber of limitations, as described by Blood and Curtis (1 995) in their evaiuation of media for Enterobactenaceae:
"An ideal medium for use in food rnicrobiology, whether broth or agar, would selectively grow only the target organisrn or group of organisms. With a group of organisms so closely reiated as the Enterobacteriaceae it is unlikely that such a medium could be developed for a single genus such as Escherichia or even the less well defined coliform group and recourse must be had to the incorporation of substrates to provide the means of differentiating the target from the unwanted strains."
However. such recourse could be had by the use of bacteriophages. As has already been discussed at length in this thesis. bacteriophages have an effective and imate ability to identiQ and infect their bacterial host, leaving non-target bacteria unaffected. The changes to the replicative statu of a virally-infected ce11 offer an ideal means of differentiating bacteria based on metabolic activity.
Turbidimetric detection is an ideal metabolic detection method for investigating such an interaction. Detection is based on measurable turbidity changes associated with increasing numbers of bactena growing in a medium. The incorporation of phages into such a detection system would produce a "reversed rapid method" for confirmation of bacterial identity, where their ability to infect and reduce host numbers results in delayed detection. By testing the same sample, in the absence and presence of phage, detection cm be confirmed since any differences observed are a result of phage activity. If test and control detection times are identical, then either the target species was not present or was not infected by the bacteriophage employed.
With problems associated with the use of highly selective media, particularly an inability to detect metabolically injured bacterial cells, there appears to be a trend towards the use of simplified media, with inhibitor levels either reduced or eliminated (Curtis & Lee,
1995). Effectively a compromise, by reducing both selective pressure and thus the time required for enrichment, they are as a result more prone to false-negative results.
Incorporating phages into such a detection system may be a means of overcoming this issue, by balancing decreased selectivity with phage specificity. Bacteriophages charactensed in Chapter Two were therefore evduated in their ability to detect their propagating host using turbidimeûic measurements. Their subsequent application in microbiological analysis of food samples was investigated in conjunction with selective media requirements for specific bacterial detection.
4.2. MATERIALS AND METHODS
1.2.1. Instrument set-up
Al1 turbidimetric data were generated using the Bioscreen C Microbiology plate reader
(Labsystems, Helsinki, Finland). Experimental set-up involved inputing sample information, followed by plate incubation and reading parameters. The following parameters were used for al1 expenments: Single, wide band wavelength; 37°C inciibation temperature; 10 minute preheating time; kinetic measurement; measurement time of 24-48 hours; 20 minute interval between readings; low intensity shakiiig for 5 seconds before measurements.
Afier parameters were set, stede disposable 100-well honeycomb plates (Fisher
Scientific, Mississauga, ON) were aseptically filled in accordance with the sarnple information previously inputed. The reaction mixture in ail cases, uniess specified, consisted of 280 pL sterile medium, 35 pL of sarnple, and 35 PL of phage or control (see section 2-23. for preparation) to give a total of 350 PL. Although this is 50 pL below maximum well capacity, a slightly reduced volume ensures that contents do not spill over into surrounding wells as a result of shaking, thereby increasing the risk of contamination). In al1 cases, media controls were added to plates at the end of sarnple inoculation to confrm that sterility had been maintained throughout the preparatory steps.
.411 samples were tested in duplicate, allowing a maximum of 100 samples (including controls) to be assayed at once, and experiments were duplicated. Generated optical density data were analysed using the Bioscreen C data processing software version 5.26
(Labsystems. Helsinki, Finland) and expressed as the mean value of four readings..
1.2.2. Establishing the detection limits of the assay in pure culture
Listeria monocyfogenes ATCC 23074 was grown in sterile Trypticase Soy broth (TSB;
Difco Laboratories, Detroit, MI) and incubated at 37°C for 18 hours. Following input of
Bioscreen sample information and expenmental parameters, sterile honeycomb plates were filled with 3 15 PL of TSB. Ten-fold serial dilutions of cultures were prepared in sterile 0.1 % Peptone water and 35 pL voiurnes were inoculated into duplicate wells.
Dilutions were also plated ont0 Trypticase Soy agar (TSA; Difco) to confirm bacterial concentrations ernployed.
Irnmediately following completion of inoculation, plates were placed into the Bioscreen chamber and optical density measurements were made according to the protocol described in section 4.2.1. Generated turbidity information was analysed using a number of parameten available in the Bioscreen software package. Studied parameters included
initial (background) and maximum turbidity, time to reach maximum turbidity (hours),
detection time (time in hours required for an optical density change of 0.2 nits to occur),
slope of the logarithmic phase (manually estimated), tuming point time (tirne in hours required for bacterial growth to reach the middle of the log phase) and the corresponding optical density at the niming point. Correlations between these parameters and plate counts were calculated by regression analysis to assess their usehlness as predictors of initial bacterial levels.
To examine the reproducibility of individual detection times, 1: 10 and 1 :5 dilutions of 18 hour cultures of L. monocytogenes ATCC 23074 were prepared and tested as described previously. Detection times were plotted versus bacterial concentration and regression analysis was conducted to establish reproducibility.
4.2.3. Effect of phage on bacterial gowth
Bacterial growth in the presence of phage was examined for L. rnonocytogenes ATCC
23074 in the presence of Listeria phage B 1. To determine effect, optical density data were compared in both the presence and absence of phage. Phage c-oncentration versus bacterial concentration was also exarnined. Tested phage concentrations ranged fiom an initiai level.of -1 0' ' down to -1 0' PFUImL, with 10-fold serial dilutions prepared in stede A-Ca'- buffer. Bacteria were tested at initial levels of 109 to 10 CFUImL, with dilutions prepared as before (section 4.2.1 -2.). E. coli phage An0and Salmonella phage
Felix 0- 1 were tested using undiluted phage (1 0" PFU/mL) to evaluate their usefuiness in turbidimetric detection assays for E. coli G2-2 and S. Typhimwium ATCC 14028.
Correlation coefficients were cdculated for samples demonstrating positive turbidimetic detection.
42.4. EEect of contaminants on phage-based detection
The effect of contaminants on the ability of phages to detect target bacteria was evaluated using combinations of two and three different bacterial species, includuig E. coli, L. rnonocytogenes, and S. Typhimurium. For two-species combinations, bacteria at levels of
10' CFU/mL were mixed together in a 1 :1 ratio; with three bacteria a 1 :1 : 1 ratio was used. Following vigorous shaking to mix suspensions, each target ce11 mixture was exposed to either controls or corresponding phages, either individu al!^ or in combinations.
42.5. Effect of medium selectivity on detection of L. monocytogenes
Increasing the selectivity of general purpose medium for the detection of Listeria rnonocytogenes by phage BI was investigated. TSB served as the base medium and selectivity was increased by incorporation of Bacto Oxford antirnicrobic supplement
(Difco) containing acriflavin (5 mg/L); cefotetan (2 mgL); colistin sulphate (20 mg/L); cycloheximide (400 rng/L); and fosfomycin (10 mg/L.). This supplement was chosen in
preference to the modified Oxford (MC) formulation, which with high levels of
cefiazidime has on occasion failed to recover L. monocyfogenes fiom food samples due to
its very selective nature (Curtis & Lee, 1995).
Effect of incorporation of bacteriophage was monitored as before (section 4.2.1.2.) and
incubation time was increased to 48 hours to ensure that adequate tirne was provided for
detection to occur under more selective environmental conditions.
42.6. Phage-based turbidimetric detection of L. monocytogenes in milk
Pasteurised 2% milk (obtained fiom Laboratory Services Division, University of Guelph)
was inoculated with L. monocyfogenes ATCC 23074 at an initial level of 10' CFUIrnL.
Dilutions of 1 O-', 1O", and 10" were prepared in 0.1 % peptone water to take into account
the background turbidity associated with miik (recommended by Mattila & Alivehmas,
1987). Dilutions were tested in both TSB alone (TSB), and TSB plus selective
supplement (SEL), with and without phage B 1 addition (+/- B 1).
4.3. RESULTS AND DISCUSSION
The use of automated turbidimetric rnethods has, in cornmon with ATP bioluminescence, been usehl for estimating total populations of bactena in foods by analysing increases in optical density associated with increased ceIl mass (Jmgensen & Schulz, 1985; Mattiia,
1987; Mattila & Alivehmas, 1987). However, the lack of specificity of such
methodologies Iias limited its application, and the detection of particular pathogens in
foods has not been reported using this methodology. As demonstrated previously in
Chapter Three, the use of bacteriophages with innate host requirements improved the
specificity of the ATP bioluminescence reaction. These phages were therefore dso
evaluated in turbidity-based detection assays to assess their suitability for specific
bacterial detection.
4.3.1. Turbidimetric detection of L. monocytogenes: assay sensitivity
The use of optical density to estimate the population density of L. monocytogenes ATCC
23074 in pure culture was investigated. Ten-fold serial dilutions were used to test the
detection sensitivity of the Bioscreen C plate reader, and optical density data were used to
generate curves show in Figure 4.1.
Al1 samples, with the exception of one at the lowest initial concentration of 10 CFUlmL, demonstrated sirnilar metabolic activity as indicated by the characteristic and
reproducible shape of each growth curve. At very high bacteriai densities a short lag
phase was demonstrated, with optical density increasing rapidly soon derthe start of
incubation. However, as bacterial concentration decreased, the lag phage increased in duration, as the bacteria took progressively longer periods of time to reach nurnbers O 4 8 12 16 20 24 28 32 Incubation time (hours)
Figure 4.1. Optical density profiles for 10-fold serial dilutions of L. monocytogenes ATCC 23074 in Trypticase Soy broth suficient for optical density increases. Taking into account the dilution eflect produced
by the small sample size employed in this plate assay, an initial sample concentration of 1 x 1O9 CFU/mL becomes a test level of 3.5 x 10' CFU in the reaction well. This approxirnates levels of bacteria associated with logarithmic growth and significant increases in optical density (Brock. Smith & Madigan. 1984). Indeed. the use of 18 hour cultures in the exponential growth phase as inocula for Growth experiments under similar conditions will not produce a lag phase. This was also demonstrated by Jsrgensen and
Schulz ( 1985) in their automated optical density measurements of various bacteriai cultures.
While it is desirable to have fast detectable bacterid growth as demonstrated in these situations, analysis pararneters relying on the onset of the logarithmic phase could be unreliable if exponential growth was already in progress at the start of testing. This is also of importance in the preparation of samples for analysis, where extended handling time under suitable temperature conditions could allow for growth of cells before testing.
This cm produce inflated, inaccurate estimates of bacterial nurnbers if pararneters such as length or onset oflag phase are considered. In this case, the dilution of food samples during preparatory steps such as stomaching, and the use of a small sample size are useful in keeping bactenal levels below the threshold levels required for logarithrnic growth to quickly ensue. However. this can also affect the lower sensitivity levels of the assay as will be discussed next. The slopes of the curves in the logarithmic phase were similar for each bacteriai concentration, indicating similar rates of growth (Figure 4.1 .). This is expected for bacteria tested in pure culture, where there are no competing microorganisms either slowing the growth of. or growing faster than the bactena to be detected (Mattila, 1987;
Mattila & Alivehmas, 1987). Arrival of sample opticai densities at the stationary phase is not shown in Figure 4.1 ., but maximum levels between 0.8 and 0.9 density units were reached for al1 ce11 concentrations, tailing off only slightly over a 48 hour penod. Similar growth curves were demonstrated by Jmgensen and Schulz (1985) and Mattila (1987), illustrating the potential of such data to estimate numbers of bacteria present in sarnples.
Only the 10 CFU/mL concentration failed to demonstrate growth in this system, and this can be attributed to the small sample size effectively "diluting out" lower levels of bacteria.
Based on Figure 4.1, the lower limit of this assay was detemiined to be an initial concentration of 10' CFU/mL, detectable within 27 hours. This is in agreement with estimates for pure cultures (Jmgensen & Schulz 1985), raw milk (Mattila & Alivehmas,
1987), and other food samples (Mattila, 1987). In addition, it is recommended that samples with bacterial loads estimated to contain greater than 107CFU/mL be diluted as a precautionary step to ensure that fdse or inaccurate results do not occur.
4.3.2. Cornparison of analysis parameters Using the Bioscreen data processing package, a number of parameters can be estimated
from optical density (O.D.) data. Their usefulness in data analysis is obviously dependent on the experiment being conducted. with some more suitable indicators of tme differences than others. Based on the curve observations fiom Figure 4.1 , parameters associated with the lag and log phases of growth were chosen as most likely indicators of detection. These included initial and maximum O.D.,time to reach maximum O.D., the slope of the log phase (measured manually by setting of mark lines), detection time
(defined as the time in houn required for an optical density increase of 0.2 units), tum- point O.D. and turn-point time (defined as the mid-point of the log phase). Optical density data generated over a 48 hour period from the growth of different concentrations of L. monocytogenes was used to compare these parameters (Table 4.1). Regression coefficients for initial plate counts and parameters were calculated to assess the ability of each to di fferentiate initial bacterid levels.
4.3.2.1. Initial and maximum optical density
Initial optical density for al1 concentrations of L. rnonocytogenes was similar with the exception of the highest concentration. At an initial level of 1 x [O9 CFU/mL, dilution during sarnple preparation would still result in a sample containing 3.5 x 10' CFU of
Lisferia, a level high enough to have an impact on optical density. However, a Mer10-
' dilution effectively removed this turbidity difference, resulting in samples with similar optical densities. Typically, samples of such high bacterial concentrations would not be Table 4.1. Validity of Bioscreen analysis parameters for the dinerentiation of initial concentrations of L. monocytogenes ATCC 23074
II 1nitia.i 1 Bioscreen Analy sis Parameters
l First Max. Tirne to Slope Detection Turn Tuni O.D. O.D. Ma.. . of log time Point Point O.D. phase (Hours) O.D. Time (Hou=) (Hom) 1.012 8.50 N.D. 3 -62 0.562 4.50 0.946 12.00 75 6.12 0.477 7.00 0.966 15.00 77 9.12 0.540 10.50 1.012 21.50 76 15.62 0.533 17.00
N.D. Not detennined * r calculated excluding 10' CFU/mL data encountered fiequently, but there are ways of dealing with background hlrbidity. One way is to incorporate a dilution step as described previously (Matilla, 1987), or to manipulate the initial optical density readings by andysing them ail from a set starting point. In terms of detection of bacteria, initial turbidity is obviously important as it dictates the starting point of measurements, but as an individual parameter it is incapable of indicating bacterial growth and detection (r = 0.63). Maximum optical densities were also similar for bacterial concentrations, although they decreased slightly at lower bacteriai concentrations. Hcwever, at an initial level of 10 CFU/rnL, maximum optical density increased by ody 0.004 units and indeed, the corresponding curve was a Bat line as observed for a sirnilar concentration in Figure 4.1. In common with initial turbidity values, maximum optical density does not give an indication of relative bacterial concentrations (r = 0.74). However, based on the difference between the maximum and initial optical density of a sarnple, when used in conjunction with the ploned growth curve. one can determine whether growth has occurred or not. This provides an ideal means of establishing the sensitivity limits of an assay, which in this case, as previously observed in Figure 4.1, corresponded to a lower lirnit of 10' CFU/mL of L. monocytogenes.
1.3.2.2. Time to reach maximum optical density
A more useful parameter proved to be that of the time required to reach maximum opticai density. With respect to decreasing numbers of bacteria tested, the time taken to reach a
139 maximum turbidity increased in accordance with the length of the lag and log phases. A
correlation coefficient of -0.97 was calculated for initial bacterial levels and
corresponding detection times. At the lowest combination, already observed not to
produce optical density changes associated with log~thmicgrowth, the time to reach
maximum optical density was reported as 48 hours. This was a function of the software, where it assessed that a maximum optical density had not been reached during the course of the experirnent and used the last measurement time as reaching a maximum O.D. This cm therefore be a slightly rnisleading pararneter, but a useful one nonetheless. if used in conjunction with, for example, maximum optical density or slope of the log phase.
43.2.3. Slope of log phage
The slope of the log phase for each bacterial concentration was cornpared, based on the manual setting of mark lines at the start and peak of the log phase. Essentially, this value is a measure of the rate of bacterial growth based on optical density changes over time, and as demonstrated in Figure 4.1, would not be expected to Vary much in pure culture at any bacterial level (Manila, 1987). Based on manual measurement of this pararneter, values were subject to variation depending on mark line positioning, and could not be assessed for the highest bacterial concentration where exponential growth was already under way. Remaining slopes were similar in value, indicating equivalent growth rates for al1 detectable concentrations. With a correlation coefficient of 0.66, slope was not suitable for predicting bacterial levels. However, it had limited use for predicting non- detectable samples where a zero slope was calculated, indicating that bacterial growth had not occurred (Figure 4.1 .).
4.3.2.4. Detection time
Similar to the time necessary to reach maximum optical density, the detection time measures the time taken for optical density to increase by a pre-determined value; in this case 0.2 units. Detection times were in al1 cases shorter than maximum O.D. times, again with the exception of the lowest bacterial concentration where a detection time of 38 hours was obtained. This is a hction of the analysis software, where a detection tirne corresponding to the end of measurement is inserted if one is not determined during the course of the experiment. This value is rnisleading as detection of Listeria did not occur at this level, and it would be better defmed as "not detectable within 48 hours".
However. in al1 cases, detection times alone were useful for confirming levels of detectable bacteria (r = -0.98) and provided results in a more timely fashion than maximum optical density estimates.
4.3.2.5. Turning point optical density and time
Tuming point optical density and time were aiso evaluated, based on their potential as an indicator of log-phase activity. As with other optical density measurements the cornparison of tuming point optical densities could not in general differentiate levels of bacteria present (r = 0.79). OnIy in the case of the lowest tested concentration did it indicate a difference but this could only be interpreted in combination with other parameters. Turning point time proved useful in predicting initial bacterial levels where bacterial growth had been confirmed (r = -0.97) but a tum-point value of 1.50 hours for the lowest concentration was an inaccurate assessrnent of logaridunic growth, giving a false-positive result for a negative sample. In addition, turn-point times took longer to provide a detection result than detection time measurements, reducing their overall usefùlness in turbidirnetric detection methods.
In concIusion. analysis of experirnentai parameters for the turbidimetrïc detection of bactena reveaied that some were not individuaily capable of reliably assessing the presence or levels of bacteria in sarnples, while others generated misleading results and extended time required for detection. This was also observed by Mattila and Alivehmas
( 1 987), who suggested that a combined parameter approach to predicting bacteriai concentration was better than any individual measurement. However, they did not assess the usefùlness of the detection time, defined as the time required for bacterial growth to increase optical density by a user-defined amount. Evaluation of analysis parameters in this study demonstrates that the Bioscreen detection time parameter is an appropriate predictor of initial bacterial concentration in samples. This was also reported by Matilla
(1 987) with correlation coefficients ranging from -0.7 1 to -0.90. This parameter was therefore chosen to assess the use of phages in turbidimetric bacterial detection. In addition, based on initial O.D. determinations the manuai setting of initiai optical density levels was adopted to minimise the background turbidity of samples.
Generated detection times (Figure 4.2.) were anaiysed using Iinear regression and the standard error of the predicted y-values for each x-value. An r value of -0.98 (~48; p 1.3.3. Effect of phage on bacterial detection As previously demonstrated in Chapter Three, phage B 1 is capable of infecting and lysing L. rnonocytogenes ATCC 23074 in the process of replication. This lytic phenornenon is accompanied by an observable decrease in turbidity as the ce11 density of suspensions infected with phage is reduced. It is therefore feasible that the action of bacteriophages during the growth cycle of bacteria can also be monitored by the measurement of optical density. Interpretation of results would however be slightly different in this case. based not on optical density decreases of turbid ce11 suspensions but on delays to increased turbidity of inoculated media. Based on previous analysis of parameten, mean detection tirne was used to assess both bacterial growth and the effect of incorporating specific bacteriophage into the turbidimetric assay. Figure 4.3. demonstrates the effect of addition of phage to actively growing suspensions of L. rnonocytogenes ATCC 23074. Both the concentration of phage and bacterial density 1 2 3 4 5 6 7 8 9 IO log CFUlmL Listeria Figure 4.2. Reproducibility of detection tirnes for 10-fold senal dilutions of L. monocytogerzes ATCC 23074 in Trypticase Soy broth Figure 4.3. Turbidimetric detection of L. monocytogenes ATCC 23074 in Trypticase Soy broth: Influence of bacterial and viral levels on detection times were altered to examine their impact on metabolic detection. The relationship observed between bacteriai numbers and optical density measurements has been reported by both Jergensen & Schulz (1 985) and Mattila ( 1987), and cm be used to constnrct standard curves and regression equations for estimation of bacterial load. When listerial concentrations are considered, an increase in mean detection time was apparent as the initial bacterial concentration decreased from 109to 10 CFUImL. In general mean detection times demonstrated an inverse linear Uicrease associated with the serial dilution of bacterial numbers. However. as previously discussed regarding analysis parameters. a 48 hour detection time was obtained for an initial bacterial concentration of 10 CFU/mL. This was descnbed as "Not detectable within 48 hours" by this method, and was indicative of negative detection of Listeria rnonocytogenes. The incorporation of phage into this system resulted in an increase in mean detection times, when compared to bactenal level in the absence of phage. Presumably the phage acts by intedering with bacterial metabolisrn, in the process of prornoting viral replication. and effectively reduces the population such that the time required to increase the optical density by 0.2 units is extended. At high PFU levels. the time required to reach optical detection was in general greater than at lower viral concentrations. At a phage concentration of 10' PFUImL, the majority of detection times were greater ai bacterial levels between 1O4 and 10' CFU/mL, as compared to higher viral concentrations of 109-1 0'' PFUlrnL. This can be explained in two ways: either the interaction of phage and bacteria is optimal at these levels, or an inaccuracy in the preparation of the 10" PFU/mL phage concentration occurred resulting in a higher phage titre than intended. As subsequent dilutions continued to demonstrate greater detection thes than some obtained using a phage titre of 109 PFU/mL, it is Iikely that the latter explanation is correct. Regardless, at ail concentrations of phage used, the time required for detection of L. monocytogencs was greater in the presence of phage as opposed to uninfected control samples. At very low initial levels of bacteria, a detection time was obtained for control samples. but a time was not achieved within 48 hours in the presence of phage. This can be best explained by the phage maintaining levels of bactena below the threshold for optical density detection. Based on information regarding viral replication thresholds, bacterial replication probably occurred until numbers were sufficiect for subsequent viral replication, reducing the bacterial population back down to low levels (Wiggins & Alexander, 1985). The ability of phage to either extend or completely elirninate detection is therefore a promising means of increasing the specificity of bacterial detection using turbidimetry. Although al1 phage concentrations tested produced similar effects on host bactenal cells. the greatest detection differences were observed at the highest bacteriophage concentration. Maximum titres of bacteriophages were therefore employed in al1 subsequent research- 4.3.4. Cornparison of phages Bacteriophages AT20 and Felix 0-1, previously demonstrated to be active agauist E. coli G2-2 and S. Typhimurium ATCC 14028 respectively, were compared to Listeria phage ATCC 23074-B i in their relative abilities to detect the presence of host bacteria. Results for mean detection times in the presence and absence of phage for each of the three species at initial concentrations of 1o9 to 10' CFUImL are shown in Table 4.2. As in al1 previous experiments, initial bacterial levels of 10' CFU/mL were not detected. regardless of the species of bactena employed. Indeed, at around 16 CFU!mL. L. rnonocytogenes was not detected either. However, incubation was in this case teminated afier 18 hours, before the typical detection time previously estimated for this level of Listerio (27 hours), and so it is likely that detection would have occurred had incubation been prolonged. Low levels of both E. coli and S. Typhimurium were respectively detected within 9 and 1 1 hours of test initiation in the absence of bacteriophage, while detection of L. rnonocytogenes took approximately two to three times longer to reach the optical detection iimit at al1 tested concentrations. These detection time data adequately demonstrate the difference in growth rates associated with these bacteria at temperature conditions optimal for al1 three species. Relative growth rates were confirmed by slope estimates of the log phase, with E. coli demonstrating the steepest log siope ( 157). followed by S. Typhimurium (136) and L. rnonocytogenes (43). In al1 cases, the addition of undiluted bacteriophages (at levels of 10" PFU/rnL) resulted in either delayed detection of bacteria, or no detection at dl. In particular, phage AT20 Table 43. Evaluation of phage-based turbidimetric detection of pure cultures of E. coli, S. Typhhurium, and L. nionu~ogenergrown in Trypticase Soy Broth 1 1 Bioscreen mean detection time (hours) Initial , L. 1 inoculurn E. coii S. TyphirnUnum monocytogenes (CFU/*) -phage + phage - phage + phage - phage + phage AT20 AT20 Felix 0- 1 Felix 0- 1 BI BI 1x10~ 0.66 14.58 1 .O8 6.42 3.33 4.50 1x10~ 1.42 16.25 2.25 7.83 4.83 6.75 1x10' 2.42 N.D. 3.42 1 0.25 7.17 9.92 1x106 3.42 N.D. 4.75 13.58 9.50 12.25 - - 1x10~ 4.66 N.D. 6.00 N.D. 1 1.83 14.66 lxloJ 5.75 N.D. 7.17 N.D. 14.33 16-50 lxlo3 6.75 N.D. 8.66 N.D. 16.75 N.D. 1x10' 8.75 N.D. 10.25 N.D. ND. N.D. 1x10' N.D. N.D. N.D. N.D. N.D. N.D, N.D. Not detectable within 18 hours * r calculated only for detected levels produced the greatest detection differences between samples containing E. coli G2-2, with only the two highest concentrations capable of producing a 0.2 optical density unit increase. Al1 lower concentrations were not detectable within 18 hours of testing. In contrast, Salmonella phage Felix 0-1 produced approximately three- to four-fold differences in detection times between infected and uninfected sarnples, with initial levels of 1o5 CFUimL and below undetectable in the presence of phage during the 18 hour incubation period. Listeria phage B 1 appeared to be the least efficient of the three phages tested, producing delayed detections of onIy two to three hours, with levels of 10' CFU/mL and below undetectable in the presence of phage. These are rather interesting data when compared to the information gleaned about these phages from the lytic ATP bioluminescence reaction described in Chapter three. Based on the assumption that a large amount of ATP released from cells was the result of a highly efficient lytic phage, Listeria phage ATCC 23074-B 1 was assurned to be the most efficient lytic agent (comparing similar phage numbers) despite being identified as lysogenic in nature in Chapter Two. In contrast AT20 and SJ2, previously identified as lytic phages (also Chapter Two), appeared to release smaller amounts of ATP over a longer penod of tirne, particularly in the case of AT2O. From this information, it could be speculated that the E. coli phage, and to a lesser degree the Salmonella phage, is capable of continuously infecting and lysing a nurnber of bacterial cells suficient to keep levels below detection thresholds. As previously descnbed, this would also depend on viral replication thresholds for each of the phages (Wiggins & Alexander, 1985). Replication would continue either until al1 cells were destroyed (at lower bacterial levels) or sufficient numbers of cells remained uninfected to push levels over the threshold of detection. In Chapter Three, phage AT20 was determined to take longer to lyse cells than phage B 1. Also taking into consideration the faster growth rate of E. coli, it is possible that replication of uninfected cells during and between viral replication cycles produced excess numbers of bacteria to accommodate both the next round of viral replication plus further bacterial replication. With such a fine balance between bacterial and viral replication, it seems feasible that it would take some thefor bacterial levels to reach the detection threshold. This theory assumes both that viral replication of phage AT20 is indeed slow, and theretbre does not radically change the phage titre over time. The scenario associated with the Listeria phage might similarly be explained in relation to the ATP bioluminescence profiles observed in Chapter Three. The Listeria phage B 1, when exposed to its propagating strain, very quickly lysed a proportion of the bacterial population thereby reducing ce11 nurnbers below the detection thresho Id. However, subsequent infection was not detected by ATP bioluminescence over a 2-3 hour penod of incubation . Given that phage B 1 was identified as lysogenic in Chapter Two, it is likely that a high proportion of progeny phage adopts a temperate as opposed to virulent life cycle following replication, thereby depleting the numbers of virulent phage in the system. 15 1 This serves two purposes: first, Listeria cells harbouring the prophage would be immune to reùifection by phage B 1, and would be able to replicate unhindered above and beyond detection threshold levels. Second, the depletion of lytic phage levels wouid effectiwly reduce the incidence of viral infection of uninfected bacterial cells, allowing them to also contribute to increasing optical density levels and subsequent detection. This wouid account for smailer differences between detection times for infected and uninfected ce11 concentrations as compared to the action of the E.coli phage. This would be particularly true at lower bacterial concentrations, where the high multiplicity of infection would favour Iysogeny and fewer concentrations would go undetected in the presence of phage (demonstrated in Table 4.2.). The activity of Salmonella phage Felix 0-1 falls in between that descnbed for E. coli phage AT20 and Listeria phage B 1 . Salmonella phage Felix 0-1 demonstrated lytic activity within 30 minutes of exposure to susceptible bacterial hosts, but detectable ATP concentration remained stable over a two to three hour period of time. With the Sulmonrlla phage assessed to release the lowest levels of bacterid ATP by ce11 lysis. it would appear in this case that a more equitable balance between viral and bacterial replication is being displayed. At higher ce11 concentrations, the lytic activity of Felix O- 1 is adequate to delay detection but is not suficient to keep levels below the detection threshold in the same marner as E. coli phage AT20. As bactend concentrations are reduced however, lytic activity would likely supercede ce11 multiplication, and numbers would be kept below the threshold of detection. This contrats with the lysogenic 153 immunity conveyed by Listeria phage B 1, which effectively prornotes the replication of bacteria over phage particles. Regardless of proposed modes of action. dl phages were capable of confiigdetection of their propagating hosts as detemined by turbidimetric analysis and they offer the potential to increase the specificity of this and other rnetabolic methodologies. 43.5. Effect of contaminants on phage-based bacterial detection The relatively slow growth rate of L monocytogenes (Table 4.2.) becarne the primary focus of subsequent research in the development of a phage-based turbidimetric assay for its detection. Both the ability to compete with other bacterial contarninants, and the ability of phage B 1 to confirm detection of Listeria were Merinvestigated The detection of combinations of bactena was evaluated, both in the absence and prescnce of one or more phages to get an idea of both the competitiveness of bacterial species at similar levels and their subsequent detection by specific phage. Resulting mean detection tirnes are demonstrated in Table 4.3. Detection times for E. coli, S. Typhimuriurn, and L. monocytogenes tested at initial levels of 10' CFUlmL were included for cornparison of subsequent detection times when more than one species was present. As demonstrated previously, growth rate was fastest for 153 Table 43. Effect of contaminants on phage-based turbidimetric detection of E. coli, S. Typhimuriurn, and L. monocytogenes grown in Trypticase Soy Broth Bioscreen mean detection the(hours) Combinations of bacteria (at 3.5~10) CFU/rnL) 1 AT20 1 Felix 1 BI 1 BI 1 phages fi 5.75 N.D. I . 7.17 N.D. E. c and S.T S.T and Lm N.D. Not detectable within 18 hours E.c E.coliG2-2 S.T S. Typhimurium ATCC 14028 L.m L. rnonocytogenes ATCC 23074 Exoli, followed by Salmonella and then the slowest growing Listeria isolate. while phage addition resulted in an inability to detect both E. coli and S. Typhimurium, and a two hour delay in detection of L. rnonocytogenes. When E. coli was added in equal concentration as a contaminant of the Listeria detection assay, detection was only 0.2 units later than that for E. coli alone. This suggested that Listeria contributed little to increasing the optical density of the growth medium by ds inability to compete with the faster growing E. di. In the presence of phage AT20. a detection time cf almost 14 hours was observed, while detection was slightly faster in the presence of phage B 1 as cornpared to the control. With L. monocytogenes only, these data suggest that Listeria was able to produce a detection time only when phage AT20 had effectiveiy eliminated the cornpetitive E. coli levels. In the presence of phage B1, detection was related both to growth of E. coli and removal of a limited nurnber of listeriae by Iytic activity of the lysogenic phage. This combined effect produced a slightly faster detection than that observed in the absence of phage. Although L. monocytogenes was present in this system, faster growing competition and lysogenic activity of the phage prevented a positive detection of Lisferia to be confirmed. Exactly the sarne scenario was observed when S. Typhimurium was added as a contaminant of the Listeria assay, where Felix 0-1 was capable of confirming detection of Salmonella but phage B 1 failed to detect Listeria. With Salmonella and E. coli present in the same detection system, detection took one hour longer than that observed for E. coli alone, but was the same as for Salmonella alone. This is probably due to slightly greater competition from Salmonella (due to its faster growth rate than that observed for Listeria). Confurnation of E. coli detectim was achieved using phage AT20, where detection was delayed due to lysis of E. coli and a slower growth rate for Salmonella. Felix 0-1 was however unable to confirm detection of S. Typhimuriurn, as phage activity presumably removed the cornpetitive pressure of Salmonella, giving E. coli an uninhibited environment to grow in. The combination of ail three species in equal nurnbers produced predictable results based on those discussed above. The detection time obtained for ail three species kvas identical to that previously determined for E. coli and S. Typhimuriurn together, while addition of phage B 1 could not confirm the presence of Lisleria in this combination. The addition of phage AT20 produced a delay in detection time associated with confirmation of the presence of E. coli The further incorporation of Felix 0-1 resulted in an even larger detection time likely attributed to both detection of Salmonella and slower growth of Listeria following the reduction of both E. coli and Salmonella. What the results of this expenment effectively demonstrate is that competing bacteria can have a negative impact on the use of bacteriophages to confirm the detection of target organism controls. In particular, faster growing bacteria can mask the presence of target bacteria, as was the case here with Listeria monocyfogenes ATCC 23074. This strain is slower growing and unable to compete well in such a system, and phage infection and replication was ultimately afTected resulting in an inability to contiirm the presence of target cells. These results also suggest that assays for the detection of Listeria monoqtogenes in foods would have to rely more heavily on selective media to minimise contaminants than assays for highly competitive bacteria such as E. coli 4.3.6. Effect of increased selecîivity on phage-based bacterial detection Concems regarding the slow growth and poor cornpetitiveness of L. rnonocytogenes ATCC 23074 prompted the incorporation of selective agents used in Listerin Oxford agar into the phage assay. Detection times for decreasing concentrations of listenae were determined as before in TSB. In addition the efiects of (i) increased selective pressure of the medium (SEL), (ii) addition of phage (+/-B l), and (iii) a combination of these two factors were assessed by the cornparison of mean detection times obtained over a 48 hour incubation period (Figure 4.4). Typical growth of initial bacterial concentrations of 10? to 1o9 CFU1m.L in TSB in the absence of phage B 1 was observed by a concomitant drcrease in rnean detection times. As demonstrated previously, initial concentrations of 10' CFU/mL were not detectable using turbidimetry, as levels were below the detection sensitivity of the assay. A similar trend was obtained for bacterial growth in the presence of more selective media pressure, although detection was, as expected, slightly delayed. As before, the addition of phage B 1 to the assay Merincreased the time required to reach an optical density detection level, with 1o2 CFU/mL undetectable within 48 hours of incubation. The combination of selective supplement and phage B 1 however, had the greatest impact on detection, with 9 log [CFUlmL] listeriae Figure 4.4. Turbidimetric detection of L. monocytogenes ATCC 23074: Interaction of phage B1 and increased media selectivity al1 tirnes greater than those observed for the effect of phage or increased selectivity alone. Indeed, initial concentrations of 10' CFU/rnL and below were no longer detectable within 48 hours. The combination of increased selectivity and phage infection acted in a synergistic fashion: bacteriophage action reduced bacteriai numbers, while increased selectivity slowed both the growth of uninfected and surviving cells~subsequently delaying detection. The only exception was the 109 CFUImL concentration, which did not demonstrate a difference in detection time based on increased selectivity, in combination with phage. It would appear that phage activity played a more crucial role in influencing detection, while increased selectivity of the medium had little effect on bacteria already in the logarithrnic growth phase. Using this synergistic approach, detection of initial Listeria levels as low as IO' CFU/mL could be positively confirmed. Based on the poor cornpetitive nature of L. monocytogenes and the inability of phage B 1 to confirm detection (Table 4.3.) this approach appeared to be the most feasible way of detecting listeriae in a mixed microflora. This more selective system was thus examined Merfor the phage-based turbidimetric detection of L. monocytogenes in milk. 43.7. Phage-based detection of L. monocytogenes in pasteurised milk Pasteurised milk, with an initial count of 10" CFUIrnL, was chosen as a more appropriate milieu in which to evaluate the phage-based detection of L. monocytogenes, as opposed to a higher and more diverse raw mik microflora. This decision was based prirnarily on dic poor competitiveness displayed by L. rnonocyfogenesATCC 23074 in the presence of coliform contaminants (Table 4.3 .) and the selectivity of the medium employed. To expand on media selectivity further, the following information is important. When the antimicrobic supplement is used in combination with Oxford medium base, these agents are reported to completely inhibit Gram-negative organisms and most Gram-positive organisms (Curtis & Lee, 1995). However, this inhibition relies not only on the incorporation of antimicrobial cornpounds but on the presence of hi& concentrations of Lithium Chloride (15 g/L). The hi@ salt tolerance of L. monocyfogenes dlows it to grow in such a medium while other bacteria such as enterococci are inhibited. Using the supplement in a medium such as TSB with a significantly lower salt concentration wili therefore effectively decrease the overall selective potential of the system. As a result this could allow the cornpetitive growth of non-listeriae such as enterococci as post- pasteurisation contaminants (Frank et al., 1990). The extent of this growth will thus contribute to determining the ability of listeriae to compete and grow to threshold levels required for detection. Milk was artificially contaminated with a high concentration of listeriae, at a level of 10' CFUlrnL, ten times higher than the background count. While this is an unrealistic level as compared to typical levels reported in raw and pasteurised milk (Ryser, 199 1), this level was used in an attempt to address a number of limitations of this assay. First, L. rnonocytoger;es was uncornpetitive in the presence of equal volumes of contaminants (Table 4.3.) so a level the same or lower than the background microflora wouid likely not be detected. This also impacts on the minimum levels of bacteria required for viral replication reported by Wiggins and Alexander (1 985), estimated to be approximately 1O4 CFU/rnL. Second, the decreased sensitivity of the assay associated with srnaIl sample volume (35 PL) and the sample dilutions used (IO-' to IO-') to elirninate turbidity (Matilla, 1987) would also impact on specific detection of L. monocytogenes at low levels. Results for turbidimetric detection of L. rnonocytogenes in miik under these optirnised conditions are show in Table 4.4. Ten-fold dilutions from 10-' to 1 O-;, corresponding to L. monocytogenes levels of approximately 350,35 and 3.5 CFU per well, were tested. Initial optical densities and subsequent dilution are included to illustrate the effect of the milk sarnple on background turbidity. Detection times were calcufated using a 0.2 unit increase in optical density as before. Using a 1O-' dilution, background turbidity was measured at an unacceptable level of 2.124-2.205 units. Typical growth curves were not observed using this dilution as turbidity was already too high for subsequent growth to make a difference. In addition, although turbidity did not change much fiom its initial high level, larger fluctuations in measurements resulted in unreliable detection times. [n the case of selective broth, turbidity levels fluctuated less and L. monocytogenes, at a level of about 350 CFU, was not detected at dl. Table 4.1. Phage-based turbidimetric deteetion of L. rnonocytogenes in artif~cially contaminated 2% milk using non-selective (TSB) and selective (SEL) media - - - Medid Mi& dilution factor (concentration of Listeria) phage O-; BI 1O-' 1O-' 1 Conditions (350CFU/well) (35CFUfwell) (3.5CFU/well) Initial D.T. InitiaI D.T. Initial D.T. O.D. (hours) O.D. (hours) O.D. (hours) TSB/-BI 2.145 14.50 0.706 9.12 0.199 9.75 TSB/+Bl 2.125 11.50 0.669 7.86 0.189 8.12 SEL/-B1 2.205 N.D. 0.739 14-62 0.260 15.62 SEL/+Bl 2.124 N.D. 0.684 35.87 0.249 38.62 N.D. Not detectable within 48 hours Even after a 1 dilution step, a hubidity of 0.684-0.739 units was measured. However, in this case and at a subsequent IO-' dilution, growth curves were obtained and detection times could be detennined. However, Matilla &Alivehmas ( 1987) recommended the use of the IO'j dilution based on varying bacterial levels observed for raw milk sarnples, and better growth curves at this dilution level. A dilution of 1 O^' was not recornmended as lower bacterial leveIs would not be detected, and indeed this dilution was not tested here as it was estirnated to resuit in ce11 nurnbers c3.5 CFU after dilution. Looking at detection times in TSB, excluding the 1O-' dilution data, the time to reach the optical density detection threshold was between 9.12 and 9.75 hours. However, the addition of phage B 1 was unable to codum that the detection times obtained were for L. monocytogenes. hdeed, in the presence of phage, detection times were lower than controls, but the difference between them was consistent. As previously described for the detection of L. monocytogenes in the presence of coliforms (Table 4.3.) it is likely that initial lytic activity by phage B 1 gave faster growing non-listeriae a more competitive advantage. In this case, with an initial microflora of 1 x 10' CFU/rnL, a 18' dilution followed by testing of only 35 pL would effectively reduce the background microflora to about 35 CFU. The Bioscreen is capable of measurùig optical density changes originating fiom the presence of single ceils (Stephens et al., 1998) and thus, the inaccuracy of dilutions could allow for at least one ce11 or more to be present in the 10-3 dilution. With the slow growth of L. monocyfogenes,present at only a 10-fold higher concentration, it is likely that smaller detection times in the presence of phage were due to the background microflora. Increasing the selectivity of the medium gave contrasting results, particulady in combination with phage B 1. Detection tirnes for milk dilutions of 1 and 10" occurred approximately 5 and 6 hours Iater than those observed in TSB, while the presence of phage increased detection tirne drarnatically, by a factor of almost five as compared to TSB with no added phage. It is quite clear fiom these data that combining Lisferia phage B 1 with a more selective medium greatly enhanced the turbidimetric detection and confirmation of L. rnonocytogenes in pasteurised milk. This is probably the most feasible approach to detection. certainly with this host-phage combination, given the less than desirable lysogenic nature of phage B 1 and the poor competitiveness of L. rnonocytogenes ATCC 23074. In conclusion, a number of possibilities for assay optimisation are available. These would include the use. where possible, of lytic as opposed to lysogenic phages. Lytic phages have been identified for Lisferia. although they are less commonly isolated (Loessner, 1994). However, based on the use of other lytic phages in this research, it is possible that they would be more effective than the lysogenic phage B 1 in detecting listenae. Certainly for detection of Listeria spp., if not al1 food-borne pathogens, a non-selective enrichment step both for detection of low numbers (aO/g),and recovery of injured cells 164 would be a usefùl addition to turbidimetric detection. The concept of dual purpose non- selective/selective broths such as Lisleria repair broth (Curtis & Lee, 1995), could be an ideal way of non-selectively increasing numben. followed by a morc selective entichent system coupled with addition of phage to both detect and confithe presence of target cells. In the case of fluid samples such as milk, a filtration step (similar to that used in ATP bioluminescence applications) prior to enrichment would be a useful means of concentrating bacterial cells and removing background turbidity. The concept of combining npid methodologies is also applicable here, where sensitivity and specificity could be improved by separating low concentrations of target cells fiom food samples or e~chmentbroths by immunomagnetic separation. This wouid also serve to eliminate both the interfering microflora and the background turbidity associated with food samples. Alternatively, the use of clarification or centrifugation techniques (Jmgensen & Schulz. 1985) to remove food particulate would be usehl in reducing or eliminating background turbidity . CHAPTER FIVE ALTERNATW PHAGE-BASED BACTERIAL DETECTION METHODS : IMPEDANCE MICROBIOLOGY AND REFLECTANCE COLORIMETRY S. 1. INTRODUCTION As discussed in previous chapters, the application of phages in bacteriai detection systems is govemed by a number of factors, concering in paaicular the interaction of the phage with its target host population. For example, the use of a lysogenic Listeria phage was shown in Chapters Three and Four to be limited in its application in specific detection methods since infection produced immunity in the target ce11 population, preventing subsequent infection. In contrast, lytic Salmonella phages demonstrated to identiQ target cells to both the genus and species levels (Chapter Two) have great potential for application in both generic and species-specific assays. Metabolic detection methods relying on selective media as their primary means of targeting particular groups of microorganisms are only as specific and time-consuming as the medium dictates. As mentioned in Chapter Four, there is no such thing as an "ideal" medium that will recover and grow al1 target cells at the expense of other groups of organisms (Blood & Curtis, 1995). Additionally, sensitivity is dependent on the mode of detection, with sarnple size and preparatory steps crucial in determining the upper and lower limits. The techniques of ATP bioluminescence and turbidimetry described previously are both useful methods for bacterial detection. However, thek drawbacks in terms of sensitivity will still have to be addressed even with the inclusion of bacteriophages. Other metabolic rnethods for the analysis of bacterial growth are available, including impedimetric and colorimetric assays, and in contrast to the previously described methods, they can detect the presence of one or more cells with greater flexibility in terms of media selection, sample size and format (Firstenberg-Eden & Klein, 1983; Manninen & Fung, 1992). According to their evaluation in biolurninescent and turbidimetnc assays, phages are well suited for confirmation of specific bacterial detection. Their ability to influence detection data by infecting and reducing numbee of target cells present in a test sample suggests that they can also be usefül in impedimetric and colorimetric detection to confïrm the presence of particular bacteria. These technologies offer a more sensitive assay in terms of test sample size (1 00 pL - i rnL), and are not influenced by background turbidity as was the case with turbidimetric detection. Indeed, reflectance colorimetry requires a particulate sample to produce enough reflectance to colour changes to be measureable above signal drift. The use of bacteriophages in impedimetric assays was first suggested by Pugh & Arnott (1 989) for the detection of Salmonella species, but the incorporation of highly selective 167 media increased the tirne required for detection and confirmation of isolates. The use of semi-selective and differential media was therefore investigated in this study, in conjunction with bacteriophages for the impedirnetric detection of L. monocyfogenes, Salmonella spp. and E. coli in various food cornmodities. Phage-based colorimetric detection of E. coli was also investigated as an alternative detection method based on the claims made by Richardson (1 988) that reflectance colorimetry suffered fewer technical dificulties than electncal measurements. 5.2. MATERIALS AND METHODS 52.1. Establishing the detection limits of the impedimetric assay Listeria monocytogenes ATCC 23074 was grown in stede Trypticase Soy broth (TSB; Difco Laboratories, Detroit, MI) incubated at 37OC for 18 hours. Sample information and experimental parameters were first imputed into the Bactometer processing software. One of these parameters - the test code - required information input regarding the choice of electrical signal to be used. For Listeria work, a capacitance signal was used based on previous research by Griffiths & Phillips (1989) and Rodrigues et al. (1995). Sterile 16 well plastic modules were filled with 900 pL of TSB. Ten-fold serial dilutions of cultures were prepared in sterile 0.1 % peptone water and 100 pL volumes were inoculated into duplicate wells. Experiments were conducted in duplicate and results were expressed as the mean of four readings. Dilutions were also plated ont0 Trypticase Soy agar (TSA; Difco) to confibacterial concentrations employed. Media and phage suspensions were also included as controls to ensure sterility of reagents. Immediately following completion of inoculation, modules were placed into the Bactometer processing unit and electrical measurernents were taken every six minutes for up to 48 hours. Impedance information was generated using the Bactometer software package, and the impedance detection time (IDT), defined as the time required for an initial bacterial population to reach the Bactometer threshold level, was caiculated for each well. Correlation between detection times and plate counts was calculated by regression analysis to indicate its usefulness as a predictor of initial bacterial levels, and the reproducibility of duplicates was analysed by calculating the variation of values about the regression line. 5.2.2. Effect of phage on impedimetric detection The effect of phage on bactenal growth and detection of L. monocytogenes ATCC 23074 was subsequently evaluated. Detection time data were compared in both the presence and absence of Listeria phage ATCC 23074-8 1. Bacterial dilutions were prepared in 0.1 % peptone water, and 100 pL volumes were tested in duplicate at initial levels of IO8 to 10' CFUhL. Listeria phage B 1 was used undiluted at a titre level of 10" PFU/mL. Correlation coefficients were calculated between bacterial counts and impedimetnc 169 detection times. 5.2.3. Evaluation of media setectivity on impedimetric detection Detection of Listeria monocytogenes was evaiuated at 37OC in pure culture using both TSB, and TSB plus selective supplement (SEL) previously used in Chapter Four. The parameters of cuve quality (e.g. baseline drift signai fluctuations, % signal increase), detection times, and correlation with bacterial concentrations were used as indicators of media suitability. Alternative media for the elecûical detection of E. coli and Salmonella were not sought as the differential coliforrn medium (CM; bioMerieux Vitek Inc., Hazelwood, MO) is recommended specifically for Bactometer use. Based on previously reported information. a conductance signal was selected for detection of both E. coli and Salmonella species (Suhren & Heeschen. 1987; Fryer & Forde, 1989; Petitt, 1989; Prentice et al., 1989; Pugh & Amon. 1989). Results using this medium were compared to the Listeria media. 5.2.4. Deiection of L. monocytogenes in raw milk Raw milk was studied in this protocol rather than pasteurised milk, as sensitivity limitations encountered in the turbidimetric analysis of milk, including background turbidity and sample size, were less critical in impedimetric analysis. Pooled raw milk 170 samples obtained from Laboratory Services Division, University of Guelph were temperature abused at 2 1OC overnight to yield a bacterial concentration of about 1O6 CFU/mL. This high background level was used to evduate the selective pressure of SEL in cornparison with the non-selective TSB. Nine mL aiiquots of milk were artificially contaminated with 1 mL volumes of L. monocyfogenesATCC 23074 dilutions to give final concentrations of 10' to 1O6 CFU/mL. Uninoculated raw rniik from the same sample was used as a control. Sterile module wells were prepared in duplicate with 800 pL of either TSB or SEL broth, plus 100 pL of Listeria phage B 1 or phage-fiee suspension. One hundred pL volumes of inoculated and control milks were added in duplicate to test and control wells. Modules were incubated in the Bactometer processing unit at 37°C for 48 hours. and mean detection times were compared for the two media. 5.2.5. Detection of Salmonella spp. in skim milk powder Eighteen hour cultures of Salmonella serovars Enteritidis and Braendemp grown in TSB at 37°C were senally diluted in peptone water. Twenty five gram quantities of skim milk powder were suspended in appropriate peptone dilutions to achieve a final concentration of approxirnately 10' CFUig. Uninoculated skim milk powder was also suspended in sterile peptone water prior to testing to be used as a control. Stenle module wells were prepared in duplicate with 800 pL of liquid CM (bioMerieux Vitek, Inc.), plus 100 pL of either (i) Salmonella phage Felix 0-1, (ii) Salmonella phage SJ2, or (iii) phage-free suspensions. Uninoculated and inoculated skim milk powders were then added to 172 modules wells in duplicate 100 pL aliquots, in the absence and presence of phage. Modules were incubated in the Bactometer processing unit at 37OC for 24 hours, and mean detection times were used to compare detection of each of the Salmonella isolates. Uninoculated skim milk powders were also plated on MacConkey agar to confimi the presence or absence of coliforms in these controis. 5.2.6. Cornparison of colorimetric and impedimetric detection of E. coli 5.2.6.1. Colorimetric instrument set-up Colorimetric measurements were made using the LabSMARFMcolorimetric software system (LabSMART LLC, Logan UT) and a flatbed colour scanner (Hewlett Packard ScanJet). Although typically used at roorn temperature. the scanner could be piaced inside an incubator for higher temperature work. In this case, a 37OC temperature was chosen for most expenments, but it was possible to use higher temperatures of up to 44.j°C with minimal deterioration of data quality. Sample information and experimental parameters were entered according to the Setup & Start instructions of the User Manual (Version 2.0; Richardson, 1997). Colour measurements of microtitre plate wells were determined using the manufacturer's recommendation of a 35 x 35 pixel spacing, while test tubes piaced in racks were set up with 73 x 75 pixel dimensions. Colorimetric data was read every six minutes for up to 24 hours. Colour changes were also determined visually following measurement termination. Endpoint (EP) values, using the formula EP = initial b*+ x, were investigated as a suitable parameter for evaluating bacterial detection. EP is defmed as the time taken, in hours for an initial media colour value b* to increase by a user-defmed arnount x. The b* value evaluates the purple (-b*) to yellow (+b*) colour change of a pH indicator system (bromcresol purple) incorporated into a growth medium. Fermentation of lactose, also in the medium, decreases the pH by the production of iactic acid. When pH decreases beiow a particular level, the pH indicator will change fiom purple to yellow. In the case of bromcresol purple, it is useful over the pH range 5.2 - 6.8. Using the endpoint formula expressed above, x values of five unit increments were evaluated up to a maximum of 33 units, based on b* colour profiles and visually determined colour changes. 5.2.6.2. Colorimetric and impedimetric detection of E. coli GZ-2 in raw milk E. coli G2-2 was inoculated into Trypticase Soy Broth (TSB) and incubated at 37T for 18 houn. Ten-fold senal dilutions were prepared in O. 1% peptone water and 1 mL volumes of each dilution were added to 9 mL aliquots of raw milk (Laboratory Services Division). After thorough shaking, duplicate 25 pL volumes of each artificially contaminated rnilk sample were added to pre-prepared microtitre plate wells conraining Coliform Medium (CM; bioMeriew Vitek. Inc.). Duplicate 100 pL volumes were added to prepared Bactometer wells also containing CM, and dilutions were also plated on MacConkey agar (Difco). Sarnples were tested both in the presence and absence of phage AT20. The completed microtitre plate was immediately placed on top of the flatbed scanner, and the LabSMART program was initiated. Media colour changes were monitored at 37OC and following termination of measurement after 24 hours, endpoint andysis was conducted. Both colonmetric endpoint and impedimetrk detection times were compared to E. coli plate counts to assess the suitability of these values for predicting initial bacterial levels. 5.2.7. Colorimetric detection of E.coli G2-2 in raw minced beef E. coli G2-2 was grown in TSB (Difco) at 37OC for 18 hours, and 1 O-fold serial dilutions were prepared in 99 mL volumes of 0.1 % peptone water. Raw minced beef purchased from a retail store was split into 25 g portions in stomacher bags, and each was artificially contaminated with the entire volume of one of the ten prepared dilutions. Bag contents were stoniached for 2 minutes. I mL aliquots were then removed and added to tubes, containing 8 mL of sterile CM (bioMeneux Vitek, Inc.) supplemented with 1 rnL of phage AT20 or phage-free suspension. Tubes were immediately inserted into test tube racks positioned and secured on top of the flatbed scanner. Media colour changes at 37OC were monitored and analysed as described previously. Plate counts were performed on MacConkey agar (Difco) incubated for 24 hours at 37OC io confirm levels of E. coli G2-2 inoculated into food samples. 5.3. RESULTS AND DISCUSSION 5.3.1. Interpretation of impedimetric detection times Figure 5.1. illustrates a typical capacitance curve obtained for an initiai L. monocyiogenes concentration of 1 x 1O6 CFU/mL inoculated into TSB. This curve initidly demonstrated a slight drift reducing the percentage (%) signai change during the first 7 hours of incubation. T'his is a generally undesirable trait of an impedance medium, as it can make the assessrnent of detection time more difficult if signal acceleration is low (Firstenberg- Eden & Klein, 1983). However, in this case the signal change increased rapidly afier about 7 hours of incubation. This rapid increase is due to bacteriai metabolism and multiplication of ce11 levels to a threshold Ievei of approximately 1O6 to 10'. affer which changes to the electrical properties of the medium cm be detected (Firstenberg-Eden & Klein, 1983). Detection times generated by the Bactometer software are calculated as the time corresponding to the intersection of the baseline slope with the signal acceleration slope. In this case, a detection tirne of 7.4 hours was generated. A 60% change in signal occurred after this point. peaking after approximately 13 hours of incubation. Incubation Time (Hours) Figure 5.1. Capacitance detection cuwe for L. monocytogenes ATCC 23074 growing in Trypticase Soy broth Capacitance changes ranging from 65-90% were reported by Phillips & Griffith (1 989), with differences attributed to medium composition. A characteristically shaped cuve response similar to the curve show in Figure 5.1. produced by capacitance was also reported by these researchers. While this should not be used alone to assess impedimetric data, it is a usefùl parameter to use in conjunction with detection times, and colour changes where indicators are incorporated into growth media- 5.3.2. Detection limits of the impedance detection assay The detection of L. rnonocytogenes ATCC 23074 was first evaluated as a means of correlating the impedance detection time with initial bacterial concentrations. By plotting IDT versus incubation time, the upper and lower limits of detection could be determined, and the repeatability of duplicates was assessed. Figure 5.2. demonstrates the relationship between IDT and different initial levels of L. rnonocyrogenes ATCC 23074 grown in TSB. An inverse relationship is apparent, where detection takes longer for samples at lower concentrations. For example, an initial concentration of 10' CFU/mL takes approximately 5 hours to reach a detection threshold, while a concentration of 5 x 1O3 CFU/mL takes 15 hours before detection occurs. The correlation coefficient was calculated to be an r value of -0.99 (n=72; p<0.001). This demonstrated the ability of detection times to accurately reflect the initial bacterial concentration. Similar standard cuves generated for pasteurised milk and cream, ice O 5 IO 15 20 25 30 Mean lmpedance Detection Time (Hours) Figure 5.2. Relatiooship between bacterial count and Impedimetric detection time (IDT) for L. monocytogenes ATCC 23074 growing in Trypticase Soy broth crearn mix. and raw milk by Firstenberg-Eden & Klein (1983), gave correlations of -0.91 to -0.95. As detection time is affected by both bacterial concentration and growth rate, it is likely that the lower correlations were a result of differing concentrations of bacteria with varying gowth rates. A higher correlation would thus be expected in this case, where al1 bacterial levels were prepared fiom a single exponentially-growing bacterial culture. Standard error of the predicted y-values for each x-value was caiculated to be 0.253, demonstrating the repeatability of the assay. Based on both this and the correlation coefficient. subsequent sarnples were tested in duplicate as opposed to triplicate. Initial bacterial levels of <10 CFU/mL were not detected, due to the 1 :10 sample dilution associated with inoculation of wells. Contrastingly, levels of 1O8 were either detected very quickly or not detected at al1 as the bacterial concentration was aiready near to or beyond the threshold required for detection. These values were thus more Iikely to be unreliable. and were deemed beyond the sensitivity of the assay. Based on the information in Figure 5.2.. this impedance assay could reliably detect L. monocyiogenes in a broth system at initial concentrations between 10' and 10' CFU/mL. This improves detection by one log as compared to turbidimetry which had a lower sensitivity of 10' CFU/mL (Chapter Four), but it should be remembered that sample volume was four times larger in the impedimetric assay allowing for an increased inoculum per well. 53.3. Effect of phage on impedimetric detection Listeria phage B 1 was added to module wells containhg its propagating strain L. monocyfugenes ATCC 23074 to assess its impact on detection. Figure 5.3. graphicdly illustrates the delay in detection time associated with phage activity. In this case the detection of an initial bacterial concentration of 1x 1o6 CFU/mL was delayed by exactly 7 hours, before an IDT of 14.4 hours was determined. However, the capacitance cuve retained its characteristic form in the presence of phage with a similar % signal change to that observed for the control. The difference between sample detection times in the presence and absence of phage is demonstrated in Figure 5.4. In this case, the time delay between detections was considerably larger (with the exception of the 10' CFU/mL data) than previously demonstrated for turbidimetric assays (Table 4.3.) but equally hi& correlation coefficients of -0.99 (p B 1. It is possible that the larger differences found here are a result of difiering detection thresholds for the two methods. Detection of low levels of L. monocyrogenes in the presence of phage indicated either that phage infection did not occur, or that it was lysogenic in nature as a consequence of the high multiplicity of infection. Infected cells would thus remain metabolically active and produce a detection time when initially low levels had reached the detection threshold. minus phage BI: IDT 7.4 Hr > ---- 1 1 plus phage 61: IDT 14.4 Hr ----. O 4 8 12 16 20 24 28 32 36 40 44 48 Incubation Time (Hours) Figure 5.3. Effect of phage B1 on the capacitance curve for I. monocytogenes ATCC 23074 growing in Trypticase Soy broth 10 20 30 40 Mean lmpedance Detection Time (Hours) Figure 5.4. Effect of phage B1 on IDT9s for L. monocytogenes ATCC 23074 182 5.3.4. Effect of media selectivity on impedimetric detection The use of TSB with the addition of Oxford supplement (SEL) was studied in Chapter Four, and was demonstrated to improve the turbidimetric detection of L. monocyfogenes in combination with Listeria phage B 1. Its use was therefore also considered for impedimetric detection of L. monocytogenes. Data for mean detection times of bacterial dilutions tested in the presence md absence of phage, in both TSB and SEL broths are tabulated in Table 5.1. Similar to the trend observed using turbidimetric analysis (Figure 4.3.) impedimetric detection times for L. monocyrogenes in the absence of phage were slightly shorter in the selective SEL broth than in TSB. As with turbidimetry, increased selective pressure \vas responsible for slowing down the metabolic rate of the bacteria, resulting in an increase in the time required to reach threshold detection levels. Al1 tested bacterial concentrations in the absence of phage were detected within 24 hours as they were within the previously detined detection limits of the assay. In the presence of phage, bacterial concentrations grown in TSB demonstrated detection delays of 7.3-12.5 hours, related to initial infection and lysis of a proportion of listeriae. Bacterial levels as low as 10 CFUImL were detected in this case. as opposed to turbidimetric analysis which failed to detect an initial level of 1 O' CFU/mL. This was again due to the larger sarnple size and correspondingly higher inoculurn possible in the impedimetric assay. When the increased selectivity of SEL broth was combined with Listeria phage B 1 however, impedimetric detection times Table 5.1. Effect of increased media selectivity on phage-rnediated impedimehic detection of L. monocytogenes ATCC 23074 in pure culture - -- IrrMU detection time murs) II Listeria 1 concentration I I I 6 7.7 1 15.0 8.0 25.0 5 10.9 18.5 10.9 30.6 4 13.8 22.0 14.6 37.0 3 16.9 26.6 18.2 42.2 2 19.5 28.9 2 1.4 N.D. 1 22.4 34.9 24.2 N.D. TSB Trypicase Soy Broth SEL TSB plus Oxford selective supplement N.D. Not detectable within 48 hours ; increased dramatically. Initial bacterial concentrations of 10' to 1O6 were detectable within 25-42 hours, with detection delayed by 17-24 hours as compared to detection in the absence of phage B 1. Initial concentrations of 1@ and 10 CFU/mL were not detected within a 48 hour incubation period, confirming that the bacteria initially detected were listeriae. Bacterial capacitance curves were similar for sarnples grown in the presence and absence of phage (data not shown), and in an increasingly selective environrnent (SEL). In al1 cases a sharp acceleration of the capacitance signal was observed corresponding to bacterid metabolism, and detection times were accurately assessed as a result. SEL was thus considered to be appropriate for subsequent detection of L. monocytogenes in raw milk. Based on turbidimetrîc data, it has been demonstnted in Chapter Four that L. monocytogenes has a slower growth rate than both E. coli and S. Typhirnurium in TSB. Figure 5.5 demonstrates a similar phenomenon observed for impedirnetric detection of each of these bacterial species, this time in differential (CM) and semi-selective (SEL) media. Both E. coli G2-2 and S. Typhimuriurn demonstrated similar growth rates wiîh calculated dope values of -0.90 and -0.87 respectively. The slope value for L. rnonocyfogenes was -0.303, illustrating a slower growth rate than the previously mentioned bacteria. The correlation coefficients for impedimetric detection of different levels of E. coli, S. Typhimurium, and L. monocytogenes were -1.00, -0.99, and -0.99 O 5 10 15 20 25 30 Mean lmpedimetdc Detection Time (Hours) Figure 5.5. Cornparison of IDT's for E. cdG2-2, L monocytogenes ATCC 23074 and S. Typhimurium ATCC 14028 respectively. demonstrating once again the strong predictor value of IDT. 5.3.5. Detection of L. monocytogenes in raw milk Raw milk was deliberately temperature-abused to a level approaching the detection limit of the impedimetric assay to evaluate the combined use of SEL broth and Listeria phage B 1. previously determined to be most effective in combination. This level of non- listeriae was described by Blivet et al. (1998) as a "hard" test condition, and was suitable for adequately testing the selectivity of the test medium. Detection thes obtained for L. rnonocytogenes levels of 10' to 106CFU1m.L in both TSB and SEL, in the presence and absence of phage, are shown in Table 5.2. It was apparent fiorn the very short detection times obtained for artificially contaminated milk and milk controls that the background microflora count was 1o6 CFU/mL or greater. As previously demonstrated both in Chapter Four (Figure 4.3.) and in this chapter (Figure 5.4.) the effect of phage on detection times for higher bacterial levels was less pronounced than at lower bacterial levels. Indeed, in this case where al1 saimpie counts were close to the detection threshold, phage had no effect on detection times when used in conjunction with TSB. Using SEL however, delayed detection times ranging from 2.2-8.8 hours were observed for sarnples with L. monocytogenes levels of 1O4 to 1O6 CFUlrnL. Below these levels, Table 5.2. Effect of media selectivity on phage-mediated impedimetric detection in raw rnük artificially contaminated with L rnonocytogenes ATCC 23074 Lister ia Mean detection time (hours) concentration CFU/mL ) 6 5 4 3 1) Milk control L - SEL TSB plus Oxford selective supplement N.D. Not detectable within 48 hours detection times were sirnilar to that observed for milk controls, indicating that the background flora eventually outgrew the lower levels of Listeria present and the detection threshold was reached. Increased selectivity alone was thus capable of slowing the growth of the background microflora sufficiently to allow higher levels of Listeria to be detected. As in al1 previous experirnents, initial bacteriai levels of 10' CFU/mL were not detected. With the addition of phage to increase the selective pressure, initiai levels of L. monocytogenes between 1o4 and 106CFUIrnL were again detectable, despite the high background microflora with detection delayed by 0.9-4.4 hours. Below this, the addition of phage B 1 did not delay detection times as the background Bora was suficientiy high to reach a detection threshold level alone. At this point, levels of Listeria were outnumbered by approxirnately 2: 1 or mure, making it more physically dificult for phage and host ce11 interaction to occur. In addition, the increasing multipliciiy of idection between phage and target cells would likely influence the direction of the phage B t infection cycle towards lysogeny. This would prevent the lytic destruction of the host population, and given the slow growth rate of this particular strain of L. monocytogenes, would not have an overail effect on detection times. The combined use of SEL and phage B 1 was thus demonstrated to be usehl in detecting and confirming the presence of L. monocpogenes at initiai levels r 104 CFU/rnL, in raw milk with a background flora of 1O6 CFU/mL. The selectivity of SEL was not sufficient to completely inhibit the microflora of raw mik, although it did improve detection as compared to raw milk grown in non-selective TSB. The activity of phage B 1 was also usehl in confming the presence of Listeria above initial levels of 1O3 CFU/mL. At counts 1 O3 CFU/mL however, the presence of Listeria could not be confmed, and this was a direct result of the high background microflora. Hancock el rd. (1 993) reported the use of a glucose-emiched nutrient broth with added inhibitors, and detedned the lirnit of detection to be 10' CFU/g of L. monocytogenes in various spiked cheeses. Inhibition of Gram-positives was achieved in their study by the use of proflavine hydrochloride (21 mg/L) while Gram-negatives were inhibited by rnoxalactarn (20 mg/L). Although this SEL-phage combination was not tested with lower count raw milk samples, it was used to detect L. monocyfogenes in pasteurised milk (data not shown) with a corresponding detection limit of 10' CFUImL. The usefulness of such a detection system will thus be dependent on the levels of the pathogen to be detected, the background flora, and the efficiency of the phage empioyed. The levels of L. monocytogenes found in raw milk are typically low (40 CFU/mL) as compared to the total indigenous population (Ryser, 199 1 ), and would not be detectable by this phage- based impedimetric method. Further work is therefore required in this area to establish more optimum selection conditions, including pre-enrichment, based on the types of food samples to be tested. 5.3.6. Detection of Salmonella spp. in skim milk powder Salmonella phages Felix 0-1 and SJ2 were demonstrated in Chapter Two (Table 2.4.) to 190 have different infection spectra, with Felix 0-1 infecting most of the tested isolates while SJ2 infected only serovars Enteritidis and Typhimurium. S. Braenderup was chosen as it was susceptible to Felix 0-1 infection, but not SJ2. Isolates were tested individually and in combination to evaluate the ability of these phages to differentiate isolates. Table 5.3. demonstrates the use of CM to detect both serovars of Salmonella at a concentration of 10' CFU/g within 10.7 hours. Uninoculated skim milk powder was also tested and detection times were not produced within a 24 hour penod. This was aiso confirmed by a negative plating result on MacConkey agar. The presence of phage Felix 0-1 resulted in a delay in detection times of 0.9-1.3 hours for the isolates individually tested and in combination. The magnitude of this delay was also reported by Pugh & Amott ( 1989) in their irnpedimetric assay for Salmonella. Only the S. Enteritidis isolate was undetectable by the impedimetric assay afier the addition of phage SJ2. S. Braenderup produced detection times comparable to the control isolates minus phage, while the combination of isolates took 2 hours longer to detect. In the case of S. Enteritidis, which was already demonstrated to be sensitive to phage SJ2, non-detection was due to phage activity keeping initial levels of Salmonella below the detection threshold. However, in the case of S. Braendenip, which was not sensitive to phage SJ2, a detection time similar to that obtained in the absence of phage was observed. This indicated that bacterial growth was not afTected by phage SJ2, and the detection time was produced by an isolate other than S. Enteritidis (or S. Typhirnurium). With both Table 5.3. Influence of phage specificity on impedimetnc detection of SuIlilonelfa serovars Entetitidis and Braendemp in skim milk powder Skim milk Inoculum - 1 Felix 0-1 SJ2 I L II S. Enteritidis ( 10.7 ( 11.7 1 N.D. S. Braenderup 10.5 11.4 10.7 S. Enteritidis and S. 10.5 11.8 12.5 Braendemp Skimmilkpowder N.D. N.D. N.D. ND. Not detectable within 24 hours isolates present, a delayed detection tirne of 1.3 hours was observed when they were exposed to phage Felix 0-1. In the presence of phage SJ2, detection was delayed by 2 hours as compared to the control in the absence of phage. This can be explained by the effective action of phage SJ2 to keep S. Enteritidis at levels where it did not contribute to the impedimetric detection threshold, while S. Braenderup was able to increase in hours as compared to the control in the absence of phage. This can be explained by the effective action of phage SJ2 to keep S. Enteritidis at levels where it did not contribute to the impedimetric detection threshold, while S. Braenderup was able to increase in numbers uninhibited. Detection took longer than the control, based on the lower level of S. Braendemp originally present in the inoculum (jx 1O' CFU). This research demonstrates that phage Felix 0-1 is able to detemiine the presence of both isolates alone or in combination, while phage SJ2 can differentiate between Enteritidis and non-Enteritidis isolates in skim milk powder. 5.3.7. Cornparison of colorimetric and impedirnetric detection of E. coli 5.3.7.1. Interpretation of endpoint analysis data The use of endpoint detection as a simple means of anaiysing colorimetric data was examined using the LabSMART software package. A typical b* profile for the growth of E. coli G2-2 in CM is demonstrated in Figure 5.6. After approximately 6 hours of CM plus E. coli I C. 0246 8 IO 12 14 16 18 20 22 Incubation time (hours) Figure 5.6. Influence of endpoint selection values on colorimetric detectioo of E. coli G2-2 incubation at 37"C, b* values start to ascend rapidly (similar to the turbidimetry log phase demonstrated in Figure 4.1 .) as the pH of the medium changed corresponding to growth of bacteria. In this case, a peak b* value of approximately 27 was reached within 17 hours of incubation. Sirnilar to the use of detection time as an analysis parameter, the endpoint (EP) time can be detemiùied by the amount of time required for bacterial growth to increase the initial b* value by a set arnount. Using x = 5, 10, and 15, endpoint times of 6, 7, and 8 hours would be obtained from this colour profile. However. what must be taken into account when establishing an appropnate endpoint is the fluctuation or drift associated with the colour change. In this particular scenario, for the control sample, approximately a 10 unit signal drift occurred which did not correspond to a purple to yellow colour change upon visual inspection of the microtitre plate. Using an x value of 5 or 10 units, this sample would have been deerned positive with endpoint times of ! 1 and 20.5 hours respectively. Using x=lj as a cut-off point, this sample would more correctly be assigned an endpoint time of zero hours (Le. not detected by endpoint analysis). Thus the interpretation of endpoint values would be negative for coliforms (Exarnple 1a), but positive in Exarnple 1b (Table 5.4). 5.3.7.2. interpretation of phage-based endpoint information A number of potential scenarios exist regarding interpretation of results when phage is incorporated into colorimetric endpoint detection assays (Figure 5.7.). Where coliforms 195 Table 5.4. Examples of interpretation of colorimehic endpoint data in the absence and preseoce of bacteriophage Endpoint The (Hours) Example No. ' CM Interpretation Minus phage Plus CoIour phage la 0.0 Purple Coliforms absent 1b 8.6 Yellow Coliforms ~resent 2 5.0 Yeilow Coiiforms present 5.0 Yellow E. coli absent 3 2.9 Yellow Coliforms present 8.2 Yellow E. coiï present 4 7.9 Yellow Coliforms present 0.0 Pumle E. coli present - - + phage AT20 b*+l O w-*--* .,,,,,,,,,,,,,--,,,------*------Q---.--*------.------.------*---*-*-*-* ------a... a@ a b*+5 --*------*------.---*-*---*-*------.------.------b* QQQ Q..." --*--P. ------.------*-----*---.----..-.--.------m - - O 2 4 6 8 10 12 14 16 18 20 22 Incubation time (hours) Figure 5.7. Influence of phage AT20 and endpoint selection values on colorimetric detection of E. coli G2-2 have been positively detected by endpoint analysis and associated media colour changes (Table 5 -4.) no di ffierences between control (-phage) and test (+phage) endpoint values would indicate that E. coli is not present in the sample (Example 2). However, either a larger endpoint (Example 3) or zero value (Example 4) in the presence of phage would indicate that al1 or a proportion of the coliform population is represented by E. coli. Employing endpoint analysis as descnbed above, the detection potential of phage AT20 was evaluated in the coIo~rnetricdetection of E. coli G2-2 in artificially contaminated raw milk and minced beef samples. Preliminary studies in pure culture were not performed as the initially clear media solutions did not give adequate reflectance for b* values to be measured. 5.3.7.3. Colorimetric and impedimetric detection of E. coli in raw milk Data collected for the colo~metricdetection of E. coli G2-2 in raw miik is shown in Table 5.5. Estimated numbers of E. coli were calculated for each sample and tabulated along with observed media colour changes and endpoint times generated using selected x values. Using an x value of 5 to establish endpoint detection, positive results were indicated for both the raw milk control and milk samples artificially contaminated with E. coli G2-2. Increasing x to 10 units, milk control detection became negative, corresponding to the Table 5.5. Colorimetnc endpoint detection of E. coii 62-2 in raw milk using Coliforrn Medium (CM) and phage AT20 Purple Endpoint @ours) to EP = b* + x, where: Yellow colour 1LqzQzF change? 1 ;- - - Milk ctrl. Yes 0.40 3.70 4-00 Yes ( 0.40 1 4.60 1 540: Yes 1 0.40 1 6.00 1 &60 Milk ctrl. Yes 1 0.40 1 1.O0 1 11-90 Yes 10.40 1 7.40 116.60 Yes 1 0.40 1 5.60 1 8.20 absence of colour change visually observed. Al1 remaining samples continued to give positive detection results. When the effect of phage AT20 addition was taken into consideration however, the x = 10 value was deemed inappropriate for accurate detection as a number of samples with positive endpoint detections did not demonstrate colour changes when visually inspected. .eialysis of scatter plots for milk controls reveaied that the fluctuation of b* values for negative samples was approximately 12 units (data not shown). When the endpoint was reassessed using an x value of 15, the previously positive results became zero endpoint values indicating that growth and colour change had indeed not occurred. Regression analysis of data points in the absence of phage gave a predictor r value of -0.99, indicating the ability of endpoint analysis to accurately predict initial differences in bacterial Ievels. There were a couple of outlying data points, at the 2.5 x 10' (minus phage) and 2.5 x 1O6 (plus phage) CFU/mL levels, probably caused by dilution baccuracies, but in both cases they were positively detected and confïrmed by phage AT20 activity. As the b* value increased quite considerably in this microtitre plate system, endpoint detections remained accurate at .u values of 20-3 0 units. At a 3 5 unit increase, some levels of E. coli were deemed undetectable as the x value exceeded their maximum b* value. Based on endpoint detection, using x = 15, the estimated sensitivity of detection of E. di 200 G2-2 in raw milk was < 2.5 CFUfmL, detectable within 10 hours, with phage AT20 able to subsequently confidetection in al1 cases. This is in agreement with detection Iimit estimates of 4 CFUImL for dairy products (Richardson et al, 1988), but this is typically achieved using a pre-enrichment step pnor to analysis (Richardson et al., 1994). It was not expected, based on the small sample size tested, that such an extremely low detection rate would be achievable. Indeed, it was anticipated that the detection limit would be similar to that of the Bioscreen. However, two factors should be considered here. First, an exponentially growing pure culture of E. coli kvas employed in these experiments, giving it a competitive advantage over the raw rnilk rnicroflora. In addition, the counts at lower levels were estimates, and subject to variation during the preparation of serial dilutions. It is therefore possible that these were underestimates of the E. coli population present. Indeed, the similarity of detection times between sampies at 10' and 10' CFU/mL would indicate that the lower levels of bacterial contamination used here are probably a log higher in value. Regardless, this colorimetric method was successful in detecting low levels of E. coli G2- 2 in raw milk within a short testing period. Confirmation of E. coli detection by phage AT20 was very effective. with al1 sample detections in the presence of phage occumng later than in al1 the test samples without phage. Based on these results, using a phage specific for E. coli, its presence could be confirmed within 15 hours of a positive coliform test. Table 5.6. compares the results obtained for colorimetric detection of E. coli in raw milk with impedimetric detection times generated for the sarne samples. The detection data for both methods were ver-similar (r = 0.99; p<0.001) although the impedimetric detection times were slightly shorter. This was due to the larger sample size (100 pL versus 25 pL) and thus siightly higher inoculum levels in module wells. Regardless, detection Iimits were almost identical, indicating that either method is suitable for phage-mediated detection of host bacteria in raw mik. 5.3.8. Phage-based colorimetric detection of E. cofi in raw minced beef Colorimetnc detection of E. coli G2-2 in raw rninced beef using CM and phage AT20 was also investigated using the LabSMART system (Table 5.7.). In contrast to the data obtained for E. coli detection in raw mik using a microtitre plate format (Table 5.5.) b* colour value increases were much smaller in this tube-based system. With inoculated samples, visible colour changes were oflen not detected using an x value of 10 units, and ail but two samples gave zero endpoint times when analysed using a b* increase of 15 units. This meant that data had to be analysed using an x value of 5 (values greater than 5 but less than IO were not available within the software). Using scatterplots of data fiom control sarnples, a signal fluctuation of around four b* units was observed (as opposed to twelve units for the raw milk plate assay). This irnmediately suggested that fluctuations in the b* signai could be problematic in data analysis. Table 5.6. Cornparison of impedimetric and colorimetric data for detection of E. col, E. coli (CFUImL) Colour IDT Colour change @ours) change 7 Milk ctrl. No N.D. No N.D. 2.5x108 Yes 1 .O Yes 2.5x107 Yes 1.4 Yes - 1 1 1 2.5~1O6 Yes 2.3 Yes P 2.5~1Os Yes 3.3 Yes H A 2.5x104 1 Yes 1 4.0 1 Yes G - E 2.5x103 1 Yes 5.2 1 Yes 2.5~1O' Yes 6.2 Yes 2.5~10' Yes 7.5 Yes 2.5x10° 1 Yes 1 8.5 1 Yes Yes Mitk ctrl. 1 No 1 N.D. 1 No N.D. Yes 1 7.0 1 Yes + Yes No N.D. No N.D. P 1 1 W 1 N.D. 1 No N.D. A G No 1 N.D. 1 No N.D. E N.D. 2.5x101 1 No N.D. No N.D. 2.5~10° No N.D. No N.D. - ~2.5 1 No 1 N.D. 1 No N.D. ND. Not detectabIe within 24 hours Table 5.7. Colorimetric endpoint detection of E. coli 62-2 in niw minced beef using ~olifo&Medium (CM) and phage AT20 Endpoint (hours) EP=b*+x where: - - Beef ctrl. Yes ,13.00 0.00 0.00 4x10~ Yes 5-00 6.50 0.00 4x 1 o6 Yes 4.50 6.50 0.00 4x 10' Yes 5.75 8.25 0.00 I 4x 1O4 Yes 7.00 1 8.75 0.00 1 4x10) 1 Yes 1 8.75 1 0.00 1 0.00 4x 1O' Yes 10.25 0.00 0.00 4x10' Yes 12.25 0.00 0.00 4x10' 1 Yes 1 14.50 1 0.00 1 0.00 - - - <4 Yes 15.75 0.00 0.00 Beef ctrl. Yes 13.00 0.00 0.00 ~ 4x 1O' Y es 4x 1O6 Yes 4x 1O* Yes 4;; 1o4 Yes 4x10) Y es <4 1 Yes 1 12.50 1 0.00 1 0.00 234 Endpoint values were likely lower as a result of the less homogeneous suspension of ground beef in CM as compared to the more consistent and fïner particulate rnik matrix. The reflectance of ground beef suspensions would thus be less than that of milk, and b* values would be lower as a result. The effect of using microtitre plates versus test tubes was not assessed for the different food commodities, but tube-based research by Richardson et al. (1988) demonstrated variable colour values depending on the product rested. suggesting that this is a more important factor than the test format. Analysing first the colour data obtained in the absence of phage AT20, an x value of 5 colour units resulted in endpoints for al1 samples. These were in agreement with visual colour assessment, and a correlation coefficient value of -0.98 (~~0.05)was obtained when endpoint data were compared to E. coli plate count estimates. Non-inoculated minced beef used as a control also gave a positive coliform result by both visual inspection and endpoint analysis. Comparing these results to data generated for the sarne samples in the presence of phage, interpretation of data was more complex. In the presence of phage, control beef samples also produced a positive coliform result, with endpoint detection at the sarne time as controls minus phage. For samples artificially inoculated with E-coli G2-2 at levels of 4 x 10' to 4 x 10' CFU/g, endpoint detection occurred later in the presence of phage. Differences between endpoints for samples in the presence and absence of phage increased as the E. coli concentration decreased. This was likely due to the combination of decreasing numbers of E. coli due to phage activity and competitive growth of background coliforms. As the initial inoculum of E. coli G2-2 was reduced so the phage was able to keep these numben below threshold detection levels, and thus the background Bora began to play a more major role in producing an endpoint detection. It should also be remembered that target E. coli suspensions used in these expenments were obtained from a healthy exponentially-growing culture. These cells in optimum metabolic condition would likely have a competitive advantage over the indigenous beef microflora. Thus detection wodd become progressively slower as fewer healthy E. coli contributed to changes in b* values. It would be expected in such a case that eventually the sarnples would produce simiiar detection times to that of the control, as only the background microflora would be involved in producing colour changes. This was the case with estimated levels of E. coli at <4 to 40 CFU/g. However, an initial level of 400 CFU/g of E. coli did not produce a detection time in the presence of phage, aithough the sarne sample was positive in the absence of phage. This could be due to an uneven distribution of cells within food clunips in the stomached sarnple, resuiting in their inclusion in one medium tube but not in another. As far as possible, pipetting of sarnples into test tubes attempted to produce representative samples, but it was more dificult to standardise with such heterogeneous particulate matter. Additionaily, uneven distribution of cells could have resulted in the presence of indigenous bacteria sensitive to phage AT20. Screening of the background microflora would be required to confirm this. Based on these results, the sensitivity level of this test was estimated to be between 4 x 206 10' and 4 x lo3 CFU of E. colilg of minced beef, detectable within 16 hours of incubation. Correlation kvas better than that achieved for the determination of total bacteria in minced beef using a redox dye (0.80 to 0.82), but with a comparable time required to achieve results (Manninen & Fung, 1992). Although sensitivity was slightly higher than that observed for raw milk, due to increased background microflora levels, the combination of Coliform medium and phage AT20 was again successful in developing a specific colorimetric method for detection of E. coli G2-2. The only dificulty encountered was the decreased reflectance attributed to the sample suspension. which necessitated the use of a lower b* value for endpoint analysis. However, fluctuations in signal were also srnall and in this case seemed not to cause fdse-positive results. It would nonetheless be desirable to increase the reflectance to improve the b* colour levels. This could be achieved by more effective sample preparation, such as blending, to create a more homogeneous rnatrix with uniform distribution of bacteria. This would also allow the phage greater access to potentially susceptible bacteria. Additionally. a controlled increase in reflectance could be facilitated by the incorporation of boiling chips or solutions of metal oxides in media tubes (Personai communication with Dr. G. Richardson). This would also allow the use of this technique for pure culture work, which was not possible in this case. The potential application of this technology for detection of other pathogens will likewise rely on media selectivity, along with the incorporation of a pH indicator. Shelef and Eden (1 996) have demonstrated the use of selective media using the Biosy sTMoptical instrument, which when coupled with redox indicators (resanirin), or pH indicaton (brorncresol purple) can detect for exampie L. monocytugenes in the presence of Salmonella Enteritidis. The potentiai also exists for multi-testing, where different reactions cmbe monitored based on the combined use of diflerent colour indicators (Shelef & Eden, 1996). CHAPTER SIX CONCLUSIONS Bactenophages were first scientificaliy acknowledged 83 years ago. In the period since their discovery. the study of bactend vimses has been responsible for many advancements in the field of molecular biology. The recognition of their innate specificity for their bacterial hosts has prompted their application in a varîety of rnicrobiological endeavours including bacterial typing and cloning. Not surprisingly, their specificity has also been exploited in numerous techniques for the detection of food- borne pathogens, outlined in Chapter One. Approaches to phage-based detection have been of a predominantly molecular nature, included luminescence, ice nucleation activity, and cloning of enzymes responsible for lysis of host cells to release progeny phage. However, the requirernent of phage for metabolically active hosts in order to replicate prompted the decision in this study to use metabolic assays for phage-based detection tests including turbidimetry, colorimetry and electrical irnpedance. In addition. the specific lytic nature of phages was exploited for use in the ATP bioluminescence reaction as a biological extractant capable of selectively infecting target bacteria. Given that the bioluminescence method is non-specific when chemically- mediated lysis is employed, it was proposed that the use of lytic phages could improve the specificity of the assay. Chapter Two focused on the characterisation of selected bacteriophages specific for the genera Escherichia, Salmonella, and Listeria, specifically to determine their potential suitability in phage-mediated detection methods. Salmonella phage SJ2 and E. coli phage AT20 were both isolated fiom environmental sources, while Salmonella phage Felix 0-1 and Listeria phage ATCC 23074-8 1 were obtained from commercial supplien. Phages were evaluated by their performance in bacterial typing (including plaque morphology) and their classification by transmission electron rnicroscopy (TEM). Listeria phage B 1 was determined to have a broad host range as it was capable of infecting isolates belonging to al1 five tested species of the genus. The overall typability of isolates using this phage was 57.6%, which was within published ranges for other Listeria phages. However, it was found to be more successful for typing of L. innocua (80%) and L. welshimeri ( 100%) than for L. monocyfogenes (30%), the only recognised human pathogen of the genus. This suggests that it may have greater potential use in detection of indicator listeriae as opposed to direct detection of the pathogen. In addition. infection of susceptible cells by phage B 1 was demonstrated to be serovar-independent, confirming previous reports in the literature. This suggests that the phage receptor is different fiom the O antigenic determinant. The use of TEM identified phage B 1 as a lysogenic syphovirus, confirmed by the production of small turbid plaques following its exposure to the propagating host L. monocytogenes ATCC 23074. Snlrnonella phages Felix 0-1 and SJ2 demonstrated contrastingly different infection patterns. Feiix 0-1 was capable of infecting al1 but one of the group B, C, and D, isolates tested, with a 95.5% typing rate comparable to previously reported estimates. Phage SJ2, however infected only isolates identified as S. Typhimuriurn or S. Enteritidis. Both phages were identified as lytic rnyoviruses based on morphology and clear plaque formation. Information available regarding receptor sites of other Salmonella phages lead to the speculation that the differences in host specificity are related to the composition and length of the O-side chain of lipopolysaccharide, and to the involvement of a secondary unknown receptor. The host specificity of E coii phage AT20 was investigated by assessing its ability to infect verotoxigenic and non-toxigenic isolates. Interestingly, al1 non-toxigenic isolates were observed to be susceptible to infection by the phage, while al1 toxigenic strains were apparently resistant. As the verotoxigenic phenotype is only produced following lysogenic conversion, it cm be speculated that al1 verotoxigenic isolates harboured a prophage which rendered them immune (as opposed to resistant where the receptor is absent or mutated) to superinfection by other phages. Further testing of isolates would be required to confirm that immunity was responsible. Phage AT20 was also identified as a lytic T4-like rnyovirus, exhibiting the largest plaque morphology of al1 four phages tested. Based on the observations made in Chapter Two. it was concluded that phages Felix 0-1 and B1 were useful only for species-specific detection assays, while phage SJ2 (and to a lesser extent. AT20) could potentially be used for more specific detection purposes. The use of bacteriophages as specific lytic agents for inclusion in ATP bioluminescence- based assays was evaluated in Chapter Three. The release of cellular ATP corresponding to phage-mediated infection and lysis of susceptible bacteria was compared to ATP levels in bacterial suspensions in the absence of phage. Bioluminescence was monitored over time to evaluate the relative abilities of phages B 1, AT20, and SJ2 to identifi their specific propagating host. Resuiting profiles indicated that the bioluminescence response was related to the phage- host combination tested; specificaily the mode of infection of the phage. In the case of L monocytogenes, peak ATP readings were obtained after 30 minutes of exposure of the bacteriai host to its specific phage B 1, followed by a steady decline in ATP levels over the next 1.5 hours. In contrast, infection of E. coli G2-2 with phage AT20 resulted in an ATP profile that began to increase after 30 minutes of incubation had expired, and continued to increase for the two hou duration of exposure. The combination of S. Ententidis and phage SJ2 resulted in detectable ATP release after 30 minutes, which peaked at 90 minutes and remained stable for a Mer30 minute period. Detection limits for t!he phage-mediated assay in TSB were estimated to be >106 CFUIrnL for al1 three phage-host combinations. The effect of competitive flora on the ability of phages to infect host cells was also evaluated. In al1 cases, an increase in ATP levels attributed to phage lytic activity was observed when counts for target bacteria were twice that of the contaminating non-target species. In terms of ATP release by lytic activity, Listeria phage B 1 produced the highest increase in ATP levels, followed by E. coli phage AT20, and then S. Ententidis phage SJ2. ïhe poor sensitivity of the assay prompted the evaluation of three alternative means to improve detection limits. i) For E. coli, tested at a bacterial level one log lower than the detection threshold, increasing the length of incubation to 3.5 hours effectively increased the arnount of total ATP released by a factor of almost four. ii) The use of filtration to concentrate Listeria cells prior to exposure to phage however, did not improve sensitivity. This was probably due to large numben of cells being trapped in the filter rnatrix and thus inaccessible to phage as compared to bactena in suspension. Thus, filtration using the particular filter system tested was concluded to be ineffective. Nevertheless, investigation of alternative filter matrices with larger surface areas could prove to be more useful in improving sensitivity. iii) Finally, the ATP amplification method proposed by Hawronskyj et d (1 995) was explored as a means of increasing ATP levels for improved detection sensitivity. Following a chernical extraction to release ATP from Listeria cells, the effect of the amplification assay on bioluminescence was evaluated. By comparing individual components, and various combinations, it was observed that only phosphoenolpyruvate (PEP) was responsi ble for increasing ATP bioluminescence outputs. This seemed to be a consequence of naturally occurring amplification components released into the medium afier ce11 lysis. Presence of excess PEP allowed amplification of the ATP released fiorn the cells. It was thus concluded that the proposed ATP amplification system was not a viable option for improvement of the phage-based assay. Further examination of the role of PEP in ATP amplification is required to assess its usehlness in this protocol. Overall, the incorporation of bacteriophages in ATP bioluminescence tests did increase the specificity of detection, but sensitivity was reduced by a level of 2-3 logs. Another alternative for improving sensitivity is the adenylate kinase (AK) assay (Blasco et ai, 1998) which was not examined at this time due to the need for extrernely pure reagents. These researchers also employed a proprietary exmctant to release vesicle-bound AK, an equivalent of which would have to be identified to optimise the sensitivity of the assay. Another possibility could be the use of pre-enrichment particularly for samples containing stressed andlor low numbers of target cells. For analysis of food sarnples, it wodd be usefui to first pre-e~ch,then use a coarse filtration or centrifugation step to remove interfering (and potentially signal quenching) food particdate. A second filtration step to concentrate bacteria followed by a washing step to reduce remaining background ATP could also be usefùl in cleaning up the sample for subsequent phage- based analysis if appropriate filters were identified. Chapter Four explored the potential of bacteriophages to improve the sensitivity of turbidimetric methods for bactenal detection. The limit of detection of the assay in pure culture using TSB was first established to be s 10' CFU/mL. This relatively high detection limit was related to the small sample capacity and corresponding inoculum size (35 PL). A number of parameters for the evaluation of turbidimetric data were compared by linear regression. It was concluded that detection time (the time required for the optical density to increase by 0.2 units) was the most appropriate predictor of initial bacterial levels (r = -0.967). In addition, the rate of growth was demonstrated to be different for each of the diree species tested. E. coli reached the detection threshold quicker thm S. Typhimurium, and these two were respectively growing three and two times faster than L. monocytogenes. The effect of bacteriophage was then evaluated as a means of confkng the specificity of bacteriai detection. The addition of specitic phage was observed to delay detection times for E. coli, S. Typhimurium, and L. monocytogenes, as compared to controls with phage absent. Phage AT20 produced the longest delays in detection, followed by the Felix 0-1 phage, while the shortest delay were observed for Listeria phage B 1. For E. coli and Salmonella phages, it was aiso shown that extension if detection time resulting from the addition of phage increased as the initial bacterial concentration decreased. In some cases host bacteria in the presence of phage were not detectable because the activity of the phage for its own replication kept bacterial levels below the threshold required for both bacterial detection to occur. The effect of contarninants on turbidimetric detection was evaluated. In the presence of E- coli, the growth of listeriae and salmonellae could not be confirmed by the addition of pliage. However, salmonellae could be codïrmed in the presence of listenae. When al1 three species were tested in combination, the growth of Listerin could not be confirmed by phage B 1 addition, but meraddition of phages AT70 and Felix 0-1 could differentiate the presence of their host strain in the combination. This failure of the phage-based detection of Listeria was attributed to the inability of L. monocytogenes to compete effectively in non-selective media with contarninants. Therefore more selective media for the turbidimetric detection of listenae were investigated. SEL (TSB plus Oxford selective supplement) in combination with phage B 1 was demonstrated to have a synergistic effect on detection times. L. monocyfogenes grew rnuch slower in SEL than in non-selective media and this appeared to act synergistically with phage B 1. The combination produced the biggest differences in detection tirnes between samples in the absence and presence of phage. However, this combination increased the detection iimit of the assay to 10' CFU/mL. The more selective nature of this combination was tested in pasteurised milk contaminated with L. monocytogenes at an inflated level of 10' CFUImL. This high level had to be used to take into account the decreased sensitivity in combination with the need to dilute samples at least one hundred times to reduce background turbidity. At an approximate concentration of 35 CFU per well following dilution, L. monocytogenes was detected within 15 hours of inoculation. With SEL and phage B 1 empioyed in combination, detection of L monocytogenes occurred approximately 2 1 hours later. In conclusion, the use of turbidimetry for the detection of bactena in foods is relatively 216 insensitive due to the need to dilute the samples and the small sampie volume allowed in the autornated turbidirneter. However, the use of a semi-selective medium in combination with phage B 1 was demonstrated to be an effective way of improving the specificity of pathogen detection. Further improvements rnight involve a pre-enrichment step. and the elimination of background turbidity in sarnples by centrifugation or clarification techniques. The development of media for use in turbidimetric detection is more flexible than for other rnethods and the use of serni-selective pressure is a possibility, particularly in conjunction with the use of a specific bacteriophage. This concept was also applied to another metabolic detection method -electncal impedance - whose sensitivity was not as limited by sample size or by background turbidity. Phage-mediated colonmetric detection of E. coli in raw mik was compared to impedimetnc detection. Colonmetric detection was also evaluated for the detection of E. coli in raw minced beeE The detection limit of the impedance assay for bacteria tested in pure culture in TSB was initially determined to be 5 10 CFU/mL using a capacitance signal. Based on the larger inoculum size this detection limit is ten-fold lower than that for turbidimetric detection.. Detection time was inversely proportional to initiai bacterial concentration. and a correlation coefficient of -0.987, indicated its suitability in predicting initial bacterial levels. As previously demonstrated for turbidimetric analysis the incorporation of phage into the assay resulted in delayed detection times. However, even at low initial levels of 10 CFU/rnL in the presence of phage B 1, a detection time was obtained. Relative rates of bacterial growth could be predicted by comparing the dopes of regression lines for detection and bacteriai numbers. As previously identified by turbidimetric data, E. coli exibited the fastest growth rate, followed by S. Typhimuriurn, and the slowest growïng L. monocyf ogenes. The suitability of SEL medium for detection of L. monocytogenes was evaluated in the absence and presence of phage B 1. As previously noted with turbidimetric detection times. a synergistic effect was observed when SEL was combined with phage B 1. Ail bacterial levels tested produced a detection in the absence of phage, while growth of listenae at 10 - 100 CFU/mL levels was not detectable within 48 hours when phage was present. The combination of SEL medium and phage B 1 was tested Merby its use in detecting various levels of L. monocytogenes inoculated into temperature abused raw milk (background flora approximately 1O6 CFU/mL). In non-selective TSB, in both the absence and presence of phage, very short detection tirnes were observed as the raw milk microflora reached the detection threshold quickly. Using SEL plus phage, initial Lisierin levels of 1 O" to 1O6 CFU/mL were confïrmed by the action of phage. However, at Listeria counts s IO' CFU/mL, presence of L. rnonocytogenes could not be confirmed by phage addition, because the low numbers of slow growing Listeria was outcompeted by the vigorous background flora. Lmpedimetric detection proved to be a useful means of metabolically detecting and confirming the growth of particular target rnicroorganisms. The application of Salmonella phages Felix 0-1 and SJ2 for the differential detection of serovars Enteritidis and Braendemp inoculated into skirn rniik powder was examined using a conductance signal and coliform medium (CM). At an initial inoculum level of approximately 10' CFU/g, growth of both salmonellae individually and in combination was detectable within 10.7 hours of inoculation. Phage Felix 0-1 caused delayed detection of both strains in al1 samples, while phage SJ2 deiayed only detection of S. Enteritidis. However, phage SJ2 did not infect S. Braenderup, resulting in a detection time similar to the control when only S. Braenderup was present. In addition, the ability of phages such as SJ2 to be more specific in their infectivity thus has potential use in detection methods where combinations of phages could be used to screen positive samples for sensitivity, in much the sarne way as a typing scheme. More isolation and screening of phages is thus required to locate and identifi such desirable viruses. The suitability of reflectance colorimetric rneasurements for the detection of E. coli in raw milk was compared with impedance detection. The same Bactometer coliform medium was used to produce impedance signals and colour endpoint data corresponding to bacterial fermentation of lactose. Both the colorimetric microtitre plate assay and the Bactometer method were capable of detecting < 2.5 CFU/mL of E cdi without phage within 10 hours of inoculation. E. coli levels s 2.5 x 10' CFUIrnL were undetectable in the presence of phage within a 24 hour incubation penod. The correlation between impedimetric and colorimetric data was 0.982, with a siope value of 0.972, indicating the similarity of the two methods. Using a test tube-based format, artificially contaminated raw minced beef was analysed for E. coli using phage AT20. Colour values in this case were smdler than for milk because the more heterogeneous suspension of beef produced less reflectance. However, false-negatives were not obtained although background signal scatter was close to the endpoint value used for analysis. Using this format a detection limit of s 1O3 CFUIg of E. coli in minced beef was achieved. Growth of coliforms was detected within 8.75 hours. and the presence of E. coli G2-2 confied by phage AT20 within 3.5 hours of the positive coliform test. Similar to turbidimetric detection, the use of lab-designed media in this colorimetric assay is more flexible than that for impedimetric detection. nie simple incorporation of redox and pH indicators to selective or semi-selective media can provide usehl growth media for the detection of a variet-of bacteria. Coupled with the use of bacteriophages for confirmation of detection. this is a sensitive and specific method for microbiological analysis of foods. As the four selected bacteriophages were used and evaluated in various detection methods it becarne clear that they al1 have diffenng modes of host infection. The infection pathway of E. coli and Sdmonella phages was determined to be lytic in nature, while the Listeria phage was lysogenic. The lysogenic mode of infection proved not to be very effective as a means of confkming the presence of the host. For exarnple, the bioluminescence assay produced only one apparent lytic infection cycle, indicated by an ATP peak at 30 minutes. This peak was followed by a steady decrease of ATP levels as presumable fewer or no cells were infected and lysed, and ATP was enzymatically degraded. This abrupt halt to ATP release was attributed both to lysogenic immunity of cells. preventing them fiom being re-infected by the sarne phage, and the nature of the balance between bacteriai numbers and viral replication. At hi& phage concentrations in the presence of low levels of target cells, the effect of phage on detection times would likely be negligible. Such a high multiplicity of infection would promote lysogeny, allowing the continued multiplication of the bacterial host. Confirmation of detection by cornparison with an uninfected control would likely not yield a difference in detection times. Convenely, in a situation where higher levels of listeriae were present in a faster growing microflora, the initial infection process could reduce the listend population enough to give the indigenous flora a cornpetitive advantage, after which it too would contribute to growth and detection, producing a detection time ahead of the control. Unfominately, most reported Listeria phages are lysogenic in nature, presumably induced from the chromosome of listerial isolates. However, lytic L. innocua phages have been identified and they may be a better option for such assays assurning that they are similar in specificity. The other disadvantage of this type of assay is the slow growth rate of Listeria isolates. Not only will detection take longer, but pre-enrichrnents will be slow particularly if a high background rnicroflora is preçent. This presents problems in the use of selective media where detection sensitivity could be comprornised as a result of slower growth. E coli phage AT20 appeared to be at the opposite end of the infection spechum when its 22 1 lytic activity was assessed in metabolic detection methods. Considering first the release of ATP fiom bacterial cells, it seemed to be an ongoing occurrence as ATP levels gradually increased throughout the duration of the assay. This cm be explained by the fiequent infection of small numbers of bacteria, slowing down detection at high bacterial levels and maintainhg lower levels below the detection threshold. Suimonellu phages placed somewhere in between phages B 1 and AT20 in terms of their lytic activity. Their bioluminescence profile involved an initial increase in lysis followed by a leveling of ATP indicative of a more balanced viral and bacterial multiplication equilibrium. Indeed, data indicated that this balance was lost only at lower bacterial levels where levels of cells were below the threshold required for multiplication of virus. The presence of a multiplicity of infection would thus keep the bacterial population low for the duration of the experiment, resulting in an inability to detect growth. An understanding of the balance between viral and bacterial multiplication is thus important in the application and interpretation of these phages. In conclusion, lytic phages AT20, Felix 0-1, and SJ2 were identified as more effective biological tools for the confirmation of bacterial detection, while the lysogenic phage B1 had limited usefulness. The slower growth rate of listeriae was also a disadvantage when trying to detect them by metabolic means. These assays have not as yet been tested with stressed bacteria, but it would be important to incorporate a non- or semi-selective e~chmentin order to circurnvent both metabolic injury and low numbers in foods. In terms of the efficiency of metabolic detection methods, both impedance and colorimetry were demonstrated to be equally sensitive methods for detection of viable microoqanisms. Indeed, colorimetry had a slight advantage both in the flexibility of test format and media selection. There are some limitations associated with development of impedimetric media, particulary when high sait concentrations are used (which is a usehl way of selecting for lidenae due to their high salt tolerance). The use of indirect impedance may be the best solution to achieving a balance in terms of specificity and selectivity of detection. Both turbidimetry and bioluminescence were prornising techniques but inherent sensitivity problems limited their application, particularly in food. Regardless of the detection method employed, this research has demonstrated that the choice of bacteriophage ultimately dictates the subsequent success of bacterial detection. For further improvements to these and other phage-mediated detection methods. future research must focus on the issue of phage specificity. In particular, a more thorough understanding of the interaction between phage tail fibres and bactenal receptoe is required to explain differences in phage specificity. Information pertaining to the structure of tail fibres and their corresponding receptor(sj will ultimately lead to specific genetic manipulations of phage tail genes. nie opportunity would therefore be available to alter the structure of tail fibres, such that they subsequently recognise different receptors. This could ultimately lead to improved specificity, where genetic manipulations could be tailored to produce tail fibres recognising, for example, serovar- specific receptors. A continued search for environmentai phages with specific modes of infection and host requirements is also important, as it is a less expensive and time-consuming endeavour than molecular gene manipulations. In addition the occurrence of phages in close proximity to their propagating hosts provides an ideal reservoir in which to search for phages specific for, as an example, more unusual isolates that are not identifiable by other detection systems. As a result, the opportunity to upgrade phage-based methods is always availabie and this could serve to promote phage-based methods as ultimately a more superior means of detecting rnicroorganisms. CaAPTER SE= REFERENCES Ahrned 9 Sankar-Mistry P, Jackson S. Ackermann H-W, Kasatiya SS. &icilbcs cerm phage typing as an epidemiologid twl in outbreaks of food poisonîng. Jounial of Clinid Miaobiology 1 995;33:636-40. Anonymous. Suicide bombing, Badus style. Discover 1995(12):28. Arslan T, Mamaev SV, Mamaeva N.V., Hecht SM. Structudy modifiai &&y luciferase. Effects of amino acid substitution at position 286. Journal of the American Chernical Society 1997;1 19: 10877-87. Arvanitis N, Vargas C, Tegos G, Perysinakis 4 Neto JJ, Ventosa 4 Drainas C. Deveiopment of a Gene Reporter System in Moderately Halophilic Bacteria by Ernploying the Ice Nucleation Gene of Pseudomo~syringoe. Applied and Environmental Microbiology 1 WS;6 1 :3 82 1-5. Baker JM, Gri&iths MW, Collins-Thompson DL. Bacterial Bioluminescence: Applications in Food Microbiology. Journal of Food Protection 1992;s5 :62-70. Barrett TI, Lior EX, Green JH, Khakhria 9Wells JG, Bd BP, Greene KD,Lewis J, Gr& PM-Laboratory investigation of a multistate food-borne outbreak ofEsCherichia wZz 0 157:H7 by using pulsed-field gel ekctrophoresis and phage typing. Journal of Clinid Micro biology 1 994;3 2:3 O 13-7. 225 Bautista D& Mchtyre L, Laleye L, GdEths MW.The application of ATP bioluminescence for the assasment of milk quality adfkctory hygïene. Joud of Rapid Metbods and Automation in Miaobiology 1992; 1 :179-93. Bennett AR, Davids FGC, Vlahodimou S, Banks JG, Betts RP. The use of bacteriophage-based systems for the separaiion and concentration of Sdhonelh. Journai of Appiied Microbiology l997;83:259-65. Bir4 J. A; Easter, M. C.; Gaye Hadfield, S.; May, E.; Stringer, M. FXapid SaIrnonellà detection by a combination of conductance and immunological techniques. Ln: Stannard, C. J.; Petitt, S. B.; Sber, F. A, editon. Rapid Microbiological Methods for Foods, Beverages and Phamaceuticals. Society for Applied Bacterîology Technical Series; UR: Blackweli Scientific Publications; 1989. pp. 165-83. Bishop IR, White CH, Firstenberg-Eden R A rapid impedimetnc method for determining the potentid sheK-Me of pasteurked whole milk. Journal of Food Protection 1984;47:47l-S. Blasco R, Murphy MJ, Sanders MF, Squirreli DI. Spdcassays for bacteria using phage mediated release of adenylate kinase. Joud of Applied Microbiology 1998;84:661-6. Blivet D, Salvat G, Humbert F, Colin P. Development of a new dture medium for the rapid detection of Sàlmonelkà by indirect conductance meaSuTements. Journal of Appüed Microbiology 1998;84:399-403. Block T, Miller R, Komgold R, Jungkind D. A phage-linked immunosorbant system for the detection of pathologically relevant antigens. Biotechniques l989;7:756-61. Blmmberg GV, OToole GA Lugtenberg BJJ, Kolter R Green Fluorescent Protein as a Marker for Pseud'omonar spp. Applied and Environmental Microbiology l997;63 :4543-5 1. Blwd RM,Curtis GDW.Media for 'total' enterobactenaceae, coliform and Escherichia coli. International Jodof Food Microbiology lWS;26:93- 1 15. Blouin K., Waiker SG, Smit J, Turner RFB. Characterization of in vivo reporter systems for gene expression and biosensor applications based on lwrAB luciferase genes. Applied and Environmental Microbiology l996;62:2O13-2 1. Boevink P, Santa Cruz S, Hawes C, Hams N, Oparka KJ. Vis-mediated delivery ofthe green fluorescent protein to the endoplasmic reticulum of plant cells. The Plant Journal 1996;10% 5-41, Bolton FI. An imrestigation of indirect conductimetry for detedon of some food-borne bacteria Journal of Apptied Bacteriology l990;69:655-61. Boonrnar S, Bangtralailnonth 4 Pomrunangwong P, Taagima J, Watanak 33, Kaneko K-1, Masuo O. Epidemiological analysis of SaIrnonella idlates from humans and broiler chickens in Thailand by phage typing and pulsed-field gel electrophoresis. Joumal of Clinical Microbiology 1 998;36:971-4. Borrego JJMMq de Vicente 4 Cbrnax & Romero P. Coliphages as an indicator of faecai pollution in water, its relationship with indicator and pathogenic microorganisms. Water Resources l987;21 :1473-80. Brock, T. D.;Smith, D. W.;Madigan, M. T. Biology of Microorganisms Fourth ed. Englewood CWs, N.J. : Prentice-Ha& Inc.; 1984. Bronstein 1, Fortin J, Stanley PE, Stewart GSAB, Kricka U.Cherniluminescent and Bioluminescenet reporter gene assays. Analytical Biochernistry l994;î 1 9: 169-81. Brutton 4 Desiere F, d'AMco N, Guerin J-P, Sidoti J, Huni B, Lucchini S. Brussow H Moleailar ecology of Streptococcus thophilus bacteriophage infions in a cheese factory. Applied and Environmental Microbiology l997;63 :3 144-50. Buriage RS, Yang ZK, Mehlhorn T. A transposon for green fluorescent protein transcriptionai frisions: application for bacterial transport expeximents. Gene 1996;173 :53-8. Busse M. Media for saimonelia International Journal of Food Microbiology 1995;26:1 17-3 1. Carminati D, Neviani E. Application of the conductance measurement technique for detection of Sbep0ctxcu.s ~~~Iivan'usssp. thennophilus phages. Joumal ofDaky Science 199 l;74:1472-6. Carrière C, Riska PF, Zimhony O, Kriakov J, Bardarov S, Burns J, Chan J, Jacobs WqJr. Conditiodiy replicating luciferase reporter phages: improved sensitivity for rapid detection and assessnent of drug susceptibility ofA@cobacterium tubercuJo.ns. Joud of Clinical Microbiology 1997;35 :3232-9. Casper SJ, Holt CA Expression of the green fluorescent protein-encuding gene fkom a tobacco mosaic virus-based vector. Gene 1996; 173 :69-73. Chalfie M, Tu Y, Euskirchen G, Ward WW,Prasher DC. Green Fluorescent Protein as a Marker for Gene Expression. Science 1994;263 :802-5. Chen J, Clarke RC, GrifEths MW. Use of luminescent strains of LicrmomICa enteritidis to moaitor contamination and swvival in eggs- Jodof Food Protection 19%;59:9 15-21. Chen J, Griffiths MW.SzIrnoneZla ddetection in eggs using LdBacteriophageses Journal of Food Protection 19%;59:908- 14. Christensen BB, Sternberg C, Molin S. Bacterid plasmid conjugation on semi-solid hcesmonitored with the green fluorescent protein (GFP) f?om Auporeu victoria as a marker. Gene 1996;173 :59-65. Colquhoun KO, Timms S, Fricker CR Detection of Escherichia cdin potable water using direct impedance technology. Journal of Applied Bactenology 1995;79:635-9. Cormack BP, Valdivia RH, Fallcow S. FACS-optimized mutants of the green fluorescent protein (GFP).Gene 1996; 173 :33-8. Cubitt AB, Heim q Adams SR, Boyd AE, Gross LA, Tsien RY. Understanding, improving and using green fluorescent proteins. Trends in Biochemical Science 1995;20:448-55. Curtis GDW, Lee WH. Culture media and methods for the isolation of Lisien'a monocytogenes. Intemational Journal of Food Miaobiology l995;26:1- 13. d'Hereiie, F. The bacteziophage and its behaviour. Smith, G. H, traaslator. Baltimore: Wiams & Wh;1926. Day M. TB jab deiay rnay kili eighty million. New Scientist 1998; 1 S8(2 l33):23. de Wet Wood KV, Helinski DR, DeLuca M Clonhg of firefly luciferase dlNA and the expression of active luciferase in Escherichia coü. Proceedings of the National Academy of Science USA 1985;82:7870-3. Drainas C, VarthoIomatos G, Panopoulos NI. The Ice Nucleation Gene fiom Pseudomoms syringcre as a Sensitive Gene Reporter for Promoter Analysis in Zymomonas mobilis. Applied and Environmental Microbiology l995;6 1:273-7. Dmggan, P.;Forsythe, S. J. ; Siiiey, P.Indirect impedance for microbial screening in the food and beverage industries. In: Kroli, R G.; Gilmour, A; Sussman, M., editors. New Techniques in Food and Beverage Microbiology. Society for Appiied Bactenology Technical Series; UK: Blackweil Scientific Publications; 1993. pp. 1 15-30. Duckworth, D. H. History and Basic Properties of Bacterial Vises. In: Goyai, S. M.; Gerba, C. P.; Bitton, G., editors. Phage Ecology. New York: 1987. pp. 143. Eaton TJ, Sheannan C& Gasson MJ. The use of bacterial luciferase gaies as reporter gaies in Lactococcils: regdation of the Lacr~~txctislQctis subsp. lacris lactose genes. Journal of Generai Microbiology 1993;139: 1495-501. Elowitz MB, Surette MG, Wolf P-E, Stock J, LeiMer S. Photoactivation turns green fluorescent protein red. Current Biology 1997;7:80% 12. Estela L& Sofos IN. Cornparison of conventional and reversed phage typing procedures for identification of Listeria spp. Applied and Environmentai Microbiology 1993;59:617-9. Estela LA, Sofos JN, Flores BB. Bacteriophage typing of Listeria niomcytogenes cultures isolated fiom seafoods. Journal of Food Protection 1992;55:13-7. Fauquet, C. M. Taxonomy and classification - general. In: Webster, R G.; Granog A, editors. Encyciopedia of Vir010gy. London: Acadernic Press; 1994. pp. 13 96-410. Femer, F. History of V~ology.In: Webster, R G.; Granoff. A, editors. Encyclopedia of Viology. London: Acadernic Press; 1994. pp. 627-34. Firstenberg-Eden R, Klein CS. Evaiuation of a rapid impedimetric procudure for the quantitative estimation of colifomis. Jounial of Food Science l983;48:1307-1 1. Fratamiw PM, Deng MY, Strobaugh TP, Palumbo SA Construction and Characterization ofEscherichia coïi 0 157:W Strains Expressing Fireily Luciferase and Green Fluorescent Protein and Their Use in Suwkai Studies. Joud of Food Protection l997;6O:1 167.73. Fryer, S. M.; Forde, K.Eiectnca1 sneening of powdered dairy products. In: Stannard, C. J.; Petitt, S. B.; Skinner, F. A, editors. Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticals. Society for Applied Bacteriology Technid Series; UK: Blackwell Scientific Publications; 1989. pp. 143-53. Gailey PC, Miller EJ, Grifnn GD. Low-cost system for red-tirne monitoring of luciferase gene expression. Biotechniques 1997;22:528-34. Goodridge, L. D. A fluorescent bacteriophage assay for detection of ficherichia coïi 0 15 7:H7 in ground beef and raw mik University of Guelph; 1997. Gottesman, M.; Oppenheim, A Lysogeny and prophage. In: Webster, R G.; Gram@ A, editors. Encyclopedia of Vuology. London: Academic Press; 1994. pp. 8 14-23. Graham J. Timely test spots TB kb'.New Scientist 1996;15 1(2M3):Z 1. 233 Grifnths MW.Applications of Biolwninescence in the Dahy Industry. Jodof Dajr Science 1993;76:31 18-25. Griffiths MW. The role of ATP bioluminescence in the food industry. Food Techwlogy l996;50:62-73. Griffiths, M. W.; McIntyre, L.; Sdy, M.; Johnson, IBnumeration of bacteria in milk. In: Stanley, P. E.; Kricka, L. J., editors. Bioluminescence and Cherniluminescence: Current Status-UK: John Wdey & sons; 199 1. Grimm LM, Goldofl M., Kobayashi J, Lewis JH, Alfi D,Perdichim AM, Tarr PI, Ongerth JE, Moseley SI, Samadpour M. Moldar epidemiology of a fm-food restaurant-asociated outbreak of Escherichia coli 0 157:H7 in Washington State. Joud of Ciinid Microbiology 1995;3 3 :2155-8. Gumarson 4 Hurrell B, Thal E. Recent experiences with salmonella-0-1 phage in routine diagnostic work Zentdbl. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt. Orig. 1. Reihe A 1977;237:222-7. Gurian-Sherman D, Lindow SE. Bacterial ice nucleation: signincance and moleailar basis. FASEB Journal 1993;7:133843 . Hancock I, Bointon BM, McAthey P. Rapid detection of Listenu speQes by selective impedimetnc assay. mers in Applied Miaobiology 1993; 16:s 114. Hawronslyj, LM.;Chittock, R S.; Wharton, C. W.; HOMI,JLow level bacterial contamlliation measured wing a novel bioluminescent assay. In: Campbeli, A El; Kncka, L. J.; Stanley, P. E., editors. Bioluminescence and Chedurninescence: Fundamentah and Applied Aspects.UK John Wdey & Sons; 1995. pp. 411-4. Hawronskyj J-M Holah J. ATP: A universai hygiene monitor. Trends in Food Science and Technology 1997; 8 :79-84. Heim R, Cubitt AB, Tsien RY. Improved green fluorescence. Nature 1995;373:663-4. Heim R. Prasher DC, Tsien RY. Wavelength mutations and posttranslational autoxidation of green fluorscent protein. Proceed'mgs of the National Academy of Science USA l994;9 1: 1250 1-4. Hennes KI?,Suttle CA Direct counts of viruses in natural waters and labonitory cultures by epifiuorescence microscopy. Limnology and Oceanography l995;4O:1050-5. Hennes KP, Suttle CA, Chan AM. Fhorescentiy Labeled Virus Probes Show that Naturai VisPopulations Can Control the Structure of Marine Miaobial Comrnunities. Applied and Enwonmentai Microbiology 19%;6 1: 3 623-7. 235 HehgS, Bockehann W, Teuber M. hification and characterisation of the lytic acfivity induced by the proheheaded bacteriophage POO 1 in Lactmtxms ktis. Journal of Applied Microbiology l997;82:233-9. W PJ, Rees CED, Wuwn Stewart GSAB. The application of lux genes. Biotechnology and Applied Biochemistq 1993;17:3-14. HüI PJ, Stewart GSAB. Use of lux genes in applied biochemistry. Joud of Bioluminescence and Chemiluminescence 1 994;9:2 1 1-5. Hirsh DC, Martin LD. Detection of SaIrnonella spp. in milk by using Fek 0-1 bacteriophage and high-performance liquid chromatography. Applied and Environmental Microbiology l983;46:1243-5. Hirsh DC,Martin LD. Rapid deteaion of Salmonella spp. by using Fek 0-1 bacteriophage and hi&-performance liquid chromatography. Applied and Environmental Microbiology 1983;45:260-4. Hof H, Rocourt J. 1s any strain of Listena monaytogenes detected in food a health risk? International Journal of Food Microbiology 1992; 16: 173-82. Holmes, R K; Schmitt, M. P. Bacteriophage t0xi.m and disease. In: Webster, R G-; GranofS A, editon. Encyclopedia of Viology. London: Academic Press; 1994. pp. 1014. Hooper CE, Aasorge RE, Browne HM, Tomkins P. CCD Imaging of luciferase Gene Expression in Single Mamrnalian Ceils. Journal of Biolumiaescence and Cherniluminescence 199û;S: 123-30. Hudson LM,Chen J, W AR, GriflEiths MW. Bioluminescence: A Rapid Indicator of Ewherichiu coli 0157337 in Selected Yoghurt and Ch- Varieties. Jodof Food Protection l997;6O:89 1-7. Irving, T. E.; Stanfield, G.;Hepbum, B. W. T-Electrid methods for water qualit)' testing. h: Stannard, C. J.; Petitt, S. B.; Skinner, F. A, editon. Rapid Microbiologicd Methods for Foods, Beverages and Pharmaceutids. Society for Appiied Bacteriology Technical Senes; UK: BlackweU ScientSc Publications; 1989. pp. 1 19-3 0. Jacobs WJr, Barletta RG, Udani q Chan J, Kaht G, Sosne G, Kieser T, Sarkis GJ, HamGF, Bloom BR Rapid assessrnent of dmg susceptiiilities of~cobacteriuum rubercziIms by means of luciferase reporter phages. Science l993;26û:8 19-22. Jay, J. M. Modem Food Microbiology Fourth ed. New York: Van Nostrand Reinhold; 1992. 23 7 Jmgensen a,Schulz E. Turbidunetric measurernent as a rapid method for the determination of the bacteriological quality of minced meat. International Journal of Food Microbiology l985;2: 177-83. Kallings LO. Sensitivity of various salmonella strains to Felix-0-1 phage. Acta Pathol. Miaobiol. (Scand.) 1967;70:446-54. Kemedy JE, Jr, Oblinger IL, Bitton G. Recovery of whphages nom chicken, pork sausage and dekatessen meats. Jodof Food Protection 1984;47:623-6. Kennedy JE, Jr, Wei CI, Oblinger JL. Distribution of coliphages in various foods. Joud of Food Protection l986;49:944-5 1. Klein D, Indraccolo S, von Rombs K, Amadon 4 Sahons B, Gunzburg WH. Rapid identification of viable retrovhs-transduced ceUs using the green fluorescent protein as a marker. Gene Therapy 1997;4: 1 256-60. Kodikara CP, Crew Stewart GSAB. Near on-line detection of entenc bactena using hrs recombinant bactenophage. FEMS Microbiology Letters 199 l;83 126 1-6. Kyriakides, A L.; Thurston, P. Aconductance techniques for the detection of contaminants in beerdtannard, C. J.; Petitt, S. B.; Skinner, F. A, editors. Rapid Microbiological Methods for Fwds, Beverages and Pharmaceuticais. Society for AppEed Bactenolog-Techaical Series; UK: Blackweli Scientific Publications; 1989. pp. 10 1-1 7. Lederberg J. Srnailex fieas ... ad hfhitum: Therapeutic bacteriophage redun PrOceedin~ of the Natiod Academy of Science USA l996;93 :3 167-8. Leff LG, Leff AA Use of Green Fluorescent Proteh to Monitor SurvivaI of Genetically Engineered Bacteria in Aquatic Emrironments. Applied and Environmental Miaobiology 1996;62:3486-8. Lwin BR, Bull JJ. Phage therapy revisited: The population biology of a bacteriai infection and its treatrnent with bactenophage and antibiotics. American Naturalist 1996;147:881-98. Lewis PJ, Emington J. Use of green fluorescent protein for detedon of d-s-c gene expression and subcellular protein localization during sponilation in Bacillus subtilis. Microbiology 1996;l42:73MO. Li J, Lee T-C. Bacterial ice nucleation and its potentiai application in the food industry- Trends in Food Science and Technology 1995;6:259-65. Lindberg Aq Holme T. Muence of O side chah on the attachent of the Feik 0-1 bactenophage to Salmonella bacteria. Journal of Bacteriology 1 %9;99: 5 13-9. 239 Lwsnier UT. Improved procedure for bacteriophage typiog of Listeria strallrs and duation of new phages. Applied and Environmental Miaobiology 199 1;57:882-4. Loessner MJ. TaxonoMcal classification of 20 newiy isolated Lisferia bacteriophages by electron rnicroscopy and protein analysis. Intervirology 1994;37: 3 1-5. Loessner MJ, Busse M. Bacterio phage typing of Listena species. Applied and Environmentai Microbiology 1WO;56: 19 12-8. Loessner UT, Maier SK, Daubek-Pwa H, Wendlinger G, Scherer S. Three Bru:ilIus cereus Bactenophage Endolysh Are Unrelated but Reveal High Homology to Cell Wd Hydrolases from Different Bacilli. Journal of Bactenology 199%; 179:2845-5 1. Loessner UT, Rees CE, Stewart GSAB, Scherer S. Construction of Luciferase Reporter Bacteriophage AS 1 1: :Id for Rapid and Sensitive Detection of Viable Listeria Cells. App lied and Environmental Microbiology 1996a;62: 1 13 3-40. Loessner MJ, RudoKM, Scherer S. Evaluation of Lucifierase Reporter Bacteriophage AS 11 ::luvlB for Detection of Listeria monocytogenes in Contaminated Foods. Applied and Environmental Microbiology l99%;63 :2961-5. Loessner MJ, Schneider 4 Scherer S. A New Procedure for Efficient Recovery of DNA, RNA, and Proteins fkom Listeria Ceils by Rapid Lysis with a Recombinant Bacteriophage Endolysin Applied and Environmentai Microbiology lWSa;6 1: 1 150-2. Loessner UT, Schneider A, Scherer S. Modified Listeria Bacteriophage Lysin Genes @&) AUow Efficient Overexpression and One-Step Mcation of Biochemidy Active Fusion Proteins . Appiied and Environmental Microbiology 1998;62:3057-60. Loessner MJ, Wendhger G, Scherer S. Heterogeneous endolysins in LiMa monocytogenes bacteriophages: a new class of enzymes and evidence for woserved holin genes withui the siphoWal lysis cassettes. Moleailar Microbiology 1995b;1 6: 123 1-4 1. Mamaev SV, Laikhter AL, Arslan T, Hecht SM. Firefly luciferase: Alteration of the color of emitted light resulting fkom substitutions at position 286. Journal of the Amencan Chernical Society 1996;118:7243-4. Maniloc J.; Ackerman, R-W.; Janis, A Bacteriophage Taxonomy and Classification. In: Webster, R G.; Granoff.A, editors. Encyclopedia of Vuology. London: Academic Press; 1994. pp. 93-100. Manninen MT, Fung DYC.Estimation of microbial numbers fiom pure bacterial cultures and fiom minced beef samples by reflectance colorimetry with OmnispecN 4000. Journal of Rapid Methods and Automation in Microbiology 1992; 1:4 1-58. 24 1 Margantis 4 Bassi AS. Principles and Biotechnological Applications of Bacteriai Ice Nucleation. Critical Reviews in Biotechnology 1991 ; 1 1 :277-95. Marquet-Van Der Mee N, Audurier A Proposais for optimization of the international phage typing system for Listeria momcytiogens: Combined analysis of phage lytk spectnim and variability oftyphg results. Applied and Environmental Microbiology l995;61:303-9. Martin CS, PA, Dobretsova A, Bronstein 1. Dual luminescence-based reporter gene assay for luciferase and /%galactosidase.Biotechniques 19%;2 1: 520-4. Mattila T. Automated Turbidometry - A Method for Enurneration of Bacteria in Food Samples. Jomal of Food Protection 1987;50:640-2. Mattila T, Alivehmas T. Automated hirbidometry for prediciting colony fonning units in raw milk Internationai Journal of Food Miaobiology l987;4:157-60. McClure PJ, Cole MB,Davies KW,Anderson WA The use of automated turbidimetric data for the construction of hetic models. Journal of Industrial Microbiology 1993; l2:277-85. Meighen EA Molecular Biology of Bacterial Bioluminescence. Miaobiological Reviews l99l;S5: 123-42. 242 Meighen EA Bacterial bioluminescence: organiraton, regdation, and application of the lux genes. The FMI33 Jod1993;7: 1016-22. Med Biswas B, Cadton Jensen NC, Creed GJ, Zulo S, Adhya S. Long-circulating bacteriophage as antibacterial agents. Proceedings of the National Academy of Science USA l996;93 :3 188-92. Misteli T, Spector DL. Applications of the green fluorescent protein in ce11 biology and biotechnology. Nature Biotechnology 1997; 15 :961-4. Murphy, M. J.; Squirreii, D. J.; Sanders, M. F.; Blasco, RThe use of adenylate kinase for the detection and idendfication of low nrunbers of microorganisms. In: Campbell, A K.; Kricka, L. J. ; Stanley, P. E., editors. Bioluminescence and Chemiluminescence: Fundamentals and Applied AspectsUK: John Wdey & Sons; 1995. Murray, N. E. Bacteriophages as cloning vehicles. In: Webster, R G.;GranoE A, editors. Encyclopedia of Virology. London: Academic Press; 1994. pp. 113-6. O'Keefe B, Green J. Coliphages as indicators of faecal pollution at three recreationai beaches on the Firth of Forth. Water Resources 1989;23 :1027-30. Ohmiya Y, Hirano T, C1hash.i M. The structural ongin of the wlor differences in the bioluminescence of firefly luciferase. FEB S Letters l9%;384: 83-6. 243 Olsvik 0, Popovic T, Skjerve E, Cudjoe KS, Homes E, Ugelstad J, Uhlén M Magnetic Separation Techniques in Diagnostic Miaobiology. Clinical Miaobiology Reviews 1994;7:43-54. ûrser C, Staskawicz BI, Panopoulos NJ, Dahlbeck D, Lindow SE. Cloning and Expression of Bacterial Ice Nucleation Genes in Escherichia di.Jod of Bacteriology 1985; l64:359-66 . Pagotto F, Brovko L, WthsMW. Phage-mediated detection ofSfcilr~Zucucmazuez~~ and Escherichia coli 0157:H7 using bioluminescence. Baderiological Quality of Raw Milk 1996;IDF Special Issue 9601:152-6. Park SF, Stewart GSAB, Kroll RG. The use of bacterial lucifefase for monitoring the environmental regulation of expression of genes encoding virulence factors in Listena nronocytogenes. Journal of General Microbiology 1992;13 8:26 19-27. Pearson RE, Jurgensen S, Sarkis GJ, Hatfûfl. G.F., Jacobs WJr. Construction of D29 shuttie plasrnids and luciferase reporter phages for detction of mycobacteria Gene 1996;183: 129-36. Petitt, S. B-Aconductance screen for Enterobacteriaceae in foods. In: Stannard, C. J.; Petitt, S. B.; Skinner, F. A, editors. Rapid Microbiological Methods for Foods, Beverages and Pharmaceuticais. Society for Applied Bacteriology Technical Saies; UK: Blackweil Scientific Publications; 1989. pp. 13 1-4 1. Pfeifer CG, Finlay BB. MonitoMg gene expression of Salmonella Uiside rnamrnalian ceh: cornparison of luciferase and f3-galactosidase fùsion systerns. Journal of Micmbiological Methods 1995;24: 155-64. Phillips JD, Griffith MW. Bioluminescence and irnpedimetnc methods for assessing sheElife of pasteurized rnik and cream. Food Microbiology 1985;2:39-5 1. Phillips JD, Griffith MW. An electrical method for detecting Listena spp. Letters in Applied Miaobiology l989;g: 129-32. Pless P, Futschik K, Schopf E. Rapid detection of Salmoneilae by means of a new irnpedance-splitting method. Journal of Food Protection 1994;57: 369-76. Poteete, A R Bacteriophage. In: Webster, R G-;GranofS A, editon. Enqclopedia of Vuology . London: Academic Press; 1994. pp. 1009- 13. Prentice, G. A; Neaves, P.; Je~s,D. 1.; Easter, M. C.An Uiter-iaboratory evduation of an electrometric method for detection of Salmonellas in milk powders. In: Stannard, C. J.; Petitt, S. B.; Skinner, F. A, editors. Rapid Miaobiological Methods for Foods, Beverages and Pharmaceuîicais. Society for Applied Bacteriology Technical Series; UK: Blackweil Scientific Publications; 1989. pp. 155-64- Prescott, L. M H J. P.; Klein, D. A Microbiology Second ed. Dubuque, Iowa, USA: Wm. C. Brown Pubiishers; 1993. pp. 346-66. Pndmore A., Sacy P. Evaluation of Proprietary Culture Media for the Shultaneous Detection of Coüforms and E. coli using the RABIT System. Special Publications of the Royal Society of Chemistry 1997; 19 1:30-9. Pugh, S. J.; Arnotî, M. L.Automated conductimetric detection of Salmonellas in confectionery products . In: Stannard, C. J.; Petitt, S. B.; Skinner, F. A, editors. Rapid Microbiologicai Methods for Foods, Beverages and Pharmaceuticds. Society for Applied Bacteriology Technical Series; UK:Blackweii Scientinc Publications; 1989. pp. 185-20 1. Radetsky, P. The Invisible Invaders: Vimses and the Scientists Who Pursue Them. Boston: Little, Brown and Company; 199 1. Radetsky P. The Good Vis. Discover 1996;17(11):50-8. Richardson G. LabSMARF software user manual. 1997;Version 2.0. Richardson GH, Grappin q Yuan TC. A re&xtance colorimeter uistNment for measurement of microbial and enzymatic activities in mik and products. Journal of Food Protection 1988;s 1 :778-85. Richardson GE& Yuan JTC, Sisson DV. Enurneration of Total Bacteria in Raw and Pasteurized Milk by Reflectance Colorimetry: Collaborative Study. Journal of AOAC Intexnational l994;77:623-7. Riska PF, Jacobs WJr, Bloom BR, McKitrick J, Chan J. Special identification of A@co&ctenzll>t tuberculos~swith the luciferase reporter mycobacteriophage: use of pNitro-c~-A~tyIamino-~HydroxyPropiophenone . Journal of Chical Microbiology 1997;35:3225-3 1. Rocourt J. Bactériophages et bactériocines du genre Listenu. Zentralbl. Bakteriol. Hyg. l986;261 :12-28. Rodrigues MJRM#,Capell CJ, Kirby RM. A new capacitance medium for presumptive detection of listeiia spp. nom cheese samples. Joud of Microbiological Methods l995;23 :291-6. Roelant CH, Burns DA, Scheirer W. Accelerating the pace of luciferase reporter gene assays. Biotechnigues 1996;20:914-7. Ryser, E. T.; Mirth, E. H Listeria, Listeriosis and Food Safi.New York: Marcel Dekker, Inc. ; 199 1. Salvat G, Rudeiie S, Humbert F, Colin P, Lahelleç C. A seleaive medium for the rapid detection by an impedance technique of Psarcdomom spp. associateci with poultry meat. Journal of Applied Microbiology l997;83 :456-63. Samadpour M, Grimm Desai B, Aln D, Ongerth J, TmPI. Molenilar epiderniology of Escherichia coli 0157:H7 strains by bacteriophage lambda restriction fiagrnent length polyrnorphism analysis: Application to a multistate foodbome outbreak and a day-care center cluster. Journal of Clinid Microbiology 1993;31 :3 179-83. Sanders C4Yajko DM, Nassos PS, Hyun WC, Fulwyler MT, Hadley WK. Detection and analysis by dual-laser flow cytometry of bacteriophage T4 DNA inside Escherichia coli . Cytomeîry 1991;12:167-71. Sanders, M. E. Bacteriophages in industrial fermentations. In: Webster, R G.; Granoe A, editors. Encyclopedia of Virology. London: Academic Press; 1994. pp. 116-2 1. Sanders, M. F.A rapid bioluminescent technique for the detection and identification of Listena ntonocytogenes in the presence of Listeria immuCarnpbeli, A K.; Kricka, L. J. ; Stanley, P. E., editors. Bioluminescence and Cherniluminescence: Fundamentais and Applied Aspects.UK: John Wdey & Sons; 1995. 248 Sad& GJ, Jacobs vr,JMfU GF. LS lucifenise reporter mycobacteriophages: a sensitive tool for the detedon and assay of live mycobacteria MoIecular Miaobiology 1995;15: 1055-67. Schauer AT. Visualizing gene expression with luciferase fusions. TIBTECH 1988;6:23-7. Schmid D, Pridmore D, Capitani G, Battistutta q Neeser J-R, Jann A. Moledar organisation of the ice nucleation protein InaV from P~ez~domonassyringae. W S Letters 1997;4I4:590-4. Seeger Y Griftiths MW. Adenosine Triphosphate Bioluminescence for Hypjene Monitoring in Health Care Institutions. Journal of Food Protection l994;57:509-1 2. Sheehan MM, Garcia IL, Lopez R, Garcia P. The lytic aiyme of the pneumococcal phage Dp- 1 : a chimeric lysin of intergeneric origin. Molecular Microbiology l997;2517 17-25. Shelef LA, Eden G. Optical instrument rapidly detects and enurnerates microorganisrns in food. Food Technology 1996:82-5. Skyttii E, Mattila-Sandholm T. A quantitative method for assessing bacteriocllis and other food antirnicrobials by automated turbidometry. JoudofMicrobioiogica1 Methods 1991;14:77-88. 349 Soothill JS. Treatment of experirnental infections of mice with bacteriophages Journal of Medical Microbiology l992;37:258-6 1. Squirrell, D. J.; Mwphy, M. J.Adenylate kinase as a ce11 mark in bioluminescent assayscampbeli, A K.; Kricka, L. J.; Stanley, P. E., editon. Bioluminescence and Cherniluminescence: Fundamentals and Applied Aspects-UK: John Wdey & Sons; 1995. Steidler L, Yu W,Fiers W,Remaut E. The expression of the Photims~isluderase gene in Stc'phyZ~~oc~tl~mre2ls Cowan 1 allows the development of a live ampiifiable tool for irnmunodetection. Applied and Environmental Microbiology l9%;62:2356-9. Stephens PJ, Joynson 14 Davies KW, Holbrook R, Lappin-Scott Humphrey TJ. The use of an automated growth analyser to measure recovery times of single heat-injureci SaZrnoneh cells. loumai of Applied Microbiology l997;83 :445-55. Stewart GSAB. Challenging food microbiology fiom a molecular perspective. Microbiology 1997; 143 :2OW-108. Stewart GSAB, Smith T, Denyer S. Genetic Engineering for Bioiuminescent Bacteria. Food Science and Technology Today l989;3 :19-22. Stewart GSAB,Williams P. Ziagenes and the applications of bactexid bioluminescence. Journal of General Microbiology 1992;1 38: 1289-3ûû. Stewart GSAB, Williams P. Shedding New Light on Food Microbiology. ASM News 1993;59:241-6. Suarez Giittler A, Stratz M, Staendner LH, TWsKN, Gunian CA Green fluorescent protein-based reporter systems for genetic analysis of bacteria including monocopy applications. Gene 1997;196:69-74. Suhren G, Heeschen W. hpedance assays and the bacteriological testhg of milk and m. products. Milchwissenschaft 1987;42:619-27. Tanaka H, Shji(Iwano) T, Sawada K, Monji Y, Set0 S, Yajima M, Yagi O. Development and application of a bioluminescence AT'assay method for rapid detedion of colifonn bacteria. Water Research 1997;3 1 :19 13-8. Tarahovsky YS, Ivanitsky GR Khusainov AA. Lysis of Escherichia coli ceils induced by bacterio phage T4. EMS Microbiology Letters 1994; 122: 1 95-200. Thompson IF, Geoghegan KF, Lloyd DB, Lanzetti AT, Magyar Rq Anderson SM, Branchini BR Mutation of a protease-sensitive region in firefiy luciferase &ers light emission properties. The Journal of Biological Chemistry l997;272:18766-7 1. 25 1 Tïmms S, Colquhoun KO,Fricker CR Detection of Eschenchia coli in potable water using indirect irnpedance technology. Jodof Microbiologicai Methods 1996;26:125-32. Tombolini R, Unge 4 Davey ME, de Bmijn FJ, Jansson K. Fiow cytometrîc and rnicroscopic analysis of GFP-tagged Psezuibmonasj7uorescens bacteria FEMS Microbiology Ecology 1 99î;Z:17-28. Tomicka AT, Chen J, Barbut S, mthsMW. Sumival of Bioluminescent Escherichia coli 01 57S7 in a Model System Representing Fermented Sausage Production. Jodof Food Protection l997;6O:1487-92. Turpin PE, Maycroft Kq Bedford J, Rowlands CL, Wellington EMH. A rapid luminescent-phage based MPN method for the enmeration of &ImoneIh typhimurium in environmentai samples. Letters in Applied Microbiology 1993;16%-7. Ugarova NN. Bioanaiytical applications of firefly luciferase (review). Applied Biochemistry and Microbiology 1993;29:13 5-44. Ulitzer, S.; Kuhn, J.Introduction of ltrx genes into bacteria, a new approach for specinc determination of bactena and their antiviotic susceptibility. In: Schoimerich, J.; Andreesen, R; Kapp, A., et al., editors. Bioluminescence and Cherniluminescence, new perspectives.Toronto, Canada: John Wiley & Sons; 1987. pp. 463-72. 252 Valdivia RH, Hromockyj AE, Monack D, Ramabishnan L, Fakow S. Applications for green fluorescent protein (GFP)in the study of host-pathogen intdom. Gene 1996;173 :47-52. Vasavada PC. Rapid Methods and Automation in Dairy Microbiology. Journal of Dajr Science l993;76:3 101-13. VR, Bechara EH. Bioluminescence of Brazilian fiefies (Coleoptera: Lampyridae): spectral distribution and pH effect on luciferase-elicited wlors. Cornparison with elatend and phengodid luciferases. Photochemistry and Photobiology l995;62:49û-S. Waes GM, Bossuyt RG. Impedance measurements to detect bacteriophage problems in cheddar cheesernaking. Journal of Food Protection 1 984;47:349-51. Walker AJ, Jasshn SAA, Holah JT, Denyer SP, Stewart GSAB.Bioluminescent Listria monocytgenes provide a rapid assay for measuring biocide efficacy. FEMS Microbiology Letters 1W2;9 1 :25 1-6. Ward De Sa J, Rowe B. A phage typing scheme for Lialmonella enterifidis. Epidemiology and Infection l987;99:2914. Warren G, Wolber P. Moiecular aspects of microbial ice nucleation. Moidar Microbiology 199 1;S :U9-43. Watanabe y Arai S. Bacterial Ice-Nucleation Activity and its Application to Freeze Concentration of Fresh Foods for Modification of their Properties. Jodof Food Engineering 1994;22:453-73. Welkos S. Schreiber M, Baer H Identification of Salmonella with the 0-1 Bactenophage. Applied Micro biology 1974;28: 6 1 8-22. White PJ, Squirrell DI, Amaud P, Lowe. C.R, Murray JAH. Improved therrnostabiiity of the North American flrefly luciferase: saturation mutagenesis st position 354. Biochernistry Journal 1W6;3 l9:343-50. W~gginsBA, Alexander M. Minùnum bacterid density for bacteriophage replication: implications for signifiaince of bacteriophages in nadecosystems. Applied and Environmentai Microbiology l985;49: 19-23. Wolber, P. K ;Green, R L-Rapid Detection of Saïmonella Contamination of Foods by Transduction of Bacterial Ice Nucleation Genes. Presented at: 95th General Meeting of American Society for Microbiology 1 995; Washington, DC. Wolber PK, GrmRL. Detection of bacteria by transduction of ice nucleation genes. Trends io Biotechnology l990;8:276-9. Yang T-T, Kain SR Kitts P, Kondepudi A, Yang MM, Youvan DC. Duai wlor microscopie imagery of ceUs expressing the green fluorescent protein and a rd-Wed variant. Gene 1996;1 73 :1 9-23. YeL, Buck LM, Schaeffer HJ, Leach FR. Cloning and sequaicing of a cDNA for firdy luciferase corn Photwisperm~yhtanica.Biochimica Et Biophysica Acta 1997;13 39:3 9-52. Young R Bacteriophage Lysis: Mechanism and Regdation. Miaobiologid Reviews 1992;56:430-81. Young q Bliisi U. Holins: fom and firnction in bacteriophage lysis. FEMS Microbiology Reviews 19%; 17:191-205. IMAGE EVALUATION TEST TARGET (QA-3) APPLIED IWGE . lnc 1653 East Main Street --.- - Rochester. NY 14609 USA ---- Phone: 71 6482-0300 7------FU: 71W88-5989