How novel methods can help discover more information about foodborne pathogens

Presented at the BIOP ’99 Symposium, Ryerson Polytechnic University, Toronto, Ontario, February 25, 1999 Department of Food Science, University of Guelph, Guelph, Ontario Correspondence and reprints: Dr Mansel W Griffiths, Department of Food Science, University of Guelph, Guelph, Ontario N1G 2W1. Telephone 519-824-4120 ext 2269, fax 519-824-6631, e-mail mgriffit@uoguelph.ca MW Griffiths. How novel methods can help discover more information about foodborne pathogens. Can J Infect Dis 2000;11(3):142-153.

B ioluminescence involves the emission of visible light by living organisms through a series of enzyme-catalyzed reactions.Luminescent organisms are found in several ecological niches and comprise species of bacteria, dinoflagellates, fungi, protozoa, sponges, jellyfish, squid, starfish, worms, fish and beetles (1).The biochemistry and genetics of the light reaction have been widely studied, but other potential applications of luminescence are now being explored by scientists in other disciplines.This review will explore the ways in which bioluminescence has been used to gain insights into the behaviour of foodborne pathogens.
The genes responsible for bioluminescence in a wide variety of organisms have been identified and cloned (Table 1), and the molecular biology of bacterial bioluminescence has been described in a series of reviews by Meighen (2)(3)(4)(5), and Meighen and Dunlop (6).The relation between the genes and enzymes of the lux operon and the bioluminescence reaction is diagrammatically represented by Figure 1.
The lux genes have been transferred and expressed in several microorganisms, including foodborne pathogens (Table 2).Luminescent phenotypes have been obtained for Gram-positive and Gram-negative bacteria.However, light output obtained from Gram-positive bacteria is usually about 100 times lower than that from comparable Gram-negative genera, and this is probably the result of poor gene expression (7).

REPORTER OF GENE EXPRESSION
The 'reporter' gene concept arose from an inability to easily monitor the products of many genes.A gene with a readily assayed product (the reporter) is fused with the gene of interest so that when the latter is expressed, the product of the reporter gene is synthesized and can be detected.Schauer (20) and Nordeen (21) reviewed the use of luciferase genes for visualizing gene expression, and the former stated that this not only enabled easy detection of transcription or translation, but also allowed localization of the event in cells or organelles.Using photon imaging techniques, O'Kane et al (22) showed that it was possible to detect Rhizobium species carrying the luxAB genes in single cells from infected plants, and these techniques allow gene expression to be visualized in single mammalian cells (23) and bacterial cells (24).The use of the luxAB gene as a molecular reporter is not without limitations, and Blouin et al (25) have cautioned against using the system without appropriate protocol design.
More recently, attention has been focused on the green fluorescent protein gene of Aequorea victoria as a marker for gene expression (26,27).It does not require any substrates or additional co-factors to fluoresce, but the GFP fluorescence phenotype does not indicate the metabolic status of cells.By using a dual gfp-luxAB marker system, it is possible to monitor cell number and metabolic activity of specific bacterial populations simultaneously (28).Expression of toxin and virulence genes: Gene fusions of bioluminescent genes and toxin-producing or virulence  factor-associated genes can be made so that expression of the latter will be accompanied by light emission.Sheehan et al (29) used this approach to study the epidermolytic toxin (eta) gene of Staphylococcus aureus, which encodes a toxin causing the exfoliation of skin in neonates.They fused the promoter region of the eta gene with the Photobacterium (Vibrio) fischeri luxAB gene and found that expression of the eta gene was growth-phase dependent, being more rapidly transcribed during the late exponential to early stationary phase.The eta gene was also shown to be under the control of an accessory gene regulator.Furthermore, the effects of environmental conditions, such as osmotic strength, on the eta promoter could be assessed by this method.This technology provides a unique opportunity to study, in real time and nonintrusively, the control of epidermolytic toxin synthesis during the disease process using in vivo experimental models (16).A luciferase-based reporter system has also been used to study the expression of the toxic shock syndrome toxin-1 (tst) gene of S aureus (30), and the results suggested that a reduction in specific growth rate was the major factor controlling tst expression.
A haemolysin (listeriolysin O) and phospholipase C, encoded by genes designated hlyA and plcA, respectively, have been shown to play a role in the virulence of Listeria monocytogenes (31,32).These genes, along with others located in the virulence chromosomal region, are coordinately regulated by the product of the prfA gene (33).Park et al (34) constructed mutants of L monocytogenes in which the promoter sites of the hlyA and the plcA genes were fused with the V fischeri luxAB gene.They found that light emission was much higher in strains containing an intact copy of prfA, providing confirmation that the activity of these promoters was dependent upon the transcriptional activator, PrfA.They were also able to show that synthesis of listeriolysin O and phospholipase C was induced by heat shock (Figure 2), whereas oxidative stress had no effect on the expression of these virulence factors.In addition, the medium composition had a marked effect on the expression of the virulence genes.For example, it has been reported that the hlyA and plcA genes are repressed in the presence of cellobiose (35).
In the future, this technique will provide valuable information on how food composition and the environment in which  food is stored affect the pathogenicity of organisms responsible for foodborne illness, as well as shed light on events that occur when the pathogen enters the gastrointestinal tract.
Reporter of germination and sporulation: Luciferase has been used as a marker for gene expression in Bacillus species (36,37).Spores of lux recombinant bacteria do not luminesce until germination occurs, and this provides a valuable tool for the study of gene expression during both germination and sporulation.Another application for bioluminescent constructs of Bacillus megaterium involves the determination of the heat resistance of spores (38), which can be predicted because the level of light output from these mutants is directly proportional to the heat treatment received (39).
Reporter of cellular injury: Because the bacterial luciferase reaction requires a reduced form of flavin mononucleotide, any stimulus that affects the intracellular production of this substrate will result in a change in light output.This makes bioluminescence an effective tool for monitoring sublethal injury and subsequent recovery from the physiological insult (13,14,40).As an example, recovery from cellular injury caused by freezing has been studied using a luxAB recombinant of Salmonella typhimurium (41).Cells of the bacterium (10 5 /mL in peptone water) were frozen at -20°C and, after thawing, they were compared with a culture of non-frozen cells by measuring bioluminescence and plate count.The data obtained by both methods were equivalent, but the bioluminescence measurements were obtained in real time.Interestingly, freezing brought about a 40-fold reduction in viable cells, whether measured by bioluminescence or by plate count, after a 2 h recovery period in Luria broth.However, the bioluminescence data indicated that, immediately after thawing, there was only a fivefold difference between control and frozen cultures (14).From these figures, Stewart (14) concluded that 20% of the S typhimurium population was able to survive the freeze-thaw cycle, with an intracellular biochemistry sufficiently intact to allow immediate light production, but only 2.5% of the cells were capable of division.In other words, a large proportion of cells could survive freezing with a functional metabolic system but these cells were not viable.Thus, bacteria that are sublethally injured in this manner could continue to produce toxigenic compounds, even though they remain nonculturable.The exact mechanism for induction of this so-called 'viable but nonculturable' (VNC) state is poorly understood, but vital information on the metabolic activity of VNC organisms can be provided by studying bioluminescent phenotypes (42,43).Duncan et al (44) followed the response of Vibrio harveyi and bioluminescent strains of Escherichia coli and Pseudomonas fluorescens to starvation, and found that quantification of luminescence changes enabled measurement of both culturable and VNC cells.Such information may allow the hypothesis of Bloomfield et al (45) to be tested; they suggest that the VNC phenomenon is the result of an imbalance in metabolism set up when stressed cells are suddenly transferred to a nutrient-rich medium.This leads to an almost instantaneous production of superoxide and free radicals which these nonadapted cells cannot detoxify, and, as a consequence, some or all of these cells die.
Ellison et al ( 41) also reported on the use of bioluminescence to elucidate the mechanism of freeze injury.They found that the rate of freezing affected the survival of S typhimurium.Survival rates in supercooled (-70°C) cultures were significantly lower than those in cultures frozen to -20°C at the same freezing rate.An osmoregulated proU promoter fused with luxAB (46) was used to confirm that freeze injury was the result of membrane damage, resulting in leakage and not accumulation of potassium ions (41).
Chen and Griffiths (47) used a luminescent strain of Salmonella enteritidis, with the luxAB genes chromosomally located, to monitor recovery from acid and heat shock.A culture of the luminescent salmonella was subjected to stress and allowed to recover in a nonselective medium at room temperature for 20 h.The cells could recover from exposure to either hydrochloric acid (pH 1.8) or acetic acid (pH 3.9) for 2 mins, and recover from heating at 55°C or 65°C for 60 mins.This approach was further adapted by Bautista et al (48).In their work, a luminescent strain of Salmonella hadar was constructed and used to inoculate turkey breast meat samples.These were then treated with lactic acid (4.25%, pH 3.0) at 40°C for 10 s, before storage at -12°C, 0°C, 5°C and 10°C.Viability of the salmonella was measured as light emitted from the bacterial cells directly on the turkey breast, and recovery from injury was estimated by monitoring light output after incubation at 22°C for 10 h.Unexpectedly, the lowest recovery rate was observed after storage at 5°C and the fastest recovery was on turkey breasts that had been stored frozen at -12°C.The ability to observe recovery of bacteria from insult directly on foods can provide much more information than studies performed in culture media.
During the early stdationary phase, bacterial cells acquire greater resistance to environmental stressors, such as heat, low pH and osmotic pressure, due to the induction of specific sets of genes by an RNA polymerase sigma factor (RpoS) encoded by the rpoS gene (49).There are several mechanisms at the transcriptional, post-transcriptional and post-translational levels for regulation of RpoS in E coli and Salmonella species (50).Thus, a reporter of rpoS transcription would not be a reliable indicator of intracellular levels of active RpoS.In some Salmonella serovars, it has been shown that a virulence determinant transcribed by the gene spvA is under the control of the product (SpvR) of the spvR gene (51).The spvR gene is, in turn, under the control of RpoS.With this in mind, Swift and Stewart (52) developed a spv::lux bioluminescence reporter of RpoS activity, and used this reporter to show that events under the control of RpoS could be induced in the presence of a competitive microflora at levels of 10 6 cells or more.Thus, in the presence of high levels of competing cells, salmonella acquired resistance to several environmental stresses (53).Stewart's group also used this reporter to demonstrate that they could monitor RpoS levels in cells in model food systems (54).
Bioluminescence has also provided a novel method for studying stress responses in foods at the molecular level.Lux gene fusions with the promoter sequences of genes involved in stress responses have been constructed in E coli, so that the organism emits light when it is under a stress that triggers transcription of the stress promoter:luxCDABE gene fusion (55-60) (Figure 3).Using these constructs, it has been shown that chlorine triggers an oxidative stress response, inducing activation of both heat shock and soxRS regulons within 1 s of exposure to free chlorine (57).Mutants of E coli carrying gene fusions of the luxCDABE genes with DNA damage-inducible promoters recA, uvrA and alkA may help in further elucidating the mechanisms involved in DNA repair (60).It is not difficult to envisage how powerful this technique will be when used in conjunction with imaging techniques that allow light emission to be detected directly in foods (48,61) and from single bacterial cells (24).

REPORTER OF BIOCIDE EFFICIENCY AND ANTIBIOTIC SUSCEPTIBILITY
Bioluminescence can be used to test rapidly the efficacy of sanitizers and disinfectants used in the food industry.Light emission from a lux recombinant organism is dependent on a functional intracellular metabolism, and any substance that interferes with these intracellular processes will result in a decrease in light production.These responses occur very rapidly, usually within 15 mins, giving a real time estimate of the efficiency of the sanitizing or disinfecting procedures (13,18,19,62,63).It is possible to produce biosensors consisting of a genetically engineered bioluminescent reporter organism interfaced with an integrated circuit to report on the presence of microbial inhibitors (64).Indeed, bioluminescent bacteria have been used to assess the ability to clean food processing equipment effectively (65).
Walker et al (66) used bioluminescent constructs of L monocytogenes to measure the antilisterial activity of phenol and chlorhexidine diacetate.After determining the time for a log 10 reduction in count or light output (D-value) at different biocide concentrations and plotting a double log plot of D-value against biocide concentration, the concentration exponents were calculated from the slope of the best fit line.Similar results were obtained when concentration exponents were assessed using plate count (8.3 for phenol and 2.13 for chlorhexidine) or the bioluminescent strains of L monocytogenes (7.3 for phenol and 2.63 for chlorhexidine).However, the results were obtained in 60 mins by the latter method.The lux recombi-nant L monocytogenes strains were also used to study the efficacy of the disinfectant Virkon (67), as well as other biocides including hypochlorite (68).Bacteria with bioluminescent phenotypes can also be used to measure antimicrobial activity against cells present as a biofilm, as well as the presence of environmental toxicants (14,18,69,70).
Agents that affect the integrity of cell membranes can also be assayed using bioluminescence.The click beetle luciferase gene has been cloned into several microorganisms, such as E coli, Bacillus subtilis and Spodoptera frugiperda.The luciferase enzyme encoded by the gene requires luciferin as a substrate.The diffusion of this substrate through cell membranes is slow at physiological pH and, therefore, a change in membrane permeability is seen as a change of in vivo luminescence of the cells (71).This technique has value for the assay of nisin and other membranolytic biocides and antibiotics.
Bioluminescence can be used to monitor the response of cell-wall deficient (L-forms) of L monocytogenes to biocides (72).These L-forms are difficult to culture, and the use of bioluminescence offers a near real time method for determining viability.L-form colonies required at least three days and as long as 21 days to become visible on agar plates, but they could be detected by bioluminescence after only 8 h of incubation.
The use of a lux recombinant bacterium for determining antibiotic susceptibilities was first described by Ulitzur and Kuhn (73).Since then, reporter strains of bacteria expressing firefly or bacterial luciferase have been used to evaluate the activities of antibiotics against Mycobacterium tuberculosis and S typhimurium in whole animal studies (74,75) and against mycobacteria sequestered in human macrophages (76) (Figure 4).This technique may also be useful when evaluating the actions of antibiotics against other intracellular pathogens such as L monocytogenes.

REPORTER OF ADHESION AND BIOFILM FORMATION
The adhesion of bacteria to surfaces has been studied by in vivo bioluminescence ( adhesion by a P fluorescens strain containing the V fischeri lux gene operon.Both biofilm and bulk phase biomass could be assayed with a detection limit of 2´10 5 attached cells/cm 2 .Light production was related to biofilm lipid synthesis per unit area.The method has potential for the study of bacterial adhesion to cells during infection and as an indicator of the bactericidal activity of host defence mechanisms (82).The use of bioluminescence may help us understand the mechanisms of biofilm formation at the molecular level.A bioluminescent reporter plasmid, pUTK50, of alginic acid biosynthesis has been transconjugated into strains of Pseudomonas putida, P fluorescens and Stenotrophomonas maltophilia (83).When stimuli that increased alginate synthesis were applied to these strains, increased bioluminescence was observed, but exopolysaccharides other than alginate were produced.This suggests that the strains possessed homologous promoter sequences which control the genes necessary for the production of exopolysaccharides in response to environmental stimuli.
Bioluminescence also has application in the study of biofilms in the food industry.Dhir and Dodd (69) studied the resistance of substratum-attached, detached and planktonic cells of S enteritidis phage type 4 to biocides using in vivo bioluminescence.An excellent correlation between bioluminescence and classical plate count data was obtained when attachment profiles, biocide concentration exponents and thermal inactivation (D-values) were compared.There is a clear difference in the susceptibility to biocides of attached cells and those present in biofilms.Following treatment with a nonfoaming acid sanitizer or a liquid hypochlorite sanitizer, bioluminescent P putida cells present as a biofilm on n-buna rubber were able to recover after a 16 h resuscitation period and grow to levels approaching those of nontreated controls (84).Bioluminescence has also been used to demonstrate that L-forms of L monocytogenes are capable of forming biofilms on stainless steel and other surfaces (85).
Bioluminescence may also be applied in studies on the adhesion of bacteria to food.Siragusa et al ( 86) used a luminescent strain of E coli O157:H7 to study adhesion of the organism to beef carcass tissue.They found that retention of the bioluminescent signal was higher on lean fascia-covered tissue than on adipose fascia-covered tissue following a water rinse and that the luminescence was strongly correlated with bacterial counts.

REPORTER OF THE MICROBIAL ECOLOGY OF FOODS
The use of luminescence-based systems for studying microbial ecology has gained widespread acceptance by environmental microbiologists (87), and these techniques are now being applied to food.As well as providing quantitative information, imaging of luminescent bacterial cells can also provide information on their spatial distribution in a food.Chen et al (61) studied the ability of a lux recombinant of S enteritidis to penetrate the egg shell membrane and to grow in eggs under different storage conditions.This strain of S enteritidis was unable to penetrate the egg shell membrane but was able to grow in the space between the shell and membrane.The lumi-nescent Salmonella strain could remain metabolically active in eggs after storage at 4°C for four weeks (61).The technique is nondestructive, gives real time results and can be carried out directly on the food of interest.
Luminescent and fluorescent strains of L monocytogenes and E coli O157:H7 have been used to monitor the survival of the organisms in yogurt and cheese (88,89), during a simulated fermented sausage manufacturing process (90), and in apple and orange juice (91) (Figure 5).
These survival studies indicate that bioluminescent reporter microorganisms have great potential as tools for developing quantitative risk assessment models.Already, thermal inactivation models for S typhimurium derived using bioluminescent strains have been shown to be equivalent to models calculated from plate count data (92,93) (Figure 6).This enables models to be generated in the presence of high levels of background microflora (92) and in real food systems (48).

REPORTER OF PATHOGENESIS
Several recent reports have illustrated the usefulness of bioluminescence to study pathogenicity (5).For example, Mettenleiter and Graewe (94) used a recombinant herpes virus carrying the firefly luciferase genes to monitor activity in single virus-infected cells.Verocytotoxicity of E coli cultures has been measured by transformation of Vero cells with a plasmid containing an intronless, firefly luciferase gene (95).Verotoxin could then be assayed by measuring the decrease in light emission with incubation time (Figure 7).The time taken for the assay could be reduced from three days for the conventional Vero cell cytotoxicity assay to 6 h for the bioluminescent cell technique, with no loss in sensitivity.
Arguably, the most exciting development is the ability to use bioluminescence to monitor bacterial pathogens in a living host.Contag et al (75,96,97) converted three strains of S typhimurium, which differed in their virulence for mice, to a bioluminescent phenotype through transformation with a plasmid conferring constitutive expression of bacterial luciferase.Using low light imaging, they were able to detect photons transmitted through tissues of the animals infected with the luminescent salmonella, and this allowed localization of the bacteria to specific tissues.Patterns of bioluminescence were observed Methods discover information about foodborne pathogens Figure 5) Survival of luminescent phenotypes of Escherichia coli 0157:H7 in apple juice (91), orange juice (91) and yogurt (88), determined by bioluminescence that suggested that the caecum played a vital role in salmonella pathogenesis.It was concluded that real time, noninvasive analyses of pathogenic events and pharmacological monitoring could be performed in vivo.
Bacteria are able to sense and respond to their own population densities by releasing and sensing pheromones.This phenomenon is referred to as 'quorum sensing' (98).In Gram-negative bacteria, quorum sensing is under the control of the LuxR family of transcriptional regulators, and the pheromone molecules (synthesized by members of the LuxI family of proteins) are acyl homoserine-lactone (AHSL) analogues.This form of cell-cell communication plays an important role in the regulation of expression of virulence factors involved in pathogenesis and luminescence and may be an important tool for studying these regulatory mechanisms (99).Because pathogenic bacteria use AHSL signals to regulate virulence genes, an understanding of the underlying mechanisms of quorum sensing may lead to the development of novel antivirulence drugs (100).The link between virulence and cell density was demonstrated using a bioluminescent reporter, where it was shown that S enteritidis strains that grew to a higher cell density than the wild type underwent significant morphological changes and became more virulent (101).

BACTERIAL DETECTION USING MOLECULAR BIOLUMINESCENCE
Ulitzur and Kuhn (73) proposed an elegant method for the detection of bacteria in foods based on the innate specificity of bacteriophage for an appropriate bacterial host.They cloned the lux genes into host-specific phages so that, on infection, the bioluminescent genes would be transferred to the host bacterium where they would be expressed, causing light to be produced.Light is not produced until after infection by the phage because the phage does not possess the intracellular biochemistry to express the genes and the host does not contain the lux genes until transduction occurs.This detection method can be quantitative if the phage is present at optimal concentrations because the amount of light emitted is proportional to the number of infected bacteria.This methodology has been applied to the detection of a variety of bacteria and can even be used for the detection of bacteria directly in foods (102).

Detection of E coli:
In their original work, Ulitzur and Kuhn (73) used cell concentrations of E coli W3110 in the range 10 1 to 10 4 colony forming units (cfu)/mL.The E coli cells were concentrated by membrane filtration and the membranes were placed in scintillation vials.Following addition of a lux modified l L4 phage (7´10 7  ple, and as few as 10 cells in the sample could be detected within 100 mins.The lux l L4 phage (2´10 8 pfu) was also added to E coli W3110 cells in sterile milk and luminescence was measured after incubation at 25°C for 18, 27 and 40 mins (73).Again, initial numbers of E coli cells in milk were directly correlated with light production, and the assay could detect 10 E coli cells/mL of milk within 30 mins.The method was also capable of detecting and enumerating E coli cells in urine samples.It has been estimated that there are over 100,000 cases of illness caused by enterohemorrhagic E coli annually in the United States, resulting in nearly 80 deaths (103).This has made it desirable to develop simple and rapid methods for the detection of this organism, and bacteriophage-based methods may prove to be valuable.A bacteriophage broadly specific for E coli O157, designated AR1, has been isolated (104).A Luciola mingrelica luc recombinant AR1 has been obtained by simultaneously infecting wild type E coli O157:H7 with a pBluescript II SK (±) phagemid containing the luc gene (Promega Corporation, Madison, Wisconsin) and AR1 (105).The transducing phage was capable of detecting 10 6 E coli O157:H7 cells/mL.Detection of Salmonella species: A lux + P22 phage has been engineered and used in experiments to infect S typhimurium LT2 (19).When light emission following infection was measured with a luminometer, as few as 10 2 S typhimurium cells could be detected, even when they were present in mixed culture at a ratio of one salmonella cell to 10 6 cells of other bacteria.S typhimurium could be detected with the lux-modified P22 phage in environmental samples including water, soil and sewage sludge (106).These researchers adopted a most probable number technique based on a 15 tube test method, consisting of five tubes containing 10 mL, 1 mL or 0.1 mL of sample in buffered peptone water.After overnight incubation at 37°C, subsamples from each tube were transferred to Luria broth in scintillation vials and 6.9 x 10 9 pfu of the lux-modified P22 phage added.Light output was measured with a luminometer after 90 mins incubation at 30°C.There was an excellent correlation between the most probable number obtained by the luminescence method and plate count for all samples, and results were achieved within 24 h.No false positive or false negative results were obtained with the lux + phage method in any of the samples tested.
Chen and Griffiths (102) obtained three recombinant phages carrying the luxAB genes from V harveyi, either by infection of luminescent strains of Salmonella with wild type bacteriophage or by bacteriophage induction from luminescent, lysogenic bacterial cells.Using these phages, it was possible to detect, by photon imaging techniques, light emitted after infection of group B, D and some group C Salmonella species directly in milk, on the surface of chicken meat and even in whole eggs.Eggs contaminated with about 10 3 cfu/egg became luminescent after 16 h incubation, but the detection limit could be decreased to as few as 63 salmonella cells/egg when the incubation period was extended to 24 h.Chen and Griffiths (102) also described a technique whereby Petrifilm (3M, St Paul, Minnesota) and the lux + modified phage could be combined for the detection of Salmonella species using x-ray autoradiography for observing light emission.
Detection of L monocytogenes: Loessner et al (107) isolated a bacteriophage, designated A511, which is a Listeria genusspecific, virulent myovirus that infects 95% of L monocytogenes one-half and four serovars.A recombinant phage was constructed by homologous recombination, which carried the gene for a fused V harveyi LuxAB protein inserted immediately downstream of the major capsid protein gene (cps).Transcription of the luxAB gene was initiated by the cps promoter at 15 to 20 mins after infection, and infected cells were detectable by their bioluminescent phenotype.Following infection and a 2 h incubation period, between 10 2 and 10 3 L monocytogenes cells/mL could be detected using a simple luminometer.Loessner and his colleagues (108) subsequently evaluated the use of the A511::luxAB phage for testing contaminated foods and environmental samples for the presence of viable listeria cells.With a short pre-enrichment step of 20 h, the phage was capable of detecting very low initial contamination rates in several foods artificially contaminated with L monocytogenes Scott A cells.In ricotta cheese, chocolate pudding and cabbage, less than one cell/g of food could be detected by comparing the light emission of phage-infected samples to that of controls without the lux + phage.In foods having a large and complex microbial background flora, such as ground beef and soft cheese, at least 10 cells/g were necessary to produce a positive bioluminescence signal.Of 348 potentially contaminated natural food and environmental samples, 55 were found to be listeria positive by the lux phage assay versus the 57 positive samples detected by the standard plating procedure.The lux phage procedure detected more positive samples among dairy products and environmental samples, whereas the plating procedure revealed more contaminated meat and poultry samples.Overall, both methods were equally sensitive.However, the minimum time required for detection of listeria with the lux phage assay was 24 h, as opposed to the four days needed by the standard plating method.The phage could also be used in a most probable number technique to provide rapid enumeration of low levels of listeria cells in foods against the background of a competing microflora.Detection of M tuberculosis: Although foodborne tuberculosis infection has been eradicated in Western countries, the numbers of reported tuberculosis cases are again on the rise, mainly due to the emergence of antibiotic-resistant strains of the organism.Culture methods for M tuberculosis require several weeks, and the need for more rapid testing systems prompted Jacobs et al (109) to investigate bioluminescence.Shuttle phasmid vectors from a number of mycobacteriophages can efficiently deliver recombinant DNA into mycobacteria by infection, and these phasmids are also amenable to genetic manipulation in E coli.A shuttle phasmid from a mycobacteriophage (TM4) that formed plaques, not only on the slow growing M tuberculosis but also on the faster growing species Mycobacterium smegmatis, was constructed by inserting the firefly luc gene downstream of a strong hsp60 promoter in an E coli cosmid (109).The luc + phage infected Bacillus Calmette-Guérin (BCG) vaccine strains of mycobacteria, as well as M tuberculosis and M smegmatis.When mixed in approximately equal numbers (5´10 7 cfu/mL bacteria and 5´10 7   Can J Infect Dis Vol 11 No 3 May/June 2000 pfu of phage), light was produced within minutes of infection, although the limit of detection appeared to be 10 4 cfu/mL.The light output was increased 1000-fold following incubation for 2 h at 37°C but was still much lower than the light emitted by an equal number of cells transformed with the luc + plasmid.This was probably due to less efficient gene expression in phage-infected cells or phage-mediated inhibition of the cellular metabolism in the host bacterium, resulting in lower intracellular ATP concentrations.The slower growth of M tuberculosis compared with M smegmatis was probably not caused by differences in gene expression because the kinetics of light production were the same in both species (109).It has been claimed that as few as 10 cells of M smegmatis can be detected in 40 h using L5 luciferase reporter mycobacteriophage (110).pfu of phage), light was produced within minutes of infection, although the limit of detection appeared to be 10 4 cfu/mL.The light output was increased 1000-fold following incubation for 2 h at 37°C but was still much lower than the light emitted by an equal number of cells transformed with the luc + plasmid.This was probably due to less efficient gene expression in phage-infected cells or phage-mediated inhibition of the cellular metabolism in the host bacterium, resulting in lower intracellular ATP concentrations.The slower growth of M tuberculosis compared with M smegmatis was probably not caused by differences in gene expression because the kinetics of light production were the same in both species (109).It has been claimed that as few as 10 cells of M smegmatis can be detected in 40 h using L5 luciferase reporter mycobacteriophage (110).
The versatility of molecular bioluminescence was demonstrated by Jacobs et al (109), who used the bioluminescent M tuberculosis cells obtained by transduction to determine their antibiotic susceptibility.An antibiotic-sensitive strain, an isoniazid-resistant strain and a multiple drug-resistant strain of M tuberculosis were grown and, following incubation with antibiotics, the transducing particles were added to give a multiplicity of infection of 1000.Luciferin was added after a further incubation step and light production was measured in a luminometer.Resistance to an antibiotic was demonstrated by light production, whereas strains sensitive to the test drug did not luminesce.This methodology allowed a dramatic reduction in the time needed to determine the antibiotic susceptibility of M tuberculosis from two to three months by cultural methods to two to three days by bioluminescence.
Thus, one assay permits both rapid detection and antibiotic susceptibility of the actual causative agent of the disease.This is of considerable importance given the increase in antibioticresistant organisms and the emergence of a number of lifethreatening foodborne pathogens, such as S typhimurium DT104, which possess increased drug resistance.

Detection of S aureus:
In an independent study, the author's research group at Guelph has adopted the same strategy as Jacobs et al (109) to detect and determine the antibiotic susceptibilities of mastitis-causing agents, especially S aureus (105).A battery of five phages were identified that could infect all strains of S aureus tested, whether they were isolated from mastitic cows or from foods implicated in foodborne illness.A plasmid incorporating the luxAB genes fused to a staphylococcal cadmium-resistant gene promoter was introduced into staphylococcal phages by homologous recombination.Using these recombinant phages, it was possible to detect 10 6 S aureus cells/mL in broth cultures.The luminescent strains ob-tained after transduction could be used to test for antibiotic susceptibility.

Multiplex assays:
The luciferases encoded by different organisms have distinct properties (1,111,112) which can be used to develop systems that can detect, and differentiate between, more than one bacterium in a single assay.For example, the luciferase encoded by the bacterial lux genes has a requirement for a long chain aldehyde and emits light at a wavelength of about 490 nm, whereas the firefly luc gene encodes an enzyme which uses luciferin and produces light at a wavelength near 560 nm (Table 3).Thus, the light produced from recombinant bacteria containing either the lux or luc genes can be easily distinguished, either on the basis of substrate specificity or emission wavelength.Lux-modified phages specific for Salmonella species and S aureus, and luc + phages that infect only E coli O157 strains have been engineered (102,105).With the aid of a 500 nm cut-off filter, it is easily possible to differentiate between bacteria expressing the lux genes and those expressing luc by photon imaging.
This method of differentiating between microorganisms has great potential for studying microbial interactions directly in food systems, and work in this area is being conducted in the author's laboratory.Advantages and limitations of the method: Lux-modified phage-based detection methods are attractive to the food microbiologist because they offer a nondestructive, rapid and easy-to-use alternative to other technologies.A particularly attractive feature of these techniques is their ability to detect the presence of luminescent microorganisms directly on food surfaces through instrumentation that allows low light imaging.Although Ulitzur and Kuhn (73) have claimed that it is possible to detect as few as 10 bacterial cells/mL directly in milk using transduction of lux genes, no research has been published that confirms this observation.However, even if the method is unable to detect directly the low number of pathogens that may be present in food, the system is sensitive enough that dramatic reductions in enrichment times can be achieved (106,113).As discussed previously, recent work in the author's laboratory has shown that simple techniques can be used to differentiate between organisms expressing luciferase genes from different sources (eg, lux and luc genes).The application of this method to food opens the door for the development of 'single shot' assays capable of detecting more than one type of bacterium.The economy of this approach, in terms of both reagents and labour, would be appealing to food microbiologists.
Some of the difficulties that have to be overcome are associated with propagation and stability of the phage, as well as poor expression of the luciferase genes in the host.This is especially true for Gram-positive bacteria, but the use of luxAB gene fusions may go some way to alleviating this problem.The identification of suitable gene promoters to regulate expression of the luciferase gene is also a key factor in optimizing the phage-mediated assay.
The specificity of the method is only limited by the identification of suitable bacteriophages capable of only infecting the target bacterium.Bacteriophages generally have a narrow host range, which is limited in most cases to single bacterial species.For certain bacteria, such as Salmonella species, phages are well characterized, but in other species, phages have still to be identified that infect an acceptable proportion of strains.Even when suitable bacteriophages are available, the lack of information on the phage genome can be a hindrance to the development of suitable assays.

CONCLUSIONS
The phenomemon of bioluminescence can be used in several ways to provide data that can help the food microbiologist achieve the goal of safer food.It can provide a way of detecting potential pathogens in food through the use of lux-modified phage.By creating organisms with a bioluminescent phenotype, it is possible to gain information on the way foodborne pathogens survive in food and interact with their environment.The potential of the technique has not been fully realized, but interest in applying molecular bioluminescence to solve problems related to the microbiological safety of foods is increasing.
Can J Infect Dis Vol 11 No 3 May/June 2000 Methods discover information about foodborne pathogens

GriffithsFigure 6 )Figure 7 )
Figure 6) Use of luminescent phenotype of Salmonella typhimurium to assess heat resistance.A Comparison of viable count and bioluminescence, data from reference 92; and B Monitoring recovery from heat treatment using bioluminescence, data from reference 93