1336_C09.

fm Page 267 Monday, November 1, 2004 10:07 AM

9 Phage for the Detection

of Pathogenic Bacteria
Catherine E.D. Rees1 and Martin J. Loessner 2
1
School of Biosciences, University of Nottingham, Sutton
Bonington Campus, Loughborough, Leicestershire, UK
2
Institute of Food Science and Nutrition, Swiss Federal
Institute of Technology, ETH Center, Zürich, Switzerland.

CONTENTS

1. Introduction ...................................................................................................267
2. Recombinant Phage .......................................................................................268
2.1. Luciferase Reporter Phage (LRP) .......................................................268
2.2. Other Reporter Phage ..........................................................................272
3. Phage Display ................................................................................................273
4. Dual Phage ....................................................................................................275
5. Phage Amplification Assays ..........................................................................277
6. Detection via Phage-Mediated Cell Lysis ....................................................279
7. Use of Phage and Phage Products as Binding Reagents .............................280
Acknowledgements ...............................................................................................281
References ............................................................................................................281

1. INTRODUCTION
The specificity of the interaction of a virus with its host cell immediately lends itself
to methods for the identification of bacteria, in particular the pathogens. While many
other procedures based upon antibodies (ELISA) or nucleic acid amplification (PCR)
have been artificially developed to allow differentiation of bacterial cell structures,
here we have a naturally evolved system in which the bacteriophage specifically
recognizes and binds only to its own host cells. This interaction has been exploited
in a number of different methods for the specific detection and differentiation of the
individual host bacteria. One of the first uses of phage was in typing schemes, where
a panel of phages with different lytic spectra is used to discriminate between different
isolates of a bacterial species or genus, according to their ability to infect the isolate
and form plaques. Differences in infectivity reflect differences in a number of cellular
characteristics, such as cell surface receptors, the presence of restriction modification
systems and related prophage, plasmid carriage, etc. When using an identical set of
phages, typing results are highly reproducible between different laboratories, although

0-8493-1336-X/01/$0.00+$1.50
© 2005 by CRC Press LLC 267
1336_C09.fm Page 268 Monday, November 1, 2004 10:07 AM

268 Bacteriophages: Biology and Applications

it is important that the phages used are propagated under specified conditions. The
ease of use and inexpensiveness of phage typing also contribute to the fact that it is
still among the most widely used methods for strain identification of many bacteria—
in particular, Salmonella, Listeria, and Staphylococcus.
Other detection techniques have been developed to directly harness the specific
binding ability of bacteriophage. These include the production of fluorescently
labelled antibodies directed against specifically adsorbing phage particles (Watson
and Eveland, 1965), or fluorescently labelled phage prepared by cross-linking dyes
to the phage surface (Hennes et al., 1995; Goodridge et al., 1999). In both of these
methods target cells are detected following adsorption of phage to the host cells and
then identification of this interaction by detection of the bound fluorescent signal.
Although this basic host-binding identification step in phage-based assays remains
the same in the various methods discussed in this chapter, our increased understanding
of both phage genetics and phage structure has allowed new genetically engineered
reagents to be developed.

2. RECOMBINANT PHAGE
2.1. LUCIFERASE REPORTER PHAGE (LRP)
The idea that phage can transduce genes into their host cell is not new. However
the idea that reporter genes could be introduced into a phage so that the phage
infection event could be readily detected was first proposed as a detection method
by Ulitzur and Kuhn (1987). In this case the reporter of choice was the bacterial
bioluminescence (lux) genes. Expression of these genes from a modified lambda
phage was rapidly and sensitively detected using a luminometer (see Fig. 9.1).
Initially a phage-based cloning vector containing the complete lux operon was used
to demonstrate that as few as 10 E. coli cells could be detected following phage
infection (Ulitzur and Kuhn, 1989). This group went on to produce luciferase reporter
phage by random Tn10::luxAB mutagenesis of wild-type phage genomes and recovered
recombinant luxAB+ phage from bioluminescent plaques. These simple constructs
termed Luciferase Reporter Phage (LRP) have been used to successfully detect the
presence of enteric bacteria in meat (Kodikara et al., 1991) and Salmonella in eggs
(Chen and Griffiths, 1996).
A more rational approach to the construction of an LRP was used by Loessner
et al. (1996). In this case, genetic analysis of the virulent Listeria phage A511
(Loessner et al., 1994) allowed identification of the late region of the phage genome
which encoded the major capsid and tail sheath proteins (Loessner and Scherer,
1995). As this construct was to be expressed in a Gram-positive bacterium, the fused
luxAB genes from Vibrio harveyi (Hill et al., 1991) were used and introduced by
recombination into the operon encoding the phage structural genes without disrupting
any of the existing gene structures. This was achieved by first introducing a plasmid
containing the modified phage operon into L. monocytogenes host cells. After plating
the phage on these plasmid-containing hosts, the desired LRP recombinant clone
was recovered from bioluminescent plaques, detected by exposing the plates to
aldehyde vapour. Although the centre of the plaque is non-bioluminescent due to
cell lysis, the infected cells at the edge of the plaque still express the phage-encoded
 rep
1. 2. 3.
Phage Phage Signal
infection replication detection

rep

Reporter
p rep rep Reporter
re
1336_C09.fm Page 269 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria

Reporter gene (rep) Reporter phage Reporter gene Reporter protein

inserted into phage only infects host cells expressed from accumulates &
genome & packaged phage genome signal detected

FIGURE 9.1 Reporter phage assay.

Reporter genes (e.g. lux, luc, ina, or gfp) are introduced into the phage genome and recombinant DNA packaged into phage particles (see text for details).
Reporter phage are mixed with the sample but only infect susceptible hosts, removing the need for purification of the organism before detection. Following
infection, reporter phage genomes are replicated and the reporter gene expressed. Once sufficient reporter protein has accumulated the expression of the
reporter gene can be detected.
269
1336_C09.fm Page 270 Monday, November 1, 2004 10:07 AM

270 Bacteriophages: Biology and Applications

lux genes and the substrate for the luciferase enzyme—a volatile aldehyde—diffuses
freely into the cells (Blisset and Stewart, 1989). LRP were recovered at a high
frequency (5 × 10−4; Loessner et al., 1996) probably because recombination between
plasmid and phage sequences is favoured due to the high copy number of each of
these DNA molecules during phage infection. When used to infect known numbers
of Listeria host cells, the bioluminescent signal from these Listeria LRP was found
to be proportional to the number of cells present, with a limit of detection of
approximately 500 cells without enrichment (Loessner et al., 1996).
The Listeria lux phage was evaluated for its ability to detect the presence of
L. monocytogenes in different types of food samples (Loessner et al., 1997). The
three foods chosen for this study were cabbage leaves, soft cheese, and minced meat.
The cabbage leaves have a flora comprising typical phylloplane microbes, predom-
inantly pseudomonads and Bacillus species. Numbers of organisms present will be
variable, depending on weather conditions, soil type, and the nature of fertilizers
applied, and will be 105 to 107 cfu per gram. Using cabbage samples as few as
0.1–1 cell per gram of food were detectable within 20 hours. In more complex
samples, such as soft cheeses or minced meat, the limit of detection was 10–100 cells
per gram. In the cheeses the predominant flora are the lactic acid bacteria at high
numbers (above 106 cfu per gram) while the meat will contain a wide variety of
bacterial types, including enteric bacteria and Gram-negative psychrotrophic rods
(again pseudomonads and related genera, in the range of 106 cfu per gram). Thus,
as for many rapid methods applied to food samples, the nature of the food matrix
and the competing microflora is critical to the sensitivity of the test. However, this
study demonstrated the main advantages of a phage-based test; the specificity of the
host-phage interaction allows the detection of low numbers of pathogens in the
presence of much higher numbers of natural competitors without the need to carry
out successive rounds of enrichment and selective plating.
Even though this provides a saving of both time and consumable costs, there is
considerable resistance in many parts of the world to the use of recombinant phage
in food analysis because these lux phage constitute Genetically Modified Organisms
(GMOs). This is despite the fact that the use of virulent phage effectively prevents
survival of infected cells, and should therefore also minimize the potential for the
establishment of any phage-encoded reporter genes in replicating bacterial cells—
one of the major concerns about the use of GMOs. While these tests are designed
to be carried out only within the laboratory so that containment is easily achieved,
in many countries the cost of registering labs for the use of GMOs is often prohibitive
within the wider context of routine food testing.
Recently, Kuhn et al. (2002) have described the construction of a genetically
“locked” lux phage derived for the Salmonella phage Felix 01 that cannot be prop-
agated on wild-type host cells. The modified phage contain two amber mutations
and will only grow on supF+ Salmonella strains, such as K772, which can suppress
such mutations. Since these phage are double amber phage mutants, reversion
rates were very low (10−8 to 10−9 per generation). To allow these phage to infect
wild type Salmonella cells, a supF gene was introduced into the phage genome on
a DNA fragment in association with the luxAB genes and during the insertion process
an essential gene was deleted from the phage genome. These phage could now be
1336_C09.fm Page 271 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 271

grown in a supF − host modified to express the deleted Felix-01 gene from a plasmid
construct to provide the essential phage functions in trans. As long as these phage
maintain the luxAB-supF segment, they will infect wild type Salmonella and produce
bioluminescence following infection; however, no viable phage particles are pro-
duced at the end of the infection cycle due to the deletion in the essential gene.
Interestingly, when tested against a range of different Salmonella isolates, it was
found that a bioluminescent signal was produced from cells that will not propagate
the wild type phage (i.e. the infection host range is wider than the replication host
range; Kuhn et al., 2002), and this increases the versatility of the assay as the number
of strains that can be detected is greater than predicted.
Although this strategy requires a more detailed understanding of the genetic
organization of relevant regions of individual phage, the fact that the reporter phage
produced are nonviable (and therefore cannot replicate to be released into the
environment at the end of the assay) should address any concerns about the use of
such GMOs in in vitro assays.
Although both the Listeria lux phage and the Salmonella lux phage have been
successfully developed, they have not been extensively evaluated. In contrast, the
recombinant phage developed for detection of Mycobacterium tuberculosis has
received much attention. The reason for this is that the use of the reporter phage
circumvents the need to culture this slow-growing organism before detection and
therefore brings a huge benefit, reducing detection times from weeks to days. There
also seems to be less concern about the use of GMOs in the context of medical
diagnosis, but no commercialized test using LRP has yet been introduced.
The LRP developed for the detection of Mycobacteria contain the firefly
luciferase (luc or Fflux) as the reporter gene (Jacobs et al., 1993). Initially, the lytic
phage TM4 was used but the rapidity of the lytic cycle meant that only limited
amounts of luciferase protein were produced and the limit of detection was 104
mycobacterial cells (Jacobs et al., 1993). When the same workers constructed a
reporter phage based on the temperate phage L5, prolonged expression and accu-
mulation of the luciferase protein was seen because the luc genes became associated
with a constitutive chromosomal promoter when the phage integrated into the
genome. This amplification of the signal allowed the limit of detection to be reduced
to a minimum of 102 cells after a 40 h incubation period, or 103 cells after 20 hours
incubation (Sarkis et al., 1995). However the limited host range of the L5 phage
meant that it was not widely applicable for the detection of M. tuberculosis in a
practical test. Further work has been carried out to improve the TM4-based LRP,
including isolating various spontaneous mutants of the parent phage and changing
both the site of insertion of the luc gene in the phage genome and the promoter used
to expressed the gene. The most recently described TM4-based phage, phAE142,
uses the PLeft promoter from phage L5 to express the luc genes to high levels, but
is genetically unstable; propagation in wild type M. smegmatis leads to rapid loss
of the luc genes. Hence a special propagating strain that constitutively expresses the
L5 protein Gp71, the repressor of the PLeft promoter, is required for phAE142
(Bardarov et al., 2003). This promoter had previously been shown to express the
luc gene to toxic levels in Mycobacteria (Brown et al., 1997), and the high level of
luc expression in this phage construct has fortuitously created phage that again are
1336_C09.fm Page 272 Monday, November 1, 2004 10:07 AM

272 Bacteriophages: Biology and Applications

unlikely to cause concern as a viable GMO. These phage have been used in combi-
nation with further refinements of the assay conditions, in particular processing steps
for the clinical sputum samples to remove inhibitors of phage infection, and the assay
has been shown to successfully detect mycobacteria in smear-positive sputum samples
within 24–48 hours (Riska et al., 1997). It also has been shown to perform favourably
in detection of positive mycobacterial specimens when compared with other standard
microbiological testing methods such as MGIT (Bardarov et al., 2003).
The majority of work has been carried out with bioluminescence reporter genes,
which, depending on their origin, have certain features that must be considered
carefully. Bacterial LuxA and LuxB subunits from marine bacteria and fusions of
the luxAB genes that create a single functional LuxAB polypeptide were found to
be rather sensitive to temperatures above 30°C (Hill et al., 1991, Loessner et al.,
1996), although the temperature stability of the enzymes varies depending on the
host strain in which they are expressed (Mackey et al., 1994). This fact can limit
their application to bacteria capable of growing and synthesizing the enzyme in a
narrow temperature range. The problem can be addressed by using other luciferases
from bacteria such as Xenorhabdus luminescens (Li et al., 1993).
The need for the delivery of specific substrates may also be a limiting factor,
especially when using the firefly luciferases, which require rather large luciferin
molecules that do not readily diffuse across bacterial membranes. In contrast, the
small and lipophilic aldehyde used by the bacterial enzyme does freely diffuse across
bacterial cell membranes but is toxic in high concentration to bacterial cells. Hence
there is a need to deliver these substrates, which can also be short lived, at an optimal
time point following phage infection to maximize the detection of the reporter gene
being expressed from the phage.

2.2. OTHER REPORTER PHAGE

The incorporation of other reporter genes into phage has been described. These
reporters were chosen so the signal they generate can be detected without background
noise from the sample, and hence the signal can be amplified to increase sensitivity.
For instance the bacterial ice nucleation protein (ina) has been used to construct
reporter phage for the detection of Salmonella (Wolber and Green, 1990). Very low
levels of gene expression are sufficient to generate a detectable signal, and it has
been reported that only one molecule of ice nucleation protein needs to be synthe-
sized to allow a positive test result (Wolber, 1993). The Ina protein is detected using
a phase-sensitive fluorescent dye that changes colour as the buffer freezes. When
samples are cooled appropriately, only those containing infected Salmonella cells
presenting Ina on their surface freeze, causing the dye to change colour, and this is
readily detected using a spectrophotometer. This assay allowed the detection of
samples containing only 2 cells/ml of S. enteritidis in a buffer system within 3 hours
and the sensitivity was further increased by using salmonellae-specific immunomag-
netic bead separation in combination with phage detection (Irwin et al., 2000).
A lambda-based reporter phage carrying the gene for green fluorescent protein
(Gfp) has also been described and used to detect the presence of E. coli (Funatsu
et al., 2002). As for the ina gene, no exogenous substrate is required, which simplifies
the assay, and the fluorescent signal can be sensitively detected following excitation
1336_C09.fm Page 273 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 273

of the Gfp protein. Detection requires 4–6 hours of incubation since time for correct
folding and oxidation of the Gfp polypeptide into the functional fluorophore is a
rate-limiting step. However a benefit of this reporter is that, once formed, the Gfp
protein is stable and the signal will be relatively long-lasting; the protein released
from cells lysed during the phage infection will actually yield a better signal.

3. PHAGE DISPLAY
The development of phage display (Smith, 1985) has revolutionized the isolation of
peptides with binding affinity for particular substrates. In this system, libraries of
peptide variants (often produced using mutational PCR methods) are cloned into
filamentous phage vectors as gene fusions to coat proteins. The cloned sequences
are expressed along with the phage coat proteins and incorporated into the mature
phage particle and are thus “displayed” on the surface of the mature virion. The size
of the peptides that can be expressed are limited as phage structure cannot be
disrupted so that the recombinant phage remain viable. The key to their success is
that libraries of clones are easily screened because the peptides encoded by the
cloned gene fragment are available for analysis (see Smith and Petrenko, 1997). For
instance, clones expressing peptides with a particular affinity can be recovered from
libraries of phage particles by biopanning experiments, where the target ligand is
bound to a solid surface and then incubated with phage particles. Phage that do not
bind tightly to the ligand are washed away, and the remaining bound fraction of the
library, which is enriched for clones expressing peptides that bind to the target ligand,
is recovered. Repeated rounds of biopanning lead to rapid identification of specific
clones expressing peptides with the highest binding affinity (see Hayhurst and
Georgiou, 2001, and Fig. 9.2).
This approach has led to the development of Phage Antibody technology where
the peptide displayed is the antigen-binding domain of an antibody molecule. A
library of such phage includes billions of clones expressing antibody-derived peptides
with different binding affinities that can then be used in an affinity selection process
(see Rader and Barbas, 1997). The most significant advance provided by this method
for selecting antibody fragments is that there is no requirement for the target ligand
to be immuno-reactive in a given host. Similarly, toxic substances, large amounts of
which cannot be delivered into a living tissue to raise an antibody, can also be used
as the target ligand.
Generally this type of technology has been used for medical or pharmaceutical
applications, with the phage display system being used solely to identify a peptide
with the desired binding characteristics. The cloned gene is then transferred to high-
level expression systems to produce the antibody fragments in usable quantities. For
use in detection assays, the molecule must also be labelled in some way so that the
binding event can be quantified. This approach has been successfully used to develop
a range of detection assays, including the identification of a variety of viruses (see
Petrenko and Vodyanoy, 2003), the differentiation of Candida species in clinical
specimens (Bliss et al., 2003) and to develop an assay for the species-specific
detection of Bacillus spores (Zhou et al., 2002; Turnbough, 2003). A peptide-mediated
magnetic separation technique has also been reported by Stratmann et al. (2002)
 274

III / VIII
ATCGTATTCGG

1.
2. 3.
Library
Washing to Further
Peptides mixed with
remove rounds of
bound target
unbound phage biopanning
ligand
1336_C09.fm Page 274 Monday, November 1, 2004 10:07 AM

Library of genes Displayed peptides Bound phage isolated Phage with highest
encoding peptides interact with bound and enriched binding affinity
cloned into phage target ligand purified & sequenced
genome & packaged

FIGURE 9.2 Phage display/biopanning.

Fragments of peptide libraries are cloned as protein fusions with either the Fd phage coat protein gene III (minor coat protein) or VIII (major coat
protein). If peptides are fused with gene III protein, fewer copies of the peptide are displayed on the phage but functional phage are produced without
the need for complementation. If peptides are fused with gene VIII protein, wildtype gpVIII must be provided in trans in the host cell to allow synthesis
of viable phage particles which contain a mixture of native gpVIII and gpVIII fused to peptide sequences. Phage libraries are exposed to the target
ligand bound to a solid support and unbound phage removed by washing. Bound phage are eluted and grown to enrich the number of phage present
with the desired binding affinity. The biopanning process is repeated, using increasingly stringent washing procedures to finally isolate phage with the
highest binding affinity. These are plaque purified and the cloned insert sequenced.
Bacteriophages: Biology and Applications
1336_C09.fm Page 275 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 275

who used phage display to identify a peptide that specifically binds to Mycobacte-
rium avium subspecies paratuberculosis. The peptide was coated onto paramagnetic
beads and then used for magnetic capture of M. paratuberculosis from milk samples
before carrying out PCR-based identification of the organism.
However, it has also been recognized that instead of purifying the peptide and
labelling with a detectable marker, such as alkaline phosphatase, a fluorescent protein
or dye, the phage particle itself can be used as a scaffold protein to stabilize the
peptide in a detection assay. In this case the phage surface can be conjugated with
dyes such as Cy5 or Alexa, either directly or via a fluorescently labelled monoclonal
antibody raised against phage surface proteins. Ideally, the peptide is expressed as a
fusion to the phage fd major coat protein, pVIII, so that up to 4000 copies of the
target ligand binding peptide are present on the surface of the phage particle. This
direct usage of phage display particles has been shown to be able to detect staphy-
lococcal enterotoxin B with an efficiency comparative to that achieved for a conven-
tional ELISA assay (Goldman et al., 2000). In addition, Turnborough (2003) showed
that both the purified, labelled peptide and a labelled phage-display particle could be
used to specifically identify Bacillus spores using fluorescent microscopy. However
it was noted that the complex surface structure of the phage filament could interfere
with the binding of the labelled phage-display particle to some types of spore samples,
and this may represent a limitation of the direct utilization of peptide-display phage
in detection assays.

4. DUAL PHAGE
A new application of phage in detection assays is Dual Phage technology (Wilson,
1999; Fig. 9.3). Like phage display detection methods, this again does not use the
specificity of the host-phage relationship to achieve the detection event, but uses
phage to detect the successful binding of an antibody to a specific antigen. Antibodies
specific for the target molecule are covalently linked to the surface of two different
transducing phages, each one conferring resistance to a different antibiotic. Antibody-
labelled phage are mixed with the sample and allowed to bind to the antigen. Given
the polyvalent nature of antibody binding to many targets, if this occurs the two
different transducing phage become physically linked together. Host cells for the
phage are then added and plated out on media selective for both antibiotic resistance
genes; a low multiplicity of infection of phage is used so that the probability of co-
infection by the free phage particles of different types is very low. The appearance
of many doubly resistant colonies on the plates indicates that the phage particles
were located close together while bound to the same antigenic particle, and this dual
infection signals the binding of the antibodies to their cognate antigen.
In this assay, the amplification of the signal comes through the growth of a single
doubly transduced cell into a visible antibiotic-resistant colony containing more than
108 cells, so the test has good sensitivity as each detection event becomes a visible
signal. The ability to use two different antibodies—a different antibody can be linked
to each of the transducing phage—also provides a high degree of specificity. This
Dual Phage assay has the advantage that no phage engineering is required and it can
be applied to any type of antigen for which an antibody already exists. Trials have
 276

2.
1. 3.
Transducing
Both phage Infected cells
phage
mixed with grown on agar
mixed with
target antigen plates
host cells
Positive result
1336_C09.fm Page 276 Monday, November 1, 2004 10:07 AM

Negative result

Antibody molecules Antibody binding to Only phage bound Only co-infected cells
chemically cross-linked same target antigen together via antigen can grow on media
to 2 different transducing molecule cross-links molecule co-infect cells containing both
phage phage particles antibiotics

FIGURE 9.3 Dual phage detection assay.

In this assay two transducing phage are used which transduce resistance genes to two different antibiotics into a host cell. Antibodies to the target antigen
are chemically bound to transducing phage; if antibodies to different epitopes of the target antigen are used this increases the specificity of the assay.
The antibody-labelled phage are mixed with the target antigen and allowed to bind. The sample is mixed with host cells at a suitable dilution rate to
ensure that random co-infection of host cells is a rare event. Infected host cells are plated on media containing both antibiotics to select for cells that
have been doubly transduced. This only occurs if phage are located close together while bound to the same antigen particle (colonies on agar plate
indicate a positive result).
Bacteriophages: Biology and Applications
1336_C09.fm Page 277 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 277

been reported showing that the assay could sensitively detect HIV particles in blood
samples, and it is also intended to develop sensitive dual phage assays for prions
(Wilson et al., 2002), as prion-specific antibodies have been produced, presumably
based on differences in peptide epitope conformations.

5. PHAGE AMPLIFICATION ASSAYS

Another method that has been established to detect pathogens using bacteriophage
is the Phage Amplification technology. This assay does not use any modified phage,
and the endpoint of detection is the formation of a plaque. Since no gene engineering
is required, there are no GMO containment concerns, and new tests can be rapidly
developed in response to the emergence of new problem bacteria. The technology
also uses standard microbiological methods, avoiding the need for high level training
in molecular methods and allowing the method to be easily established as part of
more traditional microbiological testing regimes.
Such a test has been described by two separate groups and is termed either the
Phage Amplification assay (Stewart et al.,1992; 1998) or Phage Amplified Biologically
assay (PhaB; Wilson et al., 1997, McNerney et al., 1998). The word amplification
refers to the fact that the initial infection of the target pathogen is allowed to develop
into a plaque, thus providing amplification of the signalling event (see Fig. 9.4). The
test begins by adding a phage specific for the target bacterium to a test sample. The
samples are then incubated long enough for the phage to infect host cells and enter
the eclipse phase. At this point a virucide is used to destroy all exogenous unadsorbed
phage; a variety of virucidal compounds have been described, including acetic acid,
ferrous ammonium sulphate, plant extracts, and essential oils. A short period of incu-
bation is allowed to ensure complete killing of all the free phage particles, before the
virucide is neutralized. The sample is mixed with plating bacteria of a propagating
strain for the phage used (termed the helper bacteria or sensor cells) and the mixture
is spread over the surface of an agar plate using a soft agar overlay. The replication
cycle of the internalized phage can now be completed, with the formation of one
plaque from each infected cell in the sample tested. Hence each plaque represents the
presence of one target bacterium in the original sample, detected by using them to
protect the phage from the action of the virucide. When testing samples that are likely
to contain a high number of organisms overgrowth of the lawn can be a problem and
some decontamination steps may be required to selectively enrich the target bacteria
(see Mole and Maskell, 2001). However, because of the dilution of the initial sample
throughout the assay, and the high number of helper cells added, problems are only
encountered when the competitive microflora reaches a level of about 106 cfu per ml
in the initial sample.
This technology has been shown to effectively detect bacterial pathogens such
as Listeria, Campylobacter, Pseudomonas, Salmonella, and Escherichia coli
(Stewart et al., 1998) but it has been most extensively exploited for the detection
of Mycobacterium tuberculosis (FASTPlaqueTB; for a review see Mole and Maskell,
2001). In this case, the phage amplification assay provides a great advantage over
conventional testing methods since M. tuberculosis is a slow-growing organism
taking up to 8 weeks to grow into visible colonies. In the FASTPlaqueTB assay,
 278

3.
1. 2.
Infected cells
Phage Virucide
mixed with
infection added
helper cells
1336_C09.fm Page 278 Monday, November 1, 2004 10:07 AM

Phage mixed Only host Phage in eclipse Infected cells lead to

with sample cells infected phase protected formation of plaques
from virucide

FIGURE 9.4 Phage amplification (or PhaB) assay.

No modification of phage are required for this assay. Phage for the target bacterium are mixed with the sample and allowed to infect susceptible cells.
Once phage have entered the eclipse phase, a virucide is added which does not affect the target bacterium. External phage which have not infected cells
are destroyed and the only viable phage which survive are those which successfully infect a host cell. The virucide is then neutralized and the sample
containing the infected cells mixed with helper cells that will support phage replication (these do not have to be the same cell type as the target bacterium).
The mixture is spread over agar plates and incubated to allow the lawn to develop. Phage within infected cells finish their replication cycle, lyse the
host cell and progeny phage infect helper cells in the lawn, leading to formation of a plaque.
Bacteriophages: Biology and Applications
1336_C09.fm Page 279 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 279

the helper organism used is the fast-growing Mycobacterium smegmatis, which is

also sensitive to the phage but can form lawns within 12–48 hours. As for the
detection of pathogens in food samples, the biggest challenge is dealing with the
inhibitory effect of the sample matrix itself, in this case most commonly clinical
sputum samples taken from patients. However, advances are being made to overcome
this problem, and Park et al. (2003) report a new method that allows the detection
of less than 100 viable M. tuberculosis cells in 1 ml of sputum sample.

6. DETECTION VIA PHAGE-MEDIATED CELL LYSIS

A commonly used rapid hygiene test for the detection of bacterial contaminants is
the measurement of ATP released from bacterial cells by firefly luciferase (see Stanley,
1989). The linear relationship between the number of ATP molecules hydrolysed and
the production of light by the enzyme allows the levels of ATP in a sample to be
determined. Since the amount of ATP per bacterial cell in a given growth condition
is quite constant (approximately 10 −15 g per cell) the amount of light can be correlated
to the number of bacterial cells present in a sample. These assays include a general
lysing reagent to break open the bacterial cells and release the intracellular ATP; thus
the results only give a measure of microbial load rather than the presence of specific
pathogens within the microflora. By using either intact phage or recombinantly
produced phage endolysins to replace the general lysing agent, the desired specificity
can be added to this test. This was demonstrated using both intact phage (Sanders,
1995) and purified phage lysin (Stewart et al., 1996) for the detection of Listeria
monocytogenes. More recently, Schuch et al. (2002) have used a cloned phage lysin
for the ATP-based detection of Bacillus anthracis spores within 10 minutes of induc-
ing germination; the assay could detect as few as 100 spores within 60 minutes of
addition of the lysin.
Due to high backgrounds, the practical limit of detection for ATP assays of
environmental or food samples is approximately 104 bacterial cells, and such high
numbers of pathogens should not be present in food samples. However, using
measurements of released adenylate kinase rather than ATP (see Corbitt et al., 2000)
in combination with phage-mediated cell lysis to add specificity, the limit of detection
has been reduced to less than 103 E. coli and Salmonella cells (Blasco et al., 1998;
Wu et al., 2001). This compares favorably with other rapid methods of bacterial
enumeration, such as immunoassays.
The use of phage to release a detectable intracellular marker has also been used
to develop other rapid methods of detection. Chang et al. (2002) used a phage specific
for E. coli O157:H7 to add specificity to a rapid method of detection based on
conductance. Impedance is a rapid method that enables detection of bacterial growth
by measuring the change in the electrical conductivity of the culture due to the
transformation of uncharged or weakly charged molecules into highly charged spe-
cies (e.g. conversion of polypeptides into individual amino acids). Detecting this
change can be achieved either directly, by measuring the change in the conductivity
of a liquid culture medium, or indirectly, by monitoring the change in the electrical
conductivity of a reaction solution due to absorption of gases from the inoculated
bacterial culture (for reviews see Silley and Forsythe, 1996; Wawerla et al., 1999).
1336_C09.fm Page 280 Monday, November 1, 2004 10:07 AM

280 Bacteriophages: Biology and Applications

The critical parameter when using such methods is the “time to detection” (i.e. when
sufficient change in the medium has occurred that a difference in the impedance can
be reliably measured). When phage were used in combination with this type of
detection method, a positive detection event was a change (increase) in the time to
detection for a growing culture when the phage was added to the sample. Simply,
if the cells present in the test sample are lysed by the phage they do not grow and
therefore do not change the conductance of the growth medium. By comparing a
phage-treated sample with an untreated sample, it is therefore possible to identify
the type of cells being detected by the conductance method. Again, the simplicity
of this is that it uses standard microbiological methods and can quickly be modified
to accommodate new bacterial groups of interest.
A similar approach has been taken by Neufeld et al. (2003) who used bacteriophage
as a specific lysing agent to allow the detection of a cytoplasmic marker protein (in this
case ß-galactosidase) by an amperometric method. This detection method is not an
impedance technique per se, but uses a specialized biosensor to specifically detect the
end product of the enzymic conversion of the ß-galactosidase substrate p-aminophenyl
ß-D-galactopyranoside (PAPG) into p-aminophenol (PAP). Using this technique, they
were able to detect E. coli cell numbers as low as 1 cfu/100 ml in less than 8 hours.
Another variation of this idea has been reported by Takikawa et al. (2002). Here, no
appropriate cellular marker protein was present, so cells of Enterobacter cloacae were
transformed with a chitinase gene. Release of this enzyme from phage-sensitive cells
was detected within 30 minutes by monitoring the hydrolysis of the chitinase substrate,
4-methylumbelliferon, using a simple spectrophotometric assay. This method of detec-
tion was proposed to track cells released into the environment for studies of bacterial
ecology.

7. USE OF PHAGE AND PHAGE PRODUCTS

AS BINDING REAGENTS
As knowledge of phage structure and function increases, more applications of phage
and phage products can be used for the detection of pathogens. Phages have evolved
to specifically recognize and bind to structures on the bacterial cell surface, and this
interaction can be exploited to develop new reagents with specific binding charac-
teristics. In its simplest form, the Salmonella phage Felix 01 (also referred to as phage
Sapphire) has been used as a biosorbant to selectively separate cells from suspensions
containing other related bacterial cell types (Bennett et al., 1997). Here, phage Felix 01
was attached onto polystyrene surfaces (both dipsticks and microtitre plates) by
passive immobilization to produce biosorbent materials. These materials were then
soaked in cultures of Salmonella and, after washing to remove nonspecifically bound
cells, the presence of Salmonella cells bound to their surface was detected either
directly (by fluorescent microscopy) or indirectly (by PCR). In both cases, Salmonella
cells were found to be specifically bound to the biosorbent surfaces, and using the
microtitre plate assay the Salmonella could be selectively enriched in the presence
of equal numbers of competing Enterobacteriaceae. Strains of Salmonella that were
not sensitive to this phage did not bind to the biosorbents. One of these strains
(Salmonella enterica serovar Arizonae) is known to synthesize incomplete LPS
1336_C09.fm Page 281 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 281

molecules, supporting the conclusion that the specificity of the assay was due to
phage-host interactions as the LPS forms the major part of the receptor for phage
Felix 01. As more detail is discovered about the specific interactions that occur
between phage and the host cell surface (see Ch. 9), phage proteins can be used to
generate recombinant binding proteins, or protein complexes, with the desired bind-
ing properties—possibly even contributing to the development of novel families of
affinity binding proteins. One such product, a tail fiber protein fragment of E. coli
phage T4, is now commercially available (EndoTrap). The protein specifically rec-
ognizes and binds to a highly conserved component of the LPS of gram-negative
bacteria. In a liquid chromatography format with the recombinant T4 tail spike
proteins immobilized on a cross-linked agarose, it is used for endotoxin removal
from medical and biological fluids.
The possibilities afforded by the recognition specificity have also been realized
to some extent through the study of phage endolysins and an understanding of the
detailed function of these proteins. The endolysins that destroy the bacterial cell
wall at the end of bacteriophage multiplication are composed of at least two distinct
domains, as discussed in Chapter 11. Besides the enzymatically active region of the
protein, highly specific domains target the enzymes to sites within the bacterial cell
wall. These cell wall binding domains (CBD) mediate tight (but non-covalent) and
extremely specific binding of the enzymes to their substrate molecules (Loessner
et al., 2002). This is an excellent example of how specific and fine-tuned the inter-
action of bacterial host and phage actually is. The extraordinary specificity and high
affinity of these polypeptides renders them ideal tools for decoration of the target
cells by fluorescently labelled CBD polypeptide domains (Loessner et al., 2002),
and can also be employed for immobilization. Paramagnetic particles coated with
anti-Listeria CBD molecules are very efficient tools for the separation and recovery
of Listeria cells from various foods; model experiments indicated recovery rates of
more than 90% even from dilute suspensions and mixed bacterial flora (Jan Kretzer
and Martin Loessner, unpublished data).

ACKNOWLEDGEMENTS
Thanks are due to past and present members of our laboratories, who so enthusias-
tically turned to the “phage side”; we are especially grateful to Siegfried Scherer
and the late Gordon Stewart for their support and encouragement and to Dr. Christine
Dodd for helpful contributions to the manuscript.

REFERENCES
Bardarov S, Dou H, Eisenach K, Banaiee N, Ya S, Chan, J, Jacobs, WR, and Riska, PF (2003).
Detection and drug-susceptibility testing of M. tuberculosis from sputum samples
using luciferase reporter phage: Comparison with the Mycobacteria Growth Indicator
Tube (MGIT) system. Diag. Microbiol. Infect. Dis., 45: 53–61.
Bennett AR, Davids FGC, Vlahodimou S, Banks JG, and Betts RP (1997). The use of
bacteriophage-based systems for the separation and concentration of Salmonella.
J. Appl. Microbiol., 83: 259–265.
1336_C09.fm Page 282 Monday, November 1, 2004 10:07 AM

282 Bacteriophages: Biology and Applications

Blasco, R, Murphy, MJ, Sanders, MF, and Squirrell DJ (1998). Specific assays for bacteria
using phage-mediated release of adenylate kinase. J. Appl. Microbiol., 84: 661–666.
Bliss JM, Sullivan MA, Malone J, and Haidaris CG (2003) Differentiation of Candida albicans
and Candida dubliniensis by using recombinant human antibody single-chain variable
fragments specific for hyphae. J. Clin. Microbiol., 41: 1152–1160.
Blissett SJ, and Stewart GSAB (1989). In vivo bioluminescent determination of apparent
Kms for aldehyde in recombinant bacteria expressing luxAB. Letts. Appl. Microbiol.,
9: 149–152.
Brown KL, Sarkis GJ, Wadsworth C, and Hatfull GF (1997). Transcriptional scilencing by
the mycobacteriophage L5 repressor. EMBO J., 16: 5914–5921.
Chang TC, Ding HC, and Chen SW (2002). A conductance method for the identification of
Escherichia coli O157: H7 using bacteriophage AR1. J. Food Protect., 65: 12–17.
Chen J, and Griffiths MW (1996). Salmonella detection in eggs using lux+ bacteriophages.
J. Food Protect., 59: 908–914.
Corbitt AJ, Bennion N, and Forsythe SJ (2000). Adenylate kinase amplification of ATP
bioluminescence for hygiene monitoring in the food and beverage industry. Letts.
Appl. Microbiol., 6: 443–447.
Funatsu T, Taniyama T, Tajima T, Tadakuma H, and Namiki H (2002). Rapid and sensitive
detection method of a bacterium by using a GFP reporter phage. Microbiol. Immunol.
46: 365–369.
Goldman ER, Pazirandeh MP, Mauro, JM, King, KD, Frey, JC, and Anderson, GP (2000).
Phage-display peptides as biosensor reagents. J. Molec. Recog., 13: 382–387.
Goodridge L, Chen J, and Griffiths M (1999). Development and characterization of a fluorescent-
bacteriophage assay for detection of Escherichia coli O157:H7. Appl Environ
Microbiol., 65: 1397–1404.
Hayhurst A, and Georgiou G (2001). High-throughput antibody isolation. Curr. Opin. Chem.
Biol., 5: 683–689.
Hennes KP, Suttle CA, and Chan AM (1995). Fluorescently labeled virus probes show that
natural virus poulations can control the structure of marine microbial communities.
Appl Environ Microbiol., 61: 3623–3627.
Hill PJ, Swift S, and Stewart GSAB (1991). PCR Based gene gggngineering of the Vibrio
harveyi lux operon and the Escherichia coli trp operon provides for biochemically
functional native and fused gene products. Molec. Gen. Genet., 226: 41–48.
Irwin P, Gehring A, Tu SI, Brewster J, Fanelli J, and Ehrenfeld E (2000). Minimum detectable
level of Salmonellae using a binomial-based bacterial ice nucleation detection assay
(BIND). J. AOAC Int., 83: 087–1095.
Jacobs WR, Barletta RG, Udani R, Chan J, Kalkut G, Sosne G, Kieser T, et al. (1993). Rapid
assessment of drug susceptibilities of Mycobacterium tuberculosis by means of
Luciferase Reporter Phages. Science, 260: 819–822.
Kodikara CP, Crew HH, and Stewart GSAB (1991). Near on-line detection of enteric bacteria
using lux recombinant bacteriophage. FEMS Microbiol. Letts., 83: 261–266.
Kuhn J, Suissa M, Wyse J, Cohen I, Weiser I, Reznick S, Lubinsky-Mink S, et al. (2002).
Detection of bacteria using foreign DNA: The development of a bacteriophage reagent
for Salmonella. Int. J. Food Microbiol., 74: 229–238.
Li Z, Szittner R, and Meighen EA (1993). Subunit interactions and the role of the luxA
polypeptide in controlling thermal stability and catalytic properties in recombinant
luciferase hybrids. Biochim Biophys Acta., 1158: 137–145.
Loessner MJ, Kramer K, Ebel F, and Scherer S (2002). C-terminal domains of Listeria
bacteriophage peptidoglycan hydrolases determine specific recognition and high
affinity binding to bacterial cell wall carbohydrates. Molec. Microbiol., 44:335–349.
1336_C09.fm Page 283 Monday, November 1, 2004 10:07 AM

Phage for the Detection of Pathogenic Bacteria 283

Loessner MJ, Krause IB, Henle T, and Scherer S (1994). Structural proteins and DNA
characteristics of 14 Listeria typing bacteriophages. J. Gen. Virol., 75: 701–710.
Loessner MJ, and Scherer S (1995). Organization and transcriptional analysis of the Listeria
phage A511 late gene region comprising the major capsid and tail sheath protein
genes cps and tsh. J. Bacteriol., 177: 6601–6609.
Loessner MJ, Rees CED, Stewart GSAB, and Scherer S (1996). Construction of luciferase
reporter bacteriophage A511::luxAB for rapid and sensitive detection of viable Listeria
cells. Appl. Environ. Microbiol., 62: 1133–1140.
Loessner MJ, Rudolf M, and Scherer S (1997). Evaluation of luciferase reporter bacteriophage
A511::luxAB for detection of Listeria monocytogenes in contaminated foods. Appl.
Environ. Microbiol., 63: 2961–2965.
Mackey BM, Cross D, and Park SF (1994). Thermostability of bacterial luciferase expressed
in different microbes. J. Appl. Bacteriol., 7: 149–154.
McNerney R, Wilson SM, Sidhu AM, Harley VS, Al Suwaidi Z, Nye PM, Parish T, and Stoker
NG (1998). Inactivation of mycobacteriophage D29 using ferrous ammonium sul-
phate as a tool for the detection of viable Mycobacterium smegmatis and M. tuber-
culosis. Res. Microbiol., 149: 487–495.
Mole RJ, and Maskell TWO’C (2001). Phage as a diagnostic—the use of phage in TB diagnosis.
J. Chem. Technol. Biotechnol., 76: 683–688.
Neufeld T, Schwartz-Mittelmann A, Biran D, Ron EZ, and Rishpon J (2003). Combined phage
typing and amperometric detection of released enzymatic activity for the specific
identification and quantification of bacteria. Anal. Chem., 75: 580–585.
Park DJ, Drobniewski FA, Meyer A, and Wilson SM (2003). Use of a phage-based assay for
phenotypic detection of mycobacteria directly from sputum. J. Clin. Microbiol., 41:
680–688.
Petrenko VA, and Vodyanoy VJ (2003). Phage display for detection of biological threat agents.
J. Microbiol. Meth., 53: 253–262.
Rader C, and Barbas CF (1997). Phage display of combinatorial antibody libraries. Curr.
Opin. Biotech., 8: 503–508.
Riska PF, Jacobs WR, Bloom BR, McKitrick J, and Chan J (1997). Specific identification
of Mycobacterium tuberculosis with the luciferase reporter mycobacteriophage:
Use of p-nitro-alpha-acetylamino-beta-hydroxy propiophenone. J. Clin. Microbiol.,
35: 3225–3231.
Sanders MF (1995). A rapid bioluminescent technique for the detection and identification
of Listeria monocytogenes in the presence of Listeria innocua. In: Campbell,
A.K., Kricka, L.J., and Stanley, P.E. (Eds), Bioluminescence and Chemilumi-
nescence: Fundamental and Applied Aspects. John Wiley & Sons, Chichester.
pp. 454–457.
Sarkis GJ, Jacobs WR, and Hatfull GF (1995). L5 Luciferase reporter mycobacterio-phages:
A sensitive tool for the detection and assay of live Mycobacteria. Molec. Microbiol.,
15: 1055–1067.
Schuch R, Nelson D, and Fischetti VA (2002). A bacteriolytic agent that detects and kills
Bacillus anthracis. Nature, 418: 884–889.
Silley P, and Forsythe S (1996). Impedance microbiology - A rapid change for microbiologists.
J. Appl. Bacteriol., 80: 233–243.
Smith GP (1985). Filamentous fusion phage—novel expression vectors that display cloned
antigens on the virion surface. Science, 228: 1315–1317.
Smith GP, and Petrenko VA (1997). Phage display. Chem. Rev., 97: 391–410.
Stanley PE (1989). A review of bioluminescence ATP techniques in rapid microbiology.
J. Biolumin. Chemilumin., 4: 375–380.
1336_C09.fm Page 284 Monday, November 1, 2004 10:07 AM

284 Bacteriophages: Biology and Applications

Stewart GSAB, Jassim SAA, Denyer SP, Park S, Rostas-Mulligan K, and Rees CED (1992).
Methods for rapid microbial detection. Patent WO 92/02633.
Stewart GSAB, Loessner MJ, and Scherer S (1996). The bacterial lux gene bioluminescent
biosensor revisited. ASM News, 62: 297–301.
Stewart GSAB, Jassim SAA, Denyer SP, Newby P, Linley K, and Dhir VK (1998). The
specific and sensitive detection of bacterial pathogens within 4 h using bacteriophage
amplification. J. Appl. Microbiol., 84: 777–783.
Stratmann J, Strommenger B, Stevenson K, and Gerlach GF (2002). Development of a peptide-
mediated capture PCR for detection of Mycobacterium avium subsp paratuberculosis
in milk. J. Clin, Microbiol., 40: 4244–4250.
Takikawa Y, Mori H, Otsu Y, Matsuda Y, Nonomura T, Kakutani K, Tosa Y, et al. (2002).
Rapid detection of phylloplane bacterium Enterobacter cloacae based on chitinase
gene transformation and lytic infection by specific bacteriophages. J. Appl. Microbiol.,
93: 1042—1050.
Turnbough CL (2003). Discovery of phage display peptide ligands for species-specific detection
of Bacillus spores. J. Microbiol. Meth., 53: 263–271.
Ulitzur S, and Kuhn J (1987). Introduction of lux genes into bacteria; a new approach for
specific determination of bacteria and their antibiotic susceptibility. In: Slomerich,
R., Andreesen, R., Kapp, A., Ernst, M., and Woods, W.G. (Eds), Bioluminescence
and Chemiluminescence new perspectives. John Wiley & Sons, New York, pp 463–472.
Ulitzur S, and Kuhn J (1989). Detection and/or identification of microorganisms in a test
sample using bioluminescence or other exogenous genetically introduced marker. US
Patent 4,861,709.
Watson BB, and Eveland WC (1965). The application of the phage-fluorescent antiphage
staining system in the specific identification of Listeria monocytogenes. J. Infect.,
115: 363–369.
Wawerla M, Stolle A, Schalch B, and Eisgruber H (1999). Impedance microbiology: Appli-
cations in food hygiene. J. Food Protect. 62: 1488–1496.
Wilson SM (1999) Analytical method using multiple virus labelling. PCT Patent WO99/63348.
Wilson SM, Al Suwaidi Z, McNerney R, Porter J, and Drobniewski F (1997). Evaluation of
a new rapid bacteriophage-based method for the drug susceptibility testing of Myco-
bacterium tuberculosis. Nat. Medicine., 3: 465–468.
Wilson SM, Lane A, Oliver J, and Stanley C (2002). Investigation of the potential of “Dual
Phage,” a novel ultra-sensitive detection label. Clin. Chem., 48: B89 Part 2 Suppl. S.
Wolber PK (1993). Bacterial ice nucleation. Adv. Micro. Physiol., 34: 203–237.
Wolber PK, and Green RL (1990). Detection of bacteria by transduction of ice nucleation
genes. Trends Biotechnol., 8: 276–279.
Wu Y, Brovko L, and Griffiths MW (2001). Influence of phage population on the phage-
mediated bioluminescent adenylate kinase (AK) assay for detection of bacteria. Letts.
Appl. Microbiol., 33: 311–315.
Zhou B, Wirsching P, and Janda KD (2002). Human antibodies against Bacillus: A model
study for detection of and protection against anthrax and the bioterrorist threat. Proc.
Natl. Acad. Sci. USA., 99: 5241–5246.