Bacteriophage-based biosorbents coupled with bioluminescent ATP assay for rapid concentration and detection of Escherichia coli

Journal of Microbiological Methods 82 (2010) 177–183

Contents lists available at ScienceDirect

Journal of Microbiological Methods j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j m i c m e t h

Bacteriophage-based biosorbents coupled with bioluminescent ATP assay for rapid concentration and detection of Escherichia coli O. Minikh a,b, M. Tolba a, L.Y. Brovko a,⁎, M.W. Griffiths a a b

Canadian Research Institute for Food Safety, University of Guelph, 43 McGilvray St., Guelph, Ontario, N1G 2W1 Canada Department of Chemical Enzymology, Lomonosov Moscow State University, 1/3, Leninskiy Gory, 119991 Moscow, Russia

a r t i c l e

i n f o

Article history: Received 19 May 2010 Accepted 23 May 2010 Available online 1 June 2010 Keywords: Immobilized T4 bacteriophage Escherichia coli ATP bioluminescence assay

a b s t r a c t Wild type T4 bacteriophage and recombinant T4 bacteriophages displaying biotin binding peptide (BCCP) and cellulose binding module (CBM) on their heads were immobilized on nano-aluminum fiber-based filter (Disruptor™), streptavidin magnetic beads and microcrystalline cellulose, respectively. Infectivity of the immobilized phages was investigated by monitoring the phage-mediated growth inhibition of bioluminescent E. coli B and cell lysis using bioluminescent ATP assay. The results showed that phage immobilization resulted in a partial loss of infectivity as compared with the free phage. Nevertheless, the use of a biosorbent based on T4 bacteriophage immobilized on Disruptor™ filter coupled with a bioluminescent ATP assay allowed simultaneous concentration and detection of as low as 6 × 103 cfu/mL of E. coli in the sample within 2 h with high accuracy (CV = 1–5% in log scale). Excess of interfering microflora at levels 60-fold greater than the target organism did not affect the results when bacteriophage was immobilized on the filter prior to concentration of bacterial cells. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Prompt and rapid detection of pathogens in food and in the environment is an important part of a food safety management strategy. Conventional methods usually take 2–5 days to detect and confirm the presence of a pathogen. A wide range of rapid technologies have been developed; with most being based on antigen and nucleic acid detection, such as ELISA and PCR (Abubakar et al., 2007). In general, however, immunoassays require 104–105 cfu/mL of the pathogen to be present in the sample for reliable detection (Gracias and McKillip, 2004; Squirrell et al., 2002) and for PCR detection, at least 200 cells per sample are needed (Fung, 2002). Furthermore, the sensitivity of both antigen– antibody assays (Gracias and McKillip, 2004) and PCR (Abubakar et al., 2007) is greatly limited by the complexity of the sample matrix and uneven distribution of the target pathogen in the sample. One of the ways to improve the sensitivity of pathogen detection is to separate and concentrate the target bacterium from the sample prior to assay. Immunomagnetic separation (IMS) is widely used as a pretreatment method for pathogen detection. However, the efficiency of separation is also affected by the complexity of the sample matrix (Tatavarthy et al., 2009). Bacteriophage-based techniques were used successfully to rapidly detect pathogens (Blasco et al., 1998; Edgar et al., 2006; Favrin et al.,

⁎ Corresponding author. Tel.: + 1 519 824 4120x58301; fax: + 1 519 763 0952. E-mail address: [email protected] (L.Y. Brovko). 0167-7012/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.mimet.2010.05.013

2003; Goodridge et al., 1999; Squirrell et al., 2002; Wu et al., 2001). They are based on the ability of phages to efficiently and specifically attach to the host bacterium and subsequently cause its lysis. Additional advantages of using bacteriophages for rapid and specific bacterial detection include: i) availability of bacteriophages specific to a wide variety of bacteria of interest that can be easily produced in large quantities; ii) increased stability of bacteriophages to temperature, pH and ionic strength when compared with antibodies; and iii) ability of bacteriophage to recognize and detect only live bacteria. Several phage-based techniques have been proposed for rapid bacterial detection and identification (Mandeville et al., 2003). One such method, which is based on the enumeration of progeny phages released following infection of the target bacterium, is called the phage amplification technique. Progeny phages can be assayed by common cultural methods (Favrin et al., 2003) or by real-time PCR (Tolba et al., 2010) with the latter method being much faster. Phagemediated lysis of the target bacterium can be also detected by monitoring the release of intracellular components into the medium. Several compounds have been proposed as indicators of cell lysis including ATP (Sanders 1994, Brovko et al., 2007) and adenylate kinase (AK) (Blasco et al., 1998; Wu et al., 2001). Using bacteriophage specific to Salmonella Typhimurium coupled with a bioluminescence ATP assay it was possible to detect the presence of the target bacterium (106 cfu/ml) in the presence of 5-fold excess of other Salmonella serovars (S. hadar, S. infantis) within 1 h (Brovko et al., 2007). The bioluminescent AK-assay allowed reliable detection of 104 cfu/mL of E. coli or Salmonella Newport within 1–2 h (Blasco et al., 1998). The results were not affected by the presence of competing

178

O. Minikh et al. / Journal of Microbiological Methods 82 (2010) 177–183

bacteria (Wu et al., 2001). Despite their great potential these methods did not demonstrate the expected sensitivity. This can be partially explained by the fact that the indicator compounds (ATP and AK) are present in substantial amounts in the sample matrix, thus compromising the sensitivity of the assays. Phage-based biosorbents have the potential to easily isolate/ concentrate pathogens from complex food matrixes. Biosorbents based on passive immobilization (Bennett et al., 1997) and on immobilization of chemically biotinylated phage (Sun et al., 2001) have been investigated. However, the efficiency of pathogen capture did not exceed 20%; probably due to the low concentration of fully infective bacteriophages on the surface of the biosorbent. As a way to improve this, several strategies were investigated for oriented bacteriophage immobilization through the head, thereby ensuring that the tails are made available for bacterial sensing. Genetic modifications of phages were performed so that they display affinity tag molecules on their heads that were able to interact with surfaces modified with the respective receptors (Tolba et al., 2008, 2010). Nucleic acid stain bound through a spacer to silica microspheres has been used to attach phage particles through their head (Boss and Lieberman, 2009). The use of positively charged modified silica particles has also been proposed to ensure strong attachment of the negatively charged bacteriophage head to the surface and hence leaving phage tails free to interact with bacterial receptors (Cademartiri et al., 2010). In all cases it was shown that oriented immobilization of bacteriophage through the head resulted in the construction of efficient biosorbents capable of separating target bacteria from complex sample matrices; thus decreasing the background concentration of the lysis-indicator compounds. The constructed biosorbents have demonstrated some lytic activity. However, the effect of phage immobilization on phage infectivity was not investigated sufficiently in many of these studies. Therefore it is difficult to assess the applicability of the constructed phage-based biosorbents for simultaneous capture and detection of target bacteria. The goal of the present work was to investigate infectivity of immobilized bacteriophages and to explore their use for rapid capture and bioluminescent detection of the host bacterium. As a model system T4 bacteriophage with its host bacterium E. coli B were used. Three different biosorbents were constructed: i) genetically modified T4 bacteriophage displaying biotin binding peptide (BCCP-T4) was immobilized on streptavidin-coated magnetic beads; ii) genetically modified T4 bacteriophage displaying a cellulose binding module (CBM-T4) was immobilized on microcrystalline cellulose beads; and iii) wild type T4 was immobilized on a novel capture medium based on alumina nanofibers (Disruptor® filter media). Disruptor® is a commercially available highly electropositive nonwoven filter media with a zeta potential of N50 mV at pH7.2. It has excellent loading capacity due to the availability of more than 42,000 m2 of electropositive fiber surface area per square meter of filter medium. It was shown to capture 98% of MS2 bacteriophage from water (Komlenic, 2007). Due to the high positive charge and high surface area it was expected to ensure bacteriophage capture through electrostatic interactions with the negatively charged head, similarly to modified silica particles (Cademartiri et al., 2010), but with greater efficiency. The infectivity of immobilized bacteriophages was investigated and possible applications of the constructed biosorbents for bacterial detection were explored. 2. Materials and methods

bacteriophages displaying biotin binding peptide (BCCP-T4) and cellulose binding module (CBM-T4) were obtained previously in our laboratory (Tolba et al., 2010). Bacterial strains were cultured at 37 °C, with shaking at 200 rpm (NBS Benchtop Incubator Shaker, New Brunswick Scientific Co., Inc., New Brunswick, Canada) for 14–16 h in Luria–Bertani broth supplemented with 100 µg/mL of ampicillin in the case of bioluminescent E. coli B. For ATP assay, Minimal Broth Davis (BD Diagnostic Systems, Sparks, MD) supplemented with 1% dextrose (Sigma-Aldrich, Oakville, ON) was used for bacterial cultivation in order to have lower background ATP levels. Enumeration of bacterial cells was performed by plate count and presented in colony forming units (cfu). Propagation of bacteriophages was performed using the soft agar overlay technique (Sambrook and Russell, 2001). Obtained bacteriophages were purified by dialysis against phosphate buffered saline (PBS, 10 mM, pH 7.2, Fisher Scientific, Oakville, Canada), enumerated by plaque assay and results were presented as plaque forming units (pfu). 2.2. Bacteriophage immobilization Immobilization of the wild type and recombinant bacteriophages was performed by incubation of phage sample with the appropriate amount of the solid support at room temperature (RT) and with gentle mixing. To immobilize wild type T4 bacteriophage on Disruptor® filters (diameter 1 cm, Ahlstrom Corp., Mt. Holly Springs, PA), 50 µL of phage (1.7 × 1010 pfu/mL) was placed on top of the filter and left at room temperature for 15–40 min before use. Unbound phage particles were removed by filtering 3 × 1 ml aliquots of PBS through the membrane. The biotinylated BCCP-T4 bacteriophage (1 ml, 109 pfu/mL) was incubated overnight at room temperature with streptavidin-coated magnetic beads (200 µL of Dynabeads® MyOne™ Streptavidin C1, Dynal, Raleigh, NC). The CBM-T4 bacteriophage (1 ml, 109 pfu/mL) was incubated at the same conditions with a 1 mL suspension of microcrystalline cellulose beads (100 mg/mL, CP-102, Asthi Kasei Chem. Corp., Tokyo, Japan). In both instances, non-bound phage was removed by washing 3 times with 1 ml of PBS supplemented with 0.05% Tween-80 (PBS Tween (Fisher Scientific, Pittsburgh, PA) and then twice with PBS alone. Magnetic beads were collected by a magnetic particle separator (Boehringer Manheim Canada, Laval, Quebec, Canada) and cellulose beads were separated by sedimentation followed by re-suspension in 1 ml of PBS. These phage-based biosorbents were used immediately for bacterial capture. 2.3. Bacteria capture by biosorbents To evaluate bacterial capture by phage-coated magnetic and cellulose beads 1 ml aliquots of bacterial suspensions in PBS in the concentration range 1 × 103 to 1 × 107 cfu/mL were mixed with 0.1 mL of the respective biosorbent and incubated at room temperature with gentle mixing for 30 min. After incubation bacteria were enumerated by plate counting of both the initial suspensions (cfu/mLinitial) and the supernatants (cfu/mLsupernatant) and efficiency of capture was calculated using the following equation (Tolba et al., 2010). %capture =

ðcfu=mLÞinitial −ðcfu=mLÞsupernatant ðcfu=mLÞinitial

× 100

2.1. Bacterial strains and bacteriophages E. coli B (ATCC 11303), Salmonella Typhimurium strain C1058, bacteriophage T4, were obtained from the culture collection of the Canadian Research Institute for Food Safety (CRIFS). Bioluminescent E. coli B (lux) carrying entire luxCDABE operon from Photorhabdus luminescens was obtained as described in Sun et al. (2002). Recombinant T4

To evaluate the number of bacteria captured by phage-coated Disruptor® filter, 10 mL of the prepared bacterial suspensions in the concentration range 10–1 × 107 cfu/mL was passed through Disruptor® filters that were placed in 1 cm diameter filter holders (Millipore, Billerica, MA). The total time of contact was ∼2 min. Bacteria in the initial suspension and filtrate were enumerated by plate count. The

O. Minikh et al. / Journal of Microbiological Methods 82 (2010) 177–183

efficiency of bacterial capture was calculated using the equation given above. In all cases the respective support without the attached phage was used as a control.

179

to 3 × SD (standard deviation) of the average background value. The accuracy of the assay was presented as a coefficient of variation (CV) calculated as the average ratio of SD (in log scale) to the mean log RLU for each bacterial concentration.

2.4. In vivo bioluminescence In vivo bioluminescence was used to monitor bacterial growth in the sample. It was measured by placing a tube with 100–300 µL aliquots of bacterial suspension into a BG-P tube luminometer (GEM Biomedical, Hamden, CT) and recording bioluminescence intensity in Relative Light Units (RLU) using a 10 s signal integration time. When the sample contained magnetic or cellulose beads with captured bacteria, it was thoroughly mixed before measuring to ensure even distribution of beads within the sample volume. Filter membranes with captured bacteria were placed to the side of the test tube opposite the detector in order to prevent interference with light recording. The background signal was recorded in triplicate under identical conditions but without phage and bacteria present. The average background signal was subtracted from the sample reading prior to further data processing. 2.5. ATP bioluminescence assay for monitoring phage-mediated bacteria lysis Bacterial samples (0.3 mL) mixed with bacteriophage suspensions or phage-based biosorbents with captured bacteria in 0.1 mL fresh Minimal Broth Davis supplemented with 1% dextrose were incubated at 37 °C with gentle shaking. Fifty microliter aliquots were taken from the sample during incubation and the extracellular ATP released due to bacterial lysis was measured using the bioluminescent assay. For the measurements of ATP, a bioluminescent luciferin–luciferase reagent (BLR) Aqua Trace (Biotrace, Bridgend, UK) was used according to manufacturer's instructions. The sample (50 µL) was mixed with 50 µL of BLR and the intensity of bioluminescence in relative light units (RLU) was recorded immediately using a BG-P tube luminometer (GEM Biomedical, Hamden, CT) with 10 s integration time. After that 10 µL of standard ATP solution was added to the same tube and the resulting bioluminescence intensity was recorded. The obtained signals were corrected for background using the method mentioned above. Internal standardization was used to calculate the concentration of ATP in the sample according to Romanova et al. (2003). 2.6. Detection of E. coli B using phage-based biosorbent and bioluminescent ATP assay For the detection of E. coli B, cell suspensions in PBS were prepared in the concentration range of 102–107 cfu/mL. Three different experimental formats were used to assess the sensitivity and reproducibility of the bioluminescent phage-mediated detection of E. coli. In the first, 10 mL of bacterial samples was filtered through Disruptor® filters with bacteriophage immobilized on it. Filters with the captured bacteria were placed into 50 μL of Minimal Broth Davis for 2 h at 37 °C. ATP bioluminescence was measured for all samples using the method described above and calibration curves of RLU vs cfu/mL were constructed. In the second format, 10 mL of bacterial samples was filtered through the Disruptor® filter that did not contain immobilized phage. After that the filter was treated with 50 μL of phage solution (1010 pfu/mL), incubated for 2 h at 37 °C and ATP bioluminescence was measured. In the third case, 50 μL of the bacterial sample was treated with 50 μL of phage solution (1010 pfu/mL), incubated for 2 h at 37 °C and ATP bioluminescence was measured. Using the obtained data the calibration curves log RLU vs log cfu/mL were constructed. The sensitivity and accuracy of the assays were assessed using the method described by Harris (2007). The limit of detection (LOD) for the assay was estimated as the level of bacteria corresponding to the signal equal

2.7. Data analysis All experiments were performed at least three times, and in each experiment duplicate/triplicate measurements were taken. From the combined set of data the average values and standard deviation (SD) were calculated using the Data Analysis Tool in Microsoft Excel. The significance of treatment differences was estimated using a t-test at 95% level of confidence. 3. Results 3.1. Inhibition of growth of E. coli B (lux) by wild type T4, BCCP-T4 and CBM-T4 phages To compare infectivity of recombinant bacteriophages with the wild type, growth inhibition of E. coli B (lux) in the presence of bacteriophages was investigated using in vivo bioluminescence as an indicator of the number of live E. coli cells present in the sample. E. coli B (lux) cells (2 × 104 cfu/mL) were mixed with the respective bacteriophage at a multiplicity of infection (moi) of 300 and bacterial growth was monitored for 5 h by measuring in vivo bioluminescence. The results are presented in Fig. 1. For the control sample without phage added a lag period of approximately 2 h was observed followed by fast exponential growth of bacteria. Addition of wild type T4 bacteriophage totally inhibited E. coli growth for at least 5 h. Both T4BCCP and T4-CBM recombinant phages significantly inhibited growth of E. coli, however after 5 h incubation a significant bioluminescent signal was recorded, indicating a substantial increase in the number of live cells present in the samples as compared to the initial inoculum. These results confirm previous observation that the genetic modification of T4 bacteriophage had an effect on infectivity by reducing the burst size of the virus and prolonging the latent period (Tolba et al., 2010). The reported latent period for the wild type T4 bacteriophage was 25 min, with a burst size of 137 pfu, while both BCCP-T4 and CBM-T4 demonstrated significantly lower burst size of 35 and 19 pfu, respectively, and the latent period was slightly longer (30 and 35 min, respectively). 3.2. Phage-mediated lysis of E. coli B by wild type and recombinant T4 bacteriophages in suspension Phage-mediated lysis of E. coli B by wild type and recombinant T4 bacteriophages in suspension was monitored by the release of ATP. It

Fig. 1. Growth curves of E. coli B (lux) (2 × 104 cfu/mL) measured by in vivo bioluminescence in the absence (○) and the presence of wild type T4 bacteriophage (×), CBM-T4 (♦), BCCP-T4 (▲) at moi 300.

180

O. Minikh et al. / Journal of Microbiological Methods 82 (2010) 177–183

was shown that for wild type T4 bacteriophage there was a 1 h lag period followed by a significant increase of extracellular ATP during incubation of host cells with the phage at a moi of 140 and this increase reached a maximum after 3 h (Fig. 2). This was similar but slightly slower when compared with the time course of ATP release reported for other bacteriophages (Sanders, 1994, Brovko et al., 2007). For both Salmonella and Listeria-specific bacteriophages the maximum phage-mediated ATP release was reached after 90–120 min of incubation. On the contrary, the time course of ATP release from E. coli cells treated with biotinylated BCCP-T4 phage showed a 3 h lag period and the amount of the released ATP after 3.5 h incubation was 3-fold lower than that obtained with wild type T4 phage. No saturation of the signal was observed within the 4 h period. There was no significant difference in the amount of extracellular ATP released by E. coli cells treated with CBM-T4 phage and that from the non-treated bacterial culture. The results indicated that the recombinant phages exhibit lower infectivity, which resulted in less efficient cell lysis.

higher number of live cells captured by the Disruptor™ filter was due to the fact that the pore size of this material is less than 2 µm and it is positively charged, thus allowing physical capture of N99% of bacterial cells present in the sample. The initial level of bioluminescence observed for magnetic and cellulose beads was N100 fold less, which partly could be explained by the lower capture efficiency and, to some extent, by interference of beads with bioluminescence detection (Sun, et al., 2001). Nevertheless, for all biosorbents there was no significant difference in initial bioluminescence between phage-based biosorbents and their respective controls, indicating that non-specific absorption played a significant role in capture of bacterial cells. Growth of the cells captured by phage-based biosorbents was inhibited when compared with growth of the cells non-specifically absorbed by the respective support (Fig. 3). The more cells captured by the phage-based biosorbent the more significant was the growth inhibition observed. For the Disruptor™

3.3. Efficiency of immobilized T4-bacteriophages in capture and growth inhibition of host bacteria In our previous publication it was shown by culture methods that magnetic biosorbents based on BCCP-T4 bacteriophages were able to capture up to 90% of host bacteria present in the sample during a short 10 min contact (Tolba et al., 2008). At the same time significant nonspecific capture of E. coli cells was observed for both streptavidin magnetic beads and cellulose beads used for immobilization of BCCP-T4 and CBM-T4, respectively. It was not clear if the cells captured both specifically by immobilized phage and adsorbed non-specifically on the surface of the biosorbent would be infected by immobilized phage or by progeny phages. Growth inhibition of E. coli B (lux) cells captured by phage-based biosorbents was investigated to test the ability of biosorbents to simultaneously capture and lyse the host bacterium. T4, BCCP-T4, and CBM-T4 bacteriophages were immobilized on Disruptor® filter media, streptavidin magnetic beads and cellulose beads, respectively, using the methods described above. Streptavidincoated magnetic beads and microcrystalline cellulose beads as well as Disruptor™ filters without phages were used as biosorbents for comparison. Biosorbents containing approximately the same amount of phage and biosorbents without phage were brought in contact with one mL of E. coli B (lux) (1–2 × 106 cfu/mL) suspension for 30 min. The non-bound bacterial cells were removed by washing and the biosorbents with captured bacteria were placed in fresh nutrient medium and incubated at 37 °C. In vivo bioluminescence was monitored for 3–5 h to detect the changes in number of live E. coli cells present in the sample. At time zero, the highest level of bioluminescence was recorded for the Disruptor™ filter media (6–7 × 104 RLU), followed by cellulose beads (250–450 RLU) and magnetic beads (∼140 RLU). The significantly

Fig. 2. Time course of phage-mediated ATP release from E. coli B cells (7 × 104 cfu/mL). Conditions: cells treated with wild type T4 bacteriophage (×), BCCP-T4 (▲), CBM-T4 (♦), and control sample containing no phage (○) at moi 140.

Fig. 3. Growth of E. coli B (lux) (2 × 106 cfu/mL) captured by biosorbents. A. Growth of E. coli B (lux) captured by the Disruptor™ without phage on it (1) and with phage immobilized on the filter (2); B. growth of E. coli B (lux) captured by the cellulose beads (10 mg/mL) (1) and cellulose beads coated with CBM-T4 bacteriophage (2); C. growth of E. coli B (lux) captured by the streptavidin magnetic beads (2 × 108 beads/mL) (1) and magnetic beads coated with BCCP-T4 bacteriophage(2 × 108 beads/mL) (2).

O. Minikh et al. / Journal of Microbiological Methods 82 (2010) 177–183

phage-based biosorbent no noticeable growth of bioluminescent E. coli was observed during the 5 h incubation. The level of bioluminescence decreased 6-fold compared to the initial level indicating a significant drop in the number of live cells and, hence, efficient phage-mediated lysis. At the same time in the control sample containing cells captured by the filter in the absence of phage, the level of bioluminescence increased 20-fold due to the extensive growth of bacteria. Bioluminescence of bacteria captured by the original cellulose beads increased 10-fold during the 3 h incubation, while with the phage-based cellulose biosorbent the registered bioluminescence increase was only 7-fold, indicating significant growth inhibition. A similar effect was observed for biosorbents based on streptavidin-coated magnetic beads, on which the increase of bioluminescence for bacteria captured by magnetic beads was 11-fold. The corresponding value for phage-containing magnetic beads was only 7-fold. However, due to the lower bioluminescent signals detected in the presence of magnetic beads, which were close to the sensitivity limit, the difference between biosorbents in this case was statistically insignificant. A one hundred-fold increase in the number of phage particles used for immobilization did not result in a biosorbent with a significantly different potential for growth inhibition (data not shown). 3.4. Lytic activity of the immobilized bacteriophages detected by the release of intracellular ATP Lytic activity of the immobilized bacteriophages was analyzed by monitoring the release of intracellular ATP from the cells captured by the biosorbent using a bioluminescent ATP assay. The experimental protocol was similar to that described in the previous section, only instead of in vivo bioluminescence, extracellular ATP was measured after 2 h incubation using the bioluminescent ATP assay. For both magnetic and cellulose bead-based biosorbents the level of detected ATP bioluminescence was not significantly different from the background signal when the initial E. coli concentration used in the experiment was in the range 103–107 cfu/mL (data not shown). Thus, even for the highest E. coli concentration tested the lytic activity of the immobilized phages was either too low to detect, or the lysis was very slow and the released ATP was hydrolyzed during the incubation. On the contrary, for the phage biosorbent based on the Disruptor® filter, high levels of ATP bioluminescence were detected and the bioluminescence readings obtained were directly proportional to the initial concentration of E. coli in suspension. Based on these findings Disruptor®-based biosorbent was used in further experiments to develop the assay to detect and enumerate E. coli B. 3.5. E. coli detection using phage-based biosorbent coupled with bioluminescent ATP assay To investigate if the filter-based biosorbent could be used for simultaneous capture and detection of E. coli, the three experimental formats described in the Materials and methods section were used and compared. The obtained calibration curves are presented in Fig. 4. For the phage-based biosorbent (curve 1, Fig. 4) linear dependence of the recorded RLU values on cfu/mL was observed in double log coordinates for the entire concentration range used (R2 = 0.986). All the readings were corrected for the background signal. The limit of detection (LOD) was 3.81 log (cfu/mL), which corresponded to ∼6 × 103 cfu/mL. The average accuracy (reproducibility) of the method represented by the coefficient of variation (CV) was (5± 3)% in log scale. For the second method, when both bacteria and phage were applied subsequently to the filter, the calibration graph consisted of 2 linear regions with regression coefficients of 0.989 and 0.949, for cell concentrations b106 cfu/mL and N106 cfu/mL, respectively (curve 2, Fig. 4). The slope of the first part corresponding to the lower cell concentration was 0.954, while at higher cell concentration the observed slope was almost halved (0.474). The lower slope could be explained by

181

Fig. 4. Dependence of ATP bioluminescent signal (RLU) on initial concentration of E. coli in the sample: (2) 10 mL of bacterial suspension passed through the Disruptor™ filter and treated with 50 µL of T4 bacteriophage (1010 pfu/mL); (1) 10 mL bacterial suspension passed through the Disruptor™ filter with T4 bacteriophage immobilized on it; (3) 50 µL of bacterial suspension treated with 50 µL T4 bacteriophage in solution (1010 pfu/mL). Dashed line represents LOD for the lines 2 and 3, solid line represents LOD for the line 1.

the lower moi at higher cell concentration, which results in a slower cell lysis. Nevertheless, estimated LOD for this method was 2.75 log cfu/mL, which corresponded to ∼600 cfu/mL and was significantly lower than for the first method due to the much lower SD observed for the background signal in this case (0.1 in log scale). The average CV of the method was (1 ±1)% in log scale. When both bacteria and phage were brought together in suspension without prior concentration by the biosorbent or filter, the calibration curve also had two linear regions with regression coefficients of 0.963 and 0.994 for concentrations b105 cfu/mL and N105 cfu/mL, respectively (curve 3, Fig. 4). Unlike in the previous case, the slope for the lower concentration range was only 0.348 and only reached the limit of 1.11 at higher concentrations. Taking into consideration the SD of background signal for this method (0.1 in log scale), the LOD was estimated to be ∼1.6 × 103 cfu/mL and CV= (0.7 ± 0.4)% in log scale. However, due to the low slope of the calibration curve the quantitative assay was considered unreliable for bacterial concentrations below 105 cfu/mL. The results showed that the concentration of the cells on the biosorbent/filter prior to the ATP assay unfortunately did not result in the expected improvement in the sensitivity of detection. Though the initial bioluminescent readings for the concentrated samples were 100-fold higher than for the suspension, the background signal for these assays increased also, thus compromising the overall sensitivity. Probably further optimization of the procedure could alleviate this problem. In order to assess the selectivity of the proposed assays mixed samples containing E. coli B and Salmonella Typhimurium were prepared and analyzed using the experimental protocols described for pure culture. The concentration of E. coli in the mixtures was constant in all samples (∼7 × 104 cfu/mL) with Salmonella cells being added in the range of 0.1- to 60-fold relative to E. coli numbers. For all the samples, where E. coli cells were in excess (ratio of Salmonella to E. coli from 0.1 to 0.9) there was no significant difference in the bioluminescence signals obtained for pure cultures of E. coli when compared with the mixed cultures (data not shown). For the samples where Salmonella cells were in excess, the presence of interfering cells resulted in differences in the measured bioluminescence signal depending on the experimental format used (Fig. 5). When the assay was performed by mixing bacteria with phage in suspension and when bacterial cells concentrated on the Disruptor® filter were treated with the suspension of the phage, the

182

O. Minikh et al. / Journal of Microbiological Methods 82 (2010) 177–183

Fig. 5. Dependence of the ATP bioluminescence signal on the relative excess of Salmonella versus E. coli B cells in the sample: (1) 50 µL of bacterial suspension treated with 50 µL T4 bacteriophage in solution (1010 pfu/mL); (2) 10 mL of bacterial suspension passed through the Disruptor™ filter and treated with 50 µL of T4 bacteriophage (1010 pfu/mL); (3) 10 mL of bacterial suspension passed through the Disruptor™ filter with T4 bacteriophage immobilized on it. Concentration of E. coli B in all samples is 7 × 104 cfu/mL.

observed bioluminescent signal almost linearly depended on the level of excess of Salmonella versus E. coli (lines 1 and 2, Fig. 5). On the contrary, when bacteriophage was immobilized on the filter prior to bacterial concentration and bioluminescence detection, the observed bioluminescence signal did not change for the samples with excess of Salmonella up to 60-fold (line 3, Fig. 5). The obtained results demonstrate the high selectivity of the assay using the phage-based biosorbent. 4. Discussion Phage-based methods for bacterial detection have ‘built-in’ signal amplification steps. The extent and mechanism of amplification depend on the experimental format used for detection. In phage amplification assays each cell infected by lytic virus releases a certain number of progeny phage particles (dependant on the burst size) into the media as a result of infection. The released phages could infect the remaining bacterial cells and produce a new generation of phage particles. Thus, the increment in phage numbers is directly related to initial numbers of target cells and burst size of the phage used. With time the number of phages increases exponentially until all the bacterial cells are lysed. If the number of phage particles present initially is high enough to infect all the bacteria present, the lysis of the entire cell population proceeds rapidly (within an hour). Usually this happens at a high multiplicity of infection (moi) in the range 10–100 (Wu et al., 2001). High excess of phage particles, however, interferes with accurate quantification of the increase in phage numbers. To destroy endogenous phage, virucidal agent is usually added at the first stage of infection when phage DNA is already injected into the cell, but before lysis occurs (Stewart et al., 1998). This step requires very thorough optimization of timing and concentration, which are characteristic for each bacteriophage-target cell pair. Use of phage-based biosorbents could provide an easy means of removing exogenous phage prior to quantification, and thus facilitate development of a rapid, sensitive and simple method for bacterial detection. However, two of the three phage-based biosorbents investigated in this work were shown to have a diminished burst size (Tolba et al., 2010) and were considered not optimal for the phage amplification assay formats. When release of intracellular compounds by phage-mediated lysis is used for bacterial detection the extent of signal amplification depends on the amount of indicator compound within the bacteria, its stability in the solution and on the sensitivity of the assay employed for its detection. Bioluminescent methods are known to be among the most sensitive detection techniques, and intracellular content of ATP—the major energy source in the cell, is relatively constant and high (millimolar range). Thus, theoretically, even a single cell can be detected

using the bioluminescent ATP assay. However, in reality the sensitivity of the method is compromised by the relatively high concentration of ATP in regular nutrient medium as well as by hydrolysis of the released ATP by bacterial ATP-ases that are active even after cell lysis. To achieve the highest possible sensitivity the assay should be performed in chemically defined media with negligible ATP background, and the time course of phage-mediated ATP release should be investigated to identify the end-point of the assay that provides the maximum possible ATP bioluminescence signal. Taking this information into consideration infectivity of both wild type and recombinant T4 bacteriophages in solution and in immobilized form was investigated to assess the feasibility of use of phagebased biosorbents–biosensors for the concentration and detection of host bacteria using phage-mediated lysis assays. Immobilization of the phage on the Disruptor® filter, that provided high density, oriented attachment of the phage particles to the surface through their heads, allowed detection of as low as ∼103 cfu/mL of E. coli in the sample within 2 h with high accuracy. Both recombinant phages did not perform well as biosorbents, probably because of the lower density of the immobilized phage on the surface of the beads which resulted in insignificant capture of bacteria and, hence, undetectable lysis. An additional advantage of using a phage-based biosorbent for concentration and detection of E. coli was observed when a mixed culture was used for the assay. Excess of interfering microflora at levels 60-fold greater than the target organism did not affect detection, thus ensuring the high selectivity of the proposed assay. When both the bacterial mixed culture and bacteriophage were in suspension and when bacteria captured by the filter were subsequently treated by bacteriophage the selectivity of the assay was low, as the recorded signal increased with the increased amount of interfering bacteria. The nature of the improved selectivity for the biosorbent is not clear. One of the possible reasons could be that the bacteriophage immobilized on the positively charged matrix through the negatively charged head has an optimal orientation for the efficient capture of the host bacteria followed by fast lysis. At the same time, if the interfering bacterium comes in contact with the immobilized phage, the orientation does not support non-specific capture and lysis. On the other hand, when bacteria and phage are in solution or when a high concentration of phage is applied to the mixture of bacteria captured on the filter, the phage could possibly attach non-specifically to the non-host bacteria and cause its lysis. Though the probability of such an event is very low, at a high moi (N105) non-specific lysis could add to the concentration of the indicator compound and, hence, decrease the selectivity of the assay. Work continues to develop a method for bacteriophage immobilization that does not interfere with phage infectivity, and will allow further improvements to the sensitivity and selectivity of bacteriophage-based bacterial assays. Acknowledgments The work was partly supported by SENTINEL—Bioactive paper network (NSERC) and the Ontario Ministry of Agriculture, Food and Rural Affairs (OMAFRA). The authors would like to thank Argonide Corporation and Ahlstrom Corporation for the generous gift of Disruptor™ 4601 Filter materials. References Abubakar, I., Irvine, L., Aldus, C.F., Wyatt, G.M., Fordham, R., Schelenz, S., Shepstone, L., Howe, A., Peck, M., Hunter, P.R., 2007. A systematic review of the clinical, public health and cost-effectiveness of rapid diagnostic tests for the detection and identification of bacterial intestinal pathogens in faeces and food. Health Technol. Assess. 11. Bennett, A.R., Davids, F.G.C., Vlahodimou, S., Banks, J.G., Betts, R.P., 1997. The use of bacteriophage-based systems for the separation and concentration of Salmonella. J. Appl. Microbiol. 83, 259–265. Blasco, R., Murphy, M.J., Sanders, M.F., Squirrell, D.J., 1998. Specific assays for bacteria using phage mediated release of adenylate kinase. J. Appl. Microbiol. 84, 661–666.

O. Minikh et al. / Journal of Microbiological Methods 82 (2010) 177–183 Boss, P.A., Lieberman, S.H., 2009. Technique for orienting and binding phage for bacteria detection. United State Patent, US 7,632,637 B1. Brovko, L.Y., Romanova, N.A., Mandeville, R., Allain, B., Griffiths, M.W., 2007. Detection and control of Salmonella Typhimurium growth using specific bacteriophage. In: Szalay, A.A., Hill, P.J., Kricka, L.J., Stanley, P.E. (Eds.), Bioluminescence and Chemiluminescence. Chemistry, Biology and Applications. World Scientific, Singapore, pp. 75–78. Cademartiri, R., Anany, H., Gross, I., Bhayani, R., Griffiths, M., Brook, M.A., 2010. Immobilization of bacteriophages on modified silica particles. Biomaterials 31, 1904–1910. Edgar, R., McKinstry, M., Hwang, J., Oppenheim, A.B., Fekete, R.A., Giulian, G., Merril, C., Nagashima, K., Adhya, S., 2006. High-sensitivity bacterial detection using biotintagged phage and quantum-dot nanocomplexes. Proc. Natl. Acad. Sci. U. S. A. 103, 4841–4845. Favrin, S.J., Jassim, S.A., Griffiths, M.W., 2003. Application of a novel immunomagnetic separation-bacteriophage assay for the detection of Salmonella enteritidis and Escherichia coli O157:H7 in food. Int. J. Food Microbiol. 85, 63–71. Fung, D.Y.C., 2002. Rapid methods and automation and microbiology. Compr. Rev. Food Sci. Food Safety 1, 3–22. Goodridge, L., Chen, J., Griffiths, M., 1999. The use of a fluorescent bacteriophage assay for detection of Escherichia coli O157:H7 in inoculated ground beef and raw milk. Int. J. Food Microbiol. 47, 43–50. Gracias, K.S., McKillip, J.L., 2004. A review of conventional detection and enumeration methods for pathogenic bacteria in food. Can. J. Microbiol. 50, 883–890. Harris, D.C., 2007. Quality Assurance and Calibration. In: Harris, D.C. (Ed.), Quantitative Chemical Analysis. W. H. Freeman and Company, New York, pp. 78–92. Komlenic, R., 2007. Water filtration media. Talking about revolution? Filtr. Sep. 44 (5), 26–29. Mandeville, R., Griffiths, M.W., Goodridge, L., McIntyre, L., Ilenchuk, T.T., 2003. Diagnostic and therapeutic applications of lytic phages. Anal. Lett. 36 (15), 3241–3259. Romanova, N.A., Brovko, L.Y., Moore, L., Pometun, E., Savitsky, A.P., Ugarova, N.N., Griffiths, M.W., 2003. Assessment of photodynamic destruction of Escherichia coli

183

O157:H7 and Listeria monocytogenes by using ATP bioluminescence. Appl. Environ. Microbiol. 69, 6393–6398. Sambrook, J., Russel, D.W., 2001. Molecular Cloning:. A Laboratory Manual, Vol. 1, Cold Spring Harbor Laboratory Press, New York, pp. 5.4–5.13. Sanders, M.F., 1994. A rapid bioluminescent technique for the detection and identification of Listeria monocytogenes in the presence of Listeria inocua. In: Campbell, A.K., Kricha, L. J., Stanley, P.E. (Eds.), In Bioluminescence and Chemiluminescence. Fundamentals and Applied Aspects. John Wiley & Sons, Chichester, pp. 454–457. Squirrell, D.J., Price, R.L., Murphy, M.J., 2002. Rapid and specific detection of bacteria using bioluminescence. Anal. Chim. Acta 457, 109–114. Stewart, G.S.A.B., Jassim, S.A.A., Denyer, S.P., Newby, P., Linley, K., Dhir, V.K., 1998. The specific and sensitive detection of bacterial pathogens within 4 h using bacteriophage amplification. J. Appl. Microbiol. 84, 777–783. Sun, W., Brovko, L., Griffiths, M., 2001. Use of bioluminescent Salmonella for assessing the efficiency of constructed phage-based biosorbent. J. Ind. Microbiol. Biotechnol. 27, 126–128. Sun, W., Khosravi, F., Albrechtsen, H., Brovko, L.Y., Griffiths, M.W., 2002. Comparison of ATP and in vivo bioluminescence for assessing the efficiency of immunomagnetic sorbents for live Escherichia coli o157:H7 cells. J. Appl. Microbiol. 92, 1021–1027. Tatavarthy, A., Peak, K., Veguilla, W., Cutting, T., Harwood, V.J., Roberts, J., Amuso, P., Cattani, J., Cannons, A., 2009. An accelerated method for isolation of Salmonella enterica serotype Typhimurium from artificially contaminated foods, using a short preenrichment, immunomagnetic separation, and xylose–lysine–desoxycholate agar (6IX method). J. Food Prot. 72, 583–590. Tolba, M., Brovko, L.Y., Minikh, O., Griffiths, M.W., 2008. Engineering of bacteriophages displaying affinity tags on its head for biosensor applications. NSTI Nanotech 2008 conference, Vol. 2, pp. 449–452. Boston, USA. Tolba, M., Minikh, O., Brovko, L.Y., Evoy, S., Griffiths, M.W., 2010. Oriented immobilization of bacteriophage for biosorbent applications. Appl. Environ. Microbiol. 76, 528–535. Wu, Y., Brovko, L., Griffiths, M.W., 2001. Influence of phage population on the phagemediated bioluminescent adenylate kinase (AK) assay for detection of bacteria. Lett. Appl. Microbiol. 33, 311–315.