Long-amplicon propidium monoazide-PCR enumeration assay to detect viable Campylobacter and Salmonella

Authors


Correspondence

Michele I. Van Dyke, Department of Civil and Environmental Engineering, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L 3G1, Canada. E-mail: mvandyke@uwaterloo.ca

Abstract

Aims

The effect of amplicon length on the ability of propidium monoazide-PCR (PMA-PCR) to reliably quantify viable cells without interference from dead cells was tested on heat- and ultraviolet (UV)-killed Salmonella enterica and Campylobacter jejuni, two important enteric pathogens of concern in environmental, food and clinical samples.

Methods and Results

PMA treatment followed by quantitative PCR (qPCR) amplification of short DNA fragments (<200 bp) resulted in incomplete signal inhibition of heat-treated Salm. enterica (3 log reduction) and Camp. jejuni (1 log reduction), whereas PCR amplification of a long DNA fragment (1·5 and 1·6 kb) completely suppressed the dead cell signal. PMA pretreatment of UV-irradiated cells did not affect PCR amplification, but long-amplicon PCR was shown to detect only viable cells for these samples, even without the addition of PMA.

Conclusions

The long-amplicon PMA-PCR method was effective in targeting viable cells following heat and UV treatment and was applicable to enteric pathogens including Salmonella and Campylobacter that are difficult to enumerate using culture-based procedures.

Significance and Impact of the Study

PCR amplicon length is important for effective removal of the dead cell signal in PMA pretreatment methods that target membrane-damaged cells, and also for inactivation mechanisms that cause direct DNA damage.

Introduction

The quantification of pathogenic micro-organisms is important in food, environmental and clinical samples to identify their contribution to public health. The health risk caused by pathogens requires reliable and sensitive detection methods, with the additional challenge to specifically detect viable cells. Classical growth-based methods can underestimate the viable cell count. Many types of bacteria can enter a viable but nonculturable (VBNC) state and are unable to grow on culture media, and yet VBNC cells can maintain metabolic activity and resuscitate to a virulent state (Colwell and Grimes 2000). To circumvent this problem, molecular-based methods such as polymerase chain reaction (PCR) amplification have been developed. However, PCR can result in an overestimation of targeted live (active and VBNC) cells owing to DNA persistency after cell death (Josephson et al. 1993). One suggested approach to address this problem is to block the availability of DNA originating from dead cells for PCR amplification. This is the principle of a relatively recent approach, which applies a photoreactive dye [propidium monoazide (PMA) or ethidium monoazide (EMA)] that can enter dead cells with a broken membrane but not live cells with an intact membrane (Nogva et al. 2003). The photoreactive dye forms an irreversible cross-linkage with DNA when exposed to visible light and prevents PCR amplification (Nogva et al. 2003; Rudi et al. 2005).

Compared with EMA, PMA was shown to be less toxic to live cells and has a higher affinity to DNA (Nocker et al. 2006). PMA-PCR has been tested on a wide variety of bacteria, protozoa, viruses and fungi including pathogenic and environmental strains (e.g. Nocker et al. 2007a,b; Vesper et al. 2008; Brescia et al. 2009; Fittipaldi et al. 2010). However, some studies have found that the PMA-PCR method was not fully effective at removing the signal from dead cells. Kralik et al. (2010) reported that not more than a 2 log decrease in PCR signal could be obtained using membrane-permeable cells of Mycobacterium avium paratuberculosis. Pan and Breidt (2007) also showed that PMA-PCR did not always eliminate the signal of heat-killed Listeria monocytogenes. Similar results showing incomplete suppression of the dead cell signal have also been reported using EMA-PCR (e.g. Flekna et al. 2007; Wagner et al. 2008; Kobayashi et al. 2009). In addition, the use of membrane-based viability assays with inactivation mechanisms that do not affect the cell membrane has always been questioned (Nocker et al. 2007a; Parshionikar et al. 2010). In many environments, bacteria can be killed by processes [e.g. ultraviolet (UV) light] that do not directly cause membrane damage but instead cause injury to the genetic material. PMA treatment, for instance, was not successful at differentiating between live and UV-killed Escherichia coli O157:H7 (Nocker et al. 2007a).

The application of longer PCR amplicon sizes may be an effective method to improve the efficacy of the intercalating dye viability assays (EMA-PCR and PMA-PCR). Long amplicon sizes were shown to improve the effectiveness of EMA-PCR for heat-killed bacteria measured using both end-point and quantitative PCR (qPCR) assays (Soejima et al. 2008, 2011). Recent studies using PMA-PCR have also shown that amplification product size was important when applied to heat-killed E. coli O157:H7 and Enterobacter aerogenes (Luo et al. 2010) and also Vibrio anguillarum and Flavobacterium psychrophilum (Contreras et al. 2011). However, there are limited studies that have quantitatively assessed the effect of amplicon size on PMA-PCR and that have also extended the evaluation to include both heat- and UV-killed cells, both of which were evaluated in the present work. PCR amplification of specific gene targets was assessed using short and long amplicon sizes to optimize the reduction in the dead cell signal, and this effect was measured using qPCR methods.

The bacteria included in this study were Salmonella and Campylobacter, which are important foodborne and waterborne enteric pathogens. Campylobacter has been reported to be the most common bacteria associated with enteric disease worldwide and is transmitted mainly through contaminated food and water (Miller and Mandrell 2005). Campylobacter is fastidious with stringent growth conditions and can readily turn into VBNC cells under environmental conditions (Bhunia 2008), which can cause monitoring difficulties using conventional culture-based laboratory methods. Salmonella is also a common cause of foodborne and waterborne disease and of major concern to public health throughout the world. Salmonella species have also been reported to enter into a VBNC state in response to environmental stresses (Oliver 2005). Culture-dependent detection of both of these pathogens can be time-consuming and their quantification can be difficult, especially for environmental samples that typically require a most-probable number method. Therefore, the development of qPCR methods that can accurately assess viable cells of these bacterial pathogens following exposure to different inactivation mechanisms is of particular importance for identifying human health risk.

Materials and methods

Bacteria and culture

Campylobacter jejuni ssp. jejuni ATCC 35920 and Salmonella enterica ssp. enterica ATCC 13311 were obtained from American Type Culture Collection. Both strains were grown from long-term stocks stored in 25% glycerol peptone medium at −80°C. Salmonella enterica was grown on nutrient agar (BD Biosciences, Mississauga, ON, Canada) plates at 37°C overnight, and cells were inoculated into 100 ml of nutrient broth (BD Biosciences) in a 250-ml Erlenmeyer flask. The culture was grown overnight at 37°C without shaking. One millilitre of the culture was harvested by centrifugation at 12 000 g for 5 min, and the cell pellet was resuspended in 10 ml of sterile phosphate-buffered saline (PBS). The Salm. enterica concentration was adjusted to 1 × 107 colony forming units (CFU) ml−1 by measuring optical density (OD) together with a standard curve comparing OD (600 nm) vs plate count for this strain. Campylobacter jejuni was grown on Mueller Hinton blood agar (BD Biosciences) at 42°C under microaerophilic conditions (CampyPak Plus System; BD Biosciences) for 2–3 days. Using a sterile swab, colonies were suspended in a sterile 0·2-μm-filtered 0·85% NaCl solution. A NaCl solution was used because PBS can interfere with the BacLight viability count, as described by the manufacturer. The Camp. jejuni suspension was stained and enumerated by direct fluorescent microscopic cell count as described by Van Dyke et al. (2010), and the concentration was adjusted to 1 × 107 CFU ml−1 in 0·85% NaCl.

Preparation of heat- and UV-killed cells

To prepare heat-treated cells, Salm. enterica and Camp. jejuni cell suspensions (each 1 × 107 cells ml−1) were incubated at 90°C for 20 min. To prepare UV-treated cells, 15 ml of Salm. enterica and Camp. jejuni (each 1 × 107 cells ml−1) was transferred into 5·0-cm-diameter glass Petri dishes and exposed to 50 mJ cm−2 low-pressure UV light using a collimated beam apparatus (Calgon Carbon Corp., Pittsburgh, PA, USA). The collimated beam apparatus was fitted with a 12-W low-pressure mercury lamp, and UV exposure times were calculated using the software and method described by Bolton (1999). Irradiance was measured using a radiometer (International Light Model IL1700) equipped with a SED 240 UV detector. The radiometer and probe were calibrated by International Light, according to the US National Institute of Standards and Technology (NIST) method. The solutions were mixed during the exposure time using a magnetic stir bar.

Untreated (live), heat-treated and UV-irradiated Salm. enterica samples were enumerated by serial dilution and viable plate count on nutrient agar. Untreated (live) and heated-treated Camp. jejuni were enumerated by fluorescent microscopy using the BacLight LIVE/DEAD Viability Kit (Invitrogen, Burlington, ON, Canada), because plate counts would be affected by rapid VBNC formation of this species. UV-treated Camp. jejuni were not enumerated because the BacLight assay, based on membrane dye permeability, cannot measure the viability of UV-killed cells (Anderson et al. 2004). Live or killed cell preparations were transferred in 0·5 ml aliquots to 1·5-ml sterile, transparent microcentrifuge tubes and placed in the dark at room temperature. Within 30 min of heat or UV treatment, control and killed cells were incubated with and without PMA (each treatment in duplicate) as described below, followed by qPCR analysis.

PMA treatment

Solid PMA (phenanthridium, 3-amino-8-azido-5-[3-(diethylmethylammonio) propyl]-6-phenyl dichloride) was purchased from Biotium Inc. (Hayward, CA, USA), and a 4-mmol l−1 stock solution was prepared in 20% (v/v) dimethyl sulphoxide. The stock solution was transferred to 1·5-mL light-impermeable microcentrifuge tubes and stored at −20°C.

PMA was added to Salm. enterica at a final concentration of 10 μmol l−1 and to Camp. jejuni at a final concentration of 15 μmol l−1. Optimal PMA concentrations were previously determined by testing a range of PMA concentrations (0–100 μmol l−1) on heat-treated cells and using the same exposure conditions described below (data not shown). The optimal PMA concentrations resulted in a maximum reduction in the dead cell signal without affecting the live cells. PMA concentrations higher than 20 μmol l−1 were found to cause toxicity to live cells as determined by viable cell count measurements. After PMA addition, the cell suspension was mixed well by vortexing, followed by incubation in the dark for 5 min with constant mixing by inversion. The sample tubes were then placed on ice to avoid excessive heating and exposed to a 500-W halogen lamp for 10 min at a distance of 20 cm with the caps open. Two hundred microlitres of each duplicate sample was removed for DNA extraction and PCR as described below. Viable cell enumerations were made from one tube of each treatment. One hundred microlitres of Salm. enterica sample was serially diluted in PBS and enumerated by spread plating onto nutrient agar as described above. Five microlitres of Camp. jejuni sample was enumerated using the BacLight LIVE/DEAD Viability Kit (Invitrogen).

DNA extraction

DNA was extracted from 200 μl of PMA-treated and non-PMA-treated (control) samples using the Qiagen Dneasy tissue kit (Mississauga, ON, Canada). Cell lysis and DNA extraction was not preceded by a centrifugation step to ensure that free DNA was also measured in the samples. For this reason, the method was modified by adding 400 μl of AL buffer to each sample, and complete cell lysis was confirmed microscopically. Following column purification (as described by the manufacturer), samples were eluted in 200 μl of AE buffer and stored at −80°C until analysis.

PCR analysis

For Salm. enterica, two sets of primers were used that targeted the invA gene (Table 1), including one that amplified a 119-bp gene fragment (Hoorfar et al. 2000) and a set that amplified a 1614-bp gene fragment (this study). For Camp. jejuni, the cpn60 gene was targeted. Three different primers sets were used (Table 1) that could amplify a 174-bp fragment (Chaban et al. 2009), an 899-bp fragment (this study) and a 1512-bp fragment (this study). The primers in this study were designed using Beacon Designer 7.7 software (Bio-Rad, Mississauga, ON, Canada) together with sequence alignment data from the National Center for Biotechnology Information (NCBI). All primers and probes were purchased from Sigma-Genosys. PCR standard curves used Camp. jejuni and Salm. enterica cultures prepared as above. The cultures were enumerated by direct microscopic cell count (Van Dyke et al. 2010), and DNA was extracted using the Qiagen DNeasy tissue kit. Purified DNA was then serially diluted (tenfold) in TE buffer (10 mmol l−1 Tris, pH 8·0, 1 mmol l−1 EDTA) and stored at −80°C.

Table 1. Target genes and primers used in this study
SpeciesGene targetPrimersPrimer sequenceProduct size (bp)Reference
SalmonellainvAStyinva-JHO-2-left5′-TCGTCATTCCATTACCTACC-3′119Hoorfar et al. (2000)
Styinva-JHO-2-right5′-AAACGTTGAAAAACTGAGGA-3′
SalmonellainvASal-1614-F5′-ACAGTGCTCGTTTACGACC-3′1614This study
Sal-1614-R5′-TACGCACGGAAACACGTTC-3′
Campylobacter (Camp. jejuni)cpn60JH00395′-GAGCTTTCAAGCCCTTATATC-3′174Chaban et al. (2009)
JH00405′-AAGAACACCGCGAAGTTTATT-3′
Campylobactercpn60JH23F5′-CAGATGAAGCAAGAAAYAAAC-3′899This study
JH921R5′-AGTTCTTCCAAGYTCTTC-3′
Campylobactercpn60JH23F5′-CAGATGAAGCAAGAAAYAAAC-3′1512This study
JH1534R5′-GCATRCTAGCTACWGAAAC-3′

For all Salmonella invA and Campylobacter cpn60 primer sets, qPCR amplification was performed using Ssofast EvaGreen Supermix (Bio-Rad). PCR amplification was performed using the Bio-Rad iCycler iQ Real-Time PCR Detection System. Each 25-μl reaction mixture contained 400 nmol of each primer, 1× EvaGreen supermix and 10 μl DNA template. Each PCR run included duplicate standard curves and negative controls. The PCR amplification conditions for the Salmonella invA 119-bp gene fragment were as follows: one cycle at 95°C for 3 min; 50 cycles at 95°C for 30 s, 60°C for 30 s, 72°C for 30 s; and one cycle at 72°C for 10 min. Conditions for the 1614-bp primers were similar but used annealing/extension conditions of 53°C for 30 s/72°C for 1·5 min. Amplification conditions for the Campylobacter cpn60 174-bp primers were as follows: one cycle at 95°C for 3 min; 50 cycles at 95°C for 30 s, 62·5°C for 30 s, 72°C for 30 s; and one cycle at 72°C for 10 min. Conditions for the cpn60 899-bp and 1512-bp primers were similar but used annealing/extension conditions of 46°C for 30 s/72°C for 1·5 min and 55°C for 30 s/72°C for 1·5 min, respectively. For all the EvaGreen qPCR runs, PCR product specificity was confirmed by melting curve analysis using a ramping rate of 0·5°C per 10 s from 55 to 95°C.

qPCR amplification products were also analysed by agarose gel electrophoresis. Ten-microlitre aliquots of amplification product were mixed with DNA-loading buffer (Bio-Rad) and analysed on 1% agarose gels at a constant voltage of 100 V in 1× TRIS–acetate–EDTA buffer (EMD, Darmstadt, Germany). The agarose gels were stained in a 0·5 μg ml−1 ethidium bromide solution and visualized with a Bio-Rad Universal Hood II transilluminator using Quantity One 4.6.2 software. The GeneRuler 100-bp Plus DNA Ladder (Fermentas Canada Inc., Burlington, ON, Canada) was used as a DNA marker.

Heat-killed Campylobacter jejuni in river water

A 500-ml sample was collected from the Grand River in Kitchener, ON, Canada. Further details on the sample location and collection method are described by Van Dyke et al. (2010). The sample turbidity was 12·6 NTU. The river water sample was inoculated with 1 × 104 cells ml−1 of Camp. jejuni, prepared as described above. The sample was then pelleted by centrifugation at 10 000 g for 30 min, washed once in PBS and resuspended in 1 ml of PBS. Three 200-μl aliquots of the concentrated water sample were removed, followed by either (i) no treatment (live control), (ii) heating at 90°C for 20 min followed by PMA treatment or (iii) heating without PMA treatment. Samples were then centrifuged for 5 min at 12 000 g, and DNA was extracted and purified using a guanidine isothiocyanate (GITC) extraction buffer followed by the Qiagen DNeasy tissue kit as described by Cheyne et al. (2010). Samples were eluted in 200 μl of AE buffer, and 5 μl was analysed using the Campylobacter cpn60 long-amplicon (1512 bp) PCR assay.

Results

qPCR assays

qPCR assays using the fluorescent intercalating dye EvaGreen were used for both the Salmonella invA and the Campylobacter cpn60 gene targets. The qPCR assays could reliably detect these two species, and all amplicons were of the expected size. Each PCR assay produced standard curves with R2 values of 0·99 or greater and with slopes that ranged between −3·6 and −3·9 as a measure of the PCR efficiency. PCR containing 10 cells consistently resulted in DNA amplification, indicating that the detection limit was below 10 cells per reaction (102 cells ml−1 of original sample). PCR products were analysed by melt curve analysis, and nonspecific products were not observed. Long amplicon (1614-bp) primers for the Salmonella invA gene were designed in this study to be specific for the Salmonella genus. The two new sets of Campylobacter cpn60 primers (899 and 1512 bp) were designed to specifically target the thermophilic species Camp. jejuni, Campylobacter lari and Campylobacter coli, which are the species most frequently isolated from humans. The specificity of the primers was assessed using the Basic Local Alignment Search Tool (Blast) software (Madden et al. 1996) and the cpnDB chaperonin sequence database (Hill et al. 2004), but was not tested against a range of closely related species.

Effect of amplicon size on PMA pretreatment of heat-killed cells

High concentrations of live and heat-killed cells (1 × 107 cells ml−1) with and without pretreatment with PMA were enumerated by qPCR. Following treatment of live cells with PMA, there was no bactericidal effect on either Salm. enterica or Camp. jejuni and qPCR results of live cells and live cells with PMA treatment were the same (Figs 1a and 2a). After heat treatment, the viable plate count (Salm. enterica) and the viable microscopic cell count (Camp. jejuni) were below the detection level for both strains (2 cells ml−1). The killed cells without PMA pretreatment showed less than 1 log reduction in qPCR signal compared with live cells, and this effect was the same for both Salm. enterica and Camp. jejuni at short and long amplicon sizes (Figs 1a and 2a).

Figure 1.

Effect of amplicon size on qPCR signal reduction in heat-killed Salmonella enterica. Live and heat-killed cells were treated with or without PMA, and qPCR analysis was carried out using primers that targeted a 119- or 1614-bp fragment of the invA gene. qPCR data (a) show the average of duplicate data points, and error bars correspond to the range of values. qPCR amplification products were also analysed by gel electrophoresis (b): Lane 1, DNA Ladder; lanes 2 and 6, live cells without PMA; lanes 3 and 7, live cells with PMA; lanes 4 and 8, heat-treated cells without PMA; lanes 5 and 9, heat-treated cells with PMA. (■) Live; (image) Live + PMA; (image) Killed; (image) Killed + PMA and (▵) Viable count.

Figure 2.

Effect of amplicon size on qPCR signal reduction in heat-killed Campylobacter jejuni. Live and heat-killed cells were treated with or without PMA, and qPCR was carried out using primers that targeted a 174-, 899- or 1512-bp fragment of the cpn60 gene. qPCR data (a) show the average of duplicate data points, and error bars correspond to the range of values. qPCR amplification products were also analysed by gel electrophoresis (b): Lane 1, DNA Ladder; lanes 2, 6 and 8, live cells without PMA; lanes 3, 7 and 11, live cells with PMA; lanes 4, 8 and 12, heat-treated cells without PMA; lanes 5, 9 and 13, heat-treated cells with PMA. (■) Live; (image) Live + PMA; (image) Killed; (image) Killed + PMA and (▵) Viable count.

Heat-killed Salm. enterica cells treated with PMA in combination with PCR primers that amplified a long DNA fragment (1614 bp of invA gene) reduced the dead cell signal to below the detection limit (Fig. 1a). Similarly, PCR amplification of a long fragment size (1512 bp of cpn60 gene) for PMA-treated Camp. jejuni also reduced the dead cell signal to below the detection level (Fig. 2a). However, amplification of Salm. enterica using a relatively shorter DNA fragment on the same gene (119 bp) resulted in only a 3 log reduction in the dead cell PCR signal (Fig. 1a). For Camp. jejuni, PMA-treated killed cells detected using PCR primers that amplified a short fragment of the cpn60 gene (174 bp) resulted in only a 1 log reduction in the PCR signal (Fig. 2a), and amplification of an 899-bp fragment resulted in a 4 log suppression (Fig. 2a). Agarose gel electrophoresis confirmed the qPCR results (Figs 1b and 2b). Strong bands were present for live cells and dead cells without PMA, regardless of the amplicon size. PCR primers that target short gene fragments resulted in apparent but somewhat fainter bands for the heat-treated with PMA samples, and no bands were observed for heat-killed Salm. enterica and Camp. jejuni treated with PMA using long amplicon sizes.

Effect of amplicon size on PMA pretreatment of UV-treated cells

The ability of the PMA-PCR method to remove the false-positive signal from dead cells was also tested using the same high concentrations (1 × 107 cells ml−1) of UV-killed Salm. enterica and Camp. jejuni. Following irradiation by UV light at 50 mJ cm−2, the viable cell count was reduced by 3·4 log for Salm. enterica (Fig. 3a). PCR amplification of a long gene fragment (1614-bp invA gene) showed a decrease in the Salm. enterica PCR signal (3–3·2 log) for both PMA- and non-PMA-treated samples, similar to the decrease in the viable cell count. However, with a short amplicon target (119 bp), no signal reduction was observed in UV-killed Salm. enterica (Fig. 3a). Similar results were observed for UV-treated Camp. jejuni (Fig. 3b) with a 3–3·2 log PCR signal reduction using the long amplicon size for both PMA- and non-PMA-treated cells, and no decrease in the PCR signal for UV-treated cells using the short amplicon size.

Figure 3.

Effect of amplicon size and PMA on qPCR quantification of ultraviolet (UV)-treated or live (no UV) Salmonella enterica (a) and Campylobacter jejuni (b). qPCR data show the average of duplicate data points, and error bars correspond to the range of values. (A) (■) No UV −PMA; (image) No UV +PMA; (image) UV −PMA; (image) UV +PMA and (▵) Viable count. (B) (■) No UV −PMA; (image) No UV +PMA; (image) UV −PMA; (image) UV +PMA and (▵) Viable count.

PMA treatment of heat-killed Campylobacter jejuni in river water

The long-amplicon PMA-PCR method was applied to a river water sample that was inoculated with Camp. jejuni. The river water was concentrated (500×) and exposed to heat to kill the cells. qPCR of the control not exposed to heat resulted in a signal equivalent to 2·8 × 106 cells ml−1 in the concentrated water sample, and heat-treated cells without PMA resulted in 8·9 × 105 cells ml−1 (data not shown). However, heat-treated cells incubated with PMA did not result in a detectable signal by PCR.

Discussion

This study investigated how PCR amplicon size can affect the ability of PMA pretreatment methods to target only viable cells and remove interference from dead cells. Results showed that PMA pretreatment prior to PCR amplification can be effectively used as a viability assay for membrane-damaged cells, but that PCR amplicon size plays an important role in method design. PCR amplification of both long and short amplicon sizes was evaluated that targeted the invA gene in Salmonella and the cpn60 gene in Campylobacter. Heat-killed cells at an initial concentration of 1 × 107 cells ml−1 were pretreated with PMA, and PCR amplification of a 1614-bp target for Salm. enterica and a 1512-bp target for Camp. jejuni led to complete inhibition of the PCR signal. Short amplicon sizes were also tested, and PCR amplification of a 119-bp invA gene fragment resulted in only 3 log reduction in the Salm. enterica dead cell signal. PCR amplification of a 174-bp DNA fragment of heat-killed Camp. jejuni (1 × 107 cells ml−1) reduced the unwanted PCR signal by only 1 log. An intermediate Camp. jejuni amplicon size of 899 bp greatly reduced the dead cell signal from PMA-treated cells (4·2 log), but did not completely eliminate the dead cell signal. These results show that PMA treatment followed by PCR amplification of short DNA fragments (<200 bp) did not completely reduce the signal for heat-killed Salm. enterica and Camp. jejuni, but applying a long PCR amplicon size resulted in complete signal removal of heat-killed cells.

Incomplete signal reduction in dead cells using the PMA-PCR method has also been reported in other studies that targeted relatively short PCR amplicon sizes (Pan and Breidt 2007; Kralik et al. 2010). For both the EMA-PCR and PMA-PCR intercalating dye viability assays, a number of studies have also found improved viability detection with longer PCR amplicons (Luo et al. 2010; Contreras et al. 2011; Soejima et al. 2011). The findings in this paper are comparable with those of Luo et al. (2010) who assessed the PMA-PCR method on heat-killed E. coli O157:H7, Ent. aerogenes and Alcaligenes faecali. Using end-point PCR targeting the 16S rRNA gene, amplification of a 1400-bp PCR product resulted in complete band disappearance, whereas PMA had little effect on the 230-bp product. Although PMA/EMA treatment followed by short amplicon qPCR was previously shown to result in incomplete signal reduction, others have not made this same observation (e.g. Wang and Levin 2006; Cawthorn and Witthuhn 2008; Lee and Levin 2009). While the reason for this is unknown, it is possible that this may be due to the method detection limit or results may depend on the type of micro-organism tested.

qPCR assays normally use short amplicon sizes to guarantee method efficiency and for use with probe-based (i.e. Taqman) qPCR procedures. Relatively small amplicon sizes are widely used for the quantification of both Salmonella (Josefsen et al. 2007; Löfström et al., 2011) and Campylobacter (Rönner and Lindmark 2007; Josefsen et al. 2010) in different applications such as clinical, food and water quality studies. Although a 1–3 log PCR signal reduction in dead cells may be satisfactory when studying samples with low bacterial concentrations (such as surface or drinking water), incomplete suppression of the false PCR signal means that the method is not applicable to samples with higher concentrations of dead cells. High concentrations of bacterial pathogens can be present in clinical, environmental (i.e. sewage), food and laboratory studies (i.e. disinfection efficacy testing). Additionally, for unknown samples with no information on cell concentration, the results obtained using short amplicon PMA-PCR assay cannot be considered reliable in terms of live/dead cell evaluation.

These results show that PMA is able to inhibit PCR amplification of heat-treated cells more effectively when longer amplicons are targeted. The main mechanism of action of EMA/PMA for inhibiting PCR amplification is the formation of irreversible cross-linkages with DNA following exposure to light. The structural damage to DNA caused by cross-linkage with the dye is suggested to prevent strand elongation during PCR amplification and result in the elimination of the dead cell signal (Nogva et al. 2003; Rudi et al. 2005). However, studies by Soejima et al. (2007, 2008) showed that EMA cross-linkages can cause cleavage to chromosomal double-stranded DNA when exposed to visible light, which was suggested as the mechanism for PCR signal reduction. Some earlier studies have also demonstrated that ethidium is able to cleave single-stranded DNA (Deniss and Morgan 1976). Targeting longer amplicons results in a greater chance that PMA-induced damage be encountered and interfere with the PCR amplification process. The required amplicon length can be dependent on both PMA-DNA-binding and PMA-DNA-cleaving characteristics, as well as the level of DNA breakage caused by mechanical shearing during DNA extraction (Contreras et al. 2011).

This study observed a dissimilarity in the ability of PMA to suppress the PCR signal of heat-killed cells of two different types of bacteria. With short PCR amplicons of similar size (119 and 174 bp), PMA-PCR resulted in 3 log suppression for heat-killed Salm. enterica and 1 log suppression for heat-killed Camp. jejuni. This difference may be due to the selected target genes, because the invA gene was used for Salm. enterica and the cpn60 gene was used for Camp. jejuni. However, Waring (1965) reported little or no sequence preference for propidium iodide (PI), and assuming a similar behaviour for PMA, difference in PCR signal reduction between strains is likely not due to gene sequence differences. Differences in EMA inhibitory effect have also been reported by Soejima et al. (2011) between five different genera of bacteria with short amplicon sizes even when the same gene was targeted, but these disparities were eliminated when they tested PCR amplicons over 2000 bp in size. Therefore, the benefit of using a long-amplicon PCR method is that the dead cell signal is completely removed, and any differences in PMA-induced DNA damage between strains and/or gene targets are no longer important. This results in a method that is more applicable to samples containing mixed populations of micro-organisms.

PCR amplification of larger DNA fragments has been previously reported to correlate better with viable cell populations because of general nucleic acid degradation subsequent to cell death (McCarty and Atlas 1993; Aellen et al. 2006). However, in this study, both Salm. enterica and Camp. jejuni showed only a small reduction (1 log or less) in PCR signal following heat treatment without exposure to PMA. This reduction may be due to single- or double-stranded DNA breakage during heat exposure (Grecz and Bruszer 1981), but because the assay was performed within 30 min of heat treatment, further degradation of the DNA was not observed. Amplicon size resulted in a small decrease in PCR signal of dead cells without PMA for Camp. jejuni, but this effect was not observed for Salm. enterica.

Relatively low PMA concentrations of 10–15 μmol l−1 were used in this study, compared to others that have used PMA at 50 μmol l−1 (e.g. Nocker et al. 2007a; Pan and Breidt 2007). However, a maximum reduction in the dead cell signal was achieved at 10–15 μmol l−1 PMA, and higher concentrations did not result in improved signal suppression. Contreras et al. (2011) suggested that with a longer PCR amplicon, lower concentrations of PMA (10 μmol l−1) are needed to get complete suppression of the dead cell PCR signal. Because we found that PMA concentrations greater than 20 μmol l−1 resulted in cell toxicity, a combination of low PMA concentration and long amplicon size resulted in the accurate discrimination of the viable and dead cell signal.

The effect of PCR amplicon size was also evaluated on UV-treated Salm. enterica and Camp. jejuni. It was previously reported that PMA-PCR did not suppress the false-positive PCR signal of UV-killed cells (Nocker et al. 2007a). Results in this study using short amplicon sizes confirmed this, and showed that no or very low PCR signal reduction was achieved for UV-killed cells of both Salm. enterica and Camp. jejuni. This was expected because the main targets of UV damage are nucleic acids, and therefore, cells are not expected to have permeabilized membranes directly after UV irradiation. Complete dead cell signal suppression was observed when long-amplicon qPCR was applied, and PMA pretreatment had no effect on PCR signal suppression. These results support previous findings that signal reduction in UV-killed cells can be dependent on the size of fragment amplified by PCR (Süss et al. 2009; Rudi et al. 2010). Primary structural UV-induced damage to DNA includes the photoproducts cyclobutane pyrimidine dimer and 6-4 pyrimidine-pyrimidone (Moan and Peak 1989), both of which can lead to transcription blockage, replication arrest and consequently to cell death (Britt 1995). DNA breakage has also been reported to occur in UV-induced cells, not as a direct consequence of UV irradiation but as the consequence of cellular repair mechanisms (Bradley 1981). Considering the mechanisms of UV damage to DNA, PCR amplification can be prevented more effectively with an increase in amplicon size, similar to the actions caused by PMA cross-linkage to membrane-permeabilized cells.

This study highlights that amplicon length was an important factor in signal reduction in killed cells using PMA-PCR. Targeting relatively longer DNA fragments in PCR amplification can significantly improve the effectiveness of PMA-PCR method in terms of viable cell detection. This method is particularly important for live/dead cell discrimination when higher concentrations of bacteria are present. Additionally, results revealed that long-amplicon PCR can be used for viable cell determination of UV-killed cells. In a mixed population of bacteria exposed to unknown stresses, including environmental samples that will be exposed to natural sunlight, this method will be better able to target viable cells, regardless of the mechanism causing cell inactivation. This study showed that the long-amplicon PMA-PCR method could remove the PCR signal from a high concentration of heat-killed Camp. jejuni cells in a concentrated river water sample. This indicates that PMA treatment was not affected by particulate and organic material present in surface waters, and showed, as a proof-of-principle, that the assay can be applied to environmental matrices. Therefore, long-amplicon PMA-PCR is a valuable technique that can provide reliable quantification of viable micro-organisms, and may find application for a variety of sample types.

Acknowledgements

This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the partners of the NSERC Industrial Research Chair in Water Treatment at the University of Waterloo (www.civil.uwaterloo.ca/watertreatment). Funding was also provided by the Canadian Water Network and graduate scholarships held by A. Banihashemi, including an Alexander Graham Bell Canada Graduate Scholarship provided by NSERC and an Ontario Graduate Scholarship provided by the Ontario Ministry of Training, Colleges and Universities.

Ancillary