Rapid Detection of Listeria by Bacteriophage Amplification and SERS-Lateral Flow Immunochromatography

A rapid Listeria detection method was developed utilizing A511 bacteriophage amplification combined with surface-enhanced Raman spectroscopy (SERS) and lateral flow immunochromatography (LFI). Anti-A511 antibodies were covalently linked to SERS nanoparticles and printed onto nitrocellulose membranes. Antibody-conjugated SERS nanoparticles were used as quantifiable reporters. In the presence of A511, phage-SERS nanoparticle complexes were arrested and concentrated as a visible test line, which was interrogated quantitatively by Raman spectroscopy. An increase in SERS intensity correlated to an increase in captured phage-reporter complexes. SERS limit of detection was 6 × 106 pfu·mL−1, offering detection below that obtainable by the naked eye (LOD 6 × 107 pfu·mL−1). Phage amplification experiments were carried out at a multiplicity of infection (MOI) of 0.1 with 4 different starting phage concentrations monitored over time using SERS-LFI and validated by spot titer assay. Detection of L. monocytogenes concentrations of 1 × 107 colony forming units (cfu)·mL−1, 5 × 106 cfu·mL−1, 5 × 105 cfu·mL−1 and 5 × 104 cfu·mL−1 was achieved in 2, 2, 6, and 8 h, respectively. Similar experiments were conducted at a constant starting phage concentration (5 × 105 pfu·mL−1) with MOIs of 1, 2.5, and 5 and were detected in 2, 4, and 5 h, respectively.


Introduction
Listeria monocytogenes is a Gram-positive, motile, facultative anaerobic rod and the etiological agent of food-borne listeriosis. Symptoms of listeriosis include gastroenteritis, diarrhea, meningitis and bacteremia. It also contributes significantly to spontaneous miscarriages [1]. Listeriosis is responsible for approximately 1600 food-related illnesses and 260 deaths in the U.S. annually [2], and is the third leading cause of death among foodborne pathogens (behind Salmonella spp. and Toxoplasma gondii), with 94% of cases leading to hospitalization and a mortality rate of 20%-30% [2,3]. The U.S. has adopted a zero tolerance policy for Listeria on food. However, because of the protracted turnaround times (TATs) of conventional detection methods, the majority of food products potentially contaminated with L. monocytogenes are not tested before entering the marketplace, increasing the risk of widespread outbreaks. Given Listeria's natural occurrence in soil and among animal reservoirs, many processed foods are at risk of Listeria contamination. These include ready-to-eat meats and cheeses, unpasteurized dairy products, hot dogs, smoked seafood and raw produce [4]. Further adding to difficulties in outbreak prevention is Listeria's ability to grow over a wide range of temperatures, including those as low as 1˝C, which allows it to propagate even when refrigerated (~4˝C). One example of this is the Viruses 2015, 7, 6631-6641 The use of several particle types, each producing unique Raman spectra, allows for the simultaneous detection of multiple target analytes [33]. While SERS particles at high concentrations allow visual observation of LFI test lines, test line visibility can be faint and unreliable at low concentrations. It is hypothesized that the use of SERS extends sensitivity below visual levels and provides a quantifiable signal, thus eliminating the need for visual conformation. Figure S3 (Supplementary materials) displays a characteristic spectrum of the organic reporter molecule trans-1,2-bis(4-pyridyl)-ethylene, which was used in this study [33].
In this work, we describe the development of a novel SERS-LFI device utilizing anti-A511 conjugated SERS nanoparticles. This was combined with phage amplification for rapid and specific detection of L. monocytogenes. A511 amplification experiments were monitored via SERS-LFI and confirmed by spot titer assay. Raman signal quantitation and overall test sensitivity is discussed. While the aim of this study was specifically for Listeria detection, this platform can be adapted to other bacterial pathogens for which a suitable lytic phage is available.

A511 Propagation
Listeria ivanovii ATCC 19119 was obtained from the American Type Culture Collection (Manassas, VA, USA) and used for A511propagation (provided by Martin Lossener, Institute of Food, Nutrition and Health, ETH Zurich, Zurich, Switzerland) using soft agar overlays [34]. Briefly, 80 µL aliquots of A511 (10 8 pfu¨mL´1) were spotted onto lawns of L. ivanovii mixed with soft agar (0.5% agar in Brain Heart Infusion, BHI) and were incubated at 23˝C for 24 h. Resulting plaques were harvested by an addition of 3 mL of phosphate buffered saline (PBS), pH 7.4, followed by centrifugation of the resulting slurries at 9000ˆg for 15 min at 4˝C. Supernatants were filter-sterilized with 0.22 µm PES 1000 mL Rapid Flow Filter Units (Nalgene, Rochester, NY, USA). Polyethylene glycol phage precipitation (PEG 8000) (OmniPur, Gibbstown, NJ, USA) was conducted as previously described [35]. Further phage purification was conducted by cesium chloride gradient ultrafiltration [35]. Residual CsCl was removed by dialysis in PBS. All anti-phage antibodies were prepared using this purified phage.

Production and Purification of Anti-A511 Antibodies
Polyclonal rabbit anti-A511 phage IgG antibodies were prepared by Antibodies Incorporated (Davis, CA, USA). Antibodies were Protein G purified (Nab TM , Thermo Scientific, Rockford, IL, USA) and specificity was confirmed by an enzyme-linked immunosorbent assay (ELISA). Purified antibodies were dialyzed in PBS, concentrated by ultrafiltration (Amicon Ultra, 30 kDa cutoff) (Millipore, Billerica, MA, USA) and filter-sterilized with 0.22 µm PES filters (Thermo Scientific).

Nanoparticle Reporter and Control Particle Preparation
Anti-phage SERS reporter particles were prepared by conjugation of 50-60 nm diameter SERS-S440 Nanotags (SERS NPs, OD 24) (Becton Dickinson, Research Triangle Park, NC, USA) with purified polyclonal anti-A511 antibodies. The conjugation method described here is a modification found in Wang et al. [36]. The crosslinker, sulfosuccinimidyl 4-[N-maleimidomethyl] cyclohexane-1-carboxylate (sulfo-SMCC) (Thermo Scientfic) was prepared in degassed conjugation buffer, 10 mM 3-morpholinopropane-1-sulfonic acid (MOPS) (Sigma Aldrich, St. Louis, MO, USA) (pH 7.2). Sulfo-SMCC was reacted at 50 molar equivalents of purified anti-A511 antibodies for 30 min at 23˝C. Stock SERS NPs were prepared by dilution into a degassed conjugation buffer at a volume ratio of 1:1, and then reacted with antibody-crosslinker complex solution (350 molar excess antibodies to SERS NPs) at 23˝C for 3 h with continuous inversion. After primary conjugation, unreacted thiols on SERS NPs were blocked by a solution of N-ethylmaleimide (NEM) (Thermo Scientific), and prepared in a degassed conjugation buffer (650,000 molar excess NEM to SERS NPs). Concomitantly, Blocker™ Casein in PBS (Thermo Scientific) was added to block the surface of the SERS NPs. Blocking was performed at 23˝C for 2 h with continuous inversion. Unreacted sulfo-SMCC malimide groups were quenched with 2-mercaptoethanesulfonic acid (MP Biomedicals, Santa Ana, CA, USA) at 23˝C with continuous inversion for 45 min. Excess reagents were removed by centrifugation (1000ˆg for 10 min), and the supernatant was removed and replaced with a storage buffer, 50 mM sodium borate (Fisher Scientific, Fairlawn, NJ, USA), 1% v¨v´1 gelatin (telostean gelatin from cold water fish skin, Sigma Aldrich), and 0.05% w¨v´1 sodium azide (Fisher Scientific) (pH 7.5). This was repeated 4 times and conjugated particles were stored at 4˝C. Figure S4 (see Supplementary materials) shows a schematic overview of the conjugation.
Control particles were prepared by passive conjugation of blue carboxy-modified polystyrene Seradyn particles (Thermo Scientific, Waltham, MA, USA, 223 nm diameter, 2.5% solids) to ImmunoPure biotinylated bovine serum albumin (Biotin-LC-BSA) (biotin BSA) (Thermo Scientific). Briefly, particles were centrifuged at 9000ˆg for 5 min, and the supernatant discarded. Particles were resuspended in 50 mM 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid (HEPES) (Research Organics, Cleveland, OH, USA) (pH 7.4) with a 2 mg¨mL´1 biotin BSA added at a 2.5:1 ratio and mixed with continuous inversion for 90 min at 23˝C. The resulting suspension was centrifuged at 10,000ˆg for 5 min, the supernatant discarded, and the pellet resuspended in HEPES. This was repeated twice with final resuspension in TNGA (0.025 M Tris base, 0.1 M NaCl, 1% v¨v´1 fish gelatin, and 0.05% w¨v´1 sodium azide) (pH 8.4) and blocked for 1 h at 23˝C with continuous inversion. Particles were pelleted by centrifugation (10,000ˆg, 5 min) with the supernatant removed and replaced with fresh TNGA. This centrifugation step was repeated twice more, then resuspended to 0.625% solids and stored at 4˝C.

LFI Device Fabrication
Nitrocellulose membranes (Millipore Hi-Flow 180, Billerica, MA, USA, SHF1800425) were prepared by applying purified anti-A511 antibodies as a test line and NeutrAvidin™ Biotin-Binding Protein (Thermo Scientific) as a control line using an IVEK Digispense 2000 striper (IVEK Corporation, North Springfield, VT, USA). Anti-A511 (2 mg¨mL´1) and NeutrAvidin (1.25 mg¨mL´1) were applied at a rate of 4 µL¨s´1 and dried for 15 min at 35˝C. Antibody/control line-stripped membranes were stored desiccated at 23˝C. Release pads (Schleicher & Schuell, Keene, NH, USA) were prepared by impregnation of glass fiber release medium with a solution of SERS reporter nanoparticles (0.02% solids) and control particles (0.01% solids) in 2 mM sodium borate, 0.1 M NaCl, 1% v¨v´1 fish gelatin, 0.05% w¨v´1 sodium azide, and 3% w¨v´1 sucrose (Baker, Phillipsburg, NJ, USA) (pH 8.4) dried for 30 min at 35˝C and stored desiccated at 23˝C until assembly. LFI devices were fabricated by mating sample pad, release pad, nitrocellulose membrane, and absorbing pad (Schleicher & Schuell) to an adhesive backboard (G&L Precision Die Casting, San Jose, CA, USA) with an overlap of approximately 2 mm between layers ( Figure S1). Assembled LFI strips were cut to a width of 3.7 mm using a programmable sheer (Kinematic Automation Inc., Twain Harte, CA, USA) and desiccated at 23˝C until use.

Determination of LFI Limit of Detection
LFI LOD was determined by a serial dilution of filter-sterilized A511 in tryptose and 1 mM CaCl 2 . Triplicate analysis of phage dilutions ranging from 1ˆ10 9 plaque forming units (pfu)¨mL´1 to 2 pfu¨mL´1 were tested, in addition to phage-free controls. Prior to application, phage samples were mixed with a running buffer at a 1:1 ratio to a volume of 100 µL and applied drop-wise to the sample pad. The running buffer consisted of 0.1 M sodium borate, 3% w¨v´1 bovine serum albumin (BSA) (Sigma Aldrich), and 1% v¨v´1 Tween 20 (Sigma Aldrich) (pH 8). LFI was conducted for 30 min, wicking pads were removed to prevent backflow, and LFI strips were dried in a desiccator for 10 min. Test lines were interrogated by Raman spectroscopy (Advantage 785, DeltaNu Inc., Laramie, WY, USA) at 785 nm with a laser power of 51 mW. Twelve measurements were collected along the test line with an interrogation time of 3 s per measurement. Whole spectra analyses of SERS samples were performed by assuming that sample spectra were a linear combination of an SERS nanoparticle reference spectrum and a nitrocellulose reference spectrum, which was then solved using least squares [28,37].

Phage Amplification and LFI Analysis
Four phage amplification experiments were performed using A511 and L. monocytogenes ATCC 19115 at a multiplicity of infection (MOI) of 0.1 with varying A511 starting concentrations (1ˆ10 6 , 5ˆ10 5 , 5ˆ10 4 , and 5ˆ10 3 pfu¨mL´1). Two additional phage amplification experiments were conducted at MOIs of 2.5 and 5 with the same starting phage concentration (5ˆ10 5 pfu¨mL´1). The A511 starting concentration was always below the determined phage LOD to avoid false positive results caused by initial infecting phage. Overnight cultures of L. monocytogenes were back-diluted and grown to an OD 620 of 0.3, corresponding to 1ˆ10 8 colony forming units (cfu) mL´1, and were subsequently diluted to the appropriate starting concentration at a 10 mL final volume. A511 (5ˆ10 7 pfu¨mL´1) was added at appropriate volumes for the desired starting phage concentrations. Aliquots were taken from amplification reactions every hour and filtered to remove bacteria. Fifty microliters of resulting filtrates were mixed with 50 µL running buffer and applied to LFI strips as described in the previous section. Time was required to run the LFI device (~30 min), dry the strip (~10 min) and interrogate the LFI strip with the Raman spectrometer (~5 min). Phage concentration was also followed in parallel for the duration of the experiment by spot titer assay as previously described [38] to confirm results of the SERS-LFI.

SERS-LFI Device Optimization
Phage biology, SERS, and LFI are independent concepts previously utilized for bacterial detection [39][40][41][42][43]. The current study combined aspects of each of these principles to overcome obstacles associated with manufacturing reliable SERS-LFI devices. Consistent, optimized antibody conjugation to SERS NPs and LFI construction were particularly important to reproducibility; thus, improving these methods was a major focus of the current study. All changes to the original NP manufacturer's conjugation protocols were intended to minimize agglomeration of conjugated NPs on the membrane. Agglomerated NPs did not consistently travel down the nitrocellulose, which created an artificially high signal with phage-free controls. Problems addressed in this paper included: the optimization of conjugation and running buffer composition; the antibody:SERS NP ratio; the blocking of non-specific phage-particle interactions; the capture antibody application; the membrane flow rate; and the SERS reporter release pad concentration. Troubleshooting these parameters led to the development of a new, robust protocol for fabrication of SERS-LFI devices. These adjustments are summarized in Table 1, and further discussion can be found in the Supplementary materials.

Determination of LFI Limit of Detection
Both visual and spectroscopic limits of detection were determined for prototype SERS-LFI. Visual positives were determined by the formation of a pink test line. As shown in Figure 1A,B, phage concentrations of 1ˆ10 9 pfu¨mL´1 and 2ˆ10 8 pfu¨mL´1 produced clearly visible pink lines. The minimum concentration for which a visually observable colored line formed was 6ˆ10 7 pfu¨mL´1 ( Figure 1C). It should be noted that this concentration produced a very weak response, which could be mistaken as a negative result. The possibility of such outcomes emphasizes the difficulty associated with qualitative visual analysis and interpretation by the untrained user.        Raman signals measured from the same serial dilutions of A511 are shown in Figure 2. These intensities represent the interquartile range (IQR) of 12 random shots along a single test line, shown by the vertical error bars. The IQR was necessary, because heterogeneous membrane pore distribution was observed to lead to areas of higher and lower Raman intensity along the length of the test line. Horizontal error bars in Figure 2 represent standard deviation of phage titers done in triplicate. The lowest phage dilution to produce a Raman signal above phage-fee controls was 6ˆ10 6 pfu¨mL´1. Thus, this phage dilution, represented by a dotted line in Figures 3B and 4B, and the resulting Raman signal (intensity of 1.37), represented by a dashed line in Figures 3A and 4B, were accepted as the phage and SERS LOD, respectively. A comparison between SERS and visual limits of detection (6ˆ10 6 pfu¨mL´1 and 6ˆ10 7 pfu¨mL´1, respectively) demonstrated an order of magnitude greater sensitivity for SERS. All IQR measurements below SERS LOD represent instrument noise.

Phage Amplification and SERS-LFI Analysis
Four phage amplification experiments at a MOI of 0.1 were monitored by SERS-LFI ( Figure 3A, Table 1) and parallel spot titer assay ( Figure 3B). The purpose of this experiment was to investigate how decreasing phage and bacterial concentrations affected the time it took to get a positive result. Phage concentrations were varied from 1ˆ10 6 pfu¨mL´1 to 5ˆ10 3 pfu¨mL´1, and corresponding bacterial concentrations were varied from 1ˆ10 7 cfu¨mL´1 to 5ˆ10 4 cfu¨mL´1. SERS detection of the highest phage concentrations (1ˆ10 6 pfu¨mL´1 and 5ˆ10 5 pfu¨mL´1) was achieved in 2 h. Additional time was required before decreased phage concentrations reached detectable levels (6 h for 5ˆ10 4 pfu¨mL´1 and 8 h for 5ˆ10 3 pfu¨mL´1). Parallel spot titer assays confirmed that all positive tests represented phage concentrations greater than the established LOD and that increased SERS intensities correlated with phage amplification.
Viruses 2015, 7, page-page 7 Raman signals measured from the same serial dilutions of A511 are shown in Figure 2. These intensities represent the interquartile range (IQR) of 12 random shots along a single test line, shown by the vertical error bars. The IQR was necessary, because heterogeneous membrane pore distribution was observed to lead to areas of higher and lower Raman intensity along the length of the test line. Horizontal error bars in Figure 2 represent standard deviation of phage titers done in triplicate. The lowest phage dilution to produce a Raman signal above phage-fee controls was 6× 10 6 pfu·mL −1 . Thus, this phage dilution, represented by a dotted line in Figures 3B and 4B, and the resulting Raman signal (intensity of 1.37), represented by a dashed line in Figures 3A and 4B, were accepted as the phage and SERS LOD, respectively. A comparison between SERS and visual limits of detection (6 × 10 6 pfu·mL −1 and 6 × 10 7 pfu·mL −1 , respectively) demonstrated an order of magnitude greater sensitivity for SERS. All IQR measurements below SERS LOD represent instrument noise.

Phage Amplification and SERS-LFI Analysis
Four phage amplification experiments at a MOI of 0.1 were monitored by SERS-LFI ( Figure 3A, Table 1) and parallel spot titer assay ( Figure 3B). The purpose of this experiment was to investigate how decreasing phage and bacterial concentrations affected the time it took to get a positive result. Phage concentrations were varied from 1 × 10 6 pfu·mL −1 to 5 × 10 3 pfu·mL −1 , and corresponding bacterial concentrations were varied from 1 × 10 7 cfu·mL −1 to 5 × 10 4 cfu·mL −1 . SERS detection of the highest phage concentrations (1 × 10 6 pfu·mL −1 and 5 × 10 5 pfu·mL −1 ) was achieved in 2 h. Additional time was required before decreased phage concentrations reached detectable levels (6 h for 5 × 10 4 pfu·mL −1 and 8 h for 5 × 10 3 pfu·mL −1 ). Parallel spot titer assays confirmed that all positive tests represented phage concentrations greater than the established LOD and that increased SERS intensities correlated with phage amplification. A second series of amplifications was conducted to investigate the relationship between MOI and detection time ( Figure 4, Table 2). As previously discussed, an MOI of 0.1 resulted in detection in 2 h. Detection times increased with increasing MOIs. MOIs of 2.5 and 5 surpassed detection limits at 4 h and 5 h, respectively. Parallel spot titer assays are shown in Figure 4B. Error bars in Figures 3A  and 4A correspond to the IQR of 12 random shots along the test line of a single LFI strip, while error bars in Figures 3B and 4B represent standard deviation of phage titers measured in triplicate. Table 2 summarizes Figures 3 and 4. A second series of amplifications was conducted to investigate the relationship between MOI and detection time ( Figure 4, Table 2). As previously discussed, an MOI of 0.1 resulted in detection in 2 h. Detection times increased with increasing MOIs. MOIs of 2.5 and 5 surpassed detection limits at 4 h and 5 h, respectively. Parallel spot titer assays are shown in Figure 4B. Error bars in Figures 3A  and 4A correspond to the IQR of 12 random shots along the test line of a single LFI strip, while error bars in Figures 3B and 4B represent standard deviation of phage titers measured in triplicate. Table 2 summarizes Figures 3 and 4.   The results displayed in Figures 3 and 4 highlight a drawback of phage amplification assays. Low concentrations of phage and bacteria decrease the chance of a phage and a bacterium meeting, irrespective of the binding efficiency of the phage to the bacterium [39]. This increases the time necessary for the phage concentrations to reach the detection limit of the detector device [44][45][46]. Hagens and Loessner have previously discussed this phenomenon [47].
In conclusion, this is the first report of phage amplification combined with the use of SERS and LFI for Listeria detection. This study focused on establishing a robust anti-phage conjugation protocol and on optimization of LFI construction, with the goal of minimizing nanoparticle agglomeration and improving reproducibility. The resulting devices are capable of detecting progeny A511 at concentrations as low as 6× 10 6 pfu·mL −1 . The shortest detection time for L. monocytogenes was 2 h, while, in a separate experiment, detection of bacteria at a concentration of 1 × 10 4 cfu·mL −1 was achieved in 8 h. While an enrichment step is still needed to obtain bacterial concentrations that allow for propagation of progeny phage to detectable levels, phage amplification eliminates the need for downstream plating on selective media, and for further biochemical or molecular tests (reducing detection by 24-48 h), while providing evidence of viable cells. SERS-LFI allows for positive identification in as little as 30 min, while traditional plaque assays can take up to 24 h.  The results displayed in Figures 3 and 4 highlight a drawback of phage amplification assays. Low concentrations of phage and bacteria decrease the chance of a phage and a bacterium meeting, irrespective of the binding efficiency of the phage to the bacterium [39]. This increases the time necessary for the phage concentrations to reach the detection limit of the detector device [44][45][46]. Hagens and Loessner have previously discussed this phenomenon [47].
In conclusion, this is the first report of phage amplification combined with the use of SERS and LFI for Listeria detection. This study focused on establishing a robust anti-phage conjugation protocol and on optimization of LFI construction, with the goal of minimizing nanoparticle agglomeration and improving reproducibility. The resulting devices are capable of detecting progeny A511 at concentrations as low as 6ˆ10 6 pfu¨mL´1. The shortest detection time for L. monocytogenes was