Robustness of Salmonella loop-mediated isothermal amplification assays for food applications

Authors


Abstract

Aims

Loop-mediated isothermal amplification (LAMP) assays have been developed recently for Salmonella detection. This study aimed at evaluating the robustness of two Salmonella LAMP assays in comparison with PCR and real-time quantitative PCR for food applications.

Methods and Results

Performance of the assays was examined under abusive preparation conditions, running temperatures and pH, and with the addition of various inhibitors and food rinses. LAMP achieved robust detection under abusive assay preparation conditions (holding at 22 and 37°C for up to 30 min) and running temperatures (57–68°C). With a hot-start DNA polymerase, PCR obtained comparable results under these temperature ranges. However, PCR performed markedly poorer under abusive pH. LAMP also showed greater tolerance to potential inhibitors than PCR. When food rinses including meat juice, chicken rinse, egg homogenate and produce homogenate were added at 20% of the reaction mix, PCR amplifications were completely inhibited, but LAMP reactions were not.

Conclusions

Our results demonstrated that LAMP is a robust alternative to PCR in Salmonella detection for food applications.

Significance and Impact of the Study

This study filled important knowledge gaps regarding the robustness of Salmonella LAMP assays. The findings will help bring Salmonella LAMP assays closer to wider applications in food testing.

Introduction

As a leading food-borne pathogen in the USA, Salmonella is estimated to cause 1 million cases, 19 336 hospitalizations and 378 deaths each year (Scallan et al. 2011). In 2012, close to 40% of laboratory-confirmed bacterial and parasitic infections captured in FoodNet surveillance areas were due to Salmonella (CDC 2013a). Salmonella also accounted for 30% of food-borne disease outbreaks with a single laboratory-confirmed aetiologic agent during 2009–2010 (CDC 2013c). A wide variety of food commodities have been implicated in Salmonella outbreaks, including meat, poultry, eggs, dairy, peanut butter and fresh produce (CDC 2013b; Painter et al. 2013).

To identify potential contamination problems during the production, processing and distribution of these high-risk foods, rapid, reliable and robust detection methods for Salmonella are urgently needed. Conventional culture methods are reliable but require a great deal of time and labour (Andrews et al. 2011). Immunological methods, for example enzyme-linked immunosorbent assay (ELISA), have the drawback of low specificity (Eriksson and Aspan 2007). Molecular-based methods such as PCR and real-time quantitative PCR (qPCR) have been widely used to detect Salmonella and demonstrated to be rapid, specific and sensitive (Rahn et al. 1992; Malorny et al. 2004; Cheng et al. 2008; Chen et al. 2010). However, both methods require a sophisticated thermal cycling instrument, and the commonly used Taq DNA polymerase is rather susceptible to PCR inhibitors present in food samples (Abu Al-Soud and Radstrom 1998).

Recently, multiple loop-mediated isothermal amplification (LAMP) assays have been developed for Salmonella detection and shown to be rapid, specific and sensitive (Hara-Kudo et al. 2005; Wang et al. 2008; Li et al. 2009; Lu et al. 2009; Chen et al. 2011). LAMP uses four to six specially designed primers and a strand-displacing Bst DNA polymerase to produce a stem–loop DNA structure during initial steps, followed by subsequent cycling amplification of this structure, which results in 109 copies of target DNA within an hour (Notomi et al. 2000). One desirable feature of LAMP is that the reactions are carried out under isothermal conditions, circumventing the need for a thermal cycling instrument (Notomi et al. 2000). LAMP was also found to be more tolerant than PCR against various clinical samples, such as serum, blood, urine and stools (Kaneko et al. 2007; Francois et al. 2011). Several of the Salmonella LAMP assays have been successfully applied in food samples, particularly eggs and produce (Ohtsuka et al. 2005; Li et al. 2009; Ueda and Kuwabara 2009; Techathuvanan et al. 2010; Chen et al. 2011; Ye et al. 2011; Zhang et al. 2011; Techathuvanan and D'Souza 2012; Yang et al. 2013). However, robustness of the assays, that is, the capacity to remain unaffected by small but deliberate variations in assay parameters (AOAC International 2012), has not been examined closely for food applications.

This study aimed at evaluating the robustness of two representative Salmonella LAMP assays in comparison with PCR and qPCR for food applications, specifically, under abusive assay conditions (elevated temperatures, pH fluctuations) and in the presence of inhibitors likely encountered during food testing (culture media, potential inhibitors and food rinses). One LAMP assay chosen for evaluation was the first Salmonella LAMP assay reported (Hara-Kudo et al. 2005), and the other one was developed recently by our research group (Chen et al. 2011). Both assays targeted the Salmonella invasion gene (invA), which possesses a broad specificity for Salmonella serovars (Rahn et al. 1992).

Materials and methods

Bacterial strain and DNA template preparation

Salmonella enterica serovar Typhimurium strain LT2, the principle strain for cellular and molecular biology in Salmonella (McClelland et al. 2001), was used in this study. The strain was cultured on trypticase soy agar (BD Diagnostic Systems, Sparks, MD, USA) at 35°C overnight. Ten-fold serially diluted DNA templates were prepared from stationary-phase cultures as described previously (Wang et al. 2012). The exact cell numbers were determined by standard plate counting. Aliquots of the templates were used for assay robustness tests described below.

LAMP assays

Two LAMP assays, developed by Hara-Kudo et al. (2005) and Chen et al. (2011), were evaluated and designated LAMP1 and LAMP2, respectively. Both assays employed six specially designed LAMP primers, two outer, two inner and two loops that targeted specific regions of the Salmonella invA gene (GenBank accession number M90846).

The assays were run as described previously (Chen et al. 2011) using 1 μl of DNA template prepared above at 8·5 × 107 CFU ml−1 (i.e. 8·5 × 104 CFU reaction−1). The reactions were conducted at 63°C for 40 min and terminated at 80°C for 5 min in a real-time turbidimeter (LA-320C; Eiken Chemical Co., Ltd., Tokyo, Japan). Turbidity readings at 650 nm were acquired every 6 s, and the time threshold values (Tt; in min) were determined when the turbidity increase measurements (differential value of moving average of turbidity) exceeded 0·1.

PCR and qPCR assays

As a comparison, Salmonella invA-based PCR and qPCR assays described previously (Rahn et al. 1992; Cheng et al. 2008) were carried out. The PCR mix (25 μl) consisted of 1 × PCR buffer, 1·5 mmol l−1 MgCl2, 0·2 mmol l−1 each deoxynucleoside triphosphate (dNTP), 0·4 μmol l−1 each primer (Integrated DNA Technologies, Coralville, IA, USA), 0·625 U of GoTaq Hot Start polymerase (Promega, Madison, WI, USA) and 1 μl of DNA template. The PCR programme included initial denaturation at 95°C for 2 min followed by 30 cycles of denaturation at 95°C for 30 s, primer annealing at 64°C for 30 s, extension at 72°C for 30 s and a final extension at 72°C for 5 min in a Bio-Rad C1000 thermal cycler (Hercules, CA, USA). PCR products (10 μl) were analysed by gel electrophoresis and documented by a Gel Doc XR system (Bio-Rad).

The qPCR mix (25 μl) contained 1 × PCR buffer, 1·5 mmol l−1 MgCl2, 0·2 mmol l−1 each dNTP, 1 μmol l−1 each primer, 0·25 μmol l−1 probe, 1·5 U of GoTaq Hot Start polymerase and 1 μl of DNA template. The reactions were carried out in an iQ5 real-time PCR detection system (Bio-Rad) using 95°C for 2 min then 50 cycles of denaturation at 95°C for 15 s and annealing/extension at 60°C for 30 s. The cycle threshold values (Ct) were obtained when fluorescence readings exceeded 30 units.

Abusive assay condition tests

Various abusive assay conditions (temperature, pH) were tested. Assay preparation temperature abuse was performed by holding all reagents at room temperature (22°C) or 37°C for 10 and 30 min, followed by flash preparation (<5 min) at room temperature. Holding and preparing the reagents at 4°C was included as a control. Assay running temperature abuse included running the assays at 57, 60, 63, 65, 68 and 70°C, which in the case of PCR and qPCR were annealing temperatures. The experiments were repeated three times.

For pH abuse tests, 4 μl of 1 mol l−1 Tris buffers (Sigma-Aldrich, St. Louis, MO, USA) at adjusted pH (6·8, 7·3, 7·8, 8·3, 8·8, 9·3 and 9·6) were added into the reaction mix. The experiments were repeated three times.

Inhibitor tests

Potential inhibitors including culture media used for enrichment and dilution (buffered peptone water, trypticase soy broth, 0·1% peptone; BD Diagnostic Systems), humic acid (0·005%, w/v; Sigma-Aldrich), a plant polysaccharide (polygalacturonic acid, 1·25%, w/v; Sigma-Aldrich) and soil (10%, w/v) were tested. The soil sample was obtained from a local agricultural field and mixed with buffered peptone water at a ratio of 1:10. These potential inhibitors were added in certain proportions (0, 1, 2, 5, 10, 20 and 30%) of the final reaction mix. The experiments were repeated three times.

Food rinse tests

Ground beef, ground pork, whole chicken, shell eggs, peanut butter and various produce (cantaloupe, jalapeño pepper, alfalfa sprouts and tomato) were obtained from a local grocery store and processed within 2 h of collection. To prepare food rinses, the whole chicken sample was manually mixed with 400 ml of buffered peptone water for 5 min. For all other food types, 25 g of samples were mixed with 225 ml of buffered peptone water and manually mixed for 2 min. All of the samples were examined for the presence of Salmonella following procedures in the Microbiology Laboratory Guidebook (USDA 2013). Only rinses from confirmed Salmonella-negative samples were used, which were added in certain proportions (0, 1, 2, 5, 10, 20 and 30%) of the final reaction mix. The experiments were repeated three times.

Data analysis

Means and standard deviations of Tt for LAMP and Ct for qPCR were calculated by Microsoft Excel (Microsoft, Seattle, WA, USA). These values sorted by abusive assay conditions or potential inhibitors were compared using the analysis of variance (SAS for Windows, version 9.2; SAS Institute Inc., Cary, NC, USA). Differences between the mean values were significant when the P value was <0·05.

Results

Assay performance under abusive temperature and pH

Table 1 summarizes the assay performance under various temperature and pH values. For LAMP and qPCR, the result was considered positive whenever a threshold value (Tt or Ct) is presented. Overall speaking, all of the four assays achieved positive detection of Salmonella Typhimurium LT2 across a wide range of temperatures, but were negative at several pH values. Noticeably, many Tt and Ct values observed in positive LAMP and qPCR assays were below 30 min and 30 cycles, respectively (Table 1), suggesting robust detection.

Table 1. Performance of LAMP1, LAMP2, PCR and qPCR under abusive assay conditions
ConditionLAMP1 Tt (min)LAMP2 Tt (min)qPCR Ct (cycle)PCR result
  1. Three independent repeats were conducted. In each column within each assay condition, mean Tt or Ct values followed by different upper case letters are significantly different (< 0·05).

  2. a

    Two of three repeats for qPCR generated Ct values at this pH level.

Holding temperature/time
4°C12·3 ± 0·4 (A)16·9 ± 0·7 (A)21·5 ± 0·1 (A)+
22°C for 10 min12·7 ± 0·1 (A)17·4 ± 1·0 (A)22·3 ± 0·1 (B)+
22°C for 30 min13·7 ± 0·4 (B)17·8 ± 1·1 (A)23·6 ± 0·2 (C)+
37°C for 10 min13·0 ± 0·1 (AB)18·3 ± 0·8 (A)23·3 ± 0·2 (C)+
37°C for 30 min15·0 ± 0·6 (C)18·5 ± 1·0 (A)23·4 ± 0·1 (C)+
Running temperature
57°C30·3 ± 0·8 (D)27·6 ± 2·1 (D)30·8 ± 0·0 (C)+
60°C20·1 ± 1·7 (B)20·3 ± 0·2 (C)26·4 ± 0·2 (A)+
63°C12·4 ± 0·5 (A)17·0 ± 0·5 (AB)28·0 ± 0·8 (B)+
65°C12·0 ± 0·0 (A)15·3 ± 0·4 (A)28·3 ± 0·1 (B)+
68°C22·4 ± 0·0 (C)18·4 ± 0·1 (BC)+
70°C
pH of the added Tris buffers
6·8
7·329·4 ± 6·1a
7·820·5 ± 1·5 (A)
8·319·5 ± 0·6 (A)22·1 ± 1·6 (A)
8·824·9 ± 6·3 (A)23·8 ± 0·0 (A)
9·3
9·6

Under abusive assay preparation temperatures, the mean Tt values ranged from 12·7 to 15·0 min for LAMP1 and from 17·4 to 18·5 min for LAMP2, while the mean Ct values for qPCR were between 22·3 and 23·4 cycles. These values were not significantly different (> 0·05) from the 4°C control for LAMP2, but qPCR was significantly different. For LAMP1, holding at room temperature or 37°C for 10 min did not increase the Tt values significantly, but holding for 30 min did (< 0·05). PCR consistently gave positive results under all of the preparation temperatures.

Among different running temperatures, both LAMP assays had the lowest Tt values at 65°C, which were not significantly different from those at 63°C (> 0·05). The lowest Ct values for qPCR were observed at 60°C. As the running temperature deviated from the optimum, both Tt and Ct values increased significantly (< 0·05). At 70°C, all of the four assays failed to yield any amplification product. In fact, qPCR stopped amplification at 68°C (Table 1).

The effect of adding Tris buffers with various pH values on the assay performance was more pronounced (Table 1). The functional pH ranges for LAMP1 and LAMP2 were 7·8–8·8 and 8·3–8·8, respectively. While for qPCR, the only functional pH was 7·3. Negative results were observed for all PCR.

Assay performance in the presence of potential inhibitors

Table 2 summarizes assay performance in the presence of six potential inhibitors. Peptone (up to 30% of the reaction mix) did not affect the positive detection of Salm. Typhimurium LT2 by any of the four assays (Table 2). Humic acid (0·005% w/v) had the strongest inhibitory effect, with 20, 5 and 2% or more of the reaction completely inhibited LAMP assays, qPCR and PCR, respectively. Except for soil (30% by LAMP2 only), positive LAMP amplifications were still observed when they were added in up to 30% of the reaction. In contrast, all qPCR amplifications stopped when these inhibitors occupied 30% (20% for soil) of the reaction mix. The performance of PCR was more obviously affected at 20% or above. Based on the data in Table 2, the overall ranking of the inhibitory effects were humic acid (0·005%) > soil (10%) > polygalacturonic acid (1·25%) > buffered peptone water ≥ trypticase soy broth > peptone (0·1%), while the assay tolerance to these inhibitors were LAMP1 ≥ LAMP2 > qPCR > PCR.

Table 2. Performance of LAMP1, LAMP2, PCR and qPCR in the presence of inhibitors
SubstanceProportion in reaction (%)LAMP1 Tt (min)LAMP2 Tt (min)qPCR Ct (cycle)PCR result
  1. Three independent repeats were conducted. In each column within each inhibitory substance, mean Tt or Ct values followed by different upper case letters are significantly different (< 0·05).

  2. a

    One of three repeats for qPCR generated Ct values at this level.

Buffered peptone water011·0 ± 0·0 (A)16·3 ± 0·0 (A)21·3 ± 0·3 (A)+
111·1 ± 0·1 (A)16·7 ± 0·3 (AB)21·8 ± 0·1 (AB)+
211·2 ± 0·3 (AB)17·1 ± 0·6 (ABC)22·6 ± 0·9 (AB)+
511·6 ± 0·2 (B)17·4 ± 0·5 (BC)23·0 ± 1·1 (AB)+
1012·2 ± 0·2 (C)17·9 ± 0·4 (C)23·2 ± 0·8 (B)+
2013·8 ± 0·1 (D)19·5 ± 0·1 (D)25·0 ± 0·5 (C)
3017·3 ± 0·3 (E)23·3 ± 0·1 (E)
Trypticase soy broth011·8 ± 0·5 (A)a16·5 ± 0·3 (A)20·1 ± 1·9 (A)+
111·8 ± 0·4 (A)16·8 ± 0·7 (A)21·2 ± 1·1 (A)+
212·0 ± 0·3 (A)16·8 ± 0·5 (A)22·2 ± 0·5 (AB)+
512·3 ± 0·2 (A)16·4 ± 0·1 (A)22·5 ± 0·4 (AB)+
1012·8 ± 0·0 (A)16·8 ± 0·4 (A)22·9 ± 0·5 (AB)+
2015·1 ± 0·4 (B)18·3 ± 0·5 (B)24·4 ± 1·5 (B)
3019·4 ± 1·0 (C)21·3 ± 0·5 (C)
0·1% peptone011·0 ± 0·1 (A)16·8 ± 0·4 (A)21·3 ± 1·6 (A)+
111·1 ± 0·2 (AB)17·0 ± 0·5 (AB)22·6 ± 1·1 (A)+
211·3 ± 0·2 (AB)17·5 ± 0·4 (ABC)22·8 ± 1·2 (A)+
511·9 ± 0·5 (ABC)17·7 ± 0·2 (BC)22·8 ± 1·2 (A)+
1012·1 ± 0·3 (BC)18·0 ± 0·4 (C)22·9 ± 1·3 (A)+
2012·4 ± 0·5 (C)18·1 ± 0·2 (C)23·0 ± 1·1 (A)+
3012·9 ± 0·8 (C)18·3 ± 0·3 (C)23·1 ± 1·1 (A)+
Humic acid (0·005%, w/v)011·0 ± 0·0 (A)17·2 ± 0·5 (A)22·7 ± 0·3 (A)+
112·3 ± 0·0 (B)18·1 ± 0·3 (AB)24·3 ± 0·5 (B)+
212·9 ± 0·2 (B)19·0 ± 0·0 (B)24·5 ± 0·5 (B)
514·5 ± 0·1 (C)21·5 ± 0·1 (C)
1019·3 ± 0·8 (D)27·3 ± 0·9 (D)
20
30
Polygalacturonic acid (1·25%, w/v)011·0 ± 0·0 (A)15·9 ± 0·8 (A)22·2 ± 0·0 (A)+
111·0 ± 0·0 (A)17·3 ± 0·2 (B)23·1 ± 1·2 (A)+
211·6 ± 0·1 (AB)18·4 ± 0·1 (C)23·2 ± 1·0 (A)+
512·5 ± 0·0 (AB)19·4 ± 0·1 (D)23·5 ± 0·8 (A)+
1013·7 ± 0·0 (B)20·7 ± 0·4 (E)23·3 ± 1·3 (A)+
2016·9 ± 0·3 (C)25·8 ± 0·0 (F)23·8 ± 1·0 (A)+
3026·0 ± 2·9 (D)40·8 ± 0·1 (G)
Soil (10%, w/v)011·5 ± 0·4 (A)15·5 ± 0·3 (A)21·5 ± 1·1 (A)+
112·0 ± 0·3 (AB)15·7 ± 0·4 (A)22·2 ± 1·3 (A)+
212·5 ± 0·4 (AB)15·9 ± 0·3 (AB)22·6 ± 1·2 (A)+
513·4 ± 0·5 (B)16·6 ± 0·3 (B)22·9 ± 1·3 (A)+
1016·3 ± 1·0 (C)18·7 ± 0·1 (C)24·3 ± 2·0 (A)
2023·2 ± 0·1 (D)25·2 ± 0·4 (D)
3034·7 ± 1·5 (E)

The vast majority of positive LAMP and qPCR reactions had Tt and Ct values below 30 min and 30 cycles, respectively, suggesting robust detection. With the increase in inhibitor concentrations, increasing trends in both Tt and Ct values were observed, some were statistically significant (< 0·05). For example, polygalacturonic acid (1·25% w/v) in 30% of the reaction resulted in Tt values for both LAMP1 and LAMP2 more than doubled those without inhibitors, while qPCR and PCR failed to detect Salmonella at this concentration.

Assay performance with the addition of food rinses

Table 3 summarizes the assay performance with the addition of nine food rinses. LAMP1 showed the best tolerance to various food rinses (up to 30% of the reaction) except for shell eggs (20 and 30%) and jalapeño pepper (30%). LAMP2 demonstrated comparable results as LAMP1 in rinses from chicken, peanut butter, cantaloupe, alfalfa sprouts and tomato, but was less robust in other food rinses. Negative qPCR results were obtained with a wide range of food rinses and concentrations, including ground beef and whole chicken from 2 to 30%, ground pork from 10 to 30%, shell eggs and peanut butter at 20 and 30%, and all of the produce items at 30%. For PCR, negative results were observed when alfalfa sprouts were added from 10 to 30% of the reaction mix and all other food rinses at 20 and 30%. Overall, the strongest inhibitory effect, where >2% of the food rinse in the reaction mix would generate negative results, was observed in shell eggs for LAMP2 and ground beef and whole chicken for qPCR (Table 3).

Table 3. Performance of LAMP1, LAMP2, PCR and qPCR in the presence of food rinses
Food typeProportions of food rinses in the reaction mix (%) with positive assay results
LAMP1LAMP2qPCRPCR
  1. Three independent repeats were conducted.

Ground beef0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 200, 10, 1, 2, 5, 10
Ground pork0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 100, 1, 2, 50, 1, 2, 5, 10
Whole chicken0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 20, 300, 10, 1, 2, 5, 10
Shell eggs0, 1, 2, 5, 100, 10, 1, 2, 5, 100, 1, 2, 5, 10
Peanut butter0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 20, 300, 1, 2, 5, 100, 1, 2, 5, 10
Cantaloupe0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 200, 1, 2, 5, 10
Jalapeño pepper0, 1, 2, 5, 10, 200, 1, 2, 5, 100, 1, 2, 5, 10, 200, 1, 2, 5, 10
Alfalfa sprouts0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 200, 1, 2, 5
Tomato0, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 20, 300, 1, 2, 5, 10, 200, 1, 2, 5, 10

The vast majority of positive LAMP and qPCR reactions had Tt and Ct values below 30 min and 30 cycles, respectively, suggesting robust detection (data not shown). As with the potential inhibitors, increasing trends in both Tt and Ct values were observed with the increase in rinse concentrations and some were statistically significant (< 0·05). For example, egg homogenates at 10% increased the Tt (for LAMP1) and Ct values from 11·4 to 48·3 min and from 22·2 to 39·5 cycles, respectively (data not shown), while LAMP2 failed to detect Salmonella at this concentration.

Discussion

Molecular-based methods such as PCR, qPCR and more recently LAMP have been applied to Salmonella detection, owing to their rapidity, specificity and sensitivity (Rahn et al. 1992; Hara-Kudo et al. 2005; Cheng et al. 2008; Wang et al. 2008; Lu et al. 2009; Chen et al. 2010, 2011). To expand their applications in food testing, assay robustness needs to be evaluated under abusive conditions and in the presence of potential inhibitors likely encountered during food testing (Ge and Meng 2009). Although a recent study (Francois et al. 2011) evaluated the robustness of LAMP for clinical diagnostic applications, this is the first study that comparatively evaluated the robustness of two LAMP assays, one PCR and one qPCR for Salmonella detection in food applications.

A major difference observed between the two LAMP assays is that LAMP1 was faster but more prone to generating false positives, confirming our previous findings (Yang et al. 2013). Both LAMP1 and LAMP2 demonstrated robust detection of Salm. Typhimurium strain LT2 under abusive assay preparation conditions (holding at room temperature and 37°C for up to 30 min) and running temperatures (57–68°C). In the study by Francois et al. (2011), a commercial Salmonella LAMP kit (Eiken Chemical Co.) based on LAMP1 (Hara-Kudo et al. 2005) was also capable of detecting Salmonella after sample pre-incubations for up to 30 min at room temperature and 37°C and across a wide range of amplification temperatures (57–67°C), corroborating findings in the present study. Due to the use of a hot-start DNA polymerase, PCR and qPCR had comparable results under abusive preparation conditions and were also shown to be robust throughout the range of tested annealing temperatures.

The pH ranges for the successful amplification of target Salmonella DNA by LAMP1 and LAMP2 were 7·8–8·8 and 8·3–8·8, respectively. These ranges were narrower than those reported recently where LAMP was shown to be robust across two pH units (7·3–9·3) (Francois et al. 2011). Two reasons may account for the difference. In the earlier study, 4 μl of positive control Salmonella DNA in the LAMP kit (purified DNA with unspecified concentration) was used while 1 μl of directly boiled DNA template (8·5 × 104 CFU reaction−1) was used in the present study. The amount of Tris buffer added was 4 μl in the present study, but 1 μl in the earlier study, which would result in a different pH in the final reaction mix. Nonetheless, PCR and qPCR tested in parallel performed markedly poorer than the two LAMP assays. As the normal pH of PCR or qPCR reaction used in this study was around 8·5, it is surprising that none of the PCR reactions worked when 4 μl of 1 mol l−1 Tris buffers with pH between 6·8 and 9·6 were added, suggesting the inhibitory effect of excess Tris buffer to PCR or qPCR.

One major challenge associated with pathogen detection in food is the presence of assay inhibitors in complex food matrices (Ge and Meng 2009). Previous studies reported that humic acid, plant polysaccharides and soil were major factors in plant-based foods that inhibited PCR reactions (Demeke and Adams 1992; Tsai and Olson 1992; Abu Al-Soud and Radstrom 2000), but their effects on LAMP have not been evaluated. We also chose to evaluate buffered peptone water, trypticase soy broth and 0·1% peptone because they are commonly used enrichment broths and diluents for preparing samples for food testing (Ge and Meng 2009; USDA 2013). A previous study (Kaneko et al. 2007) reported that some media such as saline, phosphate buffered saline and Eagle's minimum essential medium (MEM) inhibited a PCR assay for herpes simplex virus at 30, 20 and 2%, respectively, while showing no effect on LAMP performance at up to 30%. This superior tolerance of LAMP to potential inhibitors was confirmed in the present study, where buffered peptone water and trypticase soy broth did not inhibit LAMP, but inhibited PCR at 20 and 30% and qPCR at 30%. Our finding that 0·1% peptone was noninhibitory to either LAMP, qPCR or PCR also corroborated findings in a previous study on PCR (Rossen et al. 1992). The strong inhibitory effects of humic acid (0·005%), and to a lesser extent, soil (10%) and polygalacturonic acid (1·25%) were clearly demonstrated in the present study; highlighting the critical role sample preparation plays during food testing (Ge and Meng 2009). Nonetheless, as shown in Table 2, both LAMP assays demonstrated greater tolerance to these inhibitors than either PCR or qPCR.

The superior tolerance of LAMP to inhibitors than PCR was also demonstrated when actual food rinses were tested. Due to the complex nature of food matrices, inhibitors in a particular food to molecular assays are not fully characterized (Rossen et al. 1992; Witham et al. 1996). We chose to test meat, poultry, egg, peanut butter and produce as they have been frequently involved in Salmonella outbreaks (CDC 2013b; Painter et al. 2013). As expected from previous studies (Rossen et al. 1992; Witham et al. 1996), the inhibitory effects of meat and poultry rinses to qPCR were rather obvious and PCR was completely inhibited by all of the food rinses at 20%. Again, both LAMP assays were more robust when food rinses were added at similar levels. It is noteworthy that among all food rinses tested, shell eggs demonstrated the strongest inhibitory effect to these assays. The high lipid and protein content, and the presence of lysozyme in egg white and certain compounds in egg yolk were shown previously to inhibit PCR (Spanova et al. 2000; He et al. 2007).

The findings in this study are subject to several limitations. First, the study tested a single Salmonella serovar (Typhimurium), and it is unknown how the results relate to other serovars. However, because Salmonella invA gene is genus specific, similar assay performance may be expected for other serovars as long as this gene is present. Second, the study does not address the sensitivity of these assays under these abusive conditions because a fixed Salmonella concentration (8·5 × 104 CFU reaction−1) was used. In practice, enrichment is commonly used in routine food testing to increase the cell numbers to levels comparable to that tested in the present study. Third, DNA extracted from pure culture was used in food rinse testing rather than DNA extracted from food samples, the latter may better reflect natural or more practical conditions. We chose to add food rinses directly to the reaction mix because this represents the worst case scenario, as most food rinses would be removed through the DNA extraction step during routine food testing. It is therefore reasonable to assume better assay robustness under those practical conditions.

In conclusion, both LAMP assays possessed superior tolerance over PCR or qPCR to abusive assay conditions (elevated temperature, pH fluctuations) and the presence of assay inhibitors likely encountered during food testing. Coupled with the rapidity, specificity and sensitivity of LAMP shown in several recent studies (Ohtsuka et al. 2005; Li et al. 2009; Ueda and Kuwabara 2009; Techathuvanan et al. 2010; Chen et al. 2011; Ye et al. 2011; Zhang et al. 2011; Techathuvanan and D'Souza 2012; Yang et al. 2013), these LAMP assays are robust alternatives to PCR-type assays and may be widely adopted for routine Salmonella testing in food.

Acknowledgements

We would like to thank Siyi Chen and Feifei Han for valuable technical advice and helpful discussion. We also thank Kelly Jones for suggestions and critical reading of the manuscript.

Conflict of Interest

The authors declare no conflict of interest.

Ancillary