Development of a qPCR for the detection and quantification of Salmonella spp. in sheep feces and tissues

Abstract

Salmonella spp. are common causes of disease in intensive livestock production systems, and contamination of foodstuffs is of significant concern for public health. Therefore, the identification and quantification of Salmonella spp. is important for monitoring the level of fecal shedding or tissue colonization in infected animals and animal products. We developed and evaluated a quantitative PCR (qPCR) method on spiked sheep tissue and fecal samples for the detection and quantification of Salmonella spp. Without the use of a pre-enrichment step, the qPCR limit of detection (LOD) results for sheep fecal (4 × 10⁴–6 × 10³ cfu/g) and tissue (4 × 10⁵–4 × 10³ cfu/g) samples were not adequate for detection purposes. With the inclusion of a 6-h pre-enrichment step in buffered peptone water (BPW), the LOD was 9 cfu/g (2.57 × 10¹ copies/g) in sheep feces, and 5.4 cfu/g (3.22 copies/g) sheep tissue. Comparison of the 6-h BPW qPCR method with a 24-h mannitol–selenite–cystine broth enrichment culture method using spiked samples revealed a sensitivity of 91% and 92%, respectively, and a specificity of 100% for both methods. The correlation was significant between the quantity (copies/mL) of Salmonella spp. in BPW at 6 h and at 0 h, allowing semiquantitative analysis. Our results demonstrate that, following inclusion of a 6-h pre-enrichment step in BPW, qPCR is semiquantitative with improved LODs of Salmonella spp. in sheep fecal and tissue samples.

Introduction

Salmonella spp. are common causes of disease in intensive livestock production systems. Clinical manifestations vary in severity and duration depending on strain virulence, dose, nutrition, age of animals, and management practices.²² Common clinical signs include diarrhea, loss of appetite, fever, depressed mentation, and mortality. Less common clinical manifestations include arthritis, abortions, stillbirths, and meningitis.^16,22,29 In sheep, Salmonella enterica subsp. enterica serovar Typhimurium (termed Salmonella Typhimurium herein) and Salmonella Bovismorbificans are the most frequently isolated serovars.²⁹ Salmonella Typhimurium and Salmonella Dublin are the most commonly reported serovars to cause disease in cattle.¹⁶ Although Salmonella spp. infection has important animal welfare implications, it is also a common zoonotic infection frequently linked to the consumption of contaminated foodstuff.¹⁸

Identification and quantification of Salmonella spp. is important for monitoring the level of fecal shedding in animals or tissue colonization in infected animals and animal products. The 2 most common culture-based quantitative detection methods for Salmonella spp. are the most-probable-number (MPN) test and standard plate counts.¹⁴ Although the MPN method is useful for determining low concentrations of Salmonella, it is a labor-intensive process. Hence, the standard plate count method is often used as a quantitative alternative; it relies on the use of selective and differential media such as xylose–lysine–deoxycholate (XLD) agar and Rappaport–Vassiliadis soya (RVS) medium.^1,25 Alternatively, if quantification is not required, selective enrichment cultures provide a lower limit of detection (LOD). The process of isolation and biochemical testing to identify suspect colonies is expensive and time-consuming, taking up to 4 d for a definitive answer.²

The use of PCR for detection of Salmonella spp. is becoming more common and provides rapid detection.^4,13,30,31 However, tissue and fecal samples present a unique set of challenges for PCR assays. First, samples naturally infected with Salmonella spp. may not contain sufficient numbers for detection by PCR. Therefore, a pre-enrichment step is often necessary to allow detection limits comparable to culture.^7–9,19 Second, these sample types can contain PCR inhibitors, requiring extensive DNA extraction and purifications methods, often resulting in a dilution of the sample, thus hampering the potential LOD.^8,23

Few studies have evaluated PCR-based quantification of Salmonella spp. in livestock samples without the use of a pre-enrichment step. Of those that have, the DNA extraction methods described are not ideal for high-throughput sample analysis, typically relying on spin-column methods that require a high degree of sample handling.^30,31 Furthermore, little information is available on the expected LOD, and how this compares to traditional culture methods. A significant increase in PCR sensitivity has been observed in studies utilizing a pre-enrichment step,^17,28 with results comparable to enrichment culture.^9,17,27 However quantification of the Salmonella spp. load in the original sample is not achievable with these published molecular protocols. Our aim was to develop a fast, simple, and effective method for quantifying Salmonella spp. in sheep feces and tissues. Our objectives were to (1) develop a qPCR for the detection and quantification of Salmonella spp., (2) evaluate its quantitative use on sheep samples without the use of a pre-enrichment step, and (3) investigate its use with a shortened pre-enrichment step to improve the LOD while maintaining quantification attributes.

Material and methods

Analytical specificity and sensitivity

The analytical specificity of the qPCR was assessed against a panel of bacterial isolates (Table 1). Isolates were obtained from field samples submitted to the Livestock Veterinary Teaching and Research Unit at the University of Sydney.

Table 1.

Salmonella enterica isolates used for analytical specificity analysis. All organisms are field isolates unless specified.

Organism	qPCR result
Salmonella Amsterdam var 15+	+
Salmonella Anatum (2 isolates)	All +
Salmonella Bovismorbificans (3 isolates)	All +
Salmonella Dublin (3 isolates)	All +
Salmonella Infantis (2 isolates)	All +
Salmonella Kiambu (2 isolates)	All +
Salmonella Kottbus	+
Salmonella Mbandaka	+
Salmonella Muenster	+
Salmonella Newport	+
Salmonella Orion var 15+, 34+	+
Salmonella Tennessee	+
Salmonella Typhimurium (3 isolates)	All +
Salmonella Zanzibar (2 isolates)	All +

The following isolates were also tested, with all results being negative: Acholeplasma granularum, Corynebacterium spp., Enterococcus faecalis, Escherichia coli (4 isolates), Klebsiella pneumoniae (2 isolates), Klebsiella spp., Mycoplasma bovigenitalium, M. bovirhinis, M. bovis, M. bovoculi, M. californicum, M. capricolum, M. dispar, M. leachii, M. ovipneumoniae, Nocardia spp., Plesiomonas spp., Streptococcus uberis, S. dysgalactiae, S. agalactiae, and Staphylococcus aureus (ATCC 25923 and 1 field isolate).

The analytical sensitivity was determined from the standard curves of 7 Salmonella field isolates including Salmonella Typhimurium (n = 3), Salmonella Dublin (n = 2) and Salmonella Bovismorbificans (n = 2). A 10-fold dilution series of genomic DNA was prepared for each isolate from 1 ng/reaction to 1 × 10⁻⁵ ng/reaction, with intermediate concentrations at the lower end of the curve including 5 × 10⁻⁵ ng/reaction and 5 × 10⁻⁶ ng/reaction. Each isolate had 3 standard curves analyzed in duplicate reactions, for a total of 21 standard curves.

Isolate preparation for sample spiking

Salmonella spp. field isolates recovered from calf fecal samples were inoculated onto XLD agar (Edwards Group) and incubated at 37°C for 24–48 h. A single colony was selected and inoculated into 4 mL of lysogeny broth (LB; Becton Dickinson) and incubated at 37°C overnight. Following overnight incubation, 4 mL of fresh warmed LB was added, and the broth incubated for a further 2 h ± 15 min. The quantity of Salmonella spp. in the LB (cfu/mL) was determined by a 10-fold dilution series in sterile phosphate-buffered saline (PBS) and plating onto XLD agar in duplicate 10-µL volumes, followed by incubation at 37°C for 24 h and colony counting.

Sheep samples

Sheep fecal samples used were collected fresh from pasture immediately following excretion from sheep belonging to The University of Sydney. Sheep tissue samples were obtained from Wollondilly Abattoir. All samples were transported to the laboratory on ice and were negative for Salmonella spp. as assessed by selective enrichment culture using mannitol–selenite–cystine broth (MSCB; Edwards Group) and XLD agar.

Limit of detection without pre-enrichment

The LOD of the qPCR without the use of a pre-enrichment procedure was determined for sheep fecal and tissue samples (liver, lung, spleen, and mesenteric lymph node [MLN]) using field isolate Salmonella Typhimurium 37 (ST37). ST37 was spiked into homogenized feces or tissues across an 8-step, 10-fold serial dilution series. For each dilution, 100 µL was inoculated into 900 µL of MSCB, 10 µL was then inoculated onto XLD agar, and DNA extractions were performed, in duplicate. All XLD agar plates were incubated overnight followed by colony counting. All MSCB suspensions were incubated at 37°C overnight followed by plating in 10-µL volumes onto XLD agar and overnight incubation. The LOD experiments were performed twice using the high-throughput (HT) DNA extraction method, and once using the low-throughput (LT) method for comparison.

Establishing pre-enrichment conditions

Growth curves of ST37 in pre-enrichment broth were performed to establish the optimal broth type and duration of pre-enrichment incubation for Salmonella spp. to reach concentrations adequate for PCR analysis. This was performed on a PBS-only sample matrix and a sheep feces sample matrix (1:10 w/v with PBS). Each sample matrix was spiked with ST37 to give 3 different starting concentrations (~ 5 × 10⁴, 5 × 10², and 5 × 10⁰ cfu/mL) within each matrix. These were analyzed in 2 different enrichment broths—MSCB and buffered peptone water (BPW; Becton Dickinson)—over an 8-h period.

Each sample matrix was prepared and inoculated with the appropriate level of ST37. From this, 4 mL was transferred into 36 mL of the appropriate broth. Broths were then incubated at 37°C for 8 h. At 0-, 2-, 4-, 6-, and 8-h incubation, duplicate aliquots were removed for DNA extraction and culture. For aliquots removed for culture, a 10-fold dilution series in sterile PBS was performed followed by plating onto XLD agar in duplicate 10-µL volumes. All XLD plates were incubated at 37°C overnight followed by colony counting. This was repeated in a second independent experiment.

Validating the 6-h BPW qPCR method

The final 6-h BPW qPCR method was further validated and compared to a standard culture method (MSCB and colony counts on XLD agar), for the detection of Salmonella spp. in sheep fecal and tissue samples. Fecal samples collected from 15 sheep, and tissue samples collected from 12 sheep (3× liver, 3× spleen, 3× lung, and 3× MLN) were used. For spiking fecal samples, 9 Salmonella spp. field isolates were used: Salmonella Typhimurium (n = 3 including ST37), Salmonella Dublin (n = 3), and Salmonella Bovismorbificans (n = 3). For spiking tissue samples, 6 Salmonella field isolates were used: Salmonella Typhimurium (n = 2 including ST37), Salmonella Dublin (n = 2), and Salmonella Bovismorbificans (n = 2).

Fecal homogenates of all samples were prepared by vortexing 1:10 w/v feces in sterile PBS at room temperature until samples were evenly dispersed throughout the liquid. Thirty-eight fecal homogenate samples were inoculated with Salmonella spp., and 15 fecal homogenates were prepared with sterile PBS (negative controls).

All tissue homogenate samples were prepared 1:4 w/v with sterile PBS followed by homogenization at room temperature (Ultra-Turrax T25 disperser; IKA Works) to form a smooth paste. Thirty-two tissue homogenates samples were inoculated with Salmonella spp., and 12 tissue homogenates samples were inoculated with sterile PBS (negative controls).

Pre-enrichment broths were prepared in 9-mL volumes containing either BPW (n = 167) or MSCB (n = 167). To improve the LOD, all broths were pre-heated to 37°C. For spiked sheep fecal (n = 38) and tissue (n = 32) homogenate samples (n = 70 total), duplicate 1-mL volumes were transferred into both BPW and MSCB. This gave a total of 140 spiked BPW and MSCB broths. For negative sheep fecal (n = 15) and tissue (n = 12) homogenate samples (n = 27 total), single 1-mL volumes were transferred into both BPW and MSCB. This gave a total of 27 negative BPW and MSCB broths. All broths were placed in the incubator at 37°C.

For BPW samples, a 2-mL aliquot was removed for DNA extraction processing at 0 h and 6-h incubation. For MSCB samples, 10 µL was transferred onto XLD agar at 0-h and 24-h incubation for colony counting. All XLD plates were incubated at 37°C for 24 h and colonies counted. For all BPW samples, the growth rate (GR) was determined using the following equation²⁴:

$$\mathrm{GR}=\frac{\log \left(N_t\right)-\log \left(N_0\right)}{t}$$

in which N_t and N₀ are the concentration (cfu/mL or copies/mL) at time t and time zero, respectively.

DNA extractions

Sample preparation and extraction modifications specific for each sample type are described in the supplementary material (Suppl. Protocol 1). A blank control (PBS) was included in each batch of extractions.

All HT DNA extraction methods were performed using the BioSprint 96 one-for-all vet kit (Qiagen) and the associated reagents unless specified, following the manufacturer’s instructions for purification of viral nucleic acids and bacterial DNA from animal tissue homogenates, serum, plasma, other body fluids, swabs, and washes. DNA elution was performed in 200-µL volumes for fecal samples, and 100-µL volumes for PBS, tissue, and pre-enrichment broth samples. The kit was run on the MagMAX Express-96 (Thermo Fisher Scientific) using the BS96 Vet instrument protocol for Animal Tissue and Other Sample Types (Qiagen).

All LT DNA extraction methods were performed using the DNeasy blood & tissue kit (Qiagen) and their reagents unless specified, following the manufacturer’s instructions for the purification of total DNA from animal tissues (the spin-column protocol). DNA elution was performed in 100-µL volumes for all sample types.

Quantitative PCR

A TaqMan qPCR assay was designed and optimized for the detection of Salmonella spp. in sheep fecal and tissue samples. The primers and probe were designed and ordered (Design software; LGC Biosearch Technologies) to amplify a 134-bp product of the DNA adenine methylase (dam) gene of Salmonella Typhimurium (GenBank accession NC016863). Primer and probe specificity were validated using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Primer and probe concentrations were optimized from 250 nM to 750 nM and 125 nM to 375 nM, respectively. The annealing temperature was optimized over an 8-step gradient from 58°C to 63°C. The following reaction conditions were chosen as optimal for a 10-μL reaction volume: 1× SsoAdvanced universal probe supermix (Bio-Rad), 750 nM of each primer (damF: 5′-GCAGAAAAAGCGCAGAATGC-3′; damR: 5′-TACGCTGTGAAGTTAGCCGT-3′), 375 nM of probe (damP: 5′-FAM-TCCGCCTTATGCGCCGTTGTC-BHQ-1-3′), and 2 μL of extracted DNA. The qPCR was performed on a CFX96 Touch real-time PCR detection system (Bio-Rad) with cycling parameters of 95°C for 3 min, followed by 40 cycles of 95°C for 10 s and 60°C for 30 s. All DNA samples were run neat in duplicate reactions and diluted (1:5 with elution buffer) in single reactions to assess possible inhibition. Each PCR run included a no-template control (DNA-free water) and a 7-step standard curve of genomic DNA from field isolate ST37 in a 10-fold dilution series from 10 ng/reaction to 1 × 10⁻⁵ ng/reaction. Samples were considered positive if a cycle threshold (Ct) <40 was achieved. Using the standard curve included in each PCR run, results were expressed in ng/reaction. This was then converted to copies/reaction, with 1 ng being equal to 192,297 genome copies based on an estimated genome size of 4,817,868 bp¹² using an online DNA Copy Number Calculator (https://www.thermofisher.com/au/en/home/brands/thermo-scientific/molecular-biology/molecular-biology-learning-center/molecular-biology-resource-library/thermo-scientific-web-tools/dna-copy-number-calculator.html).

To validate DNA extraction from ovine sources, a separate PCR assay was designed as described above to amplify a 109-bp product of the cytochrome B gene of Ovis aries (GenBank accession KY366508.1). Reactions were performed in 10-µL volumes and contained 1× SsoAdvanced universal probe supermix, 500 nM of each primer (CB_ovisF: 5′-GGCACAAACCTAGTCGAATGAATC-3′; CB_ovisR: 5′-CGAGGGCTGCGATGATGAATG-3′), 250 nM of probe (CB_ovisP: 5′-CAL Fluor Orange 560-TGAGGAGGATTCTCAGTAGACAAAGC-BHQ-1-3′), and 2 µL of extracted DNA. Each PCR run contained a no-template control (DNA-free water) and a positive O. aries control from sheep fecal DNA extract and were run using the same conditions and instrument as described above. Single reactions were performed for the O. aries assay.

In order for a sample to be considered positive, a Ct < 40 needed to be achieved in both the Salmonella spp. PCR and the O. aries PCR.

Statistical analyses

Descriptive analyses were performed including graphical summaries and summary statistics. For all statistical analyses, statistical significance was declared at p ≤ 0.05. For comparison of the 6-h BPW qPCR method with the standard culture method, a 2-sample binomial test was performed on all samples. For the final 6-h BPW qPCR method, the 0 h copies/mL values were plotted against the 6 h copies/mL values and the linear dynamic range established. Continued analysis was then conducted on values within the linear dynamic range.

First, a simple linear regression analysis was performed (GenStat 16th ed.) with 0 h copies/mL (log₁₀) as the response variable (Y) and 6 h copies/mL (log₁₀) as the explanatory variable (X). Using the formula generated, the 0-h log₁₀ copies/mL in BPW was predicted for each sample and analyzed against the measured 0-h log₁₀ copies/mL using a paired 2-sample t test (GenStat 16th ed.).

Second, to assess the possible effect of Salmonella spp. serovar and sample type on growth rate, a linear mixed REML (restricted maximum likelihood) analysis was performed (GenStat 16th ed.), with growth rate as the dependent variable, serovar and sample type as the explanatory variables, and sheep of origin and 0 h copies/mL (log₁₀) fitted as the random effects.

Results

Analytical specificity and sensitivity

All Salmonella spp. isolates produced appropriate amplification against the specificity panel (Table 1). No amplification was observed for all other isolates. The sensitivity was determined as 1 × 10⁻⁵ ng/reaction, defined as the lowest DNA concentration in which a minimum of 50% of replicates were detected, as suggested previously.²⁰ With an estimated genome size of 4,817,868 bp,¹² this equated to 1.9 copies/reaction. The average (±SE) amplification efficiency and R² values as determined by the 21 standard curves from 7 Salmonella spp. isolates were 99.7% (±2.3) and 0.99 (±0.002), respectively.

Limit of detection without pre-enrichment

For sheep fecal samples, the LOD across the HT experiments was 4 × 10⁴–6 × 10³, 4 × 10³–6 × 10², and 6 × 10¹–4 × 10¹ cfu/g for qPCR, culture, and broth enrichment, respectively. The qPCR R² value and efficiency were 0.99–1.00 and 88.58–88.67, respectively. For the LT experiment, the LOD was 1 × 10³, 1 × 10², and 1 × 10² cfu/g for qPCR, culture, and broth enrichment, respectively. The qPCR R² value and efficiency were 0.99 and 90.14, respectively. PCR inhibition was not detected in any samples.

For sheep tissue samples, the LOD for any tissue sample across the HT experiments ranged from 4 × 10⁵–4 × 10³, 4 × 10³–4 × 10¹, and 4 × 10¹–1.6 × 10¹ cfu/g for qPCR, culture, and broth enrichment, respectively. The qPCR R² value and efficiency were 0.98–1.00 and 88.10–91.11, respectively. For the LT experiment, the LODs were 4 × 10⁴–2.8 × 10³, 4 × 10²–1.6 × 10², and 4 × 10¹–1.6 × 10¹ cfu/g for qPCR, culture, and broth enrichment, respectively. The qPCR R² value and efficiency were 0.99–1.00 and 87.28–90.69, respectively. Given tissue availability, MLN tissue was not examined in HT experiment 2. PCR inhibition was not detected in any samples.

Establishing pre-enrichment conditions

Growth curves were plotted for ST37 PBS and the ST37 sheep fecal sample matrixes in pre-enrichment broths (Figs. 1, 2). Each sample matrix was spiked with ST37 to give 3 different starting concentrations (~ 5 × 10⁴, 5 × 10², and 5 × 10⁰ cfu/mL) within each sample matrix. From these solutions, 4 mL was transferred into 36 mL of the appropriate pre-enrichment broth. The ST37 starting concentrations in the pre-enrichment broths were determined by culture (cfu/mL) and by PCR (copies/mL). For starting concentrations below the LOD, the concentration was estimated based on the higher concentration(s) and the appropriate dilution.

Figure 1.

Phosphate-buffered saline sample matrix trials 1 (A) and 2 (B). Growth curves of Salmonella Typhimurium 37 in pre-enrichment broth following low-, medium-, and high-dose inoculation. PCR data derived from duplicate DNA extractions analyzed in duplicate reactions (n = 4 results/concentration). Culture data derived from duplicate plating (n = 2 results/concentration). Dashed horizontal line at 2.00 = theoretical limit of detection for culture and quantitative PCR. BPW = buffered peptone water; MSCB = mannitol–selenite–cystine broth.

Figure 2.

Sheep feces sample matrix trials 1 (A) and 2 (B): growth curves of Salmonella Typhimurium 37 (ST37) in pre-enrichment broth following low-, medium-, and high-dose inoculation. PCR data points derived from duplicate DNA extractions analyzed in duplicate reactions (n = 4 data points/concentration). Accurate ST37 quantitative culture data could not be obtained because of the growth of coliform colonies and so were excluded. Dashed horizontal line at 2.00 = theoretical limit of detection. BPW = buffered peptone water; MSCB = mannitol–selenite–cystine broth.

For graphical reasons, if the concentration of ST37 was below the LOD at any sampling point, the starting concentration was assigned. For trials 1 and 2 with the ST37 sheep feces matrix, culture results were excluded because of coliform growth, making accurate quantification of ST37 not possible. For BPW and MSCB, PCR inhibition was not detected in any samples.

All inoculation levels were detectable by PCR in both broths and matrixes by 6-h incubation. The culture method allowed identification in both broths by 6 h for the PBS matrix; however, the culture method was not successful for the feces matrix because of coliform growth. Samples inoculated into BPW had a noticeably higher growth rate compared to MSCB up to 6 h. At 8 h, the growth rate appeared similar for BPW and MSCB.

Validating 6-h BPW PCR method

The breakdown of enrichment broths by isolate and sample type are displayed in the supplementary material (Suppl. Table 1). The concentrations of Salmonella spp. isolates in spiked BPW at 0 h were determined by PCR (copies/mL). Starting concentrations were created using known dilutions of Salmonella spp. isolates. Therefore, for samples that were below the LOD at 0 h, the starting concentrations were estimated according to the calculated concentrations in BPW above the LOD, multiplied by the appropriate dilution factors. The number of cfu/mL of Salmonella spp. in spiked BPW and MSCB at 0 h were also estimated according to the number of cfu/mL present in the initial LB culture used for spiking the broths, multiplied by the appropriate dilution factors.

For fecal homogenates in broth, at 0 h, the LOD for spiked BPW samples by PCR and spiked MSCB samples by culture was 3.7 × 10² cfu/mL (5.08 × 10² copies/mL) for both methods. This equates to 3.7 × 10⁴ cfu/g (5.08 × 10⁴ copies/g) in sheep feces for both methods. Following incubation, the LOD was 9.0 × 10⁻² cfu/mL (2.57 × 10⁻¹ copies/mL) for both methods. This equates to 9 cfu/g (2.57 × 10¹ copies/g) in sheep feces for both methods.

For tissue homogenates in broth, at 0 h, the LOD for spiked BPW samples by PCR and spiked MSCB samples by culture was 1.35 × 10³ cfu/mL (8.05 × 10² copies/mL) and 1.42 × 10² cfu/mL, respectively. This equates to 5.4 × 10⁴ cfu/g (3.2 × 10⁴ copies/g) and 5.7 × 10³ cfu/g in sheep tissue. Following incubation, the LOD was 1.35 × 10⁻¹ cfu/mL (8.05 × 10⁻² copies/mL) and 7.5 × 10⁻² cfu/mL, respectively. This equates to 5.4 cfu/g (3.22 copies/g) and 3 cfu/g in sheep tissue.

For all sample types, the 6-h BPW PCR method detected 69 of 140 (41%) broth samples as positive prior to incubation, with a sensitivity and specificity of 49% and 100%, respectively. Following incubation, 128 of 167 (77%) broth samples were detected as positive, with a sensitivity and specificity of 91% and 100%, respectively. The MSCB culture method detected 70 of 167 (42%) broth samples as positive prior to incubation, with a sensitivity and specificity of 50% and 100%, respectively. Following incubation, 129 of 167 (77%) broth samples were detected as positive, with a sensitivity and specificity of 92% and 100%, respectively. There was no significant difference between the proportion of samples identified as positive by the 2 methods both prior to incubation (p = 0.912) and following incubation (p = 0.897). All negative samples were negative for Salmonella spp. by both methods at both pre- and post-incubation (n = 15).

The linear dynamic range of the correlation between the concentrations of spiked Salmonella spp. sheep samples in BPW at 0-h and at 6-h incubation was from 2.03 × 10⁵ to 1.45 × 10² copies/mL at 0 h (Fig. 3). A significant association was demonstrated between 6-h log₁₀ copies/mL and the 0-h log₁₀ copies/mL for fecal (p < 0.001; 95% CI: 1.05–1.18; R² = 0.96) and tissue (p < 0.001; 95% CI: 0.96–1.20; R² = 0.85) samples when analyzed separately, and when analyzed together (p < 0.001; 95% CI: 1.02–1.16; R² = 0.89). When analyzed together, the Salmonella serovar had a significant effect on growth rate (p < 0.001). Sample type (feces or tissue) did not have a significant effect (p = 2.94); however, the interaction between serovar and sample type did have a significant effect (p < 0.001; Table 2).

Figure 3.

Correlation between the concentration (copies/mL) of Salmonella-spiked sheep samples in buffered peptone water at 0-h and at 6-h incubation within the linear dynamic range (n = 112).

Table 2.

The predicted mean growth rates of Salmonella serovars in sheep feces or tissue samples following enrichment in buffered peptone water for 6 h. Data points derived from samples within the linear dynamic range (n = 112).

Sample type	Predicted mean growth rate (expressed as PCR log copies/mL)
	Salmonella Bovismorbificans	Salmonella Dublin	Salmonella Typhimurium
Feces	0.67 (±0.02)a	0.64 (±0.02)a,b	0.66 (±0.02)a
Tissue	0.64 (±0.02)a,b	0.59 (±0.03)b	0.74 (±0.02)

^a,bPredicted mean values with different letters are significantly different (p ≤ 0.05). Standard errors in parentheses.

From the regression analysis of all samples within the linear dynamic range, the following formula was generated to estimate the 0-h log₁₀ copies/mL in BPW:

$$0-hlo{g}_{10}copies/mL=1.0892x\_4.540$$

in which x is 6-h log₁₀ copies/mL in BPW. Using this formula, the 0-h log₁₀ copies/mL in BPW was predicted for each sample and plotted against the measured 0-h log₁₀ copies/mL in BPW (Fig. 4). No significant difference was demonstrated between the predicted and measured 0-h log₁₀ copies/mL in BPW (p = 0.998; 95% CI: −0.13 to 0.13).

Figure 4.

Distribution of the predicted and measured 0-h concentration (copies/mL) of Salmonella in buffered peptone water (BPW)-enriched sheep tissue and fecal samples. Data derived from samples within the linear dynamic range (n = 112).

Discussion

Our initial investigations were made into the use of qPCR for the direct detection of Salmonella spp. in sheep samples without the use of a pre-enrichment step. In sheep fecal samples, the qPCR LOD was 6 × 10³ and 1 × 10³ cfu/g for the HT and LT method, respectively. This was higher than that of culture (6 × 10² and 1 × 10² cfu/g) and pre-enrichment (4 × 10¹ and 1 × 10¹ cfu/g) when compared directly. This can be expected because, given the complex nature of sheep fecal samples, several dilution steps in various buffers and reagents are necessary to combat the effects of qPCR inhibitors. Our results are comparable with previous studies on Salmonella spp.–spiked pig fecal samples with a reported LOD of 10³ cfu/g¹⁹ and poultry fecal samples with an LOD of 4 × 10³ cfu/g.²⁴ Values as low as 1.25 × 10³ Salmonella spp. organisms/g of sheep feces have been reported based on a standard curve of plasmid DNA.³¹ Using a high-throughput PCR system for the detection of Mycobacterium avium subsp. paratuberculosis in sheep and cattle fecal samples, the LOD was determined to be as low as 10¹–10² organisms/g of feces.²⁰ However, this application involved a multi-copy gene.²⁰

We found similar LOD results in spiked tissue samples as for the fecal samples. Although dilution of inhibitors is less of an issue in tissue samples, the quantity of target DNA can be an issue. The extraction kits that we used have a recommended upper limit of tissue to be used given the DNA-binding capacity of the kits. For the BioSprint 96 one-for-all vet kit (Qiagen), this was 10 mg. For the DNeasy blood & tissue kit (Qiagen), this was 25 mg. This limitation in sample quantity immediately reduces the possible LOD. Comparisons of these results with previous studies are difficult because little work has been done on direct PCR without pre-enrichment for the detection of Salmonella spp. in tissue samples. For other organisms, an LOD of 2.6 × 10⁴ cfu/g for Mycoplasma bovis in cattle lung samples has been reported.³

Although the LOD without pre-enrichment in our study may allow identification of Salmonella spp. in fecal and tissue samples from animals during the clinical stages of disease, there is the potential that low fecal shedding and tissue colonization may go unidentified. In sheep challenged with Salmonella Typhimurium, the mean shedding ranged from ~ 10³ to 10⁶ cfu/g of feces over 13 d post-challenge.¹⁵ In the same study, the mean tissue colonization at d 13 post-challenge ranged from ~ 10¹ to 10³ cfu/g depending on tissue type.¹⁵ Sampling of sheep during slaughter at 2 Australian abattoirs found that the mean count of Salmonella spp. in positive fecal samples was 2.7 × 10¹ MPN/g.⁵ Hence, in order for the qPCR to detect down to these levels, DNA extraction without a pre-enrichment step would not be adequate compared to standard culture techniques.

When exploring the inclusion of a shortened pre-enrichment step, despite MSCB reducing coliform growth, BPW still produced consistently higher growth rates of ST37 at 6 h and 8 h of incubation. By 6 h, all 3 inoculation levels were detectable by PCR. Between 6 h and 8 h, the growth curve appeared to demonstrate the beginning of stationary phase characteristics for the higher inoculation levels and, thus, would not be ideal for linear correlation analysis. We therefore determined that 6 h of pre-enrichment in BPW was optimal. Although many studies utilize a pre-enrichment time of 16–24 h prior to PCR analysis,^6,9,27 reduced pre-incubation times for the detection of Salmonella spp. have been investigated as a more rapid approach.^11,17 When analyzing Salmonella spp.–positive chicken samples in BPW by PCR every 2 h of incubation, no samples were detectable prior to 8 h of pre-enrichment, with 18 h required for 100% detection.¹⁷ However, pre-enrichment of 8 h in BPW followed by PCR for the detection of Salmonella spp. in pork and chicken meat resulted in relative sensitivity and specificity of 98% and 100%, respectively.¹¹ Upon evaluation of our 6-h BPW qPCR method, at 6-h incubation we achieved sensitivity and specificity of 91% and 100%, with an LOD of 9 cfu/g (2.57 × 10¹ copies/g) in sheep feces, and 5.4 cfu/g (3.22 copies/g) in sheep tissue. This was a substantial improvement from initial evaluations of sheep samples without pre-enrichment.

The use of a pre-enrichment step has been explored to improve the quantification limit of Campylobacter spp. in chicken rinses, while still allowing semiquantitative results to be obtained.¹⁰ Following selective enrichment and DNA extraction, a correlation was observed between PCR Ct values and the initial concentration of Campylobacter spp. in cfu/mL of spiked samples.¹⁰ Similarly, a significant correction between PCR Ct values and the initial concentration of Morganella spp. in fish samples following pre-enrichment has been reported.²¹ In our study, this concept was explored further for Salmonella spp. A significant correlation was observed between the concentration of Salmonella spp. in BPW at 0 h and 6 h of pre-enrichment, with a linear dynamic range observed from 2.03 × 10⁵ to 1.45 × 10⁻² copies/mL at 0 h. Hence, samples with a starting concentration > 2.03 × 10⁵ copies/mL in BPW would be unlikely to give accurate quantification at 6 h. To overcome this, it may be necessary to first quantify all samples in BPW at 0 h. Only those that are below the limit of quantification (1.28 × 10³ copies/mL) at 0 h would then be quantified at 6 h, with the 0-h concentration estimated using the developed regression formula. This would allow samples within a wide range of starting concentrations to be quantified.

Serovar and the interaction of serovar with sample type (tissue vs. feces) were found to have a significant effect on the growth rate of isolates within the linear dynamic range. Salmonella Typhimurium isolates had a significantly higher growth rate in tissue samples than the isolates of all other serovars in tissue and feces. However, sample type alone had no significant effect on growth rate and, within fecal samples, there was no significant differences between serovars. This could therefore suggest that the differences observed may be the result of specific tissue types (i.e., liver, lung, spleen, and MLN) rather than serovars, possibly because of differences in available nutrients. Given sample number constraints, all serovars could not be analyzed in all tissue types, with lung tissue being the only tissue type to include all serovars.

Typically, in diagnostic laboratories, samples would be homogenized straight into the enrichment broth.^8,9 In our study, we first homogenized fecal and tissue samples in PBS before transferring into enrichment broth, creating an additional dilution step that should be taken into consideration when interpreting these results. Although this may increase the LOD theoretically, it also creates several positive allowances. First, only one sample homogenate preparation is required, which can then be inoculated into multiple different enrichment broths or agar plates if trying to quantify multiple bacterial species that have different growth requirements. Second, by creating a homogenate in PBS, the samples are better preserved until inoculated into broth. With sample homogenization being one of the more time-consuming steps (particularly for the tissue samples), this allows samples to be prepared and inoculated in batches, creating a more streamlined process.

Although the 6-h BPW qPCR method demonstrated a substantial improvement in the LOD while remaining semiquantitative, our initial work has only been performed on artificially inoculated samples. Hence, the bacterial cells are less likely to have experienced sublethal injury that could result in delays in multiplication.²⁶ Although such a phenomenon would also affect any quantification method that does not involve broth enrichment, further validation of the 6-h BPW qPCR method is required against the standard plate count method in samples naturally infected with Salmonella spp. to fully evaluate its potential use.