Sensitivity of Mycobacterium avium subsp paratuberculosis, Escherichia coli and Salmonella enterica serotype Typhimurium to low pH, high organic acids and ensiling



Kimberly Cook, USDA-ARS, 230 Bennett Lane, Bowling Green, KY 42104, USA.




To evaluate the persistence of Mycobacterium avium subsp paratuberculosis (Mycoparatuberculosis), Salmonella enterica serotype Typhimurium (Salm.Typhimurium) and a commensal Escherichia coli (E. coli) isolate under the low pH and high organic acid (OA) conditions of ensiling of forages.

Methods and Results

Decay rates and the time required to obtain a 90% reduction in cell concentration were calculated following (i) exposure to buffered OA (pH 4·0, 5·0, 6·0 or 7·0) (ii) exposure to silage exudates and (iii) survival through ensiling of forage materials. Salm. Typhimurium had higher decay rates in silage exudates (−0·5601 day−1) than did E. coli (−0·1265 day−1), but both exhibited lower decay rates in silage than in OA or silage exudates. Mycoparatuberculosis showed no decrease in silage and decay rates in silage exudates were significantly lower (2–12 times) than for the other two organisms.


Escherichia coli, Salm. Typhimurium and Mycoparatuberculosis exhibit marked differences in response to acidity. All three organisms show acid resistance, but Mycoparatuberculosis in particular, if present in manure and applied to forage grasses, may survive the low pH and high OA of the ensilaging process; silage may therefore be a potential route of infection if ingested by a susceptible animal.

Significance and Impact of Study

This information contributes to the understanding of potential risks associated with silage preservation and contamination of livestock feed with manure-borne pathogens.


Although the majority of bacteria associated with manures are beneficial and/or innocuous, the potential for the contamination of agricultural environments, livestock and crops with manure-borne pathogens necessitates greater knowledge of their persistence. Studies have shown that livestock water and feed sources serve as reservoirs for pathogenic organisms and as a vehicle for the transmission of pathogens through herds (Ryser et al. 1997; Davis et al. 2003; Johansson et al. 2005; Wagner et al. 2005; Cook et al. 2010). As a result, there has been increased emphasis on understanding how manure-borne pathogens are spread between and within farms, focusing on contamination sources on the preharvest side of food production (i.e. on the farm environment) using the ‘farm-to-fork’ approach (Davis et al. 2003; CDC 2011; Doyle and Erickson 2012).

Silage is an important source of feed for livestock (McEniry et al. 2010). The ensiling process is used to preserve livestock feed via anaerobic fermentation of water-soluble carbohydrates by lactic acid bacteria (LAB) (Chen et al. 2005; Stevenson et al. 2006; Duniére et al. 2011). Die-off of undesirable microbial populations in ensiled materials results from the production of weak organic acids (OAs, including lactic, acetic and propionic) and concomitant decreases in pH (around pH 4·0). Pathogens of concern in silage therefore include those that are resistant to acids. Many enteric bacteria, including Escherichia coli (E. coli) and Salmonella spp., commonly endure extremely low pH values due to gastric and volatile fatty acids in the stomach, intestine and faeces (Bearson et al. 1997; Álvarez-Ordóñez et al. 2012). Intracellular pathogens such as Salmonella and Mycobacteria must also tolerate pH values below 5·0 inside of activated macrophage (Bearson et al. 1997; Vandal et al. 2008). Therefore, many of these organisms have acid resistance or acid tolerance mechanisms that enable them to persist under moderate to extreme acid stress (Diez-Gonzalez and Russell 1997; Foster 2004; Álvarez-Ordóñez et al. 2012).

Commonly ensiled crops include corn and grass forage, both of which are often fertilized with livestock manure spread by distributing onto the soil or by spray irrigation (Johansson et al. 2005). Pathogen contamination may result from the application of contaminated manure or wash waters to silage crops. In particular, recent studies have evaluated the survival of foodborne pathogenic E. coli strains in silage materials (Byrne et al. 2002; Pedroso et al. 2010; Duniére et al. 2011). In those cases, die-off generally occurred within three to five days. However, persistence of other pathogens (Listeria monocytogenes (L. monocytogenes), Salm. enterica, enterococci and streptococci) in silage has been more variable (Ryser et al. 1997; Martínez-Gamba et al. 2001; Johansson et al. 2005; Ramírez et al. 2005; Petersson-Wolfe et al. 2011). In a study to determine routes of transmission of L. monocytogenes on dairy farms producing cheese from raw milk, Schoder et al. (2011) found that the organism was three to seven times more likely to occur on farms where silage was fed than on those not feeding silage. These studies have made it apparent that more information is needed to understand factors important to forage quality for the sake of livestock health and public safety.

Salmonella enterica serovar Typhimurium (Salm. Typhimurium) and Mycobacterium avium subsp paratuberculosis (Mycoparatuberculosis) are two pathogens that are of concern to food and animal safety due to their ability to withstand harsh conditions found in the environment and inside the host. Salmonella is one of the most common causes of foodborne illness, causing an estimated $365 million in direct medical costs annually (CDC 2011). It is cited by CDC as a targeted pathogen because its incidence has not declined significantly in more than a decade. Mycoparatuberculosis is the causative agent of Johne's disease, considered one of the most serious diseases of dairy cattle. Johne's disease has been reported worldwide and has caused devastating economic losses resulting from premature culling and reduced milk production from infected animals (Harris and Barletta 2001; Rowe and Grant 2006). Both of these organisms are known to persist in acidic environments. Although several studies have evaluated the survival of Salmonella in silage, results have been inconsistent (Martínez-Gamba et al. 2001; Johansson et al. 2005; Ramírez et al. 2005) and only two studies have been conducted on survival of Mycoparatuberculosis in silage (Katayama et al. 2000; Khol et al. 2010). Those studies verified the presence of Mycoparatuberculosis in silage and survival through the process. However, questions remain regarding response of both organisms to ensiling in forages, including their: (i) die-off rates, (ii) resistance to low pHs and (iii) susceptibility to OA produced during the process.

Monitoring contamination using indicators and in particular Escherichia coli (E. coli) has a long and well-documented history. E. coli is a dominant intestinal commensal organism and an important faecal indicator bacterium. Comparison of its survival to that of the pathogens will provide important information on the adequacy of E. coli as an indicator of silage safety in relationship to these two important pathogens.

As livestock production continues to intensify, the reliance on ensiled feed materials will also increase. Understanding risks associated with feeding ensiled materials is a necessary step in assuring food safety in these intensive production systems. Therefore, the goal of this study was to evaluate the ability of the pathogens Mycoparatuberculosis and Salm. Typhimurium and a commensal E. coli isolate to persist under the low pH and high OA concentrations encountered as part of the ensiling process. Results from these studies will provide new information on the safety of silage as feed and on the validity of using indicators to monitor pathogen contamination in ensiled materials.

Materials and methods

Bacterial culture preparation

Cultures of Mycoparatuberculosis, Salm. Typhimurium and E. coli were prepared for inoculation into forage to be used for ensiling, or into silage exudate or buffered pH solutions. Mycoparatuberculosis isolated from the ileum of a clinically infected, Johne's positive dairy cow (Bolster et al. 2008) was inoculated into flasks containing Middlebrook 7H9 broth (Remel, Lenexa, KS, USA; one-litre total volume) with 2 ml l−1 glycerol, 10% Middlebrook OADC (oleic acid–albumin–dextrose–catalase supplement; Becton Dickinson; Sparks, MD, USA) and 2 mg l−1 Mycobactin J (Allied Monitor; Fayette, MO, USA). Cells were grown for 21 days at 37°C with constant mixing. The culture was pelleted and washed three times in sterile phosphate-buffered saline (PBS, 137 mmol l−1 NaCl, 2·7 mmol l−1 KCl, 4·3 mmol l−1 Na2HPO4 and 1·4 mmol l−1 KH2PO4) by centrifugation at 4251 × g for 30 min. Cell clumps were disrupted by vortexing for two min with 2-mm sterile glass beads and filtering through an 80-μm filter to remove any large clumps. The final pellet was resuspended in 100 ml of 0·1× strength PBS. Samples were plated onto nutrient agar media and incubated overnight to confirm the lack of contaminants.

Cultures of Salmonella enterica subsp. enterica serovar Typhimurium strain ATCC 13311 and E. coli [isolated from swine slurry (Bolster et al. 2010)] were inoculated into 100 ml batches of tryptic soy broth (Fisher Scientific; St. Louis, MO, USA) and grown at 37°C for 24 h. Cultures were pelleted by centrifugation for 10 min at 3700 × g. Pellets were vortexed for 2 min and filtered through an 80-μm filter to remove clumps. Killed inocula were prepared by autoclaving for 30 min; cultures were then plated to confirm the lack of culturable cells. Triplicate samples were removed from all final inocula for cultural and molecular analysis of cell concentrations as described below.

Survival in buffered citric acid at pH 4, 5, 6 or 7

The survival of Myco. paratuberculosis, Salm. Typhimurium and E. coli was assessed in citric acid buffer adjusted to pH 4, 5, 6 or 7. This buffered OA was used after preliminary studies with Myco. paratuberculosis showed that decay in unbuffered acetic acid occurred too rapidly to enable reliable quantification. Buffered OA was prepared as previously described (McIlvaine 1921). Citric acid buffer solutions (20 ml) were inoculated in triplicate with targeted organisms, incubated at 37°C and sampled 0, 2, 7, 16, 29 and 37 days after inoculation. Initial cell concentrations averaged 8·8 ± 2·3 × 10cells ml−1, 1·2 ± 0·9 × 10cells ml−1 and 2·1 ± 1·2 × 10cells ml−1 for Myco. paratuberculosis, Salm. Typhimurium and E. coli, respectively. Culturable, live and live/intact dead cells were quantified as described below.

Harvesting, inoculation and ensiling in forages

Fresh grass/alfalfa forage was harvested at 33% dry matter (DM) from a field located at Western Kentucky University agricultural farm (Bowling Green, KY, USA) and adjusted to achieve a 19–20 mm length of cut. Forage was divided into seven batches before sterile water with added inoculum (live or dead Myco. paratuberculosis, Salm. Typhimurium or E. coli; additional water was added to the control) was sprayed onto batches of forage to achieve a targeted DM content of 30%–40%. Inocula were thoroughly mixed into forage before 350 g was transferred into plastic bags (silage bags; 20·3 × 27·9 cm) and vacuum-sealed (Food Saver®, Seal-a-Meal™, Jarden Corp., Neosho, MO, USA) to make them air tight for ensiling inside of a dark growth chamber at 25°C. For each treatment, three replicate silage bags were prepared for sampling at 0, 12, 25, 50, 75 and 150 days after ensiling (n = 126). Increasing replication may have improved robustness of statistical analyses; however, the need for sampling over time increased the total number of samples taken and therefore set the manageable number of replicates possible. Silage pH, DM and OA concentrations for uninoculated controls were measured at each sample point (Dairy One Forage Testing Lab; Ithaca, NY, USA) to determine the quality of the sample fermentation. Samples (30 g) were blended to homogenize (Hamilton Beach;Southern Pines, NC, USA) prior to taking subsamples for molecular microbial analysis as described below.

Exposure to silage exudates

To determine the susceptibility of Myco. paratuberculosis, Salm. Typhimurium and E. coli to OA produced during the ensiling process, each organism was exposed to silage exudates. A grass/alfalfa mix with no inocula was ensiled as described above for 25 days, and then, 50 g was mixed with 200 ml sterile water and soaked for 1 h before removing supernatant (silage exudates) and successively filtering to sterilize. Exudate pH (Fisher Scientific; Hampton, NH, USA) and electrical conductivity (EC; CON 11 conductivity meter (Oakton Instruments; Vernon Hills, IL, USA)) were determined. Exudates (20 ml) were inoculated in triplicate with targeted organisms, incubated at 37°C and sampled 0, 1, 2, 7, 14, 21 and 30 days after inoculation. Initial cell concentrations averaged 7·2 ± 1·8 × 10cellsml−1, 2·1 ± 0·2 × 10cells ml−1 and 3·6 ± 0·9 × 10cells ml−1 for Myco. paratuberculosis, Salm. Typhimurium and E. coli, respectively. Live and live/intact dead cells were quantified as described below.

Cell culture and propidium monoazide treatment

Culturable cells, those that contained intact cell walls or all cells of the targeted organism (live and dead) were quantified by selective plating, quantitative real-time PCR (qPCR) with propidium monoazide (PMA) treatment or qPCR without PMA treatment, respectively. Culturable cell numbers were obtained by serially diluting and plating onto eosin methylene blue (EMB) or Salmonella Shigella (SS) agars (BD, Sparks, MD, USA) to measure E. coli and Salm. Typhimurium, respectively. Myco. paratuberculosis cell concentrations were determined by plating onto 7H9 media with added OADC, Mycobactin J and agar (Remel, Lenexa, KS, USA). Media were incubated for 24 h at 37°C prior to quantification of culturable E. coli and Salm. Typhimurium and for 15–21 days for the quantification of culturable Myco. paratuberculosis.

To quantify intact cells by molecular analysis, liquid samples (500 μl) from silage exudate or buffered pH solutions were treated with PMA (Biotium Inc.; Hayward, CA, USA). PMA is a DNA-intercalating dye that binds DNA that is not protected within an intact bacterial cell wall (Nocker et al. 2007). Exposure to light makes the photoinducible azide of PMA insoluble, thereby selectively removing DNA from dead cells from any genomic DNA extractions. Samples treated with 50 μmol l−1 final concentration of PMA were incubated with shaking for 5 min on ice in the dark and then exposed to light (650-W halogen) for 2 min at a distance of 20 cm from the light source according to the method of Nocker et al. (2007). PMA-treated samples were then used for DNA extraction and qPCR analysis as described below.

DNA extraction and quantitative real-time PCR analysis of target gene sequences

DNA was extracted from silage (300 mg) or liquid (300 μl) samples in triplicate using the FastDNA® Spin kit for soils (MP Biomedical; Solon, OH, USA) according to the manufacturer's specifications. Quantitative real-time PCR (qPCR) was used to determine the concentration of total cells, Myco. paratuberculosis, E. coli and Salm. Typhimurium and LAB using primers and probes and the previously described qPCR protocols for each organism (Table 1). The primers were obtained from Sigma Genosys (St. Louis, MO, USA), and the dual-labelled Black Hole Quencher probes were prepared by Biosearch Technologies, Inc. (Novato, CA, USA). Assays were carried out in Qiagen HotStarTaq Master Mix (Qiagen; Valencia, CA, USA) in a total volume of 25 μl. The amplification mixture contained 3·0 mmol l−1 MgCl2, 600 nmol l−1 each primer, 200 nmol l−1 probe and sample DNA (1 to 10 ng) or standard (from 101 to 108 copies). Cell concentrations were calculated by dividing the copy number per g or ml of sample by the average copy number of the targeted gene as shown in Table 1. qPCR assays were run on the Bio-Rad CFX 96 Real-time PCR Detection System (Bio-Rad; Hercules, CA, USA) in a total volume of 25 μl. Baseline values were set as the lowest fluorescence signal measured in the well over all cycles. The baseline was subtracted from all values, and the threshold was set to one standard deviation of the mean. All PCR runs included duplicates of standards and control reactions without template. Standard DNA consisted of plasmid PCR 2.1 vector (Invitrogen, Carlsbad, CA, USA) carrying the appropriate insert for the given assay. DNA concentrations in each extraction were determined using the Nanodrop 100 (Thermo Scientific; Wilmington, DE, USA) according to the manufacturer's instructions.

Table 1. Sequences, target size and Tm of primers used in this study
TargetGeneOligo nameSequence (5′-3′)Tm (°C)Target size (bp)Target copy No.Reference
  1. Tm = Melting temperature.

All bacteria16 S rRNA gene1055-FATG GCT GTC GTC AGC T54·03374Harms et al. (2003)
Lactic acid bacteria recA LPla1FAGGCGCGGCTGATGTCA69·7681Stevenson et al. (2006)
Escherichia coli uidA UidA784 FGTG TGA TAT CTA CCC GCT TCG C66·5821Frahm and Obst (2003)
Salmonella enterica ttrBCA ttrA-4-sal-FAGC TCA GAC CAA AAG TGA CCA TC66·5941Malorny et al. (2004)
Mycobacterium paratuberculosis IS900 Skmap FAATGACGGTTACGGAGGTGGT66·47614Kim et al. (2002)

Decay rates and statistics

Survival of targeted populations was quantified in two ways: calculation of a decay rate constant (k) and the time required to obtain a 90% reduction, T90, in cell concentration. Decay rates were calculated using the exponential decay model Nt = N0ekt, where Nt is the concentration at time, t, N0 is the starting population at time, = 0, and k is the decay rate constant (Klein et al. 2011; Moriarty et al. 2011). The decay rate (day−1) was obtained from the slope of the exponential portion of a plot of ln (Nt/N0) with regression parameters calculated using the linear curve fit function in SigmaPlot 11.0 (Systat Software, San Jose, CA, USA). The calculated decay rates were used to determine days to reach 90% die-off where T90 was –2·303/k obtained by solving equation (1·0–0·9) Nt = N0ek T90, leading to ln(0·1) = −k T90. Tukey's post hoc tests ( were calculated to determine significance of differences in mean decay constants (slopes of plots of normalized data ln (Nt/N0) versus time in days) between Mycoparatuberculosis, Salm. Typhimurium and E. coli.


Survival in buffered citric acid at pH 4, 5, 6 or 7

To compare the survival of Myco. paratuberculosis, E. coli and Salm. Typhimurium at different pH values, decay constants were calculated and T90 values (time to 90% reduction in cell concentration) were obtained from plots of the change in cell concentrations over time as quantified by viable plate counts or by PMA-qPCR. No decay rates were calculated for cultured Salm. Typhimurium and E. coli in pH 4 solutions because both became unculturable within 7 days. Mycoparatuberculosis remained culturable in pH 4 solutions for at least 7 days (1·18 ± 0·53 × 10cells ml−1; Table 2). At all other pH values, Myco. paratuberculosis had significantly greater decay rates and lower T90 values than E. coli and Salm. Typhimurium (< 0·05; Table 2). On the other hand, the decay rates for E. coli and Salm. Typhimurium were similar at all pH values (> 0·05) except at pH 7 by culture methods (< 0·05). To evaluate the relationship between viable plate counts and PMA-qPCR, coefficients of determination for regression of data from viable plate counts and PMA-qPCR analysis were calculated for E. coli, Salm. Typhimurium and Myco. paratuberculosis cultures incubated at pH 6. Results suggest that there was a good linear relationship between the two measures (r2 = 0·0·85, 0·92 and 0·92, for E. coli, Salm. Typhimurium and Myco. paratuberculosis, respectively).

Table 2. Decay rates (k day1) and time (days) to achieve 90% reduction in culturable cells or gene copies in citric acid solution at pH 4, 5, 6 or 7
Sample Mycobacterium paratuberculosis Escherichia coli Salmonella Typhimurium
T90akb T 90 a kb T 90 a kb
  1. a

    T90 = Time in days to achieve 90% reduction in cell numbers.

  2. b

    Decay constant (k) calculated by applying the best-fit slope to plot with x and y = ln(Nt/N0).

  3. c

    No significant decay based on PMA-qPCR gene copy numbers.

Viable plate counts
pH 43·6−0·640<6·0>−0·386<4·7>−0·493
pH 53·7−0·61412·7−0·18213·4−0·172
pH 64·0−0·56713·2−0·17510·2−0·225
pH 75·1−0·45417·9−0·12810.3−0·224
qPCR: PMA treated
pH 49·1−0·25214·00·16414·8−0·156
pH 53·7−0·6218·50·2719·50·242
pH 67·9−0·29112·00·19212·7−0·181
pH 711·6−0·198cccc

Silage quality

The dry matter of the silage averaged 56·1% ± 2·3% initially, but remained at 48% ± 0·4% for the remainder of the study. Silage pH decreased from 5·9 to 4·6 within 12 days and remained between 4·3 and 4·6 for the remainder of the experiment (Table 3). Lactic acid and acetic acid were the dominant acids, while propionic and butyric acids were below the levels of detection. Concentrations of lactic acid bacteria increased significantly (P = 0·003) within 12 days of ensiling of the forage and remained a dominant portion of the total population for the duration of the experiment (Table 3).

Table 3. Silage pH, fermentation products, total and lactic acid bacterial populations
Days of ensilingpHLAAAPABALAB (log10 cells g−1)Total
  1. Values represent the average ± standard deviation of triplicate silage bags.

  2. Values are given in g kg−1 dry matter unless otherwise noted.

  3. LA, lactic acid; AA, acetic acid; PA, propionic acid; BA, butyric acid; NH3-H, ammonia-N; LAB, lactic acid bacteria.

05·9 ± 0·10·50 ± 0·000·000·000·005·28 ± 0·778·42 ± 0·23
124·6 ± 0·124·80 ± 13·358·47 ± 3·060·000·008·22 ± 0·228·39 ± 0·33
254·6 ± 0·033·07 ± 11·3213·63 ± 1·420·000·008·25 ± 0·278·72 ± 0·06
504·3 ± 0·137·90 ± 7·4012·20 ± 0·260·000·008·29 ± 0·088·67 ± 0·13
754·3 ± 0·137·37 ± 8·0416·30 ± 0·920·000·008·20 ± 0·128·55 ± 0·10
1504·3 ± 0·140·60 ± 9·7614·33 ± 3·850·000·006·51 ± 0·966·74 ± 1·14

Detection of Myco. paratuberculosis, Salm. Typhimurium or E. coli in silage

To quantify the die-off of the pathogens in ensiled material, cell concentrations were measured on days 0, 12, 25, 50, 75 and 150 by qPCR. Initial concentrations of inoculated pathogens averaged 4·0 ± 2·1 × 105, 2·8 ± 1·2 × 106 and 1·5 ± 0·6 × 10cells g−1 silage for Myco. paratuberculosis, Salm. Typhimurium and E. coli, respectively. Concentrations of Myco. paratuberculosis and E. coli increased between day 0 and day 12 to 1·5 ± 1·3 × 106 and 7·4 ± 1·4 × 106 cells g−1 silage for Myco. paratuberculosis and E. coli, respectively; therefore, calculation of die-off for those organisms was based on changes from day 12. Salm. Typhimurium cell concentrations decreased exponentially in the silage from day 0 until the end of the experiment (Fig. 1). While no Salm. Typhimurium could be detected in the silage by the end of the experiment, E. coli concentrations were still around 10% of day 12. The concentration of Myco. paratuberculosis IS900 detected in silage never dropped below the levels found on day 12 of sampling (ending at 2·75 ± 0·97 × 106 cells g−1 silage). Silage inoculated with dead cultures of Myco. paratuberculosis, Salm. Typhimurium or E. coli was also tested. As with silage inoculated with live culture, concentrations of Myco. paratuberculosis in silage inoculated with dead culture never decreased (Fig. 1; averaging 3·0 ± 1·5 × 107 and 2·2 ± 1·0 × 10cells g−1 silage at the beginning and end of the experiments, respectively). It is important to note that controls run on silage samples with no-inoculated Myco. paratuberculosis had no background, showing that the qPCR assays were not detecting nonspecific background in the silage. No Salm. Typhimurium or E. coli was ever detected in silage inoculated with dead cultures.

Figure 1.

Percentage survival of Mycobacterium paratuberculosis, Escherichia coli and Salmonella Typhimurium in ensiled forage/alfalfa over time. Triplicate values from three different silage bags are shown for each organism per sample time. (a) E. coli uidA (▲) and (b) Salm. Typhimurium ttrBCA (■) genes and (c) live (●) or dead (○) Myco. paratuberculosis IS900 sequences were quantified by qPCR. Values over 100 per cent represent increases above initial mean concentrations (t = 0 for Salm. Typhimurium and t = 12 for Myco. paratuberculosis and E. coli). No data are shown for dead E. coli or Salm. Typhimurium because there were no detectable cells from the beginning of the experiment.

Decay rates and T90 values were calculated for each organism in silage (Table 4). It was not possible to calculate decay rates for Myco. paratuberculosis as concentrations never decreased during ensiling of the forages. Decay rates and T90 values for E. coli and Salm. Typhimurium show that E. coli DNA persisted for much longer in the silage than Salm. Typhimurium (Table 4).

Table 4. Decay rates (k day1) and time (days) to achieve 90% reduction in culturable cells or gene copies in silage and silage exudates
Sample Mycobacterium paratuberculosis Escherichia coli Salmonella Typhimurium
  1. a

    T90 = Time in days to achieve 90% reduction in cell numbers; for E. coli, an additional 7 days were added to account for the initial increase in cell numbers.

  2. b

    Decay rate (k) calculated by applying the best-fit slope to plot with x = time in days and y = ln(Nt/N0).

Silage exudates47·60−0·048425·20−0·12654·10−0·5601

Detection following exposure to silage exudates

To determine the response of each organism to acids from ensiling of forages without the competition from autochthonous microbial populations, exudates were produced from forage ensiled for 25 days. Reduced background in silage exudates made it possible to use PMA to pretreat samples so that only organisms with intact cell walls were detected by qPCR. The pH (4·8 ± 0·09) and EC (9·4 ± 0·4 ms) of the filtered exudates were similar to those of the unfiltered exudates [pH (4·7 ± 0·02) and EC (9·4 ± 0·4 ms)]. Despite the low pH, E. coli cell concentrations more than doubled between day 1 and day 7 in silage exudates before decreasing at a linear rate (R2 = 0·99; Fig. 2). Otherwise, decay rates for E. coli were nearly seven times higher in silage exudates than in silage (Table 4). As was the case in silage, Salm. Typhimurium concentrations steadily declined in silage exudates (R2 = 0·99; Fig. 2); however, the decay rates and T90 values were much more pronounced (over 18 times higher) than in silage (Table 4). Although Myco. paratuberculosis did exhibit significant die-off during exposure to silage exudates (< 0·01), concentrations were still 25% of initial by day 30 (Fig. 2) and the decay rates for Myco. paratuberculosis were 2·6 and 11·6 times lower than those for E. coli and Salm. Typhimurium, respectively (Table 4).

Figure 2.

Percentage survival of Mycobacterium paratuberculosis, Escherichia coli and Salmonella. Typhimurium in silage exudates obtained from 25-day ensiled forage/alfalfa. Triplicate values are shown for each organism per sample time. All samples were pretreated with PMA and quantified by qPCR. (a) E. coli uidA (▲), (b) Salm. Typhimurium ttrBCA (■) genes and (c) Myco. paratuberculosis IS900 (●) sequences were quantified by qPCR. Values over 100 per cent represent increases above initial mean concentrations.


Silage is a valuable source of nutrients for dairy and beef cattle; its importance is increasing with intensification of livestock production (Chen et al. 2005; McEniry et al. 2010; Duniére et al. 2011). In the ensiling process, anaerobic fermentation of forage materials by LAB results in the production of large amounts of OA which rapidly reduce pH, thereby eliminating most unwanted organisms and preserving the material (Chen et al. 2005; Stevenson et al. 2006; McEniry et al. 2010). The safety of this important livestock feed depends on the susceptibility of deleterious and pathogenic organisms to the low pH and OA produced during the ensiling process. This study was conducted to evaluate the effect of ensiling in forages and the low pH associated with the process on two important pathogens, Salm. Typhimurium and Myco. paratuberculosis, and the enteric organism E. coli. The response was measured by their survival in three conditions: (i) exposure to buffered weak OA (citric acid), (ii) exposure to exudates from ensiled forages, that is, materials free of autochthonous microbial populations but with the high salts and acid levels found in silage, and (iii) survival through the process of ensiling forage materials.

In these studies, three methods were used to detect targeted cells: plate counts, qPCR analysis and PMA-qPCR analysis. Comparison of plate counts and PMA-qPCR results showed that there were strong linear relationships between the two measures. Additionally, we found that PMA effectively blocked the amplification of DNA from dead cells of all three organisms in preliminary tests (data not shown) and was useful in the particulate-free, low-background samples from both the OA tests and those with silage exudates. However, when used to detect targeted organisms in silage materials, the PMA method did not work due to binding of the compound to the complex matrix and/or uptake by the more numerically dominant populations (i.e. Myco. paratuberculosis was <5% of the total microbial population). Because PMA could not be used to quantify bacteria in silage, our measurements as determined by qPCR are overestimates of cell concentrations that include DNA from viable and nonviable cells with intact DNA. This is commiserate with our greater survival rates than those found in studies in which only culturable cells were measured (which likely provides an underestimate of survival) (Katayama et al. 2000; Martínez-Gamba et al. 2001; Chen et al. 2005; Pedroso et al. 2010; Duniére et al. 2011). Our results support those of others who suggest the need for caution when interpreting the results from regular qPCR, given the detection of both viable and nonviable cells, and/or from detection by PMA-qPCR as issues may arise when the samples being tested are highly complex or have high initial cell densities (Wagner et al. 2005; Bae and Wuertz 2012) or when cell membrane integrity is not compromised despite treatment that render cells unculturable (Løvdal et al. 2011; Yang et al. 2011). However, the value of the PCR-based techniques cannot be underestimated for ecological studies such as this one where there exists a need for highly specific analyses targeting difficult to culture organisms and/or those existing within highly complex matrices.

Organic acids have antimicrobial activities due to the fact that they readily diffuse across the cell membrane and then dissociate inside the cell, thereby generating protons and the acid anion (Diez-Gonzalez and Russell 1997; King et al. 2010; Hosein et al. 2011). Cell death occurs due to (i) removal of protons at the expense of ATP generation, (ii) build-up of acid anions inside the cell disrupting cell metabolism and/or (iii) oxidative stress caused by the generation of free radicals (King et al. 2010; Hosein et al. 2011). Initial studies were conducted in buffered weak OA (citric acid) at different pHs so that the relative acid sensitivity of Myco. paratuberculosis, E. coli and Salm. Typhimurium could be determined. Citric acid is a weak OA and common natural food preservative. Although not found as a product of the ensiling process, the likely antimicrobial mechanism of this weak OA should be similar to many of those produced in silage.

Escherichia coli and Salm. Typhimurium were found to die off in a similar manner in citric acid buffered at the different pHs. Neither organism was culturable after 7 days of incubation at pH 4, but both were culturable for more than 10 days at all other pH values and had similar die-off rates. Myco. paratuberculosis, on the other hand, was culturable for longer than the other two organisms at pH 4·0, but otherwise died off comparatively rapidly. Studies by Chapman and Ross (2009) showed that the recovery of E. coli and Salm. Typhimurium from this kind of rapid acid shift can be increased in the presence of a slightly hypertonic solution (up to 3% NaCl). They suggest that this may be due to alterations in the lipid membrane of cells in a plasmolysed physiological state causing a reduced rate of acidification of the cytoplasm. On the other hand, Sung and Collins (2000, 2003) in two separate studies found that the acid resistance of Myco. paratuberculosis was dependent on the chemical composition of the media used to produce the cells, but was independent of NaCl concentration (2% to 4%). Similarly, the buffered citric acid used in this study enhanced survival of E. coli and Salm. Typhimurium, but did not increase survival of Myco. paratuberculosis.

Salmonella Typhimurium was the only organism to have a decay rate that was higher in silage exudates than in the buffered OA at similar pH values (i.e. pH 4·0 or 5·0); this suggests that the mechanisms involved in die-off in the two solutions were quite different. Studies of the effect of LAB on pathogen inhibition suggest that in addition to reducing pH by the production of lactic acid, the compound permeabilizes the outer membrane, increasing susceptibility to antimicrobial compounds and bacteriocins (Alakomi et al. 2000; Fayol-Messaoudi et al. 2005). This could explain why Salm. Typhimurium in particular died off more rapidly in silage exudates than in buffered OA solutions and why the organism had the most rapid decline in silage. Mechanisms needed to respond to rapid acid shifts in the presence of OA alone are likely less complex than those involved in attenuating the possibly additive effects of products from LAB that are found in silage and silage exudates (Fayol-Messaoudi et al. 2005; King et al. 2010). Higher decay rates for Salm. Typhimurium in silage exudates also suggest that the active compounds may be extracellular as the exudates were filter-sterilized and bacteria free.

The swine isolate of E. coli used in this study had similar rates of decay in silage exudates (−0·1285 day−1) as in buffered OA (−0·164 day−1). However, decay rates in silage were significantly lower (−0·0190 day−1). King et al. (2010) compared the transcriptomic profiles of a commensal (K-12 MG1655) and a pathogenic strain of E. coli (O157:H7 Sakai) in lactic and hydrochloric acid at pH 5·5. They concluded that while both E. coli strains exhibited similar responses to the acids, O157:H7 may have a greater ability to survive in more complex acidic environments, such as those encountered in the host and during food processing (King et al. 2010). While our E. coli strain was nontoxigenic (Cook et al. 2011), it was a livestock isolate and not a common laboratory strain, suggesting that in some instances there may be broader resistance to acids for strains adapted to transmission through diverse environmental matrices as has been shown for Salm. Typhimurium (Foster 2004; Oliveira et al. 2011). Additionally, we used only one strain of E. coli, Salm. Typhimurium and Mycoparatuberculosis. Although some aspects of acid resistance have been found to be strain independent, it is more likely that strain-level differences are important to survival in acidified environments as has been shown in other environments (Diez-Gonzalez and Russell 1997; Foster 2004; Bolster et al. 2010; King et al. 2010). Despite potential strain-level effects, the results from this study show that there were significant differences in E. coli survival in silage relative to both pathogens (significantly slower die-off than Salm. Typhimurium and significantly faster than Myco. paratuberculosis). This suggests that its behaviour in relationship to the organism being evaluated must be taken into consideration whenever E. coli is used an indicator of faecal contamination in feedstocks.

The response of Myco. paratuberculosis to acidification was much different than that of E. coli and Salm. Typhimurium. Decay rates for Myco. paratuberculosis exposed to silage exudates were almost seven times lower than when exposed to buffered OA. The other two organisms were more sensitive to the mix of OA in silage exudates that affect the integrity and function of the cell wall. Unlike the adaptive acid response exhibited by E. coli and Salm. Typhimurium, acid resistance in Mycobacteria has been shown to occur without the need for habituation (Piddington et al. 2000). Therefore, the Myco paratuberculosis cell wall may resist exposure to some acidic environments without the need for de novo synthesis of proteins, lipids and carbohydrates (Cotter and Hill 2003). Others have also found that products of LAB have variable effects on Myco. paratuberculosis populations (Gaggìa et al. 2010; Klanicova et al. 2012). On the other hand, Myco. paratuberculosis was more sensitive to direct intracellular acidification as likely occurs in the buffered OA. Similarly, Shleeva et al. (2011) found that rapid acidification of M. tuberculosis caused cell death, while gradual acidification of its environment resulted in the formation of dormant and resistant ovoid cells. Several studies have suggested that Myco. paratuberculosis is also capable of entering a dormant state (Whittington et al. 2004; Gumber et al. 2009).

Mycobacterium paratuberculosis showed no significant die-off in silage. It is important to note that dead Mycoparatuberculosis also showed no decrease in silage, while DNA from dead E. coli and Salm. Typhimurium was never detected. Persistence of DNA from dead and live Myco. paratuberculosis also occurred in a preliminary silage study (data not shown). Attempts were made to culture Myco. paratuberculosis in that study; however, the organism either lost culturability rapidly in the silage or was unculturable due to the inhibitory effects of the decontamination procedures used to eliminate background populations. Therefore, it is likely that Myco. paratuberculosis cell walls and genomic DNA remained intact throughout the ensiling process. Khol et al. (2010) also found that although no culturable Myco. paratuberculosis were detected in baled grass silage, samples were positive based on qPCR analysis. They suggest that Myco. paratuberculosis was likely inactivated quickly, but may have entered an inactive or dormant phase (Khol et al. 2010). However, in a recent study, Klanicova et al. (2012) showed that Myco. paratuberculosis remained culturable for up to six weeks in fermented milk products with a final pH slightly above 4·0 (i.e. 4·09–4·23). They also found that LAB and associated low pH (even at pH 4) had no effect on Myco. paratuberculosis DNA (Klanicova et al. 2012). These results show that Myco. paratuberculosis retains DNA within an intact cell wall to a greater extent than most bacteria and further suggest that culturability is a poor indicator and is generally disconnected from the actual persistence of Myco. paratuberculosis in environmental samples. Alternative explanations include a viable but nonculturable stage, spore formation or dormancy.

Results from these experiments show that E. coli, Salm. Typhimurium and Myco. paratuberculosis exhibit marked differences in response to the low pH and OA produced as a result of the ensiling process. Significant differences in E. coli response to acids and exudates as compared to the pathogens suggest that caution is necessary when using E. coli as an indicator of feed safety. Although culturability is an important measure of risk, current trends in livestock and foodborne disease incidence suggest that mechanisms of survival and transmission of important pathogens in the agricultural environment have not been completely explained by culturing organisms from these systems. The cell wall of Myco. paratuberculosis in particular was extremely resistant to the effects of low pH, OA and the microbial competition in silage. These results suggest that if present in manure and applied to forage grasses, a portion of the Mycoparatuberculosis population may survive exposure to low pHs and high OA during the ensilaging process and could therefore be a potential route of infection if ingested by a susceptible animal. Given the estimated U.S. dairy herd prevalence of 70% (Lombard et al. 2013) and the high concentrations of the organism present in manure, further research is needed to better understand persistence of this organism when applied to forage crops and to quantify risk to livestock if present in feed materials.


The authors wish to thank Rohan Parekh for technical assistance. This research was part of USDA-ARS National Program 214: Agricultural and Industrial By-products. Mention of a trademark or product anywhere in this paper is to describe experimental procedures and does not constitute a guarantee or warranty of the product by the USDA and does not imply its approval to the exclusion of other products or vendors that may also be suitable.