Fate of inoculated Escherichia coli in hay
Zwi G. Weinberg, Department of Food Science, The Volcani Center, Bet Dagan 50250, Israel. E-mail: email@example.com
Aims: To monitor the fate of inoculated Escherichia coli in dry and moist hay of various types, under laboratory conditions.
Methods and Results: Wheat, vetch and clover hay were used as received or wetted to 250–300 g kg−1 moisture. The hay was inoculated at about 106 CFU g−1 with a kanamycin-resistant E. coli strain that expresses the green fluorescence protein, and was stored in small open glass jars that were covered with aluminium foil. Three jars per treatment were sampled on days 1 and 3, or 4 and 7, or 8, 20 and 50, respectively, after the initiation of the experiments, and the numbers of E. coli in the hay were determined. The results indicated that E. coli disappeared from both dry and moist hay by 7–8 days after inoculation. However, in a few cases colonies that were presumed to be E. coli developed after incubation in Luria broth medium.
Conclusions: The tagged E. coli strain usually disappeared rapidly from both the dry and the moist hay, in spite of the high level of inocula used. However, in some cases a few, possibly injured E. coli might have persisted, and could be detected after incubation in a rich growth medium.
Significance and Impact of the Study: This study is part of a risk assessment associated with sewage irrigation of forage crops in Israel. The results indicate that E. coli added to the hay is not likely to pose a health risk to cattle or to humans. Nevertheless, more research with natural strains of E. coli and other enteric pathogens that might be more adapted to forage conditions is warranted in order to ensure the safety of sewage-irrigated crops.
In modern farming, forage crops are harvested during the optimal season and are preserved for the rest of the year. The major preservation methods are ensiling and hay making. Ensiling is based on anaerobic lactic acid fermentation, which results in pH decline; hay making is based on field drying of the harvested biomass to reduce the water activity levels below that required to support microbial growth.
In the processed food industry the principles of Good Management Practices and Good Agricultural Practices are applied to obtain wholesome and safe food products. Many factors might affect the quality and safety of forage crops and of their preserved forms – silage and hay, therefore, similar practices to those in the food industry should be considered for forage crops and feedstuffs, which form the first step in the human food-production chain (Lindgren 1999).
Among the factors considered hazardous in forage crops and in their preserved products are pathogenic enterobacteria such as Salmonella and toxin-producing Escherichia coli (Wilkinson 1999). Pathogenic strains of E. coli can cause severe illness in humans and animals, and the toxigenic strain E. coli O157:H7 is of special concern. If conditions in the forage are favourable for its growth E. coli may cause intestinal disorders and mastitis in animals that consume that forage (Sheinbaum and Tromp 1982; Lindgren 1991). Cattle are a primary source of pathogenic E. coli O157:H7 (Russell et al. 2000), and this organism can be transmitted to crops and, hence to products via shedding or through fertilization of fields with manure (Lindgren 1999).
In silage, a slow decline in silage pH favours the growth of enterobacteria, whereas rapid fermentation hastens their elimination (Bach et al. 2002). Bach et al. (2002) demonstrated that E. coli O157:H7 was eliminated from barley silage during a rapid fermentation, and similar results were obtained in grass silage that was inoculated with E. coli O157:H7 (Heron et al. 1993; Byrne et al. 2002). However, Brudzinski and Harrison (1998) reported that E. coli O157:H7 might develop acid resistance through induction of an acid-tolerance response, and so might be able to survive at a pH as low as 3·4. In addition, it is known that E. coli O157:H7 can survive in the human stomach and other extremely acidic environments (Arnold and Kaspar 1995; Buchanan and Edelson 1999).
In Israel, the hygiene of forage crops and their products (hay and silage) has become an important issue because of intensive use of recycled water for irrigation. Weinberg et al. (2004) showed that fresh forage crops that are irrigated with secondary-treated sewage water, and well-preserved silages contain few or no E. coli. However, this species was detected in substantial numbers in spoiled parts of silage and in feedstuffs in which the pH was above 5·0. In a previous study (Chen et al. 2005), a recombinant E. coli strain carrying a plasmid with an antibiotic resistance marker, and expressing the green fluorescence protein (GFP) was used in experiments with wheat and corn silages. The results revealed that this strain persisted longer as the pH decline in silages of wilted wheat proceeded more slowly, but eventually it completely disappeared.
Hay making is an alternative means of forage preservation that also involves drying of forage crops. However, hay may be subjected to rain damage and spoilage, and in such wet hay coliforms may survive and grow. In Israel, wheat for forage, as well as some forage legumes, are harvested in April – a time when there is still a chance of late rain which may disturb the hay making and damage the hay. Rain may also damage stored hay during the subsequent rainy season. As many forage crops are irrigated with secondary-treated sewage water hay may be contaminated by pathogenic micro-organisms (Zuckerman 1997), and wet spots that remain in the hay may potentially favour their survival.
Because forage crops may contain pathogens, especially crops irrigated with recycled sewage water, this study was undertaken in order to follow the fate of inoculated E. coli in hay. The purpose of the present experiments was to study the fate of added E. coli in both dry and wetted hay, under laboratory conditions.
Materials and methods
The E. coli strain used in this study was ATCC 25922 containing Plasmid pWMJ1029, which expresses both a kanamycin resistance gene and the GFP gene. Escherichia coli was grown overnight at 37°C in tryptic soy broth containing kanamycin at 50 μg ml−1. The content of CFU was determined by plating serial dilutions of the bacteria on tryptic soy agar (TSA) (Becton, Dickinson and Co., Sparks, MD, USA) plates supplemented with kanamycin (Sigma, St Louis, MO, USA) at 50 μg ml−1. When the bacteria are grown on TSA, GFP fluorescence is easily observed under long-wavelength UV light.
Four separate experiments were performed with the following hay: (i) wheat (I); (ii) vetch; (iii) clover; and (iv) wheat (II) hay. The wheat (cultivar ‘Galil’) was harvested at the early milk stage of maturation. The vetch (cultivar ‘Hayovel’) and the white clover (cultivar ‘Faly’) were harvested at 20% bloom. The various hays were brought to the laboratory from a commercial feed centre and were chopped into 2–3 cm pieces with a Wintersteiger® chopper (Ried, Austria). As hay may be exposed to rain damage or may undergo insufficient drying which may result in moist spots with relatively high water activity, treatments included addition of water to the hay. The following treatments were used: dry hay (control), wetted hay (control), and dry and wetted hay inoculated with a recombinant E. coli strain at the rates given below. The wetted hay was obtained by spraying and thoroughly mixing distilled water at 160, 220, 300 and 300 ml kg−1 into wheat (I), vetch, clover and wheat (II) hay respectively. The hay of the various treatments was deposited into 0·5 l (experiments 1 and 2) or 0·25 l (experiments 3 and 4) jars, which served as experimental units. The weight of the dry and wetted hay placed in each jar was 50 and 75 g in experiments 1 and 2, and 25 and 35 g in experiments 3 and 4 respectively. The jars were covered with aluminium foil which was punctured with four 5-mm diameter holes, and they were stored at room temperature (25°C) (experiments 1 and 2) or in an incubator at 30°C (experiments 3 and 4) with no direct sunlight. There were 15 jars per treatment three of which were sampled for enumeration of E. coli and determination of pH, on days 1 and 3, or 4 and 7, or 8, 20 and 50 respectively. The 50-day sampling was applied to experiments 1, 2 and 4 only. The original hay was analysed for dry matter, pH, ash and numbers of E. coli, yeasts and moulds. The relative humidity at equilibrium of the wheat (I) and vetch hay samples was determined with a Novasina MS1 Hygro Measuring System (Defensor®, Pfaffikon, Switzerland), and that of the clover and wheat (II) samples with a hydrolog-D (Rotronic®, Bassersdorf, Switzerland).
Inoculation of the hay
Wheat (I) hay
To inoculate dry hay with E. coli, 1 kg of dry hay was sprayed with 4 ml of the E. coli suspension diluted with 16 ml of water, and was mixed thoroughly by hand. An additional treatment was included in which 200 μl of the E. coli suspension diluted with 800 μl of distilled water were applied to 50 g of dry hay for each jar individually (referred to as ‘individ.’, Table 2).
Table 2. Escherichia coli numbers in the inoculated wheat (I) hay (experiment 1) (mean ± SD)
|Dry||0||6·2||4·7|| || |
|1||6·2 ± 0·0||3·2 ± 0·5|| || |
|4||6·4 ± 0·2||1·9 ± 0·5|| || |
|8||6·5 ± 0·3||<1·0||−−−|| |
|20||6·2 ± 0·1||<1·0||−−−|| |
|50||6·6 ± 0·2||<1·0||−−−|| |
|Moist||0||6·2||5·57|| || |
|1||6·2 ± 0·0||3·7 ± 0·2|| || |
|4||6·2 ± 0·0||1·9 ± 0·5||+++||+++|
|8||6·4 ± 0·1||<1·0||+++||+++|
|20||6·3 ± 0·1||<1·0||−−−|| |
|50||6·2 ± 0·2||<1·0||−−−|| |
|Individ.||0||6·0||5·11|| || |
|1||6·0 ± 0·1||3·1 ± 0·6|| || |
|4||6·3 ± 0·1||1·3 ± 1·3|| || |
|8||6·3 ± 0·4||<1·0||+++||+++|
|20||6·3 ± 0·2||<1·0||+/−−||+|
|50||6·5 ± 0·1||<1·0||−−−|| |
To inoculate moist hay with E. coli, 4 ml of E. coli suspension diluted with 156 ml of distilled water was sprayed onto 1 kg of dry hay.
The original E. coli suspension contained 2·5 ×108 CFU ml−1, therefore, the theoretical application rate for all inoculation treatments was 1 × 106 CFU g−1.
To inoculate dry hay with E. coli, 500 μl of the E. coli suspension diluted with 500 μl of distilled water was added to 50 g of dry hay in each jar.
To inoculate moist hay with E. coli, 1 kg dry hay was sprayed with 10 ml of E. coli suspension diluted with 210 ml of distilled water.
The original E. coli suspension contained 2·5 ×108 CFU ml−1, therefore, the theoretical application rate for all inoculation treatments was 2·5 × 106 CFU g−1.
Clover and wheat (II) hay
To inoculate dry hay with E. coli, 500 μl of the E. coli suspension diluted with 500 μl of distilled water was added to 50 g of dry hay in each jar.
To inoculate moist hay with E. coli, 1 kg of dry hay was sprayed and mixed thoroughly with 300 ml of distilled water, and 500 μl of the E. coli suspension diluted with 500 μl of distilled water was added to 50 g of the resulting moist hay in each jar.
In the experiments with clover and wheat (II), the original E. coli suspensions contained 9·0 × 109 and 1 × 109 CFU ml−1, respectively, therefore, the theoretical application rates for all inoculation treatments were 0·9 × 108 and 1 × 107 CFU g−1 respectively.
Enumeration of E. coli in hay samples
The original hay or the samples stored during the experiment were extracted with saline–peptone solution in sterile bags in a Stomacher blender (Seward Medical, London, UK) for 3 min. Escherichia coli population in the control hay was determined by plating serial dilutions of the extracts in Chromocult TBX® agar (Merck, Darmstadt, Germany), with and without kanamycin, by using the pour-plate double-layer technique. For the inoculated hay only kanamycin-supplemented plates were used. The plates were incubated for 2–3 h at 30°C to enable resuscitation of injured bacteria, and then for 24 h at 42°C. Blue-green colonies were presumptively identified as E. coli. To confirm that the kanamycin-resistant colonies carried plasmid PWM1029, representative colonies that developed on the Chromocult TBX agar to which antibiotic had been added were transferred with sterile toothpicks to kanamycin-supplemented TSA plates for fluorescence determination. Colonies that appeared fluorescent under a UV lamp were considered to derive from the original E. coli inoculant.
To test for the presence of injured E. coli that could not grow in the selective medium an enrichment step was added: 1-ml of samples of extract were also inoculated into 1 ml of double-strength Luria broth (LB, Hy-Labs, Israel) containing kanamycin and were incubated overnight at 30°C. To test if the original (tagged) E. coli strain was present in the enrichment broth, 1 ml of LB culture was plated in kanamycin-supplemented Chromocult TBX agar and incubated for another 24 h at 42°C. Blue-green colonies were presumptively identified as E. coli. Representative blue-green colonies were transferred to TSA plates for determination of possible GFP fluorescence. Indole test sticks (Hy Laboratories Ltd, Rehovot, Israel) were used to confirm the identification of E. coli.
Each experiment was analysed as a 2 × 2 factorial (two levels of moisture and inoculation); wheat (I) was analysed as a 2 × 3 factorial (two levels of moisture and three inoculation treatments). The statistical analysis was applied to the log10 numbers of E. coli in the media with kanamycin, using the GLM procedure of SAS (Statistical Analysis System, Cary, NC, USA)
Moisture content was determined by oven drying at 105°C for 48 h. The pH was measured in water extracts 1 : 9, using a pH-meter (Mettler Toledo, Schwerzbach, Switzerland). Ash was determined after burning at 600°C for 3 h. Yeast and mould numbers were determined in all the original hay samples (day 0) and in experiments 1 and 2 also in the final samples (day 50). This was performed in the saline extracts by means of 10-fold serial dilutions on spread plate malt extract agar acidified to pH 4·0 with 10% lactic acid. Mould were differentiated from yeasts by their fuzzy mycelium.
Table 1 displays the results of the analysis of the hay that was used in the current experiments. Addition of water to the hay at 300 ml kg−1 increased its moisture content and its equilibrium relative humidity from around 0·5 to 0·874–0·91.
Table 1. Data of the hay before inoculation
|Dry wheat (I)||121 ± 1||75||6·6||44·4||3·5||4·1|
|Moist wheat (I)||252 ± 10||–||6·2||91·0||–||–|
|Dry vetch||103 ± 3||106||5·8||61·4||2·3||4·5|
|Moist vetch||257 ± 2||–||5·9||92·6||–||–|
|Dry clover||102 ± 3||128||5·9||51·5||5·2||4·8|
|Moist clover||324 ± 14||–||6·1||88·8||–||–|
|Dry wheat (II)||108 ± 3||81||6·1||51·3||2·4||5·0|
|Moist wheat (II)||331 ± 20||–||6·0||87·4||–||–|
Table 2 gives the mean values of the log10 numbers of E. coli in the inoculated wheat (I) hay (experiment 1). The pH values of the hay did not change much in any of the treatments during the storage period. No E. coli were detected in any of the uninoculated hay samples throughout the storage period. In the inoculated dry hay the recovery of the tagged E. coli immediately after their application was about 10 times less than expected from the theoretical application rate; in the moist hay it was close to the expected number. Escherichia coli numbers decreased during the first 4 days post inoculation, and on day 20 no E. coli could be detected or even recovered from the LB broth. No E. coli were detected or recovered afterwards. A similar time-course profile was obtained for the vetch hay (experiment 2, Table 3). In the dry vetch hay some E. coli were detected on day 8. Following incubation in LB broth, on days 20 and 50, more E. coli-positive samples were recovered from the dry vetch hay samples than from the moist ones, probably because the latter harboured other microbial populations, such as yeasts and moulds with which the E. coli had to compete. In these experiments, the percentage of fluorescent colonies was 25–50%, and this low rate is probably due to the instability of the plasmid-borne GFP.
Table 3. Escherichia coli numbers in the inoculated vetch hay (experiment 2) (mean ± SD)
|Dry||0||6·1||6·0|| || |
|1||6·1 ± 0·1||3·9 ± 0·15|| || |
|4||6·1 ± 0·0||2·9 ± 0·6|| || |
|8||6·0 ± 0·0||1·7 ± 0·4|| || |
|20||6·1 ± 0·1||<1·0||+++||+++|
|50||6·0 ± 0·0||<1·0||+++||+++|
|Moist||0||6·3||6·2|| || |
|1||6·2 ± 0·0||3·24 ± 0·9|| || |
|4||6·1 ± 0·0||0·7 ± 0·3|| || |
|8||6·3 ± 0·1||<1·0||+++||−−−|
|20||7·6 ± 1·6||<1·0||++||++*|
|50||7·1 ± 1·3||<1·0||+−−||−−−**|
Table 4 gives the results of the experiment with clover hay (experiment 3). In this experiment the wetting of the hay resulted in an increase in pH during storage. Mould developed in the moist hay samples from day 7 onwards, and this interfered with the enumeration and identification of the presumptive E. coli colonies. The recovery rates of the tagged E. coli from the inoculated dry and moist hay immediately after inoculation were about 1000 and 100 times less, respectively, than expected from the theoretical application rate. In this experiment the control moist hay contained small numbers of suspected E. coli that grew on Chromocult TBX medium without kanamycin; the control dry hay contained very small numbers of E. coli, and they were detected only in the samples taken on day 3. All the colonies that developed from the inoculated samples on kanamycin-supplemented medium exhibited fluorescence.
Table 4. Escherichia coli numbers in the clover hay (experiment 3) (mean ± SD)
|Control dry||0||5·9||<1·0||<1·0||−−−|| |
|3||5·8||1·7 ± 0·8||<1·0|| || |
|Control moist||0||6·1||3·5||<1·0|| || |
|1||6·0||1·8 ± 0·3||<1·0|| ||+/−−|
|3||7·6||3·6 ± 0·9||<1·0|| ||−−−|
|7||8·5||2·2 ± 1·9||<1·0||−−−||−−−|
|Inoculated dry||0|| ||ND||5·9|| ||+|
|1||6·1||ND||2·9 ± 0·5|| ||+++|
|3||6·2||ND||1·1 ± 0·2|| ||−−−|
|Inoculated moist||0|| ||ND||6·8|| ||+|
|1||6·2||ND||4·0 ± 0·4|| ||−−−|
|3||8·0||ND||1·4 ± 1·6|| ||?|
A similar pattern was observed in the wheat (II) hay (experiment 4, Table 5). In the control moist hay, suspected colonies were recovered from the enrichment broth on day 50, but they were indole negative. In this experiment, E. coli disappeared from the inoculated dry and moist samples after 1 and 7 days respectively. In this hay, addition of water resulted in a pH decrease, probably because of fungal activity. Mould also appeared in the moist wheat (II) hay samples on day 7 after initiation of the experiment. In this experiment, fluorescence was exhibited by 100% of the presumptive E. coli colonies that developed from inoculated samples in the kanamycin-supplemented medium.
Table 5. Escherichia coli numbers in the wheat (II) hay (experiment 4) (mean ± SD)
|Control dry||0||6·1||<1·0||<1·0||−|| |
|Control moist||0||6·0||<1·0||<1·0||−|| |
|1||6·1||0·2 ± 0·4||<1·0||−−−|| |
|Inoculated dry||0||6·0||ND||5·6|| ||+|
|1||5·8||ND||2·7 ± 0·1|| ||+++|
|Inoculated moist||0||5·9||ND||6·3|| ||+|
|1||6·0||ND||3·9 ± 0·1|| ||+|
|3||4·4||ND||1·2 ± 0·2|| ||−|
|7||5·4||ND||1·7 ± 0·1|| || |
The results of the statistical analysis revealed that for all four experiments inoculation resulted in significantly higher numbers of E. coli than in uninoculated hays. The effects of the moisture treatments or the interactions (moisture × inoculation) with regard to E. coli numbers were non-significant. In the wheat (I) experiment, the individual application of the bacteria did not differ significantly (P < 0·05) from the group application.
The safety and hygiene of forage crops are of utmost importance as these crops constitute the first link in the food production chain and they are liable to affect both farm animals and humans. This study was initiated in Israel, where many fields are increasingly irrigated with secondary-treated sewage water which may be a potential source of pathogenic micro-organisms and toxic chemicals (Zuckerman 1997). Escherichia coli serves as an indicator for faecal contamination and therefore was chosen as the subject of our present study. Previous studies (Weinberg et al. 2004; Chen et al. 2005) monitored the presence and survival of E. coli in fresh sewage-irrigated forage wheat and corn and their silages. Silage preservation is based on lactic acid fermentation under anaerobic conditions, and previous studies showed that E. coli cannot survive for a long time in well-managed silages in which the pH decreases rapidly; these findings were similar to those of studies with barley and grass silages (Bach et al. 2002; Byrne et al. 2002). However, E. coli were found in substantial numbers in spoiled silage samples (Weinberg et al. 2004).
The present study monitored the fate of E. coli that were inoculated into hay. Hay making is based on field drying of the crops to a moisture level that does not permit microbial growth. The minimal water activity required for the growth of E. coli is 0·95 (Frazier and Westhoff 1978). However, hay may be exposed to rain damage or may undergo insufficient drying. Either of which may result in moist spots with relatively high water activity. Therefore, treatments in which hay was wetted with water at 250–300 ml kg−1 were also included in our experiments. Such water addition supported some growth of E. coli in the control clover hay (experiment 3) which could have harboured dormant epiphytic forms of this bacteria. Increasing the amount of added water did not result in a linear increase in the equilibrium relative humidity in the hay of various crops. This may be due to intrinsic properties of the plant particles, such as cell wall constituents, protein and mineral contents that could affect water activity. In the moist clover and wheat (II) hays, to which 300 ml water per kg was added, moulds developed during storage. The change in pH (increase in experiment 3, decrease in experiment 4) observed during storage of the moist hay in these experiments could have resulted from their activity. Different types of wetted hay varied in their support of fungal development according to their epiphytic microbial populations and according to their water activity and composition. In addition to spoiling of the hay, these moulds interfered with the detection and enumeration of the E. coli colonies, because their mycelium covered the E. coli enumeration medium. In order to avoid such problem, in future studies we will consider to add a fungal inhibitor such as potassium sorbate to the E. coli selective enumeration medium.
For the dry wheat (I) hay (Table 2), there was no significant difference (P < 0·05) between batch application of the inoculant and individual application to every jar. However, in the subsequent experiments we preferred to apply the inoculant to each jar separately in order to ensure its presence in every inoculated jar. The inoculation rate in the present experiments was high (106–108 CFU g−1), so as to simulate a severe case of contamination and to enable log-reduction measurements of bacterial counts.
The results of all four experiments indicate that E. coli disappeared from all hay samples within 7 days after inoculation or addition of water. However, in some cases, such as in the inoculated vetch dry hay or the control moist wheat (II) hay presumptive E. coli colonies developed after enrichment. These results indicate that small numbers of E. coli cells remained viable during the hay-making process and resumed growth when favourable conditions returned. The difference between the various types of hay in terms of survivability of E. coli might be due to differences in composition between them, for example, higher protein content in leguminous hays.
In the present study, a kanamycin-resistant E. coli strain that expresses GFP was used as a marker for survival of faecal coliforms in hay. This strain has also been used for studying the fate of added E. coli in silage (Chen et al. 2005). The loss of fluorescence in some of the experiments may indicate the loss of the plasmid encoding for GFP. However, as the control hay samples contained very small numbers of epiphytic E. coli, the use of the tagged E. coli was efficient in monitoring the survival of this species in hay.
Further survival studies of other E. coli strains that might be better adapted to forages are planned in our laboratory.
The tagged E. coli strain rapidly declined in numbers in both dry and moist hay, in spite of the high inoculation rate. However, in some cases presumptive E. coli colonies developed after growth in LB broth, which indicates that some bacteria are capable of surviving the hay-making process, and then developing in the hay when conditions become favourable.
This work was contributed from the Agricultural Research Organization, the Volcani Center, Bet Dagan, Israel, no. 460/06-E series. This work was supported by a grant from the Chief Scientist of the Israeli Ministry of Agriculture and Rural Development.