Thermal Inactivation of Desiccation-Adapted Salmonella spp. in Aged Chicken Litter

Abstract

Thermal inactivation of desiccation-adapted Salmonella spp. in aged chicken litter was investigated in comparison with that in a nonadapted control to examine potential cross-tolerance of desiccation-adapted cells to heat treatment. A mixture of four Salmonella serovars was inoculated into the finished compost with 20, 30, 40, and 50% moisture contents for a 24-h desiccation adaptation. Afterwards, the compost with desiccation-adapted cells was inoculated into the aged chicken litter with the same moisture content for heat treatments at 70, 75, 80, 85, and 150°C. Recovery media were used to allow heat-injured cells to resuscitate. A 5-log reduction in the number of the desiccation-adapted cells in aged chicken litter with a 20% moisture content required >6, >6, ∼4 to 5, and ∼3 to 4 h of exposure at 70, 75, 80, and 85°C, respectively. As a comparison, a 5-log reduction in the number of nonadapted control cells in the same chicken litter was achieved within ∼1.5 to 2, ∼1 to 1.5, ∼0.5 to 1, and <0.5 h at 70, 75, 80, and 85°C, respectively. The exposure time required to obtain a 5-log reduction in the number of desiccation-adapted cells gradually became shorter as temperature and moisture content were increased. At 150°C, desiccation-adapted Salmonella cells survived for 50 min in chicken litter with a 20% moisture content, whereas control cells were detectable by enrichment for only 10 min. Our results demonstrated that the thermal resistance of Salmonella in aged chicken litter was increased significantly when the cells were adapted to desiccation. This study also validated the effectiveness of thermal processing being used for producing chicken litter free of Salmonella contamination.

INTRODUCTION

Chicken litter is a waste by-product of poultry production and is comprised of feces, wasted feeds, bedding materials, and feathers (1). More than 14 million tons of chicken litter is produced annually in the United States (2). Chicken litter is usually recycled as an organic fertilizer or soil amendment for direct application to agricultural land (3). However, chicken litter may contain loads of human pathogens, such as Salmonella spp., that have great potential to directly or indirectly contaminate fresh produce and cause food-borne disease outbreaks (1). Currently, high-temperature processing is the most commonly applied method to reduce or eliminate potential pathogens in chicken litter (1, 4).

Some microorganisms become acclimatized to desiccation stress in a dry environment, and induction of the desiccation stress response in bacterial cells makes them more resistant to the dry condition in which they are present (5). Most importantly, exposure to a single stress is found to be associated with the development of cross-tolerance to multiple unrelated stresses (6). Using laboratory models, various researchers have demonstrated that the desiccated cells exhibit increased thermal resistance (6–8). Previous thermal-inactivation studies on bacterial pathogens in chicken litter have used only nonstressed cells (1, 4). Therefore, to simulate real-world conditions, thermal inactivation of desiccation-adapted cells should be evaluated, as they are present in chicken litter during stockpiling.

A population of pathogens subjected to sublethal heat treatment undergoes metabolic injury (9). Nonetheless, these sublethally injured cells present the same threat to food safety as their noninjured counterparts, because they can be resuscitated and their pathogenicity restored under suitable conditions (10). Hence, the significance of injured pathogens should not be ignored. In a nonselective medium, the injured cells are usually repaired and then become functionally normal; however, this medium cannot differentiate a target pathogen from a mixed population. On the other hand, the injured cells may fail to resuscitate when plated directly onto media with selective compounds (11). A repair step on nonselective media is therefore necessary before exposure to a selective medium for enumeration. Several studies have revealed that the presence of exogenous pyruvate in recovery media enhanced the repair of heat-stressed pathogens (12–15). Busch and Donnelly (14) reported that tryptic soy broth supplemented with 1% sodium pyruvate facilitated the repair of heat-injured Listeria monocytogenes cells. As heat-injured Salmonella cells in chicken litter are potentially pathogenic, appropriate recovery methods should be incorporated into enumeration procedures. In the current study, different recovery procedures were compared and employed to allow the resuscitation of heat-injured cells and to ensure that an accurate microbiological analysis be obtained.

As far as we are aware, no reports studying the thermal inactivation of desiccation-adapted pathogens in compost and manure are available. We herein hypothesize that desiccation stress during storage under stockpiling conditions may increase the cross-tolerance of Salmonella cells in chicken litter to subsequent exposure to high temperatures. The objective of this study was thus to investigate the potential cross-tolerance of desiccation-adapted Salmonella spp. in aged chicken litter to heat treatment.

MATERIALS AND METHODS

Sample preparation.

Fresh chicken litter was collected from a chicken barn for Bovan laying hens raised at Morgan Poultry Center, Clemson, SC, whereas the aged chicken litter was sourced from Cobb broiler chickens (Organic Farms, Livingston, CA). To prepare the aged chicken litter, the litter inside the chicken house was removed annually, followed by a partial windrow composting of ∼7 to 10 days. After composting, the litter was screened to remove rice hulls and readied for subsequent heat treatment. Commercially available dairy compost and poultry compost (both from Black Gold Compost Co., Oxford, FL) were purchased from a local supermarket. All the compost samples were dried under a fume hood until the moisture contents were reduced to less than 20% and then screened to a particle size of less than 3 mm using a sieve. Sufficient samples were collected for the entire experiment and stored in a sealed container at 4°C until use.

Bacterial strains.

Salmonella enterica serovars Enteritidis H2292 and Heidelberg 21380 (provided by Michael Doyle, University of Georgia, Griffin, GA), Senftenberg ATCC 43845, and Typhimurium 8243 [genotype, thyA deo polA2 zie-3024::Tn10(dTc)zag-1256::Tn10(dKm) [Tn10(dTc) and Tn10(dKm) are derivative, defective transposons with tetracycline and kanamycin resistance genes, respectively, as selectable markers for transduction], derived from S. Typhimurium LT2 by Russell Maurer, Case Western Reserve University, Cleveland, OH, and provided by Roy Curtiss III, Washington University, St. Louis, MO)] were used for the thermal-inactivation study. S. Typhimurium 8243 was used for the optimization of recovery media for heat-injured cells and selection of compost for desiccation adaptation. All the strains were induced to rifampin resistance (100 μg of ml⁻¹) using the gradient plate method (15).

Inoculum preparation.

Each Salmonella strain was grown overnight at 37°C in tryptic soy broth containing 100 μg of rifampin ml⁻¹ (TSB-R). The overnight cultures were washed three times with sterile 0.85% saline, and the final pelleted cells were resuspended in 0.85% saline to desired cell concentrations by measuring the optical density at 600 nm.

Optimization of recovery media for heat-injured Salmonella.

Fresh chicken litter was used for the optimization of recovery media for heat-injured Salmonella. The moisture content of chicken litter was adjusted to 30% (water activity [a_w], 0.932) with sterile tap water. The overnight culture of Salmonella was washed and resuspended in 0.85% saline to ca. 10⁹ CFU ml⁻¹ (an optical density at 600 nm of ca. 0.6 to 0.7). Chicken litter was inoculated with Salmonella cells (1:100, vol/wt) using a sterile spray nozzle and thoroughly mixed to a final concentration of ca. 10⁷ CFU g⁻¹.

Samples (about 20 g) were distributed evenly inside an aluminum pan (internal diameter, 10 cm), placed in three different locations (close to the door, center, and far away from the door) on the shelf of a controlled-convection oven (Binder Inc., Bohemia, NY), and then exposed to 75°C for up to 1 h. Temperature was monitored constantly using T-type thermocouples (DCC Corporation, Pennsauken, NJ), with one cord kept inside the oven chamber and others inserted into the bottom of litter samples of three different locations. The temperature was initially set at a higher set point of 80°C to minimize the time to the target temperature. When the interior of the litter reached the desired temperature, the temperature setting of the oven was readjusted to maintain it at the designated experimental temperature. Samples were taken out at 0.5 and 1 h and placed immediately in an ice-water bath to cool down the samples rapidly and minimize further cell death. Samples were then diluted serially with 0.85% saline and transferred in triplicate to different media to evaluate the recovery efficiencies with these media. Samples taken at the beginning of heat treatment (0 h) were used to determine the initial populations. These experiments were performed in three separate trials.

Tryptic soy agar (TSA) was used as a nonselective medium, while TSA with 100 μg of rifampin ml⁻¹ (TSA-R) and xylose lysine Tergitol-4 agar with 100 μg rifampin ml⁻¹ (XLT-4-R) were used as selective media.

The following media and methods were used for the recovery of heat-injured Salmonella cells.

1. TSA supplemented with 100 μg rifampin ml⁻¹.

2. XLT-4 supplemented with 100 μg rifampin ml⁻¹.

3. A modified two-step overlay (OV) method (OV/TSA-R and OV/XLT-4-R) (11). Heat-injured cells were plated directly onto TSA. After incubation at 37°C for 3 h to allow recovery of injured cells, 7 ml of TSA-R or XLT-4-R was laid over the TSA. Plates were incubated at 37°C for another 21 h, and then colonies were counted.

4. A modified thin agar layer (TAL) method (TAL/TSA-R and TAL/XLT-4-R) (11). After solidification of 25 ml TSA-R or XLT-4-R in the plate, 14 ml of melted TSA (48°C) was laid over the plates. Heat-injured cells were plated onto TAL media, which were then incubated at 37°C for 24 h.

5. TSA supplemented with 100 μg rifampin ml⁻¹ and 1% sodium pyruvate (P/TSA-R).

6. TSB supplemented with 100 μg rifampin ml⁻¹ and 1% sodium pyruvate (P/TSB-R). After heat treatment, 1 ml of heat-injured cells was transferred into 9 ml of P/TSB-R, and the plate was incubated at 37°C for 3 h, followed by plating onto TSA-R and incubation for another 21 h.

Selection of compost matrix for desiccation adaptation of Salmonella.

To select the compost matrix for desiccation adaptation of Salmonella, changes in bacterial populations before and after desiccation adaptation in dairy compost, fresh poultry compost, old poultry compost, and aged chicken litter were compared. The overnight-grown S. Typhimurium inoculum was washed and resuspended in 0.85% saline to ca. 10⁹ CFU ml⁻¹. To prepare the inoculum for desiccation adaptation experiments, this resuspended culture was further concentrated 100 times (ca. 10¹¹ CFU ml⁻¹) by centrifugation. Afterwards, the culture was added separately (1:100, vol/wt) into 300 g of the four different composts described above with a moisture content of 30% (a_w values of dairy compost, fresh poultry compost, old poultry compost, and aged chicken litter were 0.980, 0.916, 0.938, and 0.943, respectively) at a final concentration of ca. 10⁹ CFU g⁻¹ for a 24-h desiccation adaptation. Before and after desiccation adaptation, samples were homogenized and serial dilutions of homogenates were plated in duplicate onto XLT-4-R for enumeration. Two trials were conducted for each experiment.

Thermal inactivation.

Compost selected for desiccation adaptation and aged chicken litter samples were all adjusted to the desired moisture contents, namely, 20% (a_w, 0.873), 30% (a_w, 0.943), 40% (a_w, 0.975), and 50% (a_w, 0.986), with sterile tap water. The four Salmonella serotypes were grown separately overnight at 37°C, washed, and then mixed in equal volumes as inocula. Both desiccation-adaptation and inoculation were performed as described above. The thermal-inactivation study was carried out as described in Fig. 1. The temperatures used for this study were 70, 75, 80, 85, and 150°C. For heat treatments at the above-named temperatures, for up to 6 h at predetermined time intervals, duplicate samples were taken out, homogenized, and plated on recovery media. The samples that were negative by the direct plating recovery method (detection limit, 1.30 log CFU g⁻¹) were preenriched in universal preenrichment broth (UPB), followed by a secondary enrichment in Rappaport-Vassiliadis (RV) broth supplemented with 100 μg rifampin ml⁻¹. After 24 h of incubation at 42°C, enriched samples were then plated onto XLT-4-R. Presumptively positive colonies on XLT-4-R were further confirmed as Salmonella using the immunolatex agglutination test (Oxoid Ltd., Basingstoke, Hampshire, United Kingdom). For heat treatment at 150°C, duplicate samples were withdrawn every 10 min for up to 60 min and enriched in UPB directly to test if the Salmonella was alive. To serve as controls, washed Salmonella cultures (ca. 10⁹ CFU ml⁻¹) kept at room temperature for 24 h were aseptically added to aged chicken litter (20% moisture content) at a ratio of 1:100 (vol/wt) and exposed to the above-named temperatures, as the desiccation-adapted cultures were.

Fig 1.

Flow chart of the experimental procedure.

PFGE.

Bacterial colonies with the longest survival (defined as the last sampling time at which Salmonella could be detected by direct plating) were randomly selected from the TAL recovery media in chicken litter with the 20% moisture content after exposure to the heat treatment at 80°C. The selected colonies (n = 12) for each sample were transferred onto TSA two times and then characterized by pulsed-field gel electrophoresis (PFGE) as described by CDC/PulseNet (16). A control was prepared as described above (“Thermal inactivation”). The band patterns of these isolates were compared with the genetic profiles of four serotypes used in this study.

Moisture content, pH, and ammonia.

Moisture content was determined with a moisture analyzer (model IR-35; Denver Instrument, Denver, CO). The pH value and ammonia content in compost were measured according to the methods described by the U.S. Composting Council (17) and Weatherburn (18), respectively.

Thermal-inactivation kinetics.

An exponential model was used to describe the thermal inactivation of the control and desiccation-adapted Salmonella spp. in aged chicken litter. The equation for the model is given by the formula log₁₀[N_i(t)] = α + βe^−λt + ε_i, where N_i(t) is the number of bacteria at time t for the ith observation, α is the long-term (t → ∞) log count of the bacteria, β is the long-term reduction in the log count of the bacteria, λ is the decay rate, and ε_i is the error for the ith observation.

It was assumed that the distribution for the number of bacteria at time t followed a log-normal distribution, while the possible tailing effect was taken into account. Thus, the errors were assumed to follow a normal distribution with a mean of zero and a variance of σ².

Statistical analysis.

Plate count data were converted to log numbers of CFU g⁻¹ in dry weight. SigmaPlot 12.3 (Systat Software Inc., San Jose, CA, USA) was used for data analysis. Analysis of variance (ANOVA), followed by the least significant differences (LSD) test, was carried out to determine whether significant differences (P < 0.05) existed among different treatments.

For the thermal-inactivation kinetics study, the parameters for the exponential model were estimated using maximum likelihood, which accounted for censored observations that were not detectable by plating. The censored observations were within the interval of 0 to 1.30 log CFU g⁻¹. Separate regression models were used for each moisture-temperature combination. Because of the censoring, a pseudo-R² was calculated for each regression model as described by Magee (19). The pseudo-R² was calculated as pseudo-R² = 1 − exp[(x₂ − x₁)/n], where x₁ is the −2 log likelihood (the model with no independent variables), x₂ is the −2 log likelihood (the current model), and n is the sample size.

Linear contrasts were used for all comparisons, and the type I error rate was controlled at a P of 0.05 using the Bonferroni method. The NLMIXED procedure of Statistical Analysis System 9.3 (SAS Institute Inc., Cary, NC, USA) was used for all calculations in the thermal-inactivation kinetics study.

RESULTS

Fresh chicken litter, aged chicken litter, dairy compost, and poultry compost used in this study were all made free of Salmonella by following the microbiological detection method described by the U.S. FDA's Bacteriological Analytical Manual (20).

Eight media were tested for recovering heat-injured S. Typhimurium, and the largest populations of this pathogen were enumerated on P/TSB-R (P < 0.05) (Table 1). However, the high levels of microbial counts on P/TSB-R not only may be attributed to the repair of injured cells but also may reflect the multiplication of noninjured cells during the 3-h incubation. For the 0.5- and 1-h heat treatments, no significant differences (P > 0.05) in the enumerations of heat-injured S. Typhimurium cells occurred among TSA-R, XLT-4-R, OV/TSA-R, OV/XLT-4-R, TAL/TSA-R, TAL/XLT-4-R, and P/TSA-R. Among media containing TSA-R, TAL/TSA-R medium recovered slightly more Salmonella cells, namely, 3.96 ± 0.35 and 2.80 ± 0.39 log CFU g⁻¹ at 0.5 and 1 h, respectively. With respect to media containing XLT-4-R, higher numbers of S. Typhimurium cells were observed on OV/XLT-4-R medium, namely, 3.80 ± 0.22 and 2.75 ± 0.45 log CFU g⁻¹ at 0.5 and 1 h, respectively.

Table 1.

Comparison of recovery media for heat-injured S. Typhimurium cells in fresh chicken litter

Recovery medium	Population of S. Typhimurium cells (log CFU g−1) after exposure to 75°C fora:
	0 h	0.5 h	1 h
TSA-R	6.71 ± 0.06 b	3.83 ± 0.38 b	2.77 ± 0.24 ab
XLT-4-R	6.14 ± 0.29 c	3.74 ± 0.45 b	2.42 ± 0.70 ab
OV/TSA-R	6.69 ± 0.15 b	3.82 ± 0.52 b	2.20 ± 0.51 b
OV/XLT-4-R	6.61 ± 0.02 b	3.80 ± 0.22 b	2.75 ± 0.45 ab
TAL/TSA-R	6.68 ± 0.12 b	3.96 ± 0.35 b	2.80 ± 0.39 ab
TAL/XLT-4-R	6.19 ± 0.42 c	3.66 ± 0.40 b	2.44 ± 0.68 ab
P/TSA-R	6.69 ± 0.18 b	3.78 ± 0.47 b	2.73 ± 0.35 ab
P/TSB-R	7.59 ± 0.33 a	4.78 ± 0.32 a	3.33 ± 0.71 a

^aData are means ± standard deviations (SD) of results from three trials. Means with different letters in the same column are significantly different (P < 0.05).

As shown in Table 2, S. Typhimurium counts in all four composts decreased during the 24-h desiccation adaptation at room temperature. The populations in fresh and old poultry composts and aged chicken litter decreased more rapidly than those in dairy compost (P < 0.05). The reduced levels of Salmonella cells in dairy compost, fresh poultry compost, old poultry compost, and aged chicken litter were 0.45, 2.99, 2.18, and 2.87 log CFU g⁻¹, respectively. The levels of ammonia (average of 820.64 μg NH₄-N g⁻¹) and pH (average of 8.77) in fresh poultry compost, old poultry compost, and aged chicken litter were much higher (P < 0.05) than in dairy compost (ammonia content of 22.64 μg NH₄-N g⁻¹ and pH of 7.70) (Table 2).

Table 2.

Populations of S. Typhimurium cells before and after desiccation adaptation in different composts and their pH and ammonia levels

Sample	Population of S. Typhimurium cells (log CFU g−1)a		pH	NH4-N concn (μg g−1)
	Before desiccation adaptation	After desiccation adaptation
Dairy compost	8.81 ± 0.01 a	8.36 ± 0.13 a	7.70 ± 0.05 d	22.64 ± 3.22 d
Fresh poultry compost	8.78 ± 0.06 a	5.79 ± 0.14 c	9.09 ± 0.02 a	1,142.55 ± 100.27 a
Old poultry compost	8.76 ± 0.04 a	6.58 ± 0.03 b	8.26 ± 0.01 c	465.82 ± 3.86 c
Aged chicken litter	8.75 ± 0.05 a	5.88 ± 0.25 c	8.97 ± 0.04 b	853.55 ± 72.64 b

^aData are means ± SD of results from two trials. Means with different letters in the same column are significantly different (P < 0.05).

The times required to heat aged chicken litter with different moisture contents to 70, 75, 80, 85, and 150°C ranged from 0.42 to 2.53 h (data not shown). Our results showed that the higher the initial moisture content of the chicken litter, the longer the heating-up time. At 70, 75, 80, and 85°C, Salmonella levels in aged chicken litter decreased in all samples during heat treatment; however, the difference in the populations of control and desiccation-adapted Salmonella cells was significant (P < 0.05) (Fig. 2). For example, at 70°C, in aged chicken litter with a moisture content of 20%, the control cells survived for ∼1.5 to 2 h, as determined by enrichment, whereas the desiccation-adapted cells survived for more than 6 h of heat exposure (Fig. 2A). The desiccation-adapted cells were inactivated much faster when the moisture content of chicken litter was increased. For example, at 80°C, cell counts were still more than 2 log CFU g⁻¹ in chicken litter with a 20% moisture content after exposure to heat treatment for 6 h, whereas Salmonella cells survived for less than 3 h at the 50% moisture content, as determined by enrichment (Fig. 2C).

Fig 2.

Survival of control and desiccation-adapted (DA) Salmonella spp. in aged chicken litter with a 20, 30, 40, or 50% moisture content (MC) at 70°C (A), 75°C (B), 80°C (C), and 85°C (D). The dotted lines indicate that Salmonella was detectable only by enrichment (detection limit by plating, 1.30 log CFU g⁻¹).

At 150°C, desiccation-adapted Salmonella cells still displayed extended survival compared to the nonadapted control cells (Table 3). Control and desiccation-adapted cells in aged chicken litter at a 20% moisture content were detectable after enrichment for up to 10 and 50 min, respectively. Desiccation-adapted cells in chicken litter had shorter durations of survival with an increase in moisture content. Viable Salmonella cells in chicken litter could still be detected up to 50 min at the 20 and 30% moisture contents, whereas they were detectable for only 40 min at the 40 and 50% moisture contents.

Table 3.

Survival of control and desiccation-adapted Salmonella spp. in aged chicken litter at 150°C

Sample	Moisture content (%)	Survival with an exposure time (min) ofa:
		10	20	30	40	50	60
Control	20	+	−	−	−	−	−
Desiccation-adapted cells	20	+	+	+	+	+	−
	30	+	+	+	+	+	−
	40	+	+	+	+	−	−
	50	+	+	+	+	−	−

^a+, detectable by enrichment; −, not detectable by enrichment.

All the parameter estimates obtained from fitting the experimental observations into an inactivation model and the pseudo-R² values are shown in Table 4. The exponential model used in this study was appropriate for fitting all the inactivation curves and permitted the modeling of an extended tail for desiccation-adapted Salmonella, which was supported by the good performance of goodness-of-fit experiments (pseudo-R²). α values of desiccation-adapted cells were higher than those of control cells (P < 0.05), which was attributed to the tailing part in the inactivation curves and thus reflected a higher level of population remaining viable at the end of thermal treatment. Meanwhile, β values of desiccation-adapted cells were lower than those of control cells (P < 0.05), which suggested a lower population reduction and that cells were more heat resistant at longer exposure times. For desiccation-adapted cells, as temperature increased from 70 to 85°C and moisture content increased from 20 to 50%, there seemed to be a trend in temperature and moisture content dependencies for these two parameters, as α values decreased and β values gradually became higher. Interestingly, the moisture content threshold to achieve a long-term log count (α value) of zero for desiccation-adapted cells decreased with an increase in temperature. At 70°C, there was no moisture content threshold, but at higher temperatures, including 75, 80, and 85°C, the moisture content thresholds were 50, 40, and 30%, respectively. It should be remarked that the λ values (decay rates) obtained from the inactivation model exhibited no dependencies on either nonadapted control or desiccation-adapted cells, temperature, or moisture content, and it is thus difficult to draw a definite conclusion from this parameter.

Table 4.

Parameter estimates of the inactivation model for control and desiccation-adapted Salmonella spp.^a

Temp (°C)	Sample	Moisture content (%)	Long-term log count (α)	Long-term reduction in log count (β)	Decay rate (λ)	Pseudo-R2
70	Control	20	−0.11 ± 0.18 B	6.98 ± 0.32 A	1.06 ± 0.11 B	0.96
	Desiccation-adapted cells	20	3.58 ± 0.05 Aa	3.35 ± 0.12 Bb	1.95 ± 0.19 Aa	0.97
		30	2.98 ± 0.08 b	3.95 ± 0.18 ab	1.67 ± 0.18 a	0.96
		40	2.50 ± 0.21 b	4.43 ± 0.42 ab	1.70 ± 0.50 a	0.83
		50	1.50 ± 0.51 b	5.00 ± 0.51 a	0.83 ± 0.35 a	0.83
75	Control	20	−0.05 ± 0.08 B	7.00 ± 0.18 A	1.49 ± 0.09 A	0.98
	Desiccation-adapted cells	20	2.39 ± 0.09 Aa	4.51 ± 0.19 Bb	1.60 ± 0.17 Aa	0.96
		30	1.62 ± 0.35 a	4.95 ± 0.46 ab	1.07 ± 0.37 a	0.84
		40	1.11 ± 0.36 a	5.43 ± 0.46 ab	1.02 ± 0.30 a	0.86
		50	0.00 ± 0.38 b	6.29 ± 0.49 a	0.75 ± 0.17 b	0.88
80	Control	20	−0.02 ± 0.01 B	7.06 ± 0.03 A	2.19 ± 0.02 A	0.99
	Desiccation-adapted cells	20	2.28 ± 0.15 Aa	4.54 ± 0.30 Bc	1.46 ± 0.27 Ba	0.92
		30	1.78 ± 0.20 a	5.14 ± 0.39 bc	1.73 ± 0.40 a	0.87
		40	0.04 ± 0.23 b	6.55 ± 0.38 ab	0.96 ± 0.14 a	0.92
		50	−0.02 ± 0.12 b	6.94 ± 0.25 a	1.26 ± 0.11 a	0.96
85	Control	20	−0.02 ± 0.01 B	7.11 ± 0.03 A	2.79 ± 0.04 A	0.99
	Desiccation-adapted cells	20	1.77 ± 0.11 Aa	5.18 ± 0.24 Bb	1.53 ± 0.19 Bab	0.95
		30	0.63 ± 0.36 b	5.98 ± 0.50 ab	1.08 ± 0.32 b	0.87
		40	−0.06 ± 0.05 b	7.12 ± 0.11 a	1.51 ± 0.06 b	0.99
		50	−0.06 ± 0.04 b	7.09 ± 0.09 a	2.06 ± 0.07 a	0.99

^aData are means ± standard errors (SE) of results from two trials. At each temperature, for desiccation-adapted cells, means with different lowercase letters in the same column are significantly different (P < 0.05), while means for control and desiccation-adapted cells (20% moisture content) with different uppercase letters in the same column are significantly different (P < 0.05).

Colonies that exhibited the longest survival based on growth on TAL recovery media (0.5 h, and 6 and 24 h for control and desiccation-adapted cells, respectively) at 80°C were characterized using PFGE (Table 5). For the nonadapted control, 3 and 6 of 12 isolates were identified as S. Senftenberg and S. Typhimurium, respectively, whereas 7 and 3 of 12 isolates from desiccation-adapted cells were identified as S. Senftenberg and S. Typhimurium, respectively. Our results also showed that Salmonella could still be detected by enrichment after 24 h at 80°C in chicken litter with a 20% moisture content, and all 12 isolates were identified as S. Senftenberg.

Table 5.

Characterization by PFGE of Salmonella spp. in aged chicken litter with a 20% moisture content after thermal inactivation at 80°C

Sample	Serotypes (no. of colonies)a
Control (after a 0.5-h heat treatment)	S. Enteritidis (n = 2), S. Heidelberg (n = 1), S. Senftenberg (n = 3), S. Typhimurium (n = 6)
Desiccation-adapted cells (after a 6-h heat treatment)	S. Enteritidis (n = 2), S. Heidelberg (n = 0), S. Senftenberg (n = 7), S. Typhimurium (n = 3)
Desiccation-adapted cells (after a 24-h heat treatment)	S. Enteritidis (n = 0), S. Heidelberg (n = 0), S. Senftenberg (n = 12), S. Typhimurium (n = 0)

^aTwelve colonies were tested from each sample.

DISCUSSION

In order to accurately enumerate the bacterial populations after heat treatment, eight media were compared for recovery of heat-injured Salmonella cells. Although the traditional OV method is useful for enumeration of heat-injured cells, it is difficult to isolate pure colonies on OV/XLT-4-R that grow on this selective medium. A modified TAL method (TAL/TSA-R) was also adopted to allow the isolation of colonies for further characterization by PFGE. Therefore, TAL/TSA-R and OV/XLT-4-R were selected as recovery media for subsequent thermal-inactivation experiments with Salmonella spp. For data analysis, counts of Salmonella populations on these two media were averaged in order to simplify the data.

Dairy compost, fresh poultry compost, old poultry compost, and aged chicken litter were compared to select the compost matrix for desiccation adaptation. Our results indicated that dairy compost was much better at maintaining Salmonella cells at high numbers than other tested composts, which contained higher ammonia contents; this possibly accelerated the death of Salmonella in these samples. Ammonia has been reported to cause a significant reduction in the number of human pathogens, such as S. Typhimurium, Escherichia coli O157:H7, and L. monocytogenes, in chicken manure (21). Therefore, dairy compost was selected as the matrix for the desiccation adaptation of Salmonella in subsequent thermal-inactivation experiments.

In the present study, Salmonella populations in aged chicken litter decreased during exposure to all tested temperatures, with shorter survival times at higher temperatures. These findings are in general agreement with other published data of the thermal inactivation of Salmonella spp. in chicken litter (1, 4). Wilkinson et al. (1) could not detect any S. Typhimurium in poultry litter (30, 50, and 65% moisture contents) after 1 h at 55 or 65°C in a water bath. Kim et al. (4) found that a 7-log reduction in the number of Salmonella species cells in fresh chicken litter (30, 40, and 50% moisture contents) could be achieved by heat treatment at 80°C for 44.1 to 63.0 min. The above-described results on temperature-time combination requirements for eliminating Salmonella varied among studies. This difference may be explained by differences in the compositions and moisture levels of compost material, Salmonella strains, and also heating sources.

Our results clearly revealed that bacterial cells became less heat resistant when the moisture content of aged chicken litter increased from 20 to 50%, which is in agreement with the results of Kim et al. (4), who used fresh chicken litter. Within the same duration of exposure, moist heat is known to kill microbial cells at a lower temperature than dry heat (22). Exposure to moist heat generally denatures enzymes and other essential proteins, as well as membranes and nucleic acids. Interestingly, our finding is in contrast with the results obtained by Kim et al. (4), who reported that Salmonella survived longer in aged chicken litter with 40 and 50% moisture contents than in litter with a 30% moisture content. This difference may be attributed to different chicken litter samples with various chemical compositions and physical characteristics, which can result in different heat transfer kinetics. As suggested by Kim et al. (4), aged chicken litter used in their study had high levels of heavy metals, which might become more soluble at higher moisture contents, resulting in the stronger heat resistance of Salmonella. Additionally, our regression model analysis of thermal-inactivation data for desiccation-adapted cells clearly suggests a moisture content threshold at higher temperatures. With the boosted killing rate of Salmonella as temperature is increased, the impact of moisture content on pathogen inactivation may reach a plateau. This information is especially critical when designing thermal processes for chicken litter with varied moisture levels.

Previously published studies on the survival of Salmonella exposed to heat treatment in composts have used nonstressed cultures (1, 4). Under real-world stockpiling conditions for chicken litter, Salmonella cells may have been exposed to a dry environment for a long period of time. It is well known that bacteria can be induced to change under stress, such as osmotic stress (a dry environment) (5), and adaptation to a stress may also cross-protect bacterial cells against other lethal stresses. In support of this notion, other studies have demonstrated the thermal tolerance of desiccation-adapted Salmonella spp. in a variety of matrices. Hiramatsu et al. (8) reported that desiccated Salmonella spp. in dried paper disks acquired the remarkable ability to survive for 5 h of exposure to 70 and 80°C. Similar results have been reported by Gruzdev et al. (6), who found that desiccated Salmonella cells in sterile deionized water showed high tolerance to dry heat at 60°C, with no significant population change within 1 h, in comparison to a 3-log reduction in the number of nondesiccated cells under identical conditions. Also in the work of Mattick et al. (23), Salmonella cells that habituated at a low a_w of 0.95 for 12 h increased their heat resistance in tryptic soy broth at 54°C, with a >4-fold increase in D values at this temperature.

To our knowledge, there is no information available regarding the cross-tolerance of desiccation-adapted pathogens in compost and manure to further thermal treatment. Nevertheless, the heat shock response at sublethal temperatures by simulating the mesophilic phase of composting has been shown to provide multiple means of protection from composting stresses to Salmonella. In the field study of Shepherd et al. (24), heat-shocked E. coli O157:H7 and S. Typhimurium in dairy manure cocomposted with vegetable wastes had extended survival times in the summer composting trial. They demonstrated that the heat shock treatment may have induced the cross-resistance of these pathogens to desiccation stress. Additionally, results of Singh et al. (25) suggested that heat-shocked E. coli O157:H7, Salmonella, and L. monocytogenes at 47.5°C survived longer than non-heat-shocked cells at composting temperatures (50, 55, and 60°C). Noticeably, Singh and Jiang (26) found that acid adaptation, a stress response produced during exposure to animal gastric acidity, provided E. coli O157:H7 only with cross-protection to thermal inactivation in saline solution; however, this protection was lost in fresh dairy compost during a simulated optimal mesophilic phase of composting.

In this study, we have examined the cross-tolerance of desiccation-adapted and nonadapted Salmonella cells to heat treatment. Thermal-inactivation curves for control and desiccation-adapted Salmonella cells were all nonlinear regardless of temperature and moisture content. However, desiccation-adapted cells had significantly higher populations throughout heat treatment and also survived much longer than control cultures. Compared to the control, there were ca. >3-, >4-, ∼4- to 10-, and >6-fold increases in the exposure times required for reducing by 5 logs the numbers of desiccation-adapted Salmonella cells at 70, 75, 80, and 85°C, respectively. Presumably, one reason for this high level of thermal tolerance may be due to the fact that Salmonella cells were well adapted to survive under dry conditions, which resulted in induced cross-protection against thermal inactivation. Besides, it should be noted that pronounced tailing was observed in the survival curves of desiccation-adapted Salmonella at 70, 75, 80, and 85°C, which was further confirmed by our thermal-inactivation kinetics study. This finding is consistent with the data published by Kim et al. (4), who reported that thermal-inactivation curves of Salmonella in aged chicken litter showed an extensive tailing effect. A possible explanation for the tailing is that heat-sensitive cells were inactivated at a relatively high rate but that survivors with higher resistance were left behind. Our observation implies that desiccation-adapted cells from the tailing in survival curves should be considered sufficiently when applying thermal treatment to chicken litter. Otherwise, inadequate thermal processing will lead to the survival of a few heat-resistant desiccation-adapted cells, and when heat-treated chicken litter is used as organic fertilizer or soil amendment, these desiccation-adapted cells might contaminate produce in the field.

Although the mechanism of heat resistance of desiccation-adapted bacteria is still not clearly understood, the “water replacement hypothesis” is a possible explanation. In this hypothesis, some nonreducing disaccharides, such as trehalose, are considered to play a pivotal role in stabilizing the structures of membranes and proteins under desiccation conditions (27). There are two genes that are supposed to be involved in trehalose biosynthesis, the otsA gene and otsB gene, which encode trehalose-6-phosphate synthase and trehalose-6-phosphate phosphatase, respectively (28). In the study of Howells et al. (29), the otsA and otsB genes were isolated from S. Typhimurium and their nucleotides sequenced. Howells et al. also found that after incubation at 50°C for 7 min, the number of viable cells of a S. Typhimurium otsA mutant decreased to 0.6% of the initial population, whereas the wild type still had an 8% survival rate. Further systematic experiments are warranted to explore the underlying mechanism of thermal resistance of desiccation-adapted Salmonella cells in chicken litter.

It is generally recommended that multiple strain composites of well-characterized pathogens should be utilized in the challenge studies to evaluate the efficacy of heat-based hurdles (30). Moreover, as also proposed by others (30, 31), we hold the position that when a bacterial cocktail is constructed for experiments related to food safety, strains with robust stress resistance behavior should be included. In the present study, the thermal inactivation of Salmonella spp. was thus investigated using four serotypes, including three serotypes (S. Enteritidis H2292, S. Heidelberg 21380, and S. Typhimurium 8243) most frequently isolated from chicken litter (32) and one heat-resistant serotype (S. Senftenberg ATCC 43845) (33). An obvious variability in heat resistance among Salmonella serotypes was observed during thermal exposure of aged chicken litter, since S. Senftenberg and S. Typhimurium exhibited higher resistance profiles than the other two serotypes. To characterize the serotypes that could be detected by enrichment only after a longer period of time, we also carried out a 24-h heat treatment at 80°C for chicken litter with a 20% moisture content. S. Senftenberg, the most heat-resistant Salmonella serotype, could even survive for up to 24 h at 80°C. At this point, a significant practical consequence is that serotypes with robust thermal-inactivation characteristics, such as S. Senftenberg, may be used as indicator microorganisms to ensure microbial risk assessment of the worst-case scenario when the thermal processing of chicken litter is evaluated in further heat challenge studies. In the work of Mocé-Llivina et al. (34), S. Senftenberg was selected as a suitable bacterial indicator when the thermal treatment of dewatered sludge and raw sewage was assessed. As also reported by Murphy et al. (35), if a particular thermal treatment destroys S. Senftenberg, it will also be preferably effective against more-common Salmonella spp. In their study, S. Senftenberg was thus used as a test organism to determine the thermal inactivation of Salmonella spp. in meat products.

In summary, our results demonstrated that Salmonella cells that adapted under desiccation conditions survived substantially longer in aged chicken litter than nonadapted control cells exposed to the same heat treatment. The reduced moisture levels in chicken litter contribute to the better survival of Salmonella during heat treatment. It needs to be recognized that if these surviving cells are present in chicken litter as organic fertilizer, there is a great risk for introducing cross-contamination of fresh produce before harvest. Additionally, comparison of heat resistance characteristics among four Salmonella serotypes suggested that S. Senftenberg can be used as an indicator microorganism for validating thermal processing of chicken litter. Our research has important implications for chicken litter processors for validating and modifying their heating processes in order to eliminate Salmonella cells that may have been subjected to dry stress.