A complete factorial design (3 × 3 × 8 × 3) with three variables – moisture level, storage temperature and brand of dog food - was used to examine Salmonella survival/growth in eight commercial dog food brands. Temperatures of 18, 22, and 28 °C were selected to mimic the range of temperatures that might be encountered during different seasons. Added moisture levels were set at 20, 35, and 50% added water.
Analysis of background microflora of commercial dry dog foods
A 5-lb bag of each of the eight brands of dog food was purchased at a local supermarket after careful visual examination to ensure the packages were undamaged. Upon opening of the bags in the laboratory, samples of the dog food were analyzed for Salmonella and aerobic plate counts using the methods described below in conjunction with the standard cultural and confirmatory techniques from the FDA Bacteriological Analytical Manual were used to screen for Salmonella (Andrews et al. 2016). The brands were also tested for water activity and pH (see below).
Bacterial strains used and preparation of inocula
In all tests, dry dog food was inoculated with a cocktail of two S. enterica strains: Salmonella enterica serovar Typhimurium CVM98 (animal isolate) and Salmonella enterica serovar Enteritidis KPL13076 (clinical isolate originally obtained from CDC). These strains were selected based on their ability to survive for extended periods in dry environments (Oni et al. 2015). An exception was one growth kinetics study at 30 °C which was conducted using S. Typhimurium CVM 98 and S. Typhimurium LT-2, due to the loss of S. Enteritidis KPL13076 stock culture during a power failure in our −80 °C freezer. S. Typhimurium LT-2 is a laboratory strain used extensively for studying Salmonella detection and behavior in various foods including dry foods undergoing extended dry storage. The strains were acquired from the culture collection of the Department of Nutrition and Food Science, University of Maryland. After activating individual stock cultures of both Salmonella strains by streaking onto Brain Heart Infusion Agar (BHIA) plates (Becton Dickinson, Sparks, MD) and incubating at 37 °C for 18 to 24 h, a single colony of each strain was selected from each plate, streaked onto separate plates of Xylose Lysine Desoxycholate Agar (XLDA) (Becton Dickinson), and incubated at 37 °C for 24 h. Single black colonies were selected from the XLDA plates and used to inoculate five 10-ml tubes of BHI broth (Becton Dickinson), which were then incubated at 37 °C for 24 h. The five 10-ml tubes were combined in a sterile 50-ml centrifuge tube (BD Falcon, Franklin Lake, NJ), and centrifuged at 3,000 × g for 10 min at 7 °C. Cell pellets were washed three times with 5 ml of sterile 0.1% peptone water and re-centrifuged, and the final cell pellet was re-suspended in 3 ml of sterile 0.1% peptone water. Equal volumes of each strain were combined, re-centrifuged, and re-suspended in 1 ml of sterile 0.1% peptone water to produce the two-strain cocktail with a final concentration of approximately 109 CFU/ml.
Preparation of samples and measurement of Salmonella growth/survival at specified moisture levels
The pH and water activity (aw) of each dry dog food brand was measured upon opening each bag. pH was determined at 25 °C by weighing 1 g portions, pulverizing with a wooden mallet, hydrating with distilled water (1:2.5 g/v), and using a pH meter (Orion pH electrode 9165 BN, Orion Research, Boston, MA, USA) to take measurements. A water activity meter (Novasina IC-500, AW-LAB, Switzerland) was used to measure water activity using the manufacturer’s specifications.
Three 140-g portions of each brand of dog food, one for each target moisture level, were weighed into sterile plastic bags. The appropriate amounts of sterilized distilled/deionized water were added to rehydrate the dog food samples to 20, 35, 50% added water as calculated based on the initial weight of the dog food. No attempt was made to have all the brands have the same percent moisture or water activity after rehydration. Working under a biosafety hood, 10 μl (0.01 ml) of the concentrated Salmonella cocktail was transferred to the water to be added to the dog food. After vortexing for 20 s, the diluted inoculum was gradually added to the corresponding dog food sample. With each addition, the bag was gently massaged and shaken to ensure a homogeneous distribution. Portions (~10 g) of inoculated dog food were transferred to triplicate labeled plastic containers and stored for 72 h at 18, 22 and 28 °C. Three 10-g samples of each moisture level were analyzed immediately to determine the initial Salmonella population densities. After incubation, each 10-g sample was transferred to a Whirlpak bag (Nasco, Fort Atkinson, WI), mixed with 90 ml of sterile 0.1% PW, and stomached for 30 s. The supernatant was used to make serial dilutions, after which 50 μl aliquots of appropriate dilutions which were spiral-plated in duplicate on BHIA (total aerobic bacteria) and XLDA (S. enterica). This dual media plating system was used to allow estimation of the degree of injury of salmonellae recovered from the samples (Oni et al. 2015). Plates were incubated at 35 °C for 48 h, and enumerated at 24 and 48 h using an automated colony counter (Neutec Group Inc., Farmingdale, NY).
Growth kinetics of Salmonella in dry dog food
Two additional studies to more closely examine the growth kinetics of the S. enterica were carried out with dog food brands #2 and #4 at the 35% rehydration level. A rehydration level of 35% was selected as the level most likely used by consumers, based on an informal survey of pet owners in our laboratory. In the first study, the strong temperature dependency of S. Typhimurium CVM98/S. Enteritidis KPL13076 growth in brand #4 (Fig. 1b) was followed over 72 h using four temperatures: 15, 20, 25, and 30 °C. Growth was measured over 72 h by periodically taking duplicate samples and quantifying Salmonella population densities as described above. The second related study examined the growth kinetics of S. Typhimurium CVM98/S. Typhimurium LT-2 in brand #2 rehydrated to 35% and incubated at 30 °C as described above. This brand was selected because it was one of the brands that supported substantial growth at this rehydration level. Salmonella counts obtained from XLDA plates were log-transformed to Log(CFU/g), and the time series fitted to the three-phase linear model (Buchanan et al. 1997) using the IPMP-2014 software (USDA/ARS, Wyndmoor, PA).