Bacterial preparation
Vibrio alginolyticus strain HY9901 [11] (tdh gene positive) was originally isolated from spoiled Lutjnaus erythropterus which cause food poisoning. This strain was kindly provided by the College of Fisheries, Guangdong Ocean University (Zhanjiang, China). The strain was confirmed using PCR by amplification of a hypervariable region of the 16S rDNA gene and preserved in Tryptone Soya Broth ( Huangkai, Guangzhou, China) supplemented with 2 % NaCl and 20 % glycerol at - 80 °C. The strain was selected on thiosulfate citrate bile salt sucrose (TCBS) agar and grown with agitation at 30 °C for 24 h in Luria-Bertani (LB) medium (Beijing Land Bridge Technology Co., Beijing, China) supplemented with 2 % (w/v) NaCl. The bacterial cells were centrifuged at 2,500 g for 5 min and re-suspended in LB medium. The bacterial concentration was confirmed by plate count and adjusted to 105 CFU/mL prior to inoculation of different food matrices.
Food matrices preparation and bacterial inoculation
Four raw seafood [shrimp (Litopenaeus vannamei), briny tilapia (Oreochromis mossambicus), scallop (Argopecten irradians), oyster (Crassostrea gigas)] and three raw non-seafood [pork, chicken, freshwater fish (Ctenopharyngodon idellus)] types were obtained from a local supermarket in Zhanjiang, China, and stored at -20 °C. The meat (muscle) from these animals/fish was used in the study. For cooked foods, egg-fried-rice was prepared by mixing 50 g of egg with 50 g of boiled rice and cooked at 85 °C for 10 min, and the thawed pork and chicken separately added to boiling water and left for 20 min according to the method of Xie and co-workers [47]. Then the cooked egg-fried-rice, pork and chicken were transferred into a biosafety hood and left to cool to room temperature before subsequent treatment.
The number of each food matrix used in the study was eleven (n=11). Test portions, 10 ± 1 g each of raw briny fish, shrimp, scallop, oyster, pork, chicken and freshwater fish, were separately soaked in sterile water containing 100 ppm chlorine at 15 °C, gently shaken for 5 min and washed 10 times with sterile water to inactivate the native bacteria [48]. Salt at 2 % was added to all sterilized raw and cooked food matrices and transferred to sterile Erlenmeyer flasks. Next, each sample was inoculated with 1 ml of V. alginolyticus and mixed thoroughly in a vortex mixer (XW-80A, Qilinbei, Haimen, China) for 10 min to ensure uniform distribution of ~104 CFU/g V. alginolyticus in the samples.
Experimental procedure
Construction of the primary growth predictive model
The inoculated samples were stored at 30 ± 0.1 °C in isothermal temperature incubators. Samples were taken at 0, 2, 4, 6, 9, 12, 18, 24, and 36 h for the growth characteristic study. At each time point, 10 g of each food matrix was mixed with 90 mL of sterile 0.85 % physiological saline and vortexed (XW-80A, Qilinbei, Haimen, China) for 2 min. The homogenates were diluted (10×) in sterile 0.85 % physiological saline and 100 μL homogenate was plated onto thiosulfate citrate bile salts sucrose agar (TCBS, Beijing Land Bridge Technology Co., Beijing, China) in triplicate and incubated at 30 °C for 18-20 h. Colony forming units (CFU) were counted manually to determine the density of viable cells in each sample (CFU/g). The plate counting test was repeated three times.
Model fitting
The experimental data obtained at different time points at 30 °C and conditions often used to describe the bacterial growth curves of foods were fitted to a modified Gompertz model [Equation 1] and a logistic model using Origin Pro. 9.0. The best fitting model was used to describe the growth characteristics of V. alginolyticus on different food matrices.
where Nt :cell density at a particular storage time (lg CFU/g); N0: initial microbial cell density (lg CFU/g); λ: lag time (h); μmax : maximum specific growth rate (h-1).
Evaluation of model performance
In order to evaluate the goodness-of-fit of the modified Gompertz and logistic models, coefficient of determination (R2), bias factor (BF), accuracy factor (AF) and the mean square error (MSE) were calculated. Goodness-of-fit of primary model was evaluated using the adjusted R2. The MSE was used to evaluate the difference between the growth data estimated by the model with that measured experimentally, with the MSE values approaching zero indicating a closer fit of the data for the model. Besides, validation experiments were carried out to evaluate the models by AF and BF. AF indicates the spread of the results around the predicted values. BF measures the relative average deviation of the predicted and observed V. alginolyticus growth. In the study, the predictions exceeding observed data and < 10% on average in terms of lg (CFU/g), were considered to be accurate. That is, 1.0 < BF < AF < 1.1 was defined as a satisfactory limit. R2, MSE, BF and AF were defined by the following equations 2-5 [49, 50, 51, 52, 53].
In the above equations, Npredicted is the predicted bacterial number, Nobserved the observed bacterial number (lg CFU/g), μobserved the observed specific growth rate, μpredicted the predicted specific growth rate (h-1), n the number of observations and m the number of parameters of the model.
Hemolysin measurement
All food samples were incubated at 30 °C for 24 h [54, 55] following which 10 g of each food matrix was separately washed with 10 mL of sterile 0.01 M phosphate buffered saline solution (PBS, pH 7.2) and the solution centrifuged (Thermo Lynx 6000, Thermo Scientific, Waltham, MA) at 12,000 rpm for 20 min at 4 °C. The supernatants were filtered (0.22 μm, Millipore, Billerica, MA) and stored at -20 °C until use [56, 57, 58]. The blank control of the different food matrices were subjected to the same procedure except they were not inoculated with V. alginolyticus. The V. alginolyticus cultured in LB medium was used as the control group.
TDH activity test
Kanagawa phenomenon was tested as previously described by Takeda [59] with slight modifications. The test plate consisted of 5 % rabbit red blood cells (RBCs), 5 μg/mL chloramphenicol and 0.1 % arabinose in Wagatsuma agar medium (Beijing Land Bridge Technology Co., Beijing, China). Once the blood agar solidified, holes were punched in the agar plate. Then, 50 μL of the sample was added to each hole and cultured overnight at 37 °C. The diameter of the β-hemolysin zone around each well was measured, an indicator of a positive reaction for TDH activity.
Hemolytic titer test
Total hemolytic products in the food samples were measured using a hemolytic titer assay [60]. Briefly, rabbit RBCs were extracted by centrifugation of blood (3,500 rpm for 10 min, Thermo Lynx 6000) three times (washing with PBS each time) and diluted to 2 % with PBS. The filtrate (100 μL) of each food matrix was mixed with 100 μL of PBS (0.01 M, pH 7.2) followed by serial 2-fold dilution up to 1: 4096 in a volume of 250 μL/well in a 96-well U-bottom microtiter plate. Twenty-five microliters of 2 % RBCs was added to each diluted solution in wells and incubated for 1 h at 37 °C, then RBCs were resuspended and the plates transferred to 4 oC for 2 h. The dilution that gave 50 % hemolysis was counted. The reciprocal value of this dilution indicated the hemolytic titer in the undiluted sample. The sample are defined as the filtrates of briny tilapia, shrimp, scallop, oyster, freshwater fish, raw pork, raw chicken, cooked pork, cooked chicken and egg fried rice, which were contaminated by V. alginolyticus. The rabbit used in this study was obtained from the animal center of Guangdong Province.
Statistical analysis
Results are expressed as mean ± standard error. The differences between groups were determined using the LSD test in SPSS software, version 19.0 (SPSS). A p value of < 0.05 was considered as statistically significant.