Journal Pre-proof The inhibitory effects of spice essential oils and rapidly prediction on the growth of Clostridium perfringens in cooked chicken breast Yaodi Zhu, Yangyang Ma, Jiaye Zhang, Miaoyun Li, Longgang Yan, Gaiming Zhao, Yanxia Liu, Yanyan Zhang PII:
S0956-7135(19)30567-5
DOI:
https://doi.org/10.1016/j.foodcont.2019.106978
Reference:
JFCO 106978
To appear in:
Food Control
Received Date: 3 July 2019 Revised Date:
26 October 2019
Accepted Date: 31 October 2019
Please cite this article as: Zhu Y., Ma Y., Zhang J., Li M., Yan L., Zhao G., Liu Y. & Zhang Y., The inhibitory effects of spice essential oils and rapidly prediction on the growth of Clostridium perfringens in cooked chicken breast, Food Control (2019), doi: https://doi.org/10.1016/j.foodcont.2019.106978. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.
1
The inhibitory effects of spice essential oils and rapidly prediction on the growth
2
of Clostridium perfringens in cooked chicken breast
3
Yaodi Zhu#1, Yangyang Ma#1, Jiaye Zhang1, Miaoyun Li*1, Longgang Yan1, Gaiming
4
Zhao1, Yanxia Liu1, Yanyan Zhang2
5
1. College of Food Science and Technology, Henan Agricultural University, No.95
6
Wenhua Road, Zhengzhou, 450002, P.R. China
7
2. Henan Key Laboratory of Cold Chain Food Quality and Safety Control, Zhengzhou
8
University of Light Industry, Zhengzhou, P. R. China
9
#.The authors contributed equally to this study.
10
*Corresponding author. Tel.: 86-371-63558150; fax: 86-371-63558150.
11
E-mail:
[email protected] (Miaoyun Li)
12
1
13
Abstract
14
In this study, the effects of spices essential oils on the growth of C. perfringens
15
were investigated. The in vitro inhibitory activities and minimum inhibitory
16
concentrations (MIC) of different essential oils on C. perfringens were determined
17
using the Oxford cup and two-fold dilution method. And two models (the parameter-
18
adjusted Gompertz kinetic model, and least square support vector machine (LS-SVM)
19
model) were used to rapidly predict the relative growth/survival of C. perfringens in
20
cooked chicken breast under different essential oil concentrations. The results
21
indicated that cinnamon essential oil exhibited a notable inhibitory effect on C.
22
perfringens in vitro. Moreover, cinnamon essential oil had the lowest inhibitory
23
concentration, and the growth of C. perfringens ceased completely at 19.12 mg/mL
24
and 22.72 mg/mL, respectively. By comparing, the LS-SVM model is the optimum
25
model. The prediction accuracy of the LS-SVM model was greater than 0.99, and the
26
degree of fitting is higher than adjusted Gompertz kinetic model. The results present
27
illustrate that the model using LS-SVM has high prediction accuracy and can be
28
employed to predict the number of C. perfringens under different essential oils. This
29
work could contribute to control and rapid assessment of food safety hazards of C.
30
perfringens during meat processing.
31
Key words: Clostridium perfringens; Spice essential oils; Inhibitory effects;
32
Prediction model; Cooked chicken breast
33
34
35
2
36
1. Introduction
37
Clostridium perfringens (C. perfringens) spores are ubiquitously present in the
38
environment, and can be found in the intestinal tracts of humans and animals, soil,
39
water, and many ingredients and raw materials used to make meat and poultry
40
products (London and Payne et al., 2017). It is known for forming capsules and
41
producing a large amount of gas through the breakdown of carbohydrates in muscles
42
and connective tissues(Makino and Xu et al., 2019). Moreover, the spores of C.
43
perfringens can survive during food preservation processes, and upon germination
44
outgrowth, it can cause food spoilage as well as safety risks (Lindstrom and
45
Heikinheimo et al., 2011). Therefore, it is not only highly pathogenic, but can also
46
result in great economic losses for the meat industry (Grass and Gould et al., 2013;
47
May and Polkinghorne et al., 2016).
48
C. perfringens has been widely detected in ready-to-eat cooked meat products in
49
China (Koziel and Kukier et al., 2019). Due to the difficulties in sterilization,
50
handling and controlling C. perfringens in meat products, many developed countries
51
have considered C. perfringens as an important factor for formulating HACCP and
52
related food safety policies for meat products (Acheson and Bell et al., 2016). Though
53
being one of the research hotspots on foodborne pathogenic bacteria (Alnoman and
54
Udompijitkul et al., 2017; Li and Lillehoj et al., 2017), at present, the control of C.
55
perfringens in foods is mainly achieved by microbiological testing of products. This is
56
not only time-consuming and laborious, but can also suffer from frequent delays in
57
obtaining test results. Thus, rapid predicting and efficient inhibiting of C. perfringens
58
are of great importance to completely guarantee food safety.
3
59
Spice essential oils are aromatic oils extracted from spice plants, which are safe,
60
natural and non-toxic with the antimicrobial activity and health benefits (Chan and
61
Gan et al., 2018). Therefore, it holds broad prospects of spice essential oils for further
62
development to replace antibiotics for the inhibitory effects against various bacteria.
63
So far, extensive researches have been done. For example, it was shown that clove,
64
forsythia and cinnamon essential oils had clear inhibitory activities against bacteria,
65
yeasts and molds (Renyong and Weichang et al., 2008). And thyme, cinnamon, clove,
66
almond, Ligusticum striatum and marjoram essential oils were found to exert strong
67
and comprehensive inhibitory effects against the 25 strains tested (Dorman and Deans,
68
2004; Costa and Bounatirou et al., 2006). Besides, it was reported that addition of
69
oregano essential oil greatly influenced the proliferation of S. typhimurium and
70
prolonged the shelf life of fresh meat (Skandamis and Nychas, 2002). However, there
71
are few studies about the inhibition of C. perfringens by spice essential oils. In China,
72
various spices (star anise, dictamnus, laurel, malabathrum, etc.) are often added
73
during the manufacturing process of sauced and braised meat products (Renyong and
74
Weichang et al., 2008). Therefore, investigation into the inhibitory effects of spice
75
essential oils on the growth of C. perfringens is of great significance.
76
Predictive microbiology is a science that directly assesses food safety by
77
predicting the growth, survival and death of microorganisms, which combining the
78
knowledge of food science, microbiology and statistics with the aid of computers and
79
mathematical models (Akkermans and Nimmegeers et al., 2018; Gonzalez and Possas
80
et al., 2018). Investigating the proliferation characteristics and prediction models of C.
81
perfringens could provide a reference for the control of this pathogen in meat.
82
Modified Gompertz model is widely applicable and can best describe the growth of
83
bacteria under different temperature in food products (Zwietering and Jongenburger et 4
84
al., 1990).. Support vector machines (SVM) is developed by Vapnikhave’s group
85
(Vapnik, 1998) which been successfully applied in classification and regression.
86
Details of the least square support vector machine (LS-SVM) method were reported
87
previously elsewhere. SVMs or support vector networks were introduced to
88
nonlinearly map input vectors to a very high dimensional feature space in which a
89
linear decision surface is available. The details of them can be found in literature
90
(Wang and Zhang et al., 2019). The LS-SVM, which is a common prediction model
91
for small data sets, considers equality type constraints instead of inequalities as in the
92
classic SVM approach. This reformulation greatly simplifies a problem such that the
93
LS-SVM solution follows directly from solving a set of linear equations rather than
94
from a convex quadratic program (Anuragi and Sisodia, 2019). Finding an optimal
95
kinetic model for predicting the survival of C. perfringens is beneficial to directly
96
assess food safety.
97
Herein, the antibacterial effectiveness of a range of essential oils (cinnamon,
98
black pepper, wormwood, fennel and ginger) against C. perfringens (standard strain
99
ATCC 13124 and isolated strain C1) and the MIC were determined. Furthermore, the
100
optimal prediction model for C. perfringens survival was identified and selected by
101
comparing the parameter-adjusted Gompertz survival kinetics model, and LS-SVM.
102
This work might help to provide technical support for the control and rapid prediction
103
of C. perfringens during meat processing.
104
2. Materials and methods
105
2.1 Preparation of spice essential oils
106
The essential oils, including cinnamon (cinnamal dehyde as the main ingredient),
107
black pepper (pinene as the main ingredient), fennel (anethol as the main ingredient), 5
108
ginger (citronellal as the main ingredient) and wormwood (thujone and camphor as
109
the main ingredients), were purchased from Shanghai Moellhausen International
110
Trade Co., Ltd. Spices were irradiated at 42 kGy for 6 h to ensure their sterility. After
111
distillation and extraction, the concentration of refining essential oil was 0.88 g/ mL.
112
Tween-80 was used as an emulsifier, and the five types of essential oils were all
113
thoroughly mixed uniformly in a ratio of oil: tween-80: sterile distilled water 4:1:5,
114
respectively. Essential oil suspensions of different concentrations (Table 1) were
115
prepared according to analysis of the minimum inhibitory concentrations (MIC)
116
results.
117
118
Insert Table. 1 goes here 2.2 Strains and media
119
In this paper, through sampling of 249 samples in Henan Province (roast chicken,
120
marinated chicken legs, salt pheasant, etc.), the C. perfringens was screened in 42
121
samples (Table.S1). Moreover, most of them were identified as C. perfringens C1,
122
which indicating the representive and universal of C1. Strain of C. perfringens ATCC
123
1312 was from China general Microbiological cultural collection center (CGMCC).
124
The C1 strain of C. perfringens was isolated and identified from vacuum- packed bag
125
of salted chicken in the microbiology laboratory of the College of Food Science and
126
Technology of Henan Agricultural University.
127
Culture media used included the following: tryptose sulfite cycloserine (TSC)
128
agar, fluid thioglycollate (FTG) medium, sulfite polymyxin sulfadiazine (SPS) agar
129
medium (Beijing Luqiao Biotechnology) and tryptic soy agar (TSA) medium (Beijing
130
Aobox Biotechnology).
6
131
2.3 Antibacterial activity assay
132
The Oxford cup method was used to determine the antibacterial activity of the
133
essential oils (Vasilijevic and Mitic-Culafic et al., 2019). A sample (200 µL) of the
134
initial bacterial suspension was pipetted onto TSA medium and streaked evenly three
135
times with an applicator. Oxford cups of a uniform size were placed on bacterial agar
136
plates using sterile tweezers, followed by the addition of 100 µL essential oil
137
suspension (352 mg/mL). After standing for 5 to 10 min, the plates were
138
anaerobically cultured at a constant temperature (37 °C) for 24 h. Three parallel
139
assays and a blank control were performed. During the incubation, the size of the
140
inhibition zones was observed, photographed and recorded. The antibacterial activity
141
of different essential oils on C. perfringens were determined by measuring the
142
diameter of inhibition zones (criteria for rating the antibacterial effect: inhibition zone
143
diameter > 20 mm was extremely sensitive, 15 ~20 mm was highly sensitive, 10~15
144
mm was moderately sensitive, 7~9 mm was slightly sensitive, and absence of
145
inhibition zone was insensitive).
146
2.4 Determination of the minimum inhibitory concentrations (MIC)
147
The two-fold dilution method was used to determine the MIC of spice essential
148
oils on C. perfringens. The initial concentration of essential oil was 352 mg/mL.
149
Firstly, 1 mL of FTG liquid medium was added to 15 test tubes labeled 1 to 15
150
respectively. Then, 1 mL of essential oil was added to tube 1 and thoroughly mixed,
151
and 1 mL of the mixture from tube 1 was transferred to tube 2. In the same way, serial
152
dilutions were performed for tubes 1 to 12. Tube 13 was the positive control. As a
153
blank control for determining whether the spice essential oil was contaminated, 1 mL
154
of the mixture was pipetted from tube 12 to tube 14. Tube 15 was the Tween-80 7
155
control. Afterwards, a 100 µL sample of the initial bacterial culture was added to test
156
tubes 1-13 and 15, respectively, followed by thorough mixing. The test tubes were
157
placed in an anaerobic environment and cultured at 37 °C for 18-24 h. Thereafter, a
158
200 µL sample from each tube was evenly streaked onto TSA media, and the plates
159
were anaerobically cultured at 37 °C for 18-24 h followed by observation of results. C.
160
perfringens appeared as white colonies with irregular edges on the TSA plates, and
161
the highest essential oil concentration that corresponded to absence of bacterial
162
growth on plates was the MIC. All the experiments were repeated three times.
163
2.5 Meat preparation
164
Chicken breast was purchased at local supermarkets and held at 0-4 °C until used.
165
A sample was used to check for the absence of natural contamination by C.
166
perfringens. After adding 3500 mL of water, each 1.5 kg of chicken breast was
167
simmered at high heat (1600 W) for 5 min and at low heat (800 W) for 90 min. Then,
168
the cooked chicken breast was pulled into strips according to the texture, sterilized by
169
steam for 15 min, and dispensed (60 g) into bags for use. Samples of the essential oil
170
and the bacterial suspension were added, followed by homogenization for 120 s in a
171
homogenizer to allow extensive contact between the essential oil, bacterial suspension
172
and the chicken breasts. Finally, the mixture was clamped with a sealing clip, placed
173
in an anaerobic workstation, and cultured at 37 °C for 18 h.
174
2.6 Bacterial counting
175
The samples were diluted with 0.1% peptone water at 1:100 (g/mL), and then
176
200 µL of samples were added to the TSC plates for microbial testing. Three parallel
177
tests were performed for each gradient. After streaking and anaerobic incubation at
178
37 °C for 18 h, colonies of C. perfringens on the plates were inspected and counted. 8
179
2.7 C. perfringens survival kinetics model
180
C. perfringens survival kinetics model was constructed using Log N or Ln (N/N0)
181
as the y-axis and the essential oil concentration as the x-axis, which reflecting the
182
relationship between the surviving counts of C. perfringens and the essential oil
183
concentration. The survival curve of C. perfringens under different essential oil
184
concentrations was fitted according to the parameter-adjusted Gompertz model, and
185
LS-SVM model. The fitted results of the models were compared and the inactivation
186
kinetics model of C. perfringens with different essential oil concentrations was
187
established.
188
(1) Construction of the parameter-adjusted Gompertz model
189
i Gompertz model:
190
Y = N + A ∗ exp{− exp[((B ∗ 2.718)/ a ∗ (C − X ) + 1]}
191
In Eq (1), X is the essential oil concentration (mg/mL); Y is the C. perfringens
192
counts (CFU/g) when the essential oil concentration was X mg/mL; A is the relative
193
maximum inhibition rate; B is the essential oil concentration required to achieve the
194
relative maximum inhibition rate (mg/mL); C is the essential oil concentration when
195
the tangent line intersects the X-axis at the maximum inhibition rate; a is the constant;
196
N0 is the initial value of C. perfringens (CFU/g).
197
ii Expression of the parameter-adjusted Gompertz model:
198
log
N = C exp( − exp( A + B t )) − C exp( − exp( A)) N0
9
(1)
(2)
199
Where N0 is the initial value of C. perfringens (CFU/g); N is the C. perfringens
200
counts (CFU/g) when the essential oil concentration was X mg/mL; A, B and C are as
201
described of above; t is the constant.
202
(2) LS-SVM model construction and algorithm optimization
203
LS-SVM is a data analysis method that requires a small number of data set for
204
training and validation. In order to build an LS-SVM model, the first step was to
205
import training data. The second step was to enter the commands type = ‘function
206
estimation’ and kernel = ‘RBF_kernel.’ Here, the Gaussian radial basis function (RBF)
207
is a nonlinear kernel widely used in nonlinear problems as a kernel function to reduce
208
computational complexity of training procedures. The third step was the initialization
209
of the model by ‘initlssvm’ function. The details of them can be found in literature.
210
For a LS-SVM classifier, in the primal space it takes the form:
211
y ( x) = sign ( wT x + b )
212
where b is a real constant. For nonlinear classification, the LS-SVM classifier in
213
the dual space takes the form:
214
N y ( x) = sign ∑ ak yk ( x, xk ) + b k =1
(3)
(4)
215
where ak are positive real constants and b is a real constant, in general, K (x, xi)=
216
〈ϕ(x), ϕ(xi) 〉 , and ϕ(x) is the nonlinear map from original space to the high-
217
dimensional space. For function estimation, the LS-SVM model takes the form. When
218
radial basic function (RBF) kernels are used, two tuning parameters (γ, σ2) are added.
219
The regularization parameter gam (γ) determines the tradeoff between minimizing the
220
training error and minimizing model complexity. The parameter sig2 (σ2) is the 10
221
bandwidth and implicitly defines the nonlinear mapping from input space to some
222
high dimensional feature space. Kernel function can map the data which are not
223
linearly separable in input space into a higher dimensional feature space, where
224
becomes linearly separable.
225
2.8 Reliability evaluation of C. perfringens survival kinetics model
226
By comparing the correlation coefficient (R2) (Eq (5)) and the root mean square
227
error of prediction (RMSEP) (Eq (6)) values from different models, the optimal
228
prediction model for an inactivation kinetics model of C. perfringens in cooked
229
chicken breast with different essential oil concentrations was identified. Accuracy
230
factor (Af) and bias factor (Bf) were calculated and examined to compare the
231
parameter-adjusted Gompertz model, and LS-SVM model against the experimental
232
data. Among them, Af was used to check the fluctuation range of the predicted value.
233
Bf was used to check the difference between predicted and measured values. The
234
specific calculation formulas were shown in Eq (7) and (8), respectively.
∑ =1− ∑
n
235
236
R
2
i =1 n
RMSEP =
( y′i − yi ) 2
(5)
( y′i − y ) 2 i =1
∑
n i =1
( yi − yi′)
(6)
n
237
A f = 10[∑ N pre − N obs ] / m
(7)
238
B f = 10[∑ N pre − N obs ] / m
(8)
239 240
In Eq (5) and (6), y
' i
is the predicted value of the ith sample; y is the measured i
value of the ith sample; n is the number of samples. In Eq (7) and (8), Npre is the 11
241
predicted value of the microbiological counts. Nobs is the measured value of the
242
microbiological counts. m is the time of the test.
243
2.9 Data analysis
244
The MATLAB 2016 software was used to perform calculations of the parameter-
245
adjusted Gompertz model and LS-SVM model, and to fit the survival curves of C.
246
perfringens against different essential oils in cooked chicken breast.
247
3. Results and Discussion
248
3.1 The antibacterial activity of spice essential oils
249
The antibacterial activity of different essential oils on two C. perfringens strains
250
were determined using the Oxford Cup method. It could be seen that the five different
251
spice essential oils had good bacteriostatic effects on the ATCC 13124. The addition
252
of cinnamon and wormwood resulted in the strongest inhibitory effects (large
253
inhibition zones > 20 mm); while the antibacterial effects of black pepper, fennel and
254
ginger essential oils on C. perfringens were weaker (the inhibition zones 10~15 mm).
255
In comparison, for the isolated strain C1, cinnamon, wormwood and black pepper
256
essential oils showed strong inhibitory effects (the inhibition zones 15~20 mm), while
257
fennel and ginger essential oils demonstrated weaker inhibitory effects (the inhibition
258
zones 10~15 mm). These results indicated that the antibacterial activities on C.
259
perfringens were in the order of cinnamon, wormwood, black pepper, fennel and
260
ginger essential oil, which were from strong to weak (Table. 2). It was in accordance
261
with the observation that spices oil show varying levels of antimicrobial activity.
262
263
Insert Table. 2 goes here 3.2 The MIC of different spice essential oils on C. perfringens 12
264
Based on the results presented in section 3.1, the MIC of different essential oils
265
on C. perfringens were determined. For the strain ATCC 13124, the concentration of
266
cinnamon essential oil was determined in test tubes 1~12, respectively. The solution
267
of the 14th test tube (control group) was clarified, which showed the cinnamon
268
essential oil was not contaminated. However, the solution of test tubes 15 (Tween-80
269
control group) and 13 (positive growth control group) were feculent, which showed
270
the growth of C. perfringen. In addition, the solution of Tween-80 did not
271
demonstrate inhibiting effect. In order to gain further insight into the results, the
272
growth of C. perfringens in different concentrations of essential oils were determined
273
by plate cultural methods (Table. 3). The results indicated that cinnamon essential oil
274
exhibited the strongest inhibitory activity against the standard strain ATCC 13124,
275
with a MIC of 2.75 mg/mL. And the MIC of the spice essential oils on the strain
276
ATCC 13124 showed the following increasing order: cinnamon essential oil < black
277
pepper essential oil < wormwood essential oil < fennel essential oil < ginger essential
278
oil. In comparison, the MIC against strain C1 showed the following increasing order:
279
cinnamon essential oil < wormwood essential oil < black pepper essential oil < fennel
280
essential oil < ginger essential oil. The results indicated that cinnamon essential oil
281
exhibited the strongest inhibitory activity against both the two strains of C.
282
perfringens, with the MIC of 2.75 mg/mL and 5.5 mg/mL, respectively. And the
283
antibacterial activities of black pepper and wormwood essential oils were relatively
284
lower.
285
286
Inset Table. 3 goes here 3.3 Construction of parameter-adjusted Gompertz model
13
287
As the concentration of different spice essential oils increased, the surviving
288
count of C. perfringens decreased, and when the concentration increased to a certain
289
value, growth of C. perfringens ceased. To rapidly obtain the survival state of C.
290
perfringens at different concentrations, the parameter-adjusted Gompertz model and
291
LS-SVM model were used for nonlinear fitting to plot the survival curve of C.
292
perfringens in cooked chicken breast. According to the above results, cinnamon, black
293
pepper and wormwood essential oils, which had obvious inhibitory effects, were
294
selected and constructed the survival prediction model of C. perfringens.
295
The best fitted equations with the parameter-adjusted Gompertz model were
296
shown in Table. 4, and the survival curves of C. perfringens at different essential oil
297
concentrations were obtained (Fig. 1). For the strain ATCC 13124 (Fig. 1 a ~ c), with
298
the concentration of cinnamon essential oil increasing, the C. perfringens declined
299
reached the maximum at 5.5 mg/mL. Then the number of C. perfringens tends to
300
stabilize. The prediction curve of black pepper was similar with the decline of C.
301
perfringens reaching to the maximum at 42 mg/mL. While for the wormwood
302
essential oil, the declining of C. perfringens reached the maximum at 30 mg/mL to
303
44 mg/mL. The results showed that the prediction model with cinnamon essential oil
304
was the optimum model, and R2 and RMSEP values were 0.8912 and 0.3041 log
305
CFU/g, respectively.
306
As to the C1, as the concentration of different spice essential oils increased, the
307
surviving count of C. perfringens gradually decreased (Fig. 1 d ~ f). It was
308
noteworthy that the trend of prediction curve decline was obvious for cinnamon
309
essential oil. This indicated that cinnamon essential oil exerted the most pronounced
310
inhibitory effect on C. perfringens proliferation. And it was shown that the prediction 14
311
model with added cinnamon essential oil was the optimum model, and R2 and RMSEP
312
values were 0.8979, 0.2918 log CFU/g, respectively (Table. 5).
313
Fig. 1
314
Inset Table. 4 goes here
315
3.4 Construction of LS-SVM prediction model
316
Before implementing LS-SVM, the tuning parameters must be tuned. In this
317
paper, the tuning parameters are regularization parameter and decision variable
318
related to RBF (γ and σ2). These starting values are then passed to another
319
optimization algorithm which is the simplex method to perform a fine-tuning step.
320
Then the model performance is estimated using leave-one-out crossvalidation. The
321
tuning process must be repeated for each case. After finding the suitable values of
322
tuning parameters, the model parameters (γ and σ2) can be obtained. The values of γ
323
and σ2 calculated were 19611.28 and 0.062, respectively.
324
Prediction for the survival of C. perfringens at different essential oil
325
concentrations using the LS-SVM model were shown in Fig. 2. For the ATCC 13124,
326
when the concentration of essential oil was ≤ 5.5 mg/mL, with three essential oil
327
concentration increasing, the number of C. perfringens decreased. Among them, the
328
rate of descent was the biggest with the cinnamon essential oil. When the cinnamon
329
essential oil concentration was ≥ 15 mg/mL, the curve tended to stabilize for the C1.
330
For ATCC 13124, When the cinnamon essential oil concentration was ≥ 10 mg/mL,
331
the curve tended to stabilize. The minimum surviving counts of the ATCC 13124 and
332
the C1 were 4.13 lg(CFU)/g and 4.05 lg(CFU)/g under wormwood essential oil,
333
respectively. The results showed that for each concentration range, the survival of C. 15
334
perfringens with cinnamon essential oil was lower than that with the other two
335
essential oils. What is more, for the cinnamon essential oil, the R2 values of prediction
336
model were 0.9975 and 0.9802, respectively. The RMSEP were 0.2854 lg(CFU)/g,
337
0.2619 lg(CFU)/g, respectively.
338
339
Fig. 2 3.5 Comparison of statistical indices
340
The parameter-adjusted Gompertz mode, and LS-SVM model were compared
341
against the experiment data. In general, the calculated values of the Bf for LS-SVM
342
model close to 1 indicated no systematic bias. The calculated performance indices (R2,
343
RMSEP, Af and Bf) were shown in Table. 5. It could be seen that the prediction
344
accuracy of the LS-SVM model was greater than 0.99, and the degree of fitting was
345
highest. For the ATCC 13124 and the C1, the optimum prediction models based on
346
LS-SVM were cinnamon essential oil. For the ATCC 13124, the Af and Bf were
347
1.0101, 1.001, respectively. For the C1, the Af and Bf were 1.0122, 1.0002,
348
respectively. Therefore, it was inferred that cinnamon essential oil exhibited the
349
lowest inhibitory concentration and the strongest antibacterial activity on C.
350
perfringens. And also all the evaluation results of the prediction model constructed by
351
LS-SVM were best. Hence, in this study, LS-SVM model was selected to predict the
352
numbers of C. perfringens under different essential oil concentrations.
353
354
Inset Table. 5 goes here
4.Conclusion
355
This study confirmed the potential inhibition of spice essential oils on the C.
356
perfringens in cooked chicken breast. It was shown that cinnamon essential oil had 16
357
the lowest inhibitory concentration and the strongest inhibitory effects, followed by
358
wormwood essential oil, while ginger essential oil displayed the weakest. In addition,
359
through evaluation and comparison, it was shown that LS-SVM model could
360
accurately predict the C. perfringens survival against essential oils. The R2 values of
361
prediction model were 0.9975 and 0.9802, respectively. The RMSEP were 0.2854
362
lg(CFU)/g, 0.2619 lg(CFU)/g, respectively. This work demonstrates the feasibility of
363
rapid prediction on the growth of C. perfringens against spice essential oils, and it was
364
expected to provide some technical support for the control of C. perfringens during
365
the manufacture and processing of meat products.
366
Acknowledgment
367
This work was financially supported by the National Natural Science Fundation
368
of China (31571856, 31801671), the Henan Science and Technology Major Project
369
(NO. 161100110800), Science and technology innovation talent support program of
370
Henan province (NO. 18HASTIT036), Henan Key Laboratory of Cold Chain Food
371
Quality and Safety Control (CCFQ2018-YB-13), and Innovation Fund of doctoral
372
scientific research of Henan agricultural university (NO. 30601677).
373 374
References
375
Acheson, P. and V. Bell, et al. (2016). "Enforcement of science-using a Clostridium
376
perfringens outbreak investigation to take legal action." Journal of Public Health,
377
38(3): 511-515.
378
Akkermans, S. and P. Nimmegeers, et al. (2018). "A tutorial on uncertainty
379
propagation techniques for predictive microbiology models: A critical analysis of
380
state-of-the-art techniques." International Journal of Food Microbiology, 282 (6): 1-8. 17
381
Alnoman, M. and P. Udompijitkul, et al. (2017). "Chitosan inhibits enterotoxigenic
382
Clostridium perfringens type A in growth medium and chicken meat." Food
383
Microbiology, 64: 15-22.
384
Anuragi, A. and D. S. Sisodia(2019). "Alcohol use disorder detection using EEG
385
Signal features and flexible analytical wavelet transform." Biomedical Signal
386
Processing and Control, 52: 384-393.
387
Chan, C. and R. Gan, et al. (2018). "Polyphenols from selected dietary spices and
388
medicinal herbs differentially affect common food-borne pathogenic bacteria and
389
lactic acid bacteria." Food Control, 92 (17): 437-443.
390
Costa, M. M. and S. Bounatirou, et al. (2006). "Essential oils from anethum
391
graveolens, Levisticum officinale and pimpinella anisum hairy root cultures:
392
Composition, antibacterial and antioxidant activities." Planta Medica, 72 (11): 1065-
393
1065.
394
Dorman, H. and S. G. Deans(2004). "Chemical composition, antimicrobial and in
395
vitro antioxidant properties of Monarda citriodora var. citriodora, Myristica fragrans,
396
Origanum vulgare ssp hirtum, Pelargonium sp and Thymus zygis oils." Journal of
397
Eessential oil Research, 16(2): 145-150.
398
Gonzalez, S. C. and A. Possas, et al. (2018). "'MicroHibro': A software tool for
399
predictive microbiology and microbial risk assessment in foods." International journal
400
of food microbiology, 290: 226-236.
401
Grass, J. E. and L. H. Gould, et al. (2013). "Epidemiology of Foodborne Disease
402
Outbreaks Caused by Clostridium perfringens, United States, 1998-2010." Foodborne
403
Pathogens and Disease, 10(2): 131-136.
404
Koziel, N. and E. Kukier, et al. (2019). "Clostridium perfringens - epidemiological
405
importance and diagnostics." Medycyna Weterynaryjna-Veterinary Medicine-Science 18
406
and Practice, 75(5): 265-270.
407
Li, C. and H. S. Lillehoj, et al. (2017). "Characterization of Clostridium perfringens
408
Strains Isolated from Healthy and Necrotic Enteritis-Afflicted Broiler Chickens."
409
Avian Diseases, 61(2): 178-185.
410
Lindstrom, M. and A. Heikinheimo, et al. (2011). "Novel insights into the
411
epidemiology of Clostridium perfringens type A food poisoning." Food Microbiology,
412
28(2SI): 192-198.
413
London, A. E. and J. A. Payne, et al. (2017). "Outbreak Caused by Clostridium
414
perfringens Infection and Intoxication at a County Correctional Facility." Journal of
415
Environmental Health, 80(1): 8-13.
416
Makino, A. and J. Xu, et al. (2019). "Detection of Clostridium perfringens in tsunami
417
deposits after the Great East Japan Earthquake." Microbiology and immunology, 43
418
(17):125-131.
419
May, F. J. and B. G. Polkinghorne, et al. (2016). "Epidemiology of bacterial toxin-
420
mediated foodborne gastroenteritis outbreaks in Australia, 2001 to 2013."
421
Communicable diseases intelligence quarterly report, 40(4): E460-E469.
422
Renyong, G. and F. Weichang, et al. (2008). "Research on the antibacterial and
423
antioxidant effects of cinnamon essential oil." Food research and development, 24
424
(10): 29-32.
425
Skandamis, P. N. and G. Nychas(2002). "Preservation of fresh meat with active and
426
modified atmosphere packaging conditions." International Journal of Food
427
Microbiology, 79(1-2SI): 35-45.
428
Vapnik, V.(1998). The support vector method of function estimation, Boston: Kluwer
429
Academic Publishers.
430
Vasilijevic, B. and D. Mitic-Culafic, et al. (2019). "Antibacterial effect of Juniperus 19
431
communis and Satureja montana essential oils against Listeria monocytogenes in vitro
432
and in wine marinated beef." Food Control, 100: 247-256.
433
Wang, H. T. and R. Y. Zhang, et al. (2019). "Measurement of SSC in processing
434
tomatoes (Lycopersicon esculentum Mill.) by applying Vis-NIR hyperspectral
435
transmittance imaging and multi-parameter compensation models." Journal of Food
436
Process Engineering, 42(5):412-419.
437
Zwietering, M. H. and I. Jongenburger, et al. (1990). "Modeling of the bacterial
438
growth curve." Applied and environmental microbiology, 56(6): 1875-81.
20
TABLE 1 The concentration of essential oil added to cooked chicken breast for C. perfringens Essential oil type Cinnamon
Wormwood
Black pepper
Fennel
Ginger
Essential oil suspensions of different concentrations (mg/mL) ATCC 13124 0.6875, 1.375, 2.75, 5.5, 11, 15, 19 ,22, 44 5.5, 11, 22, 30, 38, 44, 55, 66, 77, 88, 176 2.75, 5.5, 11, 22, 44, 55, 66, 77, 88 5.5, 11, 22, 30, 38, 44, 55, 66, 77, 88, 176
C1 1.375, 2.75, 5.5, 11, 22, 28, 34, 40, 44
2.75, 5.5, 11, 22, 30, 38, 44, 88
5.5, 11, 22, 44, 55, 66, 77, 88
2.75, 5.5, 11, 22, 44, 55, 66, 77, 88, 176
11, 22, 44, 88, 110, 132,
5.5, 11, 22, 44, 88, 110, 132, 154, 176,
154, 176, 352, 704
220, 264, 308, 352
TABLE 2. The inhibitory zone diameters of different essential oils on C. perfringens The inhibition zone diameters (mm) Strains type
cinnamon
wormwood
black pepper
fennel
ginger
ATCC13124
31.00
22.81
20.99
15.82
13.65
C1
21.07
19.71
24.25
16.33
15.54
Note: The criteria for rating the antibacterial effect: inhibition zone diameter > 20 mm was rated extremely sensitive, 15 ~20 mm was rated highly sensitive, 10~15 mm was rated moderately sensitive, 7~9 mm was rated slightly sensitive, and absence of inhibition zone was rated insensitive.
TABLE 3. The MIC of different essential oils on C. perfringens MIC of different essential oils on C. perfringens (mg/mL) Spices essential oil
ATCC 13124
C1
Cinnamon
2.75
5.5
Wormwood
22
11
Black pepper
11
11
Fennel
22
11
Ginger
88
22
TABLE 4. The best fitted equation of the parameter-adjusted Gompertz model for different C. perfringens strains Essential
ATCC 13124
C1
f(x) = 8.0414+2663.0155*exp(-
f(x)= 8.4695+1623.7487*exp(-
exp(2.1522+0.0416*x))-
exp(2.0607+0.0099*x))-
2663.0155*exp(-exp(2.1522))
1623.7487*exp(-exp(2.0607))
f(x) = 7.8842+1643.4920*exp(-
f(x) = 8.4728+30.9437*exp(-exp(-
exp(2.0143+0.0046*x))-
2.0501+0.0030*x))- 30.9437*exp(-
1643.4920*exp(-exp(2.0143))
exp(-2.0501))
f(x) = 8.1903+44.6171*exp(-exp(-
f(x) = 9.7731+172.7651*exp(-
oil type
Cinnamon
Black pepper
Wormwood 5.8174+0.0424*x))- 44.6171*exp(exp(-5.8174))
exp(1.3339+0.0020*x))172.7651*exp(-exp(1.3339))
Table 5. The evaluation indices of prediction model of C. perfringens under different essential oils
Model evaluation parameters
C1
cinnamon
black pepper
wormwood
cinnamon
black pepper
wormwood
0.8912
0.8663
0.8216
0.8979
0.8568
0.8763
0.3041
0.3675
0.4932
0.2918
0.3726
0.3528
Af
1.0147
1.0156
1.0306
1.0261
1.0215
1.0215
Bf
1.0002
1.0011
1.0006
1.0010
1.0003
1.0003
R2
0.9975
0.9765
0.9389
0.9802
0.9482
0.9734
0.2854
0.3223
0.4037
0.2619
0.3368
0.4536
Af
1.0101
1.0111
1.0101
1.0122
1.0134
1.0322
Bf
1.0001
1.0001
1.0002
1.0002
1.0003
1.0004
R2 parameter-adjusted
ATCC 13124
RMSEP (lg(CFU)/g)
Gompertz model
RMSEP LS-SVM model
(lg(CFU)/g)
+ +
7 Number of C.perfringens (lgCFU·g-1)
Number of C.perfringens (lgCFU·g-1)
7.5 7
6.5
+ +
6
+
5.5
+
5
+ 5
10
+
+
5.5
20
40
10
20
30
Essential oil concentration (mg·mL ) d. Cinnamon essential oil -1
5
+ +
40
20
60
Essential oil concentration (mg·mL-1) c.Wormwood essential oil
++
7
+ 6
+ 5
+ +
98
+ +
7
6
+
5
4
+
3
3 0
+
10
+
4.5
6
0
60
+++
4 5
+
Essential oil concentration (mg·mL ) b.Black pepper essential oil
Number of C.perfringens (lgCFU·g-1)
6.5
+
7
-1
89
+
+
3 0
+ ++ +
+
4
3
15
Essential oil concentration (mg·mL ) a.Cinnamon essential oil
Number of C.perfringens (lgCFU·g-1)
++
6
-1
6
8
+ 0
7
+
+
4.5
7.5 8
9
4
5
8.5
+
Number of C.perfringens (lgCFU·g-1)
8
8
Number of C.perfringens (lgCFU·g-1)
8.5
0
50
100
Essential oil concentration (mg·mL ) e.Black pepper essential oil -1
0
10
20
30
Essential oil concentration (mg·mL-1) f.Wormwood essential oil
Fig.1 Effect of different concentrations of essential oils on the growth of C.perfringens (ATCC13124, C1)
8.5
10
9
-ATCC13124 predicted value -C1 predicted value
8
9
-Measured value
8
7.5
7
6.5
6
5.5
Number of C.perfringens (lgCFU·g-1)
Number of C.perfringens (lgCFU·g-1)
Number of C.perfringens (lgCFU·g-1)
8
7
6
5
3
4.5 0
10
20
Essential oil concentration (mg·mL-1) a.Cinnamon essential oil
6
5
4
4
5
7
3 0
20
40
Essential oil concentration (mg·mL-1) b.Wormwood essential oil
0
20
40
60
80
Essential oil concentration (mg·mL-1) c.Black pepper essential oil
Fig. 2 The correlation of measured values and predicted values by LS-SVM method under different essential oil concentration
Highlights:
The effect of spices essential oils on the survival of C. perfringens were studied; The in vitro inhibitory activities and MIC of spices essential oils were studied; The cinnamon essential oil exhibited a notable inhibitory effect on C. perfringens; The rapidly prediction model of the survival of C. perfringens is proposed; LS-SVM is the optimal model for predicting the survival of C. perfringens