Shelf-life charts of beef according to level of bacterial contamination and storage temperature

LWT - Food Science and Technology 81 (2017) 50e57

Contents lists available at ScienceDirect

LWT - Food Science and Technology journal homepage: www.elsevier.com/locate/lwt

Shelf-life charts of beef according to level of bacterial contamination and storage temperature Changhoon Chai a, So-Yeon Lee b, Se-Wook Oh b, * a b

Division of Applied Animal Science, Kangwon National University, Chuncheon 24341, Republic of Korea Department of Food and Nutrition, Kookmin University, Seoul 02707, Republic of Korea

a r t i c l e i n f o

a b s t r a c t

Article history: Received 28 September 2016 Received in revised form 1 February 2017 Accepted 12 March 2017 Available online 16 March 2017

Shelf-life charts of beef according to bacterial contamination and storage temperature were developed via mathematical regressions of the growth curves of spoilage bacteria in beef. Additionally, bacterial contamination in retail beef specimens was investigated using culture-based bacteriological tests, PCR assays, and pyrosequencing analysis. Pathogenic bacteria were infrequently detected in the 100 retail beef specimens, but spoilage bacteria were found in all of the specimens. The populations of spoilage bacteria in retail beef specimens varied from 2.27 to 6.15 log CFU$g
1. Pyrosequencing analysis of retail beef specimens suggested contamination from multiple sources, since both Lactobacillales and Pseudomonadales, which are prevalent in bovine intestine and feces and in bovine hides and beef-processing facilities, respectively, were detected. Growth of spoilage bacteria in beef at 5e25  C was predicted via mathematical characterization of experimentally determined growth curves at 5, 10, 15, 20, and 25  C using a modified-Gompertz function and subsequent nonlinear regressions of the growth characteristics, and then transformed to shelf-life charts according to the level of bacterial contamination and storage temperature. The bacterial contamination and bacteriological standards used to generate the shelf-life charts were 0, 1, 2, 3, 4, and 5 log CFU$g
1 and 5, 6, 7, and 8 log CFU$g
1, respectively. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Shelf-life Beef Mathematical modeling Bacterial contamination Bacterial distribution

1. Introduction Bacterial contamination of beef is an inevitable result of beef processing and handling. Contamination of beef occurs throughout slaughtering, boning and cutting, comminuting, packaging, and storage. Bacterial contaminants of beef are diverse and include nonpathogenic and pathogenic taxa. Beef is occasionally contaminated with pathogenic bacteria, such as Escherichia coli O157:H7, Salmonella spp., and Listeria monocytogenes (Rivera-Betancourt et al., 2004). However, most bacterial contaminants associated with beef spoilage are non-pathogenic taxa (Beuchat, 1996). The shelf-life of beef is the time that passes before it becomes unacceptable for consumption and distribution due to the growth of spoilage bacteria. Beef harboring > 7e8 log CFU$g
1 spoilage bacteria is unacceptable for human consumption, both microbiologically and organoleptically (Nollet, 2012). Beef with > 7 log CFU$g
1 bacteria is unacceptable for international trade (ICMSF, 1986). According to the EC regulations (European Commission,

* Corresponding authors. E-mail address: [email protected] (S.-W. Oh). http://dx.doi.org/10.1016/j.lwt.2017.03.023 0023-6438/© 2017 Elsevier Ltd. All rights reserved.

2005), the limit for spoilage bacteria in minced meat immediately after slaughtering is 5  106 CFU g
1. Beef producers should set the shelf-lives of their products according to their intended use to manage the microbiological safety of products and prevent rejection in international trade. However, variations in bacterial growth according to storage conditions and the level of bacterial contamination hamper estimation of the shelf-life of beef (Casaburi, Piombino, Nychas, Villani, & Ercolini, 2015). Bacterial growth follows a sigmoidal curve and can be divided into the following four phases: the lag, exponential, stationary, and death phases. The growth curve of bacteria can be characterized mathematically. The lag time (l), specific growth rate (m), initial population (N0), and maximum population (Nmax) can be calculated via regression of the curve to a mathematical model, such as a modified-Gompertz function (Shimoni & Labuza, 2000). The l, m, and Nmax at a temperature (T) can be predicted via regression analyses of plots of l, m, and Nmax versus T to appropriate nonlinear mathematical functions (Bruckner, Albrecht, Petersen, & Kreyenschmidt, 2013). Growth curves with different N0 values versus storage times (t) can be replotted by application of predicted values of N0, l, m, and Nmax to the equation (e.g., a modifiedGompertz function) used for mathematical characterization of the

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

growth curve, which enables estimation of the time required for a bacterial population to reach a certain size (Bruckner et al., 2013; Oscar, 2005). Therefore, growth curves of spoilage bacteria in beef at different T values facilitate estimation of the time required for spoilage bacteria in beef with different N0 and T values to reach the bacteriological standard. Replotted growth curves of spoilage bacteria in beef can be transformed to shelf-life curves according to N0 and T. Such beef shelf-life charts enable estimation of the shelf-lives of beef products. This study aimed to develop shelf-life charts of beef for estimation of beef shelf-life at different N0 and T values. To develop beef shelf-life charts, the growth characteristics (l, m, and Nmax) of spoilage bacteria in beef at 5, 10, 15, 20, and 25  C were calculated via regressions using a modified-Gompertz function. The l, m, and Nmax values were predicted via nonlinear regressions using exponential decay, exponential growth, and logistic functions, and growth curves of spoilage bacteria in beef at different N0 and T values were replotted by applying the predicted growth characteristics to a modified-Gompertz function. The replotted curves were transformed to charts of the time required for populations of spoilage bacteria in beef to reach 5, 6, 7, and 8 log CFU$g
1. Additionally, bacterial contamination of retail beef specimens was investigated because it influences the bacteriological safety and shelf-life of beef. Total aerobic bacteria, coliform bacteria, and E. coli were enumerated in 100 retail beef specimens. The presence of E. coli O157:H7, L. monocytogenes, and Salmonella spp. in the beef specimens was also evaluated. To detect bacterial taxa associated with beef spoilage, the bacterial populations in retail beef specimens were subjected to pyrosequencing analysis.

2. Materials and methods 2.1. Mathematical predictions of the time required for spoilage bacteria in beef to reach bacteriological standards Fresh beef purchased from a local market was divided into 10-g specimens. The specimens were stored in aseptic bags under aerobic conditions at 5, 10, 15, 20, and 25  C. Triplicate specimens were sampled at appropriate time intervals and homogenized in 90 mL of saline solution. Homogenates were plated on Petrifilm aerobic count plates (PAC; 3M Com., St. Paul, MN, USA), and incubated at 35  C for 24e48 h; colonies on PAC were then enumerated (KFDA., 2013; Linton, Eisel, & Muriana, 1997). Growth curves of spoilage bacteria in beef at 5, 10, 15, 20, and 25  C were regressed to a modified-Gompertz function (Eq. (1)) using the GraphPad Prism software v. 4.03 (GraphPad Software, San Diego, CA, USA), which enables calculation of l, m, N0, and Nmax values. The l, m, and Nmax values at 5, 10, 15, 20, and 25  C obtained from the regression were replotted versus T, and regressed with exponential decay (Eq. (2)), exponential growth (Eq. (3)), and logistic (Eq. (4)) functions using the SigmaPlot software v. 12.0 (Systat Software Inc., Richmond, CA, USA).

N ¼ A$expð
expððm$e$ðl
tÞ=A þ 1ÞÞÞ

(1)

where N is the logarithm of a bacterial population at time (t) and A is Nmax/N0.

l ¼ l0 þ al $expð
bl $TÞ

(2)

where l0, al, and bl are the regression constants.




m ¼ m þ am $ exp bm $T



where m0, am, and bm are the regression constants.

(3)

Nmax ¼ Nmax 0 þ am

.h

51

1 þ ðT=T0 Þbm

i

(4)

where Nmax 0 , am, and bm are the regression constants. The l, m, and Nmax values at T of 5e25  C were computed according to the regression equations and applied to a modifiedGompertz function to estimate the time ðtE NI Þ required for spoilage bacteria in beef to reach the bacteriological standards of interest (NI; NI ¼ 5, 6, 7, and 8 log CFU$g
1), if N0 was 0, 1, 2, 3, 4, or 5 log CFU$g
1. Based on this, the tE NI versus T according to NI and N0 were plotted to generate beef shelf-life charts. To improve their readability, the scales of the left and right y-axes of the shelf-life charts were in days and hours, respectively.

2.2. Bacterial contamination of retail beef Dices of fresh beef (approximately 1  1  1 cm), delivered to retail markets on the day of or the day before collection, were aseptically collected from six randomly selected retail markets in northern Seoul, South Korea, in June 2014. A total of 100 beef specimens were collected for bacteriological tests. Among these 100 specimens, 66 were domestic and 34 were imported from Australia. Immediately after being transported to the laboratory, the specimens were divided into four 10-g pieces and transferred to aseptic plastic bags. Ninety milliliters of saline solution was added to one of the four 10-g specimens in aseptic plastic bags and homogenized. The homogenates were diluted and plated on a PAC plate, Petrifilm coliform count plate (PCC; 3M Com), and Petrifilm E. coli/coliform plate (PEC; 3M Com.) for enumeration of total aerobic bacteria, coliform bacteria, and E. coli, respectively. Colonies of total aerobic bacteria, coliform bacteria, and E. coli formed on PAC, PCC, and PEC after incubation at 35  C for 24e48 h were enumerated as described in the Korea Food Standard Codex (KFDA., 2013). To detect E. coli O157:H7, 90 mL of modified tryptone soya broth (mTSB; Oxoid Ltd.) containing novobiocin (10 mg L
1; SigmaAldrich Co., St. Louis, MO, USA) (mTSB þ N) was added to the second 10-g beef specimen, which was then homogenized and enriched via incubation at 37  C for 24 h. The enriched culture was plated on tellurite-cefixime sorbitol MacConkey agar (TCSMAC; Oxoid Ltd.) and incubated at 37  C for 24 h. White colonies on TSCMAC agar plates were regarded as presumptive E. coli O157:H7 and collected using an inoculation loop, and streaked on tryptone bile X-glucuronide agar (TBX; Oxoid Ltd.). After incubation at 37  C for 24 h, typical E. coli O157:H7 colonies exhibiting a blue-green color on TBX agar plates were subjected to polymerase chain reaction (PCR) assay. To enrich L. monocytogenes, the third 10-g beef specimen was incubated in 90 mL of UVM-modified Listeria enrichment broth (UVM; Oxoid Ltd.) at 37  C for 24 h. Then, 0.1 mL of cultured UVM broth was transferred into 10 mL of secondary enrichment broth (Fraser Listeria broth, FB; Oxoid Ltd.) and cultured at 37  C for 24 h. A sterile cotton swab was soaked in cultured FB and streaked on Oxford agar (Oxoid Ltd.). After incubation at 30  C for 24e48 h, black colonies on the plates were sampled and subjected to PCR assay. Salmonella spp. were detected in a manner similar to that used for L. monocytogenes using the final 10-g beef specimen. The primary and secondary enrichment broths were peptone water and Rappaport-Vassiliadis broth (Oxoid Ltd.), respectively. The secondary enrichment culture was plated on MacConkey agar (MAC; Oxoid Ltd.) and xylose-lysine-deoxycholate agar (XLD; Oxoid Ltd.) and incubated at 37  C for 24 h. Colonies with no color on MAC agar plates and colonies with black centers on XLD agar plates were

52

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

subjected to PCR assay. Colonies presumed to be E. coli O157:H7, L. monocytogenes, and Salmonella spp. were transferred to 10 mL of TSB (Oxoid Ltd.) and incubated at 37  C for 18e24 h. Then, DNA from each bacterial culture was extracted using a TaKaRa MiniBEST Universal Genomic DNA Extraction Kit (TaKaRa Bio Inc., Dalian, China). PCR assays to confirm the presence of E. coli O157:H7, L. monocytogenes, and Salmonella spp. were performed using target-specific primers (Table 1). Target genes on genomic DNA of target bacteria were amplified in a 20-mL PCR reaction mixture containing 4 mL of bacterial genomic DNA, 0.25 mM of each primer (Bioneer Corp., Seoul, South Korea), and 10 mL Premix Taq DNA polymerase (Takara Bio Inc., Otsu, Shiga, Japan) using a thermal cycler (C1000 Thermal Cycler; Bio-Rad Laboratories Inc., Hercules, CA, USA) under the cycling conditions recommended in the KFDA. (2013). Amplification of target genes of target bacteria by PCR was confirmed by agarose gel electrophoresis. 2.3. Pyrosequencing analysis of bacterial communities in fresh and spoiled beef A beef specimen purchased from a local retail market was divided into two 10-g pieces and stored at 5  C. The two 10-g pieces were sampled on days 0 and 5, respectively, and homogenized in 90 mL of saline solution. Then, 1 mL of homogenate was used for enumeration of spoilage bacteria. Another 1 mL of homogenate was transferred to a 1.5-mL microtube and centrifuged for 1 min at 10,000g to collect bacteria. Genomic DNA of collected bacteria was extracted using FastDNA® SPIN Kits (MP Bio Laboratories, Carlsbad, CA, USA). Extracted bacterial genomic DNA was amplified using primers (27F and 518R; 27F: 50 -GAG TTT GAT CMT GGC TCA G-30 and 518R: 50 -WTT ACC GCG GCT GCT GG-30 ) targeting the V1eV3 regions of the 16S rRNA gene. PCR was performed in a 25-mL reaction mixture containing 1 mL of bacterial genomic DNA, 10 mM each primer, 1.25 U high-fidelity Taq polymerase (Roche Diagnostics Ltd., Indianapolis, IN, USA), 0.2 mM dNTPs, and 50 mM MgSO4 using a C1000 thermal cycler under the following cycling conditions: 95  C for 5 min; and 30 cycles of 95  C for 30 s, 55  C for 30 s, and 72  C for 30 s. Pyrosequencing was performed with a 454 GS FLX Titanium DNA sequencer (Roche Diagnostics Ltd., Penzberg, Germany) at Chunlab Inc. (Seoul, South Korea). Sequence reads from beef specimens at 5  C on days 0 and 5 were classified using barcode sequences. Pyrosequencing reads were compared with the EzTaxon-e database (Chun et al., 2007), and their taxonomies were assigned based on the similarity of each read to the database, as follows: species (x  97%), genus (97% > x  94%), family (94% > x  90%), order (90% > x  85%), class (85% > x  80%), and phylum (80% > x  75%), using the CLcommunity software ver. 3.30 (ChunLab Inc.). Sequences with similarity values lower than the dissimilarity cut-off value (3%) were considered unclassified (UC). Rarefactions of bacterial species were generated using the CLcommunity software, based on the operational taxonomic units (OTUs) obtained by analysis of the pyrosequencing data. The Chao1 index, an estimator of asymptotic species richness (Chao,

1984), and the Gini-Simpson index, an indication of biodiversity (Gotelli & Chao, 2013), were calculated using the CLcommunity software. The Gini-Simpson index reflects the probability that two randomly selected isolates belong to the same species (Gotelli & Chao, 2013). 3. Results and discussion 3.1. Mathematical predictions of the time required for spoilage bacterial populations in beef to reach bacteriological standards Bacterial growth is governed by T (Voyer & McKellar, 1993). Bacteria grow more rapidly with a shorter delay (l) as T increases, unless the T is greater than the optimal growth temperature (Koutsoumanis, Stamatiou, Skandamis, & Nychas, 2006). Spoilage bacteria in beef grew more rapidly with a shorter l as the T of storage increased from 5 to 25  C (Fig. 1A). In addition, the Nmax increased with increasing T (Fig. 1A). Since the growth of spoilage bacteria in beef followed a sigmoidal course, the l, m, and Nmax values could be determined by regression of a curve to a modified-Gompertz function (Table 2). Changes in the l, m, and Nmax values versus T are plotted in Fig. 1B, C, and D, respectively. The l, m, and Nmax values exponentially decreased, increased, and changed sigmoidally, respectively, with increasing T (Fig. 1B, C, and D). Thus, plots of the l, m, and Nmax values could be regressed to exponential decay, exponential growth, and logistic function, respectively (Fig. 1B, C, and D). The equations obtained by regression of the l, m, and Nmax values were l ¼ 1:6737 þ 312:7876,expð
0:3369,TÞ (R2 ¼ 0.999, bias factor ¼ 1.037),m ¼
0:0234 þ 0:0245,expð0:1092,TÞ (R2 ¼ 0.999, bias factor ¼ 1.036), and Nmax ¼ 7:5542 þ 0:6517=½1þ ðT=13:3586Þ
9:7135  (5  C  T  25  C) (R2 ¼ 0.997, bias factor ¼ 1.000), respectively. The l, m, and Nmax values at 5e25  C were estimated using these equations. A bacterial growth plot with a sigmoidal shape can be mathematically computed by applying the estimated l, m, N0, and Nmax values at a particular T to a modifiedGompertz function, as such growth characteristics could be analyzed by regression of a plot to the function. Equations to plot the growth of spoilage bacteria in beef versus T were computed by applying the l, m, and Nmax equations to a modified-Gompertz function. The plots were transformed to tE NI versus T according to NI and N0 to generate plots for the estimation of tE NI according to NI and N0 values (Fig. 2). The N0 of beef in retail markets ranges from 0 to 5 log CFU$g
1 depending on the product type, season, and market chain (Badoni, Rajagopal, Aalhus, & Klassen, 2012). The bacteriological standard of spoilage bacteria in beef for international trade is 7 log CFU$g
1 (ICMSF, 1986). Beef containing > 8 log CFU$g
1 spoilage bacteria is considered unsafe for consumption. The NI and N0 values applied to the plots of tE NI versus T were 5, 6, 7, and 8 CFU g
1 and 0, 1, 2, 3, 4, and 5 CFU g
1, respectively (Fig. 2).
1 Moreover, a plot of tE 8 log CFU g versus T between 14.5 and 25  C was constructed since the minimum T required for spoilage bacteria to reach 8 log CFU$g
1 was 14.46  C based on the Nmax regression equation (Nmax ¼ 7:5542 þ 0:6517=½1 þ ðT=13:3586Þ
9:7135 ).

Table 1 PCR primers used to confirm the presence of E. coli O157:H7, L. monocytogenes, and Salmonella spp. in retail beef specimens. Bacterial target and genes

Primer sequence (50 / 30 )

PCR amplicon size

E. coli O157:H7, verotoxin type 1 gene (VT1)

Forward: CTG GAT TTA ATG TCG CAT AGT G Reverse: AGA ACG CCC ACT GAG ATC ATC Forward: GCA GTT GCA AGC GCT TGG AGT GAA Reverse: GCA ACG TAT CCT CCA GAG TGA TCG Forward: ACA GTG CTC GTT TAC GAC CTG AAT Reverse: AGA CGA CTG GTA CTG ATC GAT AAT

150 bp

L. monocytogenes, L. monocytogenes hemolysin gene (hlyA) Salmonella spp., Salmonella invasion gene (invA)

456 bp 244 bp

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

53

Fig. 1. Growth curve of spoilage bacteria present in beef versus storage time fitted with a modified-Gompertz function (A) and plots of the lag time (l) (B), specific growth rate (m) (C), and maximum population (Nmax) (D) versus temperature (T) fitted with exponential decay, exponential growth, and logistic function, respectively.

Plots of tE NI versus T may facilitate estimation of the shelf-life of beef. For example, beef intended for international trade contaminated with 2 log CFU$g
1 of spoilage bacteria stored at 5  C for > 17.1 days (411.5 h) may contain > 7 log

CFU$g
1 spoilage bacteria (Fig. 2B), and would be rejected by the importing country. Accordingly, the shelf-life of this beef should not exceed 17 days to avoid rejection in international trade.

Table 2 Growth characteristics of spoilage bacteria in beef according to storage temperature obtained by regression of growth curves to a modified-Gompertz function.a Temperature ( C)

5 10 15 20 25 a b c

Growth characteristics

Regression analysis

Lag time (l) (h)

Specific growth rate (m) [(log CFU$g
1 h
1]

Maximum population (Nmax) (log CFU$g
1)

R2

59.74 (26.44e88.66) 12.14 (7.94e16.12) 4.61 (1.94e7.08) 1.73 (0.44e2.93) 1.40 (0.17e2.54)

0.014 (0.011e0.021) 0.056 (0.049e0.066) 0.104 (0.091e0.120) 0.191 (0.170e0.214) 0.354 (0.293e0.436)

7.56 (6.85e10.10) 7.59 (7.44e7.75) 8.05 (7.89e8.21) 8.17 (8.02e8.32) 8.23 (7.91e8.58)

Numbers in parentheses are the 95% confidence limits. R2: determinant coefficient. RMSE: root-mean-square error.

b

RMSEc

Bias factor

0.967

0.21

1.0038

0.995

0.09

1.0004

0.989

0.15

1.0013

0.989

0.15

1.0012

0.970

0.24

1.0025

54

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

Fig. 2. Estimated time required for spoilage bacteria in beef to reach the bacteriological standards of interest ðtE NI Þ at temperature (T) if the initial population (N0), is 0, 1, 2, 3, 4, or 5 log CFU$g
1. A, B, C, and D are plots of the tE NI for 8, 7, 6, and 5 log CFU$g
1 of the bacteriological standards of interest (NI), respectively.

3.2. Bacterial contamination of retail beef No E. coli O157:H7 or Salmonella spp. was detected in any of the 100 retail beef specimens. However, L. monocytogenes was found in 6 of the 100 specimens (Table 3). L. monocytogenes was detected in 4 of 66 (6%) domestic and 2 of 34 (6%) imported beef specimens (Table 3). L. monocytogenes is rarely present in beef; its prevalence is 1/233 (0.42%) in South Korea and 0/572 (0%) in Australia (Kim, Lee, Gi, & Lee, 2005; Phillips, Bridger, Jenson, & Sumner, 2012). All L. monocytogenes-positive domestic and imported beef specimens were from a single private butcher. Therefore, the beef was likely contaminated by L. monocytogenes during handling at this butcher. Table 3 Prevalence of E. coli O157:H7, L. monocytogenes, and Salmonella spp. in retail beef specimens. Bacterium

Positive specimens Domestic

Imported

Total

E. coli O157:H7

0/66 (0%) 4/66 (6%) 0/66 (0%)

0/34 (0%) 2/34 (6%) 0/34 (0%)

0/100 (0%) 6/100 (6%) 0/100 (0%)

L. monocytogenes Salmonella spp.

The numbers of total aerobic bacteria, coliform bacteria, and E. coli in retail beef are summarized in Table 4. Total aerobic bacteria were found in all of the 100 retail beef specimens, suggesting that beef is inevitably contaminated by spoilage bacteria during processing and handling. The number of total aerobic bacteria in beef varied from 2.27 to 6.15 log CFU$g
1. The collection and bacteriological tests were completed in one month. Moreover, we used beef that was delivered to retail stores on the day of or the day before collection. Thus, the variation in the total numbers of aerobic bacteria in retail beef might be due to differences in the extent of bacterial contamination during processing rather than to seasonal fluctuations or bacterial growth in retail outlets. Of the retail beef specimens, 33% contained 4e5 log CFU$g
1 total aerobic bacteria (Table 4). However, the level of total aerobic bacterial contamination differed between domestic and imported beef. Among the 66 domestic retail beef specimens, 22 contained 4e5 log CFU$g
1 total aerobic bacteria. Among the 34 imported beef specimens, the most frequently detected contamination level was 5e6 log CFU$g
1. Imported beef is frequently stored for a longer period than domestic beef, which may facilitate bacterial growth. Nevertheless, the prevalence of coliform bacteria and E. coli in imported beef was less than that in domestic beef. In total, 92% and 68% of 66 domestic beef specimens were contaminated with coliform bacteria (61/66) and E. coli (18/66), respectively. However, only 68% (22/34) and 3% (1/34) of imported beef specimens were

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

55

Table 4 Total aerobic bacteria, coliform bacteria, and E. coli populations in retail beef. Bacterium

Total aerobic bacteria Coliform bacteria

E. coli

a

Specimens

Domestic Imported Total Domestic Imported Total Domestic Imported Total

Bacterial population range (log CFU$g
1) 1e2

2e3

3e4

4e5

5e6

6e7

0/66 (0%) 0/34 (0%) 0/100 (0%) 14/66 (21%) 5/34 (15%) 19/100 (19%) 15/66 (23%) 1/34 (3%) 16/100 (16%)

5/66 (8%) 0/34 (0%) 5/100 (5%) 25/66 (38%) 9/34 (26%) 34/100 (34%) 2/66 (3%) 0/34 (0%) 2/100 (2%)

19/66 (29%) 7/34 (21%) 26/100 (26%) 17/66 (26%) 5/34 (15%) 22/100 (22%) 1/66 (2%) 0/34 (0%) 1/100 (1%)

22/66 (33%) 11/34 (32%) 33/100 (33%) 5/66 (8%) 3/34 (9%) 8/100 (8%) 0/66 (0%) 0/34 (0%) 0/100 (0%)

16/66 (24%) 15/34 (44%) 31/100 (31%) 0/66 (0%) 1/34 (3%) 1/100 (1%) 0/66 (0%) 0/34 (0%) 0/100 (0%)

4/66 (6%) 1/34 (3%) 5/100 (5%) 0/66 (0%) 0/34 (0%) 0/100 (0%) 0/66 (0%) 0/34 (0%) 0/100 (0%)

Prevalence (%)

Mean ± SDa(log CFU$g
1)

66/66(100%) 34/34 (100%) 100/100 (100%) 61/66 (92%) 23/34 (68%) 84/100 (84%) 18/66 (27%) 1/34 (3%) 19/100 (19%)

4.42 4.86 4.57 2.40 1.89 2.23 0.37 0.48 0.37

± ± ± ± ± ± ± ± ±

0.99 0.73 0.93 1.13 1.59 1.32 0.36 0.00 0.35

Standard deviation.

contaminated with coliform bacteria and E. coli, respectively. The greater coliform bacteria and E. coli populations in domestic beef were likely due to inappropriate handling at beef processing plants and retail stores. The level of contamination of beef with spoilage bacteria varied markedly, even in the same season (Table 4). Application of a uniform shelf-life of beef without consideration of the level of contamination with spoilage bacteria can result in spoilage prior to expiration of the shelf-life and foodborne disease. Therefore, shelflife charts for mathematical estimation of the shelf-life according to N0 and T are required to manage the safety and quality of beef. 3.3. Pyrosequencing analysis of the bacterial community in beef The total aerobic counts of the beef specimens from days 0 and 5 that were subjected to pyrosequencing analysis were approximately 4.1 and 4.9 log CFU$g
1, respectively. Based on the pyrosequencing analysis, the total bacterial 16S rRNA sequence counts of the beef specimens from days 0 and 5 were 4290 and 7402, respectively (Tables 5 and 6). Rarefaction curves, plots of OTUs versus the number of sequencing reads, became saturated with increasing number of sequencing reads (Fig. 3A). As the instantaneous slope and height of a rarefaction curve indicate species , 2012), the pyrosequencappearance (Li, Bihan, Yooseph, & Methe ing analysis carried out in this study was validated. The

instantaneous slope and height of a rarefaction curve of specimens on day 5 were steeper and greater, respectively, than those of specimens on day 0. The Chao 1 index, an estimator of asymptotic species richness, of specimens of days 0 and 5 was approximately 100 and 265, respectively (Fig. 3B), suggesting that species that were not present on day 0 were detected on day 5. The GiniSimpson index, an index of biodiversity, of the day 5 specimen was lower than that of the day 0 specimen (Fig. 3C), suggesting less marked domination by a few bacterial species on day 5 than on day 0. Newly detected species in the day 5 specimen were likely present in the day 0 specimen since a portion of the day 0 specimen was stored at 5  C for 5 days and subjected to pyrosequencing analysis. However, a very small number of species newly detected in the day 5 specimen might be present in the day 0 specimen. Genomic DNA of species not detected in the

Table 6 Taxonomic composition of bacterial populations in retail beef after storage at 5  C for 5 days. Order

Table 5 Taxonomic composition of bacterial populations in retail beef before storage at 5  C for 5 days. Order

Species

Lactobacillales

Frequency

3302 127 49 52 63 471

Brochothrix thermosphacta Other

445 26 122

Corynebacterium simulans Other Pseudomonadales

120 2 63

Flavobacteriales

20

Enterobacteriales

19

Aeromonadales

1

Caulobacterales

1

Total reads

4290

Bacillales

Corynebacteriales

Pseudomonas deceptionensis Pseudomonas fragi Pseudomonas psychrophila Pseudomonas lundensis Pseudomonas meridiana Pseudomonas panacis Acinetobacter bereziniae Other

1839 994 718 303 285 200 89 343 1445

Serratia plymuthica Serratia proteamaculans Hafnia alvei Serratia quinivorans Serratia liquefaciens Yersinia kristensenii Rahnella aquatilis Buttiauxella agrestis Other

219 217 158 148 128 119 77 50 329 589

Carnobacterium divergens Lactococcus piscium Other

426 79 84 395

Brochothrix thermosphacta Other

394 1 202

Myroides phaeus Other

196 6

Lactobacillales

Bacillales

Flavobacteriales

Total reads

Frequency 4771

Enterobacteriales

3593 Carnobacterium divergens Carnobacterium maltaromaticum Lactococcus lactis Lactococcus piscium Other

Species

Pseudomonadales

7402

56

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

Fig. 3. Results of the pyrosequencing analysis. Rarefaction curves of bacterial 16S rRNA of spoilage bacteria in beef (A). Comparison of the bacterial species richness and diversity in beef before and after storage at 5  C for 5 days using the Chao1 index, an estimator of asymptotic species richness (B), and the Gini-Simpson index, an indicator of biodiversity (C).

day 0 specimen but detected in the day 5 specimen might have been excluded during DNA preparation. Moreover, growth of these species during storage for 5 days might have enabled extraction of their genomic DNA. Lactobacillales including C. divergens, C. maltaromaticum, L. lactis, and L. piscium were dominant in the bacterial community of the day 0 specimen (Table 5). Pseudomonadales including P. deceptionensis, P. fragi, P. psychrophila, P. lundensis, P. meridian, P. panacis, and A. bereziniae, which were detected at low levels on day 0, were the most abundant species on day 5 (Table 6). Bacterial contaminants of beef originate from bovine feces, hides, and processing facilities (Galland, 1997). Lactobacillales, facultative anaerobic bacteria, are prevalent in the bovine intestine and feces (Han, Ogata, Yamamoto, Nagao, & Nishino, 2014). However, Pseudomonadales, which are aerobic bacteria, are frequently found on bovine hides and in beef-processing facilities (Nel, Lues, Buys, & Venter, 2004; Pandeeti et al., 2011). Therefore, the beef specimen used for the pyrosequencing analysis might have been contaminated from multiple sources. Bovine intestine exudates or feces and bovine hides or beef-processing facilities are likely origins of Lactobacillales and Pseudomonadales, respectively. Pseudomonadales grew during aerobic storage at 5  C and predominated in the day 5 specimen (Table 6).

4. Conclusion Mathematical modeling of bacterial growth has been studied extensively over the last two decades. It is well-acknowledged that bacterial growth in food, as well as the microbiological shelf-life of food, can be predicted using predictive mathematical tools. Mathematical modeling procedures to predict the growth of spoilage bacteria in fresh and processed meat, as well as their shelflives in different conditions, have been reported previously (Kreyenschmidt et al., 2010; Limbo, Torri, Sinelli, Franzetti, & Casiraghi, 2010). Although the procedures are descriptive and concrete, they are not being used actively in the meat industry. This is probably because mathematical modeling is difficult to understand and complicated for employees in the food industry to execute. Shelf-life charts for fresh beef were developed based on mathematical analyses of spoilage bacterial growth in fresh beef at different temperatures. Using these charts, the shelf-life of fresh beef products can be estimated easily if the spoilage bacteria population of the product is obtained. Although the charts proposed in this study are simple and easy to use for estimating the shelf-life of fresh beef, they are applicable only to fresh beef stored in aerobic conditions. However, bacterial contaminants of fresh

C. Chai et al. / LWT - Food Science and Technology 81 (2017) 50e57

beef can originate from multiple sources, including bovine intestine exudates, feces, and hides, which can contribute to contamination by aerobic bacteria, anaerobic bacteria, or both according to pyrosequencing analysis of the bacterial community in fresh beef. A large amount of fresh beef is currently distributed in modified atmosphere packages (MAPs), providing anaerobic conditions for spoilage bacteria. Thus, mathematical investigation of the growth of spoilage bacteria in MAP fresh beef, and the development of shelf-life charts for MAP fresh beef, are required to improve the microbiological safety and quality of fresh beef in the retail market. Acknowledgements This study was supported by the High Value-Added Food Technology Development program of the Korea Institute of Planning and Evaluation for Technology in Food, Agriculture, Forestry and Fisheries (IPET), funded by the Ministry of Agriculture, Food and Rural Affairs (grant no. 313032-03-2-HD020). This study was also supported by 2016 Research Grant from Kangwon National University. References Badoni, M., Rajagopal, S., Aalhus, J. L., & Klassen, M. D. (2012). The microbiological condition of Canadian beef steaks offered for retail sale in Canada. Journal of Food Research, 1(4), 124e133. Beuchat, L. R. (1996). Pathogenic microorganisms associated with fresh produce. Journal of Food Protection, 59(2), 204e216. Bruckner, S., Albrecht, A., Petersen, B., & Kreyenschmidt, J. (2013). A predictive shelf life model as a tool for the improvement of quality management in pork and poultry chains. Food Control, 29(2), 451e460. Casaburi, A., Piombino, P., Nychas, G.-J., Villani, F., & Ercolini, D. (2015). Bacterial populations and the volatilome associated to meat spoilage. Food Microbiology, 45(Part A(0)), 83e102. Chao, A. (1984). Nonparametric estimation of the number of classes in a population. Scandinavian Journal of Statistics, 11(4), 265e270. Chun, J., Lee, J.-H., Jung, Y., Kim, M., Kim, S., Kim, B. K., et al. (2007). EzTaxon: A webbased tool for the identification of prokaryotes based on 16S ribosomal RNA gene sequences. International Journal of Systematic and Evolutionary Microbiology, 57(10), 2259e2261. European Commission. (2005). Commission Regulation (EC) No. 2073/2005 of 15 November 2005 on microbiological criteria for foodstuffs. Official Journal of the European Union, L, 338, 1e26, 22.12.2005. Galland, J. C. (1997). Risks and prevention of contamination of beef carcasses during the slaughter process in the United States of America. Revue scientifique et technique, 16(2), 395e404. Gotelli, N. J., & Chao, A. (2013). Measuring and estimating species richness, species diversity, and biotic similarity from sampling data. In S. A. Levin (Ed.), Encyclopedia of biodiversity (2nd ed., pp. 195e211). Waltham: Academic Press. Han, H., Ogata, Y., Yamamoto, Y., Nagao, S., & Nishino, N. (2014). Identification of

57

lactic acid bacteria in the rumen and feces of dairy cows fed total mixed ration silage to assess the survival of silage bacteria in the gut. Journal of Dairy Science, 97(9), 5754e5762. International Commission on Microbiological Specifications for Foods (ICMSF). (1986). Microorganisms in foods 2: Sampling for microbiological analysis : Principles and specific applications (2nd ed.). Oxford: Blackwell Scientific Publications. KFDA. (2013). Korea food standards codex. Chapter 9. general methods. Retrieved from Food Standards Codex website https://www.foodsafetykorea.go.kr/portal/ safefoodlife/food/foodRvlv/foodRvlv.do. Kim, J.-Y., Lee, J.-H., Gi, N.-J., & Lee, J.-H. (2005). Microbiological quality and detection of pathogenic microorganisms in slaughtered meat in Seoul area. Korean Journal of Veterinary Service, 28(3), 215e223. Koutsoumanis, K., Stamatiou, A., Skandamis, P., & Nychas, G.-J. E. (2006). Development of a microbial model for the combined effect of temperature and pH on spoilage of ground meat, and validation of the model under dynamic temperature conditions. Applied and Environmental Microbiology, 72(1), 124e134. Kreyenschmidt, J., Hübner, A., Beierle, E., Chonsch, L., Scherer, A., & Petersen, B. (2010). Determination of the shelf life of sliced cooked ham based on the growth of lactic acid bacteria in different steps of the chain. Journal of Applied Microbiology, 108(2), 510e520. , B. (2012). Analyses of the microbial diversity Li, K., Bihan, M., Yooseph, S., & Methe across the human microbiome. PLoS ONE, 7(6), e32118. Limbo, S., Torri, L., Sinelli, N., Franzetti, L., & Casiraghi, E. (2010). Evaluation and predictive modeling of shelf life of minced beef stored in high-oxygen modified atmosphere packaging at different temperatures. Meat Science, 84(1), 129e136. Linton, R. H., Eisel, W. G., & Muriana, P. M. (1997). Comparison of conventional plating methods and petrifilm for the recovery of microorganisms in a ground beef processing facility. Journal of Food Protection, 60(9), 1084e1088. Nel, S., Lues, J. F. R., Buys, E. M., & Venter, P. (2004). Bacterial populations associated with meat from the deboning room of a high throughput red meat abattoir. Meat Science, 66(3), 667e674. Nollet, L. M. L. (2012). Shelf life of meats. In L. M. L. Nollet (Ed.), Handbook of meat, poultry and seafood quality (2nd ed., pp. 232e245). West Sussex, UK: WileyBlackwell. Oscar, T. P. (2005). Development and validation of primary, secondary, and tertiary models for growth of Salmonella Typhimurium on sterile chicken. Journal of Food Protection, 68(12), 2606e26013. Pandeeti, E. V. P., Pitchika, G. K., Jotshi, J., Nilegaonkar, S. S., Kanekar, P. P., & Siddavattam, D. (2011). Enzymatic depilation of animal hide: Identification of elastase (LasB) from Pseudomonas aeruginosa MCM B-327 as a depilating protease. PLoS ONE, 6(2), e16742. Phillips, D., Bridger, K., Jenson, I., & Sumner, J. (2012). An Australian national survey of the microbiological quality of frozen boneless beef and beef primal cuts. Journal of Food Protection, 75(10), 1862e1866. Rivera-Betancourt, M., Shackelford, S. D., Arthur, T. M., Westmoreland, K. E., Bellinger, G., Rossman, M., et al. (2004). Prevalence of Escherichia coli O157:H7, Listeria monocytogenes, and Salmonella in two geographically distant commercial beef processing plants in the United States. Journal of Food Protection, 67(2), 295e302. Shimoni, E., & Labuza, T. P. (2000). Modeling pathogen growth in meat products: Future challenges. Trends in Food Science & Technology, 11(11), 394e402. Voyer, R., & McKellar, R. C. (1993). MKES tools: A microbial kinetics expert system for developing and assessing food production systems. Journal of Industrial Microbiology, 12(3e5), 256e262.