Bacterial contaminants from frozen puff pastry production process and their growth inhibition by antimicrobial substances from lactic acid bacteria

Abstract Seventy‐five bacterial contaminants which still persisted to cleaning system from three puff pastry production lines (dough forming, layer and filling forming, and shock freezing) were identified using 16S rDNA as seven genera of Bacillus, Corynebacterium, Dermacoccus, Enterobacter, Klebsiella, Pseudomonas, and Staphylococcus with detection frequencies of 24.00, 2.66, 1.33, 37.33, 1.33, 2.66, and 30.66, respectively. Seventeen species were discovered while only 11 species Bacillus cereus, B. subtilis, B. pumilus, Corynebacterium striatum, Dermacoccus barathri, Enterobacter asburiae, Staphylococcus kloosii, S. haemolyticus, S. hominis, S. warneri, and S. aureus were detected at the end of production. Based on their abundance, the highest abundance of E. asburiae could be used as a biomarker for product quality. While a low abundance of the mesophile pathogen C. striatum, which causes respiratory and nervous infection and appeared only at the shock freezing step was firstly reported for its detection in bakery product. Six antimicrobial substances (AMSs) from lactic acid bacteria, FF1‐4, FF1‐7, PFUR‐242, PFUR‐255, PP‐174, and nisin A were tested for their inhibition activities against the contaminants. The three most effective were FF1‐7, PP‐174, and nisin A exhibiting wide inhibition spectra of 88.00%, 85.33%, and 86.66%, respectively. The potential of a disinfectant solution containing 800 AU/ml of PP‐174 and nisin A against the most resistant strains of Enterobacter, Staphylococcus, Bacillus and Klebsiella was determined on artificially contaminated conveyor belt coupons at 0, 4, 8, 12, and 16 hr. The survival levels of the test strains were below 1 log CFU/coupon at 0 hr. The results suggested that a combined solution of PP‐174 and nisin A may be beneficial as a sanitizer to inhibit bacterial contaminants in the frozen puff pastry industry.

Occasionally, microbial control strategies are inefficient to completely eliminate microorganisms. Therefore, there are many foodborne illness outbreaks have been reported by poor hygiene such as ineffective hand washing or inadequate temperature controls of food (Lambrechts, Human, Doughari, & Lues, 2014). These wrong manipulation and food nature can trigger pathogenic contamination. Nowadays, convenient food like bakery products has become more popular among Thai people reflecting the change of their lifestyle (Wongsuttichote & Nitisinprasert, 2009). Puff pastry is a kind of bakery products which is popular nowadays due to varieties of taste and ease of consumption. Its production rate had gradually increased for 5-12% during the past 3 years. Puff pastry has a laminated structure of baked layers of dough separated by thin layers of fat with varieties of fillings causing the short shelf life (Hay, 1993). Kotzekidou (2013) reported the contamination of foodborne pathogen in frozen pastries including Bacillus cereus, Salmonella spp., Escherichia coli O157:H7, Listeria monocytogenes, and Staphylococcus aureus in raw material, cross-contamination during preparation, and handling causing reduction of product shelf life. Biopreservation is another option for food safety to extend its shelf life (Stiles, 1996). Many reports have noted the success of lactic acid bacteria (LAB) for the growth inhibition of pathogens and foodborne contaminants (Muhialdin et al., 2012). LAB produces numerous antimicrobial substances (AMS) including organic acids, hydrogen peroxide, diacetyl, carbon dioxide, reuterin, and bacteriocins (Galvez, Burgos, Lopez, & Pulido, 2014), which may inhibit nonpathogens and pathogens associated with food and humans (Corsetti, Settanni, Braga, De Fatima Silva Lopes, & Suzzi, 2008). Among these, nisin is a lantibiotic antimicrobial peptide which is produced by Lactococcus lactis. Nisin exhibited bactericidal activity against foodborne bacteria including Bacillus cereus, Staphylococcus aureus, Listeria monocytogenes and recognized as safe food preservative (Wiedemann et al., 2001).
In addition, it is odorless, colorless, tasteless, and has a low toxicity (Tong, Ni, & Ling, 2014). Wongsuttichote and Nitisinprasert (2009) reported that the cell-free supernatants (CFS) from Lactobacillus plantarum KUB-KJ174 isolated from fermented rice noodles displayed spectrum inhibition activity of 100% against bacterial contaminants of 50 aerobic mesophiles, 10 Bacillus cereus, and 53 coliform tested but only 46% inhibition spectrum activity against 13 LAB isolated from bakery product namely "E' Claire". Partial purification by pH mediated cell adsorption-desorption providing an active AMS so called PP-174 displayed 100% growth inhibition activities against microbial strains of aerobic mesophile bacteria T6.14, B. cereus B6.2, coliform C4.1, LAB L2.2, and yeast Y5.1. In our previous study, AMS producing LAB have been screened from various fermented food in Thailand. The cell-free supernatants (CFS) namely  produced by four strains of KUB-KJ174, KA-FF1-4, KA-FF1-7, KA-PFUR-242, and KA-PFUR-255, respectively, grown in MRS medium exerted high inhibition activities against Staphylococcus aureus TISTR 029. These LAB strains can be source of AMS used for biopreservatives in the future.
In Thailand, these products had a short shelf life of only 1-2 days.
To maintain the fresh product, the puff pastry is maintained as a dough form at low temperature as a frozen product. They will be baked at high temperature before serving or distribution to the customer. All raw materials and equipments for three lines of dough-forming, layer and filling-forming and shock-frozen process need to be aware for food safety. Normally, all equipments used for industrial puff pastry production are properly cleaned by disinfectant agents followed by water at the end. However, sanitation test by swab test of all equipments before usage showed the occurrence of some bacterial contaminants which could be resistant to the cleaning system. Therefore, this research aimed to identify the bacterial contaminants which persisted to the cleaning system and further investigate for potential AMS for their growth inhibition.

| Bacterial contaminant sources and morphological analysis
All bacterial contaminants were collected from commercial bakery industry located in Bangkok, Thailand. The temperature during process could be about 25-37°C. They were obtained from sanitation test using swab test of all equipments before usage. Briefly, all machines and equipments were properly cleaned with detergents and water, and then left for overnight at the temperature of about 25-37°C to obtain the dry ones. Before the production started, sanitation test were performed by swab test on PCA (Merck, Germany). Each bacterial colony having different characters appear was subsequently purified on nutrient agar (NA) to obtain 8, 52, and 15 isolates from three production lines including the dough-forming line (DL), the layer-andfilling line (LFL), and the shock-frozen line (SL), respectively (Table 1).
All culture isolates were propagated in Nutrient Broth medium (NB; Merck, Germany) under aerobic conditions by shaking at 200 rpm at 37°C overnight and determined for their morphologies by an optical microscope (Primo Star, Thornwood, NY, USA).

| DNA extraction
Bacterial DNA was extracted from each overnight culture solution using the bacteria genomicPrep Mini Spin Kit (Illustra ™ , Buckinghamshire, UK) and performed according to the manufacturer's instructions. The extracted DNA was subjected to randomly amplified polymorphic DNA polymerase chain reaction (RAPD-PCR) for biotype screening and 16S rDNA analysis.

| RAPD-PCR typing
The 10-base pair primers of OPA3 (5′-AGTCA GCCAC-3′) and OPA11 (5′-CAATCGCCGT-3′) which were described by Sorokulova et al. (2003) and Baker, Crumley, and Eckdahl (2002), respectively, were used in this study. The reactions were prepared in a total volume of 25 μl containing 10-50 ng of genomic DNA, 0.4 μmol/L of each primer, 0.2 mmol/L of dNTP mix, 1.5U of Taq DNA polymerase (Fermentas, Waltham, Massachusetts, USA), and 10× PCR Buffer with MgCl 2 . The DNA was amplified using the following program: 94°C for 1 min and 45 cycles of 94°C for 1 min, 36°C for 1 min, 72˚C for 1 min, and a final extension of 72°C for 10 min. Amplified PCR products were verified using electrophoresis in 1.5% agarose gel (Mupid ® ex, Tokyo, Japan) containing ethidium bromide according to the method of Devos and Gale (1992). The DNA marker used to determine the size of amplified fragments was O'GeneRuler ™ 1 kb DNA Ladder (Fermentas, Waltham, Massachusetts, USA). Each DNA fragment was visualized using a UV transilluminator and later photographed. The clearest and most reproducible bands were chosen for determination of their presence and absence in each isolate. Faint bands which could not be systematically visualized were not taken into account.

| Cluster analysis
The observed bands in the gels were evaluated based on the presence (coded 1) or absence (coded 0) of polymorphic fragments for RAPD products. Cluster analysis was performed with the NTSYS-pc (version 2.10e) software, which is a numerical taxonomy and multivariate analysis software package (Rohlf, 2000). The similarity among digitized profiles was calculated using the SAHN. The dendrogram was constructed using an unweighted pair group method with arithmetic (UPGMA) cluster analysis.

| Determination of the detection frequency
The detection frequencies of the 75 bacterial contaminant isolates collected from the three production lines were determined. They were expressed as the percentage of isolates belonging to each genus of the total isolates, and the percentage belonging to each species of the total isolates for each production line.

| Determination of inhibition spectra
All bacterial contaminants were used as target isolates to determine their inhibition activities according to the modified method of where N inhibition is the number of bacterial contaminant isolates inhibited by each AMS and N total is the total number of bacterial contaminant isolates tested.

| Artificial contamination of conveyor belt coupon and determination of antimicrobial substance efficiency
Conveyor belt coupons (size 2 × 2 cm 2 ) were used as a model and prepared according to the modified method described by Phongphakdee and Nitisinprasert (2015). Each conveyor belt sheet was cleaned with distilled water and autoclaved at 121°C for 15 min. Twenty-one representative bacterial contaminants resistant to either PP-174 or nisin A (Table 2) were mixed and used to perform the contaminant coupons. Afterward, the coupons were transferred to conical centrifuge tubes containing 30 ml of steriled normal saline (0.85% w/v NaCl) and vortexed at the maximum speed for 1 min for cell dispersion. Aliquots 1 ml of sample were 10-fold serially diluted with 9 ml of normal saline solution and analyzed by standard plate count method using NA medium at 37°C for 24 hr to determine its survival cell/coupon.

Statistical analysis
A one-way ANOVA was performed to determine the standard deviations and statistical significance of data at the 95% confidence interval (p < .05). All treatments were performed in duplicate. All analyzes were carried out using the SPSS statistical package version 16.0.

| Morphological analysis of the bacterial contaminants
Seventy-five bacterial contaminants from three processes of puff pastry line production consisting of dough-forming line (DL), the layerforming-and-filling line (LFL), and the shock-frozen line (SL) were observed for their morphologies. These bacterial contaminants were classified into three groups including (1) gram-positive bacteria in rod shape, (2) gram-positive bacteria in coccus shape, and (3) gram-negative bacteria in rod shape as shown in Table 3.

| Grouping and identification of bacterial contaminants from puff pastry production
All 75 isolates collected from sanitation test in the puff pastry production process were first grouped by RAPD-PCR using each of the The bacterial contaminants from the two production processes, dough forming and layer-filling forming, not inhibited by either PP-174 or nisin A

| Detection frequency of bacterial contaminants
Each bacterial contaminant was collected from the three different production lines, the dough-forming line (DL), the layer-forming-andfilling line (LFL), and the shock-frozen line (SL). The detection frequencies were determined for each genus (DFG) and species (DFS) (

| Inhibition activity of antimicrobial substances PP-174, nisin A and the combination of PP-174 and nisin A, against adhering bacterial cells to conveyor belt coupons
Bakery products are now a convenience food source suiting the lifestyles of Thai people. They are generally considered as microbiologically safe due to the high temperatures (up to 170°C) used in the baking process. A reduced growth to <1 log CFU/coupon at 0 hr (Table 6).

| DISCUSSION
To investigate high efficiency of AMSs for a tropical region like Thailand, the microbial contaminants of a commercial bakery   production were collected. Seventy-five resistant bacterial contaminants were isolated from each compartment and grouped using RAPD analysis. Based on 16S rDNA sequence analysis, seven genus of bacterial contaminants were successfully identified. Usually, Bacillus species are proposed as bacterial contaminants in food production. They can be found during food preparation and processing, transportation and storage, or even in food handling by the worker (Bennett, Walsh, & Gould, 2013). In this study, B. cereus was detected with high DFS compared to other Bacilli species along the three production lines.  , 1995;Seiler, 1984). In addition, it can adhere to equipment surfaces and pipelines because of the hydrophobic character of the exosporium, and the presence of appendages on the spore surface (Heyndrickx, 2011). These characters explain why it was detected at high DFS compared to the other Bacilli species. Both B. subtilis and B. pumilus also appeared in the shock-frozen line, indicating that these two species can also tolerate low temperature. Therefore, contamination by these three Bacilli species at low temperature could be used as indicators for finish products regarding food safety considerations.
Among the six species of Staphylococci found, three S. aureus, S. epidermidis, and S. haemolyticus can produce extracellular staphylococcal enterotoxins (Lawley, Curtis, & Davis, 2008;Lindsay & Holden, 2004;Omoe et al., 2005). Only two species, S. aureus and S. haemolyticus were detected at the final step of the frozen line, whereas S. epidermidis disappeared. However, its toxin may still remain in the finish product and this should be considered for food safety. Staphylococci are normally found in the air, dust, water, food, humans and animals, and on environmental surfaces (Hait, Tallent, Melka, Keys, & Bennett, 2014). Staphylococci sp. also result from protein sources such as fresh beef, soybean curd, shells, and the yolk of quail eggs used to prepare fillings (Ananchaipattana et al., 2012;Goja, Ahmed, Saeed, & Dirar, 2013;Pyzik & Marek, 2012). Therefore, special care should be taken of these raw materials with strong control procedures.
Not only gram-positive bacteria were found, the gram-negative bacteria Enterobacter was also detected at the highest DFG of 37.33%.
Three species E. asburiae, E. cloacae subsp. Cloacae, and E. hormaechei were found in the production process. They are often found in soil, water, and plants (Asis & Adachi, 2004;Lau, Sulaiman, Chen, Yin, & Chan, 2013), and can live as normal flora in the gastrointestinal tract (Koth, Boniface, Chance, & Hanes, 2012). Among of these, E. hormaechei can be isolated from blood, wounds, or sputum. In addition, it is a pathogen causing nosocomial infections including sepsis (Townsend, Hurrell, Caubilla-Barron, Loc-Carrillo, & Forsythe, 2008). Only E. asburiae existed in all three production lines, the other two species disappeared in the shock-frozen line. E. asburiae survived in the shock-frozen line, subjected to a temperature of −60°C to −70°C for 13-18 min.
Normally, the optimum temperature for growth of E. asburiae is 30°C (Garrity, Brenner, Krieg, & Staley, 2007), but no previous studies reported the survival of E. asburiae at temperatures below 0°C. Our results indicated that this species can survive at subzero temperatures.
E. asburiae also displayed the highest abundance at 29% compared to the other species. Therefore, it can be used as another biomarker to indicate the quality of finish products, especially frozen puff pastry.
However, Both D. barathri and C. striatum appeared in the last line (SL), indicating that they could survive in low temperatures of −60°C to −70°C for 13-18 min. C. striatum is a common skin flora (Buchta  et al., 2004) and can be isolated from humans (Martínez-Martínez, Suárez, Rodríguez-Baño, Bernard, & Muniáin, 1997), or indoor air (Li et al., 2014). However, C. striatum has not been previously reported in bakery products. It is a nosocomial pathogen, with infection sites of the blood stream, lung, and central nervous system (Lee, Ferguson, & Sarubbi, 2005). Possible contamination might be via workers and air.
This is a serious issue to consider in the future. Another contaminant species, K. oxytoca and P. stutzeri are also an opportunistic pathogen in hospital or clinical setting (Lowe et al., 2012;Noble & Overman, 1994).
Additionally, D. barathri causes catheter-related blood stream infection (CRBSI) in human (Takahashi et al., 2015). Therefore, these strains should be concerned for food production.
Recently, several unit operations and treatments used during food processing such as drying, cold storage, modified atmosphere storage, heat treatment, and chemical preservatives have been applied to extend the shelf life of food products (Reis, Paula, Casarotti, & Penna, 2012;Schnürer & Magnusson, 2005). However, they may exert undesirable effects on the texture and flavor of the food. Thus, biopreservation, especially AMS produced by LAB has become more interesting.
Investigation of AMS produced by LAB in this study to inhibit the growth of those resistant bacterial contaminants after cleaning system applied resulted that the combination of PP-174 and nisin A at the concentration of 500 AU/ml each was successful to completely inhibit the growth of those contaminants at 0 hr and they were not detected up to 16 hr. Amin (2012) and Perez-Perez, Regalado-González, Rodríguez-Rodríguez, Barbosa-Rodríguez, and Villaseñor-Ortega (2006) proposed that no single antimicrobial agent covered all the requirements for food preservation. Nisin is a well-known bacteriocin which is declared to be GRAS (generally recognized as safe) and used in combination with other antimicrobial compounds (Siroli et al., 2016).
Nisin can bind to lipid II, the main transporter of peptidoglycan subunits from the cytoplasm to the cell wall, resulting in the prevention of proper cell wall synthesis, leading to cell death due to small intracellular metabolites released. Furthermore, lipid II was used as a docking molecule to initiate a process of membrane insertion and pore formation causing rapid cell death (Cotter, Hill, & Ross, 2005;Wiedemann et al., 2001). The mixture of different bacteriocin including lactocin 705 (17,000 AU/ml), enterocin CRL35 (17,000 AU/ml), and nisin A (2,000 IU/ml) inhibited 100% gram-positive bacteria Listeria monocytogenes and L. innocua (10 6 CFU/ml) in a broth and meat system after 24 hr of incubation (Vignolo et al., 2000). In addition, Twele et al.  Values are mean ± standard deviations of one-way ANOVA determinations performed in duplicate. Means with lowercase and uppercase letters within the same column and row, respectively, are significantly different (p < .05) determined using. solution of Carnobacterium maltaromaticum ATCC ® PTA-9380, C. maltaromaticum ATCC ® PTA-9381, and Enterococcus mundtii ATCC ® PTA-9382 as well as nisin at a concentration ranging from 500 to 5,000 IU/ ml used as a sanitizer to inhibit the pathogen Listeria monocytogenes in the food industry. It seemed that the combination of various bacteriocins successfully inhibited those gram-positive bacteria. However, many studies reported for the combination of bacteriocins and other chemical compounds to inhibit the growth of Gram-negative bacteria.
A combination of nisin A at 500, 800, or 1,000 IU/ml and 20% ethanol successfully inhibited the growth of gram-negative bacteria including E.coli O157:H7, Salmonella Typhimurium TISTR 292, and S. Enteritidis DMST 17368 to below 1 log CFU/ml at 15 min. The minimum inhibitory concentration of nisin A and ethanol was 500 IU/ml and 20% (v/v), respectively (Phongphakdee & Nitisinprasert, 2015). While a mixture of 25 μg/ml enterocin AS-48 and 0.1-2.0% polyphosphoric acid partially reduced or inhibited the growth in soybean sprouts of Salmonella enterica, E. coli O157:H7, Enterobacter aerogenes, Yersinia enterocolitica, Aeromonas hydrophila, and P. fluorescens to at least 2 log when stored at 15°C (Molinos et al., 2008). Wongsuttichote and Nitisinprasert (2009) also found that 0.5% partial purified bacteriocin PP-174 displayed 100% growth inhibition activities against aerobic mesophiles and Gram-negative bacteria coliform, but not to LAB. Additional low concentration of 17.42 mmol/L lactic acid and 6.91 mmol/L acetic acid to 0.5% PP-174 to form food preservative BAKE-SAFE-1 could completely inhibit the growth of LAB. In this study, it was successfully to form a combination of two bacteriocins, PP-174 and nisin A to inhibit both gram-positive and gram-negative bacteria which can be used as a disinfectant agent for food industry in the future.

| CONCLUSIONS
Seven genera of bacterial contaminants Bacillus, Corynebacterium, Dermacoccus, Enterobacter, Klebsiella, Pseudomonas, and Staphylococcus were successfully identified by 16SrDNA sequence analysis. Effective AMSs PP-174 and nisin A exhibited the growth inhibition of different target bacterial contaminants from puff pastry industry. However, their combination of PP-174 and nisin A provided 100% inhibition spectra against all those food spoilage and pathogenic bacteria, and can be used as a potential disinfectant agent for food industry in the future.