View Was Conducted in Terms of Cheese Quality and Safety Issues

The Application of Culture-Independent Methods in Microbial Assessment of Quality and Safety Risk Factors in Swiss Cheese and Oysters

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Qianying Yao, B.S.

Graduate Program in Food Science and Technology

The Ohio State University

2016

Dissertation Committee:

Dr. Hua Wang, advisor

Dr. Lynn Knipe

Dr. Melvin Pascall

Dr. Zhongtang Yu

Qianying Yao

2016

Abstract

Unwanted microorganisms greatly affect the quality and safety of the final food products. For instance, besides foodborne pathogens, quality defects in Swiss cheese ranging from unusual eyes, splits, off-flavors, to off-odors result in an estimated $24 million economic loss annually to the industry. Ohio has the largest Swiss cheese industry in the U.S., and to reveal microbial causative agents in Swiss cheese with quality defects has become a critical need to solve the problem for the industry. While conventional approaches were insufficient to identify the risk factors promptly and accurately, recent advancements in molecular techniques enabled in-depth investigation of potential causative agents and the development of rapid detection method for safety and quality control. To properly assess the split defect associated bacteria in Swiss cheese, a rapid detection platform for propionibacteria in dairy matrices was developed, and the microbial profiles of Swiss cheese with and without split defect were successfully evaluated. Results of these studies contributed to a comprehensive understanding of the microbial cause of split defect in Swiss cheese, enabled culture-independent methods in

microbial assessment for dairy products, and illustrated the power of 16S metagenomics in improving fundamental understanding of the microbiology in Swiss cheese.

In Chapter 1, an extensive literature review was conducted in terms of cheese quality and safety issues. With high nutritional values, Swiss cheese is a popular cheese type in the United States. The Swiss cheese microbiota consist of starter cultures and environmental microorganisms. The major safety risk in cheese is the contamination of pathogenic bacteria from raw milk or post-pasteurization handling. Non-pathogenic bacteria usually cause quality issues in cheese. Quite a few studies have been conducted on microbial identification and profiling in cheese using metagenomics. However, there is no study applying metagenomics to assessment of Swiss cheese safety and quality risks.

In Chapter 2, a rapid detection system for Propionibacterium in food matrices were successfully developed. In this study, one pair of genus-specific primers targeting the 16S ribosomal RNA of the genus Propionibacterium, and four pairs of species- specific primers targeting different protein coding genes of P. freudenreichii, P. acidipropionici, P. acnes, P. avidum, were designed and evaluated. This detection system showed no cross-reactivity with other dairy-related bacteria, indicating its utility in dairy industry.

In Chapter 3, the starter cultures and non-starter microbiota from split Swiss cheese blocks made by two different factories were analyzed. Result showed no significant difference in the relative abundance of starter cultures among split and non- split cheese, and the relative abundance of Enterobacteriaceae was significantly higher in

iii

split Swiss cheese than that of non-split samples, providing evidence that

Enterobacteriaceae could be a microbial causing candidate for split defects. The difference in relative abundance of starter cultures in Swiss cheese made by two factories was observed, which could be due to different recipe in production. A comprehensive assessment of the bacterial profiles potentially associated with Swiss cheese quality made by one factory was conducted in this study. The microbial profile of retail cheese with eyes, factory-made Swiss cheese with and without split defects, pasteurized milk, and starter cultures were extensively analyzed by 16S metagenomics. The microbiota of

Swiss cheese made in a local cheese factory showed consistency in both richness and diversity when compared to retail eye-forming cheese. The starter cultures were dominated by Streptococcus, Lactobacillus and Propionibacterium, and these genera were also dominant in Swiss cheese microbiota, while common commensal microbiota represented a very low proportion of the cheese microbiota. Using Principal Component

Analysis and a heat map with dendrograms, it was demonstrated that the split areas in

Swiss cheese were microbiologically different from eye areas, which could be due to inhomogeneous distribution of starter cultures.

In Chapter 4, retail oysters as a potential channel disseminating antibiotic resistance was assessed. A total of nine oysters belong to three types were collected. High prevalence of culturable bacteria resistant to tetracycline, cefotaxime, lincomycin, gentamicin, ciprofloxacin and ceftazidime were investigated using Brain Heart Infusion,

MacConkey and Phenylethyl Alcohol media. Meanwhile, shotgun metagenomics on

oyster microbiota was conducted by extracting genomic DNA and high throughput sequencing using Hiseq platform. The microbial profile, functional metabolic groups, and antibiotic resistome were analyzed from the sequencing data using MG-RAST platform.

High prevalence of multidrug-resistant and acriflavin-resistant genes were identified in oyster microbiota.

Result from this study provided critical information on the application of culture- independent molecular techniques to identifying and addressing important food safety and quality challenges.

Dedication

Dedicated to my parents, Guoqiang Yao and Ying Zhang for being a constant source of support throughout my lifetime.

Dedicated to my husband Peipei Tang, for his endless love and unconditional support in every aspect of my life.

Acknowledgement

I would like to express my sincere gratitude to my advisor Dr. Hua Wang, who is the mentor throughout my graduate study. She provides me with insightful suggestions towards my research, as well as valuable advice towards my career path. I would also like to thank Drs. Lynn Knipe, Melvin Pascall, and Zhongtang Yu for their service on my candidacy and dissertation committee. The valuable suggestions from my committee were of great value to my research. Besides my committee members, I would also like to thank Dr. W James Harper. The project is impossible without him.

I would also extend my gratitude to the former and present members of Dr.

Wang’s Lab: Lu Zhang, Ying Huang, Yu Li, Yang Zhou, and Wenfei Wang, for sharing their professional knowledge as well as experience with me. I really enjoyed the happy days we spent together.

I am grateful to China Scholarship Council for funding my stipend during my 4- year graduate study. Center for Innovative Food Technology, OARDC SEED for graduate students.

vii

Vita

2012…………………………………………B.S. Biological Sciences, Fudan University,

Shanghai, China

2012 to present……………………………...PhD student, Graduate Research Associate,

Department of Food Science and

Technology, The Ohio State University

Publications

Presentations at conferences:

Yao, Q and Wang, HH. (April 2016). Microbial Profiling Using 16S rDNA Metagenomics for Quality Assessment in Swiss Cheese. Poster presented at 2016 Ohio Agricultural Research and Development Center (OARDC) Annual Meeting. Wooster, OH

Yao, Q and Wang, HH. (July 2015). Swiss cheese microbial quality assessment by 16S rDNA metagenomics. Poster presented at Institute of Food Technologists (IFT) 2015 Annual Meeting & Food Expo. Chicago, IL

Yao, Q and Wang, HH. (April 2015). Microbial Profiling Using 16S rDNA Metagenomic Analysis for Swiss Cheese Assessment. Poster presented at 2015 Ohio Valley Institute of Food Technologists (OVIFT) Supplier’s Expo. West Chester, OH

viii

Yao, Q and Wang, HH. (June 2014). Rapid detection of Propionibacterium by 16S rDNA-targeted genus-specific PCR. Poster presented at 5th Ohio Valley Institute of Food Technologists (OVIFT) Symposium. Columbus, OH; and at Institute of Food Technologists (IFT) 2014 Annual Meeting & Food Expo. New Orleans, LA

Fields of Study

Major Field: Food Science and Technology

Table of Contents

Abstract…………………………………………………………………………………ii

Dedication………………………………………………………………………………v

Acknowledgement……………………………………………………………………….vi

Vita………………………………………………………………………………………vii

List of Tables…………………………………………………………………………...xiii

List of Figures………………………………………………………………………xiv

Chapter 1 Literature Background and Rationale of the Study

1.1 Introduction……………………………………………………………………...1

1.1.1 Bacteria in Swiss cheese production………………………………………3

1.1.2 Safety and quality challenges in Swiss cheese……………………………5

1.1.3 Microbial assessment

1.1.3.1 Research in propionibacteria identification in cheese matrices……..10

1.1.3.2 Culture-independent molecular techniques in studying bacterial

community…………………………………………………………...11

1.1.3.2.1 Denaturing gradient gel electrophoresis (DGGE) and

temperature gradient gel electrophoresis (TGGE)…...11

1.1.3.2.2 Terminal-restriction fragment length polymorphism (T-

RFLP) ………………………………………………..12

1.1.3.2.3 Automated ribosomal intergenic spacer analysis

(ARISA)……………………………………………...13

1.1.3.3 High-throughput sequencing and metagenomics in food

microbial ecology…………………………………………….14

1.2 Seafood………………………………………………………………………16

1.2.1 Antibiotic resistance in the ecosystem…………………………...18

1.2.2 Culture-independent molecular techniques to study the profiles of

antibiotic resistance genes in food products……………………..21

1.2.2.1 DNA microarray…………………………………………21

1.2.2.2 Shotgun metagenomics…………………………………..22

1.3 Rational and objectives………………………………………………………23

1.4 Bibliography…………………………………………………………………25

Chapter 2 A Rapid Detection System of Propionibacteria in Swiss cheese

on Genus and Species Levels Using Polymerase Chain Reaction

2.1 Abstract……………………………………………………………………....34

2.2 Introduction

2.2.1 Propionibacteria and Swiss cheese……………………………………..36

2.2.2 Studies on the identification of propionibacteria………………………..37

2.3 Materials and methods

2.3.1 DNA extraction of bacterial strains and Swiss cheese samples………..40

2.3.2 The design of genus- and species-specific primer pairs for

propionibacteria………………………………………………………….40

2.3.3 PCR amplification of genomic DNA…………………………………….41

2.4 Results

2.4.1 The identification of dairy bacteria from Swiss cheese………………….43

2.4.2 Genus- and species-specific PCR primer pairs design…………………..44

2.4.3 The specificity of Propionibacterium genus-specific PCR………………44

2.4.4 The specificity of P. freudenreichii, P. avidum, and P. acidipropionici

species-specific PCR……………………………………………………..45

2.4.5 Increased specificity for the identification of P. acnes…………………..48

2.4.6 Genus- and species-level of propionibacteria in Swiss cheese samples…48

2.5 Conclusions and discussions

2.5.1 The genetic markers in identification of propionibacteria……………….50

2.5.2 The significance of bacterial detection in Swiss cheese industry………..51

2.6 Bibliography………………………………………………………………....58

Chapter 3 Microbial Profiling using 16S metagenomics for quality assessment in

Swiss cheese

3.1 Abstract………………………………………………………………………62

xii

3.2 Introduction………………………………………………………………….64

3.3 Materials and methods

3.3.1 Sample collection and genomic DNA extraction………………………..66

3.3.2 16S rDNA gene library construction…………………………………….66

3.3.3 High throughput sequencing and analysis of sequencing data…………..68

3.4 Results

3.4.1 The comparison of Swiss cheese with split defect made by two factories

3.4.1.1 The microbial diversity of industry Swiss cheese samples………70

3.4.1.2 Starter cultures in Swiss cheese samples………………………...71

3.4.1.3 Non-starter bacteria in Swiss cheese samples……………………72

3.4.2 Comprehensive analysis of Swiss cheese from one factory

3.4.2.1 Sampling…………………………………………………………73

3.4.2.2 The microbial profile of factory-made Swiss cheese, starter

cultures and pasteurized milk…………………………………….74

3.4.2.3 Comparison of microbial profiles of retail cheese to factory-made

cheese…………………………………………………………….75

3.4.2.4 The investigation of microbial profiles of factory-made Swiss

cheese with and without split defect……………………………..77

3.5 Conclusions and discussions…………………………………………………79

3.6 Bibliography…………………………………………………………………93

xiii

Chapter 4 Retail Oyster as A Potential Channel Disseminating Antibiotic

Resistance

4.1 Abstract………………………………………………………………………96

4.2 Introduction………………………………………………………………….98

4.3 Materials and methods

4.3.1 The prevalence of bacteria with selected antibiotic resistance phenotype

4.3.1.1 Sampling………………………………………………………101

4.3.1.2 Bacterial enumeration…………………………………………101

4.3.2 Antibiotic resistance gene profiling by shotgun sequencing metagenomics

4.3.2.1 DNA preparation………………………………………………104

4.3.2.2 Antibiotic resistome analysis by shotgun sequencing

metagenomics…………………………………………………..105

4.3.2.3 Shotgun sequencing data analysis……………………………105

4.4 Results

4.4.1 The prevalence and abundance of bacteria with antibiotic resistance

phenotypes in oyster samples…………………………………………...106

4.4.2 The bacterial profile of oyster samples…………………………………108

4.4.3 Functional analysis of oyster microbiota……………………………….108

4.4.4 Analysis of antibiotic resistance genes…………………………………109

4.5 Conclusions and discussions………………………………………………116

4.6 Bibliography…………………………………………………………………76

xiv

Chapter 5 Summary and Conclusions…………………………………………119

List of References……………………………………………………………………127

List of Tables

Table 1.1 Multistate foodborne bacteria outbreaks caused by contaminated dairy

foods in 2010-2016…………………………………………………………9

Table 1.2 The major types and quantities (million tons) of US seafood supply in the

past 5 decades……………………………………………………………..17

Table 1.3 Outbreaks associated with Molluscs consumption in 1998-2014………17

Table 2.1 The recovery media, incubation conditions, colony morphology and

identification of dairy bacteria strains isolated from Swiss cheese……….43

Table 2.2 PCR primer sequences, the locations of target fragments, the sizes of

amplicons and their annealing temperatures used in this study…………..47

Table 3.1 Illumina Nextera Index used in this study………………………………...68

Table 3.2 The microbial profile of pasteurized milk and starter cultures…………...82

Table 4.1 Media and antibiotics used in the cultivation of antibiotic resistance

bacteria isolated from oyster……………………………………………103

Table 4.2 The origins and the taste of the retail live oyster samples………………104

xvi

List of Figures

Figure 2.1 Amplification of 16S rDNA 188 bp fragments in propionibacteria and

Swiss cheese samples using Pr16SF/Pr16SR primer……………………53

Figure 2.2 The effectiveness of P. freudenreichii, P. avidum and P. acidipropionici

identification using species-specific primer pairs………………………..54

Figure 2.3 The effectiveness of two species-specific primer pairs in P. acnes and

other dairy-related bacteria………………………………………………56

Figure 2.4 Amplification of genus- and species-specific fragments in Swiss cheese

samples using a) Prop16SF/Prop16SR and b)P-fF/P-Fr primers………...57

Figure 3.1 Bar chart demonstrating the microbial diversity of Swiss cheese samples

from two factories………………………………………………………83

Figure 3.2 Relative abundances of three starter cultures in split and non-split Swiss

cheese samples (a) and from two factories (b)…………………………...84

Figure 3.3 Relative abundances of non-starter bacteria in split and non-split Swiss

cheese samples (a) and in two factories (b)……………………………...85

xvii

Figure 3.4 The microbial diversity of Swiss cheese samples from the factory and

grocery stores…………………………………………………………….86

Figure 3.5 Relative abundances of three starter cultures in blind, eye and split area of

Swiss cheese made from the same batch………………………………...88

Figure 3.6 Principal Component Analysis of the microbial profile of factory-made

Swiss cheese samples (eye area vs. split area) from the same batch……90

Figure 3.7 A dual dendrogram and heat map showing top 21 bacterial genera across

the eye area and split area of Swiss cheese made from the same batch….91

Figure 4.1 Prevalence of antibiotic resistance bacteria in oyster samples by

demonstrating the colony forming unit of each sample………………111

Figure 4.2 The tree chart of bacteria abundance in oyster samples………………113

Figure 4.3 The bar chart for the relative abundance of functional gene groups in

oyster samples…………………………………………………………114

Figure 4.4 Antibiotic resistance gene patterns in retail oysters…………………….115

xviii

Chapter 1 Literature Background and Rationale of the Study

1.1 Introduction

Dairy products, including fluid milk, yogurt, cheese, butter, evaporated milk, frozen dairy products, and dry dairy products, have important roles in American diet.

According to Dietary Guidance for Americans 2015-2020, a healthy adult is recommended to consume 2-3 cups of dairy product per day depending on food intake patterns ranging from 1000-3200 calories. The health benefits of dairy products have been well explored by scientists. Dairy food consumption contributes to bone health, especially in children and adolescents. Dairy food is also associated with reduced risk of cardiovascular disease by providing important shortfall nutrients, such as calcium, potassium, Vitamin D, among others (Huth et al., 2006).

Cheese, as one type of dairy products, can be traced back to 5500 B.C.E. in

Poland. Since then, cheese has become a great source of nutrients and satiety. While the consumption of milk in the US is decreasing over the past two decades, cheese has gained

in popularity, with cheese consumption increasing from 18.9 pounds per capita in 1975 to

36.0 pounds per capita in 2014 (USDA, ERS, 2015). Originally from Switzerland, Swiss cheese, known as “cheese with eyes”, has become a popular dairy product in the United

States. The average consumption of Swiss cheese in Switzerland and France is 44 and 60 pounds respectively per capita in the year of 2005. In 2012, more than 360 million pounds of Swiss cheese was consumed in the US (Cheese Reporter, 2013). The main starter cultures for Swiss cheese making are Lactobacillus helveticus and Streptococcus thermophilus, both of which work synergistically to breakdown lactose and produce short chain fatty acids and volatile compounds. The eye-forming starter culture is

Propionibacterium freudenreichii, which is essential in Swiss cheese making and releases carbon dioxide while converting lactic acid into propionic acid.

1.1.1 Bacteria in Swiss cheese production

Throughout the process of Swiss cheese making, S. thermophilus outnumbers fastidious L. helveticus in the first twelve-hour stage of fermentation, due to the simple physiological requirements of S. thermophilus that are more easily met. When the pH and

Eh are sufficiently reduced, L. helveticus begins to grow, further breaking down amino acids and peptides via its proteolytic system. After the primary fermentation followed by salt brine and air dry, the cheese curds go through the second stage fermentation during which the cheese curds are stored in a warm room (20°C to 25°C) for three to four weeks, with the growth of P. freudenreichii. P. freudenreichii is capable of fermenting the rest of carbohydrates, mainly lactose that has not been utilized by the other two bacteria in the starter cultures, to form propionic acid, acetic acid, carbon dioxide and ATP. It can also metabolize lactic acid produced by the lactic starters to yield same final products. The carbon dioxide then will diffuse in cheese curd and when reaching some weak spots, the collected gas will form eyes.

A variety of volatile compounds are produced by the microbes during Swiss cheese fermentation, which contribute to the unique flavor of the product. Besides fermenting lactose to lactic acid, Lactobacillus spp. and Streptococcus sp. are involved in proteolysis, providing enzymes to release free amino acids. These amino acids later become the precursors of a rich profile of flavor compounds, including various volatile compounds. The catabolism of branched-chain amino acids, methionine, and phenylalanine results in the formation of key volatile compounds (Helinck et al., 2004).

Isobutyric acid provides sweaty aroma; 3-methylbutanal has a malty aroma; methionine and its derivatives have sulfur notes; phenylacetic acid and phenylacetaldehyde have a floral or fruity note (Rychlik et al., 2001). Yang et al. studied the volatile compounds of

Swiss cheese during ripening, and his results showed that the key volatile compounds were 2-butanone, 1-propanol, 2-butanol, and ethanol (Yang et al., 1994).

1.1.2 Safety and quality challenges in Swiss cheese

The quality of Swiss cheese is mainly determined by the quality of milk, the temperature for heating, and most importantly – the performance of the starter culture. At the beginning, Streptococcus thermophilus ferments amino acids and peptides are released from casein by Lactobacillus proteinases. After about 12 hours during which S. thermophilus reduces the pH and Eh, Lactobacillus grows, further breaking down amino acids and peptides via its proteolytic system. P. freudenreichii are initially present at low abundance in cheese, and its growth stage occurs in the warm room at temperature of 20

– 25°C for three to four weeks (Hutkins, 2006). The carbon dioxide produced by P. freudenreichii diffuses into cheese curds and when reaching some weak spots, the collected gas will form eyes. The metabolites from starter cultures provide rich and peculiar flavors, and special structure with eyes in Swiss cheese. The physical appearance of Swiss cheese, particularly the proper distribution of “perfect eyes” in the body of

Swiss cheese, is important to consumer preference and thus serves as a critical product quality parameter affecting market value. Defects in Swiss cheese have been studied for a long time. Split defect is sizable and clean-cut cracks or slits found in the body of the cheese (USDA, 2001), severely downgrading the market value of cheese product. It is thought to be the result of microbial activities with an excessive production of gas. The study of volatile compounds in Swiss cheese revealed that propionic acid was the most discriminating volatile compound in split and non-split Swiss cheese from three out of five factories (Castada et al, 2014).

One hypothesis is that the splits and cracks in Swiss cheese are caused by non- starter propionibacteria. The genus Propionibacterium, together with Luteococcus,

Microlunatus and Propioniferax, belongs to the family Propionibactericeae, suborder

Propionibacterineae, and class Actinobacteria. This genus of bacteria is Gram-positive rod with high GC content (53-68%). Species of Propionibacterium are anaerobic to aerotolerant, most of them are catalase positive. They not only share 90 – 94% 16S rDNA similarity (Stackebrandt et al., 2006), but also carry similar chemotaxonomic properties such as the pattern of polyamines (Busse et al., 1999), major menaquinones and fatty acids. So far, there are 14 species in genus Propionibacterium, among which there are

“classical Propinibacteria” and “cutaneous Propionibacteria (Stackebrandt et al., 2006).

The main “eye-forming” starter cultures involved in Swiss cheese manufacturing are P. freudenreichii subsp. shermanii and P. freudenreichii subsp. freudenreichii. Currently in the US, commercialized starter culture is P. freudenreichii subsp. shermanii. According to the NCBI database, current taxonomy of genus Propionibacterium includes P. acidipropionici, P. australiense, P. avidum, P. acnes, P. freudenreichii, P. granulosum,

P. humerusii, P. jensenii, P. microaerophilum, P. cyclohexanicum, P. propionicum, and

P. theonii. Sugar utilization of propionibacteria yield pyruvate, which will be reduced to produce propionate by transcarboxylase, or oxidized to producing acetate and CO2

(Piveteau et al., 1999). Results from many studies suggest the beneficial effects of propionibacteria in human gut, where it works with beneficial gut microbiota synergistically to promote the growth and adhesion of those microbiota to gut (Collado et al., 2007a), and inhibit enteric pathogens (Collado et al., 2007b). Propionibacteria can

survive harsh technological stress (Rossi et al., 2000) and physiological conditions

(Darilmaz et al., 2012), which makes it ideal as probiotics. Formation of splits in Swiss cheese has been attributed to non-starter propionibacteria since the 1960s (Park et al.,

1967). However, the exact causative agent(s) of this quality defect is still to be determined.

Meanwhile, besides the starter cultures, cheese products are also affiliated with other bacteria, which may affect the quality and safety of final cheese products. These microbes can be from milk, processing equipment, handlers, or the environment.

According to data from Centers for Disease Control and Prevention, pathogenic bacteria from raw milk or post-pasteurization handling contamination resulted in 99 confirmed foodborne outbreaks in the US between 2010-2016, including Listeria monocytogenes

(11 outbreaks), Salmonella spp. (15 outbreaks), pathogenic Escherichia coli (19 outbreaks), Staphylococcus aureus (2 outbreaks), Campylobacter spp. (54 outbreaks),

Clostridium perfringens (1 outbreak), Bacillus spp. (3 outbreak) and Yersinia enterocolitica (1 outbreak), including 16 multistate outbreaks (Table 1.1). Listeria monocytogenes is the most commonly seen pathogen involved in multistate outbreaks.

Besides pathogens, the microbiota in pasteurized milk include thermoduric bacteria and bacteria associated with post-pasteurization contamination, including psychrotrophic bacteria (Ternström et al, 1993; Fromm et al, 2004). Some non-pathogenic non-starter bacteria lead to severe quality issues such as late blowing caused by Clostridium tyrobutyricum, resulting in slits as well as abnormal flavor from butyric acid (Klijn et al,

1995). Previous studies in our lab found non-starter bacteria in commercial cheese products, such as Leuconostoc spp., Pseudomonas spp. (Wang et al, 2006), Enterococcus faecium, Escherichia coli (Li et al, 2010), etc. The potential involvement of microorganisms other than the starters in split defects of Swiss cheese is yet to be revealed.

Table 1.1 Multistate foodborne bacteria outbreaks caused by contaminated dairy foods in 2010-2016 (CDC, 2016). Year Pathogens States Illnesses Hospitalizations Deaths Contaminated Dairy Foods 2016 Listeria monocytogenes 2 2 2 1 Raw milk 2015 Listeria monocytogenes 10 30 28 3 Soft cheese (Karoun Dairies, Inc.) Ice cream products (Blue Bell 2015 Listeria monocytogenes 4 10 10 3 Creameries) 2014 Listeria monocytogenes 4 5 4 1 Cheese (Oasis Brands, Inc.) 2014 Listeria monocytogenes 2 8 7 1 Dairy products (Roos Foods, Inc.) Salmonella enterica Raw cashew cheese (The Cultured 2013 3 18 4 0 Stanley Kitchen, CA) Farmstead cheese (Crave Brothers 2013 Listeria monocytogenes 5 6 6 1 Farmstead Cheese, LLC) 2013 Listeria monocytogenes 6 9 8 1 Mexican style cheese, pasteurized

9 2013 Listeria monocytogenes 2 8 7 1 Latin style soft cheese

2013 Listeria monocytogenes 5 6 6 1 Cheese-le frere Imported Frescolina Marte Brand 2012 Listeria monocytogenes 14 23 21 5 Ricotta Salata Cheese Not 2011 Listeria monocytogenes 15 1 1 Blue-veined cheese, unpasteurized reported Escherichia coli, Shiga 2010 toxin-producing 5 38 15 0 Gouda cheese, unpasteurized O157:H7 Escherichia coli, Shiga toxin-producing Not 2010 4 8 Not reported Multiple cheeses, unpasteurized O157:H7 and reported O157:NM(H-) 2010 Listeria monocytogenes 3 6 4 1 Mexican style cheese, pasteurized 2010 Listeria monocytogenes 4 10 10 3 Ice cream, commercial

1.1.3 Microbial assessment

Conventional microbial assessments in Swiss cheese require cultivation of microorganisms, followed by a series of bacterial characterization, which are time and labor consuming. In addition, it is well recognized that only a small fraction of bacteria can be isolated and recovered using the available culture media. So there is no cultivation method that can recover and directly differentiate various bacteria associated with Swiss cheese.

1.1.3.1 Research in propionibacteria identification in cheese matrices

Several PCR-based methods have been reported to identify the species of propionibacteria. For instance, an Italian group used recA gene as a phylogenic marker to classify 6 species of propionibacteria (Rossi et al., 2006), but their method was not specific when the sample contained several species. In addition, several studies developed methods to identify the species of propionibacteria by 16S rDNA using ribotyping, restriction endonuclease analysis, multiplex-PCR and species-specific PCR

(Riedel et al., 1996; Riedel et al., 1998; Dasen et al., 1998; Rossi et al., 1999), but they also cannot differentiate propionibacteria from other starter cultures in Swiss cheese.

Multilocus sequence typing seems to be an effective tool in that it has unique markers composed of several genes for each species (Dalmasso et al., 2011). However, due to the limited and partial sequences for many species in NCBI database, it is not practical to

apply this method to propionibacteria. As a result, a rapid detection method based on current available sequences is needed.

1.1.3.2 Culture-Independent Molecular Techniques used to study bacterial community

Advancements in culture-independent molecular techniques in the past two decades provided sensitive and rapid methods while moving the scope from targeted bacteria to revealing unknown bacteria by illustrating and comparing the community profiles of samples.

1.1.3.2.1 Denaturing gradient gel electrophoresis (DGGE) and temperature gradient gel electrophoresis (TGGE)

DGGE and TGGE are rapid microbial profiling molecular approaches to analyze the genetic diversity of complex microbial populations. The PCR-amplified, same-sized fragments of genes (mostly 16S rRNA) are separated due to differences in GC content or

DNA sequence in DGGE or temperature in TGGE. Muyzer and colleagues (1993) first developed and applied this method to characterizing microbial populations from biofilms in 1992. Since then, both of these methods have been widely used in studying microbial diversity of complex samples, including food samples, for monitoring the growth and effectiveness of microbes in fermented foods (De Vero et al., 2006; Zhou et al., 2004;

Kim et al., 2010; Nielsen et al., 2005). Rantsiou et al. (2004) characterized lactic acid bacterial population dynamics during fermentation by rpoB-targeted DGGE. Ercolini et

al. (2004) investigated microbial diversity at different steps of mozzarella cheese manufacture using DGGE. Manzano et al. (2000) identified Listeria species from food sample using 16S rRNA-targeted TGGE. Although these molecular fingerprinting techniques surpassed conventional bacterial cultivation methods regarding microbial profiling, the limited separation power of only small DNA fragments hindered DGGE and TGGE from providing comprehensive and quantitative characterization of complex microbial communities. Vallaeys et al (1997) demonstrated that 16S rRNA segments from different methane-oxidizing bacteria cannot be separated by DGGE although they had substantial different sequences. The resolution and separation power of DGGE and

TGGE are also largely dependent on the targeted DNA region that is amplified. Although a comparative analysis of different DNA sequence candidates can be performed to choose a region with optimal separation, the unexpected bacteria in the samples are usually impossible to be evaluated.

1.1.3.2.2 Terminal-restriction fragment length polymorphism (T-RFLP)

T-RFLP is a fingerprinting technique for profiling bacterial communities based on the restriction site of a specific gene fragment. The DNA fragment of interest is firstly amplified by Polymerase Chain Reaction (PCR) from the total DNA of environmental microbe sample, and the fluorescent primer is then added to one end of the DNA segment, followed by restriction enzyme digestion, electrophoresis, and laser detection.

T-RFLP has been used in microbial fingerprinting in food samples, including dairy foods

(Rademaker et al., 2005, 2006; Quigley et al., 2011; Mounier et al., 2006; Rea et al.,

2007; Sánchez et al., 2006; Flórez et al., 2007), aquatic products (Hold et al., 2001;

Fernández et al., 2001; Carrera et al., 1999; Pontes et al., 2005; Hsieh et al., 2007), and meat products (Wang et al., 2010; Nesbakken et al., 1996). Although T-RFLP is a consistent, high-resolution approach to monitor the microbial community structure

(Hartmann et al., 2008), PCR amplification (Kopczynski et al., 1994; Wang & Wang,

1997; Qiu et al., 2001; Hugenholtz & Huber, 2003) and enzymatic restriction (Hartmann et al., 2007) induce biases.

1.1.3.2.3 Automated ribosomal intergenic spacer analysis (ARISA)

Internal transcribed spacer (ITS) is the DNA located between the small-subunit ribosomal RNA and large-subunit ribosomal RNA genes. In bacteria, the ITS is situated between 16S and 23S rRNA gene, and demonstrated high heterogeneity in both length and nucleotide sequence, making it an ideal target for microbial fingerprinting. In

ARISA, a community fingerprinting is obtained by adding fluorescent tag on PCR amplified ITS fragments in various length from total DNA and laser detection by automated sequencer. ARISA has been widely used in analyzing environmental and food samples, including water (Arias et al., 2006; Shade et al., 2007), plant (Torzilli et al.,

2006), as well as dairy foods (Arteau et al., 2010; Feligini et al.,2014; Chebeňová-

Turcovská et al., 2011). Due to the preferential amplification of shorter templates, a single organisms may contribute to more than one peak to the microbial profile. Since the differentiation among bacteria depends on ITS length, unrelated bacteria may have similar results, making it hard to differentiate.

1.1.3.3 High-throughput Sequencing and Metagenomics in Food Microbial Ecology

The diversity of microbial ecosystem is beyond the discrimination capacity of the techniques mentioned above. In recent years, the revolution of metagenomics empowered by next-generation sequencing has opened a new area of low-cost and high-throughput efficient characterization of the diversity of food microbial ecosystem.

The current application of high-throughput sequencing metagenomics mainly includes the characterization of microbial phylogenetic composition and the identification of functions of the microbial community. The study of phylogenetic composition is achieved by amplifying 16S rRNA gene and sequencing. By binning similar sequences into Operational Taxonomic Units (OTUs) and compared to available databases, the

OTUs are identified by taxonomic classifications to quantitatively characterize the phylogenetic composition. The functional groups in the community can be analyzed by shotgun sequencing of total DNA, followed by comparing to available databases. One of the advantages of high-throughput metagenomics is informative. Thanks to the rapid development of microbiology and fast-growing databases, a great number of bacterial sequences are available to the public, facilitating the comprehensiveness of metagenomics outcomes. Providing quantitative analysis is also one strength of metagenomics as compared to other culture-independent techniques. However, the interpretation of metagenomics results is solely based on our current knowledge, therefore 16S metagneomics is unable to characterize unknown bacterial species. An alternative approach to explore unknown genes is through functional metagenomics, by

ligating sheared genomic DNA fragments into vectors and phenotypically screening and subsequent PCR amplification, sequencing and annotation.

Since 2010, 16S rDNA metagenomics has been used to characterize the microbial communities in dairy foods. Most of these studies analyzed the microbial communities in milk, for example mastitic milk vs. healthy milk (Bhatt et al, 2012; Oikonomou et al,

2012), raw milk vs. pasteurized milk (Schmidt et al, 2012; Quigley et al, 2013).

Delcenserie et al investigated the microbiota in Belgian Herve cheese made by raw and pasteurized milk (Delcenserie et al., 2014). By employing high-throughput sequencing to

62 Irish artisanal cheeses, Quigley et al. revealed the presence of subpopulations of bacteria not previously associated with cheese (Quigley et al, 2012).

1.2 Seafood

Seafood, including fish and shellfish, is becoming more popular in recent years.

In 2011, the US seafood supply was 6.82 million tons, among which 1.08 million tons was Molluscs as shown in Table 1.2. Seafood provides important nutrients, such as unsaturated fatty acids, Vitamin A and minerals. However, the potential health risks of seafood consumption cannot be ignored. Because of the biomagnification of persistent toxicants in freshwater and marine food chain, seafood tends to concentrate mercury, selenium and lead in their bodies, providing an important pathway for human exposure

(Grandjean et al., 1992). Besides toxic chemical compounds, pathogenic microorganisms are another threat in that many types of seafood, like oysters, that are consumed raw with minimal processing. For example, during the year of 1998 to 2014, a total of 228 outbreaks associated with eating oysters were recorded, in which 2567 illnesses, 69 hospitalizations and 1 death (Table 1.3). Most outbreaks were caused by Norovirus and

Vibrio, but also Campylobacter and Salmonella.

It is also recognized that the food chain has an important role in transmitting antibiotic resistance to the general public. Duran and Marshall (2005) examined retail ready-to-eat shrimp and discovered various antibiotic resistant bacteria. This study investigated antibiotic resistance in oyster samples. Because oysters are mostly consumed raw, the antibiotic resistant bacteria and resistance genes detected in the samples represent those transmitted to human through food consumption.

Table 1.2 The major types and quantities (million tons) of US seafood supply in the past

5 decades (Food and Agriculture Organization of the United Nations, 2014).

Year Freshwater Fish Demersal Fish Pelagic Fish Crustaceans Molluscs 1961 0.29 0.69 0.47 0.34 0.60 1971 0.26 0.75 0.68 0.51 0.71 1981 0.32 1.06 0.76 0.57 1.18 1991 0.49 2.28 0.87 0.82 0.85 2001 0.92 2.08 0.73 1.36 1.00 2011 1.38 1.34 1.03 1.89 1.08

Table 1.3 Outbreaks associated with Molluscs consumption in 1998-2014 (CDC, 2015).

Molluscs Type Outbreaks Illnesses Hospitalizations Deaths Oysters 228 2567 69 1 Clams 45 675 33 0 Scallops 13 172 3 0 Mussels 27 114 27 0 Squids 10 55 0 0 Octopus 5 39 1 0 Total 318 3476 122 1

1.2.1 Antibiotic resistance in the ecosystem

Antibiotic resistant bacteria and resistance-encoding genes have now been detected in various samples. Targeted metagenomic analyses of rigorously authenticated ancient DNA from 30,000-year-old Beringian permafrost sediments identified a highly diverse collection of genes encoding resistance to β-lactam, tetracycline and glycopeptide antibiotics (D’Costa et al., 2011). With the discovery of antibiotics more than 70 years ago, the drug innovation and implementation in human and animal health are accompanied by the emergence of resistance microbes, although antibiotic resistance is ancient.

Antibiotic resistance in pathogens and opportunistic pathogens has received great attention and are extensively examined by researchers. The rapid emergence of antibiotic- resistant pathogens has become a major public health concern. Nosocomial infections in both Gram-positive and Gram-negative pathogens, such as methicillin-resistant

Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococci (VRE), have been widely reported (Baba et al., 2002; Enright et al., 2002; Frieden et al., 1993; Cetinkaya et al., 2000). In recent years, the crucial roles of commensal bacteria in antibiotic resistance ecology, from emergence, dissemination, amplification to persistence has been recognized, since pathogens only account for a small percentage of microbiota while 99% of microbes are commensal bacteria (Wang et al., 2009; Vartoukian et al., 2010).

One of the threats of antibiotic resistant bacteria is that the antibiotic resistance genes may be transferred from one cell to another, especially between microorganisms

with compatible genetic backgrounds. As commensal bacteria are high in abundance and genetic diversity, the chance of antibiotic resistance gene transmission between commensal bacteria and from commensal to pathogens are much higher than between pathogens. Numerous studies demonstrated that foodborne commensal bacteria are an important avenue in the dissemination of antibiotic resistance genes as well as antibiotic resistant bacteria (Perreten et al., 1997; Wang et al., 2006). High prevalence of antibiotic resistance genes and corresponding antibiotic resistant commensals were detected and identified in retail foods, including cheese (cheddar, Swiss, mozzarella and Colby), yogurt, raw milk, shrimps, meats (pork chop, deli turkey and deli beef), mushrooms and spinach, and the potential of horizontal antibiotic resistance gene transfer from these bacteria to oral residential bacterium was also confirmed (Wang et al., 2006).

The types of antibiotics are classified according to their mode of action.

Bactericidal antibiotics usually eliminate bacteria by preventing cell wall synthesis while bacteriostatic antibiotics limit the growth of bacteria by interfering with protein synthesis,

DNA replication, or other cellular metabolism (Pankey et al., 2004). As a bactericidal antibiotic, β-lactam antibiotics target the penicillin-binding proteins, which are involved in the cross-linking of the bacterial cell wall, leading to death of the bacterial cell due to osmotic instability or autolysis (Waxman et al., 1983; Tomasz et al., 1979). This class of antibiotics include penicillins, cephalosporins, carbapenems, monobactams, and β- latamase inhibitor. Aminoglycosides are another type of bactericidal antibiotics. They bind to the 30s ribosomal subunit inside the cell and cause a misreading of the genetic code, leading to the interruption of normal bacterial protein synthesis (Davis, 1987). This

class of antibiotics include gentamycin, tobramycin, streptomycin, kanamycin, among others. Quinolones is another class of bactericidal antibiotics. It bind to the DNA gyrase-

DNA complex and interrupt a process that leads to the negative supercoiling of bacterial

DNA, leading to defects in the necessary supercoiling, making the bacteria unable to multiply and survive (Hooper, 1995). As bacteriostatic antibiotics, tetracycline reversibly binds to receptors on the 30S ribosomal subunit, preventing attachment of aminoacyl- tRNA to the RNA-ribosome complex, thus inhibiting the addition of amino acids to the elongation peptide chain (Chopra et al., 2001). It works against a wide range of Gram- positive and Gram-negative bacteria. Another bacteriostatic antibiotics class is macrolides, which reversibly bind to 50S subunit of the ribosomes and inhibit transpeptidation and translocation processes, resulting in premature detachment of incomplete polypeptide chains (Mazzei et al., 1993). With a narrow spectrum, it may be bactericidal at high concentrations or against highly susceptible bacterial organisms

(Omura, 2002). This class of antibiotics include erythromycin, tylosin, spiramycin, among others.

1.2.2 Culture-independent molecular techniques to study the profiles of antibiotic resistance genes in food products

Existing methods to identify specific AR gene abundance in food samples include characterization and cultivation on media with antibiotics, amplification of AR genes by polymerase chain reaction and determination of AR gene by sequencing. These methods are time-consuming, low efficient and unable to draw a comprehensive picture of all types of AR genes in food samples. High throughput methods have been applied to study multiply even hundreds of antibiotic resistance genes (resistome) in one sample simultaneously (Hu et al., 2013; Zhang et al., 2011).

1.2.2.1 DNA microarray

In DNA microarray, a large amount of oligonucleotide probes are designed and attached to a solid support. By hybridization of a DNA sample to the probes, the specific gene expressions of the sample will be determined in a single assay. Microarray was a promising method in studying the antibiotic resistance gene profiles of microbiota.

Perrenten et al. (2005) developed a disposal microarray to detect up to 90 antibiotic resistance genes of Gram-positive bacteria. Clinically, microarray has been widely used to investigate the antibiotic resistance genes of human pathogens S. aureus (Zhu et al.,

2007; Strommerger et al., 2007), Salmonella enterica (Navarre et al., 2005; Chen et al.,

2005), E. coli (Delaquis et al., 2007; Bruant et al., 2006), Helicobacter pylori (Israel et al., 2001).

1.2.2.2 Shotgun metagenomics

By applying shotgun sequencing with metagenomics, Hu et al. (2013) investigated the antibiotic resistance genes of human gut microbiota in metagenome-wide scale, and identified 1093 antibiotic resistance genes with the highest number in Chinese individual, followed by Danish and Spanish individuals. Metagenomics has been used to study microbiota of oysters in recent years across the world. The stomach and gut content microbiomes of Crassostrea virginica were examine by King et al. (2012), and their results showed Shewanella and a Chloroflexi strain dominated oyster from two different geographic sources. Proteobacteria predominated in Pacific oyster (Crassostrea gigas) samples, while Psychrilyobacter and Bacteroidetes were also found in high amount

(Fernandez-Piquer et al., 2012). Another metagenomics assessment of Crassostrea virginicanative to Apalachicola Bay demonstrated that the oyster microbiome was predominated by Cyanobacteria and Proteobacteria, with the presence of pathogenic and symbiotic Photobacterium spp. predominated in oyster tissues (Chauhan et al., 2014).

However, there is no comprehensive investigation on the antibiotic resistome in oyster.

1.3 Rationale and objectives

Although splits and cracks in Swiss cheese cause significant financial loss to the industry and it was hypothesized that non-starter propionibacteria might have contributed to the quality problem, there is no solid evidence and proper detection methods that illustrate the microbial profile of Swiss cheese accurately and therefore the industry is still suffering from major damages without a way to detect and prevent the defects from happening. In the case of seafood safety, although emerging evidences indicated that ready-to consume seafood could be carriers for antibiotic resistant bacteria, the real magnitude and spectrum of AR affiliated with seafood products have not been fully elucidated. Therefore, the objectives of the proposed study were:

(1) To develop a rapid method for assessing targeted microorganism(s) in Swiss cheese;

(2) To use metagenomics to assess profiles of microbiota associated with Swiss cheese quality defects;

(3) To use metagenomics to assess antibiotic resistome associated with retail oysters.

This study is significant because it aims at revealing and validating the causative microbial agents against Swiss cheese split defects and food safety risk factors in seafood. Results from the study will: 1) the microbial assessment and detection platform we will develop is rapid, accurate, sensitive and economic in identifying microbial

profiles in the complex cheese matrices, potentially to be adopted by the Swiss cheese industry for quality control; 2) be the leading study using high-throughput metagenomics to characterize microbial profile in Swiss cheese as well as AR genes in oyster; 3) address a main industrial challenge with significant financial impact.

1.4 Bibliography

Arias, C. R., Abernathy, J. W., & Liu, Z. (2006). Combined use of 16S ribosomal DNA and automated ribosomal intergenic spacer analysis to study the bacterial community in catfish ponds. Letters in applied microbiology, 43(3), 287-292.

Arteau, M., Labrie, S., & Roy, D. (2010). Terminal-restriction fragment length polymorphism and automated ribosomal intergenic spacer analysis profiling of fungal communities in Camembert cheese. International dairy journal, 20(8), 545-554.

Baba, T., Takeuchi, F., Kuroda, M., Yuzawa, H., Aoki, K. I., Oguchi, A., ... & Kuroda, H. (2002). Genome and virulence determinants of high virulence community- acquired MRSA. The Lancet, 359(9320), 1819-1827.

Bhatt, V. D., Ahir, V. B., Koringa, P. G., Jakhesara, S. J., Rank, D. N., Nauriyal, D. S., ... & Joshi, C. G. (2012). Milk microbiome signatures of subclinical mastitis‐ affected cattle analysed by shotgun sequencing. Journal of applied microbiology, 112(4), 639-650.

Busse, H. J., & Schumann, P. (1999). Polyamine profiles within genera of the class Actinobacteria with LL-diaminopimelic acid in the peptidoglycan. International Journal of Systematic and Evolutionary Microbiology, 49(1), 179-184.

Carrera, E., Garcia, T., Cespedes, A., Gonzalez, I., Fernandez, A., Hernandez, P. E., & Martin, R. (1999). Salmon and Trout Analysis by PCR‐RFLP for Identity Authentication. Journal of food science, 64(3), 410-413.

Castada, H. Z., Wick, C., Taylor, K., & Harper, W. J. (2014). Analysis of Selected Volatile Organic Compounds in Split and Nonsplit Swiss Cheese Samples Using Selected‐Ion Flow Tube Mass Spectrometry (SIFT‐MS). Journal of food science, 79(4), C489-C498.

Cetinkaya, Y., Falk, P., & Mayhall, C. G. (2000). Vancomycin-resistant enterococci. Clinical microbiology reviews, 13(4), 686-707.

Chebeňová-Turcovská, V., Ženišová, K., Kuchta, T., Pangallo, D., & Brežná, B. (2011). Culture-independent detection of microorganisms in traditional Slovakian bryndza cheese. International journal of food microbiology, 150(1), 73-78.

Chopra, I., & Roberts, M. (2001). Tetracycline antibiotics: mode of action, applications, molecular biology, and epidemiology of bacterial resistance. Microbiology and molecular biology reviews, 65(2), 232-260.

Collado, M. C., Meriluoto, J., & Salminen, S. (2007a). Development of new probiotics by strain combinations: is it possible to improve the adhesion to intestinal mucus? Journal of dairy science, 90(6), 2710-2716.

Collado, M. C., Meriluoto, J., & Salminen, S. (2007b). In vitro analysis of probiotic strain combinations to inhibit pathogen adhesion to human intestinal mucus. Food Research International, 40(5), 629-636.

Dairy products: Per capita consumption, United States (Annual). (2015, September 2). Retrieved April 26, 2016, from http://www.ers.usda.gov/data-products/dairy- data.aspx

Dalmasso, M., Nicolas, P., Falentin, H., Valence, F., Tanskanen, J., Jatila, H., ... & Thierry, A. (2011). Multilocus sequence typing of Propionibacterium freudenreichii. International journal of food microbiology, 145(1), 113-120.

Darilmaz, D. O., & Beyatli, Y. (2012). Acid-bile, antibiotic resistance and inhibitory properties of propionibacteria isolated from Turkish traditional home-made cheeses. Anaerobe, 18(1), 122-127.

Dasen, G., Smutny, J., Teuber, M., & Meile, L. (1998). Classification and identification of propionibacteria based on ribosomal RNA genes and PCR. Systematic and applied microbiology, 21(2), 251-259.

Davis, B. D. (1987). Mechanism of bactericidal action of aminoglycosides. Microbiological reviews, 51(3), 341.

D’Costa, V. M., King, C. E., Kalan, L., Morar, M., Sung, W. W., Schwarz, C., ... & Golding, G. B. (2011). Antibiotic resistance is ancient. Nature, 477(7365), 457- 461.

De Vero, L., Gala, E., Gullo, M., Solieri, L., Landi, S., & Giudici, P. (2006). Application of denaturing gradient gel electrophoresis (DGGE) analysis to evaluate acetic acid bacteria in traditional balsamic vinegar. Food Microbiology, 23(8), 809- 813.

Delcenserie, V., Taminiau, B., Delhalle, L., Nezer, C., Doyen, P., Crevecoeur, S., ... & Daube, G. (2014). Microbiota characterization of a Belgian protected designation of origin cheese, Herve cheese, using metagenomic analysis. Journal of dairy science, 97(10), 6046-6056.

DuráN, G. M., & Marshall, D. L. (2005). Ready-to-eat shrimp as an international vehicle of antibiotic-resistant bacteria. Journal of Food Protection®, 68(11), 2395-2401.

Enright, M. C., Robinson, D. A., Randle, G., Feil, E. J., Grundmann, H., & Spratt, B. G. (2002). The evolutionary history of methicillin-resistant Staphylococcus aureus (MRSA). Proceedings of the National Academy of Sciences, 99(11), 7687-7692.

Ercolini, D., Mauriello, G., Blaiotta, G., Moschetti, G., & Coppola, S. (2004). PCR– DGGE fingerprints of microbial succession during a manufacture of traditional water buffalo mozzarella cheese. Journal of Applied Microbiology, 96(2), 263- 270.

Fernández, A., Garcia, T., Asensio, L., Rodriguez, M. A., Gonzalez, I., & Hernandez, P. E. (2001). PCR‐RFLP Analysis of the Internal Transcribed Spacer (ITS) Region for Identification of 3 Clam Species. Journal of food science, 66(5), 657-661.

Flórez, A. B., Álvarez-Martín, P., López-Díaz, T. M., & Mayo, B. (2007). Morphotypic and molecular identification of filamentous fungi from Spanish blue-veined Cabrales cheese, and typing of Penicillium roqueforti and Geotrichum candidum isolates. International dairy journal, 17(4), 350-357.

Fromm, H. I., & Boor, K. J. (2004). Characterization of pasteurized fluid milk shelf‐life attributes. Journal of food science, 69(8), M207-M214.

Feligini, M., Brambati, E., Panelli, S., Ghitti, M., Sacchi, R., Capelli, E., & Bonacina, C. (2014). One-year investigation of Clostridium spp. occurrence in raw milk and curd of Grana Padano cheese by the automated ribosomal intergenic spacer analysis. Food Control, 42, 71-77.

Food and Agriculture Organization of the United Nations Statistics Division. (n.d.).Retrieved June 4, 2016, from http://faostat3.fao.org

Frieden, T. R., Munsiff, S. S., Williams, G., Faur, Y., Kreiswirth, B., Low, D. E., ... & Eisner, W. (1993). Emergence of vancomycin-resistant enterococci in New York City. The Lancet, 342(8863), 76-79.

Grandjean, P., Weihe, P., Jørgensen, P. J., Clarkson, T., Cernichiari, E., & Viderø, T. (1992). Impact of maternal seafood diet on fetal exposure to mercury, selenium, and lead. Archives of Environmental Health: An International Journal, 47(3), 185-195.

Hartmann, M., Enkerli, J., & Widmer, F. (2007). Residual polymerase activity‐induced bias in terminal restriction fragment length polymorphism analysis. Environmental microbiology, 9(2), 555-559.

Hartmann, M., & Widmer, F. (2008). Reliability for detecting composition and changes of microbial communities by T-RFLP genetic profiling. FEMS microbiology ecology, 63(2), 249-260.

Helinck, S., Le Bars, D., Moreau, D., & Yvon, M. (2004). Ability of thermophilic lactic acid bacteria to produce aroma compounds from amino acids. Applied and environmental microbiology, 70(7), 3855-3861.

Hold, G. L., Russell, V. J., Pryde, S. E., Rehbein, H., Quinteiro, J., Rey-Mendez, M., ... & Rosa, C. (2001). Validation of a PCR-RFLP based method for the identification of salmon species in food products. European Food Research and Technology, 212(3), 385-389.

Hooper, D. C. (1995). Quinolone mode of action. Drugs, 49(2), 10-15.

Hsieh, H. S., Chai, T. J., & Hwang, D. F. (2007). Using the PCR–RFLP method to identify the species of different processed products of billfish meats. Food Control, 18(4), 369-374.

Hugenholtz, P., & Huber, T. (2003). Chimeric 16S rDNA sequences of diverse origin are accumulating in the public databases. International Journal of Systematic and Evolutionary Microbiology, 53(1), 289-293.

Huth, P. J., DiRienzo, D. B., & Miller, G. D. (2006). Major scientific advances with dairy foods in nutrition and health. Journal of Dairy Science, 89(4), 1207-1221.

Hutkins, R. W. (2008). Microbiology and technology of fermented foods (Vol. 22). John Wiley & Sons.

Kim, T. W., Lee, J. H., Park, M. H., & Kim, H. Y. (2010). Analysis of bacterial and fungal communities in Japanese-and Chinese-fermented soybean pastes using nested PCR–DGGE. Current microbiology, 60(5), 315-320.

Klijn, N., Nieuwenhof, F. F., Hoolwerf, J. D., Van Der Waals, C. B., & Weerkamp, A. H. (1995). Identification of Clostridium tyrobutyricum as the causative agent of late blowing in cheese by species-specific PCR amplification. Applied and Environmental Microbiology, 61(8), 2919-2924.

Kopczynski, E. D., Bateson, M. M., & Ward, D. M. (1994). Recognition of chimeric small-subunit ribosomal DNAs composed of genes from uncultivated microorganisms. Applied and Environmental Microbiology, 60(2), 746-748.

Li, X., & Wang, H. H. (2010). Tetracycline resistance associated with commensal bacteria from representative ready-to-consume deli and restaurant foods. Journal of Food Protection®, 73(10), 1841-1848.

Manzano, M., Cocolin, L., Cantoni, C., & Comi, G. (2000). Temperature gradient gel electrophoresis of the amplified product of a small 16S rRNA gene fragment for the identification of Listeria species isolated from food. Journal of Food Protection®, 63(5), 659-661.

Mazzei, T., Mini, E., Novelli, A., & Periti, P. (1993). Chemistry and mode of action of macrolides. Journal of Antimicrobial Chemotherapy, 31(suppl C), 1-9.

Mounier, J., Goerges, S., Gelsomino, R., Vancanneyt, M., Vandemeulebroecke, K., Hoste, B., ... & Cogan, T. M. (2006). Sources of the adventitious microflora of a smear‐ripened cheese. Journal of applied microbiology, 101(3), 668-681.

Muyzer, G., De Waal, E. C., & Uitterlinden, A. G. (1993). Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Applied and environmental microbiology, 59(3), 695-700.

Nesbakken, T., Kapperud, G., & Caugant, D. A. (1996). Pathways of Listeria monocytogenes contamination in the meat processing industry. International journal of food microbiology, 31(1), 161-171.

Nielsen, D. S., Hønholt, S., Tano‐Debrah, K., & Jespersen, L. (2005). Yeast populations associated with Ghanaian cocoa fermentations analysed using denaturing gradient gel electrophoresis (DGGE). Yeast, 22(4), 271-284.

Oikonomou, G., Machado, V. S., Santisteban, C., Schukken, Y. H., & Bicalho, R. C. (2012). Microbial diversity of bovine mastitic milk as described by pyrosequencing of metagenomic 16s rDNA. PLoS One, 7(10), e47671.

Omura, S. (Ed.). (2002). Macrolide antibiotics: chemistry, biology, and practice. Academic press.

Pankey, G. A., & Sabath, L. D. (2004). Clinical relevance of bacteriostatic versus bactericidal mechanisms of action in the treatment of Gram-positive bacterial infections. Clinical infectious diseases, 38(6), 864-870.

Park, H. S., Reinbold, G. W., & Hammond, E. G. (1967). Role of propionibacteria in split defect of Swiss cheese. Journal of dairy science, 50(6), 820-823.

Per Capita Cheese Consumption Hit Record 33.51 Pounds In 2012. (2013, September 27). Cheese Reporter. Retrieved February 2, 2014, from http://npaper- wehaa.com/cheese-reporter/2013/09/s3/#?article=2030100

Perreten, V., Vorlet-Fawer, L., Slickers, P., Ehricht, R., Kuhnert, P., & Frey, J. (2005). Microarray-based detection of 90 antibiotic resistance genes of gram-positive bacteria. Journal of clinical microbiology, 43(5), 2291-2302.

Piveteau, P. (1999). Metabolism of lactate and sugars by dairy propionibacteria: a review. Le Lait, 79(1), 23-41.

Pontes, T., D'Amelio, S., Costa, G., & Paggi, L. (2005). Molecular characterization of larval anisakid nematodes from marine fishes of Madeira by a PCR-based approach, with evidence for a new species. Journal of Parasitology, 91(6), 1430- 1434.

Qiu, X., Wu, L., Huang, H., McDonel, P. E., Palumbo, A. V., Tiedje, J. M., & Zhou, J. (2001). Evaluation of PCR-generated chimeras, mutations, and heteroduplexes with 16S rRNA gene-based cloning. Applied and Environmental Microbiology, 67(2), 880-887.

Quigley, L., O'Sullivan, O., Beresford, T. P., Ross, R. P., Fitzgerald, G. F., & Cotter, P. D. (2012). High-throughput sequencing for detection of subpopulations of bacteria not previously associated with artisanal cheeses. Applied and environmental microbiology, 78(16), 5717-5723.

Quigley, L., McCarthy, R., O'Sullivan, O., Beresford, T. P., Fitzgerald, G. F., Ross, R. P., ... & Cotter, P. D. (2013). The microbial content of raw and pasteurized cow milk as determined by molecular approaches. Journal of dairy science, 96(8), 4928- 4937.

Rademaker, J. L. W., Hoolwerf, J. D., Wagendorp, A. A., & Te Giffel, M. C. (2006). Assessment of microbial population dynamics during yoghurt and hard cheese fermentation and ripening by DNA population fingerprinting. International dairy journal, 16(5), 457-466.

Rantsiou, K., Comi, G., & Cocolin, L. (2004). The rpoB gene as a target for PCR-DGGE analysis to follow lactic acid bacterial population dynamics during food fermentations. Food Microbiology, 21(4), 481-487.

Rea, M. C., Görges, S., Gelsomino, R., Brennan, N. M., Mounier, J., Vancanneyt, M., ... & Cogan, T. M. (2007). Stability of the biodiversity of the surface consortia of Gubbeen, a red-smear cheese. Journal of dairy science, 90(5), 2200-2210.

Riedel, K. H., & Britz, T. J. (1996). Justification of the “classical” Propionibacterium species concept by ribotyping. Systematic and applied microbiology, 19(3), 370- 380.

Rledel, K. H., Wlngfield, B. D., & Britz, T. J. (1998). Identification of Classical Propionibacterium Species Using 16S rDNA—Restriction Fragment Length Polymorphisms. Systematic and applied microbiology, 21(3), 419-428.

Rossi, F., Torriani, S., & Dellaglio, F. (1999). Genus-and species-specific PCR-based detection of dairy propionibacteria in environmental samples by using primers targeted to the genes encoding 16S rRNA. Applied and environmental microbiology, 65(9), 4241-4244.

Rossi, F., Dellaglio, F., Peluso, M., & Torriani, S. (2000). Dairy propionibacteria: occurrence, resistance to technological stresses and antagonistic properties. Industrie Alimentari, 39(392), 553-557.

Rossi, F., Dellaglio, F., & Torriani, S. (2006). Evaluation of recA gene as a phylogenetic marker in the classification of dairy propionibacteria. Systematic and applied microbiology, 29(6), 463-469.

Rychlik, M., & Bosset, J. O. (2001). Flavour and off-flavour compounds of Swiss Gruyere cheese. Evaluation of potent odorants. International Dairy Journal, 11(11), 895-901.

Sánchez, J. I., Rossetti, L., Martínez, B., Rodríguez, A., & Giraffa, G. (2006). Application of reverse transcriptase PCR-based T-RFLP to perform semi- quantitative analysis of metabolically active bacteria in dairy fermentations. Journal of microbiological methods, 65(2), 268-277.

Schmidt, V. S., Kaufmann, V., Kulozik, U., Scherer, S., & Wenning, M. (2012). Microbial biodiversity, quality and shelf life of microfiltered and pasteurized extended shelf life (ESL) milk from Germany, Austria and Switzerland. International journal of food microbiology, 154(1), 1-9.

Shade, A., Kent, A. D., Jones, S. E., Newton, R. J., Triplett, E. W., & McMahon, K. D. (2007). Interannual dynamics and phenology of bacterial communities in a eutrophic lake. Limnology and Oceanography, 52(2), 487-494.

Stackebrandt, E., Cummins, C. S., & Johnson, J. L. (2006). Family propionibacteriaceae: the genus Propionibacterium. In The prokaryotes (pp. 400-418). Springer New York.

Ternström, A., Lindberg, A. M., & Molin, G. (1993). Classification of the spoilage flora of raw and pasteurized bovine milk, with special reference to Pseudomonas and Bacillus. Journal of Applied Bacteriology, 75(1), 25-34.

Tomasz, A. (1979). The mechanism of the irreversible antimicrobial effects of penicillins: how the beta-lactam antibiotics kill and lyse bacteria. Annual Reviews in Microbiology, 33(1), 113-137.

Torzilli, A. P., Sikaroodi, M., Chalkley, D., & Gillevet, P. M. (2006). A comparison of fungal communities from four salt marsh plants using automated ribosomal intergenic spacer analysis (ARISA). Mycologia, 98(5), 690-698.

United States standards for grades of Swiss cheese, Emmentaler cheese. (2001, February 22). Retrieved February 3, 2015, from https://www.ams.usda.gov/sites/default/files/media/Swiss_Cheese,_Emmentaler_ Cheese_Standard[1].pdf

Vallaeys, T., Topp, E., Muyzer, G., Macheret, V., Laguerre, G., Rigaud, A., & Soulas, G. (1997). Evaluation of denaturing gradient gel electrophoresis in the detection of 16S rDNA sequence variation in rhizobia and methanotrophs. FEMS Microbiology Ecology, 24(3), 279-285.

Vartoukian, S. R., Palmer, R. M., & Wade, W. G. (2010). Strategies for culture of ‘unculturable’bacteria. FEMS microbiology letters, 309(1), 1-7.

Wang, G. C., & Wang, Y. (1997). Frequency of formation of chimeric molecules as a consequence of PCR coamplification of 16S rRNA genes from mixed bacterial genomes. Applied and Environmental Microbiology, 63(12), 4645-4650.

Wang, H. H., Manuzon, M., Lehman, M., Wan, K., Luo, H., Wittum, T. E., ... & Bakaletz, L. O. (2006). Food commensal microbes as a potentially important avenue in transmitting antibiotic resistance genes. FEMS Microbiology Letters, 254(2), 226-231.

Wang, H. W., Jaykus, L. A., Wang, H. H., & Schlesinger, L. S. (2009). Commensal bacteria, microbial ecosystems, and horizontal gene transmission: adjusting our focus for strategic breakthroughs against antibiotic resistance. Food-borne microbes: shaping the host ecosystem, 267-281.

Wang, Q., Zhang, X., Zhang, H. Y., Zhang, J., Chen, G. Q., Zhao, D. H., ... & Liao, W. J. (2010). Identification of 12 animal species meat by T-RFLP on the 12S rRNA gene. Meat science, 85(2), 265-269.

Waxman, D. J., & Strominger, J. L. (1983). Penicillin-binding proteins and the mechanism of action of beta-lactam antibiotics1. Annual review of biochemistry, 52(1), 825-869.

Yang, W. T., & Min, D. B. (1994). Dynamic headspace analyses of volatile compounds of Cheddar and Swiss cheese during ripening. Journal of food science, 59(6), 1309-1312.

Zhou, J., Liu, X., Jiang, H., & Dong, M. (2009). Analysis of the microflora in Tibetan kefir grains using denaturing gradient gel electrophoresis. Food microbiology, 26(8), 770-775.

Chapter 2 A Rapid Detection Method for Propionibacteria in Swiss Cheese at

Genus and Species Levels Using Polymerase Chain Reaction

2.1 Abstract

The objective of this study was to develop a polymerase chain reaction (PCR) system by using genus- and species-specific primer pairs for rapid and specific identification of propionibacteria, mainly Propionibacterium. freudenreichii, P. acidipropionici, P. acnes, and P. avidum, in multi-bacterial food samples. P. freudenreichii is the eye-forming starter culture in Swiss cheese manufacturing. Due to its importance in the quality of “cheese with eyes”, rapid detection methods to determine the cause of microbial defects like splits and cracks by identifying the species of propionibacteria are desirable in industrial quality control. Non-starter propionibacteria are considered as one of the causing agents of Swiss cheese quality defect. Efficient and cost-effective ways of identifying non-starter propionibacteria is of great need to the cheese industry. In this study, one pair of genus-specific primer targeting 16S ribosomal

RNA for detecting the genus Propionibacterium, and four pairs of species-specific

primers targeting different constitutive genes for detecting of P. freudenreichii, P. acidipropionici, P. acnes, P. avidum, were designed and evaluated. This method provides a rapid and effective way for assessing Propionibacterium in food matrices, and it has no cross-reactivity with other common foodborne bacteria.

2.2 Introduction

2.2.1 Propionibacteria and Swiss cheese

The dominant microorganisms in Swiss cheese are starter cultures, including

Lactobacillus spp., Streptococcus thermophilus, and propionibacteria. They proliferate during cheese making, and metabolize amino acids, sugars, and proteins, which provide a rich flavor and nutritional profile in Swiss cheese. Defects in Swiss cheese include discoloration, poor eye development, unusual eyes, splits and cracks, off-flavors, and off- odors. Some are caused by unidentified microorganisms. Previous studies have identified numerous non-starter microorganisms in commercial cheese products, such as

Leuconostoc sp., Pseudomonas sp. (Wang et al., 2006), Enterococcus faecium,

Escherichia coli (Li et al., 2010), etc. Non-starter propionibacteria is also a concern to the industry. Formation of splits in Swiss cheese has been attributed to non-starter propionibacteria since 1960s (Park et al., 1967). These non-starter microorganisms may be introduced to cheese from the milk, cultures, the equipment, the environment, or workers during manufacturing, and may have an impact on the microbial interaction in cheese. They may also affect the property and quality of eyes in Swiss cheese, so the identification of microorganisms is in great need by the cheese industry.

Propionibacterium freudenreichii is the eye-forming starter culture in Swiss cheese. It releases carbon dioxide while converting lactic acid into propionic acid. The genus Propionibacterium belongs to the family Propionibactericeae, suborder

Propionibacterineae, and class Actinobacteria. Generally speaking, this genus of bacteria

is Gram positive rods with high GC content (53-68%). Propionibacterium is anaerobic to aerotolerant, and the species in this genus are catalase positive. They share 90-94% 16S ribosomal RNA similarity (Stackebrandt et al., 2006), and carry similar chemotaxonomic properties such as the pattern of polyamines (Busse et al., 1999), major menaquinones and fatty acids. There are 14 identified species in genus Propionibacterium,including

“classical propinibacteria” and “cutaneous propionibacteria” (Stackebrandt et al., 2006).

Besides the main “eye-forming” starter culture P. freudenreichii, P. acidipropionici is one of the dairy-related propionibacteria, and it was found in Leerdammer cheese samples, a Swiss-style cheese mainly from Netherland (Britz et al., 1994). The cutaneous propionibacteria – P. acnes and P. avidum are common inhabitants of the skin, and they may cause skin damage (Zoubouli, 2004; Estoppey et al., 1997).

2.2.2 Studies on the identification of propionibacteria

Due to the popularity and high economic value of Swiss cheese across the world, to make cheese with monitored propionibacteria growth is important to the food industry.

The conventional method for propionibacteria identification involved the cultivation of microorganisms followed by morphological, biochemical and serological analysis (Malik et al., 1968). This labor- and time-consuming method had difficulties in identifying among species, and it was substituted with molecular fingerprinting in 1990s. The

Restriction Fragments Length Polymorphism (RFLP) was a commonly used type of molecular markers in identifying propionibacteria (Riedel et al., 1995, 1998; Gautier et

al., 1997; Rossi et al., 1997). The mechanisms of this method depended on the assumption that different sizes of amplicons were due to different restriction endonucleases sites that were targeted to unique DNA sequences of each propionibacteria species. Another study used recA gene as molecular marker and sequenced the amplicons to determine the bacteria (Rossi et al., 2006). However, those techniques were aimed to the identification of pure isolates of propionibacteria rather than multi-bacterial food samples.

In terms of propionibacteria species, many studies have been conducted to detect dairy propionibacteria, including P. freudenreichii, P. jensenii, P. acidipropionici and P. theonii. The major approach was to design species-specific primer pair to differentiate the species from other propionibacteria. Tilsala-Timisjarvi et al. developed two pairs of primers based on the 16S rRNA gene of the bacteria. However, the primer pair PfrI/PfrII for P. freudenreichii was unable to differentiate Streptococcus spp., while the PacI/PacII for P. acidipropionici failed to eliminate the cross-reactivity of Pseudomonas spp.

(Tilsala-Timisjarvi et al., 2001). In another study, the researchers specifically identified

P. freudenreichii by designing PF/PB2 primer pair on 16S rRNA gene, and they also designed PA/PB2 for the identification of P. acidipropionici (Rossi et al., 1999).

However, PF/PB2 was unable to differentiate Lactobacillus spp., while PA/PB2 failed to eliminate the cross-reactivity of Streptococcus spp. and Enterococcus spp.

The ideal DNA fragments for the identification of bacteria on a genus-level should be conserved within the genus while different from other genera; as for species-

level identification, the ideal target sequences are supposed to be highly variable among species. Although highly variable regions in 16S ribosomal RNA are ideal targets for genus-specific identification, it is too conserved among Propionibacterium to identify each species. Constitutive genes that perform consistent function in bacteria can serve as the targets for classification scheme. By acquiring the constitutive gene sequences available from the National Center for Biotechnology Information (NCBI) database, we have found several candidates to identify propionibacteria on species level.

In this study, a rapid detection system for propionibacteria, mainly P. acidipropionici, P. acnes, P. avidum, and P. freudenreichii, using conventional PCR with genus- and species-specific primer pairs targeted to constitutive genes was designed and evaluated. The cross-reactivity of this system with other foodborne bacteria was reduced so it was applicable to food matrices containing multiple bacterial genera and species.

Finally, this rapid detection system was applied to Swiss cheese samples with and without splits and cracks to investigate the presence of non-starter propionibacteria.

2.3 Materials and Methods

2.3.1 DNA extraction of bacterial strains in Swiss cheese samples

A total of 9 dairy-related strains including 1 Lactococcus strain, 2 Staphylococcus strains, 3 Enterococcus strains, and 3 Pseudomonas strains were recovered in Brain Heart

Infusion broth at 30°C for 18 hours. Four Propionibacterium freudenreichii strains (B1-

B4), two Lactococcus strains (WL1, WL2) and 2 Streptococcus strains (WS3, WS4) were isolated from Swiss cheese samples (Table 2.1). P. acidipropionici strain of ATCC 4875 was obtained from Dr. Yang’ lab (Suwannakham et al., 2006). P. acnes and P. avidum strains were obtained from the USDA/ARS (Edwards et al., 2013). All propionibacteria and Lactobacillus strains were recovered using sodium lactate agar (1% tripticase soy broth, 1% yeast extract, 1% sodium lactate, 0.025% dipotassium phosphate, 1.5% agar, pH 7.0±0.2), and the Streptococcus strains were recovered on the Man, Rogosa and

Sharpes (MRS) medium by incubating at 30°C anaerobically for 5 days. The genus of the bacteria was confirmed by 16S ribosomal RNA sequencing using universal 16S primer pair listed in Table 2.2. The procedure described by Li and Wang was followed to extract

DNAs from all bacterial strains as amplification templates for conventional PCR analysis.

2.3.2 The design of genus- and species-specific primer pairs for propionibacteria

The 16S ribosomal RNA gene sequences of 14 Propionibacterium strains (P. acidifaciens C3M_31 EU979537.1, P. acidipropionici DH42 AY360222.1, P. acnes

2929 AB108484.1, P. australiense LCDC-98A072 NR_025076.1, P. avidum LET106

FN824486.1, P. cyclohexanicum JCM 21245 AB639145.1, P. freudenreichii ISU P59

AY533300.1, P. freudenreichii subsp. shermanii E11 NR_036972.1, P. granulosum D-34

FJ785716.1, P. humerusii P08 NZ_AF AM01000003.1, P. jensenii DSM 20535 20535

AJ704571.1, P. microaerophilum M5 NR_028778.1, P. propionicum F0230a

NR_102899.1, P. thoenii JCM 6435 AB729071.1) and 11 dairy-related bacteria

(Enterococcus faecium H2 EU887814.1, Lactococcus lactis BIHB 331 FJ859680.1,

Streptococcus thermophilus KI-1 AB812892.1, Lactobacillus rhamnosus L60

EF495247.1, Lactobacillus delbrueckii UFV H2b20 EF015468.1, Staphylococcus aureus

RKA6 EF463060.1, Leuconostoc mesenteroides KIBGE-IB19 HQ588348.1,Micrococcus antarcticus T2 NR_025285.1, Micrococcus terreus V3M1 NR_116649.1, Micrococcus lylae DSM 20315 NR_026200.1, and Mycobacterium avium AY360329.1) were obtained from NCBI GenBank database. Sequence alignments were conducted using the

DNASTAR MegAlign program, and a total of five primer pairs were designed on various regions of genomic DNA of propionibacteria (Table 2.2). The homology of the primers was also checked to ensure no cross-reactivity with other bacteria using the BLAST program in NCBI.

2.3.3 PCR amplification of genomic DNA

The PCR reactions were performed in a volume of 20 ul for identification and 50 ul for DNA sequencing. PCR conditions for each primer pair were: initial denaturation

for 3 min at 94°C followed by 33 cycles of 30s at 94°C, 30s at various annealing temperature listed in Table 2.2 followed by 30s at 72°C, with a final extension of 3 min at

72°C. Cycling conditions for general primers of 16S ribosomal RNA were: initial denaturation for 3 min at 94°C followed by 33 cycle of 0.5 min at 94°C, 1 min at 50°C and 0.5 min at 68°C, with a final extension of 3 min at 68°C. Amplification products were separated on 2.5% and 1.5% agarose gels respectively and stained with ethidium bromide.

2.4 Results

2.4.1 The identification of dairy bacteria from Swiss cheese

Four P. freudenreichii strains, two Lactobacillus strains, and two Streptococcus strains isolated from Swiss cheese are listed in Table 2.1. The identifications of each strain were confirmed by sequencing 16S ribosomal RNA using universal 16S primer pair 16SF/16SR.

Table 2.1 The recovery media, incubation conditions, colony morphology and identification of dairy bacteria strains isolated from Swiss cheese.

Isolate Recovery Incubation Isolate Source Identification ID Media Condition Morphology B1 Brownish, B2 Ripened Sodium opaque, Propionibacterium B3 cheese Lactate medium- freudenreichii B4 sized, convex WL1 30°C for 7 White, Ripened Sodium days opaque, large- Lactobacillus spp. WL2 cheese Lactate anaerobically sized, convex WS3 White, Cheese opaque, out-of- MRS Streptococcus sp. WS4 small-sized, press convex

2.4.2 Genus- and species-specific PCR primer pairs for Propionibacterium

In this study, a total of five primer pairs listed in Table 2.2 were designed for genus- and species-specific PCR. Prop16SF/Prop16SR was the genus-specific primer pair for Propionibacterium, targeting the conserved regions of the Propionibacterium 16S ribosomal RNA gene. P-acpF/P-acpR was the species-specific primer pair for P. acidipropionici designed based on adk gene which encodes adenylate kinase. The targeted region for P-anrF/P-anrR primer pair is on the recombinase encoded gene – recA, and P. acnes were the only species that exhibited a 335-bp product with these primers. P-vF/P-vR was designed to amplify partial pepN genes which encodes aminopeptidase N, and can be used to distinguish P. avidum from other propionibacteria and foodborne bacteria. For P. freudenreichii, which is commonly used in Swiss cheese manufacturing, the species-specific primer pair P-fF/P-fR was designed on the fumC gene which encodes fumarate hydratase. All the primers designed were checked using the

BLAST program in the NCBI database to minimize cross-reactivity with other foodborne bacteria.

2.4.3 The specificity of Propionibacterium genus-specific PCR

The effectiveness of the genus-specific primers was tested on 4

Propionibacterium strains and other dairy-related species stated in 2.3. At annealing temperature of 57°C, 4 strains of Propionibacterium showed a prominent band (188 bp)

(Fig. 2.1). There was no amplification of the Propionibacterium-specific fragments in non-propionibacteria strains examined (Fig. 2.1). The primer pair Prop16SF/Prop16SR showed the targeted 188-bp amplicon in all four propionibacteria species – P. freudenreichii, P. acidipropionici, P. acnes, and P. avidum. However, there is no amplification observed in in Lactobacillus spp. (WL1, WL2) and Streptococcus spp.

(WS1, WS2) isolated from Swiss cheese samples, Lactococcus sp. (YLT1-8),

Staphylococcus spp. (YLE1-3, YLT9-10), Enterococcus spp. (G6-13, 14-32, 18fT3), or

Pseudomonas spp. (YLS3-1, YLS3-3, YLS3-4).

This result showed high specificity of designed primer pair in identifying genus

Propionibacterium in food matrices, and there is no cross-reactivity of this primer pair in other common foodborne bacteria.

2.4.4 The specificity of P. freudenreichii, P. avidum, and P. acidipropionici species- specific PCR

The primer pairs P-fF/P-fR (Fig. 3.2a), P-vF/P-vR (Fig. 2.2b), P-acpF/P-acpR

(Fig. 2.2c) were designed and tested with conventional PCR and gel electrophoresis. The results of the study showed effective detection with high specificity of the primer pairs for P. freudenreichii and P. avidum respectively against Lactobacillus, Streptococcus,

Lactococcus, Staphylococcus, Enterococcus, and Pseudomonas (not shown in Fig. 2.2).

The amplification of P-acpF/P-acpR targeted fragments in propionibacteria and

foodborne bacteria (Fig. 2.2c) was not as specific as the other two primer pairs. Although some other bacteria (Enterococcus spp., Pseudomonas spp., P. freudenreichii, P. acnes) also showed amplicons in Fig. 3.2c in conventional PCR using P-acpF/P-acpR, different sizes of amplicons from P. acidipropionici (319-bp) made it feasible to specifically identify P. acidipropionici from other propionibacteria or foodborne bacteria.

Table 2.2 PCR primer sequences, the locations of target fragments, the sizes of amplicons and their annealing

temperatures used in this study

Amplicon Annealing Detection Primer Sequence (5’-3’) Size (bp) Reference position Temperature (°C) Targets Pr16SF TGAGGAAKGTAGGGGAGAATG 16S 188 57 Genus This study Pr16SR CCACACCTAGTACCCACCG P-acpF TGTGAAGTGCCACACGATGA adk 319 54 P. acidipropionici This study P-acpR CGACGAGGAGATCGGCTTC

4 P-anrF GCAGGCAGAGTTTGACATCC 7 recA 335 60 P. acnes This study P-anrR GCTTCCTCATACCACTGGTCATC P-anF GGGTTGTAAWCCGCTTTCGCCTG 16S 587 60 P. acnes Harada et al. P-anR GGGACACCCATCTCTGAGCAC P-vF ATGGACTTGCACAGCTCGAGG pepN 472 60 P. avidum This study P-vR ATGTGCGCGATCGTCTCATTGC P-fF ACCGAGCTGCAGGAGATGT fumC 195 60 P. freudenreichii This study P-fR GTCGGAGAACTTGATGCACTT 16SF AGAGTTTGATCCTGGCTCAG Universal 16S 16S 1498 50 Hummel et al. 16SR TACCTTGTTACGACTT primer

2.4.5 Increased specificity for the identification of P. acnes

Another pair of primer P-anF/P-anR (Harada et al., 2001) specific for P. acnes was evaluated, and the result was shown in Fig. 2b. The sizes of amplicons for

Pseudomonas spp. (Fig. 2.3b Lane 4-6) are different from that of P. acnes. However, P- anrF/P-anrR designed in this study showed great specificity in P. acnes (Fig. 2.3a), and there was no amplification in Pseudomonas spp. or other common foodborne bacteria examined in this study, which made it a better PCR primer pair for the detection of P. acnes with increased specificity.

2.4.6 Genus- and species-level of propionibacteria in Swiss cheese samples

Genus Propionibacterium were validated with 188-bp amplicons in both premium and defected Swiss cheese samples by the genus-specific Prop16SF/Prop16SR primer pair (Fig. 2.4a). The species-specific PCR was conducted with four primer pairs designed in this study: P-fF/P-fR, P-acpF/P-acpR, P-anrF/P-anrR, and P-vF/P-vR. Only P. freudenreichii specific 195-bp fragments were amplified in Swiss cheese using P-fF/P-fR primers (Fig. 2.4b), and other primers produced negative results (not shown in Fig. 2.4), indicating the predominant population of propionibacteria was P. freudenreichii, and the population of P. acidipropionici, P. acnes, and P. avidum were at an undetectable level.

There was no significant difference in the composition of propionibacteria in split and non-split Swiss cheese, and the only Propionibacterium species detected in both of them

was P. freudenreichii, which was the essential starter culture for Swiss cheese manufacturing.

2.5 Conclusions and Discussions

2.5.1 The genetic markers in identification of propionibacteria

In this study, a culture-independent, rapid molecular detection system for propionibacteria has been successfully developed for the identification of P. acidipropionici, P. acnes, P. avidum, and P. freudenreichii. Genus- and species-specific primer pairs targeting constitutive genes, including 16S rRNA, adk, recA, pepN, and fumC, were designed and validated.

The 16S rRNA gene has been well-recognized as the marker for phylogenetic identification (Clarridge, 2004). However, when it comes to differentiate bacteria that are genetically closely related, other constitutive genes are in need. Adenylate kinase, which catalyzes the reversible ATP-dependent phosphorylation of AMP to ADP and dAMP to dADP, can also catalyze the conversion of nucleoside diphosphates to the corresponding triphosphates. Many studies used the gene adk, encoding adenylate kinase in bacteria, for bacterial identification (Maiden et al., 1998; Corrigan et al., 2013; Helgason et al., 2014).

As an essential gene for the repair and maintenance of DNA, recA has been widely used in the identification and classification of bacteria (Payne et al., 2005; Torriani et al., 2001;

Blackwood et al., 2000). Encoding aminopeptidase N, pepN has been applied in the identification of Lactobacillus helveticus (Fortina et al., 2001). This gene encodes a highly-conserved area which is also homologous to mammalian aminopeptidase N (Tan et al., 1992). Fumerase is responsible for catalyzing the reversible hydration or dehydration of fumarate to malate (Ueda et al., 1991). The gene encoding fumerase,

fumC, has been used in the multilocus sequence typing of clinical diagnosis (Robertson et al., 2004).

2.5.2 The significance of bacterial detection in Swiss cheese industry

Swiss cheese is a popular dairy food in both Europe and the US. The average consumption of Swiss cheese in Switzerland and France is 44 and 60 pounds per capita in the year of 2005. Americans consume less cheese than Europeans do, but per capita cheese consumption in the US increased from 22.6 pounds in 1985 to 33.3 pounds in

2010. In 2012, more than 360 million pounds of Swiss cheese was consumed in the US

(Cheese Reporter, 2013), making it an important dairy food.

Results from many studies suggested the beneficial effects of propionibacteria in human gut, where they work synergistically with beneficial gut microbiota to promote the growth and adhesion of those microbiota to the gut (Collado et al., 2007a), and to inhibit enteric pathogens (Collado et al., 2007b). Propionibacteria can survive harsh technological stress (Rossi et al., 2000) and physiological conditions (Darilmaz et al.,

2012), which makes it ideal as probiotics.

Most of the studies for bacterial identification and classification were conducted due to clinical needs. In fact, with the increasing demand for culture-independent method that can deeply investigate the uncultivable bacteria, more and more genetic markers, usually housekeeping genes, have been explored and studied by researchers.

Rapid and specific identification of microorganisms is also a challenge to the food industry. Researchers in the industry have been looking for efficient and economic methods to monitor the microorganisms during dairy processing. In this study, the genus- specific and species-specific primer pairs were designed and evaluated to serve a rapid detection platform for propionibacteria in multi-bacterial food matrices. The advantage of this method is that it minimized the cross-reactivity of propionibacteria with other starter cultures and foodborne bacteria, which makes it an effective and efficient quality control tool in the identification of propionibacteria on a genus- and species-level for the food industry.

Figure 2.1 Amplification of 16S rDNA 188 bp fragments in propionibacteria and

Swiss cheese samples using Pr16SF/Pr16SR primer. M, 100 bp marker; 1, P. freudenreichii (B1) isolated from Swiss cheese; 2, P. acidipropionici ATCC 4875; 3, P. acnes; 4, P. avidum; 5, Lactobacillus sp. (WL1) isolated from Swiss cheese; 6,

Streptococcus sp. (WS1) isolated from Swiss cheese; N, negative control.

Figure 2.2 The effectiveness of P. freudenreichii, P. avidum and P. acidipropionici identification using species-specific primer pairs. a) Amplification of fumC gene 195 bp fragments in propionibacteria using P-fF/P-fR primer pair. M, 100 bp marker; 1, 2, P. freudenreichii (B1, B2) isolated from Swiss cheese; 3, P. acnes; 4, P. acidipropionici

ATCC 4875; 5, P. avidum; 6, Enterococcus sp. (G6-13); N, negative control. b)

Amplification of pepN gene 472 bp fragments in propionibacteria and other dairy-related bacteria using P-vF/P-vR primer pair. M, 100 bp marker; 1, 2, P. avidum; 3, P. acnes; 4,

5, P. freudenreichii (B1, B2); 6, P. acidipropionici ATCC 4875; N, negative control. c)

Amplification of adk gene 319 bp fragments in propionibacteria and other dairy-related bacteria using P-acpF/P-acpR primer. M, 100 bp marker; 1, 2, 16, 17, P. acidipropionici

ATCC 4875; 3, 4, Lactobacillus spp. (WL1, WL2) isolated from Swiss cheese; 5, 6,

(continued)

Figure 2.2: continued

Streptococcus spp. (WS1, WS2) isolated from Swiss cheese; 7, Lactococcus sp. (YLT1-

8); 8, 9, Staphylococcus spp. (YLE1-3, YLT9-10); 10-12, Enterococcus spp. (G6-13, 14-

32, 18fT3); 13-15, Pseudomonas spp. (YLS3-1, YLS3-3, YLS3-4); 18, 19, P. freudenreichii (B1,B2) isolated from Swiss cheese; 20, P. avidum; 21, P. acnes; N, negative control.

Figure 2.3 The effectiveness of two species-specific primer pairs in P. acnes and other dairy-related bacteria. a) Amplification of recA gene 335 bp fragments using P- anrF/P-anrR primer. b) Amplification of 587 bp fragments using P-anF/P-anR primer. M,

100 bp marker; 1, 2, P. acnes; 3, Lactococcus sp. (YLT1-8); 4-6, Pseudomonas spp.

(YLS3-1, YLS3-3, YLS3-4); N, negative control.

Figure 2.4 Amplification of genus- and species-specific fragments in Swiss cheese samples using a) Prop16SF/Prop16SR and b) P-fF/P-fR primers. M, 100 bp marker; 1,

Swiss cheese from Vat 7 with splits; 2, Swiss cheese from Vat 7 without splits; 3, Swiss cheese from Vat 8 with splits; 4, Swiss cheese from Vat 8 without splits; N, negative control.

2.6 Bibliography

Blackwood, K. S., He, C., Gunton, J., Turenne, C. Y., Wolfe, J., & Kabani, A. M. (2000). Evaluation of recA sequences for identification of Mycobacterium species. Journal of clinical microbiology, 38(8), 2846-2852.

Britz, T. J., & Riedel, K. H. (1994). Propionibacterium species diversity in Leerdammer cheese. International journal of food microbiology, 22(4), 257-267.

Clarridge, J. E. (2004). Impact of 16S rRNA gene sequence analysis for identification of bacteria on clinical microbiology and infectious diseases. Clinical microbiology reviews, 17(4), 840-862.

Collado, M. C., Meriluoto, J., & Salminen, S. (2007a). Development of new probiotics by strain combinations: is it possible to improve the adhesion to intestinal mucus?. Journal of dairy science, 90(6), 2710-2716.

Corrigan, R. M., Campeotto, I., Jeganathan, T., Roelofs, K. G., Lee, V. T., & Gründling, A. (2013). Systematic identification of conserved bacterial c-di-AMP receptor proteins. Proceedings of the National Academy of Sciences, 110(22), 9084-9089.

Darilmaz, D. O., & Beyatli, Y. (2012). Acid-bile, antibiotic resistance and inhibitory properties of propionibacteria isolated from Turkish traditional home-made cheeses. Anaerobe, 18(1), 122-127.

Edwards, H. D., Anderson, R. C., Taylor, T. M., Miller, R. K., Hardin, M. D., Nisbet, D. J., ... & Smith, S. B. (2013). Interactions between oil substrates and glucose on pure cultures of ruminal lipase-producing bacteria. Lipids, 48(7), 749-755.

Estoppey, O., Rivier, G., Blanc, C. H., Widmer, F., Gallusser, A., & So, A. K. (1997). Propionibacterium avidum sacroiliitis and osteomyelitis. Revue du rhumatisme (English ed.), 64(1), 54-56.

Gautier, M., de Carvalho, A., & Rouault, A. (1996). DNA Fingerprinting of Dairy Propionibacteria Strainsby Pulsed-Field Gel Electrophoresis. Current Microbiology, 32(1), 17-24.

Harada, K., Tsuneyama, K., Sudo, Y., Masuda, S., & Nakanuma, Y. (2001). Molecular identification of bacterial 16S ribosomal RNA gene in liver tissue of primary biliary cirrhosis: is Propionibacterium acnes involved in granuloma formation?. Hepatology, 33(3), 530-536.

Helgason, E., Tourasse, N. J., Meisal, R., Caugant, D. A., & Kolstø, A. B. (2004). Multilocus sequence typing scheme for bacteria of the Bacillus cereus group. Applied and Environmental Microbiology, 70(1), 191-201.

Hummel, A., Holzapfel, W. H., & Franz, C. M. (2007). Characterisation and transfer of antibiotic resistance genes from enterococci isolated from food. Systematic and applied microbiology, 30(1), 1-7.

Kim, J., Shen, R., Olcott, M. C., Rajagopal, I., & Mathews, C. K. (2005). Adenylate kinase of Escherichia coli, a component of the phage T4 dNTP synthetase complex. Journal of Biological Chemistry, 280(31), 28221-28229.

Maiden, M. C., Bygraves, J. A., Feil, E., Morelli, G., Russell, J. E., Urwin, R., ... & Feavers, I. M. (1998). Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proceedings of the National Academy of Sciences, 95(6), 3140-3145.

Malik, A. C., Reinbold, G. W., & Vedamuthu, E. R. (1968). An evaluation of the taxonomy of Propionibacterium. Canadian Journal of Microbiology, 14(11), 1185-1191.

Park, H. S., Reinbold, G. W., & Hammond, E. G. (1967). Role of propionibacteria in split defect of Swiss cheese. Journal of dairy science, 50(6), 820-823.

Payne, G. W., Vandamme, P., Morgan, S. H., LiPuma, J. J., Coenye, T., Weightman, A. J., ... & Mahenthiralingam, E. (2005). Development of a recA gene-based identification approach for the entire Burkholderia genus. Applied and environmental microbiology, 71(7), 3917-3927.

Riedel, K. H., Wingfield, B. D., & Britz, T. J. (1995). Justification of the “classical” Propionibacterium species concept by restriction analysis of the 16S ribosomal RNA genes. Systematic and applied microbiology, 17(4), 536-542.

Robertson, G. A., Thiruvenkataswamy, V., Shilling, H., Price, E. P., Huygens, F., Henskens, F. A., & Giffard, P. M. (2004). Identification and interrogation of highly informative single nucleotide polymorphism sets defined by bacterial multilocus sequence typing databases. Journal of medical microbiology, 53(1), 35-45.

Rossi, F., Sammartino, M., & Torriani, S. (1997). 16S-23S Ribosomal spacer polymorphism in dairy propionibacteria. Biotechnology techniques, 11(3), 159- 162.

Rossi, F., Dellaglio, F., & Torriani, S. (2006). Evaluation of recA gene as a phylogenetic marker in the classification of dairy propionibacteria. Systematic and applied microbiology, 29(6), 463-469.

Stackebrandt, E., Cummins, C. S., & Johnson, J. L. (2006). Family propionibacteriaceae: the genus Propionibacterium. In The prokaryotes (pp. 400-418). Springer New York.

Suwannakham, S., Huang, Y., & Yang, S. T. (2006). Construction and characterization of ack knock‐out mutants of Propionibacterium acidipropionici for enhanced propionic acid fermentation. Biotechnology and bioengineering, 94(2), 383-395.

Tan, P. S., van Alen-Boerrigter, I. J., Poolman, B., Siezen, R. J., de Vos, W. M., & Konings, W. N. (1992). Characterization of the Lactococcus lactis pepN gene encoding an aminopeptidase homologous to mammalian aminopeptidase N. FEBS letters, 306(1), 9-16.

Tilsala-Timisjärvi, A., & Alatossava, T. (2001). Characterization of the 16S–23S and 23S–5S rRNA intergenic spacer regions of dairy propionibacteria and their identification with species-specific primers by PCR. International journal of food microbiology, 68(1), 45-52.

Torriani, S., Felis, G. E., & Dellaglio, F. (2001). Differentiation of Lactobacillus plantarum, L. pentosus, and L. paraplantarum by recA gene sequence analysis and multiplex PCR assay with recA gene-derived primers. Applied and Environmental Microbiology, 67(8), 3450-3454.

Ueda, Y., Yumoto, N., Tokushige, M., Fukui, K., & Ohya-Nishiguchi, H. (1991). Purification and characterization of two types of fumarase from Escherichia coli. Journal of biochemistry, 109(5), 728-733.

Zouboulis, C. C. (2004). Acne and sebaceous gland function. Clinics in dermatology, 22(5), 360-366.

Chapter 3 Microbial profiling using 16S metagenomics for quality assessment in

Swiss cheese

3.1 Abstract

Swiss cheeses are susceptible to food quality issues associated with undesirable microorganisms. Quality defects ranging from splits, unusual eyes, off-flavors, to off- odors result in great economic loss annually to the Swiss cheese industry. Currently there is a lack of sensitive and culture-independent studies for the investigate of Swiss cheese microbiota. The aim of this study is to use culture-independent 16S metagenomics to analyze and compare the microbial profile in Swiss cheese with and without split defects, with long-term goal of identifying microbial causing agent(s) for split defect.

The starter cultures and non-starter microbiota from 4 split Swiss cheese blocks made by two different factories were analyzed. Result showed no significant difference in the relative abundance of starter cultures among split and non-split cheese, and the relative abundance of Enterobacteriaceae was significantly higher in split Swiss cheese

than that of non-split samples, providing evidence that Enterobacteriaceae could be a potential microbial causing agent of split defects.

Cheese samples, starter cultures and pasteurized milk from the same batch were aseptically collected. V3-V4 regions of the 16S ribosomal RNA gene segments of cheese microbiota were sequenced by Illumina MiSeq platform, then aligned and analyzed by

MOTHUR pipeline. The microbiota of Swiss cheese made in the local cheese factory showed consistency in both richness and diversity when compared to retail eye-forming cheese. 99.82% of Swiss cheese microbiota were found to be starter cultures, including

Streptococcus (56.68±5.09%), Lactobacillus (41.51±5.72%) and Propionibacterium

(1.62±0.87%). Common commensal microbiota represented a very low proportion of the cheese microbiota. Using Principal Component Analysis and a heat map with dendrograms, this study demonstrated the split areas in Swiss cheese were microbiologically different from eye areas, which could be due to inhomogeneous distribution of starter cultures.

Further validation through cheese making is underway. In this study, metagenomics in 16S rDNA provides a powerful tool in revealing microbial profiles of

Swiss cheese, assisting in further targeted analyses of microbial safety and quality risk factors.

3.2 Introduction

Defects in Swiss cheese such as splits are often due to microbes. Many studies have been conducted on the investigation of split defect in Swiss cheese. Hettinga et al.

(1974) suggested that propionbiacteria strain that were able to grow at 3.8ºC produced excessive amounts of carbon dioxide and resulted in a higher incidence of splits than the strain lacking this low-temperature growth ability. Another study conducted a 2×2×2 factorial experiment to investigate the effect of different Lactobacillus helveticus/Propionibacterium freudenreichii ssp. shermanii starter culture combinations on the occurrence of split defect in Swiss cheese, and concluded that there was no correlation between moisture, pH, fat, protein, calcium, lactose content, D/L lactate ratio or protein degradation to splits. However, seasonal variation in milk resulted in summer cheese having higher split incidences, and culture combination also influenced the formation of splits (White et al., 2003). In that study, the split incidence was random and there was no correlation between the incidence and the season. As a result, the causing agent was hypothesized to be different from the previous studies.

Splits and cracks in Swiss cheese are believed to be associated with the contamination of microorganisms, especially commensal bacteria from milk, processing equipment, handlers, or the environment, and they may affect the quality and safety of the final cheese products. Some have led to severe quality issues, like late blowing caused by Clostridium tyrobutyricum, resulting in slits as well as abnormal flavor from butyric acid (Klijn et al., 1995). Non-starter bacteria in commercial cheese products, such as

Leuconostoc sp., Pseudomonas sp. (Wang et al., 2006), Enterococcus faecium,

Escherichia coli (Li et al., 2010), were found in commercial cheese.

The culture-independent 16S metagenomics is low-cost and provides high- throughput result in characterizing the microbiota of dairy food. Metagenomics has been applied in dairy science since 2010. Most of these studies analyzed the microbiome in milk, like mastitic milk vs. healthy milk (Bhatt et al., 2012; Oikonomou et al., 2012), raw milk vs. pasteurized milk (Schmidt et al., 2012; Quigley et al., 2013). Presently, there is limited studies in using metagenomics to assess the microbial quality of Swiss cheese.

The objectives of the study was to use 16S metagenomics approach: 1) to analyze the microbiota of split Swiss cheese made by two different factories; 2) to comprehensively analyze and compare the spatial microbial distribution in Swiss cheese with and without split defects made by one factory; and 3) to compare the main starter cultures of retail and factory-made cheese with eyes.

3.3 Materials and Methods

3.3.1 Sample Collection and Genomic DNA Extraction

Pasteurized milk, starter cultures and Swiss cheese samples were collected from manufacturing facilities associated with the Swiss Cheese Consortium companies in Ohio.

Commercial cheese samples with eyes were obtained from local grocery stores. For each block of cheese, the following sites were sampled: “eye”, “blind” and “split” (if any).

“Eye” is where round or slightly oval-shaped holes can be found on the cheese. “Blind” is where there is no eye formation on the cheese body. “Split” is sizable and clean-cut cracks or slits found on the body of the cheese (USDA, 2001). The cheese samples were serially diluted with 37ºC pre-warmed 2% sodium citrate solution and homogenized by a stomacher. Total DNA of these samples were extracted from the cheese homogenates using the PowerFood® Microbial DNA Isolation Kit (MOBIO Laboratories, Inc., West

Carlsbad, CA). DNA of pasteurized milk and starter cultures were directly extracted using the same kit. The concentrations of DNA were determined with a NanoDrop spectrophotometer.

3.3.2 16S rDNA Gene Library Construction

The 16S rDNA library for each cheese sample was generated according to 16S metagenomics sequencing library preparation procedure (Illumina, Inc., San Diego, CA).

First, V3-V4 hypervariable region of 16S rRNA segment was amplified. 16S amplicon

was obtained with forward primer: 5’-

TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG -

3’, and reverse primer: 5’-

GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAA

TCC -3’. Each PCR reaction was set up with 2.5 μl of microbial genomic DNA (5 ng/μl in 10 mM Tris pH 8.5), 5 μl reverse primer (1 μM), 5 μl forward primer (1 μM), and 12.5

μl 2× KAPA HiFi HotStart Ready Mix. PCR was performed in a thermal cycler using the following program: 95ºC for 30 minutes, 25 cycles of 95ºC for 30 seconds + 55ºC for 30 seconds + 72ºC for 30 seconds, followed by 72ºC for 5 minutes. The PCR products were verified using 1% agar gel electrophoresis. AMPure XP beads (Beckman Coulter, Inc.,

Brea, CA) were used to purify the 16S V3-V4 amplicon. Freshly prepared 80% ethanol was used in this step. Clean-up products were verified on 1% agarose gel.

Second-round amplification is used to attach index sequences to each end of the amplicons for sequencing recognition. The index used were listed in Table 3.1. Each PCR reaction was set up with 5 μl of PCR products from the previous step, 5 μl Nextera XT

Index primer 1 (N7xx), 5 μl Nextera XT Index primer 2 (S5xx), 25 μl 2× KAPA HiFi

HotStart Ready Mix, and 10 μl PCR grade water. PCR was performed in a thermal cycler using the following program: 95ºC for 30 minutes, 8 cycles of 95ºC for 30 seconds +

55ºC for 30 seconds + 72ºC for 30 seconds, followed by 72ºC for 5 minutes. The result of amplification is examined by running a small amount of amplicons on 1% agarose gel.

The rest of the amplicons is purified by AMPure XP beads (Beckman Coulter, Inc., Brea,

CA). The final library was verified on 1% agarose gel and pooled together at same amount.

Table 3.1 Illumina Nextera Index used in this study.

Index Name Sequence Index Name Sequence N701 TCGCCTTA S517 GCGTAAGA N702 CTAGTACG S502 CTCTCTAT N703 TTCTGCCT S503 TATCCTCT N704 GCTCAGGA S504 AGAGTAGA N705 AGGAGTCC S505 GTAAGGAG N706 CATGCCTA S506 ACTGCATA N707 GTAGAGAG S507 AAGGAGTA N708 CCTCTCTG S508 CTAAGCCT N709 AGCGTAGC N710 CAGCCTCG N711 TGCCTCTT N712 TCCTCTAC

3.3.3 High throughput sequencing and analysis of sequencing data

All libraries were run in the same sequencing reaction and all amplicons were sequenced at 25M standard flow-cell v3, 600 bases, using Illumina MiSeq system. The high throughput sequencing was performed in the Molecular and Cellular Imaging Center

(MCIC) (OARDC, Wooster). The data analysis was conducted following the MOTHUR

SOP (accessed 12 December 2014). The 16S rDNA sequence reads were processed using

MOTHUR software package (Schloss et al, 2009). Barcodes and primers in each read were trimmed and the paired reads were assembled into full-length amplicons. The assembled reads were then screened to remove low-quality reads (Q score below 25). The quality of all the sequence reads were denoised using the PyroNoise algorithm implemented and filtered in MOTHUR. Only sequences with no ambiguous base, maximum of 8 homopolymers, and in length more than 400 base pairs were kept for downstream analysis. The screened sequences were aligned using the NAST algorithm with the Silva v102 reference set of sequences, and pre-clustered allowing 1 difference for every 100 bp of sequence (Pruesse et al, 2007). The sequences were checked for the presence of chimeric amplification using the MOTHUR implementation of UCHIME algorithm (Edgar et al, 2011). The reads were then classified by comparing to the reference sequences from Ribosomal Database Project Classifier (Cole et al, 2009),

MOTHUR-formatted version of training set 9, with a bootstrap value of 80%, and non- bacterial sequences were removed. The final reads were clustered into operational taxonomic units (OTUs) with a 0.03 distance unit cutoff. A taxonomic identity were assigned to each OTU. Cheese samples were grouped according to different site. A heat map was generated to visualize the difference in OTU abundance in all samples.

3.4 Results

3.4.1 The comparison of Swiss cheese with split defect made by two factories

3.4.1.1 Microbial diversity of industry Swiss cheese samples

A total of four blocks of Swiss cheese samples ripened for 2 months were collected from two factories. As shown in Figure 3.1a, the starter cultures consisted of more than 98% of Swiss cheese microbiota. In all samples, the average relative relativeabundances of Lactobacillaceae, Streptococceae and Propionibacteriaceae were

45.4%, 48.4% and 4.6%, respectively. Besides starter cultures, non-starter bacteria accounted for ~2% of Swiss cheese microbiota (Figure 3.1b), including

Pseudomonadaceae, Enterobacteriaceae, Planococcaceae, Enterococcaceae,

Moraxellaceae, etc. The sources of the non-starter bacteria could be milk, handler, processing environment, ripening environment, etc. Some bacteria from these families negatively impact the quality of the final cheese products. Pseudomonas fluorescens was identified as the causative agent of spoilage by producing a blue fluorescent pigment on the fresh, low-acid cheese (Martin et al., 2011). However, some bacteria in this group served as adjunct cultures in cheese production. Enterococcus faecium positively affected physical structure as well as overall sensory profile of Greek Feta cheese

(Sarantinopoulos et al., 2002).

3.4.1.2 Starter cultures in Swiss cheese samples

In terms of split and non-split Swiss cheese, there was no significant difference in the relative abundances of Lactobacillus, Streptococcus or Propionibacterium (Figure

3.2a). Lactobacillus and Streptococcus consisted 45-50% of cheese microbiota, while

Propionibacterium was around 5%.

Among two factories where samples were collected, the relative abundances of

Lactobacillus and Streptococcus in cheese microbiota were consistent with no significant differences (Figure 3.2b). However, the relative abundance of Propionibacterium in

Swiss cheese made by Factory A was significantly higher than that of Factory B, with P <

0.05, indicating the final Swiss cheese product in Factory A carried more

Propionibacterium in each block. This discrepancy might be due to differences in starter culture recipe or cheese processing; whereas no supportive evidence showed the split defect was related to the overgrowth of Propionbacterium.

3.4.1.3 Non-starter bacteria in Swiss cheese samples

The relative abundances of non-starter bacteria were investigated in split and non- split Swiss cheese as shown in Figure 3.3a. Split cheese area carried more

Enterobacteriaceae and Porphyromonadaceae than non-split Swiss cheese. In family

Enterobacteriaceae, the major bacteria genera in Swiss cheese samples included

Escherichia/Shigella, Citrobacter, and Klebsiella, all of which were coliforms producing gas when fermenting lactose. This indicated that coliforms might be involved in the formation of split in Swiss cheese and could be microbial candidates for split causing agents. Since the approach in this study didn’t exclude DNA fragments from dead cells, most coliforms could be from raw milk and were killed during pasteurization and fermentation. So the causative correlation between split defect and coliforms requires further validation.

In terms of Swiss cheese made by different factories, the relative abundance of

Pseudomonadaceae in Swiss cheese from Factory A was significantly higher than that of

Factory B. However, products made by Factory B carried more Clostridiaceae,

Fusobacteriaceae and Porphyromonadaceae, all of which were nearly undetectable in products from Factory B.

3.4.2 Comprehensive analysis of Swiss cheese from one factory

3.4.2.1 Sampling

A total of 4 types of retail cheese with eyes were investigated, including Jarlsberg

Norseland (JPE, JPB, JSE, JSB, and JSS) imported from Norway and made from part skim pasteurized cow’s milk, Atlantique Baby Swiss (AE and AB) imported from

Normandy and made from pasteurized cow’s milk, Swiss Emmenthaler Emm (SE and SB) locally produced and made from unpasteurized cow’s milk, and Parrano Uniekaas (PE and PB) which was a gouda type cheese imported from Netherland and made with pasteurized cow’s milk.

A total of 6 blocks of Swiss cheese samples were produced on May 3rd 2015, in a local Swiss cheese factory and made from pasteurized cow’s milk. The cheese samples went through ripening stage and were collected after 51-day of production. Among the

Swiss cheese samples, three blocks from Vat 9, Vat 10 and Vat 16 were premium cheese with no defects; the other 3 blocks from Vat 8, Vat 13 and Vat 15 were cheese with splits.

Using 16S metagenomic analysis, Swiss cheese blocks from grocery stores and a local factory with and without split defect, together with the starter cultures and pasteurized milk for making the Swiss cheese were investigated. The average number of sequencing reads for each sample was around 3.7×104.

3.4.2.2 The microbial profile of factory-made Swiss cheese, starter cultures and pasteurized milk

The identification of Swiss cheese starter cultures used in the local factory showed that Streptocccus thermophilus, Lactobacillus spp. and Propionibacterium freudenreichii were the starters with no contamination of non-starter microorganisms.

Table 3.2 showed the microbial profile of pasteurized milk and starter cultures used in making Swiss cheese samples. Lactobacillus sp., which was found in all finished

Swiss cheese products, counted 99.57% of the microbial population in the pasteurized milk used for making Swiss cheese samples, with small amount of Streptococcus (0.29%) and Propionibacterium (0.14%). The majority of natural bacteria found in milk might be killed during pasteurization and had little impact on the microbiota in final Swiss cheese products. The microbiota analysis on starter cultures showed agreement with our previous knowledge obtained from the factory. Our study suggested that the starter cultures were not contaminated with common environmental bacteria, which may lead to quality defects of Swiss cheese.

This result demonstrated that 16S metagenomics is a feasible and accurate method in characterizing the microbial profile of Swiss cheese and related dairy foods.

3.4.2.3 The investigation of microbial profiles of factory-made Swiss cheese with and without split defect

Starter cultures accounted for over 99% in the total reads were firstly analyzed since they had the greatest impacts on the quality of Swiss cheese final products.

Common commensal microbiota represented a very low proportion of the cheese microbiota. In the split area of Swiss cheese, all three genera of starter cultures exhibited significant differences (P<0.05) when they were compared to their counterparts in eye area and blind area, respectively. Higher abundance of Lactobacillus, and lower abundance of both Propionibacterium and Streptococcus were observed in all split area of Swiss cheese samples from the factory (Fig 3.5). When it comes to blind area and normal eye area, there was no significant difference in the relative abundance of all three starter cultures.

The high throughput sequencing data showed that, besides starter cultures, some reads were classified as commensal bacteria. To better illustrate the whole picture of

Swiss cheese microbiota, and to investigate the role of commensal bacteria in Swiss cheese with splits, Principal Component Analysis (PCA) (Fig 3.6) and heat map (Fig 3.7) were conducted. The data showed significant difference in terms of microbial profile in eye and split area. Commensal bacteria consisted of less than 0.2% of the total microbiota in Swiss cheese samples.

In Figure 3.6, two circles representing two groups of Swiss cheese samples (split vs. normal eye) were drawn based on a P-value of 0.05. There was no overlap area in the

split samples which were in green color and the normal eye samples in red color, indicating the microbial profile of split area was significantly different from that of normal eye area. This result was in agreement with the result in Fig 3.5, in which the three main starter cultures were analyzed.

The dendrogram on the Y-axis of heat map showed the clustering of samples based on their phylogenetic relations. In the dendrogram, the split and normal eye area were classified into two groups, which means they were different in microbial profiles.

The same result was observed in the heat map, where split area carried more

Lactobacillus and less Streptococcus and Propionibacterium as compared to eye area in

Swiss cheese. This discrepancy could be due to heterogeneous distribution of bacteria in the cheese body.

3.4.2.4 Comparison of microbial profiles of retail cheese to cheese from Ohio manufacturer

Result showed that the starter cultures of several types of commercial cheese with eyes were different in abundance and diversity (Fig 3.4a). The dominant genus in

Atlantique Baby Swiss was Streptococcus (~90.2%), followed by Propionibacterium

(~4.0%), Lactobacillus (~2.5%) and Lactococcus (~2.6%). Swiss Emmenthaler Emm contained Lactobacillus (~71.7%), Streptococcus (~27.6%), Propionibacterium (<1%), and Lactococcus (<1%). These two types of Swiss cheese carried similar microbes as factory-made Swiss cheese, although varied in abundance. Jarlsberg Norseland is also a

Swiss/Emmental type cheese, but the microbial profiles were different from that of other types of Swiss cheese. Lactococcus (~72.7%) was the predominant genus in Jarlsberg

Norseland, followed by Propionibacterium (~25.1%), Leuconostoc (~1.4%), and

Lactobacillus (<1%). As a Gouda type cheese, Parrano Uniekaas was found to contain

Lactococcus (~50.1%), Thermus (~44.5%), Lactobacillus (~4.9%), and Leuconostoc

(<0.1%).

The retail cheese with eyes carried more diversified microbiota than expected.

Propionibacterium, which converted lactose into propionic acid and carbon dioxide, was found as the common starter culture in the three types of retail Swiss cheese samples examined. Atlantique Baby Swiss and Swiss Emmenthaler Emm both carried

Streptococcus and Lactobacillus, which were similar to the Swiss cheese samples from the Ohio manufacturer, although the abundance of certain bacterial genera varied. As a

Swiss type cheese, the microbial profile of Jarlsberg Norseland was quite different from that of other types of Swiss cheese, with Lactococcus, Propionibacterium, Lactobacillus and Leuconostoc identified as dominant bacteria. Although there were slight variations in the relative abundance of starter cultures in each retail cheese samples, the pattern of microbial profile with in one sample was consistant.

3.5 Conclusions and Discussions

Using 16S rDNA metagenomic analysis, four Swiss cheese blocks from two factories with split and non-split areas were investigated. Result showed no significant difference in starter cultures among split and non-split cheese; however, products made by Factory A carried more Propionibacterium than Factory B. Non-starter bacteria accounted for ~2% of Swiss cheese microbiota. In both factories, the relative abundance of Enterobacteriaceae that fermented lactose in split Swiss cheese was higher than that of non-split products, providing evidence that Enterobacteriaceae could be a potential microbial causing agent of split defect.

In order to comprehensively analyze the Swiss cheese samples made in one factory in Ohio, the microbial profiles of retail cheese with eyes, factory-made Swiss cheese with and without split defects, pasteurized milk, and starter cultures were extensively analyzed. The retail cheese showed different microbial profile in different samples, while the factory-made Swiss cheese were consistent in both microbial abundance and diversity. None of Swiss cheese making materials, including pasteurized milk and several starter cultures, was contaminated with common environmental bacteria, indicating the environmental commensal bacteria have limited impact on the formation of splits. The split area in downgraded Swiss cheese contained more Lactobacillus but less

Streptococcus and Propionibacterium than Swiss cheese samples from normal eye area, suggesting the heterogeneous distribution of starter cultures in the body of Swiss cheese with split defects.

The application of metagenomics in fermented dairy foods is promising because of the easy sample preparation and informative data output. There is no cultivation step, which normally introduces bias when analyzing microbial population and diversity, is involved. However, the sampling and DNA extraction steps are critical in the culture- independent studies. Fermented dairy foods are usually high in milk fat, which will interfere with the DNA extraction step and reduce the yield. Due to different cell membrane structures, the DNA of some bacteria might be harder to extract, resulting in biases in the relative abundance of the bacterial DNA.

The investigation of microorganisms on the quality of cheese products has been conducted by many researchers. Swearingen et al. (2001) studied the nonstarter lactic acid bacteria and suggested the positive contribution of these bacteria to flavor development. Le Bourhis et al. (2007) investigated the contribution of different species of

Clostridium to the Swiss cheese quality, and concluded the presence of C. tyrobutyrucum was associated with late-blowing. They also observed the association of C. beijerinckii or

C. sporogenes to C. tyrobutyricum enhanced the butyric and propionic fermentation and the cheese defects. The study of volatile compounds in Swiss cheese revealed that propionic acid was the most discriminating volatile compound in split and non-split

Swiss cheese from three out of five factories (Castada et al, 2014).

proteolysis, providing enzymes to release free amino acids. These amino acids later become the precursors of a rich profile of flavor compounds, including various volatile compounds. The catabolism of branched-chain amino acids, methionine, and phenylalanine results in the formation of key volatile compounds (Helinck et al., 2004).

Swiss cheese during ripening, and his results showed that the key volatile compounds were 2-butanone, 1-propanol, 2-butanol, and ethanol (Yang et al., 1994).

Microorganisms could be one of the causes of split defects; however, the cheese- making process could also influence the quality of final products. The graininess, a common textural defect in fresh cheese, was introduced during post-processing such as tempering and shearing (Hahn et al., 2012). However, systematic studies involving microbiology, chemistry and engineering to assess and identify the cause of quality defects in fermented foods are in great need.

Table 3.2 The microbial profile of pasteurized milk and starter cultures.

The Relative Abundance Starter Culture Sample Name Major Bacteria Identified Lactobacillus Streptococcus Propionibacterium Pasteurized Milk 99.57% 0.29% 0.14% NA Coccus ST-M8 0.03% 99.97% NA Streptococcus #3225982 Rod Culture 100.00% NA NA Lactobacillus #SSR 126

Rod Culture Lactobacillus helveticus and 100.00% NA NA #LH 100 Lactobacillus delbrueckii Mixed culture of VP- 4 #1071520 and Propionibacteirum and 99.63% 0.01% 0.36% HoldBac LC Lactobacillus rhamnosus #4112478127 NA: not applicable.

Figure 3.1: Bar chart demonstrating the microbial diversity of Swiss cheese samples from two factories. a) Three major starter cultures (to the family level) which accounted for

~98% sequencing reads. b) Major non-starter bacteria (to the family level) which accounted for ~2% sequencing reads.

Figure 3.2: Relative abundances of three starter cultures in split and non-split Swiss cheese samples (a) and in two factories (b).

Figure 3.3: Relative abundances of non-starter bacteria in split and non-split Swiss cheese samples (a) and in two factories (b).

Figure 3.4 The Microbial diversity of Swiss cheese samples from the factory and grocery stores. a) Diversified microbiota of cheese with eyes. V17B, V17E, V20B, V20E, Swiss cheese samples made from same batch; AB, AE, Atlantique Baby Swiss; SB, SE, Swiss

(continued)

Figure 3.4: continued

Emmenthaler Emm; JPB, JPE, JSB, JSE, JSS, Jarlsberg Norseland; PB, PE, Parrano

Uniekaas. b) Consistent diversity of microbiota in Swiss cheese from the same batch in the factory.

Figure 3.5 Relative abundances of three starter cultures in blind, eye and split area of

Swiss cheese made from the same batch. *, P < 0.05.

Figure 3.6 Principal Component Analysis of the microbial profile of factory-made

Swiss cheese samples (eye area vs. split area) from the same batch. Circle was drawn based on P-value of 0.05. V15E, normal eye area in Swiss cheese sample from Vat 15;

V13E, normal eye area in Swiss cheese sample from Vat 13; V10E, normal eye area in

Swiss cheese sample from Vat 10; V9E, normal eye area in Swiss cheese sample from

Vat 9; V8E, normal eye area in Swiss cheese sample from Vat 8; V16E, normal eye area in Swiss cheese sample from Vat 16; V8S, split area in Swiss cheese sample from Vat 8;

V15S, split area in Swiss cheese sample from Vat 15; V13S, split area in Swiss cheese sample from Vat 13.

Figure 3.6

Figure 3.7 A dual dendrogram and heat map showing top 21 bacterial genera across the eye area and split area of Swiss cheese made from the same batch. V15E, normal eye area in Swiss cheese sample from Vat 15; V13E, normal eye area in Swiss cheese sample from Vat 13; V10E, normal eye area in Swiss cheese sample from Vat 10; V9E, normal eye area in Swiss cheese sample from Vat 9; V8E, normal eye area in Swiss cheese sample from Vat 8; V16E, normal eye area in Swiss cheese sample from Vat 16; V8S, split area in Swiss cheese sample from Vat 8; V15S, split area in Swiss cheese sample from Vat 15; V13S, split area in Swiss cheese sample from Vat 13.

Figure 3.7

3.6 Bibliography

Cole, J. R., Wang, Q., Cardenas, E., Fish, J., Chai, B., Farris, R. J., ... & Tiedje, J. M. (2009). The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucleic acids research, 37(suppl 1), D141-D145.

Edgar, R. C., Haas, B. J., Clemente, J. C., Quince, C., & Knight, R. (2011). UCHIME improves sensitivity and speed of chimera detection. Bioinformatics, 27(16), 2194-2200.

Hahn, C., Wachter, T., Nöbel, S., Weiss, J., Eibel, H., & Hinrichs, J. (2012). Graininess in fresh cheese as affected by post-processing: Influence of tempering and mechanical treatment. International Dairy Journal, 26(1), 73-77.

Le Bourhis, A. G., Doré, J., Carlier, J. P., Chamba, J. F., Popoff, M. R., & Tholozan, J. L. (2007). Contribution of C. beijerinckii and C. sporogenes in association with C. tyrobutyricum to the butyric fermentation in Emmental type cheese. International journal of food microbiology, 113(2), 154-163.

Martin, N. H., Murphy, S. C., Ralyea, R. D., Wiedmann, M., & Boor, K. J. (2011). When cheese gets the blues: Pseudomonas fluorescens as the causative agent of cheese

spoilage. Journal of dairy science, 94(6), 3176-3183.Oikonomou, G., Machado, V. S., Santisteban, C., Schukken, Y. H., & Bicalho, R. C. (2012). Microbial diversity of bovine mastitic milk as described by pyrosequencing of metagenomic 16s rDNA. PLoS One, 7(10), e47671.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B. M., Ludwig, W., Peplies, J., & Glöckner, F. O. (2007). SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic acids research, 35(21), 7188-7196.

Rychlik, M., & Bosset, J. O. (2001). Flavour and off-flavour compounds of Swiss Gruyere cheese. Evaluation of potent odorants. International Dairy Journal, 11(11), 895-901.

Sarantinopoulos, P., Kalantzopoulos, G., & Tsakalidou, E. (2002). Effect of Enterococcus faecium on microbiological, physicochemical and sensory characteristics of Greek Feta cheese. International Journal of Food Microbiology, 76(1), 93-105.

Schloss, P. D., Westcott, S. L., Ryabin, T., Hall, J. R., Hartmann, M., Hollister, E. B., ... & Sahl, J. W. (2009). Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Applied and environmental microbiology, 75(23), 7537-7541.

Swearingen, P. A., O'sullivan, D. J., & Warthesen, J. J. (2001). Isolation, characterization, and influence of native, nonstarter lactic acid bacteria on Cheddar cheese quality. Journal of Dairy Science, 84(1), 50-59.

Wang, H. H., Manuzon, M., Lehman, M., Wan, K., Luo, H., Wittum, T. E., ... & Bakaletz, L. O. (2006). Food commensal microbes as a potentially important

avenue in transmitting antibiotic resistance genes. FEMS Microbiology Letters, 254(2), 226-231.

Yang, W. T., & Min, D. B. (1994). Dynamic headspace analyses of volatile compounds of Cheddar and Swiss cheese during ripening. Journal of food science, 59(6), 1309-1312.

Chapter 4 Retail Oyster as a Potential Channel Disseminating Antibiotic

Resistance

4.1 Abstract

Ready-to-consume seafood products, such as shrimp, were found to be an important avenue in the transmitting antibiotic resistant bacteria. However, oysters are mostly consumed raw, but the understanding of antibiotic resistance associated with oysters is very limited. The objectives of this project were to 1) investigate the prevalence and abundance of antibiotic resistant bacteria using cultivation method, and 2) to study the microbial taxonomy, functional metabolic groups, and antibiotic resistome using high-throughput metagenomics with data analyzed by MG-RAST. Phenotypic resistant population against tetracycline, cefotaxime, lincomycin, gentamicin, ciprofloxacin, and ceftazidime were screened by conventional plating in the presence of the corresponding antibiotics. Antibiotic resistomes were further assessed by metagenomics using the genomic DNA of oysters. A diversified antibiotic resistant bacterial profile is revealed, with Proteobacteria in the highest abundance, followed by Mollicutes and Firmicutes.

Genes coding for antibiotic resistance were identified in oyster microbiota. Various antibiotic resistance genes were in high abundance, including acriflavin resistance genes, multidrug resistance genes, genes encoding transporters, and tetracycline resistance genes, among others. This study provided a comprehensive overview of the oyster microbial metagenome, which may deserve more public health concerns.

4.2 Introduction

The use of antibiotics in aquaculture has well-known positive effects on the control of bacterial infection. However, side effects that affect both aquaculture and the environment are observed. From 1973 to 2011, a total of 55 antibiotic-resistant foodborne outbreaks were identified. Based on FDA guidance, approved aquaculture drugs include florfenicol, oxytetracycline, sulfamerazine, sulfadimethoxine/ormetoprim combination, while chloramphenicol, clenbuterol, diethylstilberstrol, fluoroquinolones are prohibited to use in fish (FDA, 2011). Done and Halden performed reconnaissance of antibiotics in shrimp, salmon, tilapia and trout sold in the US, and 5 out of 47 antibiotics were detected with oxytetracycline the most common antibiotics. All samples showed compliance under current federal regulation with low level of drug residues (Done et al., 2015).

Many ready-to-eat seafood containing commensal bacteria with antibiotic resistance genes were consumed without further processing, such as heat, acids, fermentation, although these steps are common in other food products. Recent studies showed that ART bacteria are prevalent in various aquaculture products. Wang et al.

(2011) examined 171 salmon, shrimp, and tilapia samples imported from 12 countries.

Their results showed 26.3% samples contained pathogens, including Campylobacter jejuni, Escherichia coli, Listeria monocytogenes, Salmonella Typhimurium, and Vibrio parahaemolyticus. Of these pathogenic isolates, 2 (100%) C. jejuni were resistant to gentamicin; 75% of L. monocytogenes isolates were resistant to nitrofurantoin; 63.3% of

V. parahaemolyticus showed intermediate resistance to ampicillin; 13% of E. coli isolates

were resistant to trimethoprim-sulfamethoxazole (Wang et al., 2011). DuráN and

Marshall (2005) examined thirteen brands of ready-to-eat shrimp originating from four different countries and found high prevalence of ART genes that 657 (42%) of the isolates were ART bacteria, representing 131 (81%) species (DuráN et al., 2005).

Antibiotic resistance was also detected in Vibrio spp. from retail raw oyster in Louisiana

(Han et al., 2007), in V. parahaemolyticus isolated from shellfish from Georgia and

South Carolina (Baker-Austin et al., 2008), in Salmonella from the imported seafood products (Khan et al., 2009), in E. coli O157:H7 from retail shrimp from India

(Surendraraj et al., 2010), and in V. parahaemolyticus and V. alginolyticus from farmed fish from Korea (Oh et al., 2011). Previous study in our lab illustrated the antibiotic resistant bacteria associated with retail aquaculture from Guangzhou, China. All products were found with antibiotic resistant commensal bacteria. Antibiotic resistance genes including sul1, sul2, tetE, ermB, ermC, blaTEM, blaCMY were identified in the resistant bacteria (Ye et al., 2013). In the absence of selective pressure, the antibiotic resistance traits in many isolates were stable. Huang et al. (2014) investigated a total of 4747 phenotypically antibiotic resistant isolates which were originated from fish, water and pond with no antibiotic application history. The results suggested that the antibiotic resistance gene traits in many isolates were stable without the selective pressure, and the antibiotic resistance risk factors were independent from direct antibiotic exposure in aquaculture production.

A preliminary study on oysters and clams was conducted in our lab on the antibiotic resistant microbiota that were culturable on media with selective antibiotics.

Result of the study showed high prevalence of antibiotic resistant bacteria that could grow on Brain Heart Infusion plates. Based on the result of 16S metagenomics, about 50% of the isolates were classified as Pseudomonas. The preliminary study provided supportive information on the abundance of antibiotic resistant bacteria in seafood.

Although emerging evidences indicated that ready-to-consume seafood could be carriers for antibiotic resistant bacteria, the real magnitude and spectrum of AR affiliated with seafood products have not been fully elucidated. The objectives of this chapter were to 1) investigate the prevalence of bacteria with antibiotic resistance phenotypes in ready- to-eat oysters, and 2) study the antibiotic resistome in oysters using shotgun sequencing and metagenomics.

100

4.3 Materials and Methods

4.3.1 The prevalence of bacteria with selected antibiotic resistance phenotype

4.3.1.1 Sampling

A total of nine wild caught oyster samples (3 Blue Point oysters, 3 Sewansecott oysters, and 3 Virginia oysters) were collected from grocery stores. Each oyster was opened and the flesh was aseptically weighed and transferred into the stomach bag, followed by adding 4 times volume by weight of sterile 3.5% sodium chloride solution.

Oyster homogenates were obtained through light stomaching for 60 seconds assisted with hand massage for 1 minute.

4.3.1.2 Bacterial enumeration

All samples were enumerated by plating 100 μl of diluted homogenates on three media, including Brain Heart Infusion (BHI) agar (Becton, Dickinson and Company,

Franklin Lakes, NJ) for the cultivation of fastidious microorganisms, MacConkey (MAC) agar (Becton, Dickinson and Company) for the growth of Gram-negative and enteric bacteria, and Phenylethyl Alcohol (PEA) agar (Becton, Dickinson and Company) for the growth of Gram-positive bacteria. One of the following antibiotics was added to media according to Table 4.1: 16 μg/ml of tetracycline (Sigma-Aldrich, St. Louis, MO), 2 μg/ml of cefotaxime (Sigma-Aldrich), 4 μg/ml of ciprofloxacin (Sigma-Aldrich), 16 μg/ml of

101

gentamycin (Sigma-Aldrich), or 32 μg/ml of ceftazidime (Sigma-Aldrich) based on the spectrum of the antibiotics. Cycloheximde was added to all agar plates in 100 μg/ml to minimize the growth of yeast and mold. One hundred micro-liter of oyster homogenates was serially diluted and plated in duplicate on the corresponding bacterial agar plates.

The plates were then incubated under anaerobic condition using GasPak™ EZ anaerobe sachets (Bento, Dickinson and Company) with indicator at 30ºC for 48 hours before the plate counts were obtained.

102

Table 4.1 Media and antibiotics used in the cultivation of antibiotic resistance bacteria isolated from oyster.

Antibiotics Class Spectrum Mode of Action BHI MAC PEA Inhibition of the binding of aminoacyl- Tetracycline Tetracycline Broad √ √ √ tRNA to the mRNA-ribosome complex Inhibit bacterial cell wall synthesis by Cefotaxime β-lactam Broad binding to one or more of the penicillin- √ √ √ binding proteins (PBPs) Inhibition of protein synthesis in Against G+ susceptible bacteria by binding to the 50S Lincomycin lincosamide and cell wall- subunits of ribosomes and preventing √ × √

103 less bacteria peptide bond formation upon

transcription Against Irreversibly bind to specific 30S subunit Gentamicin aminoglycoside √ √ × Pseudomonas proteins and 16S rRNA Inhibition of topoisomerase (DNA gyrase) enzymes, which inhibits Against Ciprofloxacin Quinolone relaxation of supercoiled DNA and √ √ × Pseudomonas promotes breakage of double stranded DNA Interference with bacterial cell wall Against ceftazidime β-lactam synthesis and inhibiting cross-linking of √ √ × Pseudomonas the peptidoglycan

103

4.3.2 Antibiotic resistance gene profiling by shotgun sequencing metagenomics

4.3.2.1 DNA preparation

A total of eight wild caught oyster samples were collected from grocery stores.

Sample origins and IDs were listed in Table 4.2. The oyster homogenates for each sample was obtained using the same method described in 4.3.1.1. In order to remove the eukaryotic cells, for each oyster, all the homogenates were collected and centrifuged at

1085 ×g for 2 minutes. The supernatant was then centrifuged at 6040 ×g for 20 minutes.

The genomic DNA of pellets was extracted using QIAamp DNA Stool Mini Kit (Qiagen,

Germany). The quality and size of the DNA was evaluated by NanoDrop and 2% agar gel electrophoresis. Genomic DNA of each oyster sample was pooled at the same amount.

Table 4.2 The origins and the taste of the retail live oyster samples.

ID Oyster Name Origins Taste

Medium, salinity, plump and BP Blue Point Long Island Sound, CT juicy

SH Salty Hog Hog Island Bay, VA Salty, full texture

CS Choptank Sweet Choptank River, MD Sweet, with a clean finish

104

4.3.2.2 Antibiotic resistome analysis by shotgun sequencing metagenomcis

The genomic DNA of oysters was sent for sequencing by Illumina HiSeq 3000

Sequencer in Biomedical Genomics Core, Nationwide Children’s Hospital, Columbus,

OH. Unlike 16S rRNA, sequencing which relies on a PCR step to specifically amplify the hypervariable region of 16S rRNA. In this study, there is no PCR step and the genomic

DNA was sheared and sequenced directly. A total of 18.3 Gigabyte DNA sequences was obtained by high through-put sequencing.

4.3.2.3 Shotgun sequencing data analysis

The sequencing data was processed by the metagenomics RAST server (MG-

RAST) (Meyer et al., 2008) under the accession number 4705834.3. Low quality sequences were removed to ensure that quality greater than 15. The replicate sequences were removed by MG-RAST QC pipeline. After the above quality filtering, a total of

91,211,730 reads were obtained for subsequent analysis of oyster bacterial metagenome.

The quality-filtered reads were submitted to the MG-RAST for taxonomic classification and function analysis. Taxonomic analysis, functional classification, and antibiotic resistome was conducted by annotation with databases in MG-RAST.

105

4.4 Results

4.4.1 The prevalence and abundance of bacteria with antibiotic resistance phenotypes in oyster samples

Bacteria resistant to tetracycline, cefotaxime, lincomycin, gentamicin, ceftazidime and ciprofloxacin were found in all three types of retail oysters, as illustrated in Figure A.

Virginia oysters carried the most bacteria that showed resistance phenotypes against antibiotics, when comparing the abundance of antibiotic resistance bacteria among different types of retail oysters.

On BHI agar plates (Figure A-a), which were used for the cultivation of a wide variety of fastidious bacteria, lincomycin-resistant bacteria were the most abundant antibiotic resistant bacteria (103 - 105 CFU/g); while gentamicin-resistant bacteria were

102-104 CFU/g, and cefotaxime-resistant bacteria were 10-103 CFU/g. The abundance of tetracycline-, ceftazidime- and ciprofloxacin-resistant bacteria were low, which were below or around 102 CFU/g, indicating the high susceptibility of bacteria in oysters to these antibiotics.

MacConkey agar plates were a selective medium for isolating Gram-negative and enteric bacilli by using the crystal violet and bile salts to inhibit the growth of gram- positive microorganisms. The total bacteria count without antibiotic pressure on MAC plates was 1 log lower than that on BHI plates. Although there was no significant difference between the bacterial count from MAC and BHI plates with tetracycline,

106

cefotaxime, ceftazidime and ciprofloxacin, gentamicin showed great selective power on

Gram-negative bacteria (102-104 CFU/g on BHI plates and 101-103 CFU/g on MAC plates).

Another selective medium, phenylethyl alcohol agar plates were also used to investigate the antibiotic resistant bacteria in retail oysters. PEA plates supported growth of Gram-positive bacteria and inhibited or greatly reduced growth of Gram-negative bacteria by interfering with DNA synthesis by the active ingredients phenylethyl alcohol.

On PEA plates with no antibiotics, the total bacteria count reduced by 1 log as compared to the count on BHI plates, indicating the selection under by the chosen cultivation media.

107

4.4.2 The bacterial profile of oyster samples

The taxonomic classification of rRNA genes was conducted based on the SILVA

LSU database in MG-RAST and the result was shown in Figure B. Proteobacteria (49%),

Firmicutes (26%), Mollicutes (12%), Bacteroidetes (6%), and Fusobacteria (4%) were the dominant phylum in the oyster microbiota. This result was supported by a previous study indicating high abundance of Proteobacteria, Mollicutes, Firmicutes were found in the stomach and gut of oysters (King et al., 2012). In Proteobacteria, 56% was

Gammaproteobacteria, which included Vibrio (25%), Shewanella (17%), Pseudomonas

(7%), etc. In Firmicutes, Lactobacillus accounted for 57% of the reads, while 18% was

Clostridium. However, in Mollicutes, 92% was Mycoplasma.

4.4.3 Functional analysis of oyster microbiota

The annotation of functional genes was conducted based on the SEED database in

MG-RAST and the result was shown in Figure C. A total of 670,100 predicted functions were generated based on SEED level 1 subsystems database. Figure 4.3 showed the relative abundance of 27 basic metabolic categories in the oyster metagenome. The most abundant category was Clustering-based subsystems (15.5%), in which there was functional coupling evidence that genes belong together. Protein metabolism and re were the 2nd and 3rd abundant SEED subsystem representing 11.7% and 9.6% of oyster metagenome. Protein metabolism was selected for further analysis using the MG-RAST

108

pipeline. The annotated sequences of protein metabolism in oyster microbiota were assigned to 5 subsystems at level 2, in which protein biosynthesis was the most abundant

(75% of annotated sequences in protein metabolism), followed by protein degradation

(13%), protein processing and modification (6%) and protein folding (5%). It was worth mentioning that 2% of the annotated functional sequences was classified into virulence, disease and defense subsystem, where 73% was related to resistance to antibiotics and toxic compounds.

4.4.4 Analysis of antibiotic resistance genes

In order to comprehensively investigate the antibiotic resistance genes in oyster microbiota, the sequences were annotated by GenBank database implemented in MG-

RAST pipeline. A total of 1,572,735 annotations, among which 2058 annotations (0.13%) were antibiotic resistance genes, were generated. Based on their different functions, the reads were grouped into multidrug resistance genes, transporters, acriflavin resistance genes, bicyclomycin resistance genes, tetracycline resistance genes, etc. as shown in

Figure A.4. Multidrug resistance genes (53.16%) were the most abundant type of antibiotic resistance genes. Acriflavin resistance genes were the second most abundant type that consisted of 38.19% of antibiotic resistance genes. Acriflavin is a type of antibiotics that is normally used as a broad-spectrum prophylactic agent and therapeutant, and is widely used in aquaculture production (Bondad-Reantaso, 2007). Transporters reported to be 32.51% of the total antibiotic resistance genes. Bicyclomycin and

109

tetracycline resistance genes represented 4.81% and 4.23% of total annotated antibiotic resistance genes, respectively. It worth mentioning that bicyclomycin pump genes were found to be the dorminant antibiotic resistance gene type in Hatosy’s study (2015) on the antibiotic resistance gene profiles of seawater. Aminoglycoside (streptomycin and kanamycin)-, -lactam (methicillin)-, macrolide (erythromycin)-, glycopepetide

(vancomycin)-, lincosamide (lincomycin)-, bacitracin- and polymyxin-resistance genes were also found in low abundance in oyster microbiota.

110

Figure 4.1 Prevalence of antibiotic resistance bacteria in oyster samples by demonstrating the colony forming unit of each sample. A) Antibiotic resistance bacteria growth on BHI agar plates. BT, BHI with tetracycline; BX, BHI with cefotaxime; BL,

BHI with lincomycin; BG, BHI with gentamicin; BZ, BHI with ceftazidime; BP, BHI with ciprofloxacin; BC, BHI without antibiotics. b) Antibiotic resistance bacteria growth on MAC agar plates. MT, MAC with tetracycline; MX, MAC with cefotaxime; MG,

MAC with gentamicin; MZ, MAC with ceftazidime; MP, MAC with ciprofloxacin; MC,

MAC without antibiotics. c) Antibiotic resistance bacteria growth on PEA agar plates. PT,

PEA with tetracycline; PX, PEA with cefotaxime; PL, PEA with lincomycin; PC, PEA without antibiotics. SSOi, SSOii, SSOiii, 3 Sewansecott oyster individuals; BPOi, BPOii,

BPOiii, 3 Blue Point oyster individuals; Vai, Vaii, Vaiii, 3 Virginia oyster individuals.

◊ Bacteria count was below detection limit.

111

Figure 4.1

112

Figure 4.2 The tree chart of bacteria abundance in oyster samples. Color shading indicated the phylum level. Leaf weight were displayed by stacked bar. The data was compared to SILVA LSU using a maximum e-value of 1e-5, a minimum identity of 60 %, and a minimum alignment length of 15 measured bp for RNA databases.

113

Figure 4.3 The bar chart for the relative abundance of functional gene groups in oyster samples. The data was generated using a maximum e-value of 1e-5, a minimum identity of 60 %, and a minimum alignment length of 15 measured in amino acids for protein. The data was normalized to values between 0 and 1.

114

others 2.28%

tetracycline resistance 4.23%

bicyclomycin resistance 4.81%

transporter 32.51%

acriflavin resistance 38.19%

multidrug 53.16%

0% 10% 20% 30% 40% 50% 60%

Figure 4.4 The antibiotic resistance gene pattern in retail oysters. The resistance gene groups were obtained after alignment of the high-throughput sequencing reads to SEED

LSU database. The percentages represented the proportion of the reads of each gene in the total reads of all the annotated antibiotic resistance genes in oyster metagenome.

115

4.5 Conclusions and Discussions

Oysters are popular seafood worldwide. They are members of family Ostreidae, in which the edible oysters are mainly belong to the genera Ostrea, Crassostrea, Ostreola and Saccostrea. In this study, the microbiota of 4 types of selective oysters was extensively explored. Using conventional cultivation method, the prevalence and abundance of bacteria to 6 antibiotics were characterized. Using metagenomics, the taxonomic annotations for high-throughput sequencing reads revealed the diversity and abundance of the microbial profile in oyster samples. The result showed that

Proteobacteria, Mollicutes and Firmicutes dominated in the oyster metagenome.

Sequencing reads also revealed clustering-based subsystem, protein metabolism and carbohydrate metabolism were the most abundant SEED subsystem in the microbial community. Antibiotic resistance genes analysis revealed high abundance of acriflavin resistance and multidrug resistance genes followed by transporter and tetracycline resistance genes. By investigating the high-throughput sequencing of the microbial DNA of ready to consume oysters, this study provided a comprehensive overview of the taxonomic and functional analysis, as well as the antibiotic resistome in oyster microbiota.

Commensal bacteria in ready-to-consume food, due to its direct exposure to human gut microbiota after digestion, serve as a potential important avenue in transmitting antibiotic resistance genes. Consumers choose to eat raw oyster for its soft and fleshy texture as well as good flavor. However, oyster could be associated with food safety issues. Cooke examined coliforms from shellfish including oyster, and found 88.7%

116

of the isolates were resistant to at least one type of antibiotics, 62.2% of the isolates were resistant to more than one antibiotic, and some isolates resistant to streptomycin or tetracycline were capable of transferring their resistance pattern to E. coli K-12 (Cooke et al., 1975). In 1976, Morgan et al. (1976) identified ampicillin resistant E. coli, ampicillin resistant Enterobacter cloacae, as well as streptomycin, ampicillin or tetracycline resistant

Pseudomonas spp. in oyster from Chesapeake Bay. The study conducted by Brands et al.

(2005) showed 7.4% of U.S. market oysters contained Salmonella enterica serovar

Newport, which was resistant to ampicillin and tetracycline. Many studies have been conducted on the antibiotic resistance pathogens found in oysters; however, our knowledge in the prevalence of antibiotic resistance commensals and the antibiotic resistome in oysters is limited.

Previous studies in the antibiotic resistant bacterial profile in oysters were culture- dependent, which means a vast proportion of unculturable antibiotic resistant bacteria were not taken into consideration. By applying culture-independent metagenomics, we were able to explore the antibiotic resistant bacteria more extensively. The high abundance of acriflavin resistance genes including AcrB/AcrD/AcrF were found in oyster microbiota. This was supported by the fact that acriflavin was a common type of antibiotics used in aquaculture production by directly adding into water (Chua, 1996).

Multidrug resistance genes (MRPs, acrB/acrD/acrF, MarC, efflux pump, Mdt, NorM, etc.) and genes encoding transporter (ABC transporter, Bcr/CflA, EmrB/QacA, etc.) were also in high abundance among all the annotated antibiotic resistance genes.

117

It was worth mentioning that the genes encoding bacterial mobile genetic elements were found in oyster metagenome, such as transposon and integrase. The presence of these genes might facilitate the horizontal transfer of antibiotic resistance genes from antibiotic resistant bacteria to other host microbes.

118

Chapter 5 Summary and Conclusions

The detection of food microorganisms usually involves bacterial enumeration, due to the high detection limit of the cultivation and plating methods. For example, the detection limit of conventional sampling followed by serial dilution is often above 10

CFU/g of samples; not to say the biases caused by the sampling step – only a small part of the sample could be spread onto agar plates. To overcome the conventional detection hurdle, culture-independent rapid detection platform of microorganisms has received increasing attention for food quality and safety control in the food industry because of minimal sample preparation with fast and accurate results. Theoretically, PCR is able to amplify even single DNA molecules, making it an ideal tool for the detection of bacteria that is in relatively low abundance in the microbiota, such as pathogenic bacteria. The sample preparation step determines the quality of DNA, which affects the accuracy and quality of the detection results.

The application of metagenomics in the food industry offers a revolutionary tool in many fields. It can be used to improve the food quality by monitoring and identifying

119

microflora changes during the production or aging of food products, such as fermented foods. Metagenomics is also able to identify the source of microbial contaminations or the presence of pathogens.

It is well accepted that high throughput sequencing and metagenomics ensure swift and reliable characterization of millions of bacteria simultaneously in food samples.

It requires far less benchtop work than conventional methods. Additionally, the shotgun sequencing is able to obtain insights into the functional genes on microbial activities, making the metagenomics result more informative than just revealing microbial profiles.

However, there are several factors that limit the wide use of metagenomics in food industry. First, the result of high throughput sequencing cannot be interpreted without sufficient bioinformatics analysis. Although “blast” is the core of data analysis, the preprocessing of the raw data before blast step involves quality check to eliminate sequences with too many ambiguous nucleotides or in low quality. In whole genome sequencing or paired end sequencing, the processed data will be further assembled before alignment with databases. These preprocessing steps often require extensive biostatistic and computational analysis, of which most food scientists have limited knowledge.

Secondly, although the cost of high-throughput sequencing is decreasing drastically over the past years, for food microbiology research groups and food industry, the cost for sequencing service per sample as well as computer workstations for data analysis are still comparatively high. When studying the microbiological profile of food samples, there is another limitation for metagenomics which is the taxonomic resolution. Unlike Pulsed-

120

field gel electrophoresis or other molecular techniques, 16S metagenomics has limited discriminative power in differentiating bacteria at the species level. Most researchers utilized the high-throughput sequencing reads at the genus or higher level for taxonomic classification, depending on different sequencing library – 16S or whole genome.

However, when it comes to differentiating bacteria species, highly discriminative molecular typing techniques are irreplaceable currently. With the development of cost- efficient sequencing platforms, culture-independent techniques such as metagenomics will be more widely used in food microbiota studies in the coming years, although limitations must be carefully considered.

Using culture-independent molecular techniques, the study systematically investigated the microbial profiles in premium and split Swiss cheese, and developed a rapid detection platform for the identification of specific bacteria in food matrices. It is the leading study that applies high throughput sequencing and metagenomics in the quality assessment of Swiss cheese, by addressing a significant cheese industrial challenge with great economic impact. The species-level Propionibacterium detection system developed in this study is specific and cost-efficient, with high industry application potential.

121

4.6 Bibliography

Baker-Austin, C., McArthur, J. V., Tuckfield, R. C., Najarro, M., Lindell, A. H., Gooch, J., & Stepanauskas, R. (2008). Antibiotic resistance in the shellfish pathogen Vibrio parahaemolyticus isolated from the coastal water and sediment of Georgia and South Carolina, USA. Journal of Food Protection®, 71(12), 2552-2558.

Bondad-Reantaso, M. G. (Ed.). (2007). Assessment of freshwater fish seed resources for sustainable aquaculture (No. 501). Food & Agriculture Org..

Brands, D. A., Billington, S. J., Levine, J. F., & Joens, L. A. (2005). Genotypes and antibiotic resistance of Salmonella Newport isolates from US market oysters. Foodbourne Pathogens & Disease, 2(1), 111-114.

Bruant, G., Maynard, C., Bekal, S., Gaucher, I., Masson, L., Brousseau, R., & Harel, J. (2006). Development and validation of an oligonucleotide microarray for detection of multiple virulence and antimicrobial resistance genes in Escherichia coli. Applied and environmental microbiology, 72(5), 3780-3784.

Cetinkaya, Y., Falk, P., & Mayhall, C. G. (2000). Vancomycin-resistant enterococci. Clinical microbiology reviews, 13(4), 686-707.

Chauhan, A., Wafula, D., Lewis, D. E., & Pathak, A. (2014). Metagenomic assessment of the Eastern oyster-associated microbiota. Genome announcements, 2(5), e01083- 14.

Chen, S., Zhao, S., McDermott, P. F., Schroeder, C. M., White, D. G., & Meng, J. (2005). A DNA microarray for identification of virulence and antimicrobial resistance genes in Salmonella serovars and Escherichia coli. Molecular and cellular probes, 19(3), 195-201.

Chua, F. H. C. (1996). Aquaculture health management in Singapore: current status and future directions. FAO Fisheries Technical Paper (FAO).

122

Cooke, M. D. (1976). Antibiotic resistance among coliform and fecal coliform bacteria isolated from sewage, seawater, and marine shellfish. Antimicrobial agents and chemotherapy, 9(6), 879-884.

Davis, B. D. (1987). Mechanism of bactericidal action of aminoglycosides. Microbiological reviews, 51(3), 341.

D’Costa, V. M., King, C. E., Kalan, L., Morar, M., Sung, W. W., Schwarz, C., ... & Golding, G. B. (2011). Antibiotic resistance is ancient. Nature, 477(7365), 457- 461.

Done, H. Y., & Halden, R. U. (2015). Reconnaissance of 47 antibiotics and associated microbial risks in seafood sold in the United States. Journal of hazardous materials, 282, 10-17.

DuráN, G. M., & Marshall, D. L. (2005). Ready-to-eat shrimp as an international vehicle of antibiotic-resistant bacteria. Journal of Food Protection®, 68(11), 2395-2401.

Fernandez‐Piquer, J., Bowman, J. P., Ross, T., & Tamplin, M. L. (2012). Molecular analysis of the bacterial communities in the live Pacific oyster (Crassostrea gigas) and the influence of postharvest temperature on its structure. Journal of applied microbiology, 112(6), 1134-1143.

Fish and Fishery Products Hazards and Controls Guidance, Fourth Edition. (2011, April). Retrieved March 1, 2015, from Chapter 11, http://www.fda.gov/downloads/Food/GuidanceRegulation/UCM251970.pdf

Food and Agriculture Organization of the United Nations Statistics Division. (n.d.).Retrieved June 4, 2016, from http://faostat3.fao.org

Foodborne Outbreak Tracking and Reporting. (n.d.). Retrieved June 4, 2016, from http://wwwn.cdc.gov/foodborneoutbreaks/

123

Han, F., Walker, R. D., Janes, M. E., Prinyawiwatkul, W., & Ge, B. (2007). Antimicrobial susceptibilities of Vibrio parahaemolyticus and Vibrio vulnificus isolates from Louisiana Gulf and retail raw oysters. Applied and environmental microbiology, 73(21), 7096-7098.

Han, W., Liu, B., Cao, B., Beutin, L., Krüger, U., Liu, H., ... & Wang, L. (2007). DNA microarray-based identification of serogroups and virulence gene patterns of Escherichia coli isolates associated with porcine postweaning diarrhea and edema disease. Applied and environmental microbiology, 73(12), 4082-4088.

Hatosy, S. M., & Martiny, A. C. (2015). The ocean as a global reservoir of antibiotic resistance genes. Applied and environmental microbiology, 81(21), 7593-7599.

Hooper, D. C. (1995). Quinolone mode of action. Drugs, 49(2), 10-15.

Hu, Y., Yang, X., Qin, J., Lu, N., Cheng, G., Wu, N., ... & Meng, Z. (2013). Metagenome-wide analysis of antibiotic resistance genes in a large cohort of human gut microbiota. Nature communications, 4.

Huang, Y., Zhang, L., Tiu, L., & Wang, H. H. (2015). Characterization of antibiotic resistance in commensal bacteria from an aquaculture ecosystem. Frontiers in microbiology, 6.

Israel, D. A., Salama, N., Arnold, C. N., Moss, S. F., Ando, T., Wirth, H. P., ... & Peek Jr, R. M. (2001). Helicobacter pylori strain-specific differences in genetic content, identified by microarray, influence host inflammatory responses. Journal of Clinical Investigation, 107(5), 611.

Khan, A. A., Ponce, E., Nawaz, M. S., Cheng, C. M., Khan, J. A., & West, C. S. (2009). Identification and characterization of class 1 integron resistance gene cassettes among Salmonella strains isolated from imported seafood. Applied and environmental microbiology, 75(4), 1192-1196.

King, G. M., Judd, C., Kuske, C. R., & Smith, C. (2012). Analysis of stomach and gut microbiomes of the eastern oyster (Crassostrea virginica) from coastal Louisiana, USA. PLoS One, 7(12), e51475.

Mazzei, T., Mini, E., Novelli, A., & Periti, P. (1993). Chemistry and mode of action of macrolides. Journal of Antimicrobial Chemotherapy, 31(suppl C), 1-9.

Meyer, F., Paarmann, D., D'Souza, M., Olson, R., Glass, E. M., Kubal, M., ... & Wilkening, J. (2008). The metagenomics RAST server–a public resource for the automatic phylogenetic and functional analysis of metagenomes. BMC bioinformatics, 9(1), 1.

124

Morgan, R. C., Guerry, P., & Colwell, R. R. (1976). Antibiotic resistant bacteria in Chesapeake Bay. Chesapeake Science, 17(3), 216-219.

Navarre, W. W., Halsey, T. A., Walthers, D., Frye, J., McClelland, M., Potter, J. L., ... & Libby, S. J. (2005). Co‐regulation of Salmonella enterica genes required for virulence and resistance to antimicrobial peptides by SlyA and PhoP/PhoQ. Molecular microbiology, 56(2), 492-508.

Oh, E. G., Son, K. T., Yu, H., Lee, T. S., Lee, H. J., Shin, S., ... & Kim, J. (2011). Antimicrobial resistance of Vibrio parahaemolyticus and Vibrio alginolyticus strains isolated from farmed fish in Korea from 2005 through 2007. Journal of Food Protection®, 74(3), 380-386.

Omura, S. (Ed.). (2002). Macrolide antibiotics: chemistry, biology, and practice. Academic press.

Strommenger, B., Schmidt, C., Werner, G., Roessle-Lorch, B., Bachmann, T. T., & Witte, W. (2007). DNA microarray for the detection of therapeutically relevant antibiotic resistance determinants in clinical isolates of Staphylococcus aureus. Molecular and cellular probes, 21(3), 161-170.

Surendraraj, A., Thampuran, N., & Joseph, T. C. (2010). Molecular screening, isolation, and characterization of enterohemorrhagic Escherichia coli O157: H7 from retail shrimp. Journal of Food Protection®, 73(1), 97-103.

Tomasz, A. (1979). The mechanism of the irreversible antimicrobial effects of penicillins: how the beta-lactam antibiotics kill and lyse bacteria. Annual Reviews in Microbiology, 33(1), 113-137.

Vartoukian, S. R., Palmer, R. M., & Wade, W. G. (2010). Strategies for culture of ‘unculturable’bacteria. FEMS microbiology letters, 309(1), 1-7.

Wang, F., Jiang, L., Yang, Q., Han, F., Chen, S., Pu, S., ... & Ge, B. (2011). Prevalence and antimicrobial susceptibility of major foodborne pathogens in imported seafood. Journal of Food Protection®, 74(9), 1451-1461.

125

Waxman, D. J., & Strominger, J. L. (1983). Penicillin-binding proteins and the mechanism of action of beta-lactam antibiotics1. Annual review of biochemistry, 52(1), 825-869.

Ye, L., Lu, Z., Li, X., Shi, L., Huang, Y., & Wang, H. H. (2013). Antibiotic-resistant bacteria associated with retail aquaculture products from Guangzhou, China. Journal of Food Protection®, 76(2), 295-301.

Zhang, T., Zhang, X. X., & Ye, L. (2011). Plasmid metagenome reveals high levels of antibiotic resistance genes and mobile genetic elements in activated sludge. PloS one, 6(10), e26041.

Zhu, L. X., Zhang, Z. W., Wang, C., Yang, H. W., Jiang, D., Zhang, Q., ... & Cheng, J. (2007). Use of a DNA microarray for simultaneous detection of antibiotic resistance genes among staphylococcal clinical isolates. Journal of clinical microbiology, 45(11), 3514-3521.

126

List of References

Chapter 1

127

Cetinkaya, Y., Falk, P., & Mayhall, C. G. (2000). Vancomycin-resistant enterococci. Clinical microbiology reviews, 13(4), 686-707.

Dairy products: Per capita consumption, United States (Annual). (2015, September 2). Retrieved April 26, 2016, from http://www.ers.usda.gov/data-products/dairy- data.aspx

Darilmaz, D. O., & Beyatli, Y. (2012). Acid-bile, antibiotic resistance and inhibitory properties of propionibacteria isolated from Turkish traditional home-made cheeses. Anaerobe, 18(1), 122-127.

Dasen, G., Smutny, J., Teuber, M., & Meile, L. (1998). Classification and identification of propionibacteria based on ribosomal RNA genes and PCR. Systematic and applied microbiology, 21(2), 251-259.

128

Davis, B. D. (1987). Mechanism of bactericidal action of aminoglycosides. Microbiological reviews, 51(3), 341.

D’Costa, V. M., King, C. E., Kalan, L., Morar, M., Sung, W. W., Schwarz, C., ... & Golding, G. B. (2011). Antibiotic resistance is ancient. Nature, 477(7365), 457- 461.

DuráN, G. M., & Marshall, D. L. (2005). Ready-to-eat shrimp as an international vehicle of antibiotic-resistant bacteria. Journal of Food Protection®, 68(11), 2395-2401.

Fromm, H. I., & Boor, K. J. (2004). Characterization of pasteurized fluid milk shelf‐life attributes. Journal of food science, 69(8), M207-M214.

Feligini, M., Brambati, E., Panelli, S., Ghitti, M., Sacchi, R., Capelli, E., & Bonacina, C. (2014). One-year investigation of Clostridium spp. occurrence in raw milk and

129

curd of Grana Padano cheese by the automated ribosomal intergenic spacer analysis. Food Control, 42, 71-77.

Food and Agriculture Organization of the United Nations Statistics Division. (n.d.).Retrieved June 4, 2016, from http://faostat3.fao.org

Hartmann, M., Enkerli, J., & Widmer, F. (2007). Residual polymerase activity‐induced bias in terminal restriction fragment length polymorphism analysis. Environmental microbiology, 9(2), 555-559.

Hartmann, M., & Widmer, F. (2008). Reliability for detecting composition and changes of microbial communities by T-RFLP genetic profiling. FEMS microbiology ecology, 63(2), 249-260.

Hooper, D. C. (1995). Quinolone mode of action. Drugs, 49(2), 10-15.

Hsieh, H. S., Chai, T. J., & Hwang, D. F. (2007). Using the PCR–RFLP method to identify the species of different processed products of billfish meats. Food Control, 18(4), 369-374.

Huth, P. J., DiRienzo, D. B., & Miller, G. D. (2006). Major scientific advances with dairy foods in nutrition and health. Journal of Dairy Science, 89(4), 1207-1221.

130

Hutkins, R. W. (2008). Microbiology and technology of fermented foods (Vol. 22). John Wiley & Sons.

Mazzei, T., Mini, E., Novelli, A., & Periti, P. (1993). Chemistry and mode of action of macrolides. Journal of Antimicrobial Chemotherapy, 31(suppl C), 1-9.

Nesbakken, T., Kapperud, G., & Caugant, D. A. (1996). Pathways of Listeria monocytogenes contamination in the meat processing industry. International journal of food microbiology, 31(1), 161-171.

131

Omura, S. (Ed.). (2002). Macrolide antibiotics: chemistry, biology, and practice. Academic press.

Park, H. S., Reinbold, G. W., & Hammond, E. G. (1967). Role of propionibacteria in split defect of Swiss cheese. Journal of dairy science, 50(6), 820-823.

Piveteau, P. (1999). Metabolism of lactate and sugars by dairy propionibacteria: a review. Le Lait, 79(1), 23-41.

132

Riedel, K. H., & Britz, T. J. (1996). Justification of the “classical” Propionibacterium species concept by ribotyping. Systematic and applied microbiology, 19(3), 370- 380.

Rossi, F., Dellaglio, F., & Torriani, S. (2006). Evaluation of recA gene as a phylogenetic marker in the classification of dairy propionibacteria. Systematic and applied microbiology, 29(6), 463-469.

Rychlik, M., & Bosset, J. O. (2001). Flavour and off-flavour compounds of Swiss Gruyere cheese. Evaluation of potent odorants. International Dairy Journal, 11(11), 895-901.

Schmidt, V. S., Kaufmann, V., Kulozik, U., Scherer, S., & Wenning, M. (2012). Microbial biodiversity, quality and shelf life of microfiltered and pasteurized

133

extended shelf life (ESL) milk from Germany, Austria and Switzerland. International journal of food microbiology, 154(1), 1-9.

Stackebrandt, E., Cummins, C. S., & Johnson, J. L. (2006). Family propionibacteriaceae: the genus Propionibacterium. In The prokaryotes (pp. 400-418). Springer New York.

Tomasz, A. (1979). The mechanism of the irreversible antimicrobial effects of penicillins: how the beta-lactam antibiotics kill and lyse bacteria. Annual Reviews in Microbiology, 33(1), 113-137.

Vartoukian, S. R., Palmer, R. M., & Wade, W. G. (2010). Strategies for culture of ‘unculturable’bacteria. FEMS microbiology letters, 309(1), 1-7.

134

Waxman, D. J., & Strominger, J. L. (1983). Penicillin-binding proteins and the mechanism of action of beta-lactam antibiotics1. Annual review of biochemistry, 52(1), 825-869.

Yang, W. T., & Min, D. B. (1994). Dynamic headspace analyses of volatile compounds of Cheddar and Swiss cheese during ripening. Journal of food science, 59(6), 1309-1312.

Zhou, J., Liu, X., Jiang, H., & Dong, M. (2009). Analysis of the microflora in Tibetan kefir grains using denaturing gradient gel electrophoresis. Food microbiology, 26(8), 770-775.

Chapter 2

Britz, T. J., & Riedel, K. H. (1994). Propionibacterium species diversity in Leerdammer cheese. International journal of food microbiology, 22(4), 257-267.

Clarridge, J. E. (2004). Impact of 16S rRNA gene sequence analysis for identification of bacteria on clinical microbiology and infectious diseases. Clinical microbiology reviews, 17(4), 840-862.

135

Darilmaz, D. O., & Beyatli, Y. (2012). Acid-bile, antibiotic resistance and inhibitory properties of propionibacteria isolated from Turkish traditional home-made cheeses. Anaerobe, 18(1), 122-127.

Estoppey, O., Rivier, G., Blanc, C. H., Widmer, F., Gallusser, A., & So, A. K. (1997). Propionibacterium avidum sacroiliitis and osteomyelitis. Revue du rhumatisme (English ed.), 64(1), 54-56.

Gautier, M., de Carvalho, A., & Rouault, A. (1996). DNA Fingerprinting of Dairy Propionibacteria Strainsby Pulsed-Field Gel Electrophoresis. Current Microbiology, 32(1), 17-24.

Hummel, A., Holzapfel, W. H., & Franz, C. M. (2007). Characterisation and transfer of antibiotic resistance genes from enterococci isolated from food. Systematic and applied microbiology, 30(1), 1-7.

136

Malik, A. C., Reinbold, G. W., & Vedamuthu, E. R. (1968). An evaluation of the taxonomy of Propionibacterium. Canadian Journal of Microbiology, 14(11), 1185-1191.

Park, H. S., Reinbold, G. W., & Hammond, E. G. (1967). Role of propionibacteria in split defect of Swiss cheese. Journal of dairy science, 50(6), 820-823.

Rossi, F., Sammartino, M., & Torriani, S. (1997). 16S-23S Ribosomal spacer polymorphism in dairy propionibacteria. Biotechnology techniques, 11(3), 159- 162.

137

Rossi, F., Dellaglio, F., & Torriani, S. (2006). Evaluation of recA gene as a phylogenetic marker in the classification of dairy propionibacteria. Systematic and applied microbiology, 29(6), 463-469.

Stackebrandt, E., Cummins, C. S., & Johnson, J. L. (2006). Family propionibacteriaceae: the genus Propionibacterium. In The prokaryotes (pp. 400-418). Springer New York.

Zouboulis, C. C. (2004). Acne and sebaceous gland function. Clinics in dermatology, 22(5), 360-366.

138

Chapter 3

Edgar, R. C., Haas, B. J., Clemente, J. C., Quince, C., & Knight, R. (2011). UCHIME improves sensitivity and speed of chimera detection. Bioinformatics, 27(16), 2194-2200.

139

Martin, N. H., Murphy, S. C., Ralyea, R. D., Wiedmann, M., & Boor, K. J. (2011). When cheese gets the blues: Pseudomonas fluorescens as the causative agent of cheese spoilage. Journal of dairy science, 94(6), 3176-3183.

Rychlik, M., & Bosset, J. O. (2001). Flavour and off-flavour compounds of Swiss Gruyere cheese. Evaluation of potent odorants. International Dairy Journal, 11(11), 895-901.

140

Yang, W. T., & Min, D. B. (1994). Dynamic headspace analyses of volatile compounds of Cheddar and Swiss cheese during ripening. Journal of food science, 59(6), 1309-1312.

Chapter 4

Bondad-Reantaso, M. G. (Ed.). (2007). Assessment of freshwater fish seed resources for sustainable aquaculture (No. 501). Food & Agriculture Org..

Chauhan, A., Wafula, D., Lewis, D. E., & Pathak, A. (2014). Metagenomic assessment of the Eastern oyster-associated microbiota. Genome announcements, 2(5), e01083- 14.

Chua, F. H. C. (1996). Aquaculture health management in Singapore: current status and future directions. FAO Fisheries Technical Paper (FAO).

141

Cooke, M. D. (1976). Antibiotic resistance among coliform and fecal coliform bacteria isolated from sewage, seawater, and marine shellfish. Antimicrobial agents and chemotherapy, 9(6), 879-884.

Done, H. Y., & Halden, R. U. (2015). Reconnaissance of 47 antibiotics and associated microbial risks in seafood sold in the United States. Journal of hazardous materials, 282, 10-17.

Fish and Fishery Products Hazards and Controls Guidance, Fourth Edition. (2011, April). Retrieved March 1, 2015, from Chapter 11, http://www.fda.gov/downloads/Food/GuidanceRegulation/UCM251970.pdf

Foodborne Outbreak Tracking and Reporting. (n.d.). Retrieved June 4, 2016, from http://wwwn.cdc.gov/foodborneoutbreaks/

Hatosy, S. M., & Martiny, A. C. (2015). The ocean as a global reservoir of antibiotic resistance genes. Applied and environmental microbiology, 81(21), 7593-7599.

Huang, Y., Zhang, L., Tiu, L., & Wang, H. H. (2015). Characterization of antibiotic resistance in commensal bacteria from an aquaculture ecosystem. Frontiers in microbiology, 6.

142

King, G. M., Judd, C., Kuske, C. R., & Smith, C. (2012). Analysis of stomach and gut microbiomes of the eastern oyster (Crassostrea virginica) from coastal Louisiana, USA. PLoS One, 7(12), e51475.

Morgan, R. C., Guerry, P., & Colwell, R. R. (1976). Antibiotic resistant bacteria in Chesapeake Bay. Chesapeake Science, 17(3), 216-219.

143

Zhang, T., Zhang, X. X., & Ye, L. (2011). Plasmid metagenome reveals high levels of antibiotic resistance genes and mobile genetic elements in activated sludge. PloS one, 6(10), e26041.

144