Use of omics methods for the advancement of food quality and food safety

Advancing food quality and food safety is critical for improving global food security and public health metrics. Omics tools (i.e., genomics, transcriptomics, proteomics, and metabolomics) are robust tools to gain further insight into microbial communities along the food chain and their implications for human and animal health. Whole genome sequencing facilitates early detection of foodborne illness outbreaks, including smaller clusters of related illnesses spanning over longer time periods, and microbial source tracking. Omics approaches can provide insight into adaptation of microorganisms to specific niches along the food continuum and other strain-specific characteristics that affect human and animal health, including virulence genes and antimicrobial resistance. Food quality and food safety are critical components to maintaining a safe and economically sustainable food supply, and are important to improving food security across the globe (WHO, 2015). Maintaining and improving the microbiological quality as well as ensuring the safety of food require identifying the presence of known and unknown microorganisms, including spoilage and pathogenic organisms that have been introduced throughout the food chain. Utilization of omics-based techniques (i.e., genomics, transcriptomics, proteomics, and metabolomics; see Figure 1) can broaden the scope of sampling programs and can increase detection of a broad range of issues, including pathogen detection, foodborne illness outbreak detection, microbial source tracking investigations, niche adaptation, antimicrobial resistance, and product shelf-life. Using omics techniques can improve food quality, food safety, and subsequently public health metrics by developing better screening and subtyping tools for both known and unknown pathogens. Better tools can then be generated for tracing specific bacterial strains to their origin and our understanding of these pathogens can be improved to prevent future contamination events, including identification and remediation of strains that persist in a given niche such as the processing plant environment (Miller et al., 2013; Zhou et al., 2016; Forbes et al., 2017). Omics techniques extend the range of information that is collected from a single system. For instance, genomics, the study of all DNA in an organism, increases the number of genes that can be compared across strains (Jackson et al., 2016), whereas metagenomics, the study of all microorganisms present in a system, can identify the presence of known and unknown microorganisms in the overall system (Forbes et al., 2017). The investigation of these strains or systems using omics-based approaches still requires well-characterized databases and a deep understanding of the individual organisms and genes. Improvements in food quality and food safety have been and will continue to be the direct result of integrating these approaches into routine sampling and testing schemes and microbial source tracking investigations (Jackson et al., 2016). Word cloud showing various state-of-the-art “omics” technologies. The simple biological central dogma (DNA replication, transcription to mRNA, and translation to protein) is the starting point for expanding these omics techniques and can provide a robust picture of the biological or environmental system under study. Figure 2 shows a description of these processes. The development of motifs, or patterns, that can be associated with specific protein functionality can then be used to search and identify a protein’s functionality based on sequence similarity across many different strains or species. This application of specific knowledge (i.e., studying a single gene in a single organism) to a broader context increases the pace at which novel and known mechanisms of virulence, antibiotic resistance, or enzymatic ability can be identified (Chen et al., 2012; Zhao et al., 2014; Jia et al., 2017). Initial studies that sequenced individual genes and probed their function in the cell through extensive combined genetic and phenotypic characterization provide this vital information for understanding how genes are regulated, and how gene sequences correlate to protein structure and function. Understanding the structural differences between transmembrane, extracellular, or intracellular proteins can provide insight into their function without necessarily performing additional laboratory experiments (Krogh et al., 2001). DNA is transcribed to mRNA and mRNA is translated to protein. New technologies (e.g., “omics” technologies) have been developed to study changes in DNA, mRNA, and protein under varying physiological and environmental conditions. There are an estimated 3.6 million cases of bacterial foodborne illness each year in the United States, and an estimated 10.3 million cases globally (Scallan et al., 2011; Havelaar et al., 2015). The incidence and cost of foodborne illness is shown in Figure 3. Foodborne illness and food product recalls bear a significant emotional and economic burden on society globally. Foodborne illnesses affect the population at large by increasing healthcare costs and decreasing the quality and length of life for sick individuals. Estimating the number of quality-adjusted life years, or healthy years lost in terms of quality or death, can provide a measurement for determining the full impact of outbreaks and the resulting illnesses. The monetary value (see Figure 4) and the loss in quality-adjusted life years have been estimated for each of the 14 common foodborne pathogens with the combination of Listeria monocytogenes, Salmonella enterica, and shiga toxin producing Escherichia coli (STEC) amounting to more than US$6 billion annually and more than 27,000 quality-adjusted life years lost to foodborne illnesses in the United States alone (Batz et al., 2012; Hoffmann et al., 2012). In 2010, Ivanek et al. estimated the marginal benefit per prevented case of listeriosis for food processing companies to be between US$4.4 and US$8.8 million, whereas the marginal cost per prevented case of listeriosis for food processing companies to be between US$0.5 and US$1 million (Ivanek, et al., 2005). This imbalance of the cost benefit ratio could be leveraged to increase support from governmental agencies and the food industry for the reduction of listeriosis, and potentially all other pathogens overall. Incidence and cost of foodborne illness. Cost of various foodborne pathogens. Source: USDA Economic Research Service. The identification of pathogens in the food processing environment can be difficult due to the complexity of the processing network, heterogeneity of food products, and the presence of normal or natural microflora. Bacteria can be routinely introduced into a food processing facility through a variety of mechanisms, including raw materials, water, equipment/utensils, and workers. Once inside a food processing facility, bacteria can be transferred to various surfaces and postlethality-exposed product by cross-contamination of raw products or contaminated equipment (as reviewed by Pérez-Rodríguez et al., 2008). Food safety controls (i.e., cleaning and sanitation programs, process interventions and management systems, and robust environmental monitoring) are important techniques for food processing facilities to reduce spoilage organisms and control pathogens. Although the application of efficacious food safety controls appears to control low-level contamination, when these food safety systems fail a sporadic high-level contamination event may result in an outbreak of foodborne illness. Foodborne pathogens, such as STEC, S. enterica, and L. monocytogenes, are commonly identified along the food chain continuum and are capable of causing disease in both humans and animals. These pathogens must be able to survive a significant number of different stressors to be transmitted throughout the food continuum, including large temperature fluctuations, high salt concentrations, low water activity, limited free nutrients, oxidative stresses, and disinfectants. Many of these stresses are intentionally incorporated into food processing as interventions as well as in cleaning and sanitation practices to decrease bacterial loads, and prevent additional growth of spoilage and pathogenic microorganisms. Animals that symptomatically or asymptomatically carry and shed pathogens can introduce pathogens into environments associated with food production, processing, and handling, which may result in loading the human food chain with pathogens ultimately increasing the incidence of human foodborne and animal disease (McDaniel et al., 2014). Current gold standard methods of bacterial subtyping include phenotypic methods such as serotyping and band-based molecular methods such as pulse field gel electrophoresis (PFGE). Although these current gold standard methods are rapidly being augmented with omics approaches, information obtained from gold standard methods is still clinically and epidemiologically relevant. Serotyping and pulse field gel electrophoresis differentiate strains of a given bacterial species beyond the species level–based antibody/antigen interactions and endonuclease restriction of total bacterial DNA followed by fragment separation, respectively. Based on serotyping data, STEC has been shown to be comprised of more than 180 different serotypes, S. enterica includes more than 2,500 serotypes, and L. monocytogenes isolates can be differentiated into 13 serotypes (Seeliger and Höhne, 1979; Uzzau et al., 2000; Stenutz et al., 2006). However, approximately 91.7% of STEC outbreaks in the United States during the last 18 yr have been attributed to only seven serogroups, including the O157:H7 serotype and the six regulated non-O157 serogroups (i.e., O103, O111, O121, O145, O26, and O45) or the “big 6” non-O157 STEC as reported by the CDC (https://wwwn.cdc.gov/foodborneoutbreaks/). In addition, only a small fraction of all S. enterica subtypes are regularly isolated from human clinical cases and animals, and all of these serotypes belong to S. enterica subspp. enterica (as reviewed in Uzzau et al. (2000)). Only three of the 13 L. monocytogenes serotypes (i.e., 1/2a, 1/2b, and 4b) appear to be responsible for majority (approximately 90%) of human listeriosis cases (McLauchlin, 1990; Maury et al., 2016), and reviewed in Orsi et al. (2011). Serotyping can provide subspecies discrimination and relevant historical clinical reference information; however, a significantly deeper level of strain discrimination is necessary to facilitate outbreak detection, outbreak investigations, and other microbial source tracking studies. Pulse field gel electrophoresis demonstrates increased discriminatory power over serotyping and is still the current gold standard for differentiating bacterial pathogens by PulseNet at the CDC. However, this technique can be laboratory-dependent and requires a significant amount labor (https://www.cdc.gov/pulsenet/pathogens/pfge.html). In addition, some types of pulse field gel electrophoresis are very common in general and a shared subtype between a human clinical and food isolate does not necessarily imply that the food was responsible for disease without supporting epidemiological information. This technique does not have sufficient discriminatory capability to differentiate isolates belonging to certain Salmonella serotypes such as Salmonella Enteritidis and Salmonella Newport. Another major category of molecular subtyping methods involves sequencing specific gene regions of the bacteria to be subtyped, including individual genes, multiple genes, or the whole genome. For example, multilocus sequence typing (MLST) can distinguish strains based on the allelic type of a set of selected genes that are present in the core genome of a species (e.g., housekeeping genes), for instance, housekeeping genes (Feil and Enright, 2004). To accomplish this, the gene or a fraction of the gene is sequenced and the resulting nucleotide data are then compared with a database of previously identified allelic types. If the query sequence matches one of the reference sequences uniquely, then the query sequence is assigned to that allelic type. Clonal complexes have been defined as group organisms sharing 100% similarity of allelic types in six of seven housekeeping genes tested by a MLST protocol (Feil et al., 2004). MLST protocols have been developed that target different regions for the same organism. For instance, MLST schemes have been developed for L. monocytogenes using housekeeping genes only, genes from the prfA virulence gene complex only, and a mixture of housekeeping, virulence, and stress-response genes (Salcedo et al., 2003; Ward et al., 2004; Nightingale et al., 2005). The higher resolution subtyping that molecular techniques such as MLST have been shown to provide has been used for source-associated clustering of L. monocytogenes isolates (Nightingale et al., 2006). This study showed that isolates from the same source (i.e., human, animal, or food) are more closely related to each other than isolates from different sources, supporting niche adaptation of L. monocytogenes isolates to a human or animal host or nonhost environment. Once the DNA sequence–based subtyping information has been collected, the sequence data can then be used for additional bioinformatics to understand the molecular evolution and ecology of a set of isolates (den Bakker et al., 2008; Ragon et al., 2008). Recent advancements in sequencing technology (Next Generation Sequencing) have made available several omics-based techniques that can be used to improve food quality and food safety by providing rapid detection and high-resolution subtyping. The genome of an organism can be used to provide strain-level resolution of the pathogen-causing disease and to track the pathogen to its source to limit additional exposure in humans and animals (den Bakker et al., 2014; Deng et al., 2016). Bacterial subtyping by whole genome sequencing (WGS) has been performed using several different platforms that vary in the length of fragments that can be sequenced and the correctness or quality of the nucleotide attribution for each read. Shorter reads can significantly limit the ability of assembling a complete genome de novo, without a reference sequence, due to the presence of repetitive regions that can occur in the bacterial genome. Longer reads can be used to sequence across these repetitive regions with adequate coverage for de novo assembly. However, current technology limits the quality of the reads generated by long read technologies relative to short read technologies. High-quality draft (i.e., incomplete or unclosed) genomes generated using current platforms have been used for regular subtyping of several organisms and have been shown to identify the source of sporadic cases or smaller clusters of related cases over longer periods of time (Jackson et al., 2016). High-quality draft genomes can be used in multiple subtyping schemes based on previous MLST structures, novel MLST subtyping schemes, specifically on the number of single-nucleotide polymorphism (SNP) differences (Zhang et al., 2004; Cody et al., 2017; Saltykova et al., 2018). Core genome MLST focuses only on the genes that are present in all organisms identified in the species, whereas the whole genome MLST subtyping scheme uses the pan-genome, or both core and accessory genes. L. monocytogenes WGS data have shown that using either the core genome or the whole genome MLST structures results in similar phylogenetic trees (Chen et al., 2016; Jackson et al., 2016; Moura et al., 2016). As WGS becomes more routinely performed in laboratories, including the rate at which isolates can be sequenced, assembled, and compared through bioinformatics have to from systems to and outbreak detection et al., 2017). For example, the CDC is performing WGS on all L. monocytogenes human clinical isolates and the is routinely using WGS in sampling investigations by recalls or foodborne illness In the of draft genomes generated by WGS has been shown to be for outbreak detection and a similar level of discrimination by in both S. enterica and L. monocytogenes et al., Bakker et al., 2014; et al., 2017). However, are still some with on WGS data to outbreaks without the of food information. For instance, L. monocytogenes with WGS genome sequences have been in multiple environments et al., 2015). The from which these are made can be on the of reference sequence and et al., 2014). WGS can be used to between and strains of L. monocytogenes, providing facilities that process and food with information on the between cleaning and sanitation and from raw and other et al., 2015). The high discriminatory power of WGS and will be for identifying outbreaks and cases that may be the result of or multiple pathogens. than one strain of L. monocytogenes has been from both a case of listeriosis and an outbreak associated with et al., et al., In to the of WGS can provide information on the virulence and antibiotic by providing information on all of the genes in the genome (Chen et al., 2012; Jia et al., 2017). This includes genetic differences in gene genetic or gene et Bakker et al., et al., 2012). Once the genome has been assembled, many of the used methods of subtyping can be using databases and programs, such as to identify allelic types in et al., 2004; et al., 2014; and 2005). For instance, genetic for serotype identification can be used in of antibody/antigen interactions and provide increased discrimination between serotypes (Zhang and 2005). In addition, genetic have been used to group coli into to identify strains with specific virulence set of genetic for coli virulence is the presence or of a combination of the and genes. However, these genes alone are not sufficient for as bacterial virulence systems are a of and to with some a host et al., 2011; et al., the genome of the organism the ability to the mechanisms responsible for bacterial in the environment and of bacterial for different the process of can identify important in the and can be used to distinguish between and strains a given The of WGS and the other omics-based techniques can provide a in the to or due to the large amount of information that can be generated by each of the omics For instance, routine of food animals in the identification of coli to the antibiotic et al., could result in the of to coli whole sequencing was performed to identify genes that could be responsible for the antimicrobial et al., The sequence generated was then compared with closely related that not have to the antibiotic et al., this multiple genes selected and in an to the gene resistance. The or all available mRNA, can be by the strain in with and without the The identification of genes under the antibiotic provide additional support for several of the genes in resistance. In a similar gene of known can be used with whole to the of these by genes et al., 2008; et al., et al., However, have shown that the between level and protein is than in (Zhang et al., that in can be for using specific and 2012). The full provide additional of a gene or a different that may be responsible for by increased transcription that may be present for some genes. has been used to the specific known and unknown enzymatic of various et al., 2016). These omics techniques with a of and that when used that be however, the a to from a biological as to a multiple of omics-based techniques into a single data set that can a single environmental could provide an understanding of how a or pathogen in the environment. tools are available for at multiple of omics-based data at one time et al., 2014). will be for to at more complex between bacterial pathogens, the and their Although to on the that has given to these techniques is all of these omics techniques and the information obtained can provide a and rapid for understanding more of the biological in antibiotic and expanding the level of and information for bacterial pathogens and the host environment could provide for and decreasing the and rate of all bacterial pathogens. of the genes, and proteins in a a large and complex data set to of the different omics data types provide a different and can made at sequencing (i.e., or of microbial such as or the of the can provide information on the and could provide into the and of the has been used to the of antibiotic on the during the of to differences in the and changes in the et al., 2017). In addition, strain-level identification of bacterial pathogens can be generated from sequencing et al., 2016). has been used to identify significant microbial differences between and healthy from et al., 2017). These differences can be used to and improve the of these as a food source the (i.e., sequencing of the of the microbial in a given has been used to the differences in the from regularly with and without The increased of associated with and and 2018). of may as a for bacteria that can survive the increased including L. monocytogenes, STEC, and S. enterica and et al., 2012; et al., 2014). the of a bacterial isolate or microbial can provide information how to an or by sequencing all available bacterial DNA in a system, the antibiotic of the whole can be for providing information on the presence of and common bacteria that persist in a given Improvements in food quality and food safety are to the of safe and food and can subsequently increase food security across all Omics techniques can improve current food quality, food safety, and public health metrics by increasing the number and type of organisms and in a using and WGS can provide very high-resolution subtyping to differentiate strains to facilitate early detection of foodborne illness outbreaks and increase the of identifying the source of an outbreak or Figure an of and the of to track and identify microorganisms throughout the food chain. In addition, these can be used in with the standard protocols and being in routine testing in food processing and protocols can increase the of these protocols across many decreasing the to for pathogens and other The additional information through these mechanisms can then be transferred to for the of improving the and of these techniques in the and for pathogens throughout the food chain. sequencing for of and Figure on is an of in the for and a in both and Food from a in at and performed at the for Food at the of has the of and Listeria monocytogenes using whole genome sequencing and transcriptomics, and focuses on Nightingale is a in the of and Food at Nightingale in and in Food from Nightingale and in the of Food at and to understand the molecular ecology and of foodborne pathogens along the food Nightingale as a for Food and on Food Nightingale is a of which biological to control foodborne pathogens.