Edited by: Walid Alali, Hamad bin Khalifa University, Qatar
Reviewed by: Biswapriya Biswavas Misra, Texas Biomedical Research Institute, United States; Vasiliki Chini, Qatar Foundation, Qatar
*Correspondence: Shabarinath Srikumar
This article was submitted to Food Microbiology, a section of the journal Frontiers in Microbiology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The development of next generation sequencing (NGS) techniques has enabled researchers to study and understand the world of microorganisms from broader and deeper perspectives. The contemporary advances in DNA sequencing technologies have not only enabled finer characterization of bacterial genomes but also provided deeper taxonomic identification of complex microbiomes which in its genomic essence is the combined genetic material of the microorganisms inhabiting an environment, whether the environment be a particular body econiche (e.g., human intestinal contents) or a food manufacturing facility econiche (e.g., floor drain). To date, 16S rDNA sequencing, metagenomics and metatranscriptomics are the three basic sequencing strategies used in the taxonomic identification and characterization of food-related microbiomes. These sequencing strategies have used different NGS platforms for DNA and RNA sequence identification. Traditionally, 16S rDNA sequencing has played a key role in understanding the taxonomic composition of a food-related microbiome. Recently, metagenomic approaches have resulted in improved understanding of a microbiome by providing a species-level/strain-level characterization. Further, metatranscriptomic approaches have contributed to the functional characterization of the complex interactions between different microbial communities within a single microbiome. Many studies have highlighted the use of NGS techniques in investigating the microbiome of fermented foods. However, the utilization of NGS techniques in studying the microbiome of non-fermented foods are limited. This review provides a brief overview of the advances in DNA sequencing chemistries as the technology progressed from first, next and third generations and highlights how NGS provided a deeper understanding of food-related microbiomes with special focus on non-fermented foods.
It is well known that foodborne diseases cause considerable morbidity and mortality in humans particularly in immunocompromised individuals and in young children (Stein et al.,
Food, an indispensable part of everyday life, undergoes many processing steps before reaching the consumer. The total population of all microorganisms (microbiome), play important roles in any food matrix ranging from fermentation, contamination and spoilage. Deep taxonomic understanding of the microorganisms and their communities is required either to enhance desired food processes like fermentation or to mitigate detrimental occurrences like contamination and spoilage. Historically, conventional techniques including the classical Gram stain along with individual biochemical characteristics are used for the isolation, identification and characterization of bacteria from clinical, food or environmental origins. Even though considered as the “gold standard,” culture dependent techniques can only detect 0.1% of a complex community, such as that found in the human intestinal microbiota. Therefore, to extend the understanding of an ecological niche, such as food, techniques are needed to identify or characterize microorganisms and predict the functional interactions of different microbiological communities present in the sample. To this end, contemporary advances in multi-omic technologies have enabled microbial community profiling, monitoring population fluctuations in different microbial ecosystems and characterization of different microbial species in food matrices.
The rapid development of nucleic acid sequencing technologies over the past four decades has improved the capacity to characterize the microbiomes of complex matrices associated with food or environmental samples. The ubiquitous nature and specificity of nucleic acids make the molecule an ideal target for bacterial or microbiome characterization. Utilizing significant advancements in sequencing chemistries, DNA sequencing gradually evolved from low throughput DNA fragment sequencing to high throughput next generation (NGS) and third generation sequencing techniques (Loman and Pallen,
Traditionally, most NGS related food microbiome studies have focussed on fermented foods, such as cheese, kimchi and sausages (Patra et al.,
Scientific advances in whole genome sequencing proceeded through three major technological revolutions: first
The first DNA sequencing strategy was the
The advantages of NGS over Sanger sequencing can be summarized as follows (1)
A recognized limitation of NGS technologies is the requirement for a PCR amplification step, which introduces a bias in read distribution ultimately affecting the coverage. Third generation sequencing technology was designed to address this limitation. Here single DNA molecules are directly sequenced thereby reducing low error rates by avoiding amplification associated bias, intensity averaging, phasing or synchronization problems.
The first commercially released long read methodology was single-molecule-real-time (SMRT®) technology (Pacific Biosciences) (Eid et al.,
HeliScope (Braslavsky et al.,
MinION (Oxford Nanopore Technology) was released in 2014 through the MinION Access Programme (MAP). Here, electrophoresis is used to move the DNA/RNA molecule through a nanopore. This system involves the use of electrolytic solutions and the application of a constant electric field. As the nucleic acid passes through the nanopore, the change in the current pattern and magnitude is measured. During the library preparation step, double stranded DNA is sheared using a Covaris g-TUBE and fragmented DNA is repaired using a PreCR step. Blunt ended DNA molecules are then created using an end repair step before a poly A tail is added to the 3′-OH end. Two adaptors are then added to the DNA, a Y adapter (so called, due to its shape) and a hair pin adaptor. A motor protein unzips the double stranded DNA at the Y adapter and feeds the DNA as a single strand through the nanopore. Base calling is then performed and a read length of a few hundred thousand base pairs is achieved with an accuracy ranging from 65 to 88%. If information from only one strand is used, base calling is 1-dimensional (1D), otherwise it is 2-dimensional (2D) system (Lu et al.,
This is one of the most important culture independent methods used for conventional microbiome analysis. Most bacteria contain 16S rDNA gene which is made up of nine hypervariable regions flanked by conserved sequences (Neefs et al.,
Traditional Sanger sequencing allows only a smaller proportion of amplicons to be sequenced. This results in less abundant members of the microbiome population being missed, thus compromising the comprehensive description of the microbial community. The subsequent inclusion of NGS platforms in 16S rDNA sequencing increased the capacity for a more thorough identification of the bacterial members of the community by several orders of magnitude and at a much lower cost. Since only a short amplicon was sequenced, much higher coverage per sample was obtained (Claesson et al.,
Due to shorter reads obtained from NGS protocols, especially from Illumina platforms, bacterial classification using 16S rDNA sequencing often cannot be identified beyond the genus level (Claesson et al.,
Metagenomics refers to the application of high throughput techniques to sequence the entire DNA (or RNA) content found in a sample, independent of its origin. Template DNA contained in a sample of interest is subject to sequencing directly, without any prior marker gene amplification step. Metagenomic data not only provides an in-depth taxonomic identification of the microbiome but can also simultaneously compare the relative abundance of all organisms present in the microbiome. Substantial amounts of sequencing data generated using a metagenomic approach is then queried against databases, such as k-mer (Compeau et al.,
The main challenge of the metagenomic approach is the amount of sequence data generated. This procedure is expensive compared to 16S rDNA-based strategies. Moreover, data analysis requires high-end bioinformatics requiring a long term financial investment, which is possible only in specialized laboratories. The lack of specially designed reference databases also makes the use of this technology challenging when attempting to extract biological information on a routine basis.
Another concern is that metagenomics approaches cannot distinguish viable microbial populations within a microbiome (Ercolini,
Alternatively, RNA can also be used as a template to distinguish the viable population within the microbiome. Sequencing total RNA purified from a sample is the basic principle underpinning metatranscriptomic analysis. Apart from distinguishing the viable members within the population contained in a microbiome, this technique is invaluable for providing a functional characterization of the different bacterial members of the microbiome. In a complex microbiological sample, such as food, different microbiological communities interact with each other, either to degrade, spoil or ferment the organic constituents of the matrix. Sequencing RNA purified from these samples would provide a basic description of how the communities interact with each other.
Determining the optimal sequencing approach useful for the study of different food matrices depends upon the complexity of the sample to be analyzed and the depth of bacterial taxonomic information required. An initial 16S rDNA sequencing based profile would provide a broad overview of the microbial composition within a food sample. Nonetheless, this technique lacks the necessary resolution required to provide species-level/strain-level identification. Further, it will not provide an assessment of the functional capability of these organisms, contained within the sample. Therefore, for in-depth species-level or for strain level identification or detailed functional characterization of the different members in the microbiome, metagenomics and metatranscriptomics would be useful.
For the purpose of this review, a literature search was performed on the current NCBI PubMed database (
Between 2011 and June 2017, a total of 126 papers were published describing the characterization of various food microbiomes (Figure
Hygiene is an essential step in the
A comprehensive list of publications using next generations sequencing approaches to study the environmental microbiome along the food production chain.
Artisan cheese factory and cheese samples | United States | 16S rDNA sequencing (V4); qPCR | Illumina MiSeq | Facility-specific “house” microbiota play an important role in shaping site-specific characteristics in products | Bokulich and Mills, |
Wine factory equipment surface | United States | 16S rDNA sequencing (V4) | Illumina MiSeq | Winery surface microbiomes have no obvious link with spoilage microbes in wine under normal operating conditions | Bokulich et al., |
Carcass, processing environment and beefsteaks | Italy | 16S rDNA sequencing (V1-V3) | Roche 454 GS Junior | 4°C aerobic storage led to dramatic decrease in beef microbial complexity; spoilage-associated bacteria originated from carcasses and carried through the production chain to the products | De Filippis et al., |
Brewery plant environment and beer product | United States | 16S rDNA sequencing (V4, for bacteria); Fungal internal transcribed spacer (1 loci, for fungi); T-RFLP; Droplet digital PCR) | Illumina MiSeq | Most microbes found in the brewery environment originated from raw ingredients; beer-spoilage and hop-resistance genes were found throughout the brewery, but little beer spoilage occurred | Bokulich et al., |
Sausage processing environment and product | Finland | 16S rDNA sequencing (V1-V3) | Roche 454 Titanium FLX | Abundant mesophilic psychrotrophs were prevalent throughout sausage production chain microbiomes, and with different characteristic patterns of contamination for different genera | Hultman et al., |
Ready-to-eat meal plant environment and product | Not mentioned | 16S rDNA sequencing (V1-V3) | Roche 454 GS Junior | Pothakos et al., |
|
Cheese factory environment and cheese product | Italy | 16S rDNA sequencing (V1-V3, for bacteria); 26S rDNA sequencing (D1-D2, for fungi) | Roche 454 GS Junior | Coexistence of lactic acid bacteria and possible spoilage-associated bacteria was found in core microbiota of cheese factory environment and cheese samples | Stellato et al., |
Powdered Infant Formula plant environment | Ireland | 16S rDNA sequencing (V3-V4); Flow cytometry | Illumina MiSeq | Bacteria present in low, medium and high care area of a powdered infant formula plant environment were mostly associated with soil, water, and humans, respectively | Anvarian et al., |
Environment samples alone beef production chain | United States | Shotgun metagenomics sequencing | Illumina HiSeq 2000 | No antimicrobial resistant determinants (ARD) were identified in final beef products, indicating slaughter interventions may reduce ARD transmission risk | Noyes et al., |
Dairy farm agroecosystems | United States | Shotgun metagenomics sequencing | Ion Torrent Personal Genome Machine | The most abundant antimicrobial resistant genes in dairy agroecosystems were grouped under multidrug transporters | Pitta et al., |
Butchery meat and environment samples | Italy | 16S rDNA sequencing (V1-V3) | Roche 454 GS Junior platform | The type of retail (large- or small-scale distribution) had no apparent effect on initial fresh meat contamination | Stellato et al., |
Environment samples along beef production chain | United States | Shotgun metagenomics sequencing | Illumina HiSeq 2000 | Usage of standard antimicrobial interventions in beef processing system significantly reduced the diversity of remaining microbiomes | Yang et al., |
Vegetables are known to be vehicles of pathogenic microorganisms and in several cases and have led to outbreaks of foodborne illness (Buchholz et al.,
Soil is another factor that influences vegetable and fruit microbiomes. A study using 16S rDNA sequencing reported that the microbiomes of leaves, flowers and fruits shared a greater proportion of taxa with the soil microbiome in which the plants were grown (Zarraonaindia et al.,
A comprehensive list of publications using next generations sequencing approaches used in characterizing the microbiome of raw food products.
Broiler filet strips | Finland | 16S rDNA sequencing (V1-V3); T-RFLP | Roche 454 GS FLX | Marination process led to increased lactic acid bacteria in broiler meat microbiome, resulting in enhanced CO2 production and acidification | Nieminen et al., |
Broiler filet strips | Finland | 16S rDNA sequencing (V1-V3); Shotgun metagenomics sequencing | Roche 454 GS FLX; Roche 454 GS FLX; | Marination altered broiler fillet strips' microbial community by favoring the spoilage associated bacteria |
Nieminen et al., |
Spoiled retail foodstuffs | Belgium | 16S rDNA sequencing (V1-V3) | Roche 454 GS Junior | Characterization of psychrotrophic lactic acid bacteria that cause unexpected food spoilage cases in Belgian retail food | Pothakos et al., |
Store bought meat | United States | Shotgun metagenomics sequencing | Illumina Miseq | Primary characterization of viruses commonly found in US store-bought meats | Zhang et al., |
Beef burger | Italy | 16S rRNA sequencing (V1-V3); PCR-DGGE | Roche 454 GS Junior | Nisin-based antimicrobial packaging reduced the abundance of microbes that produce compounds of specific metabolic pathways related to spoilage | Ferrocino et al., |
Raw pork sausage | France | 16S rDNA sequencing (V1-V3); qPCR | Roche 454 GS FLX++ Titanium | Salt reduction, particularly when combined with CO2-enriched packaging, resulted in faster spoilage of raw sausages by lowering the overall bacterial diversity | Fougy et al., |
Raw milk | Finland | 16S rDNA sequencing (V1-V2); | Illumina MiSeq | Bacterial diversity is better preserved in bovine raw milk by additional flushing with N2 gas compared to cold storage at 6°C alone | Gschwendtner et al., |
Raw milk | United States | 16S rDNA sequencing (V4); qPCR | Illumina MiSeq | raw milk microbial community structure can be influenced during low-temperature, short-term storage | Kable et al., |
porcine musculature | Austria | 16S rDNA sequencing (V1-V2); qPCR | Roche 454 GS-FLX Titanium | Pork sample microbiota was dominated by psychrophilic spoilers; |
Mann et al., |
Raw milk | Australia | 16S rDNA sequencing (V5-V8) | Roche 454 | Spoilage bacteria growth was delayed by at least 7 days in CO2 treated raw milk sample | Lo et al., |
Bulk tank milk | United States | 16S rDNA sequencing (V4); qPCR; Flow cytometry | Illumina MiSeq | Spoilage and spore-forming bacteria were ubiquitous in all dairy farms | Rodrigues et al., |
Common carp filets | China | 16S rDNA sequencing (V3-V4) | Illumina HiSeq 2500 | Use of cinnamon essential oil extended vacuum-packaged common carp fillets shelf-life by approximately 2 days based on sensory and other analysis, but showed no significant differences in dominant microbiota composition compared with non-treated samples at the end of shelf-life | Zhang et al., |
A comprehensive list of publications using next generations sequencing approaches in characterizing the microbiome of ready to eat food.
Bagged leaf vegetables | United States | 16s rDNA sequencing; | Roche 454 GS-FLX Titanium | No significant differences found on microbial compositions between organic and conventionally grown, surface-sterilized and non-sterilized leaf vegetables | Jackson et al., |
Store-bought fruits and vegetables | United States | 16S rDNA sequencing | Roche 454 | Microbial communities of certain type product are more similar than different types, but Significant difference identified between conventional and organic product within the same type | Leff and Fierer, |
Field grown lettuce | United States | 16S rDNA sequencing (V5-V9); qPCR | Roche 454 GS-FLX Titanium | Lettuce phyllosphere microbiome are affected by seasonal, irrigation, and biological factors | Williams et al., |
Carrots | United Kingdom | Metatranscriptomics qPCR | Illumina MiSeq | Carrot yellow leaf virus are strongly associated with carrot internal necrosis | Adams et al., |
Basil leaves | Belgium | 16S rRNA sequencing (V1-V3); PCR-DGGE | Roche 454 GS-FLX Titanium | Spoilage of commercially grown basil leaves was caused by tissue injuries and visual defects rather than by specific bacterial growth | Ceuppens et al., |
Cilantro | United States | 16S rRNA sequencing (V1-V3); Shotgun metagenomics sequencing (with pre-enrichment) | Illumina MiSeq; Illumina MiSeq | A 24 h non-selective enrichment identified |
Jarvis et al., |
Bagged spinach | United States | Shotgun metagenomics sequencing (with pre-enrichment) | Illumina MiSeq | Eight h pre-enrichment and sequencing depth identified' spiked Shiga toxin-producing |
Leonard et al., |
field-grown and retail lettuce | United States | Shotgun metagenomics sequencing; Metatranscriptomics | Illumina HiSeq 2500; Illumina HiSeq 2500 | Virome of iceberg lettuce from fields and produce distribution center were dominated by plant pathogenic viruses but human and animal viruses were also identified | Aw et al., |
Oregano | United States | Shotgun metagenomics sequencing (with pre-enrichment) | Illumina MiSeq | Addition of corn oil during pre-enrichment of oregano samples led to increased overall abundance of Gram negative microorganism and a ≥50% recovery rate of |
Beaubrun et al., |
Bagged spinach | United States | Shotgun metagenomics sequencing (with pre-enrichment) | Illumina MiSeq | Shotgun metagenomics sequencing identified Shiga toxin-producing |
Leonard et al., |
Cheese | Ireland | 16S rDNA sequencing (V4-V5); Shotgun metagenomics sequencing; qPCR | Roche 454 GS-FLX; Illumina HiSeq 2000 | Carotenoid-producing bacteria, genus |
Quigley et al., |
All the above studies provide insights into the role that NGS-based strategies played in uncovering the dynamic changes in the microbiome. Careful analysis of NGS data can be used to facilitate the development of safer production processes thereby reducing risk for the consumer.
Control measures like refrigeration, modified atmospheric packaging (MAP), nisin treatment and others are often used to extend the shelf life of many perishable food products. The microbiome undergoes considerable compositional change when food products are stored under defined conditions. It is possible that these fluctuations in the microbiome may finally affect the quality of the food product. NGS techniques have been increasingly applied to study how these variations contribute to improved shelf life. The most common response noted in refrigerated food products was a reduction in bacterial diversity associated with the microbiome. A 16S rDNA sequencing approach noted this reduction in bacterial diversity in refrigerated spinach (Lopez-Velasco et al.,
Marination is another traditional treatment method frequently used during food production process. It is a process of soaking foods in a seasoned, often acidic, liquid before cooking. The derivation of the word refers to the use of brine or a water solution containing a significant amount of salt. The most common examples are used for curing, preserving, and developing flavor in foods, such as that used in the pickling process. NGS-based approaches have been used to investigate the microbiome diversity and structure alterations during marination, and to determine whether, such process can extend product shelf life. Sometimes contrary to its intended use, marination increased the speed of spoilage. In a typical example, the poultry product was rapidly spoiled when marinated with acetic acid and subsequently packaged in a modified atmospheric environment. Investigation on the bacterial composition contributing to spoilage using 16S rDNA sequencing indicated a heterofermentative lactic acid bacteria
Extended shelf life can also be achieved through the addition of the bacteriocin nisin, a polycyclic antibacterial peptide secreted by
Addition of NaCl to meat is known to improve texture, flavor and taste whilst also improving shelf life by reducing water activity. In one study using high salt concentration along with low temperature and CO2 enriched packing to improve the shelf life of sausage meat, the reduction of salt concentration was not surprisingly associated with faster spoilage in products. The 16S rDNA sequencing-based investigation revealed that reduction in salt concentration led to an overall reduction in bacterial diversity which caused faster spoilage. Improvements in sausage meat processing with higher salt concentrations combined with vacuum packaging increased the abundance of a subpopulation consisting of
The selected studies cited above demonstrate the potential of NGS approaches and facilitated a better understanding of bacterial food spoilage. These sequencing based investigations not only identified the spoilage agent in many cases but also showed how different bacterial communities interacted with each other to counteract spoilage.
Metagenomics/metatranscriptomics is also a valuable tool to understand how bacterial communities interact with each other in fermented foods. The first metagenomic study used a 454 GS FLX titanium platform to describe the microbiome of a ferment food, kimchi, with the predominant microbial population identified as
Metatranscriptomics also played a pivotal role in understanding the process of fermentation, especially the ripening of cheese. Camembert-type cheese ripening is driven by fungal microflora including
All these insights above showed that metagenomic/metatranscriptomic analysis provided an in-depth analysis of how microbial communities interacted with each other during fermentation. However, reports on the usage of these techniques in non-fermentative foods are rare, and metagenomics/metatranscriptomics approaches have not been utilized to their full potential in investigating food microbiomes of non-fermentative foods. Considering the limitations of 16S rDNA sequencing approaches, metagenomics/metatranscriptomics could be increasingly used in future to extend understanding of the functional microbiome of non-fermented food.
Many bacteria are non-culturable, either because they are unknown or they are known but not recoverable in the laboratory conditions. Genomic approaches have played a major role in understanding such non-culturable bacteria, and in some cases, have led to development of new media that can be subsequently used for their cultivation. A classic example is the case of
NGS approaches can be utilized in the identification and characterization of specific pathogenic or spoilage bacteria from food or food production microbiomes. Some examples include the identification of
Since the sequence data obtained from a metagenomics approach are detailed, these data could be used to characterize organisms other than bacteria in the microbiome. Metagenomics approaches were used to study the virome associated with store-derived beef, pork and chicken identified a novel bovine polyomavirus in beef and a novel gyrovirus species in chicken (Zhang et al.,
Dissemination of antibiotic resistance in human pathogens is a matter of global concern. Antibiotics are used in food-producing animals and aquaculture as therapeutic agents, for prophylactics and in some jurisdictions as growth promoters (Phillips,
NGS methods are highly efficient for microbiome related studies but there are still challenges and limitations to consider when applying these techniques to specific cases on food and food-related environments.
Natural environment samples, such as soil and water, along with stool and saliva, and fermented food samples, such as cheese and kimchi, contain high numbers of microorganisms. Consequently, these samples can offer sufficient template nucleic acid for subsequent analysis. In contrast, sanitary control and maintenance of strict food production environment standards have made isolation of total DNA from these environments challenging (Anvarian et al.,
Different enrichment methods can be carried out to increase nucleic acid concentrations in samples. Prior to nucleic acid purification, selective or non-selective cultivation based pre-enrichment or enrichment can be used to generate microbiomes with higher target bacterial counts (Duan et al.,
Internal controls are essential in multi-omic strategies. Often these are never considered. A sequencing control should be designed to contain DNA sequences from known bacterial species, processed in parallel with other samples during sequencing, which could finally provide an estimation of the sequencing errors during downstream bioinformatics analysis. An ideal positive control could contain DNA isolated from mixture of multiple reference strains constituting a
As with the positive, a negative control is also relevant in sequencing runs. Being high throughput, conventional amplicon based or whole genome sequencing experiments can analyze DNA at nanomolar concentrations. Trace levels of contamination present in the reagents used during DNA isolation or library preparation can introduce an undesirable bias in the analysis. This may interfere with final sequencing results giving rise to background contamination noise, especially when the target contains very few microorganisms (Salter et al.,
The utility of culture dependent approaches in the study of bacterial diversity is inherently limited in sensitivity. Nucleic acid sequencing platforms are being increasingly used to characterize bacteria and to identify bacterial communities from complex environmental matrices. With respect to food, popular sequencing technologies used to identify microbial communities include 16S rDNA sequencing, metagenomics and metatranscriptomics. 16S rDNA sequencing is currently the most commonly applied technique, while metagenomics and metatranscriptomics approaches are still underutilized to date. The latter approaches provide deep insights into the compositional and functional characteristics of a fermented food microbiome. Compared to fermented food, these technologies are rarely used to characterize the non-fermented food microbiome. In addition to discussing different sequencing chemistries currently available, this review has provided an overview of non-fermented food-related microbiome studies based on these next generation approaches. Overall, the potential benefits of multi-omic approaches in improving food safety was emphasized.
Even though, most metagenomic approaches use NGS platforms to sequence DNA, third generation sequencing (TGS) strategies are not routinely used for microbiome studies. Recently, a third generation sequencing MinION platform was used to sequence full length 16S rDNA amplicons generated from a synthetic community of ten microbial species with varying relative abundance (Li et al.,
YC, SF, and SS organized the draft and wrote the manuscript. YC, SF, SP, KJ, and SS participated in the critical revision of the manuscript. All authors read and approved the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Ankita Naithani and João Anes from University College Dublin, and Gopal Rao Gopinath and Ben Davies Tall from U.S. Food and Drug Administration for the critical reading of this manuscript.