Spring 2022

In this edition:

• Mashpit: Sketching out genomic epidemiology
• Characterization of Salmonella through poultry processing 
• Survival and fitness of Salmonella and Cronobacter on sesame seeds and in tahini paste
• Identification of infectious human norovirus enabled by human intestinal enteroids
• Rapid one day detection of fungal spoilage organisms using nanopore sequencing

Mashpit: Sketching out genomic epidemiology

Summary

We created a rapid genomic epidemiology platform called Mashpit. Users can query a genome, e.g. found in a food facility, against a database of known pathogens.

Situation

We are in the era of genomic epidemiology. For many transmissible diseases, large percentages of the pathogenic agent are being whole genome sequenced. Then, their sequences are being compared in various ways. One example is  Salmonella, where it is sequenced through the PulseNet molecular surveillance network and then uploaded to NCBI. From there, these sequences can be analyzed in a variety of ways including multilocus sequence typing (MLST). These comparisons can help track down a food vehicle and lead to actionable results such as a food recall or inspection.

We note that for some organisms like Salmonella, queries can be of a sensitive nature. For example, if a regulatory agency uncovers Salmonella in a food processing plant, then it is possible there would need to be regulatory action. However, this result would need to be confirmed conclusively such that the public would not be needlessly alerted and a business needlessly affected.

Amazingly for these organisms, at the time of this writing in January 2022, there are more than 500k Salmonella genomes. These numbers are expected to increase dramatically and therefore faster methods are needed.

There have been some major advances to increase the speed of these comparisons. We observed that a new algorithm for genomics called min-hash became available through software called Mash. Querying with Mash can be about three orders of magnitude faster than other common methods like Basic Local Alignment Search Tool (BLAST) and can have a smaller disk footprint. Therefore it can be run on more common scientific workstations. However, it does not yield metadata (e.g., date of isolation) which is necessary to gain meaning (e.g., if it is an ongoing outbreak or if it happened 20 years ago). On the other hand, NCBI has also created the Pathogen Detection (PD) pipeline and site. It combines information from at least three databases: SRA, GenBank, and BioSample. About once a day, it compares all genomes of a given taxon, separates all genomes into individual clusters using MLST, and then creates a phylogeny for each cluster using single nucleotide polymorphism (SNP) analysis. This method is quite comprehensive, but it relies on each sample being public, and it cannot be executed locally.

Response

We present Mashpit, a new rapid genomic epidemiology platform to query against these large groups of genomes on a local computer. Mashpit is a database of min-hash sketches of genomes alongside their metadata, e.g., dates of isolation. Users can add their own genomes to the database offline. Users can query genomes against this database rapidly offline and generate spreadsheets of results. For example, a Salmonella query could produce a spreadsheet of large distances against the database, indicating that the query is unrelated to any known genome. Or, it could produce a spreadsheet of small distances, indicating that it is related to a known sample. Information from the database hits such as location and date could be helpful for genomic epidemiology.

Impact 

We have provided Mashpit to the public health community. A genomic epidemiology study can filter through hundreds of thousands of genomes in seconds with Mashpit, such that only tens of genomes need be in a high-resolution analysis. This high-resolution analysis is most commonly CPU-intensive but well-validated in the public health community and can include MLST or SNP analysis. Although it is too soon for a substantial public health impact to be measured with our stakeholders, we can show that we average about 75 visits to the Mashpit website every week. We believe that a Mashpit instance in any food safety office will increase the response time for any possible pathogen response.

Fig. 1. The spreadsheet output. A single query against the Mashpit database only takes seconds and returns a helpful spreadsheet. Here, we queried with a Salmonella enterica genome SRR17383926 which has a similarity of 0.985 (distance of 0.015). We can gather that our genome is most similar to a sample from CDC of the same species and represents a SNP cluster shown in the last column. This value is truncated in the image but evaluates to https://www.ncbi.nlm.nih.gov/pathogens/isolates/#PDS000038749. This site is shown in Fig. 2.

Fig. 2. The NCBI result. The NCBI result shows three panels: the selected samples on the left; the spreadsheet of samples on the top; and the phylogeny below. The top result from Fig. 1 is highlighted in the phylogeny. In this phylogeny, there are a set of highly related samples that could be investigated further.

_______________________________________________________________________________________________________________________________________________________________

Characterization of Salmonella through poultry processing 

Nikki W. Shariat 

Summary 

Antimicrobials are used during the chilling process to reduce foodborne pathogens, including Salmonella, on poultry products. Salmonella prevalence and serovar populations were evaluated during processing at three broiler processing establishments, and serovar growth differences in an antimicrobial were also assessed. 

Situation 

Despite a reduction in Salmonella during broiler processing, human cases of Salmonella linked to poultry continue to occur. One hypothesis is that some Salmonella are more tolerant of antimicrobial interventions, such as peroxyacetic acid (PAA) or chlorine. Early work showed that prior to any antimicrobial intervention, individual broiler carcasses can harbor multiple different Salmonella serovars. This study evaluated changes in Salmonella serovar populations by CRISPR-SeroSeq through broiler processing where PAA was used during the two step chilling process (pre-chill and main chill). This study also assessed the growth of different serovars in chlorine. 

Response 

Three commercial processing plants were visited (eight independent visits) and 660 rinsates were evaluated. Salmonella isolation was performed in parallel in RV and TT selective enrichment and plated on XLT-4 and BGS agar. Salmonella incidence pre-intervention (pre-chill or preOLR) was 66.4% (146/220), 13.2% (29/220) at post-pre-chill, and 2.3% (5/220) post-main chill. Serovar population analyses were performed on all samples, though the low incidence post main chill prevented us from drawing meaningful comparisons between pre-and post-main chill. Two collections (Plant 1, collections 1 and 2) provided 10 or more positive samples in the post-pre chill. Here, there was a significant reduction in the number of serovars per carcass at pre-pre chill compared to post-pre chill from collection 1 (two-tailed t test, p <0.05) but not for collection 2 (p>0.05). Serovar Typhimurium was prevalent in Plants 1 and 3, while serovar Kentucky was prevalent in Plant 2. Overall, the average number of serovars per carcass pre-pre chill was 1.55, (range, 1-4 serovars per carcass). In each plant, the serovar populations differed between collections, suggesting that Salmonella contamination is driven by the serovars on the birds coming in (each collection corresponded to a single flock), rather than contamination by persistent Salmonella in the plant/on equipment. An example of this is shown in Figure 1. There was no difference in serovar enrichment between RV and TT selection. Ten different serovars were tested for to evaluate their growth in chlorine (0-300 ppm). Growth of all serovars was significantly attenuated at high concentrations of chlorine. At 12 hours growth in 50ppm chlorine, the average reduction in growth was 30.4% compared to growth without chlorine. Serovar Typhimurium was able to grow best at this concentration (9.3% reduction) compared to the other serovars (all >24% reduction), while serovar Heidelberg was most sensitive (37.6% growth reduction). 

Impact 

The results of this study show that Salmonella incidence is effectively reduced following carcass chilling during slaughter. The data indicate that multi-serovar populations on carcasses at processing occur frequently pre-intervention and that the identities of these serovars are driven by the incoming flock. All serovars tested were sensitive to high concentrations of chlorine, and serovar differences were revealed at lower concentrations. 

Fig. 1. Salmonella serovar populations in broiler processing. CRISPR-SeroSeq was performed on Salmonella positive carcass rinsates (32%; 38/120) collected over two different visits to Plant 3. Collection 1 represents a flock harboring four different serovars, while only two serovars were found from Collection 2. The heat map depicts the relative frequency of each serovar within an individual sample, color-coded according to the key shown.

Fig. 2. Salmonella growth in chlorine. Ten serovars commonly associated with human illness or frequently found in poultry were grown in increasing amounts of chlorine and their growth over 24 hours was evaluated.

_______________________________________________________________________________________________________________________________________________________________

Survival and fitness of Salmonella and Cronobacter on sesame seeds and in tahini paste

Issmat I. Kassem

Summary

We evaluated the survival of Salmonella and Cronobacter sakazakii on sesame seeds and associated paste (tahini). Furthermore, we determined the properties of Salmonella and C. sakazakii associated with tahini production.

Situation

Sesame seeds and derived products (SSDPs) are used as components in various foods and are prized for their nutritional benefits. However, SSDPs such as tahini (sesame seed-based paste) have been implicated in serious Salmonella outbreaks in different countries, including the United States. Furthermore, we have recently documented the emergence of multi-drug resistant C. sakazakii in SSDPs. Although the seeds are normally roasted before further processing, contamination appears to happen post-roasting or due to defective roasting procedures. However, there are currently no robust interventions for post-roasting contamination of end-products with Salmonella or Cronobacter and little information on the properties of associated strains. Given the expanding interest in sesame seed products by both consumers and the industry, investigations of the properties and survival of bacterial contaminants are critical.

Response 

We assessed the antibiotic resistance (AMR) profiles and biofilm formation ability of Salmonella enteritidis and C. sakazakii isolated from sesame seeds and tahini production. We also quantified the survival of selected strains on sesame seeds and in tahini at 4° C and room temperature. The latter was accompanied with an evaluation of the impact of persistence in the matrix on antibiotic resistance profiles.

Impact

Salmonella and C. sakazakii isolates exhibited resistance to important antibiotics, including ampicillin and tetracycline. The majority of C. sakazakii and 30% of the Salmonella were classified as multi-drug resistant (MDR). All isolates exhibited a varying ability to form biofilms. Furthermore, MDR strains were detectable at high densities for many days (> 48 days in some instances) on sesame seeds and tahini, especially at 4°C. The persistence had no impact on the AMR profiles. Our findings show that MDR Salmonella and C. sakazakii can occur in both sesame seeds and tahini paste. These strains have the ability to form biofilms which might contribute to their persistence in the tahini production environment. Furthermore, the strains can remain on the product for an extended period of time, further highlighting the risk associated with contamination.

Fig. 1: Biofilm formation ability of selected A) C. sakazakii and B) Salmonella strains.

_______________________________________________________________________________________________________________________________________________________________

Identification of infectious human norovirus enabled by human intestinal enteroids

Revati Narwankar, Amy Mann and Malak A. Esseili

Summary

The Esseili food virology lab at the UGA Center for Food Safety is optimizing the use of a newly-identified cell culture system based on human intestinal enteroids (HIE) for studying infectious norovirus in food, water and the environment.

Situation

Human norovirus is the leading cause of foodborne outbreaks in the US. The virus is also associated with waterborne outbreaks. Since the first reported outbreak of human norovirus in the 1970s, these viruses have been recalcitrant to cultivation in permissive cell lines. This resulted in the expansion of research utilizing culturable surrogate viruses and various molecular tools to estimate the virus infectivity. However, it is widely known that these methods and the surrogate viruses are not ideal and may not mimic the actual virus under field conditions. In 2016, a breakthrough occurred in culturing HuNoV in cells derived from 3D human intestinal enteroids. This cell culture system is not commercially available, is labor- and resource-intensive and requires the use of human fecal samples positive for norovirus.

Response

Before the pandemic, the PI trained on using this cell culture system at the National Calicivirus Laboratory at the CDC. During the pandemic, we were still able to acquire the human stem cells and initiate the process of 3D culturing and then differentiating these into monolayers which are then susceptible to infection with human norovirus (Figure 1). This latter process takes at least 40 days. We obtained and screened several fecal samples from collaborators at the CDC, Emory University and Canada. We successfully implemented the HIE cell culture system and identified infectious norovirus at high titer in the CDC sample so far (Figure 1E). This cell culture system will be used in studying infectious norovirus in food, water and the environmental matrices.

Impact

The food virology lab is one of the very few in the world that is adopting this expensive and labor-intensive cell culture system for studying infectious norovirus in food, water and environmental matrices. 

Fig. 1 : (A) Human intestinal enteroids inside wells of a 24-well plate shown on 3D matrigels droplets (B) Human intestinal enteroids after one month of passage in cell culture as seen under light microscopy showing the 3D structure of the mini-guts, (C) enteroids dispersed into monolayers in 96-well plates grown on collagen, (D) Differentiated Cell monolayer after 4 days in cell culture as seen under light microscopy (E) Successful replication of a human norovirus from fecal suspension as shown by significant increase in log copies of viral RNA after 3 days of incubation on differentiated HIE monolayers.

_______________________________________________________________________________________________________________________________________________________________

Rapid one day detection of fungal spoilage organisms using nanopore sequencing

Henk C. den Bakker, Amy Mann

Summary

The research team of Dr. Henk den Bakker from the UGA Center for Food Safety is working to develop a rapid detection method to find fungal spoilage organisms in foods.

Situation

Fungal spoilage organisms (i.e., yeasts and molds) can grow in a wide variety of foods and can grow in environments that are usually hostile to other microorganisms, such as foods with a low water activity, high sugar, or high salt content. Identification of fungal spoilage organisms is often time-consuming and requires knowledge of and training in mycology.

Response

For this project, we are looking to develop a protocol that consists of (i) an easy-to-use DNA extraction method for fungal DNA from foods and (ii) a rapid DNA sequencing method to identify the fungal species from the food samples. For the DNA sequencing part of the protocol, we are using Oxford nanopore sequencers (small USB drive-sized sequencers that work on a laptop) and targeting ITS, a region that is found universally in all eukaryotic species, including fungi. For the development of efficient DNA extraction methods, we are spiking foods (yogurt, fruit pulp and honey) with yeasts strains and using commercial DNA kits and modified protocols for DNA extraction with such kits to recover yeast DNA. So far, we have found most commercial kits to underperform in the retrieval of fungal DNA of the foods we have tested, however, the addition of treatment with specific enzymes to the protocols improves DNA retrieval. Initial exploration of the ITS region shows that while it can easily be amplified and sequenced, it may not be discriminatory enough to discriminate between certain species of yeast, and the exploration of other target regions is warranted.

Impact

Food spoilage (including fungal spoilage) accounts for a huge amount of food loss annually. Our project will deliver a tool to explore the ecology of fungal spoilage organisms in food ingredients and final products and may help the food industry find new ways to avoid fungal spoilage.

_______________________________________________________________________________________________________________________________________________________________