16S Metagenomics: A case study for troubleshooting spoilage sources

November 17, 2017

Understand how metagenomics can help you troubleshoot spoilage and shelf life issues.

Video Transcript

TITLE: 16S Metagenomics: A Case Study for Trouble-shooting Spoilage Sources, Joe Heinzelmann, Director of Business Development, Food Safety Genomics

Joe Heinzelmann (voice-over): Today we're gonna do a webinar.

This webinar is entitled 16S Metagenomics: A Case Study for Trouble-shooting Spoilage Sources.

My name is Joe Heinzelmann. I'm the Director of Business Development for our Food Safety Genomics business.

And, today, we're gonna jump into some data that really helps people understand how metagenomics data helps tell a story which will uncover unique issues within a processing environment.

So, three kind of takeaways that you'll see today from this data set is how we're able to see un-culturable organisms that could cause spoilage. We'll show how the metagenomics is able to help tell this story on where these issues are coming from within a processing facility and we're going to go through a dataset here and these slides to help kind of pull it all together.

As a reminder, if you don't remember, there are two additional webinars. What we've tried to do is break up these this web presentation into three kind of bite-sized chunks, depending on your application and what you really want to get after.

The second one will be more focused around the differences between whole-genome sequencing, shotgun metagenomics and 16S metagenomics. That's in December.

Last one will be metagenomics and the application for shelf-life studies and clean label formulations. So a lot of the impetus for clean label and reformulation is causing microbial spoilage concerns and issues. We can show you how metagenomics helps address some of those and arm scientists to help tackle those concerns.

All those are available on our website and, if you would like, we can send you the link after the webinar.

So, as we kind of get into a little bit of background, I like to kind of take it step back and have you think about genomic testing in three big areas.

The first one is pathogen outbreak. So we're talking about whole-genome sequencing, any type of subtyping methods, all kind of fall into this big, first bucket.

The next one is microbiome and spoilage trouble-shooting. This is for, maybe, non-pathogenic organisms, microbiome analysis for sanitation verification, all that falls into that middle bucket.

And the last one is what we call food fraud. So this is your meat species identification, identity testing, food fraud, all all these applications can leverage next-generation sequencing to help provide better data and additional insights.

So the first kind of concept I really want to drive home is this idea of a microbiome. Microbiome is just a $64,000 word to help us talk about how microorganisms relate to each other when they're close in proximity. And there's three things that I want us to think about as we think about the microbiome.

First one is "viable, but not culturable". Our ability to identify and quantitate microorganisms is often limited by our culture method. It served us well so far in terms of our progression of food safety and food quality over the decades, but this is, I think, these new technologies are going to help us understand some things that we wouldn't have been able to see with traditional media.

And that kind of leads into the next point is media biases.

Especially with anaerobic organisms, these biases that we have with tryptic soy agar or septic soy broth, really tend to lead our analyses and our investigations to a specific answer and that might not be the best approach.

[A slide titled Data set #1 is shown. A legend down the right-hand side lists different bacterium and pathogens and shows how they are color-coded in the graph. Highlighted are Staphylococcus, Brevibacterium, Pseudomonas, and Meiothermus. The graph is titled "Abundance at Genus Level". The y-axis is labelled "Abundance in %" and the x-axis is labelled "Sample ID". The first 4 bars for "Sample ID" are almost completely Stapholococcus. The fifth bar is almost completely Pseudomonas. Continuing rightward for 15 to 20 bars, the bars begin showing small amounts of Brevibacterium that increase rapidly to 60% where it levels off. The bars are complemented by various other bacterium including Brachybacterium, Staphylococcus, and other minute amounts of various bacterium. The fourth from last bar shows a large population of Pseudomonas. The next two bars are almost all Brevibacterium. The last bar is almost all Meiothermus.]

So, this is a data set from a 16S metagenomics. What you're really looking at on the x-axis are different sample type. Okay? So on the x-axis, each bar represents a different sample. Each color is a different microorganism. The size of the bar represents the amount of microorganisms in the particular sample and their relative to each other. Okay?

We're gonna get into. This is the data set we're gonna get into.

As we kind of take a step back now that we see what a data set looks like, each sample, if we were to do this through traditional methods, would have to kind of go through this generic process.

We're doing a dilution and an isolation, maybe on a tryptic soy agar or VRBGA [Violet red bile glucose agar].

Then we get into initial slant and ID. We still have interpretation.

And then there's some type of final confirmation step. Often a biochemical ID, your API strips, maybe you're doing traditional 16S typing at this point, or even MALDI-TOF [Matrix-assisted laser desorption/ionization-time of flight] for an ID.

The point is here, is that for each different microorganism, you'd have to go through this process. And it all depends on that first step. In addition to having this be an iterative process, you'd have to be able to grow it up to identify it.

And what we'll see in some of the data sets is there's kind of some unique microorganisms that we're seeing in our data that we have here today. But we're also seeing this in some of the other data sets that we've been running as a service.

One last point that I want to get into before we get into the data is: There's a difference between 16S metagenomics and shotgun metagenomics.

For 16S, what we're doing is we're amplifying one specific region of a bacterial genome which helps us identify that particular organism. This happens out of either a product, or from an environmental sponge. And what this means is that we aren't going through those dilution or isolation steps. We're going straight to the sample. We're obtaining those bacteria and then we're starting the sequencing process.

Very similar to shotgun metagenomics, the difference there is the amount of data and how we approach that from a sequencing standpoint. So, one of the things we've seen with 16S metagenomics as compared to whole-genome sequencing, shotgun genomics, is a sensitivity on clean surfaces because there is a PCR amplification. We we like what this data is telling us, especially in a food production context.

And this is how it looks is: So, we get samples in. We do the sample extraction. We do PCR. We are doing the sequencing on a MiSeq and then we get in the data analysis. Okay?

So, the case study that we have here today is from our manufacturing environment. If you're not familiar, we manufacture a Soleris®. It's a rapid microbiological test system and their aseptic fill using traditional bacterial medias.

What we did for this experimental design is we included both raw materials, which are supposed to be sterile. And what that means is any type of bacterial contamination, would be considered a spoilage event. And we went through and sampled multiple different sites, some raw materials and we're using this as part of that comparison between spoiled and non-spoiled samples.

And this is going to give us an understanding of this facility microbiome or microbiota and that's going to lead us some to some root cause analysis.

So, a little bit of background about our manufacturing facility. This is a cleanroom Class 100, but I think what you'll see here today is that, conceptually, this applies to any type of food manufacturing where you've got a product that has a bacterial profile. We were doing some routine monitoring and identification of isolates. And we saw this as a unique opportunity to use metagenomics to look at the whole facility and how bacteria were exchanging between sites and locations and the product.

[The slide titled "Data set #1" is seen again.]

So as we kind of get back into the data, there's a couple of things I'll remind you. So as we talked about before, on the x-axis, each one of these represents a different microbiological swab. The colors that you see here are different microorganism and the size of this bar represents the abundance of that microorganism.

So on the left hand side here, what you're looking at is nearly all Staphylococcus and the sample is on the left and then you can see that it's kind of sporadic throughout the facility.

[A slide titled "Step 1: contamination investigation" is shown. The bars from the graph showing the 9 bars from the graph on the slide titled "Data Set #1" is displayed. Next to the edited graph is a list: "Key Bacteria: Staph, Pseudomonas, Meiothermus, Brevibacterium".]

So the first thing that we like to do in these types of analysis is pull out the product. And so what you could imagine here is that if the product had passed our quality standards there wouldn't be any type of bacteria.

And what you see in these, is this in this contamination event, we're seeing four different key bacteria. We have Staph, Pseudomonas, Meiothermus and Brevibacterium. And there's another interesting piece of data we can pull out from this, this slide right here.

What we're looking at are two different contamination events. Specifically, these samples here.

(And I'll pull the eraser off.)

Those samples there in the middle, right here, have a mixed flora that you see in these contamination events. On the bookends here for the Staph and the Meiothermus, only one bacteria dominates that contamination event.

And that's going to help in our analysis.

[A slide titled "Step 2: Potential source of contamination" is shown.

Organism Potential Sources
Staph Humans, raw materials
Pseudomonas Ubiquitous, water sources
Meiothermus ????
Brevibacterium Actinomycetales, soil bacterium, likely from air particles, human skin


So as we kind of look into what these bacteria represent, think about Staph and Pseudomonas as pretty ubiquitous. Staph and Brevibacterium are coming in from people, whether that's soil, or hair, or whatever. The really interesting one is Meiothermus.

[A slide titled "Step 3: Source Tracking of Spoilage" shows a graph of the bars with the most Miotherus abundance from the graph on "Data Set #1". Arrows point to the largest Meiothermus population segments shown on the graph. On the right is a list: "Meiothermus - Correlates to S30, S19, and S4 - Hot Water bath, Table adjacent, and drain respectively".]

So if we if we look for Meiothermus within the data set, what you can see is that, if we start from the least abundance and move forward, you can see from S4 to S19, it grows in abundance. Then from S19 to S30, it gets even bigger. And that is kind of helping us point to S47 which represents the contaminated product.

And what this means...what this correlates to: these are environmental samples from a hot water bath, a table adjacent to the hot water bath and a drain.

And it's going from smallest to largest. So the smallest abundance was in the drain. Table was the second bar we showed. And S30 on the left-hand side is that hot water bath.

[A slide with definitions is shown titled "Step 4: Organism Review - Meiothermus".]

What's really interesting is if you look up Meiothermus, it's a thermophile. It forms biofilms. It can be resistant to certain biostatic cleaners.

All this means was that when we were in our manufacturing process, as a heat source we were using a water bath. We were cleaning and draining the water. We'd come back after our sanitation event. Re-inoculated it with water. Warm it up and we have an opportunity for the biofilm to go reproduce, proliferate and cause issues.

So what you'll see in the next data set is we went in there with some corrective actions. We pulled out the water bath. Found a new heat source and completely remove that. And the next data set is set around a sanitation.

[A slide titled Data set #2 is shown. A legend down the right-hand side lists different bacterium and pathogens and shows how they are color-coded in the graph. Highlighted are Staphylococcus, Brevibacterium,  and Lysinibacillus. The graph is titled "Abundance at Genus Level". The y-axis is labelled "Abundance in %" and the x-axis is labelled "Sample ID". The first 7 bars for "Sample ID" show 2 samples are almost completely Staphylococcus. Starting with the eigth bar and continuing rightward for 15 to 20 bars, the bars begin showing small amounts of Brevibacterium that increase rapidly to 75% where it levels off. The bars are complemented by various other bacterium including Brachybacterium, Dietzia, and other minute amounts of various bacterium.]

So there's kind of two things I want you to take away.

The first one when you see this data set is that there's no Meiothermus.

So from the first initial metagenomics run we're able to identify the source and a way to prevent it from happening again.

Alright so then this data set, what we've done is we've centered half the samples around pre-sanitation and half post-sanitation.

[A slide is shown titled "Pre and Post Cleaning" with sample bars from the graph separated into 2 columns: On the left are the bars with little or no Brevibacterium; on the right are the bars with around 75% Brevibacterium population. The left colummn is labelled "Pre-cleaning comprised of Staph, Brevibacterium, and bacillus" ant the right column is labelled "Post cleaning Staphylococcus removed. Samples have higher relative abundance of Brevibacterium.]

So if you remember, the bacteria that we were really worried about is Staph, Brevibacterium and what we're seeing is Brachybacterium and Dietzia pre-cleaning. So this is after some run-time. What we see post-sanitation is really pretty interesting because we've been able to successfully remove almost all the Staphylococcus, which is one of our key bacteria that are causing the contamination, and we're seeing Brevibacterium and Brachybacterium.

So if we look into what those bacteria are, they kind of help uncover why they are there. That's helping tell the story about the microbiome of this manufacturing facility.

[A slide with definitions titled "Brevibacterium" is shown.]

So Brevibacterium is a mesophile. It's halophilic, alkaline tolerant and it's pretty ubiquitous, like we talked about. The other thing, I mean one of the really kind of key things here, is the alkaline tolerance and the halophilic.

When we looked at the corrective actions and looked at the cleaners and sanitizers that were being used in these facilities this facility, we're using an alkaline cleaner. The phenylphenol and para-tertiary amylphenol are both alkaline in nature.

So if we look at how we're cleaning and what impact that has on the microbiome, now we have at least an understanding of where these bacteria are coming from and what we might need to do.

So we're looking at the the rotation of the acids and bases within this cleaning of bacteria. We're looking at how Brevibacterium could be introduced through people and as I said before, we've removed the water bath so our contamination source of Meiothermus has been removed.

So hopefully with this data set, I think you hopefully understand exactly how metagenomics is very different than what you might be seeing with whole-genome sequencing databases and outputs with phylogenetic trees.

And how the changing abundance and presence of bacteria throughout a manufacturing process can start to tell a story of how things are moving. What the environment's actually like and the impact that has on product quality.

[A slide titled "Salmonella and Listeria Reporting Turned Off" is shown. The slide has a graph titled "Abundance Profile At Genus Level". The y-axis is labelled "Abundance in %" and the x-axis is labelled "Detected Genus". A legend at top lists a variety of bacteria. The bars reading left to right show mostly Bacillus as the predominant bacterium. S19 and S14 are highlighted in red. S19 is mostly Bacillus. S14 is mostly Staphylococcus with some Bacillus.The graph caption reads: "Red identifies Listeria spike at high levels".]

So there's one question we get all the time is: what about Salmonella and Listeria?

So this has resolution to the genus. So we can't go beyond Listeria species.

We can, in our bioinformatics pipelines, turn off these bacteria and they...kind of our understanding is you'll have your own food safety programs in place to mitigate and monitor and address these hazards. And, for a spoilage investigation, we can turn these off for that particular experiment. And what you see in the two red boxes are samples that had been spiked with a Listeria species.

And as you can see in this reporting event, these do not show up in this analysis.

So how we typically like to approach metagenomics projects is we'd like to sit down and talk to you.

One of the key things that you'll see is the metadata is really important. So designing how many samples are coming in from product or from the environment or from raw materials, we've got some experience now -- we've done a few of these data sets -- and we can kind of help walk you through that process.

It helped coordinate the samples and then when those come off, we can help support the data interpretation and help you understand what you're seeing, because we are seeing a few these data sets.

We've got a couple of documents that, if you're interested, we can have them sent to you.

One of them is the submission guideline, we do have some recommendations for the particular type of sponges that we validated this service on. We've got some sampling recommendations and then, the other thing, are the submission forms that have a lot of the data capture elements.

So we can walk you through, kind of the key things, that you see or would want to capture to help that data interpretation become a lot easier.

So just as a reminder, what we're talking about today is a service that we are offering. We are using the Illumina MiSeq™, if we kind of want to give them the details a little bit.

We do all the extraction. You send us in both, environmental swabs or products, and what we do is we start to do that analysis. We do the extraction for you. And then 14 days after all the samples have been to our lab, you'll have a report.

This is part of a suite of products. If you're not familiar with NeoSeek™, it's a brand of Neogen® and this is the service that we run out of our Lincoln, Nebraska genomics facility.

It also includes NeoSeek™ STEC for ID and Confirmation. This is a 24-hour ID service that does the top 7 E. coli serogroups.

The other one is Salmonella stereotyping. A 3 to 5 day turnaround time and this uses the MiSeq™ as well.

We talked about 16S metagenomics.

We also do some of the food fraud aspects of this, which is meat species identification.

And we are working towards whole genome sequencing services that will be kind of above and beyond what you see out there today.

So with that, that kind of concludes this kind of brief data set, I would love to take some questions and maybe answer some of those based on what you've seen here in this data set. And if you have any other comments you can either find us on Twitter, LinkedIn, or email any of your sales reps or email me and we can start to answer these questions.

Thanks again for your time. Talk to you soon.


Category: Solution Spotlights