The role environment plays in microbial diversity

First thing’s first: I know I spent most of 2015 not blogging but I’m hoping that I’ll have more time in the upcoming year to put together at least a couple of interesting posts. 2015 was an interesting beer year for me, including a 6 week long mad brewing spree to get all of my beers ready for my August wedding, tasting new beers from around the world, and making a wild cider from a VT friend’s backyard apples (and subsequently using the culture to make wild beer). I promise that I’ll post more in 2016 than in 2015 – it’s my Brew Year’s Resolution.

Speaking of Brew Year’s Resolutions, I’ve spent a lot of time recently listening to Drew Beechum and Denny Conn’s new podcast Experimental Brewing. It’s an excellent biweekly podcast with great production value and interesting beer-related topics. I’d definitely recommend checking it out. In a recent episode, they had an interview with Nick Impellitteri (biobrewer over at reddit, beeradvocate, etc), the founder and “Chief Yeast Wrangler” over at the Yeast Bay.

I won’t go into too much detail about the interview – after all, you could just go listen to the episode – but there was one point Nick mentioned that found particularly interesting. He mentioned at one point (and I’m paraphrasing) – that it’s not worth worrying about the ratios of various microbes within a sour blend. As soon as you modify the environment of the blend by adding it to wort, everything is thrown off and the ratios of microorganisms will shift. I agree that it’s not worth worrying about it but a question remains – is it possible to predict which microbes will succeed in the final blend, and will pitching the same blend of microbes without concern for the ratios (say, with a starter vs without) result in appreciably different beers?

Coincidentally, I had the opportunity last month to attend a seminar given by Dr. Benjamin Wolfe, a scientist at Tufts University. I’d classify his work as a definite dream job. Not only is his lab doing high quality, high impact research, but he’s doing it with some of my fermented favorite foods: cheese, salami and kombucha. His lab uses these foods as a platform to study the interaction of microbes and microbial diversity.

One of the projects he discussed was data that was published in Cell in 2014. The article is Cheese Rind Communities Provide Tractable Systems for In Situ and In Vitro Studies of Microbial Diversity,1 and it’s work that (I believe) was done during his postdoc in Dr. Rachel Dutton’s lab at Harvard. He also discussed some other really interesting work but it’s beyond the scope of this blog so I’ll be focusing on a couple of aspects of this paper. The major thrust of the paper was to identify the community of organisms that are found on the surface of aged cheeses, to establish the factors that drive what species thrive in a given cheese community (queso clique? fromage family?), and to determine whether these communities form reproducibly. You can watch a video summary of the project here put together by CellPress:

Using a fungal identification method (ITS) and bacterial identification method (16S sequencing), Wolfe and colleagues obtained and characterized the genera of fungus and bacteria present on the rinds of 137 cheeses from the US and Europe.2 What they found was an incredible diversity in types of organisms present – they identified 24 different genera on these cheese rind samples, with an average of 6.5 bacteria genera and 3.2 fungal genera per sample. There are likely many times more species on a given rind. They then used a bioinformatic method (principal component analysis) to determine what traits may result in a given set of organisms occurring on any one rind. Surprisingly, there was no correlation between geographical origin of a cheese and the makeup of a community. In other words, where the rind came from had a nonsignificant effect on microbes. (I would be interested to know if this could be recapitulated in breweries engaging in spontaneous fermentation from different regions of the world!)

So what does influence the composition of a microbial community? In short, it appears be the local environment –  i.e. the cheese itself. The three factors highlighted by the paper were pH, salt concentration, and moisture content. Interestingly, they also observed that certain bacteria are more strongly associated with certain fungi, suggesting some sort of cooperative growth, or an affinity for similar environments. They also saw that bacteria were highly responsive to the fungi present while the opposite was not the case. Finally, and perhaps most relevant to my original question, they found that seeding cheese with a known collection of microorganisms resulted in a reproducible ratio between the organisms. In other words, when given an identical environment, microbes will grow in a predictable pattern resulting in a similar ratio of organisms at the end of the experiment.

All of this said, this brings me back to Nick Impelliteri’s statement about the ratios of microbes in a sour blend, and why you shouldn’t worry about it much. I wonder if these findings in cheese might apply to a beer. If they do, it would suggest that with a given collection of organisms and a given recipe (a consistent environment), a beer should be able to reliably fermented the same way every time. Of course, real life is always more complicated and there are a variety of reasons why minor variation might occur, but it does imply that a microbial community tends to stabilize over time to conform to the environment and the other species present.

I’d love to know if the same principals of microecology apply to beer as they do to cheese. If they do, it would have pretty significant implications for the reproducibility of sour beers. For example, what environmental characteristics are relevant in defining the ecological diversity of microorganisms in beer (pH, IBUs, carbohydrate sources, alcohol content, osmotic pressure)? Knowing these whether these concepts apply to beer would also lend insight into the cooperative and conflicting roles that microorganisms play in making a successful sour beer (or not)!

 

the paper can be found behind a paywall here: http://www.cell.com/cell/abstract/S0092-8674(14)00745-4 .  A 2015 commentary regarding the use of fermented foods (including beer!) as models of ecosystems by Wolfe and Dutton can be found here: http://www.cell.com/cell/fulltext/S0092-8674(15)00200-7

2Note that this strategy identified genera – i.e. Staphylococcus, Pseudomonas, Vibrio, etc, and not individual species.

Other notes:

Finally, I should note that Dr. Ben Wolfe runs a site called microbialfoods.com, a great resource for learning more about the role microbes play in food – definitely check this out!

Apologies after the fact for such a text-heavy post, but the rules on reproducing journal figures are a bit unclear to me…

The Weihenstephan Strain (WLP300/WY3068) – Saccharomyces or Torulaspora? Part 2

Note: if you haven’t read Part 1 yet, do that first! Or don’t – who am I to tell you what to do? 

On my last post on this subject I left the reader (and myself) with more questions than I did answers. Here, I’ll try to begin to resolve some of these questions without raising too many more (though I’ll likely fail at that).

As a brief recap: WLP300/WY3068, the strain that Weihenstephan uses for its hefeweizen (and has used since seemingly the beginning of time), is claimed by various sources to be one of two strains of yeast: S. cerevisiae or T. delbrueckii. There is definitely a dearth of information about this strain and whether it is one species or the other. There are a couple of scientific publications that claim that it is S. cerevisiae but don’t detail any in-depth species ID. The argument for WLP300 being T. delbrueckii is similarly weak, with all claims undocumented.

Via a cursory literature search, I found a couple of methods that might be useful in determining the species of this strain. One paper hinted that an osmotolerance experiment might be useful. However, upon a more thorough review it appears that the two strains of T. delbrueckii that were tested by this lab vary quite a bit in osmotolerance, to the point where this isn’t really a notable phenotype of the organism. So, that’s out the window – this is not a useful experiment to perform to determine the species of WLP300.

Another paper I found mentioned that T. delbrueckii will sporulate on normal nutrient rich media (YPD), while S. cerevisiae cannot (sporulation of both species can be induced using a starving the cells on a special media that is aptly named sporulation media). I decided to test this using a control strain of brewer’s yeast alongside some WLP300. I plated these cells out on YPD and let them sit for 15 days at 30C.

Unfortunately, I don’t have a T. delbrueckii strain on hand, although it would have been nice to have one to confirm the claim that this species can sporulate on YPD. At both 8 and 15 days, I pulled a small sample and checked them for the presence of sporulating yeast cells, which should look like panels A, C and D here. I’ve attached some photos of the WLP300 and a control strain on day 15 of growth on YPD (apologies in advance for the mediocre pictures taken with a dusty CCD).

CTRL

Control S. cerevisiae after 15 days on YPD. 

WLP300 - not sporulating after 15 days.

WLP300 – not sporulating after 15 days.

I looked at several hundred cells for each sample and didn’t see a single sporulating cell, which is to be expected if both strains are S. cerevisiae. While these data certainly seem to indicate that WLP300 is a S. cerevisiae and not T. delbrueckii, it’s not quite enough information – it’s really just a part of the puzzle.

To further determine the likely species of WLP300, I pulled up some published sequencing data that was pointed out to me by reddit user thewhaleshark. There are 3 genes from the WLP300 genome that have freely accessible sequence data: RNQ1, YGL108C-like protein and Trp1. These data are easily accessible from this page. From there, it’s pretty straightforward to pull up a sequence from WLP300 in GenBank, and to BLAST it* against the entire collection of reference genomes (all species in the GenBank database).

I pulled all 3 sequences, and BLASTed them against both genomes, using the BLASTn algorithm – this allows me to search for somewhat similar sequences instead of just perfect or very close matches. In other words, it reduces the stringency, allowing me to see if there is anything even semi-close to my query sequence. The results look something like this, an interactive list with the hits in order of rank. Red and longer bars are better hits, blue or black and shorter are worse:

Screenshot 2015-01-19 17.09.48

The WLP300 YGL108C-like protein gene sequence BLASTed against all the GenBank database genomes.

In addition, you get a list of the alignments that looks like this:

Screenshot 2015-01-19 17.05.55

As you can see, S. cerevisiae ranks as the closest strain by a large margin, with the highest similarity score.

And here are the results for RNQ1 and Trp1, respectively:

Screenshot 2015-01-19 16.54.25Screenshot 2015-01-19 17.52.30

So what can we conclude from this information? There is a very good chance that WLP300 is in fact S. cerevisiae and not T. delbrueckii. In fact, scrolling down the list, I didn’t observe any matches to the T. delbrueckii reference genome among the (very) distantly related hits.

Out of curiosity, I wanted to determine how closely these genes aligned to anything in the T. delbrueckii genome, so I repeated this search but specified T. delbrueckii (taxid: 4950) as the only organism to query, forcing BLAST only to compare to T. delbrueckii.

Screenshot 2015-01-20 13.21.59

WLP300 RNQ1 BLASTed against T. delbrueckii.

Screenshot 2015-01-20 13.20.48

WLP300 YGL108C-like protein gene BLASTed against T. delbrueckii.

Screenshot 2015-01-20 13.19.15

WLP300 Trp1 BLASTed against T. delbrueckii.

So what do the above images indicate? These three genes in WLP300 are (by far) best aligned with the S. cerevisiae reference genome, and they do not align to any significant degree to the T. delbrueckii reference genome.¹ It’s not a nail in the coffin, but it is very hard to envision all three of these sequences aligning so poorly if WLP300 is indeed T. delbrueckii.

While these data were pretty convincing, I decided to contact some people who surely know more about WLP300 than I do in order to get their positions on the matter.

To check out one of the more knowledgable sources on this side of the pond, I got in touch with an rep at White Labs via email, who told me:

“We are in process of doing DNA sequencing on our strains. We will be publishing that when it is available. So far though, it appears to be S. cerevisiae.

In a follow up email, they confirmed that this was based on previous characterization as well as the newer sequencing data.

At the urging of commenter Estus (in a comment on Part 1), I also contacted the Department of Brewing and Beverage Technology at TUM, research program associated with Weihenstephan, to ask them what species this strain is (known over there as St 68). They told me that St 68 is, without a doubt, S. cerevisiae.² People that have spent far more knowledge of this yeast strain than I have are convinced that it is S. cerevisiae, and I’m pretty sure that I agree with them based on the available data. 

Another brief thought is that there are many people out there on the internet and in publications that claim that (in general, not necessarily WLP300 specifically) hefeweizen yeast are T. delbrueckii. Are any of these strains really T. delbrueckii? Perhaps the most notable claim that I can find is in Randy Mosher’s Tasting Beer. It also appears in a Google preview of his 2015 book “Mastering Homebrew,” although I am not sure if that is the final version of the book. I contacted him regarding his thoughts on this and he (graciously and rapidly) responded with the following:

That was based on the best available material I had at the time, and I would definitely defer to White Labs and other genuine experts in the area.

I have to say that yeast taxonomy is maddeningly contradictory and seems to change every decade or so. Hopefully gene-sequencing work like the research White Labs is helping with will clarify the situation once and for all.

Regarding that second to last point, I absolutely agree. It seems to be hard enough for taxonomists to keep up with microorganisms to ensure that they are appropriately categorized (especially given the capability to obtain increasingly detailed genomic datasets). It’s probably even harder for brewers to keep up with the taxonomists!

So where does this leave us? Well, if we broaden the scope a bit to all hefeweizen strains, I’m very interested in what differentiates hefeweizen strains from the rest of the family of ale yeast. It’s pretty obvious that they are genetically different to, say, WLP001, but I’d really love to see what those differences are. I’d hypothesize that the hefe yeast strains cluster together in a nice little subfamily on the yeast family tree. Who knows? Maybe they would cluster enough to justify their own species/subspecies. It’s impossible to know how genetically unique these yeast strains really are without whole-genome data. However, the requisite data for elucidating these finer points will (or already does) exist courtesy of White Labs.

Ultimately, does it really matter what species WLP300 is? To most people, probably not. It makes a great beer, whether it is S. cerevisiae or T. delbrueckii. However, we’re always learning new things about the various yeast strains that we brew with, and I personally find it pretty entertaining to ask these kinds of questions. I’ll certainly be keeping an eye out for the data coming from White Labs/illumina/Synthetic Genomics – the data that come from that project should be very, very interesting.

* For those less familiar with genetics, BLAST stands for “Basic Local Alignment Search Tool.” It allows scientists to look for sequences in a standardized genome (in this case, from S. cerevisiae or T. delbrueckii). It is essentially a genomic search engine. It allows you to match up your query sequence with matching sequences in a genome, and ranks the results based on similarity to your query sequence.

¹ Even though it doesn’t appear that WLP300 is a T.delbrueckii, it surprised me that it was so different – I was expecting there to be a little bit more homology considering these two species were originally classified within the Saccharomyces genus.

² As a side note, the contact at TUM said that have encountered T. delbrueckii as an unwanted strain in beer making but also mentioned that it does have practical/volitional applications in wineries.

The Weihenstephan Strain (WLP300/WY3068) – Saccharomyces or Torulaspora? Part 1

I was flipping through some old ACS “Molecule of the Week” posts today when I came across this post about 4-vinylguaicol (4VG), one of the molecules that yeast can produce to generate a clove-like flavor in beer. This is particularly common in various yeast strains, such as Torulaspora delbrueckii (although the ACS post uses outdated nomenclature – Saccharomyces delbrueckii). This flavor is particularly desirable in wheat beers and is detectable in beers made with classic hefe yeast strains such as WLP300/WY3068. While the role of 4VG in hefeweizen wasn’t exactly news to me – many people have discussed 4VG and the best practices surrounding it’s production – it got me thinking about the strains we typically use for hefeweizens and related beers.

4VG

4VG, 2-methoxy-4-vinylphenol

I’ve always assumed that WLP300 (the Weihenstephan strain, also one of my favorite strains) was S. cerevisae – after all, S. cerevisae has the capability of converting ferulic acid into 4VG.¹ That said, I was unable to find any documentation indicating that this strain was actually S. cerevisae with the exception of this paper (though it appears to lack validation that WLP300 was indeed S. cerevisae)The authors did sequence 4 loci in all of the yeast strains studied, including WLP300, but it’s not unreasonable that they wouldn’t be able to determine the species based on only the limited sequence data from those four regions (which they were expecting to be highly conserved or highly variable between S. cerevisae strains).

On the other hand, Randy Mosher’s “Tasting Beer” claims that Bavarian Weissbier, aka hefeweizen, uses a unique Torulaspora delbrueckii yeast that produces a clove aroma, along with banana and bubble-gum fruitiness.” Even though he doesn’t directly address this specific strain, it’s hard to imagine him not including WLP300/3068 under the umbrella of authentic Bavarian hefeweizen yeast.

I also found some individuals claiming that WLP300 is indeed T. delbrueckii, in the HBD archive, the Mr. Malty yeast database, and at Eureka Brewing (WY3068), but with no backing documentation.² In fact, many references to WLP300/WY3068 being T. delbrueckii all write “Weihenstephan 68 (S. delbrueckii single strain)” leading me to think that most individuals are finding this information in a common location – perhaps the Mr. Malty yeast database.

Saccharomyces or Torulaspora? – from homebrewing.org

Either way, with the information available on the internet, it’s a dead end, with no really convincing or conclusive information. Both White Labs and Wyeast’s descriptions of the respective strains are curiously vague when describing the strains, not mentioning the what the species is. I would imagine if these strains were not S. cerevisae, that they’d make it a point to highlight that fact.³ As indicated by the Eureka Brewing post linked above, there’s no available rRNA data for this strain, making it a bit more difficult to determine the species without access to specialized equipment or services. I could get the rRNA at work but there would be no way to fund the sequencing without digging into my own pockets – not exactly realistic for this grad student! There are other potentially cheaper routes of species ID, but I’ll get into that in another post.

There is a teeny light at the end of the tunnel. I found a study comparing the osmotolerance of S. cerevisae and T. delbrueckii – it turns out that T. delbrueckii is quite a bit more osmotolerant than S. cerevisae – that is, it can survive in saltier, sugarier, etc conditions than it’s fungal relative. This can be tested (in an admittedly quick-and-dirty way) using the solid media osmotolerance testing methodology described by Hernandez-Lopez et. al, and while it couldn’t conclusively demonstrate which species WLP300 is, it might provide some useful preliminary information. So that’s where the next step lies – growing some WLP300 up and comparing it’s osmotolerance to a reference strain. Stay tuned!

Have any information regarding these strain’s species? I’d love to hear it!

Footnotes

¹As another caveat, not all T. delbrueckii strains robustly produce 4VG (they can be POF-, meaning that they lack the enzyme to produce 4VG). In other words, both S. cerevisae and T. delbrueckii can generate 4VG, but either species may also lack the capacity to do so. Therefore, ability or lack thereof to produce 4VG probably doesn’t help us much in determining what species a yeast strain is.

²I didn’t know about the Eureka Brewing yeast science blog until I googled around a bit on this subject. If you haven’t read the linked article or others on the blog, I’d highly recommend it! Very interesting information.

³And if this strain is indeed not S. cerevisae, is it really an ale yeast (as labeled by White Labs – “Hefeweizen Ale Yeast”)? It seems that most definitions of “ale yeast” specify that the strain is S. cerevisae. Sure, this is being a little pedantic, but I still think it’s an important question to ask!

Edit: I did a bit of reading and it actually might not be cost prohibitive to ID this strain via sequencing, so I may try that as well.