Brewing yeast family tree (Oct 2019 update)

Two new fantastic paper on interspecies Saccharomyces hybrids were recently published in Nature Ecology and Evolution. One from the Hittinger lab and collaborators, and the other from the Verstrepen lab and collaborators. A bunch of new commercially available brewing yeast strains were sequenced during the projects, and this meant there was yet again more data available to add to the brewing yeast tree (latest version now always available from the link https://beer.suregork.com/tree). The tree has been updated mainly with lager yeasts and a number of Wyeast strains. I’ve left out the S. kudriavzevii hybrids for now (e.g. WY1214, WLP500 and Abbaye, but these group in the Beer 2 clade). Anyways, enjoy (click for PDF):

I’ve left out a couple of strains, because they didn’t seem to make much sense (possible mixups or mislabelling?).

WLP530 and WLP775 grouped closely together off in a long branch in the wine clade, so I left these out. But it might be that they belong there?

WLP566 grouped in the Beer 1 clade, and had homozygous nonsense mutations in both PAD1 and FDC1 (meaning it should be POF-), despite being a POF+ saison strain. I left in WLP566 from the Gallone et al. 2016 paper instead.

WY1187 was sequenced in three runs, one was pure S. cerevisiae the other two were lager strains. So I don’t know if there has been a mix up or if this is a blend? I left in the S. cerevisiae run.

As revealed in both papers, WLP029, WLP051, and WLP515 are lager yeasts. WLP838 seems to be a S. cerevisiae strain.

It’s interesting to see that there are no Saaz strains among the commercially available lager yeast, they all are Frohberg.

The infamous WLP644 strain was also sequenced, and it can be found in the Beer 2 clade, close to the ‘Duvel’ strains.

For old versions, with a lot of good discussion about the strains, see here: Nov 2018, Apr 2018, and Dec 2017.

Thanks for having a look and feel free to leave a comment 🙂

Solving the Muri mystery

Last summer (2018), we published a paper on the ’Muri’ hybrid. This strain was isolated from a yeast culture that Bjarne Muri had produced when attempting to revive his grandfather’s old kveik culture. In the paper we did some genetic and phenotypic characterization of the strain (a single cell isolate from Richard Preiss). The strain turned out to be a Saccharomyces cerevisiae × Saccharomyces uvarum hybrid, with substantial contributions from Saccharomyces eubayanus as well. In addition to the characterization, we also attempted to reconstruct the hybrid through hybridization of closely related parent strains.

Interspecies hybrids have been found multiple times from beer; the most famous hybrid of course being lager yeast (S. cerevisiae × S. eubayanus). However, S. cerevisiae × S. uvarum from brewing environments have not really been reported (they have been found from wine and cider though). So this was already an interesting finding. The ‘Muri’ strain behaves very differently from other kveik strains, reaching very high attenuations (thanks to being diastatic) and producing phenolic off-flavours. Genome sequencing also revealed that the hybrid is not related to the other kveik isolates. So the question was, had this strain really been a part of the Muri family kveik culture or was this some contaminant that had unintentionally been propagated during Bjarne’s revival attempts?

As Lars mentioned in his recent article about the strain, it looked like we would probably never get the answer to this question. However, by chance, I stumbled upon an interesting finding when I was going through the recently uploaded sequence data linked to this pre-print on S. eubayanus and its hybrids from the Hittinger lab that had been deposited to NCBI-SRA. The sequence data of one strain, WLP351 Bavarian Weizen, was deposited under ‘Saccharomyces cerevisiae × Saccharomyces eubayanus × Saccharomyces uvarum’. This immediately caught my interest, and I downloaded the data. After trimming, aligning to a concatenated reference genome of S. cerevisiae, S. eubayanus and S. uvarum, and variant calling (see methodology in our 2018 paper), it became more and more evident that WLP351 might actually be Muri (or rather Muri was WLP351 or a similar strain).

First of all, based on the read coverage across the reference genomes, it appears as if the S. uvarum subgenome in both strains have the same S. eubayanus introgressions (Muri left, WLP351 right in the second image below). These are quite distinct to what has been reported in other studies for S. uvarum. If we perform phylogenetic analysis (together with other ‘Beer 2’/’Mosaic Beer’ strains) based on the SNPs present in the S. cerevisiae sub-genomes, we see that they are very similar in Muri and WLP351. Compared to the reference genomes, Muri and WLP351 share 86534 SNPs, and differ only at 470 sites.

So the evidence unfortunately points towards Muri actually being a contaminant and not a part of the original family kveik culture. Even with these new results, I still think the hybrid is very interesting, and the methods and analysis that we have performed in the paper are still relevant and valid.   

The mysteries of diastatic brewing yeast

It’s time for another long due update. I’m very excited to share our latest manuscript on diastatic brewing yeast (or “Saccharomyces diastaticus” as it was formerly known as)! Keep in mind that this is a preprint and the manuscript is currently under peer-review. Below I’ll summarize the main findings of our small project (that ended up having quite significant results).

Edit: The manuscript has now been published in Applied Microbiology and Biotechnology: https://link.springer.com/article/10.1007/s00253-019-10021-y

First of all, here is the link to the preprint: https://www.biorxiv.org/content/10.1101/654681v1

So as most of you are aware, the presence or absence of the STA1 gene is commonly used as a marker for diastatic yeast. However, as recent work by e.g. the Hutzler lab and Escarpment Laboratories have shown, not all strains carrying the STA1 gene are problematic (i.e. not all STA1+ strains can super-attenuate beer or ferment starch). This is of course a problem. While the PCR-based method for detecting diastatic yeast with the SD-5A/SD-6B primers is rapid, it can’t differentiate between the highly and poorly diastatic strains. I set out to find out the reasons behind the variation in diastatic ability that we have observed in STA1+ strains.

We started out by screening 15 strains of S. cerevisiae carrying the STA1 gene (thanks to Mathias Hutzler and Richard Preiss for sharing strains). We tested the diastatic ability of the strains using three different tests: 1) the ability to grow on starch agar, 2) the ability to grow in beer (wort force-fermented with a lager yeast), and 3) the ability to grow on dextrin as a sole carbon source. In all three tests, the 15 strains grouped into two distinct groups: the highly diastatic, and the poorly diastatic. This was already promising.

Next, we sequenced the open reading frame and the upstream region of the STA1 gene in all 15 strains. In simpler terms, we looked at the sequence coding the actual gene and the sequence in front of the gene (which usually controls how much of the gene is expressed into the protein it codes for). Here we observed something fascinating: all 10 of the poorly diastatic strains had a large 1162 bp deletion in the promoter of STA1 (i.e. in the upstream region / the sequence in front of the gene). Within this region that was deleted, is a transcription factor binding site (transcription factors are proteins that control the expression of the gene). We hypothesised that this deletion in the STA1 promoter decreases the expression of STA1 (i.e. less of the glucoamylase enzyme is produced), which in turn results in poor diastatic ability. It seemed very obvious at this point already that this deletion was the cause of the variable diastatic ability, but we still needed to confirm this.

To confirm the effect of the deletion, we performed CRISPR/Cas9-aided reverse engineering. Wyeast 3711 was the most diastatic strain out of all 15 that we tested (in all three tests). So we used CRISPR/Cas9 to replicate this same 1162 bp deletion in the promoter of STA1 in WY3711. We then repeated the three diastatic tests. We compared the original WY3711, the WY3711 variant with the deletion in the STA1 promoter, and WLP570 that naturally has the deletion in the STA1 promoter. The deletion resulted in decreased ability to grow in beer and on dextrin (WLP570 and the WY3711 variant with the deletion performed similarly poorly), confirming that it is indeed this 1162 bp deletion in the STA1 promoter which causes the poor diastatic ability in some of the STA1+ strains. To test our hypothesis that STA1 expression is affected, we measured STA1 mRNA levels with RT-qPCR in the same three strains as above, and witnessed 150-200 fold higher levels of STA1 mRNA in WY3711 compared to WLP570 and the WY3711 variant.

So what can we do in practice with this information? We designed a new set of PCR and qPCR primers that bind within the deleted region in the promoter. So these primers will only give a signal (produce an amplicon) if the strain has the full STA1 promoter (i.e. the strain is highly diastatic). These new primers can therefore distinguish between very active (full promoter) and slow (deletion in the promoter) diastatic strains. They can also be multiplexed with the commonly used SD-5A/6B primers! So this means that you can either use these new primers on their own, which will produce a signal for very active diastatic strains, and no signal for both poorly diastatic strains and strains lacking STA1 completely. Or, you can use these new primers in the same PCR reaction together with the SD-5A/6B primers, which will produce two bands for highly diastatic strains, one band for poorly diastatic strains, and no bands for strains lacking STA1 completely. If this sounds confusing, you can check out the image below (and two above) for some examples. Now we can use PCR to distinguish the highly diastatic and poorly diastatic strains! We tested the primers on some lager yeast cultures that we contaminated with various ratios of WY3711, and we could reliably detect WY3711 down to a concentration of 10^-4 with regular PCR and 10^-5 with qPCR using these new primers.

After we had managed to elucidate the cause behind the variable diastatic ability in STA1+ strains, I still wanted to explore the prevalence and evolutionary role of STA1. Brewers tend to link diastatic yeast with “wild yeast”. This turns out to be incorrect. After screening 1100+ publically available genome assemblies, we show that STA1 is prevalent in the ‘Beer 2’/’Mosaic Beer’ population, and surprisingly a population of yeast isolated from a village in French Guiana (mostly isolates from human feces). So STA1 is clearly linked to domesticated yeast (i.e. no wild strains possessed STA1). The link to ‘Beer 2’ was expected (the saison strains can be found from here), but the French Guiana link was definitely unexpected. The villagers produce and consume a traditional starch-rich beverage called cachiri, so maybe STA1 (allowing starch fermentation) has given the strains a fitness advantage there? Anyways, exploring that was outside the scope of this particular study, but it should definitely be clarified in the future.

If we go back to the ‘Beer 2’ strains, the Gallone et al. 2016 paper has some nice phenotypic data that can be used. I decided to look at the maltotriose use ability of all the STA1+ ‘Beer 2’ strains, and if we divide the strains into two groups based on whether they have the full or partial STA1 promoter, one could see that the strains with the full promoter have significantly higher maltotriose use ability. This hinted that STA1 possibly has some role in enabling maltotriose use in the ‘Beer 2’ strains. The main mechanism for maltotriose use in ‘Beer 1’ brewing strains is through the AGT1/MAL11 transporter. ‘Beer 2’ strains have a non-functional AGT1/MAL11, which means some other unknown mechanism(s) must be responsible. Could STA1 be the unknown mechanism?

I then decided to compare the wort fermentation ability of WY3711 with the WY3711 variant with the deletion in the STA1 promoter. In the first 24 hours of fermentation the strains performed identically. At this time it is mainly glucose that is consumed from the wort. After this time point, the variant with the deletion in the STA1 promoter slowed down significantly. Maltotriose use in particular slowed down. WY3711 had consumed around 80% of the maltotriose after 96 hours, while the variant with the deletion in the STA1 promoter had only consumed 12%. So these initial results already suggested that STA1 has a central role in enabling maltotriose use in STA1+ strains.

To further test this, we next used CRISPR/Cas9 to completely delete STA1 from three of the highly diastatic strains (WY3711 and two TUM strains). We repeated the wort fermentations, and again saw significantly reduced maltotriose use for all three deletion strains. The two TUM strains (TMU PI BA 109 and TUM 71) in particular, barely consumed any maltotriose at all after STA1 had been deleted. This confirmed that STA1 seems to enable efficient maltotriose use through extracellular hydrolysis! We think that the formation and retention of STA1, through the fusion of FLO11 and SGA1, is an alternative evolutionary strategy for efficient utilization of sugars present in brewer’s wort.

The WY3711 deletion strain still showed minimal maltotriose use, so that suggested that while STA1 is the main mechanism for maltotriose use in these strains, it is not the only one. To confirm this, we measured the uptake rate of maltotriose in these strains. This is done with radio-labelled maltotriose (14C). All three strains were able to transport maltotriose into the cell (though at levels lower than what is normally measured for brewing yeast), confirming that while STA1 appears to be the main mechanism for maltotriose use, the strains carry some low affinity maltotriose transporter(s) as well. We performed long-read sequencing of WY3711 using an Oxford Nanopore MinION, and the resulting genome assembly appeared to contain multiple copies of an MTT1-like transporter (which probably explain the maltotriose uptake ability and the low maltotriose use of the STA1 deletion strain). MTT1 is a transporter that has been shown to transport maltotriose.

So to conclude, we saw three major results here in this study. The first is probably of most use to the brewing community: The variable diastatic ability in STA1+ strains seems to be determined by a 1162 bp deletion in the STA1 promoter. Strains with the deletion are not very diastatic. You can use our newly designed PCR primers so differentiate between STA1+ strains with and without the deletion. The second result was that STA1 is not linked to wild yeast, rather it appears to be prevalent only in ‘Beer 2’/’Mosaic Beer’ strains, and surprisingly the ‘French Guiana, human’ strains. STA1 presumably gives a fitness advantage in starch-rich environments (e.g. beer). The third major result was that STA1 enables efficient maltotriose consumption in STA1+ strains. This appears to be the unknown mechanism that has enabled efficient maltotriose use in the ‘Beer 2’ strains (which otherwise have a non-functional AGT1/MAL11 transporter). STA1 therefore seems to be an alternative evolutionary route (‘domestication signature’) to enable efficient utilization of the sugars present in wort. Let me know if you have any questions or feedback, I’d be happy to help! Thanks again to Mathias Hutzler and Richard Preiss for sharing strains!

An updated brewing yeast family tree

I’ve been bad at updating the blog lately, as I have been busy writing on my thesis, taking care of our 1.5 year old son, and being involved in various smaller brewing research projects at work. We’ve had a number of results published during the autumn, including two papers on kveik (here and here) and one on adaptive evolution of a low-diacetyl lager strain (here). I’ll write a separate post about those if I have time. Anyways, the point of this post was to share an updated version of the brewing yeast family tree I’ve blogged about a couple times before. An interesting preprint, A polyploid admixed origin of beer yeasts derived from European and Asian wine populations by Fay et al. (link), was uploaded recently to bioRxiv. In it they propose that the Ale beer / Beer 1 strains are derived from admixture between strains of the Sake/Asian and European Wine populations. In the study they sequenced a number of commercial Wyeast, Fermentis and Lallemand strains, which I retrieved the sequence data for (Bioproject PRJNA504476) and added to the previous version of the tree consisting mainly of White Labs strains (here and here). Below you’ll find the tree in PDF format (click on the image below) together with some observations by ‘qq’ and me. For clarity I decided to keep out the non-brewing strains from the 1011 yeast genomes.

Here are some comments from ‘qq’ (with minor modification by me):

Safbrew WB-06 and Wyeast 1388 Belgian Strong (“Duvel”) – With both of them STA1+, it is no great surprise to see them both up in Beer2 near WLP570 which supposedly came to Duvel from McEwans.

Lallemand BRY-97 – Surprisingly, this strain doesn’t group with the Beer 1 US strains, but rather in the Mixed group. As supposedly one of the key strains in the story of US yeast going from East to West, what is this doing here and not in the main US group?
Muntons English – Presumably not Munton’s Gold but the “ordinary” Munton’s dry yeast, which shows up in a lot of kits. The story goes that this was the old EDME yeast related to Windsor/S-33 which is consistent with what we see here.

Brewferm Lager – not on the chart but according to Table S2 this falls in the Mixed group.

Lallemand Munich – with the other German hefeweizen strains as you’d expect
Wyeast 3068 – supposedly Weihenstephan 68, the classic German wheat (and supposedly the origin of Danstar Munich Classic?)

Wyeast 1007 German – the internet had thought this could be close to K-97 and WLP036, but WLP003 German II makes sense
Wyeast 2565 Kolsch – Makes sense that it’s close to our old friend WLP800 Pilsner.
Wyeast 3463 Forbidden Fruit – assumed to be from Hoegaarden Verboden Vrucht (link), plausible that it’s in that WLP410/510 Belgian Wit II group although actually WLP400 Belgian Wit is meant to be Hoegaarden
Wyeast 3787 Trappist HG – supposedly from Westmalle, close to that WLP400 “Hoegaarden” and WLP530 “Abbey Ale”
Wyeast 3942 Belgian Wheat – supposedly from De Dolle, a Belgian brewery not known for its wits but the yeast falls in that wit group.

Wyeast 1764 Pacman, Wyeast seem to have stopped offering Rogue’s yeast from their Private Collection but Imperial A18 Joystick is meant to be the same. Supposed to be a better-behaved derivative of Chico and this seems to confirm that ancestry.

Wyeast 1275 Thames Valley (“Brakspear”) – close to WLP023 Burton Ale which despite the name is also meant to come from Brakspear.
Wyeast 1332 Northwest – not surprising that it’s close to WLP041 Pacific (“Redhook”) in the WLP002/007 group as it’s meant to come from Hales of Seattle. Mike Hale spent a year in England at Gale’s and brought the yeast back with him (link) and a former brewer at Gale’s has specifically said Gale’s used Whitbread B (link). Supposedly Gale’s got their yeast from Brickwoods, the main brewery in Portsmouth who were bought by Whitbread in 1971. This supports the idea that the WLP002/007 group represents the Whitbread B family, perhaps the most important group of British industrial yeasts. The Gale’s yeast is now used by Marble among others.
Wyeast 1968 London ESB – bit surprising that it’s not closer to WLP002 English since the internet reckons they both come from Fuller’s. But neither of them seem to quite have the “marmalade-iness” of real Fuller’s beer, either they’ve mutated or weren’t actually from Fuller’s in the first place.
Escarpment Labs Vermont Ale – The ‘classic’ NEIPA strain is closely related to Wyeast 1968 in the Whitbread B group.
Coopers Australian Ale Yeast – presumably the dry yeast from their kits? Seems to be an outlier of the main UK Beer1 group which makes sense for an Australian yeast if somewhat distant from WLP009 Australian Ale also supposedly from Coopers.
Wyeast 1098 British Ale – Wyeast 1098 and 1099 are both meant to come from Whitbread, and you will see tables on the internet saying that 1098 is equivalent to WLP007 Dry English. It’s clearly not, it’s close to WLP017 Whitbread II  (an elusive Vault strain) and 1318 London Ale III. It’s a shame that we don’t have sequence for 1099 but its brewing numbers suggest that it’s not much like WLP007 either.
Wyeast 1318 London Ale III – Seems to be another member of that little Whitbread II subfamily.  Traditionally it’s linked to Boddington’s which I never quite believed but Boddies had all sorts of yeast problems in the 1980s and were bought by Whitbread in 1989 so it’s plausible that the original yeast was ultimately replaced by one from the yeast bank at head office (perhaps after they’d tried others?). 1318 is a super-fashionable strain that everyone seems to be using for NEIPAs and is known for hop biotransformation, so it might be interesting to test its relatives for that.
Wyeast 1945 NeoBritannia – An exclusive that Wyeast used to do for Northern Brewer before the ABInBev takeover. Close relative of 1318 in the Whitbread II group.
Wyeast 1469 West Yorkshire – Was fully expecting this to be a Beer2 strain! 1469 is meant to come from Timothy Taylor, who got their yeast from Oldham, who got their yeast from John Smith’s. The John Smith yeast also went to Harvey’s (the source of VTT-A81062, a Beer2 strain). So it’s a bit of a surprise that 1469 is in the heart of the UK Beer1 strains, closest to WLP022 Essex (“Ridleys”). So either the traditional stories aren’t true, there’s been contamination/mixups, or we’re looking at John Smith being some kind of multistrain with both Beer 1’s and Beer 2’s in it.
Wyeast 1028 London Ale (“Worthington White Shield”) and Wyeast 1728 Scottish Ale “McEwans” – wasn’t expecting them to be so close, and for 1728 to be so far from WLP028 Edinburgh (also “McEwans” but at other end of main UK group). Also interesting to see them close to the WLP011 European and WLP072 French pairing, and some way from WLP013 which is also meant to be from White Shield.
Wyeast 1187 Ringwood – as expected close to WLP005 British and NCYC1187. In general Wyeast strains seem to have diverged more than White Labs, and this is a good example.
Safale S-04 – Closely related to WLP006 Bedford (“Charles Wells”) and WLP013 London, even though internet tradition always called it a dry version of Whitbread B. It’s nowhere close to the Whitbread strains.

Hopefully this should be useful both for finding yeast substitutes and elucidating the history of these strains. As genome sequencing becomes cheaper and more accessible all the time, we will certainly be able to update the tree with more strains in the future.

Phylogenetic tree of 1011+157 yeast genomes

A couple of weeks ago the main results of the 1002 yeast genomes project (which actually ended up as 1011 yeast genomes) were published in Nature. This amazing piece of work from the J Schacherer & G Liti labs offers insights into the evolutionary history of S. cerevisiae, and is also an amazing source of data for any yeast nerd (most of the data is freely available to download here). While browsing through the paper and the supplements, I noticed there wasn’t any phylogenetic tree available where the individual strains names were visible (yes I know, such a tree would be quite messy with the number of strains). The relatedness of different brewing yeast strains has been discussed in some of my previous posts and gathered much interest from readers, so I decided to put together a phylogenetic tree myself from the genome assemblies the authors have made available. As I’m a brewing yeast guy, I decided to also expand the tree with the 157 yeast genomes from the Gallone et al. 2016 study. I’ll get into the details below, and bring up some general observations. So, here it is, a phylogenetic tree of 1168 yeast genomes (click the image below to download the PDF):

Notes:

First of all, sorry about the colors. It was difficult to find a good dark color palette (with 24+ colors) to differentiate the different strain origins and clades. I hope the tree is still readable. If not I will post a version with all the strains and branches is black.

The strains were originally named with their code names (3 letter code in 1011 yeast genomes, and XX### in Gallone et al. 2016). I’ve then replaced the code names with the strain names as listed in Supplementary Table S1 of the 1011 yeast genomes paper, and our decoded White Labs strains (only the medium to high confidence identifications). Here is a copy of the phylogenetic tree using only the original code names.

Many of the brewing strains sequenced in the 1011 yeast genomes paper are quite different from the Gallone et al. strains, but there is some overlap (e.g. Beer002, Beer003, WLP099 = Beer071, WLP570 = Beer085).

I think DBVPG6694 (Artois) and DBVPG6695 (Orval) might be mixed up in the paper, since Beer041 is reported as ‘Belgian Lager’ while Beer077 is reported as ‘Belgian Trappist’.

If CFG is Fermentis S-04 (and not S-40 as stated in the Table S1), then it interestingly doesn’t seem to cluster with the other Whitbread yeasts, but rather seems to be close to WLP006 Bedford and WLP013 London.

Fermentis S-33 and Lallemand Windsor are quite closely related.

The WLP530 isolate (CFC) sequenced in the 1011 yeast genomes paper is not at all where I was expecting it. Me and ‘qq’ were assuming Beer078 from the Gallone et al. paper would be WLP530 (which clusters together with several other Trappist beer strains), but instead WLP530 clustered together with Beer095-097 of unknown origin and WLP009 Australian Ale (Beer052). I’m not really sure what is going on here?

There are a couple of S. cerevisiae var. diastaticus strains (e.g. AEQ/CBS1782/NCYC361, YAG/YJM271, and AAQ/CLIB272_2) that cluster in the Beer 2 / Mosaic beer group (the genomes of which might be a source of good info for new identification methods).

There is probably a lot of observations I’m missing, so please feel free to comment 🙂

Quick summary of the methods:

Genome assemblies were downloaded and aligned to S288c using NUCmer through the NASP pipeline. SNPs were then called from each alignment. The resulting VCF was annotated with SnpEff, and filtered to only retain sites present in all 1167 strains, inside the coding region of a gene, and with a minor allele frequency greater than 0.25% (i.e. minor allele present in at least 2 strains). A maximum likelihood tree was then generated based on 462,842 filtered sites with IQ-TREE, using the GTR+F+R4 model and 1000 ultrafast bootstrap replicates.

Here is an archive containing the newick trees, FigTree NEXUS files, and the various strain maps (e.g. color map, code-to-strain name translation).

References

Gallone et al. 2016. Domestication and Divergence of Saccharomyces cerevisiae Beer Yeasts. Cell 166:1397 – 1410.e16
Peter et al. 2018. Genome evolution across 1,011 Saccharomyces cerevisiae isolates. Nature 556:339–344

Improving fermentation performance of yeast hybrids through adaptive evolution

As I mentioned a couple of posts ago, the research project I’ve been dedicating most of my time to the past 18 months is one based on the adaptive evolution of newly created lager hybrids. Our manuscript based on the work was accepted in late November 2017, and today the article was published in the latest issue of Applied and Environmental Microbiology. Please have a look if you are interested!

I thought I’d write up a short summary of our results. It is known that the genomes of newly created hybrids (particularly interspecies ones) tend to be quite unstable. This means that their genomes, along with their phenotypic properties, can change as the hybrid is grown for several generations. This is not very desirable in a brewing environment, where yeast is reused for multiple consecutive fermentations. We decided to exploit this instability, by growing (and adapting) some of our hybrids in media containing high ethanol concentrations for 130+ generations.

Figure_1

We started with four yeast strains, one a S. cerevisiae strain, while the other three were S. cerevisiae × S. eubayanus hybrids. The S. cerevisiae strain was A-81062, an ale strain that was a common parent strain for all three hybrids (this is Y1 in the study). The hybrids were all produced in our previous studies, and they included an allotetraploid one (Y2), an allotriploid one (Y3), and a POF- meiotic segregant, which was approximately diploid (Y4). The allotetraploid hybrid had two copies of the chromosomes from both parent strains, while the allotriploid hybrid had two copies of the S. cerevisiae chromosomes, but only one copy of the S. eubayanus chromosomes. We were interested in seeing if higher ploidy levels resulted in increased adaptation. This is because higher ploidy hybrids tend to be more unstable, and in layman’s terms ‘have more starting material’ available for adaptation.

What we then did with these four yeast strains, was grow them in 30 consecutive small-scale fermentations in two different growth media containing 10% ethanol (see (A) in image above). We took samples after 10, 20 and 30 fermentations, and isolated single cell isolates on agar (B). During the small-scale fermentations, we monitored optical density to see if there was an improvement in growth over time. This there was for all four yeast strains as seen in the image below:

Figure_2_new

We then performed high-throughput screening of all our isolates using a liquid handling robot and cultivations in 96-well plates (C). The growth media contained malt extract and was again supplemented with ethanol. We monitored sugar concentrations using HPLC and selected isolates which fermented the fast. We saw an improvement in fermentation speed compared to the original strains for the majority of the adapted isolates. We then selected a couple of the best-performing isolates from each yeast strain, and performed small-scale wort fermentations with them (D). Again, we saw that most adapted isolates performed better than the original (wild-type) strains:

Figure_4

After this, we selected two isolates from each strain (one from each growth media used during the adaptation process) and performed 2L-scale wort fermentations with them (E). We monitored these fermentations more carefully, and analysed the resulting beers afterwards for aroma compounds. We not only saw that the isolates again appeared to ferment faster (Y1-Y4: A-D):

Figure_5

The adapted isolates also appeared to produce lower concentrations of undesirable aroma compounds such as higher alcohols and diacetyl, while producing higher concentrations of many desirable aroma compounds such as esters:

Figure_6_groups

So, to sum up the results up until this point: by growing the yeast strains in an high ethanol environment, we were able to isolate adapted strains that not only appeared to ferment faster in wort, but also produced beer with a more desirable aroma profile. Interestingly, we saw this improvement with all four strains, not only the hybrids!

What we wanted to do next was try to elucidate what genomic changes had occured, and try to see if we could link any of them to the phenotype change that we observed. We sequenced the adapted strains (Illumina 150bp paired-end) and compared them with the original strains. First we looked at overall DNA content (measured with flow cytometry and SYTOX Green staining), and saw quite large changes in many of the adapted strains. I won’t copy the numbers in here (have a look in the article if you are interested), but we saw quite large decreases in DNA content in the isolates obtained from the tetraploid hybrid, while the isolates remained more similar (one exception though). We then looked at copy number changes of individual chromosomes. We saw, for example, that the hybrids on average tended to lose more of their S. eubayanus chromosomes, and that S. cerevisiae chromosomes VII and XIV were amplified in multiple isolates:

Figure_8_new

Interestingly, chromosome VII of A-81062 contains multiple maltose/maltotriose transporters, and we did actually see a positive correlation between the sugar utilization rate and estimated copy numbers of these maltose transporters:

Figure_8B

We also looked at single nucleotide polymorphisms occuring in the coding sequence of genes that resulted in amino acid changes. We found several mutations that hit common genes across the adapted isolates, such as IRA2, HSP150 and MNN4. In addition, some genes known to be linked with ethanol tolerance, such as UTH1, were affected.

So while there were no very obvious genomic changes that would explain the improved phenotype in the adapted isolates, we saw several factors that appeared to contribute to it. We unfortunately weren’t able to test them through reverse engineering as this was outside the scope of the current project, but it would definitely be interesting to do so in the future!

Our study demonstrates the possibility of improving newly created lager yeast hybrids (and also ale strains!) through adaptive evolution by generating stable and superior variants that possess traits that are desirable in beer fermentation (fast fermentation rate and desirable aroma profile). Thanks for reading!

Abstract:

Interspecific hybridization is a valuable tool for developing and improving brewing yeast in a number of industry-relevant aspects. However, the genomes of newly formed hybrids can be unstable. Here, we exploited this trait by adapting four brewing yeast strains, three of which were de novo interspecific lager hybrids with different ploidy levels, to high ethanol concentrations in an attempt to generate variant strains with improved fermentation performance in high-gravity wort. Through a batch fermentation-based adaptation process and selection based on a two-step screening process, we obtained eight variant strains which we compared to the wild-type strains in 2-liter-scale wort fermentations replicating industrial conditions. The results revealed that the adapted variants outperformed the strains from which they were derived, and the majority also possessed several desirable brewing-relevant traits, such as increased ester formation and ethanol tolerance, as well as decreased diacetyl formation. The variants obtained from the polyploid hybrids appeared to show greater improvements in fermentation performance than those derived from diploid strains. Interestingly, it was not only the hybrid strains, but also the Saccharomyces cerevisiae parent strain, that appeared to adapt and showed considerable changes in genome size. Genome sequencing and ploidy analysis revealed that changes had occurred at both the chromosome and single nucleotide levels in all variants. Our study demonstrates the possibility of improving de novo lager yeast hybrids through adaptive evolution by generating stable and superior variants that possess traits relevant to industrial lager beer fermentation.

IMPORTANCE Recent studies have shown that hybridization is a valuable tool for creating new and diverse strains of lager yeast. Adaptive evolution is another strain development tool that can be applied in order to improve upon desirable traits. Here, we apply adaptive evolution to newly created lager yeast hybrids by subjecting them to environments containing high ethanol levels. We isolated and characterized a number of adapted variants which possess improved fermentation properties and ethanol tolerance. Genome analysis revealed substantial changes in the variants compared to the original strains. These improved variant strains were produced without any genetic modification and are suitable for industrial lager beer fermentations.

 

Citation: Krogerus K, Holmström S, Gibson B. 2018. Enhanced wort fermentation with de novo lager hybrids adapted to high-ethanol environments. Appl Environ Microbiol 84:e02302-17. https://doi.org/10.1128/AEM.02302-17.

Decoding White Labs strains from Gallone et al. 2016 – Update

Thanks to some amazing contributions and discussions in the comment section of my previous post on this article (special thanks to ‘qq’), we’ve now managed to identify (its at least a good estimation) the majority of the White Labs strains from the set of brewing strains in the Gallone et al. 2016 paper.

In addition to the genotype data, the Gallone et al. 2016 paper contains a huge amount of phenotypic data on these strains, and this information could be very useful for (home)brewers as well now that we know which strains are which (at least most of them). I’ve compiled the following spreadsheet (with the help of ‘qq’) containing all the identifications and useful phenotypic data (e.g. POF, flocculation, sugar use, various tolerances, and ester concentrations). See the ‘Notes’ sheet for more information. Press the icon in the lower right corner to open the spreadsheet in a new window (recommended). Please feel free to leave feedback or suggestions for improvement if you find something unclear or missing!

[embeddoc url=”http://beer.suregork.com/wp-content/uploads/2017/12/White_Labs_Gallone_v1_1.xlsx” download=”none” viewer=”microsoft”]

 

Updated family trees

I also put together the following dendrograms displaying the White Labs codes for those interested in how they look in the family tree (click to enlarge). As an added bonus, if anyone is wondering where the lager yeasts fit in on the tree, I’ve expanded it with the S. cerevisiae sub-genomes of two Frohberg strains (‘A15’ (VTT-A63015) and W34/70; names in purple) and two Saaz strains (CBS1503 and CBS1538; names in blue). We’ve sequenced A15 ourselves, while the other sequence reads were pulled from Okuno et al. 2016.

White_Labs_Gallone_Tree_Small_1_0

In addition, the following dendrograms are also available:

‘qq’ also put together the following dendrograms (Beer 1 and Beer 2 groups separately) from Figure 1 in the original study:

gallone1

gallone2

There are some really interesting observations that can be made, e.g.

  • WLP540, the ‘Rochefort’ strain, appears to be of British origin (POF- as well), and not related to other Trappist strains.
  • The group of ‘US’ strains appears to contain one English and one Belgian strain. The Belgian one could be WLP515.
  • WLP800 and WLP320 appear closely related, and their phenotypes seem very similar. Anyone want to try making a lager with WLP320 American Hefeweizen?
  • Not related to the White Labs strains at all, but two strains used for industrial lager production (Beer039 and Beer040) appear to be closely related to various Belgian Ale strains in the Beer2 group.
  • As can be seen, the lager yeasts branch off in the beginning of the ‘Beer 1’ clade together with the Hefeweizen strains. This is similar to the result obtained by Goncalves et al. 2016.

Phenotypic diversity

I’ve also recreated the heatmap from Figure 3 in the original publication, now with the White Labs codes in place of the strain codes. Strains (rows) that are grouped closely together are phenotypically similar. You can see a nice division between the ‘Beer 1’ strains and the rest of the strains. Click to enlarge.

White_Labs_Gallone_Heatmap_1_0

Furthermore, I performed Principal Component Analysis (PCA) on the phenotype dataset, and below you can find the scores and loadings plots for ‘PC1 vs PC2’ and ‘PC1 vs PC3’. These first three principal components explained about 40% of the variation in the dataset. In layman’s terms, you can interpret the plots in the following way:

  • Strains close to each other on the ‘scores’ plot (top) are phenotypically similar.
  • By looking at the corresponding location in the ‘loadings’ plot (bottom), you can find what phenotypic traits are strongly associated with those particular strains.
  • The larger the magnitude of the phenotypic trait (i.e. the further from the origin), the more it contributes to these principal components.
  • E.g. In the top of the ‘scores’ plot for ‘PC1 vs PC2’ we find ‘wine018’, ‘wine017’ and ‘WLP050’. Looking at the ‘loadings’ plot, we see that these strains are strongly associated with high production of ethyl, isoamyl, isobutyl and phenylethyl acetate, as well as isoamyl alcohol. The particular traits have a large effect on PC2.
  • Use these plots e.g. to find similar strains (if you want to substitute one yeast for another for example), or strains associated with particular traits.

PC1 vs PC2

White_Labs_Gallone_PCA_12_1_0

PC1 vs PC3

White_Labs_Gallone_PCA_13_1_0

Presence of STA1 in the strains

Following the recent news about a potential S. cerevisiae var. diastaticus contamination in White Labs yeast, I also did a BLAST search for STA1 (GenBank: X02649.1), the gene encoding for an extracellular glucoamylase, in each of the assemblies. Interestingly, there were full hits in multiple strains (partial matches in many strains as well). The strains are:

  • Beer002
  • Beer013
  • Beer059 (probably WLP026)
  • Beer085 (very likely WLP570)
  • Beer086 (probably WLP585)
  • Beer091 (probably a WLP strain)
  • Beer092 (probably a WLP strain)
  • Wine019 (probably a WLP strain)

I also assembled the reads from the WLP570 sample in the 1002 yeast genomes project, and STA1 was present in the resulting assembly as well. While these strains appear to contain STA1, it is unclear if any of them are diastatic (this needs to be tested). Interestingly, all the STA1 sequences contain a ‘T’ insertion at position 2406, close to stop codon (position 2427-2429). With the insertion, the amino acid sequence is extended by 39 amino acids (and these amino acids are homologous to the C-terminus of the SGA1-encoded intracellular glucoamylase). All this data is in the Excel spreadsheet as well!

Other information

In addition to these identifications, the comment section of the previous post contains a lot of interesting discussion on topics such as the history and spread of different strains, and whether interspecies hybrids have been included in the study. Please have a look! This information would definitely be worth a post of its own, but I’m sure someone else would be better at putting together such a post than me.

Hopefully White Labs will now be willing to unblind their own strains. It would make this data much more valuable, and I’m not really sure what they have to gain from keeping it secret.

Briefly, for those wondering how the tree was constructed: the Illumina reads from the lager strains were aligned to a concatenated reference genome of S. cerevisiae and S. eubayanus. FreeBayes was used to call variants. A consensus sequence for each strain was created using BCFtools and the FreeBayes VCFs. Only the S. cerevisiae sub-genome was retained for subsequent analysis. In addition, regions with coverage < 5 were excluded. kSNP3 was used for SNP detection and phylogenetic analysis on the set of assemblies from Gallone et al. 2016 and the four consensus sequences. The resulting SNP matrix was fed into IQ-TREE to produce a maximum likelihood tree with 1000 ultrafast bootstrap approximations. While the tree is very similar to the one in the Gallone et al. 2016 paper, there are some minor differences resulting from the different methodology (they e.g. only looked at SNPs located in the coding sequence of genes, while this approach looks at the whole genome).

My previous post contains information and discussion about how the White Labs strain identification was done.

DISCLAIMER: These are only guesses based on a range of evidence.

Decoding some White Labs strains from the Gallone et al. 2016 paper

NOTE: This post has been updated.

The paper on brewing yeast domestication authored by the Verstrepen lab and White Labs last year is a fantastic piece of work (i.e. Domestication and Divergence of Saccharomyces cerevisiae Beer Yeasts, Gallone et al. 2016, Cell). I’m sure most of you have seen it by now, so I won’t be going into any details about it. While the article and the work that has been done is amazing, there is one big negative about it: the strains used in the study have been coded, and there is no way of knowing what strain in the paper corresponds to what strain from White Labs or the Verstrepen lab. This means that the readers can’t really benefit from the huge amount of phenotypic data that the study generated. Wouldn’t be nice, for example, to know which White Labs strains are POF+, which ones can’t use maltotriose, which ones produce high concentrations of isoamyl acetate, or which ones sporulate easily?

There is one way of finding out what the coded strains are though, but unfortunately it isn’t completely straight-forward and at the moment it can only be done with a handful of strains. The authors have released the genome assemblies for each of the strains used in the study (available here: https://www.ncbi.nlm.nih.gov/bioproject/323691). The genomes of some White Labs strains have been sequenced in other studies, and performing a search for ‘WLP*’ in the NCBI database yields some hits (https://www.ncbi.nlm.nih.gov/biosample/?term=WLP*). Most of these Illumina reads are from a recent paper focusing on wine yeasts by Borneman et al. 2016. Luckily the strains weren’t coded in this study, and we know what White Labs strains these sequence reads are derived from.

What I then did, was download all of the raw sequence reads for any White Labs strains I could find. I then aligned them to a brewing yeast reference genome (VTT-A81062). After alignment, I looked for SNPs with FreeBayes, and used the resulting VCF files to create consensus sequences for each White Labs strain using BCFtools. I then used ParSNP to perform core genome alignment on these consensus sequences and all 157 of the assemblies from the Gallone et al. paper that I had obtained previously. ParSNP outputs a core genome phylogeny (generated with FastTree2) and a SNP matrix. I additionally produced a maximum likelihood phylogenetic tree using the SNP matrix and ExaML. To identify the White Labs strains from the coded strains in the Gallone et al. paper, I then looked for the closest hits in the phylogenetic tree (this was really obvious for most strains). In cases when there wasn’t one obvious hit, I also looked at Supplementary Table S1 in Gallone et al. 2016 for the reported origin and source of each strain.

Using this approach, I was able to identify eight White Labs strains from the set of coded strains. I know it is not much, but at least it is a good start, and it already gives some interesting and valuable information. First of all here are the results:

WLP_strainsThe question mark (?) after the ‘Code in paper’ means that there were two close hits, and I chose the match based on the reported origin and source as well. First of all, my suspicions regarding WLP099 were true (see this old post), with it being Beer033, a yeast grouping in the wine clade and unable to use maltotriose. Interestingly, WLP570 (Beer085) also seems to be unable to use maltotriose. This should be the ‘Duvel’ strain, and based on this information, it should only be able to produce bone-dry beers in worts supplemented with sugars (as Duvel is, if I’ve understood correctly). Another interesting observation we can make is that WLP705 appears to be Sake002. This is the only out of the seven Sake strains that wasn’t grouped in the Asian clade, but rather in the wine clade. Is this really a Sake yeast, or is it a mislabeled wine yeast? Anyways, there is a lot of interesting data to be extracted from these strains alone (esters, ethanol tolerance, etc.)!

While writing this blog post, I also noticed that the authors have released the Illumina reads to all the strains as well (in June 2017 apparently). This should allow me to confirm these results by looking at things such as chromosome copy numbers (through coverage) and SNP heterozygosity (the assemblies are haploid sequences). Unfortunately, I probably won’t have time to do that anytime soon though. But hopefully I can have a look at those later during the winter.

I will also be keeping an eye on the NCBI databases in case more White Labs strains are sequenced in the future!

Some of our recent research

I feel bad for again having to start off a post by apologizing for my inactivity here at the blog. It has been quite a busy summer for me, as my wife and I got our first child (his name is Ludwig) in the end of June. I’m currently on parental leave from my research work, so have also been busy with wrapping up my ongoing work, so I can continue with it when I return from my leave in March. Anyways, we have been working on lots of interesting hybrid-related projects at our lab, so I thought I’d briefly sum up some of the topics we’ve been working on:

Adaptive evolution of newly created lager hybrids

This is the project I’ve been dedicating most of my time to the past 18 months. While the results haven’t been published in a peer-reviewed journal yet, we’ve submitted a manuscript based on the results for review, and have uploaded a pre-print to bioRxiv for everyone to read. We and others before us, have noticed that many hybrids (particularly interspecies ones) tend to have quite unstable genomes. This means that their genomes, along with their phenotypic properties, can change as the hybrid is grown for several generations. This is not very desirable in a brewing environment, where yeast is reused for multiple consecutive fermentations. We decided to exploit this instability, by growing (and adapting) some of our hybrids in media containing high ethanol concentrations for 130+ generations. We isolated colonies of variant strains along the adaptation process, and through a screening step, we managed to select several variant strains that outperformed the original hybrids in wort fermentations. Many of these variants also had higher ethanol tolerance than the original hybrids, and produced a more desirable aroma profile (more esters, less higher alcohols, and less diacetyl). We sequenced the genomes of these variants, and noticed that all variants had undergone gains and losses of various chromosomes. The results seemed to suggest that the S. cerevisiae sub-genome in the hybrids was ‘preferred’, while the S. eubayanus sub-genome had undergone more losses. We also identified multiple single nucleotide polymorphisms and small indels which affected the coding sequence of many genes. Mutations in some of these genes had previously been reported to affect ethanol tolerance and general fitness. In addition, these variant strains appeared stable when grown for multiple generations. All in all, our study shows that interspecific hybridization, coupled with adaptive evolution, is a powerful tool for producing new strains with tailor-designed properties.

Here is a link to the pre-print: https://www.biorxiv.org/content/early/2017/10/16/204198

Alternative Saccharomyces interspecific hybrid combinations for lager fermentations

This is a project a MSc student, Jarkko Nikulin, worked on last year. We wanted to investigate whether S. eubayanus really is that important for lager (i.e. low-temperature) fermentations. Studies have shown that there are several other cold-tolerant Saccharomyces species (though not as cold-tolerant as S. eubayanus), and the questions was: can we replace S. eubayanus in a lager hybrid with one of these alternative Saccharomyces species? We generated a range of S. cerevisiae × (S. arboricola / S. eubayanus / S. mikatae / S. uvarum) hybrids and compared them in wort fermentations. We noticed that S. eubayanus seems to be dispensable, as many of the other alternative hybrids performed just as well in low-temperature wort fermentations. There is a quite a lot of potential to generate some really diverse and unique strains using these alternative Saccharomyces species.

Here is a link to the study: http://onlinelibrary.wiley.com/doi/10.1002/yea.3246/full/

The use of Saccharomyces eubayanus and its hybrids in wine-making

This is a project my colleague and fellow PhD student Frederico Magalhaes has been working on. In his PhD project he has been looking at the use of S. eubayanus for cider and wine-making. Many wine fermentations are carried out at lower temperatures, e.g. to increase aromatic complexity and decrease the growth of contaminants, and the use of S. eubayanus would allow for that. You can read more about Frederico’s work in the following blog post:

https://vttindustrialbiotechnology.com/2017/08/30/some-like-it-cold-vtts-cold-tolerant-yeast-strains-for-wine-and-cider-production/

Here is also a link to a published study: https://doi.org/10.1093/femsyr/fox049

Review on modern yeast development strategies

Finally, here is a review on modern brewing yeast design and development that I was involved in, and was written for the ISSY33 conference that took place this summer in Ireland. Some of the topics that are covered are non-conventional brewing yeast, hybridization and genome analysis.

Here is a link to the review: https://academic.oup.com/femsyr/article/17/4/fox038/3861261

Cool things that can be done with allotetraploid interspecific hybrids

The title of this post may sound a little confusing or complex for those not familiar with the nomenclature of the field, but I’ll try my best to explain our recent results to you. As I’ve written many times previously on this blog ([1], [2], [3], [4], and [5]), I have been working with and researching yeast hybrids and their use in brewing for my PhD thesis. I’ve been focusing especially on lager yeast hybrids, i.e. Saccharomyces cerevisiae × Saccharomyces eubayanus. Research, both our own and in other labs, has shown that generating new lager yeast hybrids by breeding strains of S. cerevisiae with S. eubayanus is possible, and that these new hybrids often possess desirable properties compared to the parent strains. These properties include improved fermentation rate, better stress tolerance and a more diverse formation of aroma compounds. However, one trait that has been plaguing all de novo lager yeast hybrids, is their inherent tendency to produce phenolic off-flavours (POF, i.e. the spicy, clove-like aroma that is characteristic of wheat beers and Belgian-style beers). This is an undesirable trait that is inherited from the Saccharomyces eubayanus parent. It doesn’t matter if the other parent produces phenolic off-flavours or not, the de novo lager hybrid will always produce it, as it inherits the relevant genes (more on this below) from the S. eubayanus parent. So with this in mind, and the idea that it would be cool if you could combine properties from more than two parent strains into a hybrid, we started thinking and experimenting.

One way that it is possible to remove unwanted traits from a yeast strain, is through sporulation and meiotic recombination (note, this only works if the strain is heterozygous at the loci responsible for the trait). Sporulation would also give us the opportunity to add in features from a third parent strain, as one could mate the hybrid spore clone (having a single mating type) with a third parent strain. However, interspecies hybrids tend to be sterile (this is also true for animal hybrids such as mules or ligers), which means that they don’t form any viable spores. Therefore, one would think that this is an impossible route to go down. What is interesting though, is that studies have shown that sterility mainly afflicts allodiploid hybrids (i.e. the hybrid has inherited one copy of each chromosome from both parents, just like us humans), and that allotetraploid hybrids (i.e. the hybrid has inherited both copies of each chromosome from both parents) tend to remain fertile. With this in mind, we constructed a set of five hybrids from three parents. Two of these hybrids contained DNA from all three parent strains, and from one of these two hybrids we successfully removed POF formation. These hybrids were constructed with fertile allotetraploid intermediates (which were capable of efficient sporulation) using the following scheme:

Figure_1_trans

Before I get into the properties of this set of eight strains, I’ll quickly go through the cause of the ‘POF phenotype’. Many strains of Saccharomyces produce vinyl phenols from hydroxycinnamic acids, and these phenolic compounds are considered undesirable in many beer types (especially lager beer). The most well-studied of these vinyl phenols is 4-vinyl guaiacol, which is formed from ferulic acid. We know, thanks to work by Mukai et al., that the ability of brewing yeast to produce volatile phenols is attributed to the adjacent PAD1 and FDC1 genes, both of which are needed for a POF+ phenotype. Wild yeast strains, such as S. eubayanus, tend to have functional PAD1 and FDC1 genes, while domesticated POF- brewing yeast have nonsense or frameshift mutations in these genes, making them non-functional (e.g. see the recent Gallone et al. paper in Cell). The biological function of this mechanism is to protect the cell from the toxic effects of hydroxycinnamic acids (mainly plants produce them as antibiotics), which is why one can expect to find only POF- strains among domesticated yeasts. So if any non-functional alleles of PAD1 or FDC1 are present in the hybrid genome, it should be possible to remove the POF+ phenotype through meiotic recombination (as we demonstrate).

ferulic_4vgFor this set of eight strains, we began with three parent strains (P1-P3). Two of them are S. cerevisiae ale strains (P1: VTT-A81062, P2: WLP099) and one is the S. eubayanus type strain (P3: VTT-C12902). These strains were chosen for their varying properties. Of the three, P1 is the only strain that is able to use maltotriose during fermentation, P2 is the only strain that does not produce 4-vinyl guaiacol (i.e. it is POF-), while P3 is the cold-tolerant parent strain of lager yeast. We began by creating the three possible double hybrids (P1 × P2, P1 × P3, P2 × P3) through rare mating. Hybrid H1 (P1 × P3) was fertile (thanks to allotetraploidy), despite it being an interspecific hybrid, and it produced viable spores efficiently. So we sporulated Hybrid H1 and then mated a mixture of its spores with parent strain P2 to obtain the triple Hybrid T1 ((P1 × P3) × P2). Whole genome sequencing revealed that the sub-genome ratio in T1 was approximately 1:2:1 (P1:P2:P3). Hybrid T1 was also an allotetraploid and it was able to form viable spores. Knowing that the genome contained both functional and non-functional alleles of PAD1, we attempted to remove the POF+ phenotype from Hybrid T1 through sporulation according to the image below. What we did was sporulate Hybrid T1, isolate individual spore clones, and then screen them for the POF phenotype in media containing ferulic acid. POF+ spore clones would produce a strong clove-like aroma in the media, while POF- spore clones would not. Approximately 25% of the spore clones were POF− (as one would expect), and the best growing of these was given the name Hybrid T2, i.e. a POF− meiotic segregant of Hybrid T1.

sporulation_4vg

We wanted to compare how these 8 brewing strains would perform in wort fermentations at lager brewing conditions (15 °P all-malt wort at 15 °C). There was considerable variation in fermentation performance between the eight strains, as you can see in the figure below (A). Of the 8 strains, Hybrid T1 and S. cerevisiae P2 had the highest overall fermentation rates, but these slowed down considerably after reaching 5.8% (v/v) alcohol. We looked at the sugars present in the beer, and saw that S. cerevisiae P2 was unable to ferment the maltotriose in the wort (D). Hybrid T1 had only consumed a small amount of the initial maltotriose present in the wort (D). Of the 8 strains, Hybrid T2 attenuated the best, followed closely by Hybrids H1 and H3 (A). These three hybrids had also used maltotriose efficiently (D). The beers produced with the 8 brewing strains also varied considerably in concentrations of aroma-active compounds (B). The most ester-rich (i.e. fruity) beers were produced with Hybrids H1, H3 and T1. When we compare the aroma profiles of the beers made with Hybrids T1 and T2, we see that the meiotic segregant T2 produced lower concentrations of most esters, while its beer contained higher concentrations of most higher alcohols. This could maybe be explained by it having lower activities or expression of alcohol acetyl transferases (e.g. ATF1 and ATF2). Of the 8 strains, the POF− S. cerevisiae P2 parent strain and Hybrid T2 were the only ones that did not produce any detectable amounts of 4-vinyl guaiacol (detection limit 0.2 mg L−1), thus confirming their POF− phenotype (C). All other strains produced 4-vinyl guaiacol in concentrations above the flavour threshold of 0.3–0.5 mg L−1. We finally compared the sequences of PAD1 and FDC1 in the 8strains, and found that Hybrid T2 only carried the PAD1 allele that was derived from S. cerevisiae P2 (E). This particular allele, contained a possible loss-of-function SNP at position 638 (A>G, resulting in an amino acid substitution of aspartate to glycine). The other strains, including Hybrid T1, which Hybrid T2 was derived from, carried either or both of the functional PAD1 alleles derived from S. cerevisiae P1 or S. eubayanus P3.

 

Figure_8

So there you have it, some cool things you can do with allotetraploid interspecific hybrids. We wanted to demonstrate that it is possible to construct complex yeast hybrids that possess traits that are relevant to industrial lager beer fermentation and that are derived from several parent strains. If you are interested in some more info on the topic (e.g. some lipidomics analysis of these strains), I’m happy to announce that an article based on these results was recently published in Microbial Cell Factories:

 

Inheritance of brewing-relevant phenotypes in constructed Saccharomyces cerevisiae × Saccharomyces eubayanus hybrids

Kristoffer Krogerus, Tuulikki Seppänen-Laakso, Sandra Castillo, Brian Gibson

Microbial Cell Factories, 2017, 16:66. DOI:10.1186/s12934-017-0679-8

 

Please feel free to check it out (it is open access)! Here is the abstract:

Abstract

Background

Interspecific hybridization has proven to be a potentially valuable technique for generating de novo lager yeast strains that possess diverse and improved traits compared to their parent strains. To further enhance the value of hybridization for strain development, it would be desirable to combine phenotypic traits from more than two parent strains, as well as remove unwanted traits from hybrids. One such trait, that has limited the industrial use of de novo lager yeast hybrids, is their inherent tendency to produce phenolic off-flavours; an undesirable trait inherited from the Saccharomyces eubayanus parent. Trait removal and the addition of traits from a third strain could be achieved through sporulation and meiotic recombination or further mating. However, interspecies hybrids tend to be sterile, which impedes this opportunity.

Results

Here we generated a set of five hybrids from three different parent strains, two of which contained DNA from all three parent strains. These hybrids were constructed with fertile allotetraploid intermediates, which were capable of efficient sporulation. We used these eight brewing strains to examine two brewing-relevant phenotypes: stress tolerance and phenolic off-flavour formation. Lipidomics and multivariate analysis revealed links between several lipid species and the ability to ferment in low temperatures and high ethanol concentrations. Unsaturated fatty acids, such as oleic acid, and ergosterol were shown to positively influence growth at high ethanol concentrations. The ability to produce phenolic off-flavours was also successfully removed from one of the hybrids, Hybrid T2, through meiotic segregation. The potential application of these strains in industrial fermentations was demonstrated in wort fermentations, which revealed that the meiotic segregant Hybrid T2 not only didn’t produce any phenolic off-flavours, but also reached the highest ethanol concentration and consumed the most maltotriose.

Conclusions

Our study demonstrates the possibility of constructing complex yeast hybrids that possess traits that are relevant to industrial lager beer fermentation and that are derived from several parent strains. Yeast lipid composition was also shown to have a central role in determining ethanol and cold tolerance in brewing strains.