Spontaneous fermentation and biogenic amines

Biogenic amines (BAs) are a class of biologically-produced compounds that are found in plant and animal products such as cheeses, cured meats, and other fermented foods. The names of some specific BAs may be familiar, for example histamine, and others may illustrate the sorts of characteristics some BAs have (strong bad smells), for example putrescine and cadaverine. As the names of these latter two suggest, some BAs are associated with rotting material. In higher levels BAs can have health impacts, but I won't be discussing any of that. If you want more health-related information on this, check out this great complimentary post by Bryan at Sui Generis Brewing.

In beer, BAs can originate from raw ingredients and can also be microbially produced during fermentation. Especially in fermentations that are not pure cultures of S. cerevisiae, much of the BA content of beer can be formed during fermentation. This post will focus BAs in spontaneous beer by looking at data of BAs in lambic, with an emphasis on microbial BA production points, how this may impact the flavor and aroma of spontaneous beer, and how production can be minimized without inoculation (i.e. while maintaining a spontaneously-fermented beer). I won't go into much lambic background in this post, so if you're unfamiliar with the process and/or fermentation progression of lambic and other spontaneously fermented beers, then I recommend starting with this page on the Milk the Funk wiki and these pages (here and here) on lambic.info. You can also find all of my previous posts on lambic (history, hopping, IBUs, carbohydrates, commercial brewery/blendery visits, etc.) here.

A coolship being filled

As with my other posts on beer science, this might be a bit dense for those not so interested in science. Here is a simplified summary if the details are a bit too much and/or you're not interested in that part:

Biogenic amines seem to be produced in two different phases in spontaneously fermented beer - one from enterobacteria in the initial weeks and one from lactic acid bacteria after about 6 months. BA production in the first stage may be controlled/limited by pre-acidifying wort. Differences in BA production in the second stage may be due to differences in specific strains of lactic acid bacteria, as not all can produce BAs. There aren't enough data to be sure, but BAs produced in the first stage could impact the flavor/aroma of spontaneous beer.

Biogenic amines in lambic

The data sources

I will be primarily discussing two studies on BAs over the course of lambic fermentation - Gasarasi et al., 2003 and De Roos et al., 2018. The former followed the fermentation of un-acidified wort for 400 days, and also followed fermentation of wort which they pre-acidified with 5% of a very acidic beer. This beer was prepared from the previous brewing season by pitching wort with pure strains of lactic acid bacteria. It does not seem that this ~6 month old acidic beer was sterilized before using it in the next season at a level of 5% (contributing about 1g/L lactic acid). This second beer was followed for a total of 60 days. If this acid beer wasn't sterilized, that is unfortunate because it means the beer is no longer spontaneous. The authors also briefly mention tests with simple pre-acidifying with lactic acid, which they say resulted in a reduction of BAs, but unfortunately they do not show the data or discuss it further. The second study (De Roos et al., 2018) followed two different 660 L casks of lambic from the same brew from the time the casks were filled until they were two years old. The wort for these casks was pre-acidified to pH = 4.3. Finally I am including data from two different studies on finished lambic & geuze - Izquierdo-Pulido et al., 1996 and Loret et al., 2005.

Stage 1 - Initial days and weeks

Looking specifically at beers of spontaneous fermentation, there appear to be two periods where BAs are produced. The first point occurs early in the fermentation where enterobacteria are active, beginning in the initial day(s) after cooling and lasting for the first week(s) (e.g. Van Oevelen et al., 1977; Spitaels et al., 2014; De Roos et al., 2018). At this point in the fermentation, no significant attenuation has occurred so simple sugars are available and alcohol is absent. The pH is also higher at this early stage. During this stage the primary BAs produced are putrescine and cadaverine.

Variability is seen among available data for putrescine and cadaverine production at the start of fermentation. See, for example the differences in cadaverine and putrescine between the filled green circles (Gasarasi et al., 2003 data of lambic wort which was not pre-acidified) and the open green circles (Gasarasi et al., 2003, lambic wort which was pre-acidified). Additionally, the Gasarasi et al., 2003 un-acidified wort shows much higher putrescine than the De Roos et al. (2018) data. It should be noted that this comparison is between different batches of lambic from different years and from different producers, so there could be additional sources of variability. The difference for putrescine may be important, even if cadaverine levels are similar, as putrescine is more likely to have an impact on the taste and/or aroma of lambic at the observed concentration levels. This is discussed in more detail in the taste section below.

BAs in fermenting & packaged lambic. Data in black squares are from final products. An age was chosen
(~800 days) to include them in the figure and does not reflect their actual age (which is likely much older).

The ability of pre-acidification to limit BA levels in the initial production phase is because pre-acidification can limit and/or shorten the period in which enterobacteria are active in the wort. Both the Gasarasi et al. (2003) and De Roos et al. (2018) studies highlight the potential of wort pre-acidification to limit BA production in lambic. Gasarasi et al. (2003) note that pH > 5 and cooling to T < 15° C may result in elevated BA levels, and they say that adding 2000 mg/L lactic acid to wort reduces BAs but does not completely prevent BA production. A study on industrial lambic production where the wort was acidified to pH = 4 with lactic acid before fermentation also demonstrated that pre-acidification can eliminate the enterobacteria phase of fermentation (Spitaels et al., 2015).

A stainless steel coolship

Takeaways for brewers from the first phase

For brewers wishing to pre-acidify their spontaneous beers, these studies provide some possible starting points. Note that I haven't tried any of these myself, I'm just summarizing the advice of the papers. The specific levels mentioned here (2 g/L lactic acid in one study, pH = 4.3 in another, which took a bit more than 1 g/L lactic acid; the latter here being what one lambic producer does) may be a good starting point. Using a lower pH (pH = 4.0, as used by an industrial lambic producer) may further reduce BA production by preventing the enterobacteria step. 2 g/L lactic acid sounds like quite a bit to me, considering Van Oevelen et al. (1976) report ranges of about 2-3.5 g/L lactic acid in finished bottle fermented geuze. As a comparison, the lambic studied in De Roos et al. (2018) finished with 4-5 g/L lactic acid, and the bottled gueuze samples from Spitaels & et al., 2015 finished with about 4 g/L lactic acid or more. Depending on where the beer finishes, acidifying with 2 g/L could mean half or more of the lactic acid in the final beer is from the pre-acidification.

Finally, blending in old acidic beer around 5% (for very acidic beer) before fermentation could be a good starting point. Depending on how you feel about spontaneous fermentation, you may want to do a quick sterilization of this to prevent inoculation with microbes from the old beer. Or you may view this as beneficial for your beers, by encouraging fermentation with proven microbes, even if it may make the beer not spontaneous/less spontaneous. Finally, Gasarasi et al. (2003) note that pH > 5 and temperature < 15° C seem to favor BA production. Something else to think about when pre-acidifying is that not all metals remain inert as the pH drops (see here and here). I haven't seen data to assess what specific pH values are too low for less inert metals, but it is something you may want to look into further if you are considering this and are not using stainless to cool the wort. Or just pre-acidify after the coolship.

Another stainless steel coolship

Stage 2 - After main attenuation

A second phase of BA production can be seen in lambic. This is most evident in the tyramine and histamine data, where concentrations increase only around 100 days into fermentation. In the course of lambic fermentation, this corresponds to the acidification of the lambic by lactic acid bacteria after the main attenuation has been accomplished by Saccharomyces (e.g. Van Oevelen et al., 1977; Spitaels et al., 2014; De Roos et al., 2018). Lactic acid bacteria are known to have strain specific variability in the ability to produce BAs. Therefore, differences between producers at this stage may result from different strains of bacteria in different lambics. De Roos et al. (2018) note that there is still some uncertainty regarding this phase of BA production, as the timing of BA production in their data lagged behind the peak in Pediococcus cell concentrations during lambic acidification. This second phase does not seem to have much influence on cadaverine and putrescine levels, though the De Roos et al. (2018) data do show a slight increase in putrescine here.

Comparing with finished products

Comparing the production data and finished lambic data shows some distinct trends (see the table below). First, BAs in lambic appear to generally be higher than in other beers. Data over the course of fermentation suggest that any putrescine and cadaverine produced during the initial stages of fermentation are not appreciably removed as fermentation progresses. This means that once they are formed, they survive at more or less the same levels to the finished beer. A comparison of the histamine and tyramine data from De Roos et al. (2018) with compilation of finished products shows a similar concentration range, suggesting that the same may be true for these BAs as final blends are made and beers are packaged.

The data from Loret et al. (2005) are an average of 42 products from a total of 10 producers. These data demonstrate strong variability between producers in histamine and tyramine (also shown in their Figure 3, see also the range in values from Izquierdo-Pulido et al., 1996 below). This may be controlled by variability in specific strains of lactic acid bacteria. Comparing putrescine and cadaverine data from Izquierdo-Pulido et al. (1996), Loret et al. (2005) and the two production studies, there is also a large amount of variability here. Some of this may be explained by differences in ambient brewery resident microbes and other variability. However, the data from De Roos et al. (2018), Gasarasi et al., (2003), studies of biogenic amines in other alcohols, and data on pre-acidification of lambic and enterobacteria, it is likely that much of this can be explained by whether brewers pre-acidify the wort or not. These biogenic amines appear to be formed early in the fermentation and pre-acidifying could significantly limit their production. Finer details within the Izquierdo-Pulido et al. (1996) cadaverine and putrescine data imply an interesting feature. These data show a mean that is toward the lower end of the total observed range (cadaverine: mean = 10, range = 0.4-39.9; putrescine: mean = 6.4, range = 2.8-15.2). In order for this to happen, the data must be composed of more low values with only one or a few large values. This could reflect a few producers which do not pre-acidify and have larger values of these BAs while others do pre-acidify.

Data of BAs in lambic beer and other beers.

Can BAs impact taste in beer?

The table above shows BA levels in lambic and other beers as well as taste thresholds for some of these compounds in water. This does not directly give information on their impact in beer because water lacks the other strong tastes and smells found in beer, making the threshold for detection in beer likely higher than in water. These thresholds also do not take into account any additive and/or synergistic effects that BAs would have when found with other pungent BAs and/or other compounds in beer, potentially resulting in lower detection thresholds. But these values give a starting point for an initial assessment of the potential for BAs to be flavor-active at concentrations relevant to spontaneous beer.

I could not find data on tyramine thresholds so I won't discuss that here. The histamine data are a physiological response but not a taste or aroma. And this threshold value did not hold in triangle tests. So I think the data I've seen so far suggest that histamine does not have a strong impact on flavor/aroma at the levels found in lambic. Moving to cadaverine, it seems that the levels found in lambic are much lower than those needed to detect it. However, putrescine can be found around or above the detection threshold in water. If any BAs are going to be flavor-active, it seems that putrescine is the most likely one. And with levels in lambic potentially >2x the detection limit in water, I think it is quite possible that putrescine can make a contribution to the flavor and aroma of lambic.

Acknowledgements

While much of the data included in this post have been available for 10+ years, this is a topic that I haven't heard many people talking about. Especially not before the last few weeks. I want to thank a certainly legendary producer for bringing this to my attention a year or two ago, and for helping me to connect this scientific explanation with something I had perceived and was having trouble naming. This post was also prompted and informed by discussions with some great Milk the Funkers/brewers/writers: Dan of well-deserved Milk the Funk Wiki fame, Bryan of Sui Generis Brewing and Matt of A PhD in Beer and Patent Brewing Company.

22-Jan-19 Updates: the takeaways from the first phase section was updated about 3 hours after initially publishing the post to include context for pre-acidifying levels relative to final lactic acid in geuze and to include how much lactic acid was needed to pre-acidify to 4.3. The references were updated to incorporate new references as needed and to include links to papers at this time.

References
-De Roos et al., 2018. Wort substrate consumption and metabolite production during lambic beer fermentation and maturation explain the successive growth of specific bacterial and yeast species. Front. in Microbiol. 9:2763. (doi: 10.3389/fmicb.2018.02763).
-Izquierdo-Pulido et al., 1996. Biogenic amines in European beers. J. Agric. Food Chem. 44(10) 3159.3163. (doi: 10.1021/jf960155j).
-Gasarasi et al., 2003. Occurence of biogenic amines in beer: causes and proposals of remedies. Monatsschrift für Brauwissenschaft. 56(3) 58-63.
-Loret et al., 2005. Levels of biogenic amines as a measure of the quality of the beer fermentation process:Data from Belgian samples. Food Chem. 89(4) 519-525. (doi: 10.1016/j.foodchem.2004.03.010).
-Rohn et al., 2005. Can histamine be tasted in wine? Inflam. Res. 54(S1) S66-67. (doi: 10.1007/s00011-004-0439-x).
-Romero et al., 2003. The influence of the brewing process on the formation of biogenic amines in beer. Anal. Bioanal. Chem. 376(2). 162-167. (doi: 10.1007/s00216-003-1885-2).
-Spitaels et al., 2014. The microbial diversity of traditional spontaneously fermented lambic beer. PLoSONE 9(4) e95384 (doi: 10.1371/journal.pone.0095384).
-Spitaels et al., 2015.  The microbial diversity of an industrially produced lambic beer shares members of a traditionally produced one and reveals a core microbiota for lambic beer fermentation. Food Microbiol. 9 23-32. (doi: 10.1016/j.fm.2015.01.008).
-Spitaels & Van Kerrebroeck et al., 2015. Microbiota and metabolites of aged bottled gueuze beers converge to the same composition. Food Microbiol. 47 1-11 (doi: 10.1016/j.fm.2014.10.004).
-Van Oevelen et al., 1976. Synthesis of aroma components during the spontaneous fermentation of lambic and gueuze. J. Inst. Brew. 82 322-326. (doi: 10.1002/j.2050-0416.1975.tb06953.x).
- Van Oevelen et al., 1977. Microbiological aspects of spontaneous wort fermentation in the production of lambic and gueuze. J. Inst. Brew. 83(6) 356-360. (doi: 10.1002/j.2050-0416.1977.tb03825.x).
-Wang et al., 1975. Apparent odor thresholds of polyamines in water and 2% soybean flour dispersions. J. Food. Sci. 40 274-276. (doi: 10.1111/j.1365-2621.1975.tb02181.x).