Benchtop micro-mashing: high-throughput, robust, experimental beer brewing

Brewing science is undergoing a renaissance with the use of modern analytical chemistry and microbiology techniques. However, these modern analytical tools and techniques are not necessarily aligned with the scale and scope of brewing science. In particular, brewing processes can be time consuming, ingredient intensive, and require specialised technical equipment. These drawbacks compound with the need for appropriate numbers of replicates for adequately powered experimental design. Here, we describe a micro-scale mash method that can be performed using a common laboratory benchtop shaker/incubator, allowing for high throughput mashing and easy sample replication for statistical analysis. Proteomic profiles at both the protein and peptide levels were consistent between the 1 mL micro-mash and a 23 L Braumeister mash, and both mash scales produced wort with equivalent fermentable sugar and free amino acid profiles. The experimental flexibility offered by our micro-mash method allowed us to investigate the effects of altered mash parameters on the beer brewing proteome.


Introduction
Beer brewing is one of the oldest known biotechnologies and has heavily influenced society and science. The ancient science of beer brewing is rapidly developing, with modern analytical techniques allowing deeper understanding of fundamental aspects of its integrated bioprocesses [1][2][3][4][5][6][7][8][9][10] . The brewing process begins with harvested barley seeds being malted through a process which involves controlled, partial germination followed by kilning 11,12 . The malt is then milled to open the husks and combined with warm water in a process known as mashing.
Mashing extracts the soluble sugars and proteins from the malt which are digested by enzymes into smaller sugars and free amino nitrogen (FAN) 11 . Wort, the soluble fraction of the mash, is then boiled with hops to introduce bitterness and aromatic flavours, remove volatile offflavours, stop enzymatic activity, and sterilize 11,13 . Fermentation is performed by pitching yeast into cooled wort, where the yeast consume the sugars and FAN and produce flavour compounds, carbon dioxide, and ethanol. Finally, beer is filtered, aged, carbonated, and packaged.
Complex and dynamic protein and carbohydrate chemistry occur during mashing and the boil 1 .
These processes are very well understood from an industrial standpoint, but the underlying biochemical details are only starting to be appreciated 11,12 . Mashing extracts the soluble component from malted barley seeds, primarily starch and proteins, including enzymes 11 .
Extracted starch and proteins are then broken down by enzymes into small fermentable sugars and FAN, respectively 11 . Proteomics has been used to identify and monitor changes in the abundance of these enzymes and other proteins during the mash and boil and into the final beer product 1,[4][5][6][7][8][9][10] . These studies found that while some proteins were heat-stable and could survive being boiled, other proteins were only active and stable at low temperatures, while many proteins were covalently modified by a variety of processes during brewing 1,[4][5][6][7][8][9][10]14 .
Brewing processes at the scale of a large industrial brewery or even a smaller craft brewery are time consuming, ingredient intensive, and require specialised technical equipment not typically found in research laboratories. This large scale makes it difficult and expensive to perform appropriately designed experiments with sufficient statistical power to advise the brewing process.
Our previous investigation of the beer brewing proteome identified and measured the abundance of hundreds of proteins during the mash, and found dynamic changes in protein abundance due to solubilisation, proteolysis, and degradation 1 . Here, we re-interrogate these datasets from a 23 L mash and boil and a 1 mL benchtop micro-mash 1 . We investigate the two datasets in more detail to test the reliability and comparability of a 1 mL benchtop micro-mash to large scale brewing. We present a micro-scale mash protocol that can be performed using a benchtop shaker incubator, allowing for high-throughput mashing and easy sample replication for statistical analysis (Fig. 1).

Protein abundance profiles are similar in 23 L Braumeister and 1 mL benchtop mashes
We interrogated the abundance profile of each individual protein from a 23 L Braumeister mash and a 1 mL benchtop mash ( Fig. 2A). These protein abundance profiles showed that at both mash scales proteins increased in abundance as temperature increased and proteins were solubilised. As temperature continued to increase, the temperature-sensitive proteins reduced in abundance due to denaturation, aggregation, and precipitation ( Fig. 2A). Some subtle differences are apparent between the 23 L and 1 mL mash scales, with protein extraction being slightly faster in the 1 mL micro-scale mash than in the 23 L Braumeister mash ( Fig. 2A).
However, the profile of a given protein was generally very similar at both 23 L and 1 mL mash scales, while the profiles for different proteins were distinct.
We next directly compared the performance of the two mash scales by calculating the difference in normalized abundance, subtracting the normalized abundance of each protein at each stage between the two mashes ( Fig. 2B and Supplementary Table S1). This difference map was consistent with the overall protein abundance profiles ( Fig. 2A) and demonstrated that proteins were more efficiently extracted in the 1 mL micro-scale mash, with increased levels at the end (35 min) of the 63 °C rest, although these differences were not statistically significant (Fig. 2B). This difference map also demonstrated that the loss of protein at higher temperatures was less efficient in the 1 mL micro-scale mash than in the 23 L Braumeister mash (Fig. 2B). This difference was particularly apparent during the boil. Regardless of these minor differences in the rates of extraction and denaturation, these side-by-side and difference comparisons of the two mash scales showed their protein-specific abundance profiles were very similar overall, with very few statistically significant differences (Fig. 2).

Proteolytically cleaved forms of peptides show similar profiles in a 23 L Braumeister mash and a 1 mL benchtop mash
Proteolysis of barley seed storage proteins releases FAN for yeast to consume during fermentation and has also been shown to be a key driver of protein stability during mashing 1,11 .
Barley seeds contain proteases which can proteolytically clip barley seed storage proteins. We previously found that in a 23 L Braumeister mash, the proteolytically clipped forms of proteins tend to be less stable at higher temperatures in the mash, which results in loss of these nontryptic peptides and an increase in the relative abundance of the full-length tryptic peptides 1 .
However, this effect is site-specific, as some proteolytically clipped forms show high stability.
We therefore compared this key feature of the dynamic mashing proteome at 23 L Braumeister and 1 mL micro mash scales. This analysis showed that the abundance of site-specific proteolysis products was very similar between the mash scales, with only slight differences detected at lower temperatures early in the mash ( Fig. 3A and B).
For proteolytic forms of the peptide family V122LVTPGQCNVLTVHNAPYCLGLDI145 from α-amylase/trypsin inhibitor CMa (IAAA) we found almost identical peptide abundance profiles between the two mash scales, with few statistically significant differences ( Fig. 3 and Supplementary Table S2). The semi-tryptic forms V122LVTPGQCNVLTVH135 and V122LVTPGQCNVL132 were slightly increased in abundance at lower temperatures in the 1 mL mash compared to the 23 L Braumeister mash, but these differences converged at higher temperatures later in the mash (Fig. 3A). We also found a high level of similarity in peptide abundance profiles between the two mash scales for proteolytic forms from the peptide family K-C99NVNVPYTISPDIDCSR115-I from non-specific lipid-transfer protein 1 (NLTP1) (Fig.   3B). Full tryptic peptide K-C99NVNVPYTISPDIDCSR115-I and the semi-tryptic form N100VNVPYTISPDIDCSR115-I were slightly higher in abundance early in the 1 mL micromash, but again these differences decreased as the mash progressed (Fig. 3B). At both mash scales, both of these peptide groups showed equivalent overall profiles when comparing the dynamics of proteolysis and the resulting protein stability across the mash and boil (Fig. 3). distinction between time-and temperature-dependent changes in protein and peptide abundance (Fig. 4). This can be coupled with statistical analysis to test whether these changes are indeed time-or temperature-dependent (Fig. 4).  To determine if protein changes during the mash were time-or temperature-dependent, we previously performed a micro-scale mash that included extended incubation periods at each temperature stage 1 . This analysis showed that the cause of changes in abundance (time-or temperature-dependent) could be established for both NLTP1 and IAAA using these temperature stage extensions (Fig. 5).

Discovering if changes to peptide abundance are time-or temperature-dependent using mash stage extensions
Proteolysis is a key post translational modification in brewing, driving FAN production and controlling protein stability 11 . We previously found that many physiological proteolysis events could be measured during the mash, and that the relative abundance of the corresponding differently proteolytically clipped protein forms changed over the course of the mash 1 . The observed changes were consistent with limited proteolysis occurring in malting or early in the mash, and the subsequent proteolytic clipping affecting protein stability by altering the temperature at which the protein unfolded, aggregated, and was lost from solution. Most proteolytic events lowered protein stability, while a few had no effect or even increased stability. To investigate the mechanisms underlying these effects in more detail, we studied the effect of the temperature stage extensions on the abundance of peptides resulting from physiological proteolysis. For two highly abundant peptide families, V122LVTPGQCNVLTVHNAPYCLGLDI145 from IAAA and K-C99NVNVPYTISPDIDCSR115-I from NLTP1, the relative abundance of each peptide form was statistically compared between stages of the mash ( Fig. 6C and F, and Supplementary Tables   S3 and S4).
For proteolytic forms of the peptide family V122LVTPGQCNVLTVHNAPYCLGLDI145, we observed that some smaller semi-tryptic forms were more abundant than the full tryptic peptide at lower temperatures (52 °C and 63 °C), and that at higher temperatures two semi-tryptic forms increased in relative abundance ( Fig. 6A and B), as previously described 1 . We found that for the semi-tryptic peptide V122LVTPGQCN, the decrease in abundance after the 63 °C rest was time-dependent ( Fig. 6A, B, and C). The slightly larger form V122LVTPGQCNL increased in relative abundance at 73 °C, which was temperature-dependent ( Fig. 6A, B, and C). At the 78 °C rest, a sharp decrease in abundance of V122LVTPGQCNL occurred, which was timedependent ( Fig. 6A, B, and C). Finally, during the 78 °C rest the relative abundance of bothV122LVTPGQCNVLTVHNAPYCLGLD and V122LVTPGQCNVLTVH increased (Fig.   6A, B, and C). The increase of V122LVTPGQCNVLTVHNAPYCLGLD was temperaturedependent, while the increase of V122LVTPGQCNVLTVH was time-dependent ( Fig. 6A, B, and C). For proteolytic forms of the peptide family C99NVNVPYTISPDIDCSR115-I, we observed that smaller semi-tryptic forms, predominantly ISPDIDCSR115-I. were more abundant than the full tryptic peptide at lower temperatures (52 °C and 63 °C), and that at higher temperatures the full tryptic peptide was the predominant form ( Fig. 6D and E), as previously described 1 . We found that for the semi-tryptic peptide ISPDIDCSR115-I, the increase in abundance after the 52 °C rest was temperature-dependent (Fig. 6D, E, and F). Following this, ISPDIDCSR115-I increased in relative abundance at 63 °C, which was temperature-dependent (Fig. 6D, E, and   F). The full tryptic form C99NVNVPYTISPDIDCSR115-I appeared to increase in abundance at the 73 °C rest, but no comparison was significant (Fig. 6D, E, and F).

Wort metabolomic analysis
The micro-mash system closely mirrors the performance of a 23 L Braumeister mash when considering the dynamic proteome of the mash and boil. As well as the proteome, other key parameters relevant to mashing performance are the specific gravity, FAN or soluble amino acid levels, and sugar profiles of the final wort produced by the mashing process. We compared these parameters for the 1 L micro-scale and 23 L Braumeister mashes. Multiple Reaction Monitoring (MRM) LC-MS/MS analysis showed that the fermentable sugar profiles after the mash and boil using the two different mash methods were very similar, with no significant differences (Fig. 7A), while the wort from the 1 mL micro-scale mash had higher specific gravity (Fig. 7B). Finally, we used LC-MS/MS to measure 15 free amino acids, which showed that wort from the 1 mL micro-scale mash contained very similar concentrations of free amino acids as the 23 L Braumeister mash, with no significant differences (Fig 7C).

Discussion
We have demonstrated that a benchtop micro-mashing method can be used in lieu of standard industrial-scale brewing equipment to investigate the molecular dynamics of the brewing process. Specifically, a 1 mL micro-scale mash was not only equivalent to a full 23 L Braumeister mash but allowed increased throughput for statistical analyses. We showed that the protein abundance profiles of the 23 L Braumeister and 1 mL micro-scale mashes were highly similar (Fig. 2). The main difference between the mash scales was an increased protein abundance earlier in the 1 mL micro-scale mash compared to the 23 L Braumeister mash, consistent with rapid extraction and solubilization of proteins in the micro-mash. (Fig. 2). We found that the abundance of proteolysis products in the form of semi-tryptic peptides and their full-length tryptic counterparts were very similar between the two mashes. There were only small differences between mash scales, with proteolysis products being more abundant at the lower temperatures in the micro-mash (Fig. 3).
The temperature stage extensions coupled with statistical tests provided a model for statistically determining if changes to protein or peptide abundance during the mash were temperatureand/or time-dependent 1 . At the protein level, stage extensions provided a simple tool to determine the cause of changes in protein abundance (Fig. 5). When investigating peptide abundance and proteolysis, temperature controls provided insight into how proteolysis affected peptide abundance and protein stability (Fig. 6). Our results here showed that coupling mash stage extensions with replicate analyses and statistical tests allows for insight into the environmental and process causes of changes in the abundance and modification of proteins.
Our data showed that protein and proteolysis biochemistry in the mash is highly complex, and tools such as this micro-mash method will allow future interrogation of this complex biochemistry.
We found that the 1 mL micro-scale and 23 L Braumeister mashes produced wort with very similar concentrations of fermentable sugars and free amino acids. Glucose, maltose, and maltotriose as well as all the measured amino acids had equivalent concentrations at both mash scales (Fig. 7). The final specific gravity was higher in the 1 mL mico-scale mash (Fig. 7), possibly because of more efficient solubilization of starch or maltooligosacharides at this scale.
The micro-mash method we describe here provides a high-throughput and low-cost tool to study brewing biochemistry. The temperature extensions made possible by the micro-scale and high-throughput system provide a novel and highly practical tool to systematically unravel changes in the abundance of proteins and their post-translational modifications. This micromash system will also allow high-throughput analysis of barley varieties that are being accredited or analysed commercially.

Braumeister Mash and Benchtop Micro-Mash
Previously published 1 mass spectrometry proteomics datasets PXD010013 and PXD013177 were retrieved from the ProteomeXchange Consortium via the PRIDE partner repository 15 . The mash methods that generated the datasets were previously described 1

Sample Preparation and Mass Spectrometry
Sample preparation for previously published 1 mass spectrometry proteomics datasets PXD010013 and PXD013177 included methanol/acetone precipitation of proteins from 10 µL or 100 µL of each sample from Braumeister-or micro-scale mashes, respectively, followed by digestion with trypsin in the presence of 10 mM dithiothreitol. After desalting with C18

ZipTips, peptides were measured with Data Dependent Acquisition and Data Independent
Acquisition LC-ESI-MS/MS as previously described 1,16 .

Sugar and Amino Acid Quantification by MRM LC-MS/MS
Samples were filtered with a 30 kDa Amicon Ultra centrifuge filter unit (Millipore) and diluted 1:1000 with H2O. Sugars and amino acids were measured by MRM with UPLC-MS/MS using a Nexera ultraperformance LC system (Shimadzu) coupled to a QTRAP 5500 mass spectrometer (SCIEX). Chromatography was performed with a Kinetex 1.7 µm C18 100 mm