Enhanced winemaking outcomes and wine style diversification through provision of fit-for-purpose yeast starter cultures
Abstract
While it is common practice for winemakers to choose yeast starter cultures based on their perceived impact on wine style, as much as on their capacity to reliably complete fermentation, this choice has generally been guided by anecdotal evidence rather than objective data. In this project, replicated small-scale fermentations and pilot-scale winemaking revealed the extent to which existing yeast strains can influence composition and sensory properties of red wine. Through correlation to genome sequences, genetic markers were identified for key flavour compounds. Finally, novel Saccharomyces interspecies hybrids were generated along with variants of AWRI796 that impart ‘rose’ aromas in wine.
Summary
Once grapes have been harvested and crushed, Australian winemakers are confronted by the need to choose from more than one hundred commercially available yeast strains to start their fermentations. For some winemakers the choice is straightforward, they base their decision solely upon reliability of fermentation completion and may in fact use the same yeast strain for all wine styles. Others make use of yeast germplasm resources to accentuate desired flavours and minimise the production of off-flavours, choosing strains that are ‘fit-for-purpose’ considering the wine style they are seeking to achieve. To date there has been a lack of objective data winemakers could draw upon to make this choice, typically limited to technical information for a single supplier’s strain portfolio, or scientific studies involving a relatively small number of strains. This project was designed to generate knowledge concerning the potential flavour impacts various commercially-available strains may have on wine style, in parallel with work to better understand their fermentation performance and reliability (Project AWR1302). In other words, the aim was to better define what purpose available strains are fit for. The data sets generated were then used to guide development of novel yeast strains that offer Australian winemakers unique opportunities to craft new wine styles and achieve their stylistic goals.
Outcome 1: Knowledge of potential value that can be derived from available yeast genetic diversity to shape wine flavour profile and texture.
In previous work at the AWRI, the potential for yeast strain to shape white wine style had been demonstrated for a small number of strains chosen based upon anecdotal evidence. To provide a more systematic dataset informing winemakers on choice of yeast, for this study strains were selected based on genome sequencing performed in Project AWR1302. This ensured that available genetic diversity was represented across an even split of commercially available and Australian winery isolates.
Small-scale replicated laboratory fermentations were performed using 94 yeast strains in a standardised model grape juice and a Shiraz juice. The latter was modified first-run juice to which commercially available grape phenolic extracts were added to mimic the phenolic composition of a complete Shiraz must. This enabled comparison of flavour compound production in real juice to that observed in a model system, and also exploration of the impact of yeast strain on phenolic composition of red wine.
The results show that model grape juice is suitable for evaluating production of many key yeast-derived volatile aroma compounds, though for a subset of analytes it was evident that yeast metabolism was substantially altered in real grape juice relative to defined medium. This is consistent with other research that has examined the influence of grape composition on yeast-derived flavour compound formation. Nevertheless, when considered together, the experiments in model grape juice and real grape juice provide insight into the extent of variation between existing yeast strains with regards to their potential flavour impacts. Strains could be grouped according to their overall aromatic profiles by applying clustering algorithms. Winemakers interested in harnessing yeast that boost production of desired aroma compounds can use these datasets to make an informed choice from amongst several strains, taking into consideration their performance-related properties and any preference for a particular supplier.
A major focus in this study was to determine the extent of yeast influence on red wine style, extending prior research that suggested Shiraz quality parameters were modulated according to yeast strain. In addition to laboratory-scale fermentations exploring yeast-phenolic interactions in the absence of grape solids, winemaking trials were performed in collaboration with AWRI Project 3.1.4 (texture). A follow-up winemaking trial was performed incorporating yeast strains studied throughout this project, shown at laboratory scale to yield the greatest differences in wine composition. Data on wine volatiles confirm that profiles observed at laboratory scale were replicated in pilot-scale fermentations. Together, these datasets emphasise that choice of yeast has a marked influence on red-wine style, and amongst available yeast strains there are several groups that have distinct properties that can be harnessed by winemakers.
Outcome 2: Knowledge of the potential to control yeast flocculation late in fermentation to improve processing efficiency
Cell-to-cell binding can result in the formation of large clusters of cells that settle to the bottom of the fermentation vessel, a process known as flocculation. This process is particularly helpful at the end of fermentation, making wine clarification and filtration easier. As a proof-of-concept study, gene technologies were used to construct a wine strain where the onset of flocculation could be triggered at any point during fermentation by modifying an external signal.
To gain insight into natural variation amongst wine yeast regarding the timing and strength of flocculation, two complementary approaches were taken. First, a novel high-throughput sedimentation-rate assay was devised and applied to end of fermentation samples of yeast from model grape juice fermentations described under outcome 1. The samples were transferred into model wine solutions where known triggers of flocculation were varied, namely pH and ethanol concentration, and some samples were transferred to lower temperatures. Second, a collection of 94 barcoded wine yeast strains developed in project AWR1302 were pooled and used to perform fermentations under varied conditions (pH, ethanol, temperature). The barcodes were used to detect which strains remained in suspension throughout fermentation and which strains flocculated at different stages of fermentation. Together these datasets revealed some strains that flocculate throughout fermentation and are unaffected by fermentation conditions, and other strains that flocculate strongly at the end of fermentation.
Fermentation and processing difficulties can also be associated with strains that are overly flocculent or flocculate inappropriately. The methods developed as components of this project were ultimately used not only to characterise flocculation behaviour but to alter it. Work on low alcohol yeasts in AWRI Project 3.3.1 identified strains capable of producing wine with reduced alcohol that were excessively flocculent and had only limited application in winemaking. Non-flocculating versions of those yeast have been isolated by applying the methods originally designed to characterise yeast flocculation behaviour.
Outcome 3: Knowledge of the impact that genome composition of wine yeast interspecies hybrids has on their potential to modulate wine style
Genome composition of hybrids generated in this project was assessed using custom-designed molecular assays that detect the presence or absence of representative chromosomal regions from each parent species, and by whole-genome sequencing. Genome stability was assessed by comparison of genome structure shortly after hybrid generation with structure following fermentation completion or in some cases after passaging cultures for hundreds of generations. Due to their method of creation cerevisiae/non-cerevisiae hybrids are generally triploid as determined by flow cytometry. The hybrids generated at the AWRI have demonstrated a surprising degree of genetic stability during mitotic growth. Examples of partial chromosome loss were encountered but these were often the consequence of the hybridisation event itself rather than propagation. Once a lineage was established its genetic makeup and phenotypic profile remained unchanged and thus represent suitable source material for commercial-scale propagation.
Outcome 4: Development of novel wine yeast strains
While results generated in this project demonstrate the extent of natural variation that exists amongst wine yeasts, and their potential to impact wine style, there is no single ‘perfect’ yeast. Each exhibits a combination of desirable and undesirable attributes. Furthermore, among commonly used wine yeast strains some properties are not available. Genome sequences (AWR1302) and wine compositional data from fermentations performed in this project were drawn upon to explore the potential use of genome-wide-association (GWAS) to identify genetic markers for strain development. Several potential markers were identified for production of volatile aroma compounds, such as those responsible for ‘rose’ aroma in wine, and work was initiated to confirm their relevance.
In parallel to the generation of these foundational datasets and platforms for future strain development, in this project classical strain development approaches were used to generate new interspecies hybrids and variants of an industry ‘workhorse’ wine strain (AWRI796) that improved one of its known defects and introduced a novel property.
The portfolio of interspecies hybrids generated by the AWRI was expanded to include hybrids with recently discovered Saccharomyces species: S. arboricolus and S. eubayanus. The winemaking potential of these two species was unknown prior to initiation of this work, though the industrial relevance of S. eubayanus is well established through its role as one parent of lager yeast. The capacity of lager yeast to ferment wort at low temperature prompted investigations of the low-temperature fermentation potential of wine strain x S. eubayanus hybrids alongside hybrids with another known cold-tolerant species, S. kudriavzevii. Screening of wine yeast interspecific hybrids generated between these species identified a number of hybrid yeast strains that complete fermentation at 12°C, but with a protracted timeline (15-18 days). However, two hybrid strains showed robust fermentation properties at 12°C and completed fermentation in a timely manner (9-12 days).
Newly generated hybrids, along with others generated in earlier projects, were also evaluated for use in sparkling wine production. Some of this work was performed in collaboration with Plumpton College in the UK and this has led to considerable interest in using interspecific wine yeasts for sparkling wine production.
AWRI796 was chosen as a starting point for strain improvement activities due to its relatively simple genomic organisation (its chromosomal copies are almost identical to one another) and widespread use in the wine industry. One of the known flaws of this strain is its tendency to produce high levels of succinic acid, which can lead to an imbalance in acidity and ‘salty’ flavour in wine (Coulter et al. 2004). Mutants of AWRI796 were successfully isolated that produced 40% less succinic acid than AWRI796 in white wines in laboratory trials.
Drawing upon knowledge of genes involved in production of volatile compounds responsible for ‘rose’ aroma in wine, 2-phenylethanol and 2-phenylethyl acetate, a selection strategy was implemented to isolate variants of AWRI796 with enhanced production of these compounds. This strategy succeeded without extensive accumulation of other mutations, meaning that isolated variants were essentially identical to AWRI796 aside from a very small number of changes to their genomes (< 10 nucleotides). These changes translated to production of between 5- and 50-fold increases in ‘rose’ aroma compounds relative to AWRI796, dependent upon grape variety fermented. Pilot- and industry-scale trials of these novel strains demonstrated a profound impact of these strains on wine sensory properties. Generally, favourable profiles were seen in white varieties; however, descriptors associated with these strains in red varieties were more varied and not always positive.