Characterizing the Potential of the Non-Conventional Yeast Saccharomycodes ludwigii UTAD17 in Winemaking

Non-Saccharomyces yeasts have received increased attention by researchers and winemakers, due to their particular contributions to the characteristics of wine. In this group, Saccharomycodes ludwigii is one of the less studied species. In the present study, a native S. ludwigii strain, UTAD17 isolated from the Douro wine region was characterized for relevant oenological traits. The genome of UTAD17 was recently sequenced. Its potential use in winemaking was further evaluated by conducting grape-juice fermentations, either in single or in mixed-cultures, with Saccharomyces cerevisiae, following two inoculation strategies (simultaneous and sequential). In a pure culture, S. ludwigii UTAD17 was able to ferment all sugars in a reasonable time without impairing the wine quality, producing low levels of acetic acid and ethyl acetate. The overall effects of S. ludwigii UTAD17 in a mixed-culture fermentation were highly dependent on the inoculation strategy which dictated the dominance of each yeast strain. Wines whose fermentation was governed by S. ludwigii UTAD17 presented low levels of secondary aroma compounds and were chemically distinct from those fermented by S. cerevisiae. Based on these results, a future use of this non-Saccharomyces yeast either in monoculture fermentations or as a co-starter culture with S. cerevisiae for the production of wines with greater expression of the grape varietal character and with flavor diversity could be foreseen.


Introduction
Inoculation with active dry yeasts of Saccharomyces cerevisiae is a common practice in most wine-producing regions since the middle of the 20th century, to assure prompt and reliable fermentations and wines with a consistent and predictable quality. The main drawback of this practice is the loss of the typical distinctive characteristics found in wines obtained by spontaneous fermentation,

Phenotypic Characterization
S. ludwigii UTAD17 was screened for relevant enological features [68]. The evaluation of stress resistance and the activity of the enzymes of enological interest was performed as described in [21]. Briefly, for all assays, after growth in YPD medium until the mid-exponential growth phase, the yeast strain was inoculated in the appropriate culture media for stress tolerance and enzymatic activities evaluation. YPD agar plates without a stress agent, was used as the control. Accordingly, the following concentrations were used-6%, 9%, or 12% (v:v −1 ) of ethanol; 1, 2, or 4 mM of sulfur dioxide (SO 2 ); 0.5, 1, or 2 mM of copper, supplied as copper sulfate; and 0.25, 0.5, or 1 mM of H 2 O 2 . Yeast growth in the presence of cerulenin or 5,5 ,5 -trifluoro-d,l-leucine (TFL) was screened to evaluate the potential to produce particular flavor compounds, using agar plates and a minimal medium (YNB) supplemented with glucose (2%) and TFL (0.6 mM) or cerulenin (6 µM).

Grape Juice and Inocula Preparation
Natural grape-juice was obtained by crushing grapes of the Vitis vinifera L. cv. Touriga Nacional; after homogenization, the juice was clarified by centrifugation at 12,734× g for 10 min (Sorvall centrifuge GSA 6-Place Rotor, Marshall Scientific, Hampton, NH 03842, USA) and was carefully separated from the solid fraction. A sample of the grape-juice was collected at this point for routine analysis (Table 1). After pasteurization at 70 • C for 10 min, the grape-juice was immediately cooled on ice. For each strain, the inoculum was prepared by separately pre-growing the yeast cells in 50 mL-flasks, containing 25 mL of synthetic grape-juice medium (GJM), original recipe of [69] with minor modifications in the nitrogen composition. Nitrogen was added up to 267 mg YAN/L, supplied as di-ammonium phosphate (DAP). The flasks were incubated overnight at 25 • C in an orbital shaker (IKA KS 4000 ic Control, VWR International, Radnor, PA 19087-8660, USA) set at 150 rpm.min −1 . Both strains were inoculated in grape-juice with an initial cellular concentration of 10 6 cfu·mL −1 .

Fermentation Trials
Fermentations trials were conducted by inoculating (1) a single culture of S. ludwigii UTAD17 (Sl), (2) a single culture of S. cerevisiae (Sc), (3) a mixed culture of S. ludwigii UTAD17 and S. cerevisiae (Sl+Sc) inoculated simultaneously, prior to fermentation, or (4) a mixed culture in which S. cerevisiae was inoculated sequentially, 72 h after S. ludwigii UTAD17 (Sl_Sc). Single and mixed culture fermentations were carried out in duplicates and triplicates, respectively, using a previously described system [70] consisting of 100 mL flasks filled to 2/3 of their volume (80 mL) and fitted with a side-arm port sealed with a rubber septum for anaerobic sampling. Two flasks containing uninoculated grape-must were used as control. The flasks were maintained at 25 • C under static conditions. Fermentations were monitored daily by weight loss as an estimation of CO 2 production and were allowed to proceed until no further weight loss was observed. For the assessment of growth parameters and analytical determinations, aseptic sampling was performed using a syringe-type system. After fermentation, the wines were centrifuged (10 min at 5500 rpm, Sigma 3-18K refrigerated Centrifuge, 37520 Osterode am Harz, DE) to remove yeast cells and were kept at −20 • C until the analytical determinations were performed.

Determination of Growth and Fermentation Parameters
Growth kinetics were monitored by viable cell plate counting (cfu·mL −1 ) on YPD agar or Lysine agar medium plates incubated at 28 • C for 48-72 h. The lysine agar medium was used to directly assess S. ludwigii UTAD17 viability in mixed-culture fermentations, since S. cerevisiae is unable to grow in a culture medium in which lysine is the sole nitrogen source [71]. The maximum fermentation rate (M ax FR) was determined from the slope of the linear dependence of the steepest incline in weight (g) at the corresponding time points (h), and fermentation purity (FP) was determined as acetic acid (g L −1 )/ethanol (%v:v −1 ).

Analytical Determinations
The amount of glucose and fructose, acetic acid, as well as Yeast Assimilable Nitrogen (YAN), comprising primary amino nitrogen (PAN) and ammonium, were enzymatically determined using a Y15 autoanalyzer (Biosystems S.A, Barcelona, Spain). Total SO 2 , pH, and titratable acidity were determined according to the standard methods compiled in the Compendium of International Methods of Analysis of Musts and Wines [72].
Ethanol and glycerol concentrations were determined in a high-performance liquid chromatography system (HPLC Flexar, PerkinElmer, Shelton, Connecticut, EE. UU) equipped with the ion exclusion cation exchange column Aminex HPX-87H (Bio-Rad Laboratories, Hercules, CA, USA) and refractive index detector. The column was eluted using sulfuric acid (0.005 N) at 60 • C and a 0.6 mL/min flow rate. Samples were previously filtered through a membrane (Millipore, 0.22 µm pore size) before an injection of 6 µL. The components were identified through their relative retention times, compared to the respective standards, using the Perkin Elmer Chromera Software.

Statistical Analysis
The data are presented as mean values with their standard deviation. One-way analysis of variance (ANOVA) of the inoculation strategy on yeast growth, fermentation activity, and volatile and non-volatile compounds was performed using the JMP 7.0 software (SAS Inc., 2007). If significant differences were found with ANOVA (p < 0.05), then Student's t-test was used for the paired comparisons. Partial least squares linear discriminant analysis (PLS-DA) was performed to discriminate the wines, based on volatile and non-volatile compounds, using the MATLAB R2018b environment (The MathWorks Inc.; version 9.5.; Natick, MA, USA) state abbreviation). All data were previously standardized.
2.8. Comparison of S. ludwigii UTAD17 'ORFeome' with S. cerevisiae s288c Recently S. ludwigii UTAD17 has been sequenced and annotated with the predicted set of open reading frames (ORFeome) estimated to be about 4015 protein-coding genes [67]. This whole-genome shotgun is available in the European Nucleotide Archive (ENA) under the accession number UFAJ01000000 (contigs UFAJ01000001 through UFAJ01001360; study accession number PRJEB27462; read accession number SAMEA4945973). Herein, a supervised analysis was performed by BLASTp using the proteomes of S. ludwigii UTAD17 and S. cerevisiae S288c, looking for the presence or absence of protein-coding genes that could underlie the observed physiological traits. S. ludwigii UTAD17 proteins were considered similar to those present in S. cerevisiae S288c when the resulting alignment had an associated e-value below e −20 and a minimum identity of 30% (Table S1).

Phenotypic Characterization of S. ludwigii UTAD17
In order to assess the potential of S. ludwigii UTAD17 to be used in winemaking, a phenotypic profiling was performed for a number of oenological traits, as determined by the International Organisation of Vine and Wine (OIV) [68] ( Figure S1). Ethanol is the main metabolite produced during wine fermentation while SO 2 and copper are applied by winemakers as antimicrobial agents to control spoilage in wineries and vineyards, respectively. These compounds have recognized negative effects on yeast growth and fermentative activity which could lead to stuck and sluggish fermentations [74]. The results obtained showed that S. ludwigii UTAD17 displayed high resistance to SO 2 (4 mM), ethanol (12% v:v −1 ), and copper (2 mM) ( Figure S1). The ability to produce biogenic animes which are toxic to humans, [21] was also evaluated. Interestingly, S. ludwigii UTAD17 did not present decarboxylase activities responsible for the production of histamine, tyramine and putrescine. On the other hand, S. ludwigii UTAD17 exhibited β-glucosidase and β-lyase activities involved in the liberation of terpenes from glycosylated precursors [12] and volatile thiols from cysteinylated precursors [19]. In line with the results obtained, S. ludwigii UTAD17 presents important features for a wine yeast starter, since it is able to adjust to winemaking stress, can contribute to the improvement of wine aromatic profile and does not compromise consumers' health.

Yeast Growth Kinetics and Fermentation Profiles
To evaluate the performance of S. ludwigii UTAD17 in winemaking conditions, fermentations were conducted either in single culture or in consortium with the commercial wine strain Saccharomyces cerevisiae QA23, inoculated simultaneously or sequentially, at 72 h. In parallel, a control fermentation was carried out using Saccharomyces cerevisiae QA23 in single culture, for comparison.
The growth dynamics and fermentation profiles for each single and mixed culture trials are presented in Figure 1. All fermentations were completed, although differences in the total time of fermentation were observed. The S. ludwigii UTAD17 showed a sugar uptake preference similar to S. cerevisiae strains, consuming glucose more rapidly than fructose. While in pure culture, S. ludwigii UTAD17 displayed a significantly lower fermentation rate than S. cerevisiae, although it was able to ferment the grape must sugar to dryness (below 4 g L −1 ) within 11 days, six more days than the time required for the high fermenter strain S. cerevisiae Lalvin QA23 (Table 1 and Figure 1B). This lower fermentative activity of S. ludwigii UTAD17 was not attributable to the differences in the biomass, which is known to have a great influence in determining the fermentation activity [69,75]. Indeed, both species, inoculated at the same amount (1 × 10 6 cfu·mL −1 ), resumed growth almost immediately after inoculation and, albeit with differences in the growth rate ( Figure 1A), achieved similar maximum cell populations-1.2 × 10 8 cfu·mL −1 for S. cerevisiae (after 48 h) and 1.1 × 10 8 cfu·mL −1 for S. ludwigii UTAD17 (after 72 h). A low fermentative capacity together with a high susceptibility to ethanol is believed to underlie the reduced competitiveness of non-Saccharomyces species along wine fermentation [76]. The phenotypic screening performed showed that S. ludwigii UTAD17 is able to tolerate up to 12% (v:v −1 ) of ethanol ( Figure S1), which is above the 10.2-10.4% (v:v −1 ) achieved in the final wines, indicating that the decline in cell viability registered in the later stage of the fermentation should result from other factors. The predicted ORFeome of S. ludwigii UTAD17 showed that this species is equipped with enzymes required for ethanol production from glucose, including hexoses transporters, glycolytic enzymes, and alcohol dehydrogenases (Table S1). Recently, genomic sequencing of an H. guilliermondii wine strain revealed that one of the key factors contributing to the reduced fermentation ability of this species is the lack of genes for the biosynthesis of thiamine [77], a cofactor of the pyruvate decarboxylase enzyme that is known to play an essential role in determining the regulation of the glycolytic flux [78,79]. This is not apparently the case in S. ludwigii UTAD17, since the thiamine-biosynthesis genes could be predicted from the genomic sequence of the UTAD17 strain (Table S1) [67]. Further studies are required to understand the lower fermentation rate exhibited by the S. ludwigii UTAD17, in comparison with S. cerevisae, one of the possibilities being a low activity of critical glycolytic enzymes, as recently shown to be the case in H. uvarum pyruvate kinase [80]. Contrary to that observed for S. cerevisiae which almost entirely consumed the nitrogen available in the medium, S. ludwigii UTAD17 displayed a preferential consumption of amino acids (PAN) over ammonium and left about 60 mg/L of YAN in the final wine (Table 1). It has been proposed that differences in the efficiency of nitrogen consumption, in general, and in the ability to uptake specific nitrogen sources from the grape-must account for variations in the fermentative activity of S. cerevisiae strains [81,82]. In this context, it would be interesting to determine whether the differences observed in the fermentation performances of S. cerevisiae and S. ludwigii UTAD17 (Figure 1) are the result of differences in their nitrogen uptake capability (Table 1). Table 1. Physicochemical composition of initial grape-must and wines obtained by single-cultures of S. ludwigii UTAD17 (Sl) and S. cerevisiae QA23 (Sc) or in consortium-mixed simultaneously (Sl+Sc) and sequentially (Sl_Sc) in natural grape-juice of Vitis vinifera L. cv. Touriga Nacional at 25 • C under static conditions.   When the two strains were co-inoculated simultaneously (Sc+Sl), a decrease in both strain populations was noted to likely reflect the more competitive environment in terms of space [83] and nutrients, two factors that were previously found to determine yeast-yeast interactions [32,74]. Yet, S. cerevisiae dominated over S. ludwigii UTAD17, in line with the notion that S. cerevisiae is better adapted to grape-juice per se than non-Saccharomyces yeasts [84]. Notably, although S. ludwigii UTAD17 was able to maintain a substantial viable population throughout these Sc+Sl fermentations, the rate and total time of fermentation was mostly similar, compared to S. cerevisiae in a single culture. In the sequential mixed-culture trials (Sl_Sc), S. ludwigii UTAD17 initiated fermentation, and S. cerevisiae QA23 was later inoculated at 72 h. At this stage, about 40% of the initial sugars were fermented by S. ludwigii UTAD17 and the assimilable nitrogen concentration was significantly lowered (Table 1). Contrary to that observed in simultaneous fermentations, in this case, S. ludwigii UTAD17 dominated over S. cerevisiae. Indeed, growth of the non-Saccharomyces strain proceeded as in a single culture, exhibiting a similar growth rate and loss of viability. On the other hand, in these experiments S. cerevisiae QA23 growth was limited to a maximum population of about 1.1 × 10 7 cfu mL −1 , attained 24 h after its inoculation, most probably due to the low assimilable nitrogen available in the medium. Nevertheless, there was an increase in the fermentation rate and sequential fermentations were successfully completed in eight days, taking 72 h more than S. cerevisiae in single culture. Adjusting the YAN levels of the fermenting must at this stage, could be an option to improve the growth of S. cerevisiae and the fermentation performance.

Effect of S. ludwigii UTAD17 on Wine Composition and Aroma Profile
The primary physiochemical parameters of the wines obtained are presented in Table 1. The production of ethanol is an essential attribute to define the use of yeasts in the production of fermented beverages. S. ludwigii UTAD17 showed a similar efficiency of sugar-to-ethanol conversion to that of S. cerevisiae QA23, as the ethanol levels of the final wines, which ranged from 10.2 to 10.4% (v:v −1 ), were not significantly different. Likewise, no significant differences were found on the amount of SO 2 formed in each fermentation. Both strains produced up to 20 mg mg L −1 and thus they were considered low-sulfite-forming yeasts [85]. Overall, all fermentations resulted in lower levels of acetic acid. Slightly lower levels of acetic acid were found in wines where S. ludwigii UTAD17 was involved, as compared to the wines only fermented by S. cerevisiae and significantly lower levels of this compound were detected on the sequentially inoculated wines (Table 1). These are promising features since the amount of both metabolites is tightly limited by regulations, might depreciate wine aroma (especially acetic acid) or raise concerns about consumers' safety (SO 2 ) and, thus, they should be kept at the lowest possible levels. The lower acetic acid produced in the fermentations dominated by the non-Saccharomyces yeast was accompanied by significantly lower levels of glycerol in all fermentations, suggesting that S. ludwigii UTAD17 management strategy for NADH/NAD + recycling and maintenance of redox balance is similar to that of S. cerevisiae [86]. This connection between acetic acid and glycerol production is not as clear in other non-Saccharomyces. For instance, in single culture fermentations, Starmerella bacillaris appears to be a high glycerol and low acetic acid producer [53], while H. uvarum seems to be a low glycerol and high acetic acid producer of yeast [87].
In order to evaluate how S. ludwigii UTAD17 affected the final aroma composition, the different wines were analyzed by gas chromatography. Eighteen yeast-derived aroma compounds were quantified, five alcohols, six acids, four ethyl-esters, two acetates, and one aldehyde ( Table 2). Those compounds that were found to be significantly different (p < 0.05), along with glycerol and acetic acid (Table 1), were used for Partial Least Squares-Discriminant Analysis (PLS-DA), in order to distinguish the wines obtained with the different inoculation strategies (Figure 2). The first component accounted for 69.99%, while component 2 explained 18.83% of the total variation. Replicate experiments were well grouped on the PLS. A clear separation was observed between the wines whose fermentation were dominated by S. cerevisiae, and those governed by S. ludwigii UTAD17. Although S. ludwigii UTAD17 produced overall significantly lower levels of volatile compounds (Table 2), it should be highlighted that the wines obtained by simultaneous co-inoculation of both strains (Sl+Sc), located in the upper-left quadrant (Figure 2), were characterized by a greater diversity of flavors and complexity, as compared with those fermented by S. cerevisiae alone. Overall, fermentations conducted by S. ludwigii UTAD17 resulted in wines characterized by higher levels of 1-butanol and butyric and isobutyric acids, which were found to be 2 to 3 times higher than that in Sc and Sc+Sl wines ( Table 2). 1-butanol and the short-chain fatty acid, butyric acid, are synthetized from 2-ketovalerate following decarboxylation to the aldehyde precursor, butyraldehyde, which is either reduced or oxidized, respectively [88]. In this study we could only speculate that the high levels of both compounds produced by S. ludwigii UTAD17 could result from either an excess of the intermediate α-keto acid or the use of this redox duality of the last steps of the Ehrlich pathway to help maintain the redox balance of the cell [89]. Overall, the above-described results gave good insights on the potential use of S. ludwigii UTAD17 in winemaking. Given the low fruity/estery nature of this strain, it could be a good option to obtain wines with a greater expression of grape varietal characteristics. Nowadays, winemakers make use of a blend of wines fermented with different yeast strains/species or different grape varieties to create wines tailored to meet consumer expectations. Following this line of winemaking, and considering the trend of the use of non-Saccharomyces yeasts to obtain wines with distinct aroma profiles, it should be underlined that the aromatic characteristics of S. ludwigii, UTAD17, which greatly differs from the traditional S. cerevisiae, might confer a peculiar imprint onto the final product. Nevertheless, our results warrant further studies to evaluate whether the observed differences in chemical composition can be perceived during sensory evaluation.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1, Figure S1: Phenotypic profile of the S. ludwigii UTAD17; Table S1: Comparative analysis of the predicted S. ludwigii UTAD17 ORFeome with sets of proteins of S. cerevisiae S288c   Previous studies have put aside the use of S. ludwigii in the winemaking industry, given the large amounts of ethyl acetate and acetaldehyde produced by these strains [7,8,61]. Notably, the S. ludwigii UTAD17 strain is, apparently, a low producer of both compounds as the levels detected were significantly inferior to those obtained in both Sc and Sl+Sc fermentations. Additionally, none-to-low levels of acetate and ethyl esters, which are responsible for the pleasant fruity and floral bouquet of wines, were found to be produced by S. ludwigii UTAD17. The predicted UTAD17 ORFeome did not include proteins similar to the S. cerevisiae acetyl transferases ScATF1, ScATF2, and ScAYT1, required for synthesis of acetate esters (Table S1). As such, the minor amounts of these compounds found on Sl_Sc fermentations was likely attributable to S. cerevisiae QA23 activity. Contrarily, we found that S. ludwigii UTAD17 harbors proteins similar to S. cerevisiae ethanol acyl-coA transferases ScEHT1 and ScEEB1, responsible for ethyl ester synthesis through the condensation between ethanol and fatty acyl-CoA [90], and ScIAH1 involved in esters hydrolysis [91]. In this case, the limited production of ethyl esters by S. ludwigii UTAD17 might result from differences in the activity levels of these enzymes. A more in-depth analysis of this genomic information is being undertaken that will shed light on the molecular foundations underlying some of the intriguing physiological traits of this yeast such as high SO 2 resistance, lower fermentative power, and wine aroma. From a practical point of view, this information will be very useful for a rational application of this strain, depending on the winemaking conditions Table 2. Concentration of volatile compounds detected and quantified by Gas Chromatography equipped with Flame ionization Detector (GC-FID) in wine obtained by a single-culture of S. ludwigii UTAD17(Sl) and S. cerevisiae (Sc) or in consortium-mixed simultaneous (Sl+Sc) and sequential (Sl_Sc), in natural grape-juice of Vitis vinifera L. cv. Touriga Nacional at 25 • C, under static conditions.     Overall, the above-described results gave good insights on the potential use of S. ludwigii UTAD17 in winemaking. Given the low fruity/estery nature of this strain, it could be a good option to obtain wines with a greater expression of grape varietal characteristics. Nowadays, winemakers make use of a blend of wines fermented with different yeast strains/species or different grape varieties to create wines tailored to meet consumer expectations. Following this line of winemaking, and considering the trend of the use of non-Saccharomyces yeasts to obtain wines with distinct aroma profiles, it should be underlined that the aromatic characteristics of S. ludwigii, UTAD17, which greatly differs from the traditional S. cerevisiae, might confer a peculiar imprint onto the final product. Nevertheless, our results warrant further studies to evaluate whether the observed differences in chemical composition can be perceived during sensory evaluation.