Saccharomyces cerevisiae gene expression during fermentation of Pinot noir wines at industrially relevant scale

During a wine fermentation, Saccharomyces cerevisiae transforms grape must through metabolic activities that generate ethanol and other compounds. Thousands of genes change expression over the course of a wine fermentation to allow S. cerevisiae to adapt to and dominate the fermentation environment. Investigations into these gene expression patterns have previously revealed genes that underlie cellular adaptation to the grape must and wine environment involving metabolic specialization and ethanol tolerance. However, the vast majority of studies detailing gene expression patterns have occurred in controlled environments that do not recapitulate the biological and chemical complexity of fermentations performed at production scale. Here, we present an analysis of the S. cerevisiae RC212 gene expression program across 40 pilot-scale fermentations (150 liters) using Pinot noir grapes from 10 California vineyards across two vintages. We observe a core gene expression program across all fermentations irrespective of vintage similar to that of laboratory fermentations, in addition to novel gene expression patterns likely related to the presence of non-Saccharomyces microorganisms and oxygen availability during fermentation. These gene expression patterns, both common and diverse, provide insight into Saccharomyces cerevisiae biology critical to fermentation outcomes at industry-relevant scales. Importance This study characterized Saccharomyces cerevisiae RC212 gene expression during Pinot noir fermentation at pilot scale (150 liters) using production-relevant conditions. The reported gene expression patterns of RC212 is generally similar to that observed in laboratory fermentation conditions, but also contains gene expression signatures related to yeast-environment interactions found in a production setting (e.g., presence of non-Saccharomyces microorganisms). Key genes and pathways highlighted by this work remain under-characterized, raising the need for further research to understand the roles of these genes and their impact on industrial wine fermentation outcomes.

Saccharomyces cerevisiae is most often the dominant fermentative organism during 41 vinification. As a domesticated species, it has evolved specialized metabolic strategies to 42 assimilate sugars in grape must and transform them into ethanol, thereby outcompeting 43 other microorganisms during fermentation (1). During this process, S. cerevisiae 44 encounters a dynamic stress landscape. In early fermentation, sources of stress include 45 high sugar concentration (osmotic stress), low pH (acid stress), decreasing oxygen 46 (hypoxia), the presence of other organisms that compete for nutrients or produce 47 inhibitory compounds, and sulfur dioxide additions that are used to inhibit spoilage 48 organisms. As fermentation progresses, nutrients become limiting (starvation), 49 temperature may rise or be kept low (heat/cold stress), and ethanol concentrations rise 50 (ethanol stress). Yet, through a coordinated gene expression response, S. cerevisiae adapts 51 to these stresses and most often continues fermentation until the must is dry. 52 High throughput gene expression profiling (e.g., microarray and RNA sequencing) 53 has offered a window into the metabolic strategies used by S. cerevisiae during 54 fermentation to adapt and dominate fermentation environments. Previous research has 55 reported expression changes in >2000 genes during fermentation (2-4). In early 56 fermentation, this is marked by expression of gene products that support biosynthetic 57 processes and acquisition of abundant nutrient resources (2, 3). As fermentation 58 progresses, nitrogen limitation, phosphate limitation, and/or ethanol accumulation can 59 trigger a transition to a non-proliferative state (i.e., stationary phase), which involves Here, to begin to address the impact of an industrial wine fermentation environment 122 on S. cerevisiae gene expression, we incorporate the inherent variability found in industrial 123 fermentations and determine the S. cerevisiae RC212 gene expression program across 124 chemically and biologically diverse Pinot noir grape musts. Specifically, time series  sequencing was used to capture the gene expression profiles of RC212 during 40 126 inoculated primary fermentations at pilot scale (150 liters) using California Pinot noir 127 grapes from 10 vineyards across two vintages. Using this data, a core metabolic program 128 was defined during fermentation, which is well reflected by lab-scale fermentations, in 129 addition to gene expression patterns that deviate from expectation. In particular, we 130 observe altered gene expression that may be explained by the presence non-Saccharomyces 131 organisms and regulation of metabolic processes related to stress, oxygen, and redox 132 balance throughout fermentation. These observations suggest that the core genetic 133 programs uncovered by lab-based studies are detected in industry-relevant fermentations, 134 but production-based environmental factors induce other gene expression programs that 135 are layered on top of the core gene expression program. We expect that understanding 136 such variations in gene expression within a wine production-like environment will be key 137 to defining approaches that can be used to manage commercial fermentation outcomes. 138

Conditions and rates of fermentation 140
Pinot noir grapes were harvested from the same 10 vineyards in California during 141 the 2017 and 2019 vintages for wine production at the UC Davis Teaching & Research 142 Winery ( Figure 1A). To standardize fermentations, grapes from the same Pinot noir clone 143 and rootstock were harvested at the same ripeness (~24 Brix, total soluble solids as a 144 proxy for sugar concentration). We sampled duplicate fermentations that used the grape performing differential expression, we further intersected the differentially expressed 171 genes across vintages to determine consistent changes that were vintage-independent. 172 From this analysis, 991 genes decreased expression as Brix decreased, while 951 genes 173 increased expression as fermentation progressed ( Figure 2AB, Table S1). Each vintage 174 also showed unique differential gene expression patterns, which may occur due to vintage-175 specific differences in fermentation. However, we generated these data at different times 176 and applied newly developed methods (UMI barcoding, see methods) for sequencing the 177 2019 samples, and as such we suspect that the higher number of differentially expressed 178 genes in the 2017 vintage may reflect differences in sequencing data quality. Nonetheless, 179 the large fraction of shared differentially expressed genes suggests that a core gene 180 expression program is followed independent of vintage. 181 Of the genes differentially expressed in fermentation and shared across vintage, 182 many are known to function in wine fermentation and are central to yeast growth, 183 metabolism, and cell survival ( Figure 2B and C, Table 1). A strong signature of growth 184 early in fermentation is observed that included cellular investment in ribosome biogenesis, 185 metabolism of lipids, purines, and amino acids, as well as cell division machinery (Figures 186 2C and S1-S2). These processes, coupled with enrichment of associated pathways involved 187 in RNA transcription and transport, reflect energy use for cell growth and proliferation 188 associated with log phase growth occurring in early fermentation. Further in fermentation, 189 changes in ribosomal machinery gene expression occurred, as reported in previous studies 190 (49) (Figure 2C) (Figures 2B and 2C). For example, HXT1 encodes a low affinity glucose 196 transporter that was more strongly expressed at the beginning of fermentation when 197 glucose is abundant. HXT4 has a high affinity for glucose and is expressed when glucose 198 concentrations are low (51), which is also observed in our data as HXT4 expression 199 increased in late fermentation. Importantly, the pathways we have identified as enriched in 200 early and late fermentation align with expectations based on previous research and the 201 known biology of S. cerevisiae during fermentation (2-5, 13-16). This highlights the core 202 processes that previous research efforts have defined and provides confidence that the 203 analysis methods employed in these pilot-scale fermentations capture these biologically 204 important transitions. 205 Beyond these previously defined core gene expression patterns, gene expression 206 signatures indicative of less understood processes within these fermentations are also 207 observed, which may be linked to the industry-like environment these studies were 208 performed in. These patterns of gene expression included signatures of nutrient limitation 209 in early fermentation and polyol metabolism in late fermentation that were both consistent 210 with interactions with non-Saccharomyces organisms. We also find signatures of 211 concurrent hypoxic and anoxic metabolism that suggests differential availability of oxygen 212 for some yeast populations throughout fermentation. In association, we observe mounting 213 gene expression that is likely involved in mitigating oxidative and other stresses. Finally, 214 few vintage-specific differences can be found, but those we do identify highlight gene 215 expression patterns that could be linked to altered fermentation outcomes. We discuss 216 these observations below. 217 218

Nutrient limitation in early fermentation 219
While gene expression data supports logarithmic growth at 16 hours post-220 inoculation ( Hanseniaspora guilliermondii and Brettanomyces has also been linked to induction of genes 234 involved in vitamin biosynthesis in fermentation (54, 55), which could be indicative of 235 increased nutrient competition and depletion of some nutrients early in fermentation. We 236 similarly observe induction of genes that encode enzymes involved in biosynthesis of B 237 vitamins in early fermentation, including BIO2 (biotin biosynthesis), RIB3 and RIB4 238 (riboflavin biosynthesis), PAN6 (pantothenate synthesis), SPE3 and SPE4 (pantothenic acid 239 synthesis), and MIS1 and FOL1 (folate biosynthesis). In addition, THI21 was induced, which 240 is involved in thiamine biosynthesis. As with phosphate, this may be related to the presence 241 of metabolically active non-Saccharomyces microorganisms that have been detected in all 242 of these fermentations (56). We expect that continued work using industry-like 243 fermentations across grape varieties and yeast strains, as well as controlled fermentations 244 using reconstituted microbial consortiums, will be critical for understanding the relevance 245 of these gene expression signatures to wine fermentation outcomes. If understood, such 246 interactions could potentially be addressed through timely nutrient additions to a 247 fermentation to achieve desired outcomes. A wine fermentation is generally regarded as an anaerobic process given that the 252 carbon dioxide (CO2) produced as a byproduct of ethanol fermentation protects must from 253 dissolved oxygen (57). Yet, within anaerobia, there is an important distinction between 254 hypoxic (low oxygen) and anoxic (no oxygen) conditions. In a fermentation, it is expected 255 that molecular oxygen (O2) is introduced into the grape must by handling processes, 256 including pump overs, that may introduce small amounts of dissolved oxygen into In late fermentation, induction of pathways like glycerol degradation and proline 290 metabolism that require oxygen were also observed. Glycerol is a compatible solute 291 involved in combating osmotic stress and redox balance and is primarily produced in early 292 fermentation (64). We found induction of GCY1 which encodes a glycerol catabolic enzyme 293 used in micro aerobic conditions (65), as well as RSF2, a transcriptional regulator of genes 294 that encode proteins required for glycerol-based growth. Proline metabolism genes PUT1, 295 PUT2, and PUT4 were also expressed at the end of fermentation. Although proline is an 296 abundant amino acid in grape must, it is a non-preferred nitrogen source of yeast and 297 requires oxygen to be metabolized (66). It was further observed that PUT1 and PUT2 were 298 induced in a sealed laboratory wine fermentation, but that proline was not metabolized 299 given the absence of oxygen (2). Expression of PUT1, PUT2, and PUT4 is regulated by 300 nitrogen catabolite repression (67) and the presence of proline in the absence of other 301 nitrogen sources (68), but is not regulated by the presence of oxygen. Intracellular proline 302 accumulation also protects S. cerevisiae from reactive oxygen species associated with 303 ethanol-rich environments (69). While it possible that glycerol and proline were 304 metabolized in late fermentation with oxygen ingress, other processes like nutrient 305 limitation and oxidative stress may also explain the induction of these genes. 306 Taken together, our gene expression data raises various questions about a 307 distributed gradient of oxygen (hypoxia and anoxia) in the fermentation environment that 308 may induce varied gene expression across the cell population. This could lead to yeast sub-309 populations undergoing varied metabolic outputs or having different levels of ethanol 310 tolerance due to the role of oxygen in these processes (70, 71). In the future, single cell 311 sequencing technologies combined with continuously monitored dissolved oxygen assays 312 may help resolve these questions. From a production perspective, in industrial 313 fermentations, even those that employ pump over systems and therefore maintain mixing 314 and better homogeneity, there is a gradient of dissolved oxygen in the fermentation tank 315 wherein oxygen concentration is higher toward the top of the vessel (58). This suggests 316 that heterogeneous gene expression profiles in response to oxygen would likely exist in 317 these environments too. This is also an important fact to consider, as oxygen additions 318 during fermentation are known to influence both fermentation and sensory outcomes. For 319 example, in late fermentation, a single oxygen pulse increases the rate of fermentation 320 mediated by ergosterol biosynthesis (70). Similarly, oxygen additions at different stages of 321 fermentation differentially impact wine aroma compound formation like volatile thiols and 322 esters; however, this appears to occur in a strain-dependent manner (71). This knowledge, 323 combined with the impact of oxygen addition on fermentation outcomes, raises the idea 324 that timely addition of oxygen may be a way to control fermentations rates and formation 325 of wine aromas, which would be a tool easily accessible to winemakers. Mitochondria also fulfill other critical roles in fermentation unrelated to respiration. 366 For example, mitochondria play a role in sterol uptake and transport under strictly 367 anaerobic conditions (84), and mitochondria quench reactive oxygen species especially 368 during ethanol stress (85). While we did not observe induction of specific genes related to 369 sterol biology and found induction of different genes related to reactive oxygen species 370 than those previously identified (see next section), these processes may also be linked to 371 In early fermentation, we see induction of genes involved in the thioredoxin system, 392 such as TRX1 and TRR1. Expressed targets of TRX1 included RNR1-RNR4 (91), genes 393 encoding ribonucleotide-diphosphate reductase required for DNA synthesis and cell cycle 394 progression, as well as MET16, which encodes an enzyme required for sulfate assimilation 395 (92). We further observed genes encoding Trx1 target peroxidases (TSA1) and 396 peroxiredoxins (AHP1) constitutively expressed throughout fermentation along with 397 superoxide dismutases (SOD1, SOD2). An additional source of ROS are peroxisomes, which 398 may generate hydrogen peroxide in early fermentation via beta-oxidation of fatty acids. 399 CTA1, which encodes a peroxisomal catalase, and ANT1, which encodes a peroxisomal 400 transporter involved in beta-oxidation of fatty acids, were expressed in early fermentation. 401 A major factor used to maintain redox balance is NADPH, which provides reducing 402 potential for the thioredoxin system. It has been shown that metabolic intermediates in 403 glycolysis can be re-routed to the pentose phosphate pathway to generate NADPH in 404 response to oxidative stress (93-95). We found that the pentose phosphate pathway was 405 enriched among genes expressed in early fermentation ( Figure 2C, Figure S1-S2), which 406 includes GND1, an enzyme that catalyzes NADPH regeneration and is required for the 407 oxidative stress response. Other expressed genes that encode enzymes acting downstream 408 of GND1 in the pentose phosphate pathway included RPE1, TLK1, TLK2, and TAL1. 409 Central to the glutathione (GSH) thiol-reductase system is glutathione, an abundant 410 tripeptide conserved throughout eukaryotic and prokaryotic cells with a critical role in 411 redox control, but its physiological role is both diverse and debated (95). We observed that 412 genes encoding enzymes involved in the degradation (DUG1 and DUG2), import (OPT1), and 413 biosynthesis (GSH1 and GSH2 in the 2017 vintage) of glutathione were expressed in early 414 fermentation. Additional generation of NADPH in early fermentation may be supported by 415 the transformation of isocitrate to alpha ketoglutarate via IDP1 in the mitochondria, and 416 export via YMH2, as both genes were also expressed. Genes encoding aldehyde 417 dehydrogenases ALD5 and ALD6 are similarly expressed in early fermentation, both of 418 which may regenerate NADPH through the transformation of acetaldehyde to acetate. ALD6 419 is the dominant isoenzyme responsible for acetate production in wine (96). 420 We further observed induction of genes involved in glutathione-mediated ROS 421 mitigation in late fermentation. For example, a gene encoding cytosolic glutaredoxin 422 (GRX1) was expressed in late fermentation. Unlike glutaredoxins in other species 423 (e.g., mammals), yeast glutaredoxins do not function as deglutathionylase enzymes (97). 424 Instead, induction of GRX1 increases resistance to hydroperoxides by catalytically reducing 425 hydroperoxides through glutathione conjugation and using the reducing power of NADPH 426 (98). In addition, the cytosolic peroxidase GPX1 was expressed. GPX1 uses both glutathione 427 and thioredoxin , in combination with NADPH, for reducing power (99). GPX1 is known to 428 be expressed by glucose and nitrogen starvation (100), which coincides with peak peroxide 429 formation in yeast during wine fermentation (88). While our gene expression data support 430 a role for cytoplasmic glutathione during late fermentation, genes encoding mitochondrial 431 peroxidin (PRX1) and thioredoxin (TRX3) were also expressed. Prx1 buffers the 432 mitochondria from oxidative stress and is reductively protected by glutathione, thioredoxin 433 reductase (Trr2), and Trx3 (101). Taken together, these results suggest that cytoplasmic 434 and mitochondrial systems may be integral to combating increased oxidative stress at the 435 end of fermentation. 436 Glutathione is also important for maintenance of cellular function via other systems. 437 For example, methylglyoxal is a byproduct of glycolysis, a reduced derivative of pyruvic 438 acid, that may account for up to 0.3% of glycolytic carbon flux in S. cerevisiae (102). We 439 found that GLO2, an enzyme that catalyzes methylglyoxal degradation in a glutathione 440 dependent manner, was expressed in late fermentation, as were glutathione-independent 441 systems involved in the degradation of methylglyoxal (GRE2/GRE3). Genes that encode 442 proteins involved in glutathione homeostasis were also expressed at the end of 443 fermentation, including GEX1 that encodes a proton:glutathione antiporter (103, 104). 444 GEX1 is known to be induced during oxidative stress (103)  Mannitol is one such polyol and a non-preferred sugar that can be metabolized by S. 506 cerevisiae (115-117). In S. cerevisiae, transporters encoded by HXT13 and HXT15-17 were 507 found to facilitate mannitol and sorbitol transport (116). In our data, we observed 508 induction of the mannitol transporter HXT13 in both vintages, along with the mannitol 509 dehydrogenase MAN2, which together indicate that mannitol may be present and 510 metabolized by S. cerevisiae at the end of fermentation (Figure 4). In line with this, 511 although eukaryotic transcriptional profiling via 3´ Tag-seq was performed (see methods), 512 L. kunkeei transcripts were detected in some fermentations, which is one potential source 513 of mannitol production. These data raises the possibility of mannitol consumption by S. offers clues to the potential functions of these genes that could be explored in future work. 558

Conclusion 559
In this study, we present a gene expression analysis across 40 pilot-scale 560 fermentations of California Pinot noir wine using grapes from 10 vineyard sites and two 561 vintages. The fermentations were diverse with different kinetics, initial chemical 562 conditions, and microbial communities (56). Yet among this diversity, we detected a core 563 gene expression program by S. cerevisiae that is largely consistent with that observed at 564 laboratory scale (2-4). Given that there are many genes consistently expressed across 565 these Pinot noir fermentations from diverse vineyards, members of this core fermentation 566 gene program represent strong candidates for future study to impact wine outcomes, 567 e.g. through manipulating redox balance (107-110). Excitingly, this includes a large 568 number of genes with unknown function that through investigation may provide new 569 insights into the biology of S. cerevisiae. 570 The largest deviations from benchtop fermentations are likely attributed to 571 activities of non-Saccharomyces organisms, but more research is needed to understand 572 these complex ecological interactions and their impact on fermentation. The gene 573 expression signatures around oxygen presence and metabolic availability also warrants 574 further research, in particular into the role of the mitochondria in late fermentation (3, 6, 575 74). While we detected few vintage-specific differences between fermentations, we expect 576 there are vineyard-site specific deviations from the consistent patterns of gene expression 577 described herein. Given the variability in fermentation kinetics with respect to time of 578 sampling, new methods will likely be needed to resynchronize stages of fermentation to 579 enable cross-vineyard comparisons (4). Future work is also needed to extend these 580 observations to other grape varieties and S. cerevisiae wine strains, which will define both 581 the shared and unique facets of the core gene expression program in S. cerevisiae linked to 582 these variables. With such information, we can address the impact of an industrial wine 583 fermentation environment on S. cerevisiae gene expression and define approaches that can 584 be used to manage commercial fermentation outcomes. 585

Grape preparation and fermentation 587
The wine making protocol used in this study was described previously (23, 48). The 588 grapes used in this study originated from 10 vineyards in six American Viticulture Areas in 589 California. All grapes were Pinot noir clone 667 rootstock 101-14. We harvested grapes at 590 approximately 24 Brix and transported the fruit to University of California, Davis Pilot 591 Winery for fermentation. We performed separate fermentations for grapes from each site, 592 with two fermentations per site, totaling of 20 fermentations per vintage (40 fermentations 593 total). After harvest, we separated the fruit into half-ton macrobins on harvest day and 594 added Inodose SO2 to 40 ppm. We stored the bins in a 14°C cold room until destemming 595 and dividing of the fruit into temperature jacket-controlled tanks. We performed N2 596 sparging of the tank headspace prior to fermentation and sealed tanks with a rubber 597 gasket. We cold soaked the must at 7°C for three days and adjusted TSO2 to 40 ppm on the 598 second day. After three days, we increased the must temperature to 21°C and set a 599 programmed pump over timetable to hold the tank at a constant temperature. We 600 reconstituted S. cerevisiae RC212 with Superstart Rouge at 20 g/hL and inoculated the 601 must with 25 g/hL of yeast. At approximately 24 hours after inoculation, we adjusted 602 nitrogen content in the fermentations using DAP (target YAN -35 mg/L -initial YAN)/2, 603 and Nutristart using 25 g/hL. We only adjusted nitrogen if target YAN was below 250 604 mg/L. Approximately 48 hours after fermentation, we permitted fermentation 605 temperatures to increase to 27°C and added DAP as previously described. Fermentations 606 ran to completion when Brix < 0. We took fermentation samples for Brix measurements 607 and RNA isolation at 16, 64, and 112 hours relative to inoculation. To ensure uniform 608 sampling, we performed a pumpover ten minutes prior to sampling each tank. For RNA 609 samples, we obtained 12mL of juice was obtained and centrifuged at 4000 RPM for 5 610 minutes. We discarded the supernatant and froze the pellet in liquid nitrogen. We stored 611 samples at -80°C until RNA extraction. 612

1126
Genes expressed higher in early fermentation in both the 2017 and 2019 vintages.

1127
Wine process Gene Gene product function

HXK2
Hexokinase that phosphorylates glucose in the first irreversible step leading to glycolysis.

PFK1, PFK2
Phosphofructokinases that catalyze the first irreversible reaction specific to glycolysis, producing fructose-1,6-bisphosphate from fructose-6-phosphate Acetate metabolism ALD4-6 Aldehyde dehydrogenase isoenzymes that produce acetate as a byproduct when acetaldehyde is metabolized. ALD6 encodes the main isoenzyme responsible for acetate production in wine (96). Aldehyde dehydrogenase isoenzymes ALD4 and ALD5 are expressed when ethanol is the carbon source and are not typically associated with wine fermentation. Acetate contributes the majority of volatile acidity associated with negative organoleptic properties in wine (138).
PDR12 Plasma membrane ABC transporter. Required for development of resistance to weak organic acids, including acetate (139).

CWP1
Expressed in the S/G2 phase of cell cycle (140).

MUP1
High-affinity methionine permease that is also involved in cysteine transport.

CAR1, CAR2
Involved in arginine catabolism. Arginine is the most abundant amino acid in grape must after proline (66) and is used in protein synthesis during fermentation (141).

YPQ1
Vacuolar transporter for arginine and lysine. Unused arginine is stored in the vacuole for later use (141).
Ehrlich pathway

BAT1, ARO8
Catalyzes transamination of amino acids, the product of which cannot be redirected to central carbon metabolism and so is excreted as fusel acid or fusel alcohol (142).
Overexpression of BAT1 increases the concentration of isoamyl alcohol, its acetate ester, and isobutanol in wine (142).

1128
Genes expressed higher in late fermentation in both the 2017 and 2019 vintages.

1129
Wine process Gene Gene product function

GAT1, DAL80
Transcriptional activator (GAT1) and repressor (DAL80) of genes under nitrogen catabolite repression. Expression is inversely correlated, and the detection of both genes as induced in late fermentation likely indicates tight transcriptional regulation of nitrogen metabolism.

MEP2
Ammonia permease and expression is under nitrogen catabolite repression.

AVT3, AVT4
Vacuolar amino acid exporters that mobilize internal nitrogen stores for cell maintenance during stationary phase. Expression is under nitrogen catabolite repression.
ubiquitinmediated selective protein degradation

RPN4
Transcription factor that induces expression of proteasome genes.

TMC1
Effector of proteotoxic stress induced by nitrogen limitation, weak acid, and misfolded proteins. Target of RPN4.
Autophagy ATG2, ATG4, ATG7-ATG12, ATG14, ATG32, ATG40 Proteins involved in autophagy. Autophagy is a key response to nutritional limitation that allows cells to maintain homeostasis (143). Nitrogen starvation leads to the largest autophagic response in yeast.

Erlich pathway GRE2
Final step of pathway where fusel aldehydes are oxidized or reduced into fusel acids or alcohols (142).

Carbon limitation SNF3
Plasma membrane low glucose sensor involved in regulating glucose transport

SKS1
Serine/threonine kinase involved in the adaptation to low glucose via SNF3independent signaling

PGM2
Phosphoglucomutase. Catalyzes a key step in hexose metabolism. Induced in response to glucose limitation and ethanol stress (144).

HXK1
Hexokinase that phosphorylates glucose or fructose in the first irreversible step leading to glycolysis. Under glucose-induced repression.

HXT4, HXT6
Hexose transporters required at the end of alcoholic fermentation.