Saccharomyces cerevisiae FLO1 Gene Demonstrates Genetic Linkage to Increased Fermentation Rate at Low Temperatures

Low fermentation temperatures are of importance to food and beverage industries working with Saccharomyces cerevisiae. Therefore, the identification of genes demonstrating a positive impact on fermentation kinetics is of significant interest. A set of 121 mapped F1 progeny, derived from a cross between haploid strains BY4716 (a derivative of the laboratory yeast S288C) and wine yeast RM11-1a, were fermented in New Zealand Sauvignon Blanc grape juice at 12.5°. Analyses of five key fermentation kinetic parameters among the F1 progeny identified a quantitative trait locus (QTL) on chromosome I with a significant degree of linkage to maximal fermentation rate (Vmax) at low temperature. Independent deletions of two candidate genes within the region, FLO1 and SWH1, were constructed in the parental strains (with S288C representing BY4716). Fermentation of wild-type and deletion strains at 12.5 and 25° confirmed that the genetic linkage to Vmax corresponds to the S288C version of the FLO1 allele, as the absence of this allele reduced Vmax by ∼50% at 12.5°, but not at 25°. Reciprocal hemizygosity analysis (RHA) between S288C and RM11-1a FLO1 alleles did not confirm the prediction that the S288C version of FLO1 was promoting more rapid fermentation in the opposing strain background, suggesting that the positive effect on Vmax derived from S288C FLO1 may only provide an advantage in haploids, or is dependent on strain-specific cis or trans effects. This research adds to the growing body of evidence demonstrating the role of FLO1 in providing stress tolerance to S. cerevisiae during fermentation.

transcription of $500-1000 genes depending on the strain and conditions used (Beltran et al. 2006;Deed et al. 2015). The transcriptional response to low temperature fermentation is initiated in two steps, first via the induction of cold-specific stress genes, followed by the more generalized environmental stress response and fermentation stress response (Gasch and Werner-Washburne 2002;Beltran et al. 2006;Marks et al. 2008;Deed et al. 2015). It has been well documented that different S. cerevisiae strains vary greatly in their ability to grow and ferment at lower temperatures (Charoenchai et al. 1998;Torija et al. 2003), and it has been suggested that these phenotypic differences are due to strain differences in gene expression, particularly via variation in gene promoter regions and the expression of transcription factors (Beltran et al. 2006;Chiva et al. 2012;Treu et al. 2014;Deed et al. 2015).
We have carried out genetic linkage analysis, using a set of 121 completely mapped (.99% of the genome) F 1 progeny from a cross between haploid strains BY4716 and RM11-1a [denoted as BY and RM respectively in Brem et al. (2002)], to identify QTL with a positive influence on yeast fermentation kinetics at low temperature (12.5°). A region on chromosome I showed statistically significant genetic linkage to V max among the F 1 progeny, and gene deletions and RHA were used to investigate the causative gene within this region.
Growth and fermentation conditions BY4716 · RM11-1a F 1 progeny, parental strains, and diploid F 1 hybrids generated for RHA were fermented at 12.5°(and 25°for the RHA strains) in Sauvignon Blanc grape juice, containing $22°Brix and 281 mg L 21 yeast assimilable nitrogen (Pernod Ricard, Marlborough, New Zealand). Grape juice was sterilized via overnight incubation at 25°with 200 ml L 21 dimethyl dicarbonate and supplemented with the following amino acids: 10 · leucine (300 mg L 21 ), 10 · lysine (300 mg L 21 ), and 10 · uracil (100 mg L 21 ). Yeast cultures were propagated in yeast-peptone-dextrose medium (YPD) and incubated overnight at 28°, with orbital shaking at 150 rpm. Grape juice was mixed well before being used to make 8 ml aliquots in 13-ml ventilation cap polypropylene tubes to ensure an even distribution of grape solids. Fermentations were inoculated with 1 · 10 6 cells ml 21 and a ,0.5 mm 2 pin-hole was punctured into each tube lid to allow for CO 2 escape. Fermentations were monitored daily by measuring cumulative weight loss (g) (Bely et al. 1990). To reduce variability within triplicate fermentations, outliers were removed if they deviated from other replicates by .10% weight loss at three consecutive time points, after .50% total weight loss. Fermentations of RHA strains were performed in nonaplicates (n = 9).

Analysis of fermentation kinetic parameters
Phenotypes for five fermentation-related kinetic variables, maximal fermentation rate (V max ) (dCO 2 /dt), maximal acceleration rate (A max ) (d 2 CO 2 /dt 2 ), length of lag phase (h), final weight loss (g), and finishing time of alcoholic fermentation (AF time) (h), were determined from cumulative weight loss data, as per Marullo et al. (2006).

Linkage analysis
Quantitative phenotypic data for the five fermentation kinetic parameters were sent to J. Bloom and J. Gerke (Princeton University) for QTL mapping and identification of relevant loci. Logarithm (base 10) of odds (LOD) scores were generated for 2957 genetic markers across the 16 S. cerevisiae chromosomes using R/QTL's scanone function, and a nonparametric model to compare the likelihood of obtaining the phenotypic data if mapped loci are linked against the likelihood of obtaining the data by chance (Broman et al. 2003). GBrowse maps of chromosomal regions with LOD scores . 3 (significant with a 5% chance of error) were obtained from the Saccharomyces Genome Database (SGD) (http://www.yeastgenome.org/) to determine candidate open reading frames (ORFs) linked to V max and lag phase.
Gene deletions and RHA Deletion of candidate genes, FLO1 and SWH1, within the chromosome I QTL linked to V max , were constructed in S288C and RM11-1a using a modification of the Schiestl and Gietz (1989) lithium acetate yeast transformation protocol. The KanMX construct within the HO gene of RM11-1a was replaced with a hygromycin resistance (HGM R ) cassette, HphMX, to allow for subsequent integration of KanMX into the two candidate genes. Transformation of haploid S288C and RM11-1a was performed independently to generate mutants with KanMX insertions in FLO1 and SWH1 using constructs amplified from the BY4743 EUROSCARF strains, FLO1DYAR050W::KanMX and SWH1DYAR042W::KanMX. Deletions of FLO1 and SWH1 were confirmed by PCR (list of oligonucleotide primers in Table 1). Crosses were made between wild-type S288C, RM11-1a, and flo1 and swh1 deletion mutants in order to construct diploid F 1 hybrids for RHA (Steinmetz et al. 2002) (crosses shown in Table 2). A multiplex PCR to amplify 10 variable microsatellite markers and two mating type loci, MATa and MATa, was used to ensure that the F 1 hybrids were constructed correctly (Table 3) (Richards et al. 2009).

Data availability
All strains are available upon request. Supplemental Material: Table  S1 contains the values for the kinetic parameters for all individuals. Table S2 contains the list of ORFs identified either side of each LOD .3 peak marker. File S1 contains the LOD scores for all individuals across the five fermentation parameters. File S2 contains Clustal alignments.

RESULTS
Fermentation of 121 mapped F 1 progeny identified genes linked to fermentation rate and lag phase Cumulative weight loss was measured for 616 hr throughout the fermentation of 121 BY4716 · RM11-1a F 1 progeny at 12.5°( Figure  1). As expected, the RM11-1a parental strain demonstrated superior fermentation performance compared to the S288C parental reference. F 1 progeny demonstrated sufficient phenotypic variation for genetic mapping with fermentation curves covering the full range between both parents. Positive heterosis was also evident, with some F 1 progeny exhibiting improved fermentation performance compared to RM11-1a. After removal of outliers (triplicate fermentations that deviated by .10% weight loss), 6/121 of the F 1 progeny were analyzed only in duplicate (3D, 5B, 6A, 8G, 10G, and 3A-2), while two F 1 progeny were completely excluded from the analysis (5A-1 and 11F-1). Five fermentation-related kinetic variables were derived from the weight loss data: V max , A max , length of lag phase, final weight loss, and AF time (Table S1). These parameters were used for QTL mapping.
LOD scores were generated from the phenotypic data from the remaining 119 F 1 progeny, which resulted in the identification of three regions across the genome with LOD scores . 3 (see File S1). A region on chromosome III was linked to V max , whereas regions on chromosomes VII and XIII were linked to lag phase. No loci had significant linkages to A max , final weight loss, or AF time.
Closer inspection of the chromosome III region linked to V max on SGD indicated that the linkage was due to the inclusion of the LEU2 locus, which is deleted in RM11-1a. Removal of the effect of LEU2 on the dataset eliminated the chromosome III peak and resulted in a significant LOD score for the QTL at the subtelomeric end of chromosome I. Figure 2 shows LOD score plots for V max before ( Figure 2A) and after ( Figure 2B) the effect of LEU2 was removed by using the LEU2 genotype of F 1 progeny as a covariate in a linear model of phenotype. LOD score data shows that the advantage for the V max trait on chromosome I is derived from the BY4716 allele, and not the RM11-1a allele (File S1). This was somewhat unexpected, given that the parental fermentation data showed that S288C progressed throughout fermentation much slower than RM11-1a, although the phenomenon of "low parents" in terms of transgressive segregation has been described previously (Ehrenreich et al. 2009).
Identification of two candidate genes in the chromosome I region linked to V max and multiple ORFs on chromosomes VII and XIII in the regions linked to lag phase Genes were identified in regions with LOD scores . 3 with C.I.s set at one LOD unit drop either side of a peak marker. GBrowse maps were used to identify and visualize all ORFs within the defined areas for V max and lag phase on chromosomes I, VII, and XIII (ORFs listed in Table  S2), and the presence of nucleotide differences between the parental strains was also considered as an additional criterion for candidate ORFs. Of the six ORFs in the region linked to V max on chromosome I, two were considered to be potentially relevant to low temperature fermentation based on their respective functions: FLO1 (YAR050W), encoding a cell wall lectin-like protein that binds mannose and is involved in flocculation (Miki et al. 1982); and SWH1 (YAR042W, previously known as OSH1), encoding an oxysterol binding protein (Schmalix and Bandlow 1994). S. cerevisiae swh1 mutants exhibit phenotypes similar to viable mutants defective in sterol biosynthesis and show a reduction in membrane ergosterol levels, which also results in low temperature sensitivity in a tryptophan auxotroph (Jiang et al. 1994;Daum et al. 1999). However, sequence alignment analyses using Clustal found very few allelic differences between SWH1 in S288C and RM11-1a (99% similarity and deletion of two amino acids in S288C, see File S2). In contrast to SWH1, FLO1 has a very repetitive gene structure and the allele from RM11-1a has multiple large deletions compared to S288C (see File S2). Additionally, FLO1 is very highly expressed during n The S288C parent strain in bold were added in 100 · excess of the RM11-1a parent, as S288C did not have any selectable markers that differed from RM11-1a. The F 1 hybrid selections marked with Ã could result in the presence of the RM11-1a parent, as well as the F 1 hybrid. The R-FS · S-FS cross was included as a control. HGM R , hygromycin resistance; Kan R , kanamycin resistance.
n  (Bester et al. 2006;Shen et al. 2006;Fichtner et al. 2007). Ten ORFs were within the C.I.s near the LOD score peak for lag phase on chromosome VII (Table S2), including two genes encoding B-type cyclins involved in cell cycle progression, CLB1 (YGR108W) and CLB6 (YGR109C) (Surana et al. 1991;Schwob and Nasmyth 1993). Two neighboring peak markers with LOD scores . 3 were identified on chromosome XIII in the region linked to lag phase. Either side of these two peak markers, 34 ORFs were identified (Table S2). Genes of interest include RCF1 (YML030W), encoding a cytochrome c oxidase subunit that is required for growth under hypoxic conditions (Strogolova et al. 2012;Vukotic et al. 2012), and YOX1 (YML027W), encoding a transcriptional repressor involved in the regulation of cell cycle genes (Kaufmann 1993;Horak et al. 2002). Due to the difficulty of reproducibly phenotyping lag phase between different experiments in grape juice and the sheer number of potential candidate genes within the regions linked to lag phase, it was decided to concentrate on the identification of the locus influencing fermentation rate on chromosome I.

The FLO1 gene is linked to V max
To determine the effect of the FLO1 and SWH1 loci on V max , deletions of FLO1 and SWH1 were constructed in S288C and RM11-1a and hybrids were created to perform RHA. Fermentations at 12.5 and 25°i n Sauvignon Blanc juice were performed using the original S288C and RM11-1a strains (renamed S-FS and R-FS to indicate strain name and FLO1/SWH1 genotype), the haploid flo1 and swh1 S288C and RM11-1a deletion strains (renamed S-Fs, S-fS, R-Fs, and R-fS), and the five diploid RHA F 1 hybrids constructed by crossing combinations of S288C and RM11-1a wild-type, flo1, and swh1 deletion strains (R-FS · S-FS, R-Fs · S-FS, R-fS · S-FS, R-FS · S-Fs, and R-FS · S-fS, see Table 2).
V max data are presented in Figure 3. At 12.5°( Figure 3A), there was no significant difference between the fermentation rates of S-FS or R-FS compared to three of the deletion mutants: S-Fs, R-Fs, or R-fS. However, the V max of the S288C flo1 mutant, S-fS, was reduced by $50% compared to the wild-type S-FS strain and the other deletion mutants. This result indicates that the FLO1 allele is important for low temperature fermentation in S288C, but not in RM11-1a. There was no difference in V max between the R-FS · S-FS F 1 hybrid and the original parent strains; however, R-FS · S-FS had a slightly but significantly lower V max than three of the RHA F 1 hybrids: R-Fs · S-FS, R-fS · S-FS, and R-FS · S-fS. The significance of this result is not clear, but may n  S288C  120  174  130  240  302  296  357  407  443  415  468  -RM11-1a  121  139  146  259  260  296  364  381  419  427  -492  R-Fs · S-FS  120  139  130  240  260  296  358  381  419  415  468  492  121  174  146  259  303  296  364  407  443  427  R-fS · S-FS  120  139  130  240  260  296  358  381  419  415  468  492  121  174  146  259  303  296  364  407  443  involve uncharacterized cis or trans effects in the different strain background. At 25°( Figure 3B), there was no significant difference between the fermentation rates of S288C (S-FS) and its flo1 or swh1 mutants (S-fS or S-Fs). There was also no difference between RM11-1a (R-FS) compared to its two deletion mutants, R-fS or R-Fs. However, the V max of the S288C swh1 mutant, S-Fs, was slightly but significantly higher than that of S-Fs. These data suggest that the deletion of the flo1 locus does not have a significant effect on the maximal fermentation rate at higher temperature. The five diploid RHA hybrids showed only minor differences in fermentation rate, with no clear pattern emerging.
The reduction in maximal fermentation rate in the S288C flo1 deletion strain, S-fS, strongly suggests that FLO1 is linked to low temperature fermentation and most likely corresponds to the high LOD score region on chromosome I. However, RHA between S288C and RM11-1a FLO1 gene variants did not identify any easily explained effects on maximal fermentation rate in the cold. In particular, the R-FS · S-fS F 1 hybrid did not ferment poorly compared to the other hybrids, which would be expected if the presence of the S288C FLO1 allele was promoting a more rapid fermentation at 12.5°.

DISCUSSION
Genetic linkage analysis, using a set of completely mapped 119 BY4716 · RM11-1a F 1 progeny, identified a strong linkage between maximal fermentation rate at low temperature, and the FLO1 gene on chromosome I. Mapping data indicated that the beneficial allele was derived from the "low parent" BY4716 and not from RM11-1a. The linkage of the BY4716 variant of FLO1 to V max was validated based on a 50% reduction in V max in a cold-fermented S288C flo1 mutant; as expected, there was no difference in V max between the RM11-1a flo1 deletion strain compared to RM11-1a. However, RHA between S288C and RM11-1a FLO1 alleles did not confirm the prediction that the S288C version of FLO1 was promoting more rapid fermentation in a different strain background.
FLO1 has a role in stress tolerance during low temperature fermentation The 4.6 kb FLO1 gene encodes a cell wall surface protein that aggregates cells into "flocs" by binding to mannose sugar chains on the surfaces of other cells (Miki et al. 1982;Teunissen and Steensma 1995), and on substrates during glucose starvation (Fichtner et al. 2007). FLO1 is one of four subtelomeric and structurally similar FLO genes possessed by S. cerevisiae (the others are FLO5, FLO9, and FLO10), and together they control the flocculation phenotype of different S. cerevisiae strains (Teunissen and Steensma 1995). Previous studies strongly suggest that the floc formation by S. cerevisiae is a protective mechanism against environmental and nutritional stress, since flocculation is typically induced in response to high ethanol, antifungal agents (Teunissen and Steensma 1995;Smukalla et al. 2008;Beauvais et al. 2009), and nutrient limitation [particularly carbon and/or nitrogen, see Rose (1984), Sampermans et al. (2005) and Stratford (1992)]. FLO1-expressing industrial strains also have improved fermentation performance under acetic acid stress compared to strains not expressing FLO1 (Du et al. 2015), and consume hexose sugars more efficiently than nonexpressing strains in the presence of fermentation inhibitors (Westman et al. 2014). The subtelomeric location of FLO1 is also in agreement with the observation that an especially high proportion of variable genes located at chromosomal telomeres are involved in fermentation (Argueso et al. 2009;Cubillos et al. 2011). The protection provided by the formation of flocs is not only due to the physical shielding of the cells in the center of the floc, but also due to an increased overall resistance to stress (Smukalla et al. 2008). We hypothesize that the induction of the transcriptional flocculation response not only has a role in protecting cells from chemical stressors, but also plays a role during low temperature fermentation. Smukalla et al. (2008) have shown that S288C cells engineered to express a Flo1+ flocculation phenotype also upregulate genes involved in cell wall, lipid, and sterol metabolism, which are also induced during the stress response to low temperatures (Beney et al. 2001;Gasch and Werner-Washburne 2002;Beltran et al. 2008;Redón et al. 2011;Deed et al. 2015). Additionally, genes within the DAN/TIR and PAU gene families, which have long been associated with the transcriptional response of S. cerevisiae to low temperature (Kondo and Inouye 1991;Abramova et al. 2001;Homma et al. 2003;Schade et al. 2004), including during fermentation at low temperatures (Beltran et al. 2006;Deed et al. 2015), are also induced in flocculating cells (Smukalla et al. 2008). Low fermentation temperatures may also favor flocculation due to reduced turbulence from the lower metabolic rate and slower CO 2 formation (Soares 2011). FLO1expressing cells preferentially stick to one another, regardless of genetic relatedness across the rest of the genome, suggesting a level of cooperativeness (Smukalla et al. 2008). This cooperation toward other cells expressing the same gene suggests that FLO1 is one of very few "green beard genes" for altruistic social interactions. Rossouw et al. (2015) have taken this idea one step further by showing that FLO genes allow S. cerevisiae to form large ecological networks with non-Saccharomyces species, including both flocculant and nonflocculant strains. Additionally, different members of the FLO gene family either promote or repress certain combinations of mixed species and/or strain adhesion.
The positive influence of the S288C FLO1 allele has never before been described for fermentation rate and this effect appeared to be enhanced significantly at low temperature (see Figure 3A). It is widely assumed that FLO1 in S288C is not expressed, because its transcriptional regulator, Flo8p, is nonfunctional in S288C due to a nonsense mutation (Liu et al. 1996). In RM11-1a, the FLO8 gene is functional and Brem et al. (2002) found that one-quarter of the F 1 progeny from the BY4716 · RM11-1a cross showed a flocculation phenotype (Flo1+ and Flo8+). Gene expression microarray data from Deed et al. (2015) show that FLO1 transcripts in a M2 · S288C F 1 hybrid are dramatically upregulated during low temperature fermentation compared to the M2 parental reference. FLO1 was upregulated 73-fold at early fermentation (2% weight loss) and 182-fold at midlate fermentation (70% weight loss). Typically, FLO1-dependent flocculation requires activation by Flo8p, in conjunction with another transcription factor, Mss11p, which also coregulate the MUC1/FLO11 flocculin (Kobayashi et al. 1996;Bester et al. 2006;Fichtner et al. 2007). However, there are reports of FLO1 being activated in a Flo8p-independent manner. For example, the overexpression of MSS11, encoding a transcription factor, can overcome the flo8 deletion in S288C (Bester et al. 2006). There is evidence that MUC1/FLO11 and MSS11 have temperature-dependent regulation and can only facilitate trait expression at lower temperatures, strengthening the case for temperature-specific roles for cell surface proteins such as Flo1p Taylor et al. 2016). Additionally, Shen et al. (2006) found that another transcription factor, Gts1p, could induce FLO1 in a flo8 mutant strain of W303-1A by binding to the Sfl1p repressor. This research supports the idea that specific environmental signals, initiating a stress response, may allow for FLO1 to be induced in S288C, independent of Flo8p, via other transcriptional regulators. Since Flo1p tends to support cell-substrate interactions under specific environmental conditions (Fichtner et al. 2007), the fermentation environment may induce FLO1 in a Flo8p-independent manner, resulting in cell-substrate adhesion, rather than flocculation per se. If this is the case, the ability of the S288C FLO1::KanMX (S-fS) mutant to form attachments to substrates within the grape solids could be visualized vs. the wild type using a technique such as atomic force microscopy (Canetta et al. 2006). Increased adhesion of yeast cells to substrates such as nutrients or grape solids may result in a higher fermentation rate at low temperature, as shown by cells that ferment while immobilized onto supports made of cellulose, gluten, corn starch, or wheat grains, in numerous studies (Mallouchos et al. 2003(Mallouchos et al. , 2007Kandylis et al. 2008Kandylis et al. , 2010Lainioti et al. 2011). This finding is in line with current literature demonstrating a role of Flo1p in protecting yeast from environmental stresses and improving fermentation performance under industrial conditions, including improved resistance to ethanol, acetic acid, and antimicrobial compounds (Queller 2008;Smukalla et al. 2008;Soares 2011;Westman et al. 2014;Du et al. 2015;Rossouw et al. 2015;Cheng et al. 2016).
The positive effect on V max was not visible in the RHA F 1 hybrids constructed from crosses between RM11-1a and S288C parent and deletion strains. The R-FS · S-fS RHA strain contained one FLO1 allele from RM11-1a and the flo1 deletion from S288C. The absence of the S288C FLO1 allele was predicted to result in a lower V max compared to the other RHA hybrids that possessed the wild-type version of this same allele. However, there were no significant differences between the RHA hybrids. One possible explanation is that the positive influence on V max only occurs in haploid strains and not in diploids. The RHA F 1 hybrids were the only diploid strains analyzed, since the two parents and all the 119 F 1 progeny were haploid. Fichtner et al. (2007) state that FLO1-dependent flocculation is haploid-specific and that diploids display invasive or pseudohyphal growth via a nonsubtelomeric FLO gene, MUC1/FLO11, encoding a GPI-anchored cell surface glycoprotein required for pseudohyphal formation (Kobayashi et al. 1999;Guo et al. 2000). Haploids and diploids often differ in their tolerance to stress, even with the same genetic backgrounds, which extends to fermentation-related stressors such as ethanol (Katou et al. 2008;Li et al. 2010). Since the induction of flocculation has an impact on the ethanol resistance of S. cerevisiae, perhaps the influence of FLO1 on V max is haploid-specific and works by providing additional ethanol tolerance.
As discussed in Sinha et al. (2006), QTL architecture can be very complex. In this case, there may be the requirement for a second gene modifier compensator (X R /X S ) that works along with FLO1 in order for the benefit for V max to be present. For instance, a second gene modifier may be FLO8, encoding the transcriptional inducer of FLO1; MUC1/ FLO11; another member of the FLO gene family with high sequence homology, such as FLO5, FLO9, or FLO10; or one of three pseudogenes, YAL065C, YAR061W (merged with YAR062W), or YHR213W, with sequence similarity to other flocculin genes (Teunissen and Steensma 1995). If the advantage of S288C FLO1 for improved fermentation performance at low temperature is Flo8p-independent, perhaps the complementation of FLO8 via hybridization of S288C with RM11-1a prevented the S288C version of FLO1 from having any effect on maximal fermentation rate. However, analyses of the LOD score data for FLO8, MUC1/FLO11, and the six FLO homologs did not identify any significant peaks corresponding to V max . FLO genes are particularly difficult to work with due to their highly repetitive nature and tandem repeats, high sequence homology, complex patterns of regulation, and high genetic instability (Stratford 1994;Bidard et al. 1995;Sato et al. 2001;Verstrepen et al. 2003;Fichtner et al. 2007;Liu et al. 2009;Van Mulders et al. 2010;Yue et al. 2013). Flocculation phenotypes also differ immensely between strains (Govender et al. 2008(Govender et al. , 2010. Further research that could be performed to determine why the RHA strains had no significant differences in fermentation rate include sporulating the R-FS · S-fS RHA hybrid and measuring the maximal fermentation rate at low temperature of segregating F 1 progeny containing the S288C flo1 deletion. Additionally, FLO5, FLO9, and FLO10 could be deleted in S288C to see whether these loci influence V max at low temperature. Potential differences between the maximal fermentation rate of flo1 haploids and diploids could also be investigated.

Conclusions
We have identified a QTL linked to V max and two QTL linked to lag phase in S. cerevisiae. Deletion of candidate genes confirmed that the gene on chromosome I linked to V max in S288C is FLO1, encoding a yeast flocculin. Deletion of FLO1 in the haploid S288C strain resulted in a large decrease in fermentation rate at 12.5°, but no change at 25°. A greater understanding of the role of the FLO family in stress tolerance will allow easier manipulation and/or selection of S. cerevisiae strains to improve V max and provide growth advantages during the low temperature fermentation of foods and beverages.
Sciences, University of Auckland) for processing microsatellite samples, and Andy Frost (Pernod Ricard NZ Ltd) for supplying the Sauvignon Blanc grape juice. Rebecca Deed was a recipient of the Bright Futures Top Achiever Doctoral Scholarship and the University of Auckland Doctoral Scholarship. This work was funded in part by a grant from the Foundation for Research Science and Technology in New Zealand (contract UOAX0404), with the support of New Zealand Winegrowers. The authors declare that they have no conflicts of interest.