Assays for flor formation, mat formation, and determination of cellular hydrophobicity

The flor formation in Z. rouxii was assayed by inoculating precultured cells in glass tubes containing 10 ml of YPD or YPD supplemented with 1.8 M NaCl. The mat formation was assayed by inoculating cells from a fresh colony onto YPD soft agar plates containing 1.8 M NaCl. The flor formation in S. cerevisiae was assayed by inoculating cells from a fresh colony into glass tubes containing 10 ml of SCG ura^- medium. The cells were grown at 28° for 3–7 days under static conditions and photographed. The exact incubation periods are written in the legend for Figures 1–6. The assays for the determination of the cellular hydrophobicity were performed as previously described (Barrales et al. 2008) with little modification. Briefly, the cells were grown in SCD containing 1.8 M NaCl, and then a portion of the cells was transferred to fresh medium. The cells were grown to an OD₆₀₀ of ~1.0, and then 1.2 ml of the culture was overlaid with 600 μl of octane and vortexed for 3 min. The OD₆₀₀ of the aqueous layer was recorded, and the relative difference compared with the initial OD₆₀₀ was used to determine the percentage of cellular hydrophobicity. The detailed colony morphology was observed using a digital microscope VHX-1000 equipped with VH-Z250R/Z250W (Keyence, Osaka, Japan).

Open in a separate window

Figure 1

Spontaneous flor on the surface of soy sauce mash and in manufactured soy sauce. (A) Spontaneous flor on the soy sauce mash in a concrete tank. The white pixels on the mash are the flor. (B) Spontaneous flor on manufactured soy sauce in glass bottles. (C) Appearance of the flor formation process. Z. rouxii Z3 was inoculated into soy sauce mash or soy sauce and incubated at 28° for 3 days. Bar, 30 μm.

Open in a separate window

Figure 6

Copy number variations of FLO11D in Z. rouxii. (A) PCR amplification of the FLO11D locus. (B) Flor formation in wild-type, single, double, and triple FLO11D gene mutants. A fresh colony was inoculated in 10 ml of YPD containing 1.8 M NaCl and incubated at 28° for 5 days. (C) Mat formation in wild-type, single, double, and triple FLO11D gene mutants. A fresh colony was inoculated onto YPD plates supplemented with 0.4% agar and 1.8 M NaCl and incubated at 28° for 7 days. (D) Karyotype profiles of Z. rouxii strains. The untreated chromosome (top) and NotI-digested chromosome (bottom) pattern was analyzed by Southern blot using the Flo11 domain-coding region of the FLO11D gene as a probe. (E) Summary of the loss and gain of FLO11D allele events. The tree shows the cluster based on the NotI-digested chromosome length polymorphism in D. Each event was manually mapped on the tree.

ura3 mutant construction in Z. rouxii NBRC1877 and Z3

We attempted to identify the URA3 gene coded on the P subgenome in Z. rouxii because of the requirement for the auxotrophic mutant in the allopolyploid of Z. rouxii. The genomes from ATCC2623 and ATCC46251 were digested with BglII, SpeI, and SacI, and the fragments were analyzed by Southern blot using the coding region of URA3 from ATCC2623 as a probe. The specific signal from URA3 coded on the P subgenome was detected at ~2.2 kb in the BglII-digested ATCC46251 genome. The band was recovered and cloned into the same site of pZEAK, which was constructed from pZEU (Pribylova et al. 2007b) by removing the URA3 marker genes and disrupting the BamHI recognition site derived from pSRI. S. cerevisiae BY4743 (a uracil auxotrophic mutant) was transformed with the resulting plasmid and selected on an SCD ura^- plate. The sequence analysis of the plasmid extracted from the transformants revealed the URA3 gene coded on the P subgenome (AB665419).

Z. rouxii NBRC1877 and the Z3 ura3 mutant were obtained through a two-step deletion of the ura3 gene (XM002499041, AB665419). The primers that were used in the gene disruption are indicated in Table S1. The removal of the G418 and Zeocin resistance markers from the genomes were performed as previously described (Pribylova et al. 2007b) with little modification. The cells were transformed with pZCRE, and the transformants were grown selectively on SCD ura^- plates for 2 days and then nonselectively in liquid YPD supplemented with 0.17% uracil (YPDU) for 24 hr. The cultures were diluted and plated on YPDU to obtain single colonies. Two days later, the colonies were replated on YPDU, YPDU + G418, YPDU + G418 + Zeocin, and SCD ura^- plates, and those clones with both G418- and Zeo-sensitive and Ura^- phenotypes were selected. The removal of the G418 and Zeocin resistance markers from the genome was confirmed using PCR analysis.

Plasmid construction

To identify the flor-forming gene through functional cloning, Z3 genomic libraries were prepared. The Z3 genomic DNA was partially digested with Sau3AI, and the resulting fragments were separated by agarose gel electrophoresis. The 6- to 9-kb DNA fragments were recovered using a PCR and Gel Purification Kit (Promega, Fitchburg, WI) and cloned into the BamHI site of pZEUneo, which was constructed from pZEU (Pribylova et al. 2007b) by disrupting the BamHI recognition site derived from pSRI. Escherichia coli DH5α cells were transformed with the resulting plasmids, and the transformants were selected by Luria broth medium containing carbenicillin. The resulting Z3 genomic library was extracted from these bacteria and used for functional cloning. The FLO11D gene was amplified by PCR from the Z. rouxii Z3 genomic DNA using the primers pFLO11D_F and pFLO11D_R to append the EcoRI and SalI digestion sites onto either end of the PCR product. The PCR product was digested with the respective restriction enzymes and ligated into EcoRI- and SalI-digested pZEU (Pribylova et al. 2007a) to generate pFLO11D. To drive FLO11D under the GAL1 promoter, pYFLO11D were constructed. Briefly, the coding sequence of FLO11D was amplified by PCR from the Z. rouxii Z3 genomic DNA using the primers pYFLO11D_F and pYFLO11D_R. The resulting DNA fragments and pYES2 (Life Technologies, Carlsbad, CA) digested with BamHI and HindIII were introduced into S. cerevisiae BY4743 and then plated onto a SCD ura^- plate. After the resulting plasmids were introduced into E. coli and were amplified, BY4743 and 133d flo11Δ were transformed by the plasmid. The PCR analyses were initiated using KOD plus neo (Toyobo, Osaka, Japan) according to the manufacturer’s instructions.

DNA sequencing, alignment, and analysis

All of the DNA sequences were obtained through outsourcing (Fasmac, Atugi, Japan). All of the BLASTP and PSI-BLAST searches (Altschul et al. 1997) applied an expected value cutoff of 10⁻⁵ at the hemiascomycete yeast genome database (Sherman et al. 2009; http://www.genolevures.org/) and through the National Center for Biotechnology Information. The protein sequences were aligned using ClustalX (Larkin et al. 2007). The initial formatting of the alignments was conducted in GeneDoc (http://www.psc.edu/biomed/genedoc). For the phylogenetic inferences, we used the neighbor-joining method (Saitou and Nei 1987) with 1000 bootstrap interactions. The tree was visualized in NJplot (Perrière and Gouy 1996). The predicted N-terminal signal peptides were detected with the SignalP 3.0 algorithm (Emanuelsson et al. 2007). The C-terminal GPI anchor attachment signals were detected using the fungal big-Π prediction algorithm (Eisenhaber et al. 2004).

Southern and Northern blot analyses

For the Southern blot analysis, the genomic DNA of yeast cells was extracted using standard protocols (Sambrook and Russell 2001) and digested with the appropriate restriction enzyme. The digested DNA was separated by 1% agarose gel electrophoresis and transferred onto a Hybond-N⁺ membrane (GE Healthcare, Buckinghamshire, UK) by upward capillary transfer.

For the Northern blot analysis, the total RNA was prepared with Isogen (NIPPON gene, Tokyo) according to the manufacturer’s instructions with little modification. Briefly, the total RNA was extracted after the cells were homogenized in a Micro Smash MS-100 bead beater (TOMY, Tokyo) at 3000 × g for 300 sec with adequate amounts of glass beads (0.1 mm in diameter) and Isogen solution. A 5-μg aliquot of the total RNA was run on a gel and blotted.

The Flo11 domain-coding region, which is associated with the Flo11p N-terminal region in the Pfam database (Finn et al. 2010), was used to probe the membranes. These experiments were performed using a PCR digoxigenin probe synthesis kit (Roche, Basel, Switzerland) and detection starter kit II (Roche) according to the manufacturer’s instructions.

Chromosomal DNA isolation and pulsed-field gel electrophoresis

The chromosomal DNA in plug was prepared according to a previous study with little modifications (Pribylova et al. 2007a). Instead of Lyticase, Zymolyase 100T (SEIKAGAKU Corporation, Tokyo) was used, and the cell treatment was performed overnight at 30°. The NotI-digested plug was prepared according to a previous study (Lim et al. 1994). The 1% agarose gels were prepared from pulsed-field certified agarose (Bio-Rad, Hercules, CA) and run on a CHEF apparatus (Bio-Rad) in 0.5× TBE buffer at 10°. The running conditions for chromosomal DNA separation were as follows: switch time, 350–400 sec; run time, 240 hr; angle, 120°; and voltage, 3 V/cm. The running conditions for NotI-digested chromosomal DNA separation were as follows: switch time, 60–120 sec; run time, 24 hr; angle, 120°; and voltage, 6 V/cm.

The DNA banding patterns were analyzed with Quantity One Basic (Bio-Rad) and manually. The cluster analysis of the band patterns was performed using SPSS version 11.5 with a word algorithm.

Osmotic stress-induced FLO11D expression under glucose-free conditions

Z. rouxii forms flor in a NaCl-dependent manner, and, to the best of our knowledge, this ability is a Z. rouxii-specific characteristic. We hypothesized that this unique feature was due to NaCl-dependent FLO11D expression. To test this hypothesis, Z3 was incubated with shaking in synthetic complete dextrose medium (SCD) and was then transferred to various conditions (SCD, SCD containing NaCl, SC, or SC containing NaCl). We prepared total RNA from these cultures after 30, 60, and 90 min and then measured FLO11D expression by Northern blot analysis. FLO11D-specific signals were detected only in SC containing 1.8 M NaCl (Figure 5A), suggesting that FLO11D is repressed by glucose and induced by NaCl under glucose-free conditions. To examine other carbon sources in the repression of FLO11D, FLO11D expression was measured by Northern blot analysis in SC containing 1.8 M NaCl and other carbon sources, such as galactose, ethanol, and glycerol. As a result, these carbon sources did not repress the FLO11D expression under our experimental conditions (data not shown).

Open in a separate window

Figure 5

FLO11D expression and flor formation under osmostress conditions. (A) The regulation of FLO11D expression. The cells were grown to an OD₆₀₀ of 1.0 in SCD, washed, and transferred to SCD, SCD containing 1.8 M NaCl, SC, or SC containing 1.8 M NaCl. After 30, 60, and 90 min, the total RNA was extracted and analyzed by Northern blot analysis using the Flo11 domain-coding region of the FLO11D gene as a probe. (B) Flor formation of Z3 in YPD medium or YPD medium containing 1.8 M of various chemicals (NaCl, KCl, NaNO₃, and sorbitol). A fresh colony was inoculated into the above-mentioned medium and incubated at 28° for 7 days. (C) FLO11D gene expression in response to various osmostress conditions. The cells were grown to an OD₆₀₀ of 1.0 in SCD, washed, and transferred to SC, SC containing 1.8 M NaCl, SC containing 1.8 M KCl, SC containing 1.8 M NaNO₃, or SC containing 1.8 M sorbitol. After 60 min, the total RNA was extracted and analyzed by Northern blot analysis using the Flo11 domain-coding region of the FLO11D gene as a probe.

To demonstrate whether FLO11D expression is specifically induced by NaCl, we observed flor formation in YPD medium containing various chemicals (NaCl, MgCl₂, KCl, MgSO₄, Na₂SO₄, NaNO₃, and sorbitol). Z. rouxii Z3 formed flor in medium containing 1.8 M NaCl, KCl, NaNO₃, and sorbitol after a 7-day incubation (Figure 5B) and in the medium containing 0.9 M MgCl₂, 1.8 M MgSO₄, and Na₂SO₄ after a 16-day incubation (Figure S4). The result of the Northern blot analysis clearly demonstrated that FLO11D was induced not only by NaCl but also by KCl, NaNO₃, and sorbitol under glucose-free conditions (Figure 5C). These results indicate that FLO11D is most likely induced by osmostress under glucose-free conditions. Consequently, flor formation can be observed in the late phase of culture, when the depletion of glucose occurs and Z. rouxii cells are subjected to osmotic stress.

FLO11D copy number variations

To confirm the requirement of FLO11D for flor formation in other genetic background strains, we constructed FLO11D gene mutants in the ATCC46251, NBRC0845, and 1501 strains. This analysis also showed that FLO11D played a role in flor formation, and there were FLO11D copy number variations (i.e., Z3, ATCC46251, NBRC0845, and 1501 possessed three, three, two, and one copies of FLO11D, respectively) (Figure 6, A–C).

Using a combination of pulsed-field gel electrophoresis (PFGE) and Southern blot analyses, we investigated how many FLO11D allele amplification events occurred in Z. rouxii. With respect to the NBRC1877, 1501, NBRC0845, ATCC46251, Z3, and Z1 strains, zero, one, one, three, two, and three copies of FLO11D, respectively, were detected by chromosomal PFGE analysis, whereas zero, one, two, three, three, and three copies, respectively, were detected by NotI-digested chromosomal PFGE analysis (Figure 6D). The latter analysis is in agreement with the FLO11D gene disruption experiment, indicating that the NBRC1877, 1501, NBRC0845, ATCC2623, Z3, and Z1 strains possess zero, one, two, three, three, and three copies of the FLO11D allele, respectively. These results also indicate that the first and second FLO11D alleles are located on either the same or different chromosomes, which have the same mobility in NBRC0845 and Z3. Interestingly, the mobility of the third FLO11D allele of Z3 was different from that of ATCC46251 and Z1, indicating that either the acquisition of the third FLO11D allele occurred independently two times or a sequence alteration in the third FLO11D allele occurred in Z3. We next compared the NotI-digested karyotype profiles of various Z. rouxii strains by cluster analysis, and the FLO11D amplification events were mapped on the tree (Figure 6E). The loss or gain of FLO11D allele events can be explained by the dendrogram, which suggested that the FLO11D allele originated from a single origin and that sequence alteration in the third FLO11D allele occurred in Z3 after its acquisition.

Regulation of FLO11D

The osmostress-dependent expression of FLO11D is very unique. In S. cerevisiae, FLO11 is activated by nitrogen and carbon depletion via the MAP kinase-dependent filamentous growth pathway and the glucose repression pathway (Madhani and Fink 1997; Rupp et al. 1999; Gagiano et al. 2002; Kuchin et al. 2002; Vyas et al. 2003; Schwartz and Madhani 2004). Therefore, the formation of flor can be considered as an adaptive mechanism to nutrient starvation (Zara et al. 2011) because it ensures access to oxygen and therefore permits continued growth on nonfermentable ethanol. The glucose repression pathway between S. cerevisiae and Z. rouxii could be common because not only FLO11 expression but also FLO11D expression was repressed by glucose, suggesting that the flor formation is important to the adaption to nutrient starvation in Z. rouxii. Interestingly, a previous study showed that hyperosmotic stress inhibits the development of the fluffy colony morphology in S. cerevisiae (Furukawa et al. 2009), which are the opposite results from ours (Figure 2C). This fact raises the question of why the FLO11D-dependent morphological change is induced by osmostress in Z. rouxii. We hypothesize three possible reasons: (1) the FLO11D gene is under the HOG1 MAP kinase pathway, (2) there is signaling cross talk between the filamentous growth MAP kinase pathway and the HOG1 MAP kinase pathway because the two pathways share common components (Ste11p, Ste50p, Ste20p, and Cdc42p) in their signal transduction (Chen and Thorner 2007), or (3) there is an unknown pathway. To assess the possibility of hypothesis 1, we constructed the hog1Δ strain under the Z3 background and compared the expression level of FLO11D between Z3 and hog1Δ. As a result, FLO11D gene expression was observed under osmostress conditions in both Z3 and hog1Δ (data not shown). These results disprove hypothesis 1. We therefore infer that hypothesis 2 or 3 may contribute to the osmostress FLO11D expression. However, we cannot exclude the possibility that FLO11D activation in hog1Δ under osmostress conditions is an artificial consequence because hyperosmotic shock induces inappropriate activation of the MAP kinase-dependent filamentous growth pathway in S. cerevisiae hog1Δ (Davenport et al. 1999). Clearly, a more detailed analysis is needed to elucidate the pathway required for NaCl-dependent FLO11D expression in Z. rouxii. We speculate that the osmo-ergic FLO11D expression in Z. rouxii would be important for ecological niche segregation among yeast. For Z. rouxii, the adaptation to nutrient starvation thorough the formation of flor in a non-osmostress environment would not be effective for the expansion of Z. rouxii’s natural habitat because its competitor, S. cerevisiae, has advantages in growth rate in this type of environment. Conversely, the adaptation to nutrient starvation by forming flor observed in Z. rouxii under osmostress conditions would ensure its increased survival and thereby permit its occupancy of this type of niche.

We thank Jose I. Ibeas for the yeast strains, two referees for their constructive comments, and the members of the soy sauce lab for fruitful discussions.