For the isolation of yeasts, we obtained samples from 12 stations: at Stations 1 - 3 from the midstream of Tama River on July 12, 2017; at Stations 4 - 6 from the downstream of Tama River on May 11, 2017; at Stations 7 - 9 from the estuary of Tama River in Tokyo on August 21, 2017; and at Stations 10 - 12 from Lake Kasumigaura in Ibaraki Prefecture on October 11, 2017. To obtain these samples, we collected mud sediment from the aquatic environments with the use of an Ekman-Birge Grab sampler (Miyamoto Riken Industry Co., Osaka, Japan) except at Stn. 9, where we collected seawater. The sample was collected from bottom of depth 1 - 10 m at each sampling site and immediately transferred to sterilized plastic bottles that were maintained at low temperature while they were carried to our laboratory at the Tokyo University of Marine Science and Technology. Each sample was diluted to 10 or 100 times with saline. Aliquots (200 μL) of diluted samples were spread on YPD solid medium (D-glucose, 20 g/L; peptone, 20 g/L; yeast extract, 10 g/L; chloramphenicol, 0.2 g/L; agar, 20 g/L) and incubated for approx. 72 hrs at 25˚C. Each growing yeast-like colony was picked up, and the isolation of its microorganisms was performed. We also measured the pH with HORIBA Scientific F-71S (HORIBA, Ltd., Kyoto, Japan), NaCl concentration with HORIBA Scientific B-721 (HORIBA, Ltd.) and water temperature of each sampling area.

We assayed the fermentative and tolerant activities of each yeast strain under sorbitol and ethanol with a five-step screening, as follows. 1) The yeast fermentation activity with YPD liquid medium under stress-free conditions was assayed; 2) Using strains selected in the first screening, we evaluated the effect of sugar osmotic stress on the fermentative yeasts in experiments using the YPD liquid medium containing 30% (w/v) sorbitol; 3) Using strains selected in the second screening, we determined the effect of ethanol on the fermentative yeasts by using the YPD liquid medium containing 10% (v/v) ethanol; 4) Using strains selected in the third screening, we investigated the effects of sorbitol and ethanol on the fermentative yeasts by using the YPD liquid medium containing 20% (w/v) sorbitol and 10% (v/v) ethanol; 5) With the strains selected in the fourth screening, we used the YPD liquid medium containing 30% (w/v) sorbitol and 10% (v/v) ethanol in fermentation experiments.

The fermentation was carried out in test tubes covered with aluminum caps for 2 weeks at 25°C. The screening of fermentative yeasts was conducted using inverted Durham pipes for the visualization of the CO₂ production.

The 26S rDNA D1/D2 domain of each highly fermentative yeast identified as described above in Section 2.2 were amplified by polymerase chain reactions (PCRs) using the forward primer NL-1 (5’-GCATATCAATAAGCGGAGGAAAAG-3’) and the reverse primer NL-4 (5’-GGTCCGTGTTTCAAGACGG-3’), and Premix Ex Taq (Takara Bio, Shiga, Japan). The PCR program was performed with an initial denaturation at 95˚C for 5 min, followed by 25 cycles of denaturation at 94˚C for 30 s, annealing at 56˚C for 30 s, extension at 72˚C for 1 min and final extension at 72˚C for 4 min in a TaKaRa PCR Thermal Cycler Dice. Sequences were analyzed using the BLAST function at the U.S. NCBI website (http://www.ncbi.nlm.nih.gov/blast). The D1/D2 domain sequences of the 26S rDNA genes in the yeasts were deposited in DDBJ.

We assessed the effect of salinity on the fermentative yeasts identified as described above in Section 2.3. The YPD liquid medium containing 5.0%, 10.0%, 15.0%, or 20.0% (w/v) NaCl was used in the fermentation experiments. Fermentation was carried out for 2 weeks at 25°C, and the fermentative yeasts were screened using the same method as that described in Section 2.2.

We investigated the ethanol productivity of each of the identified yeasts. Each yeast strain was aerobically precultured with stirring at 120 rpm for 48 hrs at 25˚C in 10 mL of YPD liquid medium. The growing yeast was centrifuged at 3000 rpm for 5 min, and its precipitate was washed with saline. The centrifugation/washing procedure was conducted three times, and the cells were obtained. Then, 0.3 g of yeast pellet was anaerobically cultured for 20 hrs at 25˚C in 10 mL of YPD liquid medium with the use of an anaero Pack System (Mitsubishi Gas Chemical Co., Tokyo, Japan). The ethanol production of each yeast strain was determined by an Ethanol F-kit (Roche Diagnostics, Indianapolis, IN, USA).

Molasses was purchased from Alishan Co. (Tokyo, Japan). We evaluated the sugar consumption and the ethanol productivity of each species’ representative yeast from diluted molasses. Representative strains were selected based on the results of the procedures described in Sections 2.4 and 2.5. Each yeast was aerobically precultured for 48 hrs at 25˚C, with stirring at 120 rpm in 10 mL of YPD liquid medium including 1.0 M glucose. The growing yeast was centrifuged at 3000 rpm for 5 min, and its precipitate was washed with saline. The centrifugation/washing procedure was conducted three times, and the cells were obtained. Then, 0.3 g of yeast pellet was anaerobically cultured for 48 hrs at 25˚C in 25% (w/v) diluted molasses with an anaero Pack System. The sugar consumption and ethanol production of the selected yeasts were determined with a Sucrose/D-glucose/D-Fructose F-kit (Roche Diagnostics) and Ethanol F-kit (Roche Diagnostics).

We selected two sampling sites, Tama River and Lake Kasumigaura, where a large quantity of domestic wastewater flowed into, for the isolation of a variety species of yeasts. Figure 1 illustrates the sampling sites of Tama River and its tributaries flowing between Tokyo and Kanagawa Prefecture and the sampling sites of Lake Kasumigaura in Ibaraki Prefecture. We collected mud sediment in the aquatic areas at Stns. 1 - 8 and 10 - 12 and seawater at Stn. 9. Table 1 presents the values of pH, NaCl concentration, and water temperature at each sampling area and the isolation of yeast-like colonies. The pH ranges of the samples were around 7 - 8, and the highest pH values were observed at Stn. 3; the branch junction of the river and Stn. 12; and in the lake (pH 8.80 - 8.81). With the sampling from 12 sites, we isolated 1028 yeast-like strains.

Table 1 provides the numbers of all of the isolated yeast strains and the fermentative strains in each medium. With the first screening of strains from Stns. 1 - 3 with YPD liquid medium, 482 strains were isolated as yeast-like microorganisms, and 97 strains were isolated as fermentative microorganisms. From Stns. 4 - 6, 77 strains among all 190 isolates were fermentative. From Stns. 7 - 9, 113 strains of the 171 isolates were fermentative. From Stns. 10 - 12, 125 strains of the 185 isolates were fermentative. Thus, 412 strains from all 12 of the stations were fermentative.

The second screening with YPD liquid medium containing 30% (w/v) sorbitol was carried out, and 377 strains (from all of the stations) were fermentative. These results revealed that approx. 91.5% of the fermentative strains identified by the first screening showed tolerance against 30% (w/v) sorbitol.

With the third screening with the medium containing 10% (v/v) ethanol, only 83 strains (approx. 20.1% of the strains in the first screening) exhibited tolerance against ethanol stress. Eventually, by the fifth screening, 31 strains (approx. 7.5% of the strains from the first screening) were confirmed as having tolerance against both 30% (w/v) sorbitol and 10% (v/v) ethanol.

We further analyzed the 31 selected yeast strains and confirmed the six species of these strains. The largest number of strains (n = 10) was identified as Wickerhamomyces anomalus: AN1-46 (from Stn. 7); AN2-9, AN2-56, and AN2-64 from Stn. 8; KS28, KS51, and KS75 from Stn. 11; and KN8, KN9, and KN43 from Stn. 12. W. anomalus was previously isolated as an alkali-tolerant yeast in our laboratory [16], and in the present study, we isolated W. anomalus from Stn. 7 (pH 7.93), Stn. 8 (pH 7.81), Stn. 11 (pH 7.89), and Stn. 12 (pH 8.81).

Eight yeast strains were identified as Torulaspola delbruekii: AS183 from Stn. 1, and F2-2, F2-11, F2-24, F2-36, F2-41, F2-63, and F2-82 from Stn. 5. As shown in Table 2, both the W. anomalus strains and the T. delbrueckii strains exhibited tolerance against 15% (w/v) NaCl.

Six strains were identified as Candida glabrata: TC18 and TC140 from Stn. 2; AN1-25 from Stn. 7 and AN2-49, AN2-54, and AN2-58 from Stn. 8. Four of these strains (TC18, AN2-49, AN2-54 and AN2-58) also had high tolerance against 15% (w/v) NaCl. Although C. glabrata is known to be evolutionally very closely related to baker’s yeast (which has high fermentation activity under salt pressure), C. glabrata is a pathogenic yeast, and it was thought to need special handling [17].

Four strains were identified as Pichia kudriavzevii: TC47 and TC106 from Stn. 2 and AN1-13 and AN1-19 from Stn. 7. Strain H28 (from Stn. 3) was identified as Saccharomyces cf. cerevisiae/paradoxus, and strains F2-67 (from Stn. 5) and

++: High fermentation activity, +: Fermentation activity, −: Not detected fermentation activity. High fermentation activity: Inverted Durham tubes were filled with gas within 3 days. Fermentation activity: Confirmed gas visualized within 2 weeks.

S11 (from Stn. 10) were identified as Lachancea kluyveri. Among these yeasts, S. cf. cerevisiae/paradoxus, P. kudriavzevii and L. kluyveri had no tolerance against 15% (w/v) NaCl. None of the yeasts showed tolerance against 20% (w/v) NaCl.

We conducted the fermentation of the 31 selected strains in YPD liquid medium as shown in Figure 2. Strain F2-11 showed the highest ethanol production among the eight strains of T. delbrueckii: 14.80 g/L. Of the 10 strains of W. anomalus, the highest ethanol production was by strain AN2-56 at 13.86 g/L. Of the two L. kluyveri strains, F2-67 provided the highest ethanol production; 12.46 g/L. Of the six strains of C. glabrata, TC18 had the highest ethanol production at 12.55 g/L. Of the four P. kudriavzevii strains, AN1-13 showed the highest ethanol production; 11.26 g/L. The S. cf. cereivisae/paradoxus strain H28 produced ethanol at 7.61 g/L, which seemed to be relatively lower than those of the other species. We thus selected the strain with the highest productivity in each species (except for W. anomalus) and conducted the next fermentation experiment using molasses as described above in Section 3.5. In the case of W. anomalus, we used strain AN2-64 as described in Section 3.5 because the NaCl tolerance of strain AN2-64 was higher than that of strain AN2-56.

We conducted fermentation experiments with each representative strain of the six species in 25% (w/v) diluted molasses with the following sugar composition: sucrose 104.0 g/L, fructose 33.4 g/L, and glucose 24.8 g/L. Figure 3 illustrates the changes in the sugar concentration by each representative strain of the six species before and after fermentation in the molasses. For the S. cf. cereivisae/paradoxus strain H28, the L. kluyveri strain F2-67, the T. delbrueckii strain F2-11, and the W. anomalus strain AN2-64, the level of sucrose in the molasses decreased remarkably; the activities of invertase secreted from their cells would be high. Most notably, in S. cf. cereivisae/paradoxus H28 and L. kluyveri F2-67, most of the glucose and fructose were consumed by these yeast strains, and thus the fermentation activity may be high.

Figure 4 shows the ethanol production by each representative strain of the six species after the fermentation in molasses. Among the six strains, the ethanol production by two was high in proportion to the sugar consumption: S. cf. cereivisae/paradoxus H28 produced ethanol at 57.4 g/L, and L. kluyveri F2-67 produced ethanol at 53.9 g/L.