Biodiversity of non-Saccharomyces yeasts associated with spontaneous fermentation of Cabernet Sauvignon wines from Shangri-La wine region, China

Shangri-La is a wine region that has the highest altitude vineyards in China. This is the first study investigated the biodiversity of non-Saccharomyces yeasts associated with spontaneous fermentation of Cabernet Sauvignon wines produced from two sub-regions (Lancang River and Jinsha River) of Shangri-La. The culturable yeasts were preliminarily classified based on their colonial morphology on the Wallerstein Laboratory nutrient agar plates. Yeast species were identified by the sequencing of the 26S rRNA D1/D2 region and the 5.8S rRNA ITS region. Twenty-five non-Saccharomyces yeast species belonging to sixteen genera were isolated and identified in Shangri-La wine region. Candida, Hanseniaspora, Pichia, and Starmerella were found in both sub-regions, but the Lancang River showed more diverse yeast species than the Jinsha River. Shangri-La not only exhibited high diversity of non-Saccharomyces yeasts, and furthermore, seven species of non-Saccharomyces yeasts were exclusively found in this region, including B. bruxellensis, D. hansenii, M. guilliermondii, S. vini, S. diversa, T. delbrueckii and W. anomalus, which might play an important role in distinctive regional wine characteristics. This study provide a relatively comprehensive analysis of indigenous non-Saccharomyces yeasts associated with Cabernet Sauvignon from Shangri-La, and has significance for exploring ‘microbial terroir’ of wine regions in China.

Wine fermentation is a complex biochemical process conducted by many different microorganisms, and yeasts play a major role in this process 1 . Although wine could be produced spontaneously by wild yeasts on the surface of grapes 2 , inoculation of commercial Sacchromyces cerevisiae yeast is the most common operation in current wine production in order to avoid several potential problems (e.g. sluggish fermentation) and to achieve final products with uniform quality 3 . According to some opinions, the resulting wine, however, is more like 'industrial' products and loses its 'natural' property such as diversity and distinctive characteristics 2 .
To produce wine with distinctive characteristics, researchers and enologists tried to inoculate non-Saccharomyces yeasts during alcoholic fermentation [4][5][6][7] . In these studies, non-Saccharomyces yeasts were either coinoculated or sequentially inoculated with S. cerevisiae yeast, and the resultant wines generally exhibited more varietal characteristics and distinctive sensory attributes. In addition, non-Saccharomyces yeasts have also been used in the production of other beverages, such as beer 8 and spirit 9,10 . Although a few non-Saccharomyces yeasts are commercially available, there is an increasing interest in exploring non-Saccharomyces yeasts worldwide and their potential usage in wine production.
Some research have demonstrated that indigenous yeasts are an important part of the 'terroir' in different wine producing regions around the world [11][12][13][14] . The density and diversity of those indigenous yeasts on grape are closely related to numerous factors, such as the variety and maturity of grape, geographical location and climatic conditions of vineyard, and the practices of viticulture 11 . China has a fast-growing wine industry which includes

Results
Oenological parameters in grape juice and spontaneously fermented wine. The oenological parameters in Cabernet Sauvignon grape juice and spontaneously fermented wines from Shangri-La were shown in Table 1. Grapes from each vineyard exhibited various ripeness with the reducing sugar concentration ranging from 186.74 to 265.95 g/L, total acidity ranging from 3.42 to 8.55 g/L, and the pH ranging from 3.26 to 3.97. After spontaneous fermentation, most of wines were fermented to dryness with residual sugar concentration less than 5.00 g/L, except for wines made from L-XD (10.91 g/L) and J-BZL (14.14 g/L). Due to the variations in ripeness of grapes and the residual sugar in wines, the alcohol content of resultant wines ranged between 10.79 and 15.46%. There was an increase in the total acidity during the spontaneous fermentation. The concentration of volatile acidity varied significantly among the wines ranging from 0.60 to 1.92 g/L. The pH of final sample wines varied from 3.38 to 3.84.
Yeast isolation and identification. The detailed information of 26S rRNA D1/D2 region and 5.8S rRNA ITS region of representative isolates were listed in Table S1. Most yeast isolates were identified at high similarity (> 99.00%) of D1/D2 and ITS sequences with the corresponding type strains. The reliable identification results were also confirmed by phylogenetic analysis (Figs. 1, 2). Although a high similarity was shown in BLAST results of 26S rRNA D1/D2 region, certain differences were observed in 5.8S rRNA ITS region among several isolates and related type strains, including L-SN-25, L-AD-7, L-AD-17, L-XD-23 and L-NT-69 (Table S1). As no type strain is available for Metschnikowia fructicola, the identification of L-AD-7 was based on BLAST analysis of 26S rRNA D1/D2 region of GXZJD32 (KC160603.1) and 5.8S rRNA ITS region of AP47 (FJ919773.1) ( Table S1).

Biodiversity of non-Saccharomyces yeasts in Shangri-La region.
A total of 2168 yeast isolates were collected during spontaneous fermentation of Cabernet Sauvignon wines from seven samples within the Lancang and Jinsha River Basins of Shangri-La wine region. As shown in Table 2, twenty-five different non-Saccharomyces yeast species belonging to sixteen genera were identified by colonial morphology and molecular methods. The identified yeast species include Aureobasidium pullulans, Brettanomyces

Discussion
Compared to most wine regions of China 19 , the Cabernet Sauvignon harvested from Shangri-La exhibited greater maturity degree, which was related to the ideal viticulture environment of this region16. In most spontaneously fermented wines, the concentration of volatile acidity were much higher than acceptable level (0.7 g/L) 22 . Indeed, most non-Saccharomyces yeasts [23][24][25][26] and some wild S. cerevisiae yeasts 19,25 may produce excess volatile acidity during spontaneous fermentation. While a marked increase of total acidity in final wines suggested that some yeast species being found in present study could be used for improving the wine contained insufficient acidity 27 .
Within the sixteen non-Saccharomyces yeast genera identified in Shangri-La region, Hanseniaspora and Starmerella were the most frequent isolates. The predominant species of these two genera were H. uvarum and S. bacillaris (C. zemplinina), which is in agreement with the findings of previous research 11,12,18,28 . These two species were commonly found in grape and wine, thus more attention has been paid in vinification research and application [29][30][31][32] In addition, four members of Hanseniaspora and Starmerella also existed in Shangri-La region, namely H. opuntiae, H. vineae, S. apicola (C. apicola) and S. stellata (C. stellata).
The Pichia is another abundant genus associated with spontaneous fermentation of Cabernet Sauvignon in Shangri-La region. Among the four Pichia species isolated, P. occidentalis was the most abundant species. Interestingly, P. occidentalis had also been found in Roussanne that grown in Beijing 17 . While for other wine regions of China, the widely distributed species of this genus was Pichia fermentans 18,21 . Several studies indicated that some Pichia species, such as P. kluyveri, P. kudriavzevii(I. orientalis), P. terricola(I. terricola) and Pichia fermentans have the potential to enhance wine aroma by releasing glycosidically bound aroma precursors from grape must 4,33 or Table 2. Frequency of non-Saccharomyces yeast isolates from Cabernet Sauvignon spontaneous fermentation of Shangri-La wine region. '−' means yeast species was not detected in present work. www.nature.com/scientificreports/ producing volatile compounds during fermentation 5,34,35 . But there is little information about P. occidentalis (I. occidentalis) in wine production, which might be worth further study. Two Metschnikowia species, including M. fructicola and M. pulcherrima, have been isolated in the samples of L-AD and L-SN, respectively. M. pulcherrima was the common species that had been found on Cabernet Sauvignon among most wine regions of China [18][19][20][21] . Although M. fructicola had previously only been isolated on Roussanne in Beijing 17 , the present study indicated that it may also associated with Cabernet Sauvignon that grown in China. According to studies conducted by Floriana Boscaino 36 , Elena González-Royo 6 and C. Varela 37 , the usage of M. fructicola or M. pulcherrima during fermentation could improve the aromatic profile of resulatnt wine.

Yeast species L-NT L-XD L-LTJ L-SN L-AD J-BZL J-DR Representative isolates
In this study, P. flavescens (C. flavescens) was only accounted for a small proportion in the sample of L-LTJ. Although a previous study reported that P. flavescens (C. flavescens) was found as a common yeast species isolated from Cabernet Sauvignon, Merlot and Chardonnay in Shacheng, Changli, Wuwei and Penglai of China 18 , this species seems unlikely associate with grape due to the low frequency observed in this study, which was in agreement with the view of Brysch-Herzberg 12 .
A. pullulans (so called black-yeast) and R. glutinis, which have the capacity of pigment metabolism, were found in present work. A. pullulans is one of the most well-adapted saprophytes on grape berries 20 , however the distribution of A. pullulans was not very widespread in Shangri-La as well as other regions of China [18][19][20][21] . It was only detected in sample of L-SN and exhibited a low isolation frequency. It has been reported that various enzymes (e.g. pectinase, cellulases and β-glucosidase) produced by A. pullulan could be beneficial to the flavor of wine 38 . Additionally, due to antifungal and antibacterial activity, this species might be used to against some biological diseases of vine 39 . Although R. glutinis were accidentally isolated in two samples of Shangri-La, Rhodotorula sp. are generally considered as typical phyllospheric yeasts 40 and thus rarely associated with yeast community of grape berries in China [18][19][20][21] . Although some Rhodotorula sp. can produce β-glucosidase and α-larabinofuranosidase to release bound aroma precurcors 41,42 , they are rarely used in wine production.
Zygosacchromyces bailii was found in the sample of L-NT. It is one of the most dangerous wine spoilage species which are known to produce off-flavors and form cloudiness even under high alcohol condition 43,44 . Although Z. bailii exhibited low isolation frequency in Shangri-La, some precautions should be taken in wine production.
There were seven non-Saccharomyces yeast species which showed regional characteristics were exclusively found in Shangri-La wine region, including B. bruxellensis, D. hansenii, M. guilliermondii, T. delbrueckii, W. anomalus, S. vini and S. diversa. To the best of our knowledge, this is the first time that these non-Saccharomyces yeast species have been isolated from wine regions of China [17][18][19][20][21] . T. delbrueckii is one of the commercialized non-Saccharomyces yeast for wine production. Several positive influences of T. delbrueckii, such as enhancing the fruity aroma style, increasing glycerol concentration, reducing volatile acidity and improving foam properties, have been confirmed in the production of grape 6,45 and other fruit wines 46,47 . In this study, J-BZL-163 (T. delbrueckii) was isolated from the plateau area with unique environmental conditions. Therefore, further research will be required to evaluate the differences of fermentation characteristics between J-BZL-163 and commercial T. delbrueckii strain. Although B. bruxellensis is generally considered as spoilage microorganisms, it could also contribute distinctive attributes in spirits 9 . W. anomalus and D. hansenii have also been tested in wine-related research and shown positive effects on wine aroma quality 48,49 . Although M. guilliermondii, S. vini and S. diversa have been found in wine 50 , beer 51 and some fruits 52,53 , little study has been conducted for fermentation characteristics of these yeasts in wine or other alcoholic beverages.
Compared with other wine regions of China that have been reported [17][18][19][20][21] , Shangri-La wine region exhibited greater overall diversity of non-Saccharomyces yeast species. Within Shangri-La region, J-BZL showed more non-Saccharomyces yeast diversity than other vineyards, which is most probably related to its distinctive grapegrowing environment. Compared to other six vineyards in Shangri-La region, the vineyard of J-BZL is very close to the bank of Jinsha River, which provides a high humidity environment ideal for epiphytic microorganism growth on grapes 54 . As for L-LTJ and L-SN which located at the foot of Meili Snow Mountain, a rapid decline in temperature during mature period (October) could be one of the possibilities that reduce the diversity of non-Saccharomyces yeasts present on their grapes. According to França's research, soil samples that came from low temperature area exhibited less density of yeast population. Furthermore, low temperature may also make a few yeast become the dominated culturable species and reduce the chance that other species will be isolated 55 . These results confirmed that the geographic locations and climate conditions have great impact on yeast communities 13 and provided more evidence that indigenous yeasts should be considered as an important part of 'terroir' 14 .

Methods
Grape samples and spontaneous fermentation. Cabernet Sauvignon (Vitis vinifera L.) grape samples were collected from seven vineyards (Shangri-La Winery Co., Ltd. Yunnan) located at different altitudes within two sub-regions (Lancang River and Jinsha River) of Shangri-La wine region in Yunan province. Grape samples were harvested by hand with sterile gloves in September and October in 2017. After harvest, grape samples (10 kg) from each vineyard were stored in sterile bags at 4 °C and transported to the laboratory in Kunming within 24 h. The geographical information and the location of sampling vineyards was summarized and shown in Table 3 and Fig. 3, respectively.
Grape samples of each vineyard were destemmed and crushed by hand with sterile gloves, and the must was fermented in two sterilized 5 L glass bottles at 28 °C for 12-18 days. During the fermentation, cap management was carried out once a day until the end of fermentation. Oenological parameters in Cabernet Sauvignon grape juice (reducing sugar concentration, total acidity, and pH) and spontaneously fermented wines (residual sugar concentration, total acidity, volatile acidity, pH and alcohol content) were measured according to the National www.nature.com/scientificreports/ of alcohol was determined by density method. The concentration of total acidity was determined by potentiometric titration using standard sodium hydroxide. The separation of volatile acidity from wine was carried out by steam distillation. The concentration of volatile acidity was titrated by standard sodium hydroxide. The value of pH was determined by pH meter.

Isolation of non-Saccharomyces yeasts.
For the isolation of non-Saccharomyces yeasts, fermenting samples (10 ml) were collected every 3 days from each fermentation bottle. Due to the variation in the fermentation duration, there were 4 fermenting samples collected from L-LTJ, L-SN; 5 fermenting samples collected from L-NT, L-XD and J-DR; 6 fermenting samples collected from L-AD and J-BZL. The fermenting samples were diluted into 1:10 3 to 1:10 5 ratios with sterile physiological solution (0.85%, NaCl), plated on the Wallerstein Laboratory (WL) nutrient agar (Qingdao Hope Bio-Technology Co., Ltd.) and incubated at 28 °C for 5 days until colonies showed morphological differences. About 10 to 15 colonies showing different colonial morphology were selected from each WL nutrient agar plate, and then re-streaked on the YEPD agar plate (yeast extract 10 g/L, tryptone 20 g/L, glucose 20 g/L, agar 20 g/L, and chloramphenicol 100 mg/L) for purification. There were 2168 pure yeast cultures in total isolated and stored at 4 °C for further analysis. Glycerol (15% v/v) was added into yeast culture for long-term storage at -80 °C.
Non-Saccharomyces yeasts identification. Yeast isolates were preliminarily classified according to the morphology (color, shape, consistency, and size) of their colonies on the WL nutrient agar plate, and a total of 204 isolates were chosen for molecular identification. Genomic DNA was extracted from three colonies of each isolates by using Rapid Yeast Genomic DNA Isolation Kit (Shanghai Sangon Biotech Co., Ltd.) following the manufacturer's instructions. The genomic DNA extracted was identified by analysis of the sequence similarity of the 26S rRNA D1/D2 region and 5.8S rRNA ITS region. The primers for amplification of 26S rRNA D1/D2 region were NL1 (5′-GCA TAT CAA TAA GCG GAG GAA AAG -3′) and NL4 (5′-GGT CCG TGT TTC AAG ACG G-3′) 56 , and for amplification of 5.8S rRNA ITS region were ITS1 (5′-TCC GTA GGT GAA CCT GCG G-3′) and ITS4 (5′-TCC TCC GCT TAT TGA TAT GC-3′) 57 . Table 3. The geographical information of sampling vineyards from Shangri-La wine region. *In sample ID, Land J-indicates sample collected from Lancang River and Jinsha River in Shangri-La wine region, respectively.  The PCR products were purified and sequenced by Biomed gene technology Co., Ltd (Beijing). The sequencing results were submitted and compared with those of corresponding type strain (listed in Supplementary Table S1) in the Blast service of the National Center for Biotechnology Information (NCBI http://www.ncbi. nlm.nih.gov/blast ).
Phylogenetic analysis was performed to verify the identification of representative isolates. The phylogenetic tree was developed using the Maximum Likelihood methods and the Kimura 2-parameter model by MEGA 7.0: Molecular Evolutionary Genetics Analysis version 7.0 for bigger datasets (Kumar, Stecher, and Tamura 2015). Bootstrap values were calculated from 1000 iterations. Statistical analysis. All the analyses were performed in triplicate. One-way analysis of variance (ANOVA) was used to analyze the difference of grape and wine samples performed through SPSS 19.0 (IBM Corp., Armonk, New York, U.S.A.) employing Duncan multiple range tests at a significance level of p < 0.05. The results were expressed as the mean value ± the standard deviation.