Introduction

Cider is an alcoholic beverage produced by the fermentation of apple juice. In many countries such as Spain, France, and Ireland, this production is still performed by natural fermentation involving wild yeasts (Coton et al. 2006; Cousin et al. 2017). These yeasts constitute a complex population that includes species different to those responsible for wine-must fermentation (Cousin et al. 2017; Morrissey et al. 2004). Saccharomyces, the most important fermentative yeast, and other genera such as Hanseniaspora, Metschnikowia, and Brettanomyces as well as other “transitory” species, are involved in apple juice fermentation and influence the aromatic complexity of cider (Cousin et al. 2017). Investigations performed in the last few years have raised the importance of the biodiversity study in apple cider ecology (Coton et al. 2006; Morrissey et al. 2004; Suárez Valles et al. 2007). Interestingly, Glushakova and Kachalkin (2017) identified in apple fruit basidiomycetous and ascomycetous yeasts never before isolated in such an environment.

The species composition of yeast consortia may be different in various geographical regions and contribute to specific organoleptic properties of cider. Important volatile molecules (e.g., higher alcohols and esters) derived by yeast metabolism allowed to discriminate ciders produced in two different regions of Northern Spain and France (Haider et al. 2014; Picinelli Lobo et al. 2016). Therefore, yeasts microbiota may significantly affect the quality and typicity of cider produced in a specific location.

The identification of yeasts from apple juice and cider has been mainly performed by restriction fragment length polymorphism analysis of internal transcribed sequences (ITS PCR-RFLP) and comparison of D1/D2 region of large subunit (26S) ribosomal DNA to a database (e.g., BLAST in GenBank gene database) (Coton et al. 2006; Morrissey et al. 2004; Pando Bedriñana et al. 2010; Suárez Valles et al. 2007; Graça et al. 2015). The use of nucleotide sequence database suitable for phylogenetic analysis has provided useful information to identify yeasts, as well as the redefinition of genera understanding their ecology (Kurtzman and Robnett 1998; Kurtzman et al. 2015). The phylogenetic analysis performed on yeasts from cider has been crucial when describing new species (Kurtzman et al. 2001).

In the present study, Saccharomyces and non-Saccharomyces yeasts from apple juice for Italian artisanal cider production were identified using LSU and/or ITS sequences. Comparative analysis on nucleotide databases and phylogeny reconstruction were carried out. This work was aimed at improving knowledge on yeast diversity in this environment, which has received little attention when compared to wine production.

Materials and methods

Sampling and yeast isolation

The yeasts were isolated from apple juices for the cider production sampled in two cider houses (i.e., CHA and CHB) from Northern Italy. CHA is located in a plain area and CHB in an alpine area and apples used for cider production are cultivated in the respective areas. Both cider-making systems were artisanal and juices were obtained by means of a batch mechanical press of mixture of acidic, bitter, and sweet apples. The main varieties used in both cider houses were Golden and Red Delicious, Reinette, Fuji, Granny Smith, and Gala. Three different samples of apple juices (CHA 1–3 and CHB 1–3) for the cider-making process were collected during the spring of 2017 from each cider house. Apple juices collected immediately after the pressing were placed in sterile bottles. All the samples were kept cool during transport to the laboratory then were directly processed. The samples were diluted in ten-fold series in physiological water (0.9% NaCl) and aliquots (0.1 mL) were plated on Wallerstein Laboratory Nutrient agar (WL, Sigma-Aldrich, Saint Louis, MO) supplemented with chloramphenicol 100 mg/L (Sigma-Aldrich) to inhibit bacterial growth. This medium allows a preliminary discrimination by colony morphology and color of several yeast species generally present in fruit juices and during alcoholic fermentation (Pallmann et al. 2001; Polizzotto et al. 2016). After incubation for 3–5 days at 25 °C, five to ten individual colonies were isolated from each plate and purified through repeated streaking on WL agar. A total of 95 isolates were grouped per colony morphology and color, and cell morphology observed at light microscope (Optika B-383PHi, Optika srl, Bergamo, Italy) with camera attached. A total of 77 isolates were selected as representatives of different morphotypes and molecularly analyzed. Isolates were maintained on yeast extract peptone dextrose agar (YPD; 10 g/L yeast extract, 20 g/L peptone, 20 g/L glucose and 15 g/L agar) slants and kept at 4 °C.

DNA sequencing and phylogenetic analysis

Genomic DNA was extracted from yeast isolates as previously described by Cocolin et al. (2000). The D1/D2 domain of the large subunit (LSU or 26S) rDNA gene and the internal transcribed spacer (ITS) region were amplified using primers NL1/NL4 and ITS1/ITS4, respectively (Kurtzman and Robnett 1998; White et al. 1990). Amplicons were purified using the NucleoSpin Gel and PCR Clean-up Kit (Macherey–Nagel, Düren, Germany) following the manufacturer’s instructions and sequenced at the Eurofins Genomics (Eurofins Genomics, Edersberg, Germany) using the same primers used for amplification. The searches of sequence similarity were performed using the BLAST algorithm to compare the query sequences with sequences database on the GenBank-NCBI (https://blast.ncbi.nlm.nih.gov/Blast.cgi) and YeastIP websites (http://genome.jouy.inra.fr/yeastip/index.php).

The partial DNA sequences of 77 isolates and reference taxa of each genus were used to establish datasets. The sequences of reference strains closely related to isolates used in this study were recovered from literature (Cadete et al. 2015; Chen et al. 2009; Daniel et al. 2014; de Vega et al. 2014; Kruse et al. 2017; Kurtzman and Suzuki 2010; Limtong et al. 2012; Naseeb et al. 2017; Ouoba et al. 2015; Zalar et al. 2008). Phylogenetic analysis was conducted using sequences from the Clustal W multiple alignment output using neighbor-joining (NJ) statistical methods and maximum likelihood (ML) substitution model on the MEGA 7.0 interface. The ML trees were inferred with 1000 bootstrap replicates (BS) to ascertain the reliability of a given branch pattern in the trees.

Statistical treatments of data

Principal component analysis (PCA) was performed on data obtained from phylogenetic analysis using the statistical software package XLSTAT 2017 (Addinsoft, Paris, France).

Results

The sequence similarity of the LSU and/or ITS regions of the 77 yeasts was evaluated by alignment on the GenBank and YeastIP gene databases for species identification. The comparative analysis using both databases provided information to assign the species to most of the isolates. However, results of alignment similarity and coverage scores relative to some query sequences did not allow a reliable species attribution. Thus, all 77 strains were phylogenetically analyzed to verify their taxonomic position.

A total of 20 species belonging to ten different genera were identified according to the current taxonomic studies (Table 1). Figures 1 and 2 and Supplementary Fig. S1–S7 show phylogenetic trees containing 35 out of 77 isolates, whose sequences have been deposited in the GenBank database.

Table 1 Species identification and frequency of isolates from apple juices for cider production in two cider houses (CHA and CHB)
Full size table
Fig. 1
figure 1

The neighbor joining (NJ) trees inferred of Metschnikowia spp. from the dataset containing the Metschnikowia spp. sequences of LSU rDNA D1/D2 domains and, in bold, the isolates from apple juices. Candida asparagi CBS 9770 and Schizosaccharomyces pombe NRRL Y-12796 were used as the outgroups

Full size image
Fig. 2
figure 2

The neighbor-joining (NJ) trees inferred of Moesziomyces spp. from the dataset containing the Moesziomyces spp. sequences of LSU rDNA D1/D2 domains and ITS gene region. In bold, the isolates from apple juices. Ustilago maydis MS 115 was used as outgroup

Full size image

A wide variety of non-Saccharomyces species were predominant (83%) compared to 13% of the species Saccharomyces cerevisiae, the only species of this genus (Supplementary Fig. S1).

The comparative analyses on GenBank and YeastIP gene databases allowed the identification of five species in the genus Candida (C. oleophila, C. railenensis C. cylindracea C. bentonensis, and C. zemplinina), whose frequency was 16.9%. Phylogenetic analysis confirmed the species attribution of Candida isolates, C. oleophila and C. railenensis were recorded within Kurtzmaniella clade, C. cylindracea in Ogataea clade and C. zemplinina (sym. Starmerella bacillaris) in Starmerella clade (Supplementary Fig. S2).

BLAST analysis on GenBank gene database was scarcely useful to assign the species to six out of 21 isolates of Hanseniaspora, the most frequent genus (27.3%). The LSU gene sequences of these isolates (Y1A, Y1E, Y3B, Y1G, YR6, and YR3G) showed 99–100% similarity (99–100% query coverage) to different species mainly H. meyeri, H. opuntiae, H. pseudoguilliermondii, H. guilliermondii, and H. uvarum. On the other hand, comparative analysis on YeastIP gene database and phylogenetic analysis allowed identifying these isolates as H. meyeri (Y1A, Y1E, Y3B, and Y1G) and H. pseudoguilliermondii (YR6 and YR3G). The remaining isolates were identified as H. valbyensis and H. uvarum by both BLAST and phylogenetic analyses (Supplementary Fig. S3).

Analysis on the GenBank gene database was not reliable for the species attribution on isolates belonging to Pichia. Four species (P. nakasei, P. kudriavzevii, P. fermentans, and P. membranifaciens) were ascribed to the revised genus Pichia (15.6%) by BLAST analysis on the YeastIP gene database and phylogenetic analysis. The tree topology shows four isolates (Y2C, Y15M, YR19, and Y6D) clustered into a clade including both P. membranifaciens and the strain Pichia sp. NRRL Y-27259 (BS = 99%). These isolates were identified as P. membranifaciens since they are very close to the type strain of P. membranifaciens (Supplementary Fig. S4). Meyerozyma guilliermondii, previously Pichia guilliermondii (anamorph Candida guilliermondii), was the only species recovered among Meyerozyma isolates (5.2%) (Supplementary Fig. S5). The genus Cyberlindnera, previously Pichia, was represented by Cy. misumaiensis (1.3%) (Supplementary Fig. S6).

Comparative analyses on BLAST on the GenBank and YeastIP gene databases were unable to assign the species to the two Metschnikowia isolates (2.6%). The LSU gene sequences of Y1L and Y8M isolates showed 98–100% similarity (99–100% query coverage) to different species belonging to M. pulcherrima clade (e.g., M. pulcherrima, M. fructicola, M. ziziphicola, and M. sinensis) as well as unidentified Metschnikowia sp. strains. On the other hand, phylogenetic analysis identified these isolates as M. fructicola (Y1L) and M. sinensis (Y8M) (Fig. 1).

The BLAST analysis on GenBank gene database of LSU sequences of three isolates (YR7, YR14, and YR18G) showed 99–100% similarity (99–100% query coverage) to different Pseudozyma species, mainly P. aphidis and other species of ustilaginomycetous yeasts (e.g., Dirkmeia churashimaensis, Ustilago sieglingiae, and Triodiomyces triodiae). This analysis carried out on ITS sequences did not furnish more reliable results than those obtained by LSU sequence analysis (data not shown). On the contrary, the combined tree generated by the phylogenetic analyses of the LSU and ITS regions clearly positioned the three Moesziomyces isolates (3.9%) in the recent revised Moe. bullatus clade (Fig. 2).

All the isolates from apple juices with yeast-like fungi morphological characteristics (14.2%) belonged to the Aureobasidium pullulans complex (Supplementary Fig. S7).

PCA was used to study the relationship between cider houses and the species isolation. The two cider houses were mainly discriminated on the PC1 that explained the highest variance (36.7%) (Fig. 3). CHA 2 and CHB 3 resulted quite far from CHA 1 and 3, and CHB 1 and 2, respectively, indicating variability of yeast populations within each cider house. High loading scores (> 8.0 and < − 8.0) were observed for S. cerevisiae, C. cylindracea, S. bacillaris, H. uvarum, P. nakasei, P. membranifaciens, and M. fructicola.

Fig. 3
figure 3

Bi-plot PC1 and PC2 indicates species isolated from three different samples of apple juices collected from two cider houses (CHA 1, 2, and 3; CHB 1, 2, and 3)

Full size image

Discussion

This investigation revealed that the ecosystem of apple juice used to produce cider is consistently complex in terms of the numbers of yeast species isolated. This finding is in accordance with previous investigations carried out on apple surfaces, juice, cider, and processing plants (Coton et al. 2006; Cousin et al. 2017; Morrissey et al. 2004; Pando Bedriñana et al. 2010; Pelliccia et al. 2011; Suárez Valles et al. 2007; Wang et al. 2015; Wei et al. 2017; Glushakova and Kachalkin 2017). It provides insights into yeast populations of interest from fermentative and productive points of view.

The prevalence of non-Saccharomyces yeasts observed among isolates proves how strong an impact they may have on cider aroma profiles when produced by natural fermentation (Cousin et al. 2017).

Aureobasidium pullulans is the only representative yeast-like fungus species detected in apple orchards, fruit surfaces, and apple juice processing plants (Wang et al. 2015; Wei et al. 2017). The presence of this species on apple juices could affect the cider flavor due to its ability to produce hydrolytic enzymes that support aroma molecules release (Baffi et al. 2013). Nevertheless, its possible role on the final product should be not so relevant with respect to other fermentative species described in the present study.

Several non-Saccharomyces species such as C. oleophila, C. bentonensis, H. valbyensis, H. uvarum, P. nakasei, P. membranifaciens, P. fermentans, Mey. guilliermondi, and Cy. misumaiensis had already been isolated on apple surfaces, in juice, and during the cider-making process (Cabranes et al. 1990; Coton et al. 2006; Joshi and Attri 2017; Kurtzman 2000; Kurtzman 2005; Morrissey et al. 2004; Pando Bedriñana et al. 2010; Passoth 2016; Suárez Valles et al. 2007; Wang et al. 2015; Glushakova and Kachalkin 2017). The recovery of M. fructicola and S. bacillaris in apple juice is congruent with their isolation on the apple surface as reported by Wei et al. (2017). Pichia kudriavzevii had already been found in apple juice as well as apple leaf and fruit (Wang et al. 2015; Wei et al. 2017). Moreover, several non-Saccharomyces yeasts identified in this study were previously found in apple juice processing plants to produce commercial juice (Cabranes et al. 1990; Wang et al. 2015).

The isolation of C. railenensis, C. cylindracea, H. meyeri, H. pseudoguilliermondii, and M. sinensis is worth noting. To the best of our knowledge, these species had been isolated for the first time in apple and related products (e.g., juice and cider). Interestingly, H. pseudoguilliermondii had been previously described in orange juice concentrate (Cadez et al. 2006), while C. railenensis and M. sinensis had been previously detected in grapes, must, or wine (Brežná et al. 2010; Chovanová et al. 2011; Kachalkin et al. 2015). Therefore, the isolation of the latter three species from apple juice is congruent with their occurrence in a fruit processing environment.

The role of some non-Saccharomyces species identified in this study (i.e., S. bacillaris, H. valbyensis, H. uvarum, P. fermentas, and P. kudriavzevii) in enhancing and modulating the flavor of fermented beverages, especially wine, has been documented (Cousin et al. 2017; Varela 2016). Hanseniaspora uvarum and H. valbyensis yeasts can contribute to the fruity sensory note of cider by increasing the amount of alcohols and esters during apple juice fermentation (Cousin et al. 2017). Moreover, effects on cider aroma could be linked to the presence of M. sinensis and M. fructicola in apple juice as reported for the sibling species M. pulcherrima in wine (Varela 2016).

It was detected in Moe. bullatus, a species belonging to recently revised genus (Kruse et al. 2017). Moesziomyces bullatus sensu lato contains strains of Moe. bullatus (formerly Pseudozyma bullatus), Moe. aphidis (formerly P. aphidis), Moe. rugulosa (formerly P. rugulosa), and Pseudozyma sp.. According to this genus revision, Moe. bullatus could have already been detected in apple, since the species Moe. aphidis had been found among the epiphytic yeasts on apple surfaces in China (Wei et al. 2017). This ustilaginomycetous yeast exhibits relevant hydrolytic enzymatic activities of biotechnological interests (Johnson 2013), but its effects on apple juice fermentation and cider aroma seem to be negligible.

Earlier studies on yeasts from apple orchards, apples, and related products were mainly performed through classic molecular techniques (e.g., ITS-RFLP) and comparative sequence analysis by BLAST on the GenBank gene database (Cabranes et al. 1990; Coton et al. 2006; Joshi and Attri 2017; Kurtzman 2005; Morrissey et al. 2004; Pando Bedriñana et al. 2010; Passoth 2016; Suárez Valles et al. 2007; Wang et al. 2015). These identification methods could not be completely reliable since delimitation between close species is possible only by phylogenetic reconstruction (Kurtzman and Robnett 1998). Based on the results of comparative analysis on the GenBank and YeastIP gene databases, species attribution of some isolates analyzed in the present study could be questionable. Moreover, BLAST analysis on the YeastIP gene database does not allow the identification of basidiomycetous yeasts. The results of phylogenetic analysis of Metschnikowia spp. and Moe. bullatus isolates highlighted that the sequencing of the LSU region has a high resolution for a reliable species delimitation. Hence, a combined approach using comparative and phylogenetic analyses could play an important role in increasing knowledge about yeast diversity in an apple fruit environment.

The differences in species composition observed between the two cider houses could be attributed to different climatic conditions as well as the mixture of apple varieties used to obtain juices. Moreover, it cannot be excluded that the yeast community of each cider house could also be influenced by other important factors such as cider plant microclimate, clarification techniques, chemical treatments, hygienic conditions, and handling of fruits (Cabranes et al. 1990; Glushakova and Kachalkin 2017). Yeast microbiota, characterized by high species diversity, could contribute to increasing the aroma complexity and giving distinct regional characteristics to cider, like yeasts well-investigated in the grape/wine environment (Capozzi et al. 2015). It is plausible that the volatile profile of cider from the alpine region is more complex than that of the plains region, due to its higher yeast diversity. Further investigations are necessary to ascertain the effects of certain species, such as Pichia, Metschnikowia, Hanseniaspora, and Candida/Starmerella, on the quality and typicity of cider produced in a specific geographical area.

In conclusion, traditional culture-dependent method and molecular analyses have highlighted the complexity of yeast populations residing in apple juice used to produce cider. The use of WL medium, which allows morphotypes yeast discrimination, was an important tool in the preliminary identification of isolates from the fruit environment, as previously reported (Polizzotto et al. 2016). The finding of species that had been recovered for the first time from apple and related products points out that there is still scarce knowledge about apple microbiota. Further sampling is recommended to isolate and characterize new strains which could have predictive value for cider quality. Moreover, the link between yeast diversity and local cider-making area should be ascertained. In the future, the microbial community of apple and related products could be better explored by also using culture-independent methods to obtain a wider knowledge about the yeast populations of such environments.