Isolation and Characterization of High-Ethanol-Tolerance Lactic Acid Bacteria from Australian Wine

Lactic acid bacteria are very important in winemaking. In this study, 108 lactic acid bacteria isolates were obtained from high-ethanol-content (~17% (v/v)) Grenache wines during uninoculated malolactic fermentation (MLF). The 16S rRNA and species-specific PCR showed that 104 of these were Oenococcus oeni, three were Lactobacillus hilgardii, and one was Staphylococcus pasteuri. AFLP of HindIII and MseI digests of the genomic DNA of the O. oeni strains was developed for the first time to discriminate the strains. The results showed that the method was a suitable technique for discriminating the O. oeni strains. Based on the cluster analysis, nine O. oeni strains were chosen for inclusion in an ethanol tolerance assay involving monitoring of optical density (ABS600nm) and viable plating. Several O. oeni strains (G63, G46, G71, G39) survived and grew well in MRS-AJ with 17% (v/v) ethanol, while the commercial O. oeni reference strain did not. Strain G63 could also survive and grow for 168 h after inoculation in MRS-AJ medium with 19% (v/v) ethanol. These results suggest that O. oeni G63, G46, G71, and G39 could potentially be used as MLF starters for high-ethanol-content wines. All three L. hilgardii strains could survive and grow in MRS-AJ with 19% (v/v) ethanol, perhaps also indicating their suitability as next-generation MLF starter cultures.


Introduction
Malolactic fermentation (MLF) is a critical biochemical process in winemaking. Generally occurring at the end of alcoholic fermentation, either inoculated with starter cultures of selected strains or undertaken by indigenous bacteria. During MLF, lactic acid bacteria (LAB) transform L-malic acid into L-lactic acid and carbon dioxide [1]. This biochemical conversion decreases wine acidity, improves quality, and increases the biological stability of wine [2]. While the growth of LAB in wine can have benefits, it can also be detrimental to wine quality. This depends on which species or strain grows and the stage of the winemaking process at which growth occurs [3]. Oenococcus oeni is regarded as the principal LAB responsible for MLF, but many studies showed that Lactobacillus spp. are typically present throughout in the grape and wine environment and during MLF [4,5]. du Toit et al. [6] report that some Lactobacillus spp. are suitable as a new generation of MLF starter cultures, with a number of strains commercially available.
Opportunities to identify further strains with enhanced capabilities exist because wine is a complex, harsh environment containing sugars (glucose, fructose, riboses), ethanol, organic acids, amino acids, fatty acids, phenolic compounds, sulfur dioxide (SO 2 ), and low pH. Especially inhibitory to wine microorganisms is the combination of ethanol, low pH, and temperature along with SO 2 [1,7,8]. In warm to hot climate wine regions, the potential alcohol content of the wine can already be high, but with issues related to climate warming, ethanol contents in wine have been increasing [9]. As a result, the incidence of

Bacterial DNA Extraction
Bacterial colonies were inoculated into 5 mL of MRS-AJ medium and incubated at 30 • C in a carbon dioxide incubator for 72 h. Bacterial cells were harvested by centrifugation  (10,000× g, 5 min) from 5 mL of cultures, twice washed in 1 mL of TE (10 mM Tris-HCl pH 8.0, 1 mM EDTA), resuspended in 0.55 mL of TE containing 20 µL of 50 mg/mL lysozyme, and incubated at 37 • C for 1 h. RNAase (1.5 µL) was then added with a repeat incubation at 37 • C for 1 h, before 30 µL of 10% SDS and 15 µL proteinase K (14-22 mg/mL) were added followed by a further 3 h at 37 • C. Finally, 100 µL of 5 mol/L NaCl and 80 µL CTAB-NaCl were added, with incubation for 30 min at 65 • C. The crude DNA preparation was purified by performing 2 to 3 phenol/chloroform/isoamyl alcohol (25:24:1) and one chloroform/isoamyl alcohol (24:1) extraction. Genomic DNA was precipitated by adding two volumes of cold ethanol followed by washing with 70% (v/v) ethanol and air drying at room temperature. Finally, the DNA pellet was dissolved in TE buffer (60 µL).

Bacterial Species Identification
Bacterial isolates were preliminarily identified using standard methods, including cultural/cell morphology, Gram staining, and catalase reaction [12]. Universal primers 8F (5 -AGAGTTTGATCCTGGCTCAG-3 ) and 1492R (5 -GGTTACCTTGTTACGACTT-3 ) were used to amplify 16S rRNA. The reaction was performed in a final volume of 50 µL using the following amplification mixture: 10-100 ng DNA template, 1.5 µL DMSO, 10 µL 5× Hi-Fi buffer, 10 pmol of each primer, 1 mM dNTPs, and 1 U DNA polymerase (Velocity, Bioline). An initial 5 min denaturation at 98 • C was followed by 35 cycles of 30 s at 98 • C, 30 s at 58 • C, and 45 s at 72 • C, with a final extension of 10 min at 72 • C. Amplified products were analyzed by 1.0% (v/v) agarose gel electrophoresis. Before submission to the Australian Genome Research Facility (AGRF, Waite Campus) for sequencing, all PCR products were purified via Wizard plus SV minipreps Gel and PCR clean-up system (Promega).
Identification of O. oeni isolates at the species level was performed by PCR with speciesspecific primers On1 (5 -TAATGTGGTTCTTGAGGAGAAAAT-3 ) and On2 (5 -ATCATCGTCAAACAAGAGGCCTT-3 ) according to Zapparoli et al. [13], which allowed amplification of the mle gene (1025 bp). PCR products were resolved by electrophoresis in 1.0% (v/v) agarose gels. Species-specific primers H2 (5 -ACTNATTTGACATTAAGA-3 ) and 8623 (5 -CTGGTTCACTATCGGTCT-C-3 ) were used to identify L. hilgardii [14]. PCR reactions in a total volume of 50 µL contained 10-100 ng DNA template, 1.5 µL DMSO, 10 µL 5× Hi-Fi buffer, 5 pmol of each primer, 0.8 mM dNTPs, and 1 U DNA polymerase (Velocity, Bioline). Reactions involved an initial 95 • C for 5 min followed by 30 cycles of 95 • C for 30 s, 36 • C for 30 s, and 72 • C for 2 min, and a final extension of 72 • C for 5 min. PCR products were resolved by agarose gel electrophoresis.  Table 2). The PCR cycling conditions for these reactions were 94 • C 2 min, 10 cycles of 66 • C 30 s (down 1 • C each cycle) and 72 • C 2 min, followed by an extra 20 cycles at 94 • C 20 s, 56 • C 20 s, 72 • C 2 min, and a final extension at 60 • C for 30 min. PCR products were diluted 30 times before submission to the AGRF to carry out fragment analysis by capillary electrophoresis. Gene Mapper software version 3.7 (PE Applied Biosystems, Waltham, MA, USA) was used to automatically size and quantify individual fragments by using the internal lane standard. Peak height thresholds were set at 200. Genotyper software (PE Applied Biosystems) was set to heavy smoothing. Bands of the same size in different individuals were assumed to represent the same allele. Bands of different sizes were treated as independent loci with two alleles (present or absent). Data were reported in a binary format (1 = presence of a band/peak; 0 = absence). For clustering, fragments between 50 and 500 bp were analyzed with NTSYS software by using the Dice similarity coefficient based on presence/absence of the bands and clustered by the unweighted pair group method with arithmetic mean (UPGMA) [15].
As Abs 600 measurements do not easily detect low cell numbers [16], plating on to MRS-AJ medium was performed to more accurately determine the extent to which ethanol concentration influenced the culturability of the selected bacterial isolates. Approximately 10 7 cfu of each strain was inoculated in triplicate into 50 mL MRS-AJ medium (pH 6.5) with 15%, 17%, and 19% (v/v) ethanol. Samples (20 µL) were taken at 24, 110, 168, 210, and 309 h, serially diluted in 180 µL MRS medium (10 −1~1 0 −5 ) in a 96-well plate, and 5 µL of each dilution was drop inoculated onto solidified MRS-AJ medium and incubated anaerobically at 30 • C for 7 days. The resulting colonies were counted, and data were analyzed by GraphPad Prism 6.

Statistical Analysis
Analyses of variance were performed using StatGraphics Plus version 4.0 (Manugistics, Rockville, MD, USA). Significant differences were determined using Duncan's test at p < 0.05.

Isolation and Characterization of High Ethanol Tolerant Strains
A total of 108 isolates were recovered from four individual Grenache wines (~17% (v/v) ethanol) during uninoculated MLF. All strains were tentatively identified as lactic acid bacteria (LAB) on the basis of Gram stain positivity, catalase negativity, and cultural and cellular morphology. Identification of LAB isolates to the species level was performed by 16S rRNA sequence analysis and species-specific PCR. The data revealed that 104 strains were O. oeni, 3 were L. hilgardii, and 1 was S. pasteuri (Table 2).

Genotyping of O. oeni Strains by AFLP Analysis
The amplification efficiency obtained by the HindIII and MseI digestion was higher than that with EcoRI and MseI reported in the literature [17]. Characterization of the 104 O. oeni and 3 L. hilgardii isolates was carried out by AFLP analysis using HindIIIand MseI-digested genomic DNA. Each primer combination produced an average of 61.  Table 3). For testing the reproducibility of the primer pair MA-HT and MT-HT, three O. oeni strains and three L. hilgardii strains were selected to repeat AFLP four times. The results showed that each primer pair tested had a reproducibility higher than 99% for O. oeni strains, while it was lower than 97% for L. hilgardii strains (Table 4). UPGMA cluster analysis and a dendrogram were produced from the data obtained by the AFLP assay of 104 O. oeni isolates with the primer pair MA-HT FAM and MT-HT FAM. The genetic distance between each genotype (Figure 1) allowed the strains to be divided into six different principal clusters (A-F) at the similarity level of 71%. Two main clusters (A and B) included most of the O. oeni strains. Two major biotypes (A1 and A2) comprised 11 and 31 strains, respectively. Cluster A had variable AFLP similarity among the 42 isolates grouped therein (70%). Biotype A1 revealed an AFLP similarity ranging from 80% to 93%, whereas the similarity level of biotype A2 ranged from 77% to 95%. Cluster B also include two biotypes (B1 and B2), with biotype B1 being comprised of the most isolates (51 strains) and showing a higher AFLP similarity (82% to 95%); whereas the similarity level of biotype B2 ranged from 83% to 93%. G92 was the only strain in Cluster C, which gave an AFLP similarity of 66%. Cluster D included three strains that showed an AFLP similarity of 65%, making it the most variable. Cluster E comprised only one strain (G99) with an AFLP similarity of 64%. G71 showed the lowest AFLP similarity (57%) with other clusters. Therefore, AFLP with HindIII and MseI was successfully able to reproducibly genotype O. oeni.

Ethanol Tolerance Analysis
The growth of the O. oeni strains in comparison with a commercial reference strain (VP 41) was tested in growth studies in MRS-AJ (Figure 2a). Generally, the strains showed a steady growth that had not attained stationary phase after 80 h. O. oeni G63 had the fastest growth rate, and O. oeni G39 the slowest. After an initial lag of~40 h, G50 appeared to grow at a similar rate to VP 41 and the remainder of the tested strains. The three L. hilgardii strains grew in a similar manner to one another, with a~15 h lag followed by growth before a stationary phase commencing at~35 h (Figure 3a). oods 2022, 11, x FOR PEER REVIEW 8 of 14

Ethanol Tolerance Analysis
The growth of the O. oeni strains in comparison with a commercial reference strain (VP 41) was tested in growth studies in MRS-AJ (Figure 2a). Generally, the strains showed a steady growth that had not attained stationary phase after 80 h. O. oeni G63 had the fastest growth rate, and O. oeni G39 the slowest. After an initial lag of ~40 h, G50 appeared to grow at a similar rate to VP 41 and the remainder of the tested strains. The three L hilgardii strains grew in a similar manner to one another, with a ~15 h lag followed by growth before a stationary phase commencing at ~35 h (Figure 3a). The growth of various O. oeni (Figure 2a) and L. hilgardii strains (Figure 3a) inoculated into MRS-AJ medium containing 0%, 15%, 17%, or 19% (v/v) ethanol was characterized by Abs600. Growth was apparent for all strains. Inclusion of ethanol dramatically reduced growth and maximum optical density, with greater inhibition apparent at higher ethano  ethanol MRS-AJ medium, there was little difference among these strains before 96 h ure 2a). All three L. hilgardii strains grew well in 19% (v/v) ethanol MRS-AJ medium; h ever, maximum Abs600 was not reached for more than 400 h ( Figure S1, Supplemen Materials), compared with the 15% and 17% (v/v) ethanol treatment, where maxim Abs600 occurred after approximately 170 h (Figure 3a).   O. oeni strains G63, G46, G71, and G39 all displayed strong ethanol resistance along with the L. hilgardii strains. The L. hilgardii strains also grew in MRS-AJ medium that contained 19% (v/v) ethanol.

Discussion
In this study, the isolation and characterization of lactic acid bacteria from uninoculated malolactic fermentation of Grenache wines that had high ethanol content was carried out. In all, 104 O. oeni, 3 L. hilgardii, and 1 S. pasteuri strains were isolated from these wines. For the first time, a highly reproducible AFLP method using restriction enzyme HindIII and MseI was developed for typing within O. oeni species. The ethanol tolerance of O. oeni and L. hilgardii strains was tested to highlight strains with potentially high ethanol tolerance.
O. oeni is one of the most representative species of LAB found in must and wine, and generally, its population increases during the fermentation [1]. Isolation and screening of superior O. oeni strains from grape and wine environments is still the most important way to make MLF starters. Lòpez et al. [18] identified O. oeni in 201 out of 204 isolates from wine samples. Cappello et al. [17] recognized 87 O. oeni strains out of 220 LAB from Primitivo wine during uninoculated MLF. The use of dependable and efficient methods for the identification and discrimination of bacterial strains is essential. Importantly, O. oeni strain differentiation can only be achieved by techniques that are sensitive to minor genotypic differences. A range of molecular biology methods have been utilized for O. oeni strain differentiation. Pulsed-field gel electrophoresis (PFGE) has excellent discriminative power for O. oeni strains [19][20][21], but this method is labor intensive and technically demanding, requiring specific apparatus, thereby likely restricting it to research and reference laboratories [22,23]. Compared to PFGE, random-amplification polymorphism DNA (RAPD) is quick and relatively easily performed. However, RAPD has several drawbacks as an amplification technique, such as reproducibility between laboratories [24,25].
AFLP has shown greatest usefulness so far for intraspecies discrimination of bacteria, including LAB [23,26,27]. Until now, most AFLP research on O. oeni used EcoRI and MseI; however, the greater frequency of the adenine (A) + thymine (T)-rich recognition sequences of these enzymes in cytosine (C) + guanine (G)-poor genomes, reduces their resolving power. To address this, we modified and validated a rapid and simple method to identify O. oeni [11,13] and L. hilgardii [14] at a species level using HindIII instead of EcoRI. HindIII appeared more suited to digest genomic DNA with a G + C content of 40-50 mol%, while digestion with the 'frequent-cutter' MseI was optimal for G + C contents below 50 mol% [28]. HindIII and MseI were considered a good combination for O. oeni given a G + C content of around 40%. The results confirmed that the amplification efficiency was higher than the EcoRI and MseI combination from the literature [17], with very good repeatability. Therefore, we conclude that AFLP with HindIII and MseI is a useful technique to discriminate O. oeni strains. Notably, the repeatability of L. hilgardii was poorer (around 85%).
Due to the fact that high ethanol content often causes sluggish or stuck MLF in wine, the selection of highly ethanol tolerant strains is of interest to wine producers in hot-climate regions to help avoid such fermentation issues. Normally, O. oeni grows poorly when the ethanol exceeds 14% (v/v) [29]. In this study, all isolated strains investigated grew well in MRS-AJ containing 15% (v/v) ethanol. Some strains (G63, G46, G71, G39) could survive and grow for at least 210 h when inoculated into media containing 17% (v/v) ethanol, with G63 producing the highest quantity of culturable cells. G63 survived and grew for up to 168 h after inoculation into MRS-AJ containing 19% (v/v) ethanol. O. oeni isolates G63, G46, G71, and G39 outperformed the high-ethanol-tolerant commercial strain tested (O. oeni VP41, prepared as a fresh culture). More work is needed to determine whether these strains are also tolerant of other stresses (e.g., low pH, high SO 2 , low temperature), produce harmful metabolites, or have potential as MLF starter cultures under industrial conditions in high-ethanol wine fermentations.
Several wine LAB genera other than Oenococcus, including Leuconostoc, Lactobacillus, and Pediococcus, also decarboxylate L-malic acid into L-lactic acid and CO 2 . Emerging research is revealing that Lactobacillus spp. can be important in winemaking. Izquierdo et al. [4] and Ruiz et al. [5] found that non-Oenococcus species present during MLF consisted of L. casei, L. hilgardii, L. plantarum, and Leuc. mesenteroides. The enzyme responsible for malolactic fermentation, encoded by mleA, differs between the lactobacilli and the oenococci [30]. This may bring about differences in the acidity, microbial stability, and sensory complexity of wines produced using these and, thus, potentially influence wine quality.
A recent patent [31] relates to Lactobacillus and Pediococcus capable of completing MLF of wines at 10% (v/v) ethanol or more and high pHs, upon direct inoculation to 10 6 -5×10 7 cfu/mL, whether in a dried, frozen, or lyophilized state. These strains had been selected for tolerance to various limiting conditions, especially for high ethanol. L. plantarum strain V22 assessed over three vintages at high pH (>3.5) and ethanol (≥14% v/v) [32] was also as fast as an O. oeni starter when inoculated post-AF. Such strains are of interest as they may possess more diversity in enzymatic activities (e.g., ochratoxin A reduction, β-glucosidase, esterase, protease, phenolic acid decarboxylase) compared to O. oeni [30]. Closer to the work reported here, L. hilgardii and L. plantarum were able to complete MLF (co-inoculated or post-AF) in wines at pH~3.3, 13.5-16.5% (v/v) ethanol, and 1.2-3.2 g/L malic acid at 20 • C, at a rate equivalent to a commercial O. oeni starter culture [6]. Importantly, volatile aroma assays by an informal tasting panel revealed that these Lactobacillus strains produced different profiles to O. oeni, with no off-flavors detected [33]. Others showed that L. plantarum improved wine color during MLF [34]. Accordingly, such lactobacilli offer promise as new starters and may be more suitable for high-pH and high-ethanol wines.
As part of this study, three L. hilgardii strains were isolated as non-Oenococcus species present in high-pH (>3.5) and high-ethanol wines during uninoculated MLF. The ethanol tolerance assay (Figure 3) determined that they could grow well in MRS-AJ that contained 19% (v/v) ethanol, which was much higher than previously reported [35]. Nowadays, climate warming can mean greater grape maturity at harvest resulting in wines with higher pH and ethanol, which have a critical impact on the MLF of wines [9,36]. Therefore, it is of great importance to further understand whether these L. hilgardii strains have potential as MLF starter cultures in future wines, which may have stressful pH values and ethanol contents.
Interestingly, an S. pasteuri strain was isolated from the AU4 barrel (Table 1). There are only a few reports regarding S. pasteuri; therefore, it has not been clearly described. S. pasteuri is a coagulase-negative staphylococci, which has been isolated previously from naturally fermented Italian sausages, in which its activity impacts color and aroma development, pathogen inhibition, and shelf-life enhancement [37]. S. pasteuri is one of the few microorganisms cultured from the stratosphere [38], it is not a common skin contaminant [39,40], and has never been found responsible for human disease. Although it was found as a joint-prosthesis-colonizing organism, its pathogenic role as a cause of arthroplastic infections remains unclear [41]. Jisoo et al. [42] found that S. pasteuri produces one or more new antimicrobial substances, which are putative new bacteriocins and candidates for use in the control of food-borne pathogens, such as antibiotic-resistant S. aureus. More work is needed to understand whether S. pasteuri can be used to facilitate enhancement of the microbiological safety of food and whether it can play a role in winemaking.

Conclusions
In all, 108 lactic acid bacterial strains were isolated from high-ethanol (around 17% (v/v)) Grenache wines; 16S rRNA and species-specific PCR showed that 104 of these were O. oeni, 3 were L. hilgardii, and 1 was an S. pasteuri isolate. An AFLP technique using HindIII and MseI to digest the genomic DNA of O. oeni strains was developed for the first time to discriminate O. oeni strains. From the AFLP analysis and the growth of isolates, nine O. oeni strains and the three L. hilgardii strains were chosen for an assay of ethanol tolerance. Some O. oeni strains (G63, G46, G71, G39) survived and grew well in MRS-AJ with up to 17% (v/v) ethanol, whereas the commercial high-ethanol-tolerant reference strain did not grow. Further, O. oeni G63 could survive and grow for more than 168 h after inoculation into a 19% (v/v) ethanol medium. These results indicated that O. oeni G63, G46, G71, and G39 could have potential as MLF starter cultures for high-ethanol wines. Interestingly, all three L. hilgardii strains isolated could survive and grow well in MRS-AJ with 19% (v/v) ethanol. The potential of this species as a next-generation MLF starter culture is worthy of investigation.