Use of Dairy and Plant-Derived Lactobacilli as Starters for Cherry Juice Fermentation

Background: Lactic acid bacteria (LAB) exhibit a great biodiversity that can be exploited for different purposes, such as to enhance flavours or metabolize phenolic compounds. In the present study, the use of dairy and plant-derived LAB strains to perform cherry juice fermentation is reported. Methods: The growth ability of Lactobacillus plantarum, Lactobacillus casei, Lactobacillus paracasei and Lactobacillus rhamnosus was studied in cherry juice. Profiling of sugars, organic acids and volatile compounds was performed by GC-MS (Gas Chromatography-Mass Spectrometry), while the phenolic fraction was characterized using UHPLC (Ultra High Performance Liquid Chromatography) equipped with a linear ion trap-mass spectrometer. Results: Sucrose significantly decreased in all fermented samples as well as malic acid, converted to lactic acid by malolactic fermentation. The total amount of volatile compounds increased. Specifically, propyl acetate, an ester with fruit notes, reached the highest concentration in L. rhamnosus and L. paracasei (dairy strains) fermented juices. Phenolics were extensively metabolized: caffeic acid was converted into dihydrocaffeic acid, p-coumaric acid into 4-ethylphenol and phenyllactic acid was produced. Conclusion: Lactic acid fermentation confer fruit notes to the juice and enhance phenyllactic acids, especially employing dairy strains (L. rhamnosus and L. paracasei). The level of dihydrocaffeic acid, a compound with putative biological activity was also increased (in particular with L. plantarum).


Introduction
Sweet cherry, Prunus avium L., is a native tree of Europe and western Asia and grows wild around the world [1]. Its fruit is consumed fresh, but also after processing, as canned, dried, frozen, and in syrups and juices. The main producers of sweet cherry are Turkey, United States, Iran, Spain, Italy and Chile. In Italy the annual production is about 100,000 tonnes, and the Apulia region is the most relevant area of production [2]. Sweet cherry is characterized by a high content of micronutrients and bioactive compounds, even if these attributes strongly depend on cultivar, ripening, growth condition, pre and post-harvest treatments [1]. Sugars and organic acids in fruit have been reported in the 125-265 g/kg and 3.67-8.66 g/kg ranges on fresh weight basis, respectively [3]. The balance between sweetness (sugars, mainly glucose and fructose) and sourness (acids, mainly malic acid) is paramount for the acceptance of the product by consumers [4], as well as the fruit aroma, despite the fact that aromatic

Fermentation Process and Storage
A commercial pasteurized cherry juice (Bionaturae) was used for fermentation. The absence of microbial contamination in the juice was evaluated on Plate Count Agar at 30 • C and 37 • C. LAB strains were cultivated in MRS broth for 15 h at 30 • C for L. plantarum and at 37 • C for L. rhamnosus, L. casei and L. paracasei, to reach the late exponential growth phase, centrifuged at 10,000 g for 10 min at 4 • C, washed twice with Ringer's solution (Oxoid, Milan, Italy) and re-suspended in sterile distilled water. These cultures were individually inoculated into cherry juice to reach the final concentration of ca. 7 Log CFU/mL. The juices were incubated at 30 • C for L. plantarum and at 37 • C for L. rhamnosus, L. casei and L. paracasei for 48 h and then stored for 12 days at 4 • C. Unfermented cherry juices (not added of starter cultures) were incubated at 30 • C and at 37 • C for 48 h, then stored for 12 days at 4 • C and used as controls. All the fermentations were carried out in triplicate.

Evolution of Bacterial Growth and Acidification of Cherry Juice
Cherry juice, inoculated with starters, was analyzed before and after fermentation (48 h) and after the storage period (12 days). Cultivable cells were determined using the standard plate count agar method as follows: decimal dilutions of samples were carried out in Ringer solution (Oxoid, Milan, Italy) and plated on MRS agar, then incubated at 30 • C (L. plantarum) and 37 • C (L. rhamnosus, L. casei and L. paracasei) for 48 h under anaerobic condition. The pH of samples was measured using a pH metre (Mettler Toledo, Greifensee, Switzerland). Plate count and pH measurement was carried out in triplicate.

Sugars and Organic Acid Analysis
Sugars and organic acids were analyzed after fermentation and storage, both for fermented juices and for controls. The method reported by Cirlini et al. [20], with slight modifications, was applied. Briefly, 10 µL of samples was added to 1 mL of a solution containing two internal standards (turanose and glutaric acid 500 µg/mL each), dried under vacuum and dissolved with 500 µL of dimethylformamide. Subsequently, silylation was carried out adding 400 µL of hexamethyldisilazane and 200 µL of trimethylchlorosilane to the samples and heating for 30 min at 70 • C. Samples were analyzed by a Thermo Scientific Trace 1300 gas chromatograph coupled to a Thermo Scientific ISQ single quadrupole mass spectrometer equipped with an electronic impact (EI) source on a BP5MS capillary column (30 m × 0.25 mm, with 0.25 µm film thickness, SGE Analytical Science, Milan, Italy). Chromatographic conditions were the following: initial oven temperature, 60 • C, then increase of  20 • C/min up to 280 • C; carrier gas, (flow rate, 1 mL/min). Temperature of the transfer line was maintained at 280 • C, while the ion source was set at 230 • C. The acquisition mode was full scan (m/z: . Signals were identified on the basis of their mass spectra compared with those present in the instrument library (NIST 14). In addition, once the glucidic and organic acid fractions were recognized, proper analytical standards were used in order to confirm the identifications. The semi quantification of all detected gas-chromatographic signals was performed on the basis of the use of two internal standards: turanose for sugar quantification and glutaric acid for organic acid quantification. For each identified compound, the Response Factor (RF) was calculated and the values range between 0.8-1.2.

Characterization of the Volatile Profile
The volatile profile of fermented and unfermented samples was analyzed after 48 h of incubation and after 12 days of storage. Volatiles were characterized by HS-SPME/GC-MS (Head Space-Solid Phase Microextraction/Gas Chromatography-Mass Spectrometry) technique following the protocol reported by Ricci et al. [18]. In brief, 2 mL of cherry juice was placed in a glass vial and added to an aqueous toluene standard solution (0.25 µg/mL). Head space micro-extraction was performed for 30 min at 40 • C after 15 min of equilibration time. A SPME fiber coated with 50/30 µm of Divinylbenzene-Carboxen-Polydimethylsiloxane (DVB/Carboxen/PDMS) was used (Supelco, Bellefonte, PA, USA). The desorption of volatiles was accomplished by exposing the fiber into the GC injector for 2 min at 250 • C. GC-MS analyses were performed on a Thermo Scientific Trace 1300 gas chromatograph coupled to a Thermo Scientific ISQ single quadrupole mass spectrometer equipped with an electronic impact (EI) source. All samples were injected in splitless mode. Helium was used as carrier gas, with a total flow of 1 mL/min. The separation was performed on a SUPELCOWAX 10 capillary column (Supelco, Bellefonte, PA, USA; 30 m × 0.25 mm × 0.25 µm) with the following program gradient: initial temperature, 50 • C for 3 min, linear increase by 5 • C per minute to 200 • C, then maintained for 12 min. The transfer line temperature was 250 • C. The signal acquisition mode was full scan (from 41 m/z to 500 m/z). The main volatile compounds of cherry juices were identified both on the basis of their mass spectra compared with the library NIST 14 mass spectra, as by calculation of linear retention indices (LRI). The semi-quantification of all detected gas-chromatographic signals was performed on the basis of the use of an internal standard (toluene).

Characterization of Polyphenolic Profile of Fermented and Unfermented Cherry Juices
All fermented and unfermented samples were analyzed by an Accela UHPLC 1250 equipped with a linear ion trap-mass spectrometer (MS) (LTQ XL, Thermo Fisher Scientific Inc, San Jose, CA, USA) fitted with a heated-electrospray ionization probe (H-ESI-II, Thermo Fisher Scientific Inc, San Jose, CA, USA). Separation was performed on an Acquity UPLC HSS T3 (2.1 × 100 mm) column coupled with a pre-column Acquity UPLC HSS T3 VanGuard (2.1 × 5 mm) (Waters, Milford, MA, USA). The volume injected was 5 µL, and oven temperature was set to 40 • C. Phenolic profiling was performed following the protocol reported by Ricci et al. [12]. Briefly, the mobile phase was 0.1% (v/v) acetonitrile (phase A) and 0.1% (v/v) aqueous formic acid (phase B). Elution was performed at a flow rate of 0.3 mL/min. The gradient started with 95% B and 5% A for 0.5 min, then eluent B decreased at 49% and A increased at 51% in 9 min. After 0.5 min, the column was flushed, setting the eluent percentages at 20% B and 80% A for 11.00 min. Finally, the initial conditions were restored (total run time = 17 min). Data processing was performed using Xcalibur 2.2 software from Thermo Fisher Scientific Inc, (San Jose, CA, USA).

Statistical Analysis
To evaluate the normal distribution for each group of independent samples Shapiro-Wilk test was used. One-way ANOVA was applied to discriminate the significant differences among the samples, applying Bonferroni post hoc test and the results were considered different for values of p < 0.05. All the detected compounds (volatiles, phenolics, organic acids and sugars) were used as variables for Principal Component Analyses (PCA), which was performed by applying a correlation matrix. All the mentioned analyses were performed on SPSS Statistics 21.0 software (SPSS Inc., Chicago, IL, USA), while hierarchical clustering and heat map were carried out using Heatmapper [21].

Fermentation and Acidification of Cherry Juice
Cherry juice microflora was checked before fermentation, and no cell viability was observed by plate count. The fermentation was carried out by inoculating four strains of L. plantarum, five of L. rhamnosus, four of L. casei and one of L. paracasei. After 48 h of incubation, all L. plantarum strains were able to grow in commercial cherry juice (>1 Log cycle). In addition, L. rhamnosus 2360 and L. paracasei 4186 strains were also able to grow significantly (about 1 Log cycles). However, for most of the tested strains, an increase lower than one half Log cycles or even a decrease was observed. Cell viability remained almost unchanged after 12 days of storage at 4 • C, and only in some cases a slight decrease was observed ( Table 2). Lactic acid fermentation and storage did not affect the initial pH = 3.61 ± 0.04 (mean value ± standard deviation). All further analyses were performed considering only the strains that had shown a growth around 1 Log CFU/mL or higher, namely L. plantarum 1LE1, POM1, C1, 285, L. rhamnosus 2360 and L. paracasei 4186 ( Table 2). Table 2. Lactic acid bacteria growth. Difference in Log CFU/mL detected between cells number after 48 h of incubation and the initial value, ∆(T 48h -T 0 ), and between cells number after storage (14 days) and the initial value, ∆(T 14d -T 0 ). The isolation origin (plant/dairy products) is specified. Only the strains able to grow are reported.

Species
Strain

Principal Component Analyses: Overview on Volatiles, Phenolic Compounds, Sugars and Organic Acids
The main organic acids (lactic, malic, tartaric and citric) and sugars (fructose, glucose and sucrose) (Table S1) were identified and semi-quantified in fermented and unfermented cherry juice. Among volatile components, alcohols, acids, ketones, esters, terpenes and norisoprenoids were detected and semi-quantified (57 compounds) (Tables S2 and S3A,B). Also, several phenolic acids and flavonoids (Tables S4 and S5) were identified (34 compounds) and quantified (30 compounds). PCA was performed considering the semi-quantified organic acids, sugars and volatiles and the quantified polyphenols, to highlight differences among the analyzed samples. Considering a total of 94 variables, the total variance explained by PCA reached 59%, with component 1 and component 2 describing 35% and 24%, respectively. A defined clustering among fermented and unfermented (control) samples was observed ( Figure 1A,B). Unfermented samples were characterized based on those variables showing a negative value on component 1 and a positive value on component 2, such as sucrose, fructose and caffeic acid, at their highest concentrations. On the other hand, fermented juices were all grouped together, even if some differences could be observed. Indeed, after storage, the juices inoculated with 4186, C1, 285 were found to be correlated with slightly positive variables for both components (volatile compounds as propyl acetate) and were clustered separately from the main group. The most different sample was the juice containing L. plantarum 1LE1, which was characterized by more positive variables, especially volatile compounds (such as 4-hydroxybutanoic acid, 1-heptanol and trans-geraniol) ( Figure 1A,B).  Table S1 for organic acids (A) and sugars (S), Table S2 for volatile compounds (V), and Table S4 for polyphenols (P).

Sugars and Organic Acid Metabolism
Sugar in cherry juice was mainly fructose, glucose and sucrose. Generally, their microbial metabolism was limited ( Figure 2, Table S1). As an example, fructose, the most concentrated sugar, was only partially metabolized, especially after storage by L. plantarum strains (p < 0.05). Microbial fermentation did not change glucose concentration, but it was only metabolized during storage; whereas sucrose significantly decreased in all fermented samples. Lactic acid was the main end product in all fermentations, reaching the highest amount when L. plantarum POM1 was used as starter (9.47 ± 0.32 mg/mL). Malolactic fermentation occurred, leading to total degradation of malic acid, in particular when strains of dairy origin were used. Only one strain, L. plantarum C1, did not convert malic acid ( Figure 2 and Table S1). Tartaric acid metabolism was observed after fermentation and storage, especially by L. plantarum strains (p < 0.05, reduction of 92%), but it was also observed in L. rhamnosus and L. paracasei (Figure 2, group A2 and B Table S1). No significant differences were observed for citric acid among fermented and unfermented samples ( Figure 2 and Table S1).  Table S1 for organic acids (A) and sugars (S), Table S2 for volatile compounds (V), and Table S4 for polyphenols (P).

Sugars and Organic Acid Metabolism
Sugar in cherry juice was mainly fructose, glucose and sucrose. Generally, their microbial metabolism was limited ( Figure 2, Table S1). As an example, fructose, the most concentrated sugar, was only partially metabolized, especially after storage by L. plantarum strains (p < 0.05). Microbial fermentation did not change glucose concentration, but it was only metabolized during storage; whereas sucrose significantly decreased in all fermented samples. Lactic acid was the main end product in all fermentations, reaching the highest amount when L. plantarum POM1 was used as starter (9.47 ± 0.32 mg/mL). Malolactic fermentation occurred, leading to total degradation of malic acid, in particular when strains of dairy origin were used. Only one strain, L. plantarum C1, did not convert malic acid (Figure 2 and Table S1). Tartaric acid metabolism was observed after fermentation and storage, especially by L. plantarum strains (p < 0.05, reduction of 92%), but it was also observed in L. rhamnosus and L. paracasei (Figure 2, group A2 and B Table S1). No significant differences were observed for citric acid among fermented and unfermented samples ( Figure 2 and Table S1).  Table S1 for organic acids (A) and sugars (S), Table S2 for volatile compounds (V), and Table S4 for polyphenols (P).

Sugars and Organic Acid Metabolism
Sugar in cherry juice was mainly fructose, glucose and sucrose. Generally, their microbial metabolism was limited ( Figure 2, Table S1). As an example, fructose, the most concentrated sugar, was only partially metabolized, especially after storage by L. plantarum strains (p < 0.05). Microbial fermentation did not change glucose concentration, but it was only metabolized during storage; whereas sucrose significantly decreased in all fermented samples. Lactic acid was the main end product in all fermentations, reaching the highest amount when L. plantarum POM1 was used as starter (9.47 ± 0.32 mg/mL). Malolactic fermentation occurred, leading to total degradation of malic acid, in particular when strains of dairy origin were used. Only one strain, L. plantarum C1, did not convert malic acid (Figure 2 and Table S1). Tartaric acid metabolism was observed after fermentation and storage, especially by L. plantarum strains (p < 0.05, reduction of 92%), but it was also observed in L. rhamnosus and L. paracasei (Figure 2, group A2 and B Table S1). No significant differences were observed for citric acid among fermented and unfermented samples ( Figure 2 and Table S1).

Characterization of Volatile Profile
During the characterization of unfermented and fermented cherry juices, carried out using HS-SPME/GC-MS analysis, 57 different compounds were detected in the headspace of every sample. For their identification, the mass spectra were compared with those present in the instrument library NIST 14. Furthermore, linear retention indices (LRIs) were calculated for every signal using the retention time of a linear alkane solution analyzed in the same condition, and the LRIs were compared with those reported in literature. The identification of all detected volatile compounds was reported in Table S2.
Overall, after fermentation, an increase in the concentration of total volatile compounds was observed. The main identified chemical classes were ketones, alcohols, aldehydes, terpenes, terpenic derivatives and norisoprenoids, acids and esters ( Figure 3 and Table S3A,B). In fermented samples, an increased content of ketones was found; among them, acetoin was the most concentrated, especially when strains of plant origin were employed. This compound showed a further increase after storage, reaching the highest concentration when L. plantarum C1 was used (Table 3). Benzene methanol was the most abundant alcohol (maximum amount detected with L. plantarum 1LE1 after storage, Table 3). Aldehydes were generally not affected by fermentation and storage.

Characterization of Volatile Profile
During the characterization of unfermented and fermented cherry juices, carried out using HS-SPME/GC-MS analysis, 57 different compounds were detected in the headspace of every sample. For their identification, the mass spectra were compared with those present in the instrument library NIST 14. Furthermore, linear retention indices (LRIs) were calculated for every signal using the retention time of a linear alkane solution analyzed in the same condition, and the LRIs were compared with those reported in literature. The identification of all detected volatile compounds was reported in Table S2.
Overall, after fermentation, an increase in the concentration of total volatile compounds was observed. The main identified chemical classes were ketones, alcohols, aldehydes, terpenes, terpenic derivatives and norisoprenoids, acids and esters (Figure 3 and Table S3A and S3B). In fermented samples, an increased content of ketones was found; among them, acetoin was the most concentrated, especially when strains of plant origin were employed. This compound showed a further increase after storage, reaching the highest concentration when L. plantarum C1 was used (Table 3). Benzene methanol was the most abundant alcohol (maximum amount detected with L. plantarum 1LE1 after storage, Table 3). Aldehydes were generally not affected by fermentation and storage.

Characterization of Polyphenolic Profile
Using LC-MS n , 34 polyphenolic compounds were identified, 16 by comparison with authentic standards, while for the others, for which reference compounds were not available, tentative identification was based on the interpretation of their fragmentation patterns obtained from MS 2 and MS 3 spectra and by comparison with data reported in literature. Retention times and mass spectral data, along with peak assignments for the identified compounds, are reported in Table S4. Quantified compounds are reported in Table S5. The total polyphenolic concentration did not change upon treatment, but differences inside individual polyphenolic subclasses emerged, suggesting that specific compounds were metabolized by lactic acid bacteria.
Among hydroxycinnamic acids, caffeic acid and p-coumaric acid were mainly affected by fermentation. The conversion of caffeic acid was observed in L. plantarum strains, with a complete degradation of the compound by L. plantarum 285. The metabolism of caffeic acid led to the accumulation of dihydrocaffeic acid, the only phenylpropionic acid detected ( Figure 4, Table 4). p-Coumaric acid was totally metabolized by L. plantarum 285 and POM1, whereas for the C1 strain a partial conversion was observed after fermentation, and the compound was totally metabolized upon storage ( Figure 4, Table 4). Protocatechuic acid was also metabolized by LAB, mainly by L. plantarum (Table 4). In addition, all the strains were able to produce p-hydroxyphenyllactic acid and phenyllactic acid; the highest levels of these compounds were measured after L. plantarum 285 fermentation (1.47 ± 0.16 µg/mL and 2.15 ± 0.15 µg/mL respectively).
The adaptation to the acid environment of cherry juice is based on the ability of LAB to metabolize malic acid [24], resulting in its almost complete degradation. Actually, all tested dairy and plant strains, independently of species and with the exception of L. plantarum C1, completely converted malic acid into lactic acid. Indeed, most LAB can convert malic acid into lactic acid thanks to the malolactic enzyme, decarboxylating malate to lactate by a NAD + and Mn 2+ -dependent malolactic enzyme [23,26,27]. At low pH, the choice of malic acid as the preferred energy source over glucose had already been reported in literature and linked to the increase of intracellular pH and the increase of reducing power [28], with an associated modification of cellular permeability [23]. In the present work, LAB did not consume glucose and fructose during fermentation, whereas sucrose showed a marked decrease. The metabolic activity of the tested strains, especially in L. rhamnosus and L. paracasei, is mainly based on malolactic fermentation, aimed at the maintenance of vitality, rather than on sugars fermentation, which is mainly correlated with growth [23]. The exploitation of an organic carbon source by a microorganism strongly depends on the type of substrate. Two different works on cherry juice fermentation reported the absence of sugars metabolism in one case [23], and a slight consumption after fermentation in the other [2], the latter being in very good agreement with our results. Observed differences may be related to cultivar, to ripening time, and to the applied processing conditions, which could affect the concentration of nutrients and/or acids [1,29]. In the present work, tartaric acid was also consumed by LAB, especially by L. plantarum, in agreement with literature, where Lactobacillus spp. were reported to be able to convert it into oxaloacetic acid (and then into lactic acid, acetic acid, and CO 2 ), thanks to the tartrate dehydratase enzyme [30][31][32]. Nevertheless, the degradation of tartaric acid is not widespread in LAB, and it has been studied especially in wine. Moreover, not all LAB were able to metabolize tartaric acid in the same way; for instance, differently from L. plantarum, in Lactobacillus brevis, succinic acid can be produced instead of lactic acid [33]. The acetic acid found in several fermented fruit juices [18,23,34] is characterized by sharp, pungent and vinegar notes. However, thanks to specific bacterial metabolic traits, acetic acid can be converted to the corresponding esters [35,36], such as propyl acetate; this was especially observed after fermentation with L. rhamnosus 2360 and at the end of storage using L. paracasei 4186. The increase of β-linalool, observed in different fermented samples, could be ascribed to glycosylases produced by LAB, involved in the release of aglycones from glycosylated terpenes [18]. Acetoin was also affected by fermentation and its bio-synthesis can be derived from citrate metabolism [18,27].
Phenolic compounds have been reported to exert health benefits in humans [37], to exhibit antimicrobial activity, and to impact the flavor, taste and color [10] of plant products [38]. Their beneficial activities have been partially correlated with microbial metabolism that can occur during fermentation processes [39]. Moreover, the metabolism of phenolics may contribute to bacterial stress response when microorganisms are in hostile conditions [19,40].
In the present study, hydroxycinnamic acids were transformed by LAB, ideally through phenolic acid decarboxylases or phenolic acid reductases [19]. Caffeic acid was effectively converted into dihydrocaffeic acid, as also already reported in the literature for L. plantarum POM1, [11,12,19], with a potential impact on health, due to its more effective capacity to inhibit platelet activation than its phenolic precursor [41] and to its antioxidant effect on endothelial cells [42]. In addition to putative effects on the health-promoting activity of fermentation metabolites, the microbial transformation exerted by LAB may be relevant for other purposes. For example, p-coumaric acid was decarboxylated into p-vinylphenol and subsequently reduced, possibly by a phenolic acid reductase, to the phenolic volatile 4-ethylphenol, which contributes to aroma in fermented food [19]. The reduction of p-vinylphenol was favoured under anaerobic conditions or in absence of electron acceptors, i.e., when fructose is found at high concentration, to increase NAD + quantity [19,43]. Protocatechuic acid was also metabolized [11,12], in particular by L. plantarum 285 and POM1, reaching complete depletion, but catechol was not detected. Finally, worthy of note is that phenyllactic acids were produced ex-novo by all the tested strains, probably deriving from amino acid metabolism. Phenylalanine can be converted into phenylpyruvic acid by a transamination reaction, and finally metabolized into phenyllactic acid by hydroxyacid dehydrogenase, while p-hydroxyphenyllactic acid may originate from tyrosine metabolism [44]. Their antimicrobial activity against pathogenic strains and moulds has been well documented [44,45], even if the concentrations found in the present study were lower compared to those explaining antimicrobial activity.

Conclusions
In the present study, an overview on the metabolism of LAB strains, of plant but especially dairy origin, during cherry juice fermentation and further storage, was reported. L. plantarum 285, of dairy origin, was the most interesting strain for its ability to modulate the juice phenolic profile, while L. rhamnosus 2360 and L. paracasei 4186 were useful for modifying and increasing the presence of aromatic compounds. Based on these results, a selection of strains exploitable as starters in cherry juice fermentation can be performed based on sound analytical data. L. rhamnosus and L. paracasei conferred fruit notes to the juice; on the other hand, L. plantarum increased the levels of dihydrocaffeic acid, with putative biological activity, and of phenyllactic acids.