Gentiobiose and cellobiose content in fresh and fermenting cucumbers and utilization of such disaccharides by lactic acid bacteria in fermented cucumber juice medium

Abstract The content of cellobiose and gentiobiose, cellulose‐derived dissacharides, in fresh and fermented cucumber was evaluated along with the ability of Lactobacillus plantarum, Lactobacillus pentosus, Lactobacillus buchneri and Lactobacillus brevis to utilize them during and after fermentation. The disaccharide content in fresh and fermenting cucumbers was below the detection level (10 µM) using HPLC for analysis. Utilization of cellobiose and gentiobiose by lactic acid bacteria (LAB) was tested in fermented cucumber juice medium (FCJM), a model system for the bioconversion and postfermentation lacking glucose and fructose. Changes in the fermentation metabolites were followed using HPLC and pH measurements as a function of time. The disaccharides were utilized by L. plantarum, L. pentosus, and L. buchneri in FCJM at pH 4.7 ± 0.1, representative of the active fermentation period, and converted to lactic acid. The disaccharides were not utilized in FCJM at pH 3.7 ± 0.1, representative of the end of fermentation. While L. brevis was unable to utilize cellobiose efficiently in FCJM, they were able to remove gentiobiose at pH 4.7 ± 0.1. Some strain level differences in cellobiose utilization were observed. It is concluded that the disaccharides are absent in the fresh cucumber and the typical fermentation. The LAB prevalent in the bioconversion utilizes cellobiose and gentiobiose, if available, at pH 4.7 ± 0.1. The LAB would not remove the disaccharides, which could become available from cellulose degradation by the acid resistant indigenous microbiota, after the pH is reduced to 3.7 ± 0.1.

This study evaluated the function of disaccharides, particularly cellobiose and gentiobiose, as potential energy sources in cucumber fermentations for selected LAB. It was theorized that a greater understanding of the role of energy sources found in cucumber fermentations for LAB would enable the design of functional starter or adjunct cultures that are able to complete the bioconversion and remove secondary energy sources. In doing so, a starter/adjunct culture would be able to prevent the growth of spoilage microbes during long-term storage and postfermentation.
This study confirmed the concentration of cellobiose and gentiobiose naturally found in fresh and fermented cucumbers using HPLC analysis for quantification and evaluated the ability of selected LAB relevant to cucumber fermentations to utilize these sugars. A fermented cucumber juice (FCJ) model system was used to evaluate the ability of L. plantarum, L. pentosus, L. buchneri, L. brevis, and P. pentosaceus to utilize cellobiose and gentiobiose to simulate conditions during (pH 4.7) and after (pH 3.7) fermentation. Microbial growth and fermentation of cellobiose and gentiobiose were monitored by colony counts from deMan, Rogosa and Sharpe (MRS) agar plates and by pH and metabolite concentration measurements using a probe and HPLC analysis, respectively.

| Measurement of cellobiose and gentiobiose concentration in fresh and fermented cucumbers
Samples of four fresh, size 2B (3.18 cm to 3.79 cm in diameter), pickling cucumber lots to be fermented in commercial vessels were obtained from a local processor. The corresponding fermented cucumber samples were collected on days 3 and 38 of fermentation together with the fermentation cover brine in a 50:50 ratio. Fresh and fermented cucumbers were sliced using aseptic techniques and blended using a Waring Commercial Blender 700S (Torrington, CT, USA) equipped with a sterilized glass cup for 90 s at medium speed.
Cucumber slurries were homogenized using a Seward Stomacher 400 (Bohemia, NY, USA) in 6″ × 4.5″ filter stomacher bags for 1 min at maximum speed. The slurries were subjected to two cycles of freezing and thawing to enable the release of all the sugars from the tissue into the liquid phase. One mL aliquots of the filtered homogenate were spun at 15,294 rcf for 10 min in an Eppendorf benchtop refrigerated centrifuge 5810R (Hamburg, Germany) to remove residual particulate matter. Supernatants were used for HPLC analysis to determine the cellobiose and gentiobiose concentrations.
Aliquots of 100 µl of fresh cucumber juice, juice extracted from cucumbers fermented in commercial vessels or from experimental media were diluted to 2 ml with water spiked with 50 µl of an internal standard of lactose (Sigma-Aldrich). All solutions were filtered through Dionex OnGuard-H cartridges (Dionex Corporation) to remove free amino acids into autosampler vials. The extracts were analyzed using a Dionex BioLC (Dionex Co.) at a controlled temperature of 25°C. The system consisted of a gradient pump, an autosampler, and a Pulsed Amperometric Detector. The mobile phase was 50 mM NaOH (Thermo Fisher Scientific) at an isocratic flow rate of 1.0 ml/min. The column used was PA-1, 250 mm length and 4 mm i.d. (Dionex Co.), fitted with PA-1 Guard column (Dionex Co.). The detector was programmed to run a quadruple waveform as recommended by the manufacturer. The detector sensitivity was set to 500 nCoulombs (nC).
The injection volume was 10 µl. Each sugar was quantified by calculating a ratio of the unknown peak height to the lactose (in-

| Bioinformatic analysis of the genes coding for enzymes involved in cellobiose and gentiobiose metabolism by certain LAB
Detection of the cellobiose permease gene, celB, among the bacterial genomes sequenced to date was done using the Integrated Microbial Genomics, Find Gene function (Chen et al., 2016; https:// img.jgi.doe.gov/cgi-bin/m/main.cgi). The analysis of the putative enzymes involved in the metabolism of gentiobiose and cellobiose was conducted using the publicly available genome sequences for L. pentosus (3), L. plantarum (107), L. brevis (21), L. buchneri (6), and P. pentosaceous (7). The Joint Genome Institute-Integrated Microbial Genomes platform (Chen et al., 2016) was used to develop informative metabolic maps with regard to the abundance of certain genes in the genomes described above. The KEGG Orthology Pathways (KO; Kanehisa, Sato, Kawashima, Furumichi, & Tanabe, 2016;Kanehisa, Sato, & Morishima, 2016; http://www.kegg.jp/kegg/pathw ay.html) tool was specifically used to populate Table 2. The Metacyc  and Biocyc (Karp et al., 2017) online tools at the IMG platform were used to define and/or confirmed the reference pathways for glycolysis and β-glucosidases involved in cellobiose and gentiobiose catabolism.

| Preparation of fermented cucumber juice media (FCJM)
Size 2B (32-38 mm in diameter) fresh whole pickling cucumbers from two different lots were secured from the local retail (Raleigh, NC, USA). Fresh pickling cucumbers in good condition and free of mechanical damage were selected, washed with plain water, and packed into four one-gallon glass jars, two jars for each cucumber lot, using a 50:50 (w/v, cucumbers/cover brine) pack-out ratio. The cover brine was prepared so that it equilibrated with the cucumbers at 80 mM CaCl 2 (Brenntag), 6 mM potassium sorbate (Mitsubishi International Food Ingredients), 10.1 mM Ca(OH) 2 (Sigma-Aldrich), and 44 mM acetic acid, added as 20% vinegar (Fleischmann Vinegar), to adjust the initial pH to 4.7 ± 0.1.  Table 1) was prepared as described below and supplemented to 10 5 CFU/ml. Jars were closed with commercial metal lug caps that were heated in boiling water for 10 s to soften the plastisol liner. Each lid was equipped with a rubber septum in its center to allow for sampling of cover brine using a 10-ml syringe attached to a 18G × 1 1/2″ needle (Becton Dickinson Co.). The jars were incubated at 30°C for 10 days. The pH was measured using a Fisher Accumet pH meter (Model AR25, Fisher Scientific) combined with a Gel-Filled Pencil-Thin pH Combination Electrode (Acumet, Fisher Scientific).
The completion of the fermentation was confirmed by measuring sugars, organic acids, and ethanol in cover brine samples using the HPLC analysis conducted as described below. At the end of these fermentations, the pH was determined to be 3.3 ± 0.1 and the media contained 0.5 ± 0.2 mM and 1.69 ± 0.6 mM glucose and fructose, respectively.
The cover brine and juice from fermented cucumbers was used to prepare FCJM. Fermentation cover brines were decanted from the jars into 2-L beakers. The fermented cucumbers were passed through an automatic juice extractor (Juiceman Jr. Model JM-1, Beachwood, Ohio, USA) to separate the pulp from the liquid content on a jar volume basis. The pulp remaining in the fermented cucumber juice was removed by straining with a 100% cotton cheesecloth (grade #90, 44 × 36 threads/inch, Cartridge Plus, Inc., Riva, MD, USA) and a subsequent centrifugation at 3,750 × g for 15 min at ambient temperature using a bucket rotor (Eppendorf Centrifuge Model 5810, Hamburg, Germany). The respective clear fermented cucumber juices and fermentation cover brines were mixed to the same ratio as in the fermentation jars to make up the FCJM. The FCJM derived from each cucumber lot were independently used and supplemented with gentiobiose (Sigma-Aldrich G-3000, 85% purity) or cellobiose (98% purity) as needed.
The pH of the supplemented and un-supplemented FCJM was adjusted to 4.7 ± 0.1 or 3.7 ± 0.1 as needed using a 5N NaOH solution (Spectrum Chemicals, NJ, USA) and 3N HCl (Spectrum Chemicals). The pH-adjusted FCJM were filter sterilized using 0.2-µ filtration units (Nalgene ® -Rapid Flow™, Thermo Scientific).
Aliquots of 10 ml of each FCJM were aseptically transferred to 15-mL conical tubes for experimentation.

| LAB cultures preparation
The bacterial cultures used for experimentation are described in  Fresh cucumber juice was prepared in the same way the fermented cucumber juice was prepared (as described above). The fresh cucumber juice was also filtered-sterilized prior to inoculation. The FCJM was inoculated to 10 5 CFU/ml. The optical density at 600 nm of the MRS or fresh cucumber juice cultures was measured using a Novaspec II, (Pharmacia, Stockholm, Sweden) and used to adjust the inoculation level. A sterile 0.85% NaCl (Sigma-Aldrich) solution was used to adjust the inocula concentration as needed. Inocula with mixed cultures was prepared by combining the cells suspension in saline solution so that each strain will be at 10 5 CFU/mL in the FCJM.

| Evaluation of the ability of certain LAB to utilize gentiobiose and cellobiose in FCJM
The experimental design included the testing of the ability of three strains of L. pentosus, L. plantarum, L. brevis or L. buchneri (Table 1) TA B L E 1 Description of the lactic acid bacteria strains used in this study propionic, and butyric acids. An RID-10A refractive index detector (Shimadzu Corporation) connected in series with the diode array detector was used to measure acetic acid, lactic acid, glucose, fructose, and ethanol. External curves were also run using at least five concentrations of glucose, fructose, lactic acid, acetic acid, and ethanol for quantification purposes.

| Evaluation of the ability of single strains of L. pentosus and L. plantarum to utilize gentiobiose and cellobiose in FCJM under aerobiosis and anaerobiosis
The experimental design included the testing of the ability of three Additionally, the cultures were inoculated in the un-supplemented FCJM. All FCJM cultures were incubated at 30°C for 7 days to enable secondary fermentation. Samples were aseptically collected at the end point of incubation and plated on Lactobacilli MRS agar plates to determine colony counts as described above. The pH and fermentation biochemistry were monitored at the end point as described above.

| Cellobiose and gentiobiose concentration in fresh and fermented cucumber samples
The potential role of cellobiose and gentiobiose as energy sources in cucumber fermentation for certain LAB was defined. It was found that although cellobiose and gentiobiose had been detected in cucumber fermentation spoilage samples using two-dimensional gas   Table 2). The presence of putative cellobiase in the L.

| Scrutiny of the putative pathways for cellobiose and gentiobiose utilization by lactic acid bacteria using a bioinformatic analysis
plantarum genomes was strain dependent. L. brevis and L. buchneri had some glycolysis pathway-related enzymes missing (Table 2). buchneri was incomplete, presumably due to a slower acid production rate or the diversion of the carbons to other products, such as propanediol, that were not measured (Figure 1; Johanningsmeier & McFeeters, 2013). A slower acid production is also evidenced by the higher pH (4.53 ± 0.04) after 7 days of incubation, as compared to the other LAB cultures (Figure 1).

| Ability of certain LAB to utilize cellobiose and
All the strains tested were able to utilize gentiobiose (Figure 2a).
Except for L. plantarum, the lactobacilli converted gentiobiose into lactic acid, acetic acid, and ethanol in close to a heterofermentative ratio of 2:1:0.5 (Figure 2b). L. plantarum converted 17.81 ± 3.14 mM gentiobiose to 102.07 ± 14.07 mM lactic acid (Figure 2b). The ability to utilize the disaccharides resulted into microbial growth in the FCJM as determined by plating on MRS agar plates (Figure 2a).
Changes in pH at the end point fluctuated between 0.4 and 0.7 pH units ( Figure 2a).
Inclusion of the gentiobiose anomeric form, melibiose, in this study, was an unintended consequence associated with the use of the particular commercial preparation. The gentiobiose commercial preparation contained 15% melibiose as an impurity (~1.5 to 3 mM in the FCJM). Given the magnitude of the products concentration as the result of gentiobiose utilization, it is speculated that melibiose was also utilized by the LAB tested. Melibiose is rarely present in nature (Gänzle & Follador, 2012). However, transport systems for melibiose are suspected in L. plantarum (Tamura & Matsushita, 1992). conostocs, and Enterococcus thailandicus to utilize cellobiose (Carr et al., 2002;Hammes & Hertel, 2015;Mao et al., 2015;Sterr, Weiss, & Schmidt, 2009;Tamang et al., 2005). A cellobiose permease putative gene was not found in the LAB of interest except for L. buchneri.
Contrary to L. brevis, L. buchneri has been associated with spoilage of fermented cucumbers (Franco et al., 2012;Johanningsmeier & McFeeters, 2013 The LAB used in this study were able to utilize gentiobiose and cellobiose in FCJM suggesting that there are enough of other growth factors, such as amino acids, nucleosides, minerals, etc., in the fermented cucumber juice to sustain microbial proliferation. Furthermore, growth of L. plantarum and L. pentosus in the FCJM at pH 4.7 ± 0.1 was observed in the absence of added substrates. Two parameters were modified in the FCJM to enable microbial growth which were pH and the supplementation with an energy source. These observations confirm that the potential for spoilage in a given cucumber fermentation batch could be assessed by inoculating certain spoilage organisms in the corresponding FCJM as proposed by Fleming, McFeeters, and Thompson (1983). This approach represents a tool that could prevent important economic losses at the industrial scale production. More relevant is the understanding that batches of fermented cucumbers are primarily stable during long-term storage due to the development of an extremely acidic pH and to a lesser extent due to the lack of readily available energy sources. Both of these parameters can change as a function of microbial metabolism or enzymatic activities.

| Ability of three strains of L. pentosus and L. plantarum to utilize cellobiose and gentiobiose under aerobiosis and anaerobiosis at an initial pH of 4.7 and 3.7
None of the disaccharides were utilized by the six strains of L. plantarum and L. pentosus scrutinized when the initial pH was adjusted to 3.7 ± 0.1. This observation is supported by the lack of a significant change in pH (ANOVA test at a p > .05), a reduction in colony counts to below detection limits and no changes in the concentrations of cellobiose and gentiobiose supplemented in the FCJM (data not shown). On the contrary, both bacterial species utilized the disaccharides under aerobiosis or anaerobiosis when the pH of the FCJM was adjusted to 4.7 ± 0.1 (Figures 3 and 5). Although the utilization of the disaccharides by L. plantarum and L. pentosus in FCJM at pH F I G U R E 2 Disappearance of gentiobiose from FCJM inoculated with lactic acid bacteria. The FCJM pH was 4.7 ± 0.1. It is estimated that the gentiobiose source utilized contained at least 3 mM of the α-anomer, melibiose. The colony counts from MRS (a), pH (a), and metabolites (  3.7 ± 0.1 could occur during a prolonged incubation, β-glucosidases are inhibited at pH below 4.0 (Kim, Lee, & Ma, 2017;Takase & Horikoshi, 1988;Yeoman et al., 2010;Zhong et al., 2016). While β-glucosidases occur in many organisms, the activity of the enzymes derived from thermophilic bacteria and lactobacilli is known to be severely compromised at a pH of 4.0 with a 20% enzyme stability, as compared to a 40% stability at pH 5.0 (Kim et al., 2017;Takase & Horikoshi, 1988;Yeoman et al., 2010;Zhong et al., 2016). Growth of L. plantarum is known to stop at a pH of 3.3 with the cessation of acid production at a pH of 3.0 (McDonald, Fleming, & Hassan, 1990).
Thus, a more reasonable interpretation is that L. pentosus and L. plantarum failed to utilize cellobiose and gentiobiose in FCJM at pH 3.7 given the lack of a β-glucosidase activity.
While the strains isolated from cucumber or cabbage fermentations were able to utilize cellobiose under aerobiosis or anaerobiosis in FCJM at pH 4.7 ± 0.1, strains of L. plantarum (WCSF1) isolated from saliva and L. pentosus (ATCC8041) obtained from sauerkraut (Fred, Peterson, & Davenport, 1921) were not able to utilize cellobiose in the same medium under either condition of oxygen availability (Figures 3 and 4). Some variations in the ability to utilize

Anaerobic Aerobic
A reduction in the cell concentration of the strains unable to utilize cellobiose was observed in FCJM at pH 4.7 ± 0.1 (Figure 3).

Lactobacillus pentosus Strains
Aerobic observed after gentiobiose was utilized ( Figure 5). The increases in cell densities ranged between 2 to 3 log CFU/ml as the result of gentiobiose utilization in FCJM at pH 4.7 ± 0.1 and were 1 log CFU/ml higher than those observed for cellobiose utilization under the same conditions (Figures 3 and 5).
In summary, LAB could utilize the disaccharides in cucumber fermentation, which become available as the result of tissue degradation due to spoilage or the enzymatic hydrolysis of cellulose.
Despite some differences at the strain level, both L. pentosus and L. plantarum, which are the leading microbes in cucumber fermen-  (Table 2). The inability to predict the same for the L. pentosus species may be associated with the fact that 107 L. plantarum genomes were used for the analysis instead of three genomes for the former.

| CON CLUS IONS
The magnitude of the natural content of gentiobiose and cellobiose in fresh and fermented cucumbers and the utilization of the two disaccharides by certain LAB was determined. The plant-derived disaccharides were utilized by L. plantarum, L. pentosus, and L. buchneri to variable extents in FCJM. L. brevis was unable to utilize cellobiose efficiently in FCJM. Cellobiose was homofermentatively utilized by LAB at pH 4.7 ± 0.1 but not at 3.7 ± 0.1. Supplementation of gentiobiose to FCJM at pH 4.7 to 8 and 18 mM resulted in homo-and heterofermentations, respectively. Some strain level differences were observed with regard to cellobiose utilization, but not in the conversion of gentiobiose. L. plantarum and L. pentosus were able to proliferate in FCJM at pH 4.7 in the absence of added energy sources and produced between 15 and 46 mM lactic acid. The ability of the LAB of relevance to cucumber fermentation to utilize the disaccharides may be of industrial concern, if the disaccharides become available from the degradation of cellulose by a postfermentation and prespoilage microbiota.

ACK N OWLED G M ENTS
The authors thank Ms. Sandra Parker and Ms. Janet Hayes and Mr.