Phenolic compounds isolated from fermented blueberry juice decrease hepatocellular glucose output and enhance muscle glucose uptake in cultured murine and human cells

All cell lines used – H4IIE (rat hepatoma), HepG2 (human hepatoma) and C2C12 (murine skeletal myoblasts) – were purchased from American Type Culture Collection (ATCC; Manassas, VA, USA). H4IIE cells were grown in a high glucose Dubelcco’s Modified Eagle Medium (DMEM) containing 10% Fetal Bovin Serum (FBS) and 0.5% antibiotics (PS: Penicillin 100 U/mL, Streptomycin 100 μg/mL). HepG2 cells were grown in DMEM/F12 (50/50) medium, containing 10 FBS and 0.5% PS. On the other hand, C2C12 myocytes were cultured in DMEM medium containing 10 FBS, 10 HS and 0.5% PS then switched to DMEM medium containing 2% HS to initiate differentiation. Glucose uptake assay was performed on differentiated myotubes at the 7^th day of differentiation. All cells were cultured and incubated at 37 °C with 5% CO₂ in 12-well plates for glucose uptake and G6Pase experiments and in 6-well plates for GS experiments. Overnight treatment (16–18 h) with the different samples was initiated prior to the determinations.

Serratia vaccinii were cultured as previously described [12]. The juice was inoculated with a saturated culture of Serratia vaccinii at (7.5 ± 0.3) log CFU ml − 1 corresponding to 2% of the total juice volume. A control flask was prepared under the same conditions but without inoculation. The blueberry preparations were incubated in a Lab-Line low-temperature benchtop incubated shaker (Lab-Line Instruments, Inc, Melrose Park, IL, USA) in 250-ml flasks at 22 °C, 120 rev min − 1, under aerobic conditions. After a 4 day fermentation period, the FJ was sterilized by 0.22 μm filtration as detailed elsewhere [12].

In order to facilitate the handling and insure the stability of the blueberry preparations, they were freeze-dried to produce powdered material that was kept at −20 °C until use.

Organic solvents were purchased from Fisher Scientific (Burlington, ON, Canada), unless otherwise mentioned. Normal blueberry juice has been partially characterized elsewhere [12, 13, 20, 21]. In the present studies, we carried out further phytochemical analysis using HPLC as described below. Figure 1 shows the fractionation scheme that was carried out to identify active compounds present in fermented blueberry juice (FJ). The starting material for the fractionation process was either normal blueberry juice prepared from wild blueberries (the control Juice, CJ), or the FJ in batches of approximately 500 ml. These were placed onto 29.5 cm × 5 cm chromatography columns (pre-conditioned with 1 column volume methanol then 2 column volumes water) containing Waters preparative C18 resin (125 Å, 55–105 μ) then washed with 2 column volumes of water, which was sufficient to remove sugars and organic acids (discarded). The phenolic compounds were eluted from the column using 1.2 column volumes of 13 mM trifluoroacetic acid (Sigma Aldrich, Oakville, ON) in ethanol. The ethanol eluent was dried using rotary evaporation and lyophilized to give the Phenolic fraction (F2). The CJ underwent no further fractionation after this initial step. On the other hand, a portion of F2 was dissolved in water and further fractionated on C18 column as outlined above. The first subfraction (F2.1) was eluted using 4 column volumes of 0.16 M HCl (Ricca Chemical Company, Texas, USA) and 2.06 M ethanol in water. This subfraction was composed mainly of gallic acid, protocatechuic acid, and catechol, as confirmed by HPLC (comparing retention times and UV-vis profiles of the peaks to standards). The next subfraction (F2.2) was eluted using an additional 2 column volumes of 0.16 M HCl and 2.06 M ethanol in water and was rich in chlorogenic acid (as confirmed by HPLC). The remaining bound materials, mainly flavonoids, were eluted using 0.16 M HCl and 13.7 M ethanol in water and were called the flavonoid fraction (F2.3). The three fractions were dried using rotary evaporation and lyophilisation. Next, a portion of the F2.3 fraction was dissolved in 4.28 M ethanol and applied to a 34.5 cm × 5 cm lipophilic Sephadex column (LH-20, 25–100 μ, Sigma Aldrich, Oakville, ON, Canada). HCl was neutralized and three fractions were collected. The first fraction enriched in anthocyanins (F2.3.1) was eluted using 7 column volumes of 4.28 M ethanol. The second fraction enriched in heteropolymers (F2.3.2) was then eluted using 3 column volumes of 8.56 M ethanol. Finally, a third fraction enriched in proanthocyanidins (F2.3.3) was eluted using 3 column volumes of 9.53 M acetone. All three fractions were dried using rotary evaporation and lyophilisation.

The cytotoxicity assay was based on a lactate dehydrogenase (LDH) release kit (LDH Colorimetric Kit; Roche, Mannheim, Germany) and served to determine maximum non-toxic concentration of each sample in H4IIE, HepG2 and C2C12 cells. As was described previously [22], the cells were treated overnight (16–18 h) with CJ, FJ, fractions thereof, or pure compounds at different concentrations. The culture media were collected separately for each condition and kept on ice (representing released LDH from cells).

Then the cells were lysed by adding culture medium with 1% Triton, for 10 min at 37 °C, 5% CO₂ (representing cellular LDH). All the samples were centrifuged at 250×g, 4 °C for 10 min and kept on ice in Eppendorf tubes. The ratio of released LDH to total LDH (total LDH = released LDH + cellular LDH) was calculated for each condition and results normalized to the values obtained from cells treated with the vehicle control (DMSO), always present at a final concentration of 0.1%. Maximum non-toxic concentrations for each sample were the highest ones that yielded LDH release comparable to that of DMSO controls.

H4IIE cell line was used to measure the activity of G6Pase. Confluent cells were treated overnight (16–18 h) with DMSO 0.1% (vehicle control), insulin 100 nM (positive control), CJ, FJ, each of the seven fractions (at 5 μg/mL) or each of the four pure compounds, all at their maximal non-toxic concentrations (Additional file 1: Table S1). After the treatment, cells were rinsed with PBS then lysed using a 15 mM phosphate buffer containing 0.05% Triton and 1.3 mM Phenol (pH = 6.5). A glucose-6-phosphate-containing buffer (200 mM) was then added to the cell lysates for 40 min at 37 °C; G-6-P contained in this buffer served as a substrate for endogenous G6Pase to yield glucose. A Wako AutoKit Glucose colorimetric assay (Wako Chemicals, Richmond, VA, USA) was used to determine the quantity of glucose generated in this reaction according to manufacturers’ recommendations. The BCA method was used to determine the protein content for each condition. Results were expressed relative to vehicle control (DMSO 0.1%).

HepG2 cells were used to measure GS activity since this cell line exhibit a better expression of this enzyme compared to the H4IIE hepatocytes [23]. After achieving confluence in 6 well plates, the cells were treated overnight (16–18 h) with either vehicle control (DMSO 0.1%), the maximal non-toxic concentrations of CJ, FJ, each of the seven fractions (5 μg/mL) or each of the four pure compounds (Additional file 1: Table S1). Treatment with insulin at 100 nM for 15 min was used as a positive control. After treatment, cells were rinsed with PBS then lysed in a buffer solution containing 50 mM glycylglycine, 100 mM sodium fluoride, 20 mM EDTA, 0.5% glycogen, pH 7.4 and a complete protease inhibitor cocktail added just before the assay. The lysates were centrifuged at 1000×g for 20 min at 4 °C. After centrifugation, 30 μL of supernatant from each condition were added to 100 μL of a specific buffer solution to measure active GS (25 mM glycylglycine, 0.275 mM UDP-glucose, 0.12 μCi/mL U-¹⁴C UDP-glucose, 1% glycogen, 1 mM EDTA, 10 mM sodium sulfate, pH 7.5) and another 30 μL of supernatant were added to 100 μL of a specific buffer solution to measure total GS (25 mM Tris, 5 mM UDP-glucose, 0.12 μCi/mL U-¹⁴C UDP-glucose, 1% glycogen, 3 mM EDTA, 5 mM glucose-6-phosphate, pH 7.9). The tubes were incubated in a water bath at 30 °C for 120 min. After incubation, 90 μL of the mix for each condition was transferred on Whatman 31 ET chr 2 cm² paper. The papers were rinsed with cold ethanol 66% (4 °C) for 30 min, then twice with ethanol 66% at room temperature for 30 min. After removing the ethanol, the papers were dipped in acetone for 2–3 min and dried. The papers were then transferred into scintillation vials. The radioactivity was measured using a liquid scintillation counter (LKB Wallac 1219; Perkin Elmer, Woodbridge, ON, Canada).

C2C12 myocytes were grown in 12-well plates to 60% confluence then differentiated into myotubes over a 7-day period. On day 6 of differentiation, C2C12 cells were treated overnight (16–18 h) with either 0.1% DMSO (vehicle control), the maximal non-toxic concentrations of CJ, FJ, each of the seven fractions (12.5 μg/mL) or each the four pure compounds (Additional file 1: Table S1). Metformin (400 μM) was used as a positive control in similar conditions. After treatment, cells were rinsed twice with a warm Krebs phosphate buffer (KPB: 20 mM Hepes, 4.05 mM Na₂HPO₄, 0.95 mM NaH₂PO₄, 136 mM NaCl, 5 mM glucose, 4.7 mM KCl, 1 mM CaCl₂, 1 mM MgSO₄, pH 7.4) then incubated with KPB for 30 min at 37 °C. At this point, insulin (100 nM) was added to specific wells to act as another positive control (incubation in KPB buffer for 30 min). Cells were then rinsed twice with warm glucose-free KPB, then incubated in glucose-free KPB containing 0.5 μCi/mL 2-deoxy-D-³H-glucose (TRK-383, Amersham Biosciences, Baie d’Urfé, Canada) for 10 min at 37 °C. After incubation, cells were kept on ice and rinsed 3 times with ice-cold glucose-free KPB then lysed in 0.5 μL of NaOH (0.1 M) for 30 min. The lysates were transferred with 1 mL of water to scintillation vials, then 4 mL of liquid scintillation cocktail (Beckman Coulter, Fullerton, USA) was added to each vial and incorporated radioactivity was measured using a liquid scintillation counter (LKB Wallac 1219; Perkin Elmer, Woodbridge, ON, Canada).

After overnight treatment of H4IIE, HepG2 and differentiated C2C12 cells with CJ, FJ, each of the seven fractions or each of the four pure compounds at different concentrations, LDH released and total LDH were measured for each condition. Additional file 1: Table S1 shows the maximum non-toxic concentration determined for each sample and each cell line based on the results of LDH test.

As illustrated in Fig. 2a, FJ significantly decreased G6Pase activity (31% reduction; p < 0.05) as compared to vehicle control (0.1% DMSO). This effect represented roughly half of the effect of the insulin positive control (67% decrease). In contrast, CJ did not induce a significant change in the activity of G6Pase. After fractionation of FJ, the seven fractions were tested in the G6Pase bioassay and all of them showed varied, yet statistically significant, inhibitory effect on the enzyme’s activity that ranged from 20 to 61% when compared to the DMSO (0% inhibition reference; Fig. 2a).

Figure 2b presents results concerning GS activity. FJ induced a significant increase in the enzyme’s activity (2.3 fold increase, p < 0.001) identical to that of the positive control, insulin (2.3 fold activation). As observed with G6Pase activity, CJ was without any significant effect. In addition, all seven fractions derived from FJ were able to significantly increase the activity of GS and ranged from 1.9- to 2.7-fold when compared to the DMSO vehicle control (100% activation; Fig. 2b).

When tested for their ability to enhance glucose uptake in C2C12 muscle cells, FJ showed a 20% stimulation of this uptake (p < 0.01) whereas CJ was without any effect (similar to vehicle control, DMSO 0.1%; Fig. 3). Insulin used as the positive control stimulated glucose transport into cells by 33% (p < 0.001). The phenolic fraction (F2) along with its subfractions F2.1 and F2.2, F2.3.2 were able to enhance glucose uptake into muscle cells by 25, 18 and 15% respectively (p < 0.01). Bonferroni post hoc test did not reveal significant differences between FJ, F2, F2.1, and F2.2. The other fractions were without effect when compared to vehicle control (100% activation; Fig. 3).

Results of G6Pase and GS activity as well as glucose uptake bioassays were used to guide the identification of compounds found in the active fractions. HPLC analysis confirmed that the three fractions F2, F2.1 and F2.2 were composed mainly of four phenolic compounds: chlorogenic acid (CA), gallic acid (GA), protocatechuic acid (PA) and catechol (Cat). Compound identification was confirmed by comparing retention times and UV-vis profiles of each peak with respective standards (Fig. 4 and Additional file 1: Table S2). Calculation of phenolic compounds concentrations was based on the external standard method. The concentrations of individual phenolic compounds in FJ were as follows: CA (3.0 mg/g DW), GA (3.2 mg/g DW), PA (0.3 mg/g DW) and GA (3.2 mg/g DW).

As mentioned above, the most active fractions (F2, F2.1 and F2.2) were rich in phenolic compounds mainly CA, GA, PA and Cat. These compounds were tested at their maximal non-toxic concentration in the G6Pase and the GS assays. Two of the four compounds, namely CA and PA, did not significantly decrease G6Pase activity as compared to DMSO. In contrast, GA was able to reduce the enzyme activity by 25% (p < 0.05) whereas Cat yielded a 54% decrease in G6Pase activity (similar to the positive control, insulin) compared to the vehicle control DMSO 0.1% (p < 0.001, Fig. 5a). Games-Howell test showed that difference between Cat and GA was statistically significant (p < 0.05). In the GS assay, all of the four compounds were able to significantly increase the activity of the enzyme by about 2 fold (CA) or slightly less (GA, PA and Cat) when compared to DMSO 0.1% vehicle control (p < 0.05, Fig. 5b). These effects compared favorably with the effect of insulin (2.3 fold increase). No statistically significant differences were detected between the compounds.

Along with their ability to regulate key enzymes implicated in hepatic glucose output, CA, GA and Cat significantly enhanced glucose uptake in C2C12 cells by 15% (p < 0,01), 16% (p < 0.05) and 43% (p < 0.001) respectively. In contrast, PA had no effect when compared to the vehicle control DMSO 0.1% (Fig. 6). Games-Howell test revealed that Cat yielded the greatest activity, it differed significantly from CA (p < 0.01) and GA (p < 0.05). CA and GA were not statistically different.