Food grade resveratrol was purchased from ChromaDex (Irvine, CA). GSE was purchased from supplement Warehouse (UPC: 603573579173). Only one lot of the resveratrol and one lot of the GSE were used for this particular study. Both resveratrol and GSE have been shown to be very stable when stored at 4°C in dark. Welch Concord purple grape juice was purchased at a local market and concentrated by solid phase extraction (SPE), as previously described, to produce Concord Grape Juice total polyphenol extract (Ho et al., 2009). The polyphenol profile of this juice extract was confirmed by LC-MS analysis and previously reported (Xu et al., 2011). Total polyphenol content was determined to be 62 mg of gallic acid equivalents per mL extract, with principal components including anthocyanins, PAC oligomers, phenolic acids, and quercetin glycosides. The juice extract was stored at −20°C in the dark and was diluted to the desired concentration once every 3 days. (+)-catechin, (−)-epicatechin, quercetin-3-O-glucoside, quercetin-3-O-glucuronide, and quercetin standards were purchased from Sigma Chemical Co. (St. Louis, MO). Malvidin-3-glucoside chloride and cyanidin-3-glucoside chloride were purchased from ChromaDex (Irvine, CA). All extraction and LC solvents were HPLC certified and were obtained from J.T. Baker (Phillipsburg, NJ).

Plasma PK and brain bioavailability studies were conducted using male Sprague-Dawley (SD) rats. SD rats were obtained from Harlan Sprague Dawley Inc. (Indianapolis, IN) and placed on a polyphenol free AIN-93M diet (Dyets, Bethlehem, PA), given deionized water ad lib, and allowed to acclimate for 3 days. Following the acclimation period, rats were given a combination of resveratrol, GSE and Concord juice extract (referred to as Comb), by intragastric gavage using plastic feeding tubes (Instech FTP-15-78, Plymoth Meeting, PA) for 10 days. To reach proper dosage, rats were gavaged twice a day, with 8 h separation. The non-treatment group was gavaged with water. Two days prior to PK studies, rats were anesthetized by a dose 3–5% of isoflurane in the anesthesia chamber and maintained with a mask of 1.5–3% isoflurane. A polyethylene catheter was implanted into the jugular vein. Rats were injected with Buprenex (0.01–0.05 mg/kg) before regaining consciousness to alleviate pain and allowed to rest for 24 h after surgery. Catheters were kept patent by flushing with heparinized saline (100 units/mL) every 12 h. Prior to initiation of PK studies, rats fasted for 8 h. PK assessment was conducted on the 10th day of gavage by collecting 400 μ L of blood at baseline (prior to gavage), 0.25, 0.5, 1, 2, 4, 6, 8 h postgavage from the jugular catheter into heparinized tubes. Food was offered at 2 h following initiation of PK studies. Fresh blood was processed to plasma by centrifugation at 5500 rpm for 10 min at 4°C, acidified with saline (1% ascorbic acid wt/v) in 4:1 ratio, purged with N₂, and stored at −80°C until analysis. The day after PK, rats were given a last dosage of treatment, anesthetized, and perfused with cold physiological saline. The brains were harvested, placed in 0.2% ascorbic acid in saline, and stored at −80°C until analysis.

Acidified plasma samples were thawed and brought up to 0.5 mL with acidified saline (0.1% formic acid v/v). Polyphenol metabolites from plasma and brains were extracted using SPE as previously described (Vingtdeux et al., 2010; Wang et al., 2012; Ho et al., 2013). Polyphenol metabolites were extracted with methanol and dried under vacuum. Dried residues were reconstituted with 0.1% aqueous formic acid (v/v), loaded on the preconditioned SPE cartridge, eluted with methanol, and dried. The dried phenolic extracts were then reconstituted with 0.1% aqueous formic acid (v/v) and 0.1% formic acid in acetonitrile (v/v) in 4:1 ratio and sonicated, followed by LC-MS/MS analysis.

Analysis of all polyphenol metabolites and quercetin metabolites and PAC derivatives was performed on an Agilent 6400 Triple Quad under multiple reaction monitoring modes (MRM). Both systems were equipped with an ESI source. A Waters XTerra RP-C18 column (2.1 × 100 mm, 3.5 μm particle size) was used on all analysis. A binary mobile phase system, consisting of mobile phase A: 0.1% aqueous formic acid (v/v) and B: 0.1% formic acid in acetonitrile (v/v), was used for analysis of monomeric PACs, resveratrol, and quercetin metabolites. The column was heated to 30°C and the system flow rate was 0.3 mL/min. Gradient conditions were: 10% B at 0 min, 40% B at 5.5 min, 70% B at 7 min, 95% B at 7.5 min and back to 10% B at 8.5 min to 13.5 min. Mass spectra was obtained under negative polarity scanning between 100 m/z-1000 m/z on MS-TOF. MRM mass transitions were 479 → 303 for MeO-EC glucuronides, 465 → 289 for EC-glucuronides, 289 → 137 for EC, 403 → 227 for resveratrol-glucuronide, and 227 → 143 for resveratrol under negative polarity on Triple Quad. MRM mass transitions were 493 → 317 for MeO-quercetin glucuronide, 479 → 303 for quercetin glucuronide, and 301 → 153 for quercetin aglycone under positive polarity. Fragmentor voltage was set at 135V and collision energy was 30 eV for quercetin aglycone and 17 eV for all other mass transitions. ESI source condition was described as followed: gas temp was 350°C, drying gas flow was 11 l/min, nebulizer was 30 psi, sheath gas temp was 350°C, sheath gas flow was 11 l/min, capillary voltage was 3500 V, and nozzle voltage was 1000 V. Quantification of PAC metabolites and resveratrol were estimated using calibration curves from parent catechin, epicatechin, and resveratrol, respectively. Quantification of quercetin metabolites was accomplished using a calibration curve constructed with quercetin-3-O-glucuronide standard. For anthocyanin analysis, the binary mobile phases were A: 2% aqueous formic acid (v/v) and B: 0.1% formic acid in acetonitrile (v/v). The gradient used to elute anthocyanins was: 5% B at 0 min, 10% B at 10 min, 25% B at 30 min, 5% B at 31 min, and continue at 5% B until 35 min. Mass spectra was obtained under positive polarity. ESI source condition setting was the same as described above. MRM transitions were: 493 → 331 for malvidin-3-O-glucoside, 479 → 317 for petunidin-3-O-glucoside, 465 → 303 for delphinidin-3-O-glucoside, 463 → 301 for peonidin-3-O-glucoside, and 449 → 287 for cyanidin-3-O-glucoside. Quantification of all anthocyanin glucosides, except for cyanidin-3-O-glucoside, was estimated using calibration curves of malvidin-3-O-glucoside. Quantification of cyanidin-3-O-glucoside was achieved by a calibration curve constructed with cyanidin-3-O-glucoside standard.

Female J20 AD transgenic mice, expressing human amyloid precursor protein (APP) containing both the familial AD Swedish (K670N/M671L) and the Indiana (V717F) mutations (APPSwInd) under the human platelet-derived growth factor beta polypeptide (PDGFB) promoter (Mucke et al., 2000), were purchased from the Jackson Laboratory. All mice were housed with food and water available ad libitum and maintained on a 12:12-h light/dark cycle with lights on at 07:00° h in a temperature-controlled (20 ± 2°C) room prior to experimental manipulation. All procedures and protocols were approved by the Icahn School of Medicine at Mount Sinai's Institutional Animal Care and Use Committee (IACUC) through the Center for Comparative Medicine and Surgery. Mice were randomly grouped into 5 groups, the non-treated control group (CTRL); the resveratrol-treated group (Resv): 400 mg/kg/day mixed with food; the GSE-treated group (GSE): 200 mg/kg/day mixed with food; the juice extract-treated group (Juice): 183 mg/kg/day total polyphenol; and the group treated with the combination of resveratrol, GSE, and juice (Comb). The doses chosen for each component was based on equivalent doses used in studies that showed efficacy either in human or animal models (Lagouge et al., 2006; Wang et al., 2008, 2010; Krikorian et al., 2010; Vingtdeux et al., 2010). The treatment started at 3 months of age and the animals were treated for 7 months.

Spatial learning memory was assessed by the Morris water maze (MWM) behavioral test, as previously described (Morris, 1984; Wang et al., 2007). Mice were tested in a circular pool filled with water mixed with non-toxic white paint (Dick Blick Art Materials, IL). In the initial learning phase, mice were trained for 7 consecutive days, which allowed them to learn to escape from the water onto a hidden/submerged (1.5 cm below-water surface) escape platform (14 × 14 cm) in a restricted region of the pool using the spatial cues provided. Spatial learning memory is assessed by recording the latency time for the animal to escape from the water onto the submerged escape platform as a function of the number of learning trials during the learning phase. Twenty-four hours after the last learning session, mice were subjected to a 45 s probe trial wherein the escape platform was removed. Spatial memory retention is reflected by the percentage of time animals spent within the “target” quadrant of the pool that previously contained the hidden escape platform. Water maze activity during training and probe trials was monitored with the San Diego Instrument Poly-Track video tracking system (San Diego, CA).

Total Aβ ₁₋₄₀ or Aβ ₁₋₄₂ in the brain was quantified by sandwich ELISA (BioSource, Camarillo, CA), as previously described (Wang et al., 2005). Plaque burden was analyzed as previously described (Wang et al., 2005). Briefly, tissue was fixed with 4% paraformaldehyde in phosphate buffer and 0.125% glutaraldehyde. Fixed brain hemispheres were sectioned on a Vibratome at 50 μm and stained with thioflavin-S. Micrographs were taken and plaque burden was quantified using Image J software that converts micrograph to binary images for plaque number and plaque area assessment.

Wild type mice were sacrificed by decapitation and the brains were quickly removed. Hippocampal slices (350 μm) were made and placed into oxygenated artificial cerebrospinal fluid (ACSF) at 29°C for a minimum of 90 min to acclimatize. Slices were then transferred to a recording chamber (Fine Science Tools Inc, CA, USA) and perfused continuously with oxygenated-ACSF at 32°C. Slices were treated with 200nM oligomeric Aβ in the presence or absence of 300 nM of 3′-O-methyl-epicatechin-O-β-Gluc, 300 nM of cyanidin-3-O-Glc, or 20 μ M resveratrol for 60 min before transferring to the recording chamber. The use of 20 μ M resveratrol is based on the in vitro activation of AMPK study (Vingtdeux et al., 2010). For extracellular recordings: CA1 field excitatory postsynaptic potentials (fEPSPs) were recorded by placing stimulating and recording electrodes in CA1 stratum radiatum as previously described (Gong et al., 2004). BST was not assayed in this particular study because previous studies demonstrated that oligomeric Aβ inhibits LTP in the CA1 region of mouse hippocampal slices without affecting BST (Wang et al., 2002; Olsen and Sheng, 2012). For LTP experiments, a 40–60 min baseline recording period preceded the θ-burst stimulation. Baseline was recorded every min at an intensity that evokes a response ~35% of the maximum evoked response. LTP was induced using θ-burst stimulation (4 pulses at 100 Hz, with the bursts repeated at 5 Hz and each tetanus including three 10-burst trains separated by 15 s) and fEPSPs were monitored for up to 140 min to assess the magnitude of potentiation.

To test whether polyphenols delivered in combination would change the brain bioavailability, we treated the SD rats with Comb and assessed the plasma PK and brain polyphenol metabolites' profiles and compared these profiles with those from previous studies on individual polyphenol preparation (Vingtdeux et al., 2010; Wang et al., 2012; Ho et al., 2013). Consistent with previous findings (Wang et al., 2008; Vingtdeux et al., 2010), all metabolites were found in the plasma and brain following 10-day Comb treatment: catechin and epicatechin from GSE led to the accumulation of PAC glucuronides (Gluc) including catechin-O-β-Gluc, 3′-O-methyl-catechin-O-β-Gluc, epicatechin-O-β-Gluc, and 3′-O-methyl-epicatechin-O-β-Gluc; quercetin from juice resulted in the accumulation of quercetin-Gluc, O- methyl quercetin-Gluc, malvidin-glucoside (Glc), petunidin-Glc, delphinidin-Glc, and peonidin-Glc; cyanidin-Glc from juice led to the plasma and brain accumulation of malvidin-3-O-glucoside (Glc), petunidin-3-O-Glc, delphinidin-3-O-Glc, peonidin-3-O-Glc, and cyanidin-3-O-Glc; resveratrol led to the accumulation of resveratrol and resveratrol Gluc in the brain. See Table 1 for detailed plasma PK response and brain accumulation of polyphenol metabolites in rats treated with Comb.

We treated J20 mice with Resv, GSE, Juice, or Comb, starting at 3 months of age. Treatments continued for 7 months until mice were approximately 10 months of age. Consistent with our previous studies (Wang et al., 2008; Vingtdeux et al., 2010), we found that polyphenolic treatment from all groups were well-tolerated; we observed no adverse effects in response to long-term treatments as reflected by normal drinking, eating, grooming behavior, and normal body weight (Figure 1).

The MWM test was used to evaluate cognitive function following treatment. In the hidden platform learning trial, we found GSE treatment significantly improved the cognitive behavioral performance of J20 mice, in comparison to age- gender-matched non-treated, CTRL, J20 mice [F_{(1, 13)} = 3.726, p = 0.038 for GSE treatment effect, Two-Way ANOVA RM, Figure 2A]. Neither Resv nor Juice treatment led to significant improvements in MWM training trial (Figures 2B,C). The Comb mice also performed significantly better than the CTRL mice [F_{(1, 14)} = 6.627, p = 0.0231 for treatment effect, Two-Way ANOVA RM, Figure 2D].

Interestingly, compared to the CTRL mice, all groups of treated J20 mice, except the Juice group, performed significantly better in the probe trial phase of the MWM test 24 h after the last training session (One-Way ANOVA, p = 0.021; GSE vs. CTRL, ^*p < 0.012; Resv vs. CTRL, ^*p < 0.012; Comb vs. CTRL, ^**p < 0.001, Figure 2E and Table 2 for statistical analysis). Juice group performed better than the CTRL group, but did not reach statistical significance (p = 0.032, Figure 2E). Our data suggest that GSE, Resv, Comb, and possibly Juice treatment lead to significant improvement in spatial memory retention. In parallel control studies (Figures 2F,G), we confirmed that all five groups of J20 mice performed equally well, excluding the possibility that the polyphenol treatments might have affected non-spatial parameters, such as sensorimotor performance and motivation, which might have interfered with behavioral performance during the MWM test.

LTP is one of the major cellular mechanisms that underlies synaptic plasticity and is critical for learning and memory (Bliss and Collingridge, 1993; Cooke and Bliss, 2006). Based on our observation that Resv and Juice treatment improved memory retention in the J20 mice, we explored the possibility that accumulation of polyphenol metabolites in the brain, following chronic polyphenol treatment, might directly contribute to the cognitive benefits through the promotion of LTP. We tested three metabolites that have been identified in the brain following different dietary polyphenol treatment in modulating oAβ induced LTP deficits: 3′-O-Me-EC-Gluc from GSE, cyanidin-Glc from Concord grape juice, and resveratrol. It is well established that oAβ are potent synaptotoxins and can impair synaptic plasticity. Hippocampal slices isolated from young mice exhibited normal LTP response following tetanus stimulation (Figure 3A), while 1-h bath perfusion of 200 nM oAβ completely inhibited LTP (Figure 3A). We found that 3′-O-Me-EC-Gluc and cyanidin-Glc, applied at nanomolar concentrations, significantly protects against acute oAβ-mediated LTP deficits (Figures 3B,C). Co-treatment with 3′-O-Me-EC-Gluc resulted in significantly increased LTP, expressed as a percentage of baseline fEPSP slope compared with the vehicle treatment (196.7 ± 6.5% vs. 151.8 ± 9.5 %, p < 0.01, Two-Way ANOVA). Co-treatment with cyanidin-glc resulted in a significantly increased LTP, expressed as a percentage of baseline fEPSP slope compared with the vehicle treatment (216.7 ± 8.5 % vs. 151.8 ± 9.5 %, p < 0.01, Two-Way ANOVA). In contrast, the application of resveratrol did not protect against oAβ-induced synaptic deficits (Figure 3D). This observation revealed that select brain-targeted metabolites from GSE and Concord grape juice extract are capable of directly protecting against AD-type oAβ-mediated synaptic toxicity in the hippocampal formation and may contribute to the improved cognitive function observed in the J20 mice.

Following MWM behavior assessment, we sacrificed the animals and evaluated the impact of the different treatments on Aβ in the brain. We found that both GSE and Comb treatment led to a reduced Aβ ₁₋₄₂ content in the brains, however, only the Comb group reached statistical significance [One-Way ANOVA, p = 0.0098; Comb vs. CTRL, ^**p < 0.001, F_{(1, 14)} = 4.354, Figure 4A and see Table 2 for statistical analysis]. Neither Juice nor Resv treatment had any effect on brain Aβ ₁₋₄₂ levels (Figure 4A). Similar results were obtained for the levels of Aβ ₁₋₄₀ [One-Way ANOVA, p = 0.0067; Comb vs. CTRL, ^*p < 0.007, F_{(1, 14)} = 3.152; Figure 4B and see Table 2 for statistical analysis]. Aβ ₁₋₄₂ peptides are more prone to aggregate and it is believed the increased ratio of Aβ ₁₋₄₂/Aβ ₁₋₄₀ may contribute to the pathogenesis of the disease. We found that polyphenol treatment resulted in a reduced Aβ ₁₋₄₂/Aβ ₁₋₄₀ ratio in the brain (One-Way ANOVA, p = 0.0112, Figure 4C), and both Juice and Resv groups showed a strong trend of reduced ratio of Aβ ₁₋₄₂/Aβ ₁₋₄₀ in the brain compared to the CTRL group (p = 0.06 for both groups). Plaque burden analysis revealed that GSE treatment and Comb treatment reduced the load of plaques in the hippocampal formation compared to the CTRL group, while Resv and Juice had no effect (Figure 4D and see Table 2 for statistical analysis). Plasma amyloid peptide measurements showed that none of the treatments had any effect on the levels of amyloid peptides in the plasma (Figures 4E,F).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.