Obesity development and diabetes risks are intimately linked through alterations in insulin signaling and glucose metabolism [1
]. The metabolic stressors are often underpinned by physiological changes which ultimately lead to the development of insulin resistance, hyperglycemia, and chronic low-grade inflammation in key metabolic as well as gastrointestinal tissues [2
]. Previous studies in animal models as well as in humans indicated that dietary supplementation with anthocyanin-containing foods influenced many cardiovascular and metabolic outcomes [3
]. These findings were supported by epidemiological observations [4
] and randomized controlled trials using berries [5
] or mixtures of partially purified anthocyanins [7
]. However, not all results were consistent between the studies [8
]. While berries with primary malvidins and delphinidins generally improved metabolic and cardiovascular disease risk biomarkers [6
], cyanidins offered little or no protection in many cases [12
]. Previously, we observed that blackcurrant delphinidins were also more likely to confer favorable metabolic adaptations compared to cyanidins in the diet-induced obese (DIO) mouse model [15
]. Taken together, these findings raised the question of the major differences between various anthocyanins with respect to structure and associated bioactivity.
Anthocyanins are present in plants in different concentrations and with various degrees of hydroxylation, methylation, glycosylation, and acylation [16
]. Gastrointestinal metabolism of the primary anthocyanins partially explained a previous perception of their extremely poor bioavailability after oral administration [17
]. The majority of dietary anthocyanins were not absorbed at the upper gastrointestinal level and reached the intestinal microbiome that biotransformed them into phenolic metabolites [18
]. Parent aglycone structures and secondary modifications were critical to their stability, bioavailability, and transformation in the gastrointestinal lumen. The in vivo conversion of hydroxylated anthocyanins to their respective methylated counterparts (i.e., malvidins) via Phase II methylation reactions was necessary for enhanced tissue uptake and bioactivity [19
], while glycosylation and acylation were associated with increased stability and reduced bioactivity [20
This important gap in knowledge on the effect of structural differences of the primary anthocyanins and their biological activities represents a major hindrance to achieving consensus over the efficacy and the most effective strategy of dietary supplementation with anthocyanin-rich foods to achieve favorable metabolic and cardiovascular outcomes. The present study was therefore designed to examine changes in metabolic risk factors after consumption of six berries with structurally diverse anthocyanin profiles. Primary anthocyanins were identified in feces of the DIO mice with intact and disrupted gut microbiome and correlated to several measures of metabolic health, including body composition, hyperglycemia, insulin resistance, microbiome profile, and gastrointestinal luminal oxygen and oxidative stress.
A single serving of anthocyanin-rich berries may contribute in excess of 100–200 mg anthocyanins to a regular diet [33
], which is 5–10-fold higher than the daily intake of other flavonoids [34
]. However, anthocyanins from different plants show various degrees of hydroxylation, methylation, glycosylation, and acylation [16
]. This may explain the equivocal results from animal and human studies, since predominantly delphinidin- and malvidin-containing fruits are more likely to improve metabolic and cardiovascular risk factors (i.e., blueberry [6
], black currants and bilberry [9
], or grapes [11
]), while cyanidins offer less protection (i.e., elderberry [12
], blood orange [13
], or purple carrot [14
To understand the relationship between structural constraints and biological activity of anthocyanins in the native food matrix, we selected six berries with diverse anthocyanin compositions for this study (Table 1
). Blackberry and black raspberry both contained cyanidins as primary anthocyanins, yet with contrasting glycosylation profiles. Blackcurrant and maqui berry predominantly contained delphinidins with diglycosylation profile similar to black raspberry. Blueberry contained monoglycosylated delphinidins, malvidins, and petunidins (opposite of blackberry), while Concord grape was selected for high presence of acylated delphinidins, malvidins, and petunidins (Table 2
). Despite diverse anthocyanin profiles, all berries contained sufficient amounts of total anthocyanins to allow for normalized incorporation of berries into HFD. The final diets contained from approximately 0.5% (black raspberry) to 5% (Concord grape) of the freeze-dried whole berry powders (w
) and delivered 400 µg/g food of total anthocyanins. In this study, animals consumed an average of 2.85 ± 0.34 g food/mouse/day (71.3 mg/kg/day), thus ingesting 1.14 mg/mouse/day (28.5 mg/kg/day) of total anthocyanins. When translated to humans, this treatment was equivalent to consuming 2.4 mg/kg/day or 145 mg/day of total anthocyanins for an average adult [35
] and could be easily achieved by daily consumption of 1–2 servings of fresh berries [33
]. Normalization for anthocyanin content did not allow us to control the treatments for different sugar levels. All berries had similar carbohydrate profiles (10–17 g/100 g fresh weight (FW)), however Concord grapes contained more sugars (16 g/100 g FW) than the rest of the berries (4–10 g/100 g FW).
Dietary supplementation with monoglycosylated cyanidins (blackberry) and diglycosylated cyanidins (black raspberry) had no effect on body weight, food intake, body composition and metabolic risk factors (fasting blood glucose and insulin sensitivity) in the HFD mice (Figure 2
, Figure 3
and Figure 4
). Previously, whole powdered black raspberry supplementation did not prevent the development of obesity or improved lipid status in HFD mice [19
] and it was suggested that complex glycosylation nature of black raspberry anthocyanins precluded their ability to modulate metabolic health [36
]. A similar observation was made for complex acylated cyanidins from purple carrot [14
], diglycosylated cyanidins from elderberry [12
], acylated and diglycosylated cyanidins from blood orange [13
], and monoglycosylated cyanidins from jaboticaba [37
]. Weak antidiabetic effects were reported in animals for cyanidins from purple corn [38
], black soybeans [39
], purple sweet potato [40
], mulberry [41
], cherry [42
], chokeberry [43
] or highly purified cyanidin-3-glucoside alone [44
]. Taken together, our data strongly suggested that cyanidin-based dietary interventions would be much less effective in alleviating metabolic risk factors than previously thought.
Primary diglycosylated delphinidins from black currants (non-acylated) and maqui berry showed similar effects on body weight, body composition, and measures of insulin resistance in the HFD animals. While blackcurrant showed higher efficacy at improving the metabolic outcomes associated with obesity and diabetes, supplementation with maqui berry at physiological levels of dietary intake had limited effect on these parameters, likely due to high proportion of di- and tri-glycosylated anthocyanins in the fruit. Previously, hypoglycemic effects of maqui berry were reported for supraphysiological levels of supplementation with the anthocyanin-enriched extract (100–500 mg/kg) and purified delphinidin 3-sambubioside-5-glucoside (425 mg/kg) [45
]. Beneficial metabolic effects of black currant anthocyanins have been also described [15
]. We observed a similar outcome from consumption of blueberry monoglycosylated delphinidins, malvidins, and petunidins (non-acylated) as compared to the Concord grape (acylated). Berries with acylated anthocyanin profiles showed less biological activity when consumed in physiological concentrations. Direct comparison between monoglycosylated anthocyanins from blueberry and diglycosylated anthocyanins from black currants suggested that supplementation with monoglycosylated molecules is more beneficial to the metabolic health (Figure 2
, Figure 3
and Figure 4
). There are two possible explanations for the observed differences between cyanidin- and delphinidin/malvidin/petunidin-type anthocyanins. It is very likely that anthocyanins with the free hydroxyl groups are less stable in the gastrointestinal environment and do not depend on gastrointestinal bacteria for their metabolism and absorption (delphinidin being an exception as it can be O-
methylated to form petunidin or malvidin in vivo). This property was particular evident in animals with disrupted gut microbiome (Table 3
). At the same time, O-
methylated metabolites from malvidins and petunidins (including O-
methyl delphinidins) were characterized with increased hydrophobicity at the B-ring of the molecule that reduced the plasma residence time and increased tissue affinity [47
], thus being more biologically active. Complex glycosylation and acetylation patterns reduced bioactivity of anthocyanins, but greatly increased their stability in the gastrointestinal tract. When gut microbiome was disrupted with antibiotics, we observed 4–10-fold increase in diglycosylated and acylated anthocyanins excretes in feces. The effect was strongest in animals consuming maqui berry: dietary supplementation with 400 µg/g anthocyanins resulted in 339 µg/g anthocyanins excretion rate in feces upon suppression of gastrointestinal bacteria with the antibiotic cocktail (Table 3
). The data suggested that subjects with gut microbiomes disrupted by a disease or lifestyle modification would have decreased benefits from dietary supplementation with berries as compared to their healthy counterparts, as described for black currants previously [15
The gut microbiota in the LFD mice was dominated by Firmicutes (71%) and Bacteroidetes (15%), and the HDF diet increased this ratio to 89% and 7%, respectively, in strong agreement with previously reported analysis of the mouse gut metagenome [48
]. We observed significant shifts in the gut microbiome profiles towards greater abundance of obligate anaerobes in animals supplemented with different berries, and these effects were most prominent in blackcurrant and blueberry groups (Figure 5
). Since oxygen removal from the gut depends on food composition, microbial fermentation, and native oxidases secreted into the gut lumen [49
], the changes in the oxygen content of the luminal environment may thereby modulate the composition of the gut microbiota. Indeed, consumption of HFD was associated with increased oxygen tension in all gut compartments, and this effect was reduced by incorporating berries and berry anthocyanins into the diet (Figure 6
). Conversely, increase in host oxygenation altered luminal oxygenation in the gut and suppressed oxygen-intolerant bacterial populations [50
]. The presence of marked hypoxia within the lumen of the distal parts of the gastrointestinal tract is therefore consistent with the known abundance of anaerobic bacteria at these sites, and our study indicates that gastrointestinal oxygenation can be modulated by dietary supplementation with berry anthocyanins.
4. Materials and Methods
The cyanidin-3-O-β-glucoside (HPLC grade standard) was purchased from Polyphenols Laboratories AS (Sandnes, Norway). All other chemical reagents and solvents were purchased from Sigma (St. Louis, MO, USA). Commercial whole freeze-dried berry powders were BB, blackberry (Nubeleaf, Portland, OR, USA); BC, blackcurrant (Just the Berries, Palmerston North, New Zealand); BR, black raspberry (BerriHealth, Corvallis, OR, USA); BL, blueberry (Wild Blueberry Association of North America, Old Town, ME, USA); MB, maqui berry (Sunburst Superfoods, Thornwood, NY, USA). Whole Concord grape puree was purchased from AgriAmerica (Silver Creek, NY, USA) and freeze-dried to obtain CG powder. A commercial blackcurrant powdered extract ACE30 (Active Cassis Extract 30) was kindly provided by Eddie Shiojima (Just the Berries, Palmerston North, New Zealand).
4.2. Animals and Diets
All animal experiments were approved by the North Carolina Research Campus Institutional Animal Care and Use Committee (IACUC protocols 12-018 and 16-011) in the David H. Murdock Research Institute, the AAALAC accredited animal care facility.
Male, 6-week-old C57BL/6J mice were purchased from the Jackson Laboratory (Bar Harbor, ME, USA) and housed four animals per cage under controlled temperature (24 ± 2 °C) and light (12 h light–dark cycle, lights on at 7:00 a.m.). Immediately upon arrival, animals were allowed to adapt to new conditions for 7 days and handling the animals was performed daily during this time to reduce the stress of physical manipulation. Mice were then randomized into ad libitum access to Research Diets (New Brunwick, NJ, USA) low 10 kcal % fat diet D12450J (low fat diet, LFD, 3.85 kcal/g, n = 12) or high 60 kcal % fat diet D12492 (high fat diet, HFD, 5.24 kcal/g, n = 56) and tap water for 6 weeks. Obese mice were further randomized to control high fat diet (HFD, n = 8) or berry-supplemented treatment groups normalized to 400 µg/g total anthocyanins (berry powders incorporated into HFD by Research Diets). Mice were kept on the respective diets for an additional 12 weeks. Animal weight and food intake (accounting for spillage) were recorded weekly for the duration of the study. All animal diets were kept at −80 °C for long-term storage and stability, and freshly thawed food was dispensed to animals every 3–4 days to limit phytochemical degradation in food matrix. Body composition analysis was performed on unanesthetized mice using EchoMRI (Echo Medical Systems, Houston, TX, USA) during Week 1 and Week 12 of the study.
4.3. Antibiotic Knockdown of Endogenous Gut Microbiome
An antibiotic cocktail (0.5 g/L vancomycin, 1 g/L neomycin sulfate, 1 g/L metronidazole, 1 g/L ampicillin) previously shown to be sufficient to deplete all detectable commensal bacteria [22
] was administered in drinking water ad libitum to all animals for 1 week (between Week 8 and 9 of HFD treatment).
4.4. Oral Glucose and Insulin Tolerance Tests
For oral glucose tolerance test, mice were fasted overnight (16 h) and received oral gavage of d-glucose (1.5 g/kg body weight). For insulin tolerance test, mice were fasted for 4 h and received intraperitoneal injection of insulin (0.75 U/kg body weight, Santa Cruz Biotechnology, Santa Cruz, CA, USA). Blood glucose concentrations were measured at 0, 15, 30, 60 and 120 min after glucose or insulin challenge in blood samples obtained from tail-tip bleedings, using a glucometer (Lifescan, Johnson and Johnson, New Brunswick, NJ, USA).
4.5. Sample Collection and Oxygen Measurements
At the end of the experiment, mice were euthanized and blood was collected by heart puncture. Oxygen levels in freshly dissected digestive tracts were measured with 0.3 mm Clark-type oxygen microelectrode ET1125, EPU354 isoPod, and Pod-Vu software v5.5.20 (Edaq, Colarado Springs, CO, USA). Electrode was zeroed in water sparged with nitrogen (1 min/mL water) and calibrated in distilled water saturated with air (20.9% oxygen). The accuracy of the microelectrode to reach zero and 20.9% oxygen was rechecked periodically during experiments. Microelectrode was inserted through a small hole cut into the gut wall and all readings were completed in 15 s intervals within 4 min after dissection. Percent oxygen was measured in the duodenum, ileum, cecum, and colon at ambient temperature (23 ± 1 °C) and local barometric pressure (National Weather Service). Oxygen partial pressure (1 mmHg = 133.322 Pa = 1 torr) was calculated from percent oxygen reading recorded at −800 mV polarization.
Liver, fat, muscle, gastrointestinal tissues (stomach, duodenum, ileum, cecum and colon) and luminal digesta were collected and stored at −80 °C to determine the temporal sequence and signaling events that are responsible for changes in physiology and metabolism. Fecal samples from mice were collected, weighed, and pooled by cage at four time points, including Week 4 (1 month on HFD), Week 8 (2 months on HFD, start of antibiotic treatment), Week 9 (end of antibiotic treatment), and Week 12 (3 months on HFD) and stored at −80 °C.
4.6. Gastrointestinal Microbial Profiles
Genomic DNA was extracted from mouse fecal samples using QIAamp Fast DNA Stool Mini kits (Qiagen, Germantown, MD, USA), quantified using Take3 plate and Synergy H1 microplate spectrophotometer (BioTek, Sunnyvale, CA, USA), and adjusted to 1 ng/µL. Quantitative real-time PCR was performed on an ABI 7500 Fast (Life Technologies, Carlsbad, CA, USA) in a total volume of 20 µL containing 10 µL 2× SYBR Green PCR Master Mix, 1 µL of each primer from GUt low-density array (GULDA) Array [23
], 4.4 µL of nuclease-free water and 3.6 µL of template DNA. The amplification program consisted of 50 °C for 2 min; 95 °C for 10 min; 40 cycles of 95 °C for 15 s and 60 °C for 1 min; and a dissociation curve (95 °C for 15 s, 60 °C for 15 s, then increasing to 95 °C at 2% rate). The mean C
t-value was determined based on a set threshold value of 0.2 and using the automatic baseline correction. Differences in Ct
-values for each bacterial target (N0
-normalization) were calculated between those obtained with the universal and target-specific primers and log-transformed. Fold-changes for target amplicons were calculated as the (log 2) ratio of normalized abundances at different time points.
4.7. Anthocyanin Extraction and Quantification
Anthocyanins were extracted from berry powders and feces using 60% aqueous methanol (1% trifluoroacetic acid) following a previously reported method for rapid analysis of various anthocyanins in rats [24
]. Total anthocyanins were determined using the pH differential method [25
] by producing a rapid and reversible structural change (color shift) and quantifying the difference in absorbance at 520 nm using Synergy H1 (BioTek, Sunnyvale, CA, USA).
4.8. HPLC Analysis of Anthocyanins
Individual anthocyanins in berry powders and feces were quantified by HPLC as described in our previous publication [26
]. Briefly, filtered samples were injected (10 μL) into a 1200 HPLC system (Agilent Technologies, Santa Clara, CA, USA) equipped with a UV–vis diode array detector (DAD), controlled-temperature autosampler (4 °C), and column compartment (30 °C) using a reversed-phase Supelcosil LC-18 column, 25 mm × 4.6 mm × 5 μm (Supelco, Bellefonte, PA, USA). Standard curves were calculated using peak areas at UV of 520 nm as cyaniding-3-O
Data were analyzed by one-way ANOVA followed by Dunnett’s multiple-range tests using Prism 6.0 (GraphPad Software, San Diego, CA, USA). Body weight gain, glucose and insulin tolerance were analyzed by two-factor repeated-measures ANOVA, with time and treatment as independent variables. All data were presented as means ± SEM. Significant differences were accepted when the p-value was <0.05.