Effect of cyanidin-3-glucoside and an anthocyanin mixture from bilberry on adenoma development in the ApcMin mouse model of intestinal carcinogenesis—Relationship with tissue anthocyanin levels


  • Darren Cooke,

    1. Cancer Biomarkers and Prevention Group, Department of Cancer Studies and Molecular Medicine, University of Leicester, Leicester, United Kingdom
    Search for more papers by this author
  • Michael Schwarz,

    1. Institute of Food Chemistry, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • David Boocock,

    1. Cancer Biomarkers and Prevention Group, Department of Cancer Studies and Molecular Medicine, University of Leicester, Leicester, United Kingdom
    Search for more papers by this author
  • Peter Winterhalter,

    1. Institute of Food Chemistry, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • William P. Steward,

    1. Cancer Biomarkers and Prevention Group, Department of Cancer Studies and Molecular Medicine, University of Leicester, Leicester, United Kingdom
    Search for more papers by this author
  • Andreas J. Gescher,

    Corresponding author
    1. Cancer Biomarkers and Prevention Group, Department of Cancer Studies and Molecular Medicine, University of Leicester, Leicester, United Kingdom
    • Department of Cancer Studies and Molecular Medicine, University of Leicester, RKCSB, Leicester Royal Infirmary, Leicester LE2 7LX, UK

    Search for more papers by this author
    • Fax: +441162231855.

  • Timothy H. Marczylo

    1. Cancer Biomarkers and Prevention Group, Department of Cancer Studies and Molecular Medicine, University of Leicester, Leicester, United Kingdom
    Search for more papers by this author


Anthocyanins are dietary flavonoids, which can prevent carcinogen-induced colorectal cancer in rats. Here, the hypotheses were tested that Mirtoselect, an anthocyanin mixture from bilberry, or isolated cyanidin-3-glucoside (C3G), the most abundant anthocyanin in diet, interfere with intestinal adenoma formation in the ApcMin mouse, a genetic model of human familial adenomatous polyposis, and that consumption of C3G or Mirtoselect generates measurable levels of anthocyanins in the murine biophase. ApcMin mice ingested C3G or Mirtoselect at 0.03, 0.1 or 0.3% in the diet for 12 weeks, and intestinal adenomas were counted. Plasma, urine and intestinal mucosa were analyzed for presence of anthocyanins by high-pressure liquid chromatography with detection by UV spectrophotometry (520 nm) or tandem mass spectrometry (multiple reaction monitoring). Ingestion of either C3G or Mirtoselect reduced adenoma load dose-dependently. At the highest doses of C3G and Mirtoselect adenoma numbers were decreased by 45% (p < 0.001) or 30% (p < 0.05), respectively, compared to controls. Anthocyanins were found at the analytical detection limit in the plasma and at quantifiable levels in the intestinal mucosa and urine. Anthocyanin glucuronide and methyl metabolites were identified in intestine and urine. Total anthocyanin levels in mice on C3G or Mirtoselect were 43 ng and 8.1 μg/g tissue, respectively, in the intestinal mucosa, and 7.2 and 12.3 μg/ml in the urine. The efficacy of C3G and Mirtoselect in the ApcMin mouse renders the further development of anthocyanins as potential human colorectal cancer chemopreventive agents worthwhile. © 2006 Wiley-Liss, Inc.

There is an increasing interest in the role of dietary constituents in the prevention of cancer. Examples of dietary components, which have been implicated as cancer chemopreventive agents, are flavonoids such as genistein from soya, quercetin from onions or apigenin from leafy vegetables.1 Dietary intake of these flavonoids is estimated to be within the range of 20–25 mg/day.2 Anthocyanins are flavonoids responsible for the bright blue and red colors of many fruits and berries. They are glycosides of anthocyanidins, mainly cyanidin, delphinidin, malvidin, pelargonidin, peonidin and petunidin, conjugated with sugars, including glucose, galactose and arabinose (for structures see Fig. 1). The daily intake of anthocyanins in the US is in excess of 200 mg/day, amounts far superior to those of genistein or quercetin.3 Anthocyanins and anthocyanidins have demonstrated antiproliferative activity in some cancer cell lines with IC50 values for growth inhibition in the range of 100–800 and 18–200 μM, respectively.4, 5 Anthocyanins- and anthocyanin-rich fruit extracts decreased aberrant crypt foci in rats exposed to azoxymethane6 and reduced colonic adenocarcinoma burden in rats that had received azoxymethane, 1,2-dimethylhydrazine or 2-amino–1-methyl-6-phenylimidazo[4,5-b]pyridine.6, 7, 8 The ApcMin (multiple intestinal neoplasia) mouse, which harbors a mutation in the Apc gene, is a model of human familial adenomatous polyposis, an autosomal dominant inherited human disease characterized by the spontaneous development of hundreds of colorectal polyps that can develop to carcinomas.9 Consumption of cherry anthocyanins or isolated cyanidin by ApcMin mice in their drinking water or of dried cherries in their diet decreased adenoma numbers in the cecum, but had no effect on intestinal adenomas.10 Furthermore, a recent abstract suggests that wild berries containing abundant anthocyanins interfered with intestinal adenoma development in this model.11 We wished to reinvestigate the efficacy of dietary anthocyanins in this model, using a commercially available standardized anthocyanin extract (Mirtoselect, Indena SpA, Milan, Italy12) of the berries of Vaccinium myrtillus (bilberries, blueberries) that contains 15 anthocyanins (Fig. 2). The potential effect of Mirtoselect was compared with that of one of its components, cyanidin-3-glucoside (C3G), which constitutes the most abundant anthocyanin occurring in nature.13 A problem associated with the rational evaluation of chemopreventive intervention studies using mixtures, such as Mirtoselect, is the difficulty of assessing levels of individual mixture components, which may contribute to efficacy. We addressed this issue by testing the hypothesis that efficacy of the mixture can be related with levels of individual anthocyanin-derived species in the target organ (intestinal mucosa) and in plasma. Particular care was taken to unravel, as much as possible, the nature of anthocyanin metabolites in these biomatrices. In addition, the hypothesis was tested that anthocyanins or their metabolites can be detected in the urine. Anthocyanins are potent antioxidants.14, 15, 16 Therefore, we also explored whether C3G or Mirtoselect anthocyanins affect target tissue oxidation status in ApcMin mice as reflected by levels in intestinal adenomas of the pyrimidopurinone adduct (M1dG) of deoxyguanosine. M1dG can be generated from malondialdehyde, a mutagenic product of endogenous lipid peroxidation.17, 18 The overall aim of our study was to provide information, which may help to adjudge the suitability of anthocyanins for clinical development as colorectal chemopreventive agents.

Figure 1.

Structures of C3G and other anthocyanins contained in Mirtoselect.

Figure 2.

LC/MS/MS analysis of anthocyanins in Mirtoselect. The following species were identified utilizing MRM transitions (values in m/z) obtained by loss of the sugar moiety: 465 > 303 delphinidin-3-galactoside/delphinidin-3-glucoside (2 species), 449 > 287 cyanidin-3-galactoside/ C3G (2 species), 435 > 303 delphinidin-3-arabinoside, 479 > 317 petunidin-3-galactoside/ petunidin-3-glucoside (2 species), 419 > 287 cyanidin-3-arabinoside, 463 > 301 peonidin-3-galactoside/peonidin-3-glucoside (2 species), 449 > 317 petunidin-3-arabinoside, 493 > 331 malvidin-3-galactoside/ malvidin-3-glucoside (2 species), 433 > 301 petunidin-3-arabinoside and 463 > 331 malvidin-3-arabinoside. Chromatograms are representative of 5 analyses and consistent with the HPLC/UV–vis trace provided by the manufacturer.

Material and methods

Anthocyanin preparations

Cyanidin-3-glucoside (C3G) was purified from blackberries using a counter current chromatographic method as described previously.19 Mirtoselect, an extract of bilberry standardized for anthocyanin content, which is used in health supplements, was obtained from Indena SpA (Milan, Italy, see http://www.indena.it/pdf/mirtoselect.pdf). Mirtoselect comprises 40% anthocyanins consisting of 15 different components, i.e. glucose, galactose and arabinose conjugates of delphinidin, cyanidin, petunidin, peonidin and malvidin. The residual content comprises plant-derived polysaccharides. Identities of anthocyanin components in Mirtoselect (as provided by the manufacturer) were confirmed by LC/MS/MS analysis (for method see later) using specific multiple reaction monitoring (MRM) transitions (Fig. 2). Reagents used in the chemical analysis were purchased from Sigma Chemical (Poole, UK).

Intervention studies in ApcMin mice

Experiments were carried out under animal project license PPL 40/2496, granted to Leicester University by the UK Home Office. The experimental design was vetted by the Leicester University Local Ethical Committee for Animal Experimentation and met the standards required by the UKCCCR guidelines.20 An ApcMin breeding colony was established in the Biomedical Services Unit, University of Leicester, UK, and the ApcMin/+ genotype was confirmed by polymerase chain reaction.9 Four-weeks-old heterozygous mice were randomly assigned to control and intervention groups to receive either standard AIN 93G diet or AIN diet supplemented with Mirtoselect or C3G from weaning until animals were 16 weeks old. Group sizes were n = 12 and n = 16 in the in the first and second studies, respectively, with equal gender distribution. Dietary concentrations of anthocyanin preparations were 0.1% (w/w) in the first and 0.03, 0.1 or 0.3% in the second study. These doses are ∼0.9, 3 or 9 mg per mouse per day or ∼45, 150 or 450 mg per kg body weight per day, which equates to ∼0.26, 0.87 or 2.6 g/80 kg human when extrapolated using the dose/surface area comparison between species.21 Murine body weight was checked once weekly.

Acquisition of biomatrices and assessment of adenomas

After 12 weeks, mice were exsanguinated by cardiac puncture (under halothane-induced terminal anesthesia). Blood was centrifuged (7,000g, 5 min) to obtain plasma. Before killing, samples (20–50 μl) of urine were collected from mice on C3G or Mirtoselect at 0.3%. For mice on C3G, urine samples from 3 individual mice were processed, and for mice on Mirtoselect, urine from 3 mice was pooled. The gastrointestinal tract was flushed with phosphate-buffered saline, excised and dissected longitudinally. Multiplicity and location of adenomas were recorded, and adenomas were differentiated by size (diameter) into <1 mm, 1–3 mm and >3 mm as described before.22 Intestinal mucosa was removed for chemical analysis by gentle scraping (metal spatula). Mucosa from 2–3 mice was pooled for analysis. Biomatrices were flash-frozen (liquid nitrogen) prior to storage at –80°C until analysis.

Measurement of hematocrit

Blood samples were drawn by capillary action into heparinized microhematocrit tubes (75 mm, Richardsons, Leicester, UK). The hematocrit, which constitutes the proportion of the blood volume occupied by the erythrocytes, was determined as described previously.23

Preparation of biomatrix extracts for analysis

Aliquots of plasma (200 μl) or urine (50–100 μl) or mucosa samples (100 mg) homogenized (1:1) in KCl solution (1.15% w/v, blade homogenizer) were centrifuged (16,000g, 5 min) to remove particulates. Resulting samples were diluted to 1 ml (distilled water) prior to loading onto Oasis HLB solid-phase extraction cartridges (1 ml; Waters, Elstree, UK), which had previously been conditioned with 1 ml of acetone:formic acid (9:1) and 1 ml of water:formic acid (9:1). Samples were loaded and eluted at a flow rate of 1 ml/min. Anthocyanins were eluted sequentially with 0.2, then 0.1 ml acetone:formic acid (9:1). Pooled eluates were evaporated to dryness (stream of nitrogen). Residues were reconstituted in water:formic acid (9:1, 80 ml) and centrifuged (16,000g, 10 min, 4°C) prior to analysis by HPLC. Efficiency of extraction from human plasma for C3G was 71%, with intra- and interday variations of 9.0 and 9.1%, respectively. Extraction efficiencies for anthocyanins from plasma spiked with Mirtoselect varied by less than 15%.

HPLC analysis of anthocyanins

Anthocyanin preparations or extracts of biomatrices were analyzed for anthocyanin content by HPLC with detection by either ultraviolet–visible spectrophotometry (UV–vis) or tandem mass spectrometry (MS/MS). The HPLC system (Varian Analytical Instruments, Oxford, UK) comprised a Varian 230 pump, Varian 410 auto-sampler and Varian 325 UV–vis detector. Separation of anthocyanins was achieved on an Xterra Phenyl column (Waters; 4.6 × 150 mm2, 5 μm) with guard column (Waters, 4.6 × 10 mm2, 5 μm) at 40°C. The flow rate was 1.5 ml/min. The gradient elution system comprised 2 solvents, A: 9:1 water: formic acid (pH ≅ 1.6), and B: acetonitrile. The gradient was as follows: 99–97% A (1–3% B) for 5 min, 97–90% A (3–10% B) for 3 min, no change for 4 min, decrease of A to 70% (increase of B to 30%) within 1 min, then maintenance for 4.5 min. Spectrophotometric detection was at 520 nm. Validation of the method for C3G using spiked human plasma (0.2 ml) furnished limits of detection and quantitation of 2.5 and 7.5 ng/ml plasma (5 and 15 nM), respectively. Detection was linear over the range 1–1,000 ng/ml (r2 = 0.99); accuracy was 100% ± 9%. Amounts of anthocyanins other than C3G observed in biomatrices as described under Results should be interpreted as semiquantitative, because they were calculated using standard curves established for C3G on the basis of the assumption that the absorption coefficients of these anthocyanins are similar to that of C3G. HPLC-electrospray ionization tandem mass spectrometric (MS/MS) analysis was performed using an API2000 mass spectrometer (Applied Biosystems, Warrington, UK) with sample delivery via an 1100 series HPLC instrument (Agilent Technologies UK, South Queensferry, UK). Separation was achieved on an Xterra phenyl column (2.1 × 150 mm2, 3.5 μm) with guard (2.1 × 20 mm2, 3.5 μm). The elution was as described ealier, except that solvent system A was held at 90% for an additional 5 min, and that the flow rate was 0.31 ml/min. Mass spectrometric analyses were performed in positive ion mode under the following conditions: declustering potential 55 V, focusing potential 380 V, entrance potential 10 V, collision energy 50 V, collision energy exit potential 16 V, ion-spray voltage 5,000 V and temperature 450°C. Anthocyanins (parents and metabolites) were identified in positive ion mode by MRM of fragments generated by the loss of the sugar moiety, glucuronides by monitoring of fragments that had lost glucuronic acid. The presence of anthocyanidins was tentatively indicated by ion-selective monitoring and by MRM.

Measurement of M1dG levels in mouse adenomas

Adenomas sized ∼1 mm in diameter or larger from control mice or mice on C3G or Mirtoselect (group size 6–9) were excised. Pooled tissue from each mouse was homogenized (manual glass homogenizer) and digested with Proteinase K (Qiagen, Crawley, UK) and RNAse A (100 mg/ml; Sigma). DNA was extracted as described by the manufacturer's instructions (Quiagen handbook for “genomic DNA extraction kit”). DNA content was determined by UV absorbance. M1dG was quantified by immunoslot blot, using a primary anti-M1G antibody provided by Dr. L Marnett (Vanderbilt University, TN) as reported by Leuratti et al.24 The level of M1dG in adenomas of control mice was 5.2 ± 3.7 fmol/μg DNA (mean ± SD, n = 9).

Statistical evaluation

Evaluation of significance of values as compared to the appropriate controls was performed by one-way ANOVA with subsequent Tukey's pair-wise comparison.


Effect of C3G and Mirtoselect on adenoma development in ApcMin mice

In an orientation experiment, mice (12 per group) received C3G or Mirtoselect (0.1%) in their diet. At the end of the intervention total numbers of adenomas in mice on control diet, or diet with C3G or Mirtoselect were 62 ± 16, 51 ± 17 (82% of control, p > 0.05 versus control) and 42 ± 10 (68% of control, p < 0.05 versus control), respectively. In a repeat study, mice (16 per group) received C3G or Mirtoselect at 0.03, 0.1 or 0.3% in their diet. Both interventions reduced adenoma numbers in a dose-dependent fashion (Fig. 3). Least mean square analyses for linearity of the plots of mean adenoma number versus dose yielded r2 values of 0.877 for Mirtoselect and 0.986 for C3G, thus corroborating dose-dependency of the adenoma-retarding activity. At the highest dose (0.3%), the decrease in number of adenomas was 45 and 30% for C3G and Mirtoselect, respectively. In either study the body weight of mice, which received anthocyanins, was not significantly different from that of mice on control diet, suggesting that any effect on adenoma development was not the consequence of reduced food intake.

Figure 3.

Effect of Mirtoselect (a) or C3G (b), at concentrations of 0.03, 0.1 or 0.3% in the diet, on total numbers of intestinal adenomas in ApcMin mice. Control animals (control) received standard AIN diet. Values are mean ± SD (n = 16 per group); stars indicate that values are significantly different from controls (*p < 0.05, **p < 0.001), exact p values are indicated above some bars. For details of animal experimentation see Material and methods.

ApcMin mouse adenomas are primarily located in the medial and distal sections of the small intestine, with only very few, mostly large, lesions located in the colon. We compared the effect of the anthocyanins on number and size of adenomas in different intestinal sections. Figure 4 shows that both interventions tended to particularly reduce numbers of small-sized adenomas (<1 mm), and that they interfered with adenoma development in a broadly consistent fashion in all three sections of the small intestine.

Figure 4.

Effect of Mirtoselect (a) or C3G (b) at dietary concentrations of 0.01% (closed bars), 0.1% (open bar) or 0.3% (striped bar) on number and size of adenomas in the proximal (prox), medial (med), distal (dist) or colonic sections of the intestine of ApcMin mice. Results are expressed as the mean adenoma number in the appropriate intervention cohort minus mean adenoma number in the respective control cohort. Error bars, which have been omitted for the sake of clarity of presentation, were between 23% (adenomas <1 mm, distal section, mice on 0.1% C3G or Mirtoselect) and 400% (adenomas >3 mm, distal section, mice on 0.03% C3G or Mirtoselect) of mean values. Adenoma diameter is shown below bars. Adenoma numbers in control animals (mean ± SD) were as follows: proximal intestine <1 mm: 8.5 ± 3.1, 1–3 mm: 3.4 ± 2.6, >3 mm: 0.1 ± 0.3; medial intestine <1 mm: 12.2 ± 3.5, 1–3 mm: 2.7 ± 2.0, >3 mm: 0.1 ± 0.3; distal intestine <1 mm: 15.3 ± 4.0, 1–3 mm: 6.7 ± 4.0, >3 mm: 0.1 ± 0.5; colon <1 mm: 2.0 ± 1.3, 1–3 mm: 0.5 ± 0.9, >3 mm: 0.5 ± 0.7. Number of mice per intervention group was 16. Stars indicate that original adenoma numbers are significantly different from those in controls [*p < 0.04, **p < 0.005, ***p < 0.001, (*) indicates value approaches significance with p < 0.07]. For details of animal experimentation and analysis see Material and methods.

A major complication associated with adenoma load in ApcMin mice is gastrointestinal bleeding, which is thought to contribute substantially to morbidity and mortality.9 We measured the hematocrit in treated animals. Neither anthocyanin intervention affected packed red cell volume significantly, even though in mice on C3G or Mirtoselect at the highest dose the hematocrit value was marginally increased from 22% ± 7% in control ApcMin mice to 26% ± 7% and 25% ± 8%, respectively (results not shown).

Anthocyanin levels in biomatrices of ApcMin mice

Samples of plasma, intestinal mucosa and urine of mice, which received either C3G or Mirtoselect at 0.3% in the diet, were analyzed by HPLC with UV-vis spectroscopic or mass spectrometric detection. Analysis of plasma showed the presence of unchanged anthocyanins (Fig. 5a). Amounts of C3G in the plasma on 4 (of 16) mice on C3G were between the limits of detection and quantitation (2.5–7.5 ng/ml, 5–15 nM), 1 mouse contained 12.4 ng/ml (28 nM) C3G, while C3G levels in the residual 11 mice were below the detection limit. Four molecules were found in the plasma of mice on Mirtoselect (Fig. 5a). They were tentatively identified as malvidin glucoside and galactosides of delphinidin, cyanidin and malvidin on the basis of their retention times, when compared with those of the constituents of authentic Mirtoselect (Fig. 2). The highest total anthocyanin plasma content observed in 1 mouse, calculated on the basis of the calibration curve established for C3G, was ∼46 ng/ml. Analyte levels in the plasma were too low to lend themselves to analysis by LC/MS/MS.

Figure 5.

HPLC/UV–vis chromatograms of extracts of plasma (a), intestinal mucosa (b) and urine (c) of ApcMin mice, which received either normal AIN diet (control), Mirtoselect (0.3%) or C3G (0.3%) with their diet. Each chromatogram is representative of 3–6 separate animals. Retention time of authentic C3G was validated by cochromatography using spiked biomatrix (not shown). Bracket indicates the position of malvidin galactoside (retention time ∼9 min) and malvidin glucoside (retention time ∼10 min), assigned on the basis of order of retention times of authentic Mirtoselect constituents (provided by the manufacturer) and LC/MS/MS analysis (Fig. 2). Asterix denotes positon of peonidin-3-glucoside/methyl C3G. AU is the absorbance units (at 520 nm). For details of sample preparation and HPLC analysis see Material and methods.

Table I summarizes levels of total anthocyanins recovered from intestinal mucosa and urine of mice that received C3G or Mirtoselect. Extracts of 5 mucosa samples per intervention, each one pooled from 2–3 mice, and extracts of individual urine samples from 3 mice on C3G or from pooled urine from 3 mice on Mirtoselect were analyzed. C3G was unambiguously identified in the intestinal mucosa (Figs. 5b and 6b) and urine (Figs. 5c and 6c) of mice receiving C3G. Mucosal levels of C3G and total anthocyanins were 16 ± 22 ng (37 ± 51 pmol) per gram and 43 ng/g tissue, respectively, and the analogous urine levels were 3.9 ± 4.2 μg/ml (8.7 ± 9.4 μM) and 7.2 μg/ml (Table I). Metabolites of C3G in the mucosa of mice on C3G were identified by LC/MS/MS as methyl C3G (transition m/z 463 > 301, 2 or more peaks), C3G glucuronide (m/z 625 > 287, one peak) and cyanidin glucuronide (m/z 463 > 287, 3 peaks, Fig. 6b), consistent with retention times of peaks observed by HPLC/UV–vis analysis (Fig. 5b). The mass spectrometric search in the mucosa for cyanidin, the aglycon generated by C3G hydrolysis, using the MRM transition m/z 287 > 137 or selected ion monitoring (m/z 287), yielded 2 peaks (result not shown), neither of which could be unequivocally identified by coelution with authentic compound. In the urine metabolites of C3G were observed by HPLC/UV–vis (Fig. 5b), and MS analysis afforded the following C3G metabolites (Fig. 6b): methyl C3G (m/z 463 > 301, 2 peaks), C3G glucuronide (m/z 625 > 287, 2 peaks), methyl C3G glucuronide (m/z 639 > 301, 3 or 4 peaks), cyanidin glucuronide (m/z 463 > 287, 3 peaks) and methyl cyanidin glucuronide (m/z 477 > 301, 2 or 3 peaks). One of the 2 peaks consistent with methyl C3G was apparently the most abundant urinary metabolite of C3G (Fig. 5c). The retention time of this species, when compared with those of the anthocyanin components of Mirtoselect (Fig. 2), characterizes it tentatively as peonidin-3-glucoside.

Figure 6.

LC/MS/MS chromatograms of authentic C3G (a) or extracts of intestinal mucosa (b) and urine (c) of ApcMin mice, which had received C3G (0.3%) in their diet for their life time. The mass spectrometric detection mode utilized MRM transitions indicative of the following species (values in m/z): 449 > 287, parent C3G; 625 > 287, C3G glucuronide; 463 > 287, cyanidin glucuronide; 463 > 301, methyl C3G; 639 > 301, methyl C3G glucuronide and 477 > 301, methyl cyanidin glucuronide. Note that some species occur as two or more positional isomers. Chromatograms in B and C are representative of 3–6 mice. CPS, cycles per second. For details of LC/MS/MS analysis see Material and methods.

Table I. Levels of Anthocyanins in Intestinal Mucosa and Urine of ApcMin Mice, which Received C3G or Mirtoselect (0.3%) with Their Diet
InterventionAnthocyanin levels
Intestinal mucosa (μg/g tissue)1Urine (μg/ml)
  • 1

    Mean ± SD of 5 samples from 2 to 3 mice.

  • 2

    Mean ± SD of 3 mice.

  • 3

    Pooled urine from 3 mice.

C3G0.043 ± 0.0437.2 ± 7.72
Mirtoselect8.1 ± 7.512.33

Table II shows the retention times and LC/MS/MS characteristics of anthocyanin species recovered from intestinal mucosa and urine from mice on Mirtoselect. Malvidin glycosides were most prominent (Figs. 5b and 5c). The concentrations of malvidin-3-glucoside in the mucosa and urine were 1.8 ± 1.6 μg (3.7 ± 3.3 nmol) per gram and 1.6 μg/μl (3.2 μM), respectively. Mean levels of total anthocyanins were 8.1 μg/g mucosa tissue and 12.3 μg/ml urine (Table I). Glucosides of delphinidin and cyanidin were present in the urine of mice on Mirtoselect at concentrations near the limit of detection, considerably below those of their respective galactosides, which occurred at 0.6 ± 0.7 and 0.6 ± 0.6 μg (1.4 ± 1.5 and 1.2 ± 1.2 nmol) per gram, respectively. Mass spectrometric analyses of intestinal mucosa samples (MRM and selected ion monitoring of parent ion) afforded peaks, which hinted at the presence of the aglycons delphinidin (m/z 303 > 229), petunidin (m/z 317 > 217) and peonidin (m/z 301 > 201) (results not shown), albeit these species were not unequivocally identified. LC/MS/MS analysis of urine samples from mice on Mirtoselect (Table II) yielded the components of the Mirtoselect formulation and a range of metabolic conjugates: glucuronides of delphinidin (m/z 479 > 303), cyanidin (m/z 463 > 287), peonidin (m/z 477 > 301), malvidin (m/z 507 > 331), C3G (m/z 625 > 287), peonidin-3-glucoside (m/z 639 > 301), peonidin-3-arabinoside (m/z 609 > 301), malvidin glycoside (m/z 669 > 331) and malvidin arabinoside (m/z 639 > 331), and methyl conjugates of delphinidin arabinoside (m/z 479 > 317) and of 2 methyl delphinidin galactosides and/or glucosides (m/z 449 > 317).

Table II. Anthocyanins Identified by LC/MS/MS in Urine and Intestinal Mucosa of ApcMin Mice, which Received Mirtoselect (0.3%) with Their Diet
Anthocyanin speciesRetention time (min)MRM transition (m/z)
  • 1

    Several positional isomers.

  • 2, 2,

    Not detected in mucosa.

  • 3

    Not detected in urine.

  • 4

    In the absence of authentic metabolite standards, species may represent either glucoside or galactoside.

Delphinidin-3-galactoside3.3465 > 303
Delphinidin-3-glucoside3.8465 > 303
Delphinidin-3-arabinoside4.7435 > 303
Cyanidin-3-galactoside4.5449 > 287
Cyanidin-3-glucoside4.5449 > 287
Cyanidin-3-arabinoside6.7419 > 287
Petunidin-3-galactoside6.4479 > 317
Petunidin-3-glucoside7.8479 > 317
Petunidin-3-arabinoside9.7449 > 317
Peonidin-3-galactoside9.1463 > 301
Peonidin-3-glucoside10.7463 > 301
Peonidin-3-arabinoside11.5433 > 301
Malvidin-3-galactoside11.2493 > 331
Malvidin-3-glucoside11.9493 > 331
Malvidin-3-arabinoside12.6463 > 331
Delphinidin glucuronide11.3479 > 303
Cyanidin glucuronide13.1463 > 287
Peonidin glucuronide10.5, 8.1, 11.71477 > 301
Malvidin glucuronide11.8, 9.9, 12.61507 > 331
Cyanidin glycoside glucuronide22.6, 3.01625 > 287
Peonidin glycoside glucuronide3.3, 5.6, 6.41639 > 301
Peonidin arabinoside glucuronide35.3609 > 301
Methyl delphinidin arabinoside28.7449 > 317
Methyl delphinidin glycoside2,45.9479 > 317

Anthocyanins were also recovered from tissues other than the intestine. Extracts of kidney and liver from mice on C3G furnished cyanidin glucuronide and methylated derivatives of C3G and C3G glucuronide. Liver and kidney samples from mice on Mirtoselect contained unchanged anthocyanins and those glucuronide metabolites characterized in intestinal mucosa.

Effect of C3G and Mirtoselect on M1dG levels in mouse adenomas

DNA extracted from large adenomas was analyzed for levels of M1dG, which were very variable. Levels of M1dG in adenomas of mice on Mirtoselect or C3G were 78% ± 69% and 58% ± 35%, respectively, of those in control mice. Statistical comparison of the original data showed that the difference between the respective intervention and control groups did not reach significance.


The results described earlier show, for the first time, that dietary consumption, by ApcMin mice, of anthocyanins in the form of either a mixture or an isolated glycoside interferes with small intestinal adenoma development in a dose-dependent fashion. The results also suggest that efficacy was accompanied by measurable levels of anthocyanins in the target organ and in the urine, and by concentrations near or below the detection limit in the systemic circulation. The dietary dose, at which either agent was significantly efficacious, is 0.3%, ∼9 mg/mouse or ∼450 mg/kg, which equates to ∼2.6 g/80 kg human, when extrapolated by dose/ surface area comparison.21 On the basis of this calculation, the equivalent dose of Mirtoselect in humans (2.6 g Mirtoselect containing ∼1.0 g anthocyanins) is contained in ∼740 g bilberries,25 which suggests that an efficacious cancer chemopreventive dose of bilberry extract in the ApcMin mouse may translate into a hefty, but not impossibly high, dose in humans. In terms of absolute dose of agent, C3G was less efficacious than the Mirtoselect mixture. It needs to be stressed that Mirtoselect contains only 40% anthocyanins so that the absolute intake of anthocyanins with Mirtoselect was considerably lower than the anthocyanin dose associated with C3G. The efficacious dose in ApcMin mice is similar to the dietary doses of anthocyanin-rich extracts, which have previously been shown to delay colorectal carcinogenesis in carcinogen-induced rat models.6, 7, 8 Sources of extracts explored in these models were chokeberry and bilberry at an extract concentration of 0.4% in the diet6 and purple corn, sweet potato and red cabbage at 5%.7, 8 Results presented here in ApcMin mice contrast with a previous study in this model, in which consumption of tart cherries (20% w/w in diet), anthocyanins extracted from tart cherries (0.08% in drinking water) or isolated cyanidin (0.02% in drinking water) failed to affect intestinal adenoma number, even though they decreased adenomas in the cecum.10 In our study, cecal adenomas were only observed very sporadically, confounding assessment of effect of anthocyanins on cecal adenoma number. The discrepancy between the ApcMin results presented here and previously10 may be related to the somewhat higher dose of anthocyanins employed in our study as compared to that of Kang et al.10 Alternatively, the discrepancy may be linked to the fact that in the previous study,10 anthocyanins were administered in the drinking water and in our study with the diet. Anthocyanins tend to be unstable in aqueous solution at neutral pH,14 and Kang et al. stabilized anthocyanins by addition of ascorbic acid to the drinking water.10 It is possible that the presence of the antioxidant ascorbic acid in the drinking water decreased adenoma incidence in control mice and thus obscured any effect of the anthocyanins. Another reason for the discrepancy may be the difference in composition of anthocyanins between tart cherries and Mirtoselect. Tart cherry anthocyanins are primarily cyanidin glucosylrutinoside and cyanidin-2-rutinoside, whereas Mirtoselect contains a mixture of 15 anthocyanins predominantly glucosides, galactosides and arabinosides of delphinidin and cyanidin.

Measurement of anthocyanin species in ApcMin mouse biomatrices allows a tentative interpretation of the observed activity. Anthocyanins have modest antiproliferative properties4: for example, anthocyanins extracted from chokeberries and bilberries interfered with the growth of HT29 and NCM460 colon carcinoma cells in vitro with an IC50 value of ∼25 μg extract per milliliter cellular incubation medium.5 In comparison, the concentration of total anthocyanins achieved in the intestinal mucosa of mice receiving Mirtoselect was approximately a third of this value (8.1 μg/g tissue). This comparison suggests that consumption of Mirtoselect at 0.3% in the diet can furnish anthocyanins in the murine gastrointestinal tract at levels that approach neoplastic cell growth-inhibitory concentrations. Malvidin glycosides were the most abundant Mirtoselect constituents recovered from the intestinal mucosa. Likely explanations for this phenomenon invoke differences in rates of absorption or of conjugative metabolism between individual Mirtoselect anthocyanidins, even though one cannot exclude the possibility that these malvidin species were, at least in part, products of metabolic O-methylation of other anthocyanins contained in Mirtoselect. Malvidin is chemically characterized by 2 aromatic methoxy functions, while the analogous positions in delphinidin are occupied by phenolic hydroxyl moieties, and petunidin bears one hydroxyl and one methoxy group (Fig. 1). Malvidin glycosides are probably less prone than their hydroxylated counterparts to undergo metabolic conjugation reactions, thus more likely to survive unchanged in the intestinal milieu. The different anthocyanins contained in Mirtoselect were also characterized by differential propensities to survive in the biophase depending on the glycosidic sugar moiety. This notion is borne out by the observation that, compared with the content of components in the original Mirtoselect formulation, much lower amounts of C3G than of cyanidin-3-galactoside were recovered from the mucosa and urine. This phenomenon may be the consequence of differences between the glycosides in terms of susceptibility towards absorption, metabolism or clearance.

Anthocyanidins have been shown to be more potent inhibitors of cell growth or regulators of cell signaling than their respective glycosides.4 Therefore, the efficacies of anthocyanins in vivo are hypothesized to be mediated, at least in part, by hydrolysis to anthocyanidins in the organism.26, 27, 28, 29 Consistent with this notion, cyanidin was identified in the jejunal mucosa and cyanidin and peonidin in the plasma of rats, which had received a blackberry extract.29 The analyses described here hint tentatively at the presence in the intestinal mucosa of cyanidin in the case of C3G and delphinidin, petunidin and peonidin in mice on Mirtoselect. Furthermore, anthocyanidin glucuronide metabolites, the formation of which is likely to invoke the intermediacy of the aglycon, were identified in the intestinal mucosa and urine of mice on either anthocyanin preparation. Thus, it is conceivable that the adenoma-retarding efficacy of C3G or Mirtoselect observed here was, at least partially, mediated by anthocyanidins transiently present in the biophase after hydrolysis of their precursor glycosides and before––probably very swift––removal by metabolic conjugation.

The analysis of M1dG levels suggests tentatively that dietary anthocyanins, especially C3G, reduced oxidative DNA adduct levels in intestinal adenomas. However, the difference between mice on control diet and those on anthocyanins was not significant. Therefore, the variability in M1dG values between individual mice does not support consideration of this parameter as a pharmacodynamic marker of anthocyanin efficacy.

In summary, the activity of anthocyanins in the ApcMin mouse model demonstrated here together with previous reports of their activity in carcinogen-induced rodent models of colorectal carcinogenesis render anthocyanin-containing preparations appealing as potential colorectal cancer chemopreventive agents in humans. Anthocyanin levels associated with an efficacious dose in the ApcMin mouse were low in the systemic circulation but measurable in the gastrointestinal tract and urine, hinting at the potential use of urine to monitor adherence. In the light of some concern with regard to the unwanted side effects of nonsteroidal antiinflammatory drugs and selective cyclooxygenase-2 inhibitors when used as colorectal cancer chemopreventive agents in humans, the lack of known adverse effects of anthocyanins renders them potentially attractive alternative candidates for clinical development.


We thank Dr. P. Morazzoni (Indena SpA, Milan, Italy) for provision of Mirtoselect, Prof. L. Marnett (Vanderbilt University, Nashville) for provision of the M1dG antibody, Dr. R. Singh (Department of Biochemistry, University of Leicester) for help with the M1dG analysis and the staff of the Biomedical Services Unit, University of Leicester, for their help with animal husbandry. This work was supported by the UK Medical Research Council (A.J.G.).