Anthocyanins in Colorectal Cancer Prevention. A Systematic Review of the Literature in Search of Molecular Oncotargets

Background: Colorectal cancer (CRC) is the malignant process that surges in the terminal part of gastrointestinal tract when adenomatous polyps convert to neoplastic cells able to infiltrate the submucosa. Despite the constant progress in applying preventive measures (screening, colonoscopy) and developing new cures (surgical and chemotherapy), CRC is still one of the leading causes of cancer death worldwide. The importance of natural dietary components in CRC prevention has been recognized. Defining the precise role of the diet and its particular molecular moieties in CRC prevention is of constant scientific interest years behind. Anthocyanins (AC), phenolic phytochemicals present in pigmented plants and vegetables, have been reported to have some role in counteracting CRC carcinogenesis. Nonetheless, evidence coming out the pre-clinical, clinical, and epidemiological studies is still controversial. This review is addressing the need to better comprehend the causes of missing data and discrepancies in investigations on the role of dietary AC in modulating CRC carcinogenesis. Methods: We have analyzed the scientific literature, available in PubMed database, according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement methodology for systematic reviews. Subsequently, two selection strategies, with their screening and eligibility criteria, were applied to retain research articles reporting in vitro and in vivo studies aimed at exploring the molecular mechanisms underlying the observed effects of AC in CRC prevention. Results: From the pool of 82 identified publications, we selected 19 articles reporting experimental or observational data on the effect of AC-enriched diets in CRC prevention in humans or murine species. Furthermore, we selected 10 articles reporting about molecular mechanisms of action of pure AC in CRC experimental models. Conclusions: The major outcome of this review is that AC showed essentially no effect in human studies, whereas AC-enriched diets proved to be effective in experimental murine models of CRC. In cell culture tests, AC showed to interfere with cell signaling pathways related to cell growth and differentiation, apoptosis, oxygen stress, and inflammation response. Further molecular characterizations are required to include AC in the panel of disease-modifying agents.


INTRODUCTION
Colorectal cancer (CRC) is the third most common cancer and the fourth most common cause of cancer death worldwide. Even though there is a remarkable advancement in the field of early detection throughout programmed screenings as well as therapies like surgery and chemotherapy, there were still over 1.8 million new cases in 2018 (Bray et al., 2018).
Cancer arises in the terminal part of the intestine, triggered by mutations in oncogenes or tumor suppressor genes expressed in crypt cells. Cumulative mutations lead to uncontrolled division of mucosal cells and formation of polyps. Some of these transform to precancerous adenomas and subsequently to neoplasia. Early detectable change in mucosa is characterized by the presence of the aberrant crypt foci (ACF) (Takayama et al., 1998;Janakiram and Rao, 2008).
Risk factors are related to age, sex, or family history. What is interesting, CRC has a rising incidence connected to the changeable risk factors, i.e., those related to lifestyles and environmental exposure, like physical inactivity, diets rich in processed and red meats, smoking, and alcohol consuming. CRC is also correlated with economic and consequent nutritional transition, and it is increasing in countries that are undergoing rapid social and economic changes (Bray et al., 2012).
The scientific community has perceived these evidences and focused its attention on studies on the relation of food consumption with CRC prevention. A large portfolio of evidence associating high intake of vegetables, fruits, whole grains, and foods containing dietary fiber, dairy products, and calcium supplements with the declined risk of CRC is available (Ryan-Harshman and Aldoori, 2007). It has been hypothesized that specific phytochemicals might be the chemical agents of CRC prevention, raising the hope that these molecules may act synergistically with conventional chemotherapeutics .
Anthocyanins (AC) are natural compounds that have attracted attention in the recent years for their capacity to interact with cancer cells and modify their cellular response. Due to their phenolic structure and the presence of hydroxyl groups, AC are also recognized as a potent antioxidants. By quenching reactive oxygen species, AC prevent mutagenesis in normal cells, but may be toxic for cancer cells (Wang and Stoner, 2008;Cvorovic et al., 2010).

Objectives of the Study
We were interested in performing a systematic analysis of the literature to find information from preclinical, clinical, and epidemiological studies on the mechanism by which AC prevent the onset and progression of CRC, and the underlying molecular events. Thus, we read the scientific literature to seek information and address a series of specific questions, such as the following: 1) Are there any studies performed on human groups or populations? The aim was to provide an overview of the available knowledge about the ability of AC-containing food to prevent human colorectal cancer. 2) Are there any studies performed on animals? What animal models have been used? The aim was to survey investigations on the effects of AC-rich foods on experimental CRC in animals.
3) Are there any studies performed in cellular models? The aim was to survey in vitro studies performed in cellular models of CRC, focusing on the effect(s) of pure AC on the molecular process of carcinogenesis. 4) What molecules have been used to test anti-cancer effects?
The aim was to check what AC molecules were tested. 5) Which biological processes have been investigated? The aim was to survey the mechanisms whereby AC interfere with usual carcinogenic pathways. 6) Have specific molecular targets or markers been identified?
The aim was to survey AC molecules that may potentially serve as anticancer drugs.
It is easily recognizable that this series of questions recapitulates the approach of reverse translational medicine to drug discovery, which is also said "bedside-to-bench" (Gibbs et al., 2018). By this approach, observations about patients and their response to treatments are the starting point to design and simultaneously implement more detailed pharmacokinetic and pharmacodynamic tests, as well as more targeted, hypothesis-driven in vitro molecular studies. This approach is believed to accelerate drug development, since it focuses on a limited number of drug candidates. Furthermore, all phases of drug discovery can be run at the same time.
To search for answers to these questions, we have applied the PRISMA Statement methodology (Shamseer et al., 2015) to retrieve, select, and analyze the scientific bibliography about the role of AC in CRC by a standardized method, which can be easily replicated and will enable continuous updating of the research outcomes.

Information Sources
We have searched PubMed, regarded as the most important and complete archive of biomedical literature.

Search Strategy
We have searched for documents in PubMed database by using the query "colorectal cancer AND anthocyanins." Table 1 summarizes the search strategy details. Filters limits were not applied. The time window was up to 31.12.2018.

Selection Strategy
The articles identified during the search step were first screened to retain only research articles complying with criteria specified in Table 2.
The retained documents were analyzed by applying two parallel selection strategies. Strategy A ( Table 3) was applied to identify studies showing if AC-rich food reduces the risk of CRC. Strategy B (Table 4) was applied to identify studies highlighting the molecular mechanisms by which pure AC compounds, either alone or in pure mixtures, interact with hypothetical molecular targets involved in colorectal carcinogenesis. In respect to the experimental use of pure AC mixtures, it was decided that the purity threshold for inclusion was 90% (w/w).

3
June 2019 | Volume 10 | Article 675 Frontiers in Pharmacology | www.frontiersin.org Eligible articles were retained for full-text reading, analysis, and classification of their study features are expected to provide answers on the abovementioned research questions. Table 5 presents a synopsis of the rationale applied in this systematic review. Indeed, for each main question, a specific selection strategy (i.e., either A or B) was applied to select articles from the pool. Figure 1 is the flowchart of this systematic review. It shows that 82 articles were identified in PubMed. After pre-screening, 67 articles were screened for inclusion. The output was 19 and 10 studies, resulting from strategies A and B, respectively (Supplementary Material).
Studies were either observational (n = 3) or interventional (n = 16). The latter were done essentially only in murine species (n = 9 in mice; n = 6 in rats), whereas only one was in humans. Observational studies recorded AC intake by interviewing subjects. In interventional studies, the orally administered doses of AC ranged from 0.5 to 3 g in humans and 0.03% to 30% dietary supplement in animals. Risk reduction was observed in only one human observational study, though there may be several confounding factors. Intervention studies in animals (n = 15) showed, by contrast, effects related to the number of colorectal tumors, their size, total burden, or volume.

Animals With CRC Induced by Carcinogenic Chemicals
Preclinical in vivo experimental data have been obtained from interventions on rodent models of CRC.
Chemically induced carcinogenesis by various combinations of compounds, such as azoxymethane (AOM)/1.2-dimethylhydrazine (DMH), azoxymethane/dextran sodium sulphate (DSS), or 1,2 dimethylhydrazine/2-amino-1-methyl-6-phenylimidazo[4,5-b] pyridine (PhIP), develops mostly in colon and resembles human pathology. Animals usually present histological modification of intestinal tract, like ACF (when used only one carcinogen) or progressive grades of dysplasia and intraepithelial neoplasia (combination of two carcinogens). Disadvantage of these models is a very high exposure to chemicals, which is uncommon for humans. However, they are considered useful tools to test chemopreventive molecules, like AC in colon carcinogenesis (Femia and Caderni, 2008).

Studies With Mice
Diet with AC-containing purple-fleshed potatoes suppressed the incidence of tumors in the mouse model by elimination of colon cancer stem cells (in vitro observation). Furthermore, these animals presented milder signs of gastrointestinal toxicity (e.g., stomach/intestinal ulcers), suggesting that AC treatment may be more effective than the anti-inflammatory drug sulindac (Charepalli et al., 2015).
Anthocyanin-rich extract (ARE) from bilberries, administered as a 10% supplement to regular diet, almost abolished macroscopically visible tumors in AOM/DSS AOM-induced CF-1 mice model. Even when their diet was supplemented with only 1% ARE, tumors were smaller and fewer than that in controls (Lippert et al., 2017).
A supplementary diet with AC-enriched purple-fleshed potato (P40) extract, administered to AOM-induced CF-1 mice model, reduced the total number of large and medium, but not the small, ACF (Lim et al., 2013).
A very detailed study conducted by Shi et al. showed an important contribution of lyophilized strawberries in amelioration of numerous CRC-related parameters. AOM/DSS-induced CD-1 mice developed tumors, whose incidence and multiplicity was decreased significantly when animals were fed strawberries. Besides colorectal tumors, mice experienced an inflammatory response, as shown by increased expression of inflammatory mediators in the colon epithelium. AC-enriched strawberry powder, added to normal diet, interfered and counteracted the pathway of inflammatory mediators. Phosphorylation of proinflammatory signaling molecules (PI3K, Akt, ERK1/2, NFκB) was significantly reduced. Consequently, mRNA expression of the inflammatory mediators TNF-α, IL-6, COX-2, and iNOS was down-regulated. As it has been suggested by many studies, end products of arachidonic acid metabolism deriving from the cyclooxygenase pathway (i.e., prostaglandins, PG) can worsen colon cancer prognosis. Indeed, dietary strawberries, (2.5%, 5%, 10% by weight) affected tumor-induced increase of COX-2 and iNOS activity, detected as dose-dependent reduction of PGE2 (30%, 65%, and 70%, respectively), and reduced total nitrate and nitrite production (Shi et al., 2015).
Black raspberry ACs were able to up-regulate miRNA-24-1-5p expression in colon tissue from mice with AOM/DSS-induced CRC. Protective effect of this miRNA was manifested through the inhibition of the cell proliferation-related wnt/β-catenin pathway and consequent attenuation of cancer cell growth, migration, and survival. In vitro studies on HCT116 and CaCo-2 cell lines confirmed these observations (Zhang et al., 2018).

Studies With Rats
The dietary supplement Purple corn color (PCC, containing 21.5% cyanidin 3-glucoside) significantly decreased the number of colonic nodules in F344 rats treated with DMH and PhIP. Hyperplastic and neoplastic lesions were observed also in other organs (jejunum, pancreas, liver, prostate, and seminal vesicle), but their incidences were not affected by PCC administration, showing that the supplement was bioactive only in the colorectal tract. Animal treatment with PhIP alone induced the formation of ACF and their average number was diminished by co-administration of PCC (Hagiwara et al., 2001).
Lyophilized black raspberries (BRB containing 1.7% ACs) caused a limited reduction effect on AOM-induced rat colon carcinogenesis. BRB had the strongest impact on the initial step of carcinogenesis, characterized by ACF formation. The ACF number was significantly reduced at all given BRB concentrations. The total tumor number was also significantly reduced in the BRB-treated animals compared with controls. No significant differences in adenoma incidence or multiplicity were found among groups. The multiplicity of adenocarcinomas was significantly reduced only in the BRB group that received the highest AC supplement, i.e., 10% BRB. Nevertheless, all treatments (2.5%, 5%, and 10% BRB) showed to significantly  (Harris et al., 2001).
Rats treated with AOM to develop CRC lesions were fed with diets enriched in AC from bilberry, chokeberry, or grape. The total number of ACF was significantly reduced with all three AREs. All extracts caused significant reduction of large or medium ACF. However, none of the three extracts had significant effects on small ACF. Rats fed bilberry and grape extracts, but not chokeberry, exhibited lower expression of COX-2 mRNA levels in colonic mucosa. Reduction in urinary 8-OHdG levels was not observed for any treated group, unlike in the above-mentioned case of BRB (Harris et al., 2001). AREs-fed rats had significant increases in fecal moisture content and fecal amount as well as dramatically reduced bile acids (promoters of colon cancer) in fecal extract. Authors stressed that these processes are connected and can contribute to cancer prevention (Lala et al., 2006).
Anthocyanin-rich compounds, such as purple sweet potato color (PSPC) and red cabbage color (RCC), used as 5% supplement in regular diet, ameliorated some of the parameters of DMH-and/or PhIP-induced carcinogenesis in rats. The average number of ACF was significantly decreased in PhIP-treated when they were fed with RCC, but not with PSPC. Rats that were treated by the combination of two carcinogenic agents, such as DMH and PhIP, developed macroscopic colon nodules, mainly distributed in distal colon and rectum. Diets supplemented with PSPC or RCC caused a significant decrease in number of adenomas, adenocarcinomas, and average tumor number per rat (both benign and malignant) (Hagiwara et al., 2002).
The rat model of AOM/DSS-induced carcinogenesis was used to test the effect of AC extracts enriching sausages, so to mimic a diet containing meat. Extracts from strawberries and blackberries ameliorated some histopathological parameters, such as the number of hyperplastic Peyer patches (used as a measure of pro-inflammatory condition). These diets induced the increase of total antioxidant capacity in the blood plasma. The average number of colon polyps was significantly decreased in animals fed AC-rich sausages, suggesting a contribution of AC to protection against CRC. However, no significant reduction of the total tumor area could be observed. Detailed analysis of intestinal microbiota composition (16S ribosomal RNA sequencing) of caecum content revealed the reduced expression of pro-inflammatory Bilophila wadsworthia in AC-functional sausage group if compared to control sausage group (Fernández et al., 2018).
The palm tree fruit açaì, growing in South America, has a high content of AC (3.19 mg/g dry weight), with cyanidin 3-glucoside and cyanidin 3-rutinoside as the principal ones (Schauss et al., 2006). Lyophilized açai pulp (AP), as 5% or 7.5% supplement, reduced the total number of ACF in rats with colitis-associated colon carcinogenesis. Treatment with 5% AP decreased tumors with high-grade dysplasia more than tumors with low-grade dysplasia. AP also decreased cell proliferation in tumors (Ki-67 staining) and increased expression of tumor suppressor genes, such as Dlc1, Akt3, and the anti-inflammatory genes Ppara (Fragoso et al., 2018).

Mice With Hereditary Predilection to CRC
The adenomatous polyposis coli (APC) min mice model, a standard experimental CRC model for preclinical trials, presents a mutation in the murine homolog of the human APC tumor suppressor gene. As a result, mice develop multiple neoplasms predominantly in the small intestine. The total number of adenomatous polyps (particularly those ≤2 mm) in the small intestine significantly decreased when APC min mice were fed with a basic diet added with sweet potato flesh, skin or AC-rich extract from sweet potato. Sweet potato extracts were significantly more efficient in diminishing polyp numbers than the flesh-or skinsupplemented diet. The number of polyps in colon was even increased with skin and ARE-supplemented diets, showing their inability to counteract carcinogenesis (Asadi et al., 2017).
Non-steroidal anti-inflammatory drugs demonstrated to be effective in cancer prevention. Nevertheless, their prolonged use can cause intestinal tract damage. It has been suggested that their combined use with phytochemicals may improve CRC prevention. APC min mice fed with either lyophilized tart cherry ACs (including cyanidin 3-O-sambubioside and other complex glycosides of cyanidin) or pure cyanidin had different responses on tumor frequency and volume, depending on the intestinal tract area. In the caecal part, adenomas were significantly fewer and smaller in animals fed with either extracts or the pure aglycone with respect to mice consuming the control diet or sulindac, a non-steoidal antiinflammatory drug. The same treatment had no significant effect on colonic tumor volume or number (Kang et al., 2003). A combined diet of AC-rich tart cherry extract with sulindac given to APC min mice resulted in lower tumor number and smaller total tumor area per mouse in small intestine with respect to mice fed sulindac alone. These observations were AC dose-independent. The authors clarified that the equivalent effective dose of AC for humans would be very unusual (more than 1 kg of AC extract/day), so that further tests should determine the lowest effective dose of AC in inhibiting intestinal adenoma development (Bobe et al., 2006).
In the study where APC min mice consumed either Mirtoselect (bilberry extract containing 40% of ACs, such as glucose, galactose, and arabinose conjugates of delphinidin, cyanidin, petunidin, peonidin, and malvidin) or cyanidin 3-glucoside as 0.03%, 0.1%, and 0.3% diet supplements, a dose-dependent reduction of smallsize adenoma number was observed. Adenomas were mostly distributed in the medial and distal sections of the small intestine, with very few lesions in colon (Cooke et al., 2006).

Human Studies
Contrarily to preclinical studies, human studies did not demonstrate encouraging data on the relation between AC-rich diets and prevention of CRC.
In the interventional study performed by Thomasset et al. (2009), a group of 25 CRC patients received an AC-rich extract, mirtocyan (also said Mirtoselect) (0.5-2.0 g/day of AC) for 7 days. The authors had performed preliminary pharmacokinetic studies and identified AC, their methyl and glucuronide metabolites in plasma, colorectal tissue, and urine, but not in the liver. In treated patients, proliferation of cancer tissue was decreased by 7%. In the pharmacodynamic part of the investigation, they found that 9 June 2019 | Volume 10 | Article 675 Frontiers in Pharmacology | www.frontiersin.org the concentration of circulating procarcinogenic Insulin-like growth factor 1 (IGF-1) was not significantly reduced. Mirtocyan did not affect the levels of oxidative DNA damage biomarkers.
A prospective cohort study on middle-aged Finnish men, with mean flavonoid intake of 131.0 ± 214 mg/day, demonstrated no correlation between flavonoid consumption and CRC risk reduction. Anthocyanidins represented 5% (5.9 mg/day) of all flavonoids in food (Mursu et al., 2008).
A retrospective case-control study was performed by including both CRC patients and healthy controls. Eating habits were recorded through. The authors of this study found only borderline significant inverse association between flavonoids/AC (from fruits and vegetables) intake and colon cancer risk increase (Xu et al., 2016).
A huge prospective cohort study with 2519 registered CRC cases found no correlation between high daily flavonoid (4% AC) intake and risk of CRC. Dietary intake data were collected every 4 years through food-frequency questionnaire (Nimptsch et al., 2016).

AC-Related Anti-Cancer Molecular Mechanisms
We have retrieved 10 articles reporting molecular mechanisms of AC-related anticancer activity. Table 7 summarizes features that give information related to this topic. Most of the studies (n = 8) were performed in vitro on cell models; only one was performed in vivo. Early-stage colon cancer cells HCT-116 were used in five studies, whereas four studies used cells associated with advanced stage colon cancer, such as HT-29. Two studies reported data obtained from Caco-2 cells, regarded as a model of differentiated intestinal epithelium; one study used non-cancer colon fibroblast cells, i.e., CCD-18Co. The pure AC molecules used to test cancerpreventive molecular mechanisms were cyanidin (n = 1), cyanidin-3-glucoside (n = 3), cyanidin chloride (n = 1), delphinidin (n = 2), delphinidin-3-O-glucoside (n = 1), or AC-rich extracts from grapes (n = 2), cocoplum fruit (n = 1), and bilberries/blackcurrant (n = 1).
Concentrations of AC ranged from 0.01 µM to 240 µM or 5 to 250 µg/ml. Studies showed mostly positive actions of AC on processes involved in carcinogenesis, like apoptosis, inflammation, proliferation, tight junction modulation, oxidative stress, and tumor number.
Phenolic-rich extracts, rather than AC alone, were able to inhibit proliferation of human colon cancer cell lines, such as HT-29 and HCT-116, pointing to the existence of potential synergic actions of different polyphenols . On the other hand, the same authors concluded that two pure AC molecules, i.e., cyanidin 3-O-glucoside and delphinidin 3-O-glucoside, were more efficient than phenolic-rich extracts as inhibitors of EGFR tyrosine kinase. By targeting this receptor, both molecules showed to interfere with the delicate balance of two opposed processes, like cancer cell survival and programmed cell death. Cyanidin 3-glucoside was also more efficient in reducing adenoma numbers in small intestine than the AC-rich bilberry extract (Cooke et al., 2006).
Cancer cells have both a huge proliferative potential and impaired mechanisms of apoptosis. Some authors  have suggested that targeting these pathways can be promising for the development of new anticancer drugs. They examined the mechanism(s) by which the delphinidin aglycone is able to induce apoptosis in HCT-116 cell line. Delphinidin treatment induced activation of caspases and PARP cleavage, two authentic mechanisms of apoptotic response. Delphinidin treatment resulted in a significant increase in proapoptotic protein Bax and decrease in antiapoptotic protein Bcl-2, leading to cell cycle arrest. Delphinidin also upregulated the expression of both the tumor suppressor protein p53 and its downstream target p21. It caused downregulation of both cyclin B1 and cdc2, again leading to cell cycle arrest. They found that delphinidin treatment of HCT-116 caused downregulation of the NF-κB pathway, as shown by decreased expression of the p65 subunits of NF-κB, reduced IκB phosphorylation, and inhibited nuclear translocation of NF-κB.
In studies on the human colon cancer HT-29 cell line, some authors  suggested PGK1 as a potential biomarker of intracellular oxidative damage. Cells exposed to 50 µM H 2 O 2 for 24 h showed significant expression of PGK1. Interestingly, cells co-treated with delphinidin had attenuated expression of this protein. High levels of PGK1 are associated with tumor survival and angiogenesis. These authors suggested that the antioxidant potential of delphinidin could contribute to anti-cancer strategy.
The apoptotic signaling pathway was also explored in HCT-116 cells ). An AC mix, isolated from grape (AIMs), was used in relatively low concentrations (5-60 µg/ml), easily reachable in vivo. The authors observed decreased expression of pro-caspase-3, -8, and -9, activation of caspase-3 and PARP cleavage. AIMs reduced the expressions of anti-apoptotic proteins (XIAP, cIAP-1, and cIAP-2), so promoting cell death. They suggested that suppression of XIAP was mediated by suppression of the PI3K/Akt pathway. Other apoptotic pathways were influenced by AIMs through increased p38-MAPK phosphorylation and attenuated JNK phosphorylation.
Another AC mixture derived from bilberries and blackcurrant exhibited anti-proliferative effects in Caco-2 cells, by increasing both pro-apoptotic caspase-3 and the cell cycle-related protein p21 Waf/Cip1 . Furthermore, this AC extract caused accumulation of ROS in Caco-2 cells but not in normal fibroblast cells (NIH/3T3), thus demonstrating potentially selective cytotoxicity toward cancer cells.
Some authors suggested that the chemical structure of AC could strongly influence their biological function . They used human colon carcinoma CaCo-2 cells and two ACs, i.e., cyanidin 3-O-glucoside (CY3G) and its aglycon cyanidin chloride (CY), to examine if the substitution pattern of the β-ring could influence AC-mediated cell growth signaling. Reactive oxygen species (ROS) were found to be elevated in almost all cancers. Cancer cells balance ROS levels with antioxidative enzymes levels in favor of their own proliferation and survival. CY (25 µmol/L) was more successful in decreasing ROS level than CY3G. The typical Comet assay showed that DNA fragmentation occurred only at moderate (25-100 µmol/L) and high concentrations of CY3G (200 µmol/L). Possible intercalation of CY3G and CY molecules to DNA was explored in the atypical Comet assay performed on naked DNA; this time, dosedependent fragmentation was observed at low levels of CY3G ).

11
June 2019 | Volume 10 | Article 675 Frontiers in Pharmacology | www.frontiersin.org (25 or 50 µmol/L). In the absence of DNA repair mechanisms, CY-caused damage was even more dramatic, whereas CY3G caused damage only at high concentration (200 µmol/L). CY3G was more efficient than CY in counteracting the H 2 O 2 -induced DNA damage. Specific mutagenic lesions occurring after exposure to ROS were removed by 8-oxoguanine DNA glycosylase (OGG1) (a DNA repair enzyme). Expression of OGG1 was decreased after cell treatment with 25 µmol/L CY and increased at 200 µmol/L. On the contrary, CY3G treatment-induced OGG1 expression at the low concentration of 25 µmol/L. Heat shock proteins (HSP70) are highly expressed in CaCo-2 cells and their expression was even more increased when the cells were treated with CY3G or CY. The same pattern of increased expression of cell cycle-related proteins, such as Ataxia telangiectasia mutated gene (ATM), p53, and topoisomerase IIβ, was obtained by both CY3G and CY. The authors have proposed an explanation, i.e., that the increase in ATM triggers expression of mutated form of p53, not capable to act as a transcriptional factor for p21.
Metabolic alterations in cancer cells can lead to high metabolic activity and increased proton production. Briviba et al. (2001) demonstrated that growth factors, neurotensin and EGF could increase proton extrusion, measured as extracellular acidification, in the human cancer cell line HT-29. Proton extrusion is proposed to be involved in cell growth (Di Sario et al., 1999). This process was inhibited by cyanidin (10 µM), but by none of its glycosides, i.e., cyanin (i.e., cyanidin 3,5-O-diglucoside) and idaein (i.e., cyanidin 3-galactoside). Neurotensin-induced [Ca 2+ ]i was significantly lower in cells pre-incubated with 10 µM cyanidin. Cyanidin did not change extracellular-signal-regulated kinase (ERK) phosphorylation. The suggested role of cyanidin is the modulation of intracellular signaling between receptors and [Ca 2+ ]i, consequently affecting cellular metabolism and growth. Glucosidase, released by intestinal bacteria, could hydrolyze ACs to aglycones, able to inhibit [Ca 2+ ]i and cellular metabolism induced by neurotensin and EGF.
Cocoplum anthocyanins (CP) have been demonstrated to exercise anti-inflammatory activity in both human cancer cells HT-29 and inflamed benign cells . Inflammation is closely related with carcinogenesis. Persistent inflammation favors growth and inhibition of apoptosis in transformed cells.
CCD-18Co colon fibroblasts were selected for their low levels of ROS. Thus, experimental induction of inflammation could be monitored by measurable intracellular increases of ROS. TNF-αinduced inflammation (measured as ROS production) in these non-malignant colon fibroblasts was significantly decreased with CP extract. After this treatment, NF-κB1 mRNA expression was decreased with CP extract, along with protein expression of TNFα, IL1β, and IL-6. The authors also reported decreased mRNA expression of inflammatory mediators (NF-κB1, TNF-α, IL1β, and IL-6) in HT-29 colon cancer cells after CP treatment, whereas protein expression decrease was observed only for TNF-α.
In the studies conducted in HCT-116 , AC isolated mix (AIM) from the Asiatic grape variety V. coignetiae Pulliat (45 µg/ml) diminished the expression of the tight junction (TJ) proteins known as claudins, as well as several matrix metalloproteinases. Preservation of TJ integrity and inhibition of proteolytic digestion of the extra cellular matrix is important to prevent tumor cell mobility and invasiveness. In functional assays, anthocyanins AC also exerted potential anticancer activity. The AC mix increased TJs tightening, measured as transepithelial electrical resistance. The same AIM concentration was successful in diminishing the HCT-116 cells capability to penetrate ECM. They proposed that AIM-induced down regulation of MMP-2 and -9, as well as of claudin-3, can be partially attributed to the p38-MAPK activation and the suppression of the PI3K/Akt pathway.

DISCUSSION
Concerning the general question whether AC prevent or not CRC in vivo, the detailed analysis of the literature enabled to provide some answer to the initially defined questions, as follows: 1) Are there any studies performed on human groups or populations? We were able to identify four studies involving human subjects: one interventional study (Thomasset et al., 2009), one retrospective cohort study (Xu et al., 2016), and two prospective cohort studies (Mursu et al., 2008;Nimptsch et al., 2016). A synopsis of these studies is presented in Table 8. The reported effects of AC were, however, controversial. In fact, the single interventional study showed borderline data, showing that AC administration mildly affected tumor cell proliferation (7% decrease) only at the lowest dose of mirtocyan. No dose dependence could be found, since, two higher doses gave no significant decrease. Other carcinogenesis-related markers, such as serum IGF-I levels, oxidative DNA damage or apoptotic index, were not affected. The authors recognized that they tried elevated doses of AC, 0.5 to 2.0 g/day, corresponding to unusually high servings, ranging 350 to 1,500 g fresh bilberries. In three cohort studies, diets included several flavonoid classes, with an estimated consumption of AC of no more than 6 to 20 mg/day. While one of these studies (Xu et al., 2016) found that CRC prevention correlated with the intake of flavonoids (including AC) in fruits and vegetables, no specific AC-related effect was reported. Two prospective cohort studies found no AC effect. Taken together, these studies showed weak or no correlation between AC intake and CRC risk decrease. 2) Are there any studies performed on animals? What animal models have been used? As many as 15 intervention studies tested the role of AC in CRC prevention in different murine strains, i.e., four rat strains (F344/DuCrj, Fischer 344, Sprague-Dawley, Rattus norvegicus F344) and four mouse strains (Apc Min , A/J, Balb/c, Crj: CD-1). Animals were fed with standard rodent food in combination with supplements enriched with AC extracts or fruits. The AC supplementation ranged from µg to g. In some studies, though, the doses of AC consumed by animals could not be calculated. A synopsis of these studies is in Table 9.
With the exception of the mouse Apc Min model (a knockout for the human homologue of the tumor suppressor APC gene), all murine strains were investigated after chemical induction of CRC lesions of different tumor stages. In seven studies, AC treatments were more efficient on ACF rather than on developed cancers. In the genetic model Apc Min , adenomas/adenocarcinomas were mostly developed in the small intestine than in the colon, where they were very rare, making this model not perfectly representing CRC in humans. Here, four studies showed that the preventive effect of AC was predominantly on adenomas rather than on adenocarcinomas, suggesting that AC may be most effective at early tumor stages. These animal models have both advantages for short-term pre-clinical screening goals and disadvantages related to differences in lifespan, the diet and environmental exposure of human and murine species. Thus, experimental design in murine models of induced CRC and interpretation of results must take these differences in due consideration (Femia and Caderni, 2008).
Concerning the molecular mechanisms whereby AC prevent CRC in vivo, the analysis of the included articles enabled to give answers to our research interests, as follows: 3) Are there any studies performed in cellular models? Only nine articles complying with the inclusion criteria reported in vitro studies on the molecular mechanisms by which AC inhibit CRC progression, as outlined in Table 10.
Numbers of articles were excluded (n = 57), because using heterogeneous polyphenolic extracts, instead of pure molecules. All studies were conducted on human colon cancer cells, such as HCT-116 (n = 4), HT-29 (n = 3), and Caco-2 (n = 2). HCT-116 cells, as a model of early stage colon cancer, were used for studies of proliferation, apoptosis and tight junction modifications. HT-29 cells, as a model of colorectal adenocarcinoma, were used in studies of metabolic activation, inflammatory processes, and intracellular oxidative status. Caco-2 cells, colorectal adenocarcinoma cells, were used in studies of cell cycle/stress-related protein expression, DNA damage, proliferation and intracellular ROS level. Only three studies have used normal benign cells, like the human colon fibroblast lines CCD-33Co, CCD-18Co, and the mouse fibroblasts NIH/3T3. These cells were used to test AC cytotoxicity and inhibition of basal cellular ROS production.

4) What molecules have been used to test anti-cancer effects?
As shown in Table 10, six pure molecules were tested, i.e., cyanidin, cyanidin 3-glucoside and cyanidin 3,5-diglucoside, cyanidin 3-galactoside, cyanidin chloride, and delphinidin. Furthermore, four AC mixtures with purity >90% were used. 5) Which biological processes have been investigated? AC were shown to alter several fundamental processes that play a key role in cell cycle control and carcinogenesis, as listed in Table 11 and illustrated in Figure 2.

6) Have specific molecular targets or markers been identified?
Only two studies assessed the direct interaction of AC molecules with molecular targets, such as EGFR  and DNA , as shown in Table 7.
By contrast, all the other studies focused on downstream effects, mostly assessed as changes in the levels of signaling components regarded as markers of AC-dependent effects. By this approach, it is however impossible to infer the primary cause of molecular marker changes, because of cross talk involving several signaling pathways. Only one marker was described as being in straight correlation with delphinidin treatment of cancer cells, i.e., phosphoglycerate kinase PGK1 .

CONCLUSIONS
With this systematic review we aimed at bringing to light the actual state of the art in the research exploring AC-specific CRC prevention and the underlying oncotargets. AC have shown to protect against early-stage cancer lesions in experimental animals, but not in humans. Several studies in CRC cell models 13 June 2019 | Volume 10 | Article 675 Frontiers in Pharmacology | www.frontiersin.org have shown AC to affect cellular processes related to cell cycle and transformation. However, these effects cannot yet be ascribed to AC-specific molecular interactions with CRC oncotargets, but rather to complex effects resulting in preventing intestinal cells from entering in the cell fate of neoplastic transformation.
We believe that this overview may serve the scientific community to properly select the experimental models and conditions to further clarify the mode of action of AC and their degradation products in modulating the carcinogenetic process in CRC. 10 | List of experimental models used to identify the molecular mechanisms of CRC prevention by AC (not diets or not supplements).

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. This data can be found here: https://www.ncbi.nlm.nih.gov/pubmed/?term =Colorectal+cancer+AND+Anthocyanins.

AUTHOR CONTRIBUTIONS
SP designed the method, created and managed the database of the PubMed search output. NM carried out all phases of the systematic review. Both NM and FT independently screened and assessed the included articles. NM and SP drafted the article.

FUNDING
This work is an activity of the standard project Agrotur II (code 1473843258), funded by Interreg Italy-Slovenia 2014-2020 (European Regional Development Fund and national funds). The activity goal is to improve knowledge on the health effects of grape antioxidants.

ACKNOWLEDGMENT
Marialuce Giorgini participated in this work as a trainee student of the University of Trieste and reported the preliminary results in her BSc thesis (Biological Sciences and Technologies) in October 2018.