Diet, Genes, and Microbes: Complexities of Colon Cancer Prevention

Abstract

Colorectal cancer is one of the leading causes of cancer-related deaths in the United States, and generally, as countries climb the economic ladder, their rates of colon cancer increase. Colon cancer was an early disease where key genetic mutations were identified as important in disease progression, and there is considerable interest in determining whether specific mutations sensitize the colon to cancer prevention strategies. Epidemiological studies have revealed that fiber- and vegetable-rich diets and physical activity are associated with reduced rates of colon cancer, while consumption of red and processed meat, or alcoholic beverages, and overconsumption as reflected in obesity are associated with increased rates. Animal studies have probed these effects and suggested directions for further refinement of diet in colon cancer prevention. Recently a central role for the microorganisms in the gastrointestinal tract in colon cancer development is being probed, and it is hypothesized that the microbes may integrate diet and host genetics in the etiology of the disease. This review provides background on dietary, genetic, and microbial impacts on colon cancer and describes an ongoing project using rodent models to assess the ability of digestion-resistant starch in the integration of these factors with the goal of furthering colon cancer prevention.

Introduction

Colorectal cancer (CRC) is the third most common cancer diagnosed among U.S. men and women with ~130,000 new cases diagnosed each year. Six percent of all Americans, 1 out of 17, will develop CRC (American Cancer Society 2012), and only ~5% to 10% of these cancers are explained by a specific genetic susceptibility. A long-standing issue in cancer biology has been attributing the cause of cancer to environment and genetic factors. Key environmental factors for colon cancer include diet, as discussed below. The genetics of colon cancer were known to be complex long before the genomics age. The recent use of deep deoxyribonucleic acid (DNA) sequencing to study CRC genetics is revealing that indeed CRCs are some of the genetically most complex cancers that have been widely subjected to these investigations. The recent developments that are facilitating genomic studies of gut microorganisms are revealing new insights into the interactions between diet, cancer genomics, and gut microbial communities (microbiota) in colon cancer etiology. Consequently, the face of colon cancer prevention is becoming much more exciting, in that it may be modifiable; but daunting, because of the complexity.

Complexity of Colon Cancer Genetics

Genetic research into colon cancer has benefited from several hereditary syndromes that confer high risk for colon cancer, in particular these syndromes have allowed investigators to identify high-penetrance genes for colon cancer (de la Chapelle 2004). High-penetrance genes associated with hereditary colon cancers include the adenomotosis polyposis coli (APC) gene in familial adenomotosis polyposis, the mismatch repair mutations in Lynch syndrome, and axis inhibition protein 2 (Axin2), DNA polymerase δ (POLD), and transforming growth factor-β receptor 2 (TGFβR2) in familial CRC (de la Chapelle 2004). Identification of genes such as these that confer high risk for developing colon cancer set the stage for more extensive research into the genetic factors associated with sporadic colon cancer. It has been proposed that sporadic cancer is due to mutant low-penetrance genes that interact more extensively with environmental factors (de la Chapelle 2004). Studies into the pathogenesis of colon cancer revealed that distinct mutations are associated with biologically and clinically distinct types of colon cancer (Fearon 2011).

Genome wide sequencing technologies and strategies to interpret the massive amount of sequence data that are obtained have revealed far more mutations in cancers than were previously expected (Vogelstein et al. 2013). Colon cancer has received considerable attention in the search for mutated genes using these approaches. Vogelstein and collaborators have also been able to identify the driver genes whose alternations are required for the development of colon cancer and passenger genes that are likely not involved in cancer development and may result in part from the genomic instability in cancer (Vogelstein et al. 2013). For colon cancers with alternations in microsatellite unstable driver genes there can be 100 to 200 mutant genes, while a colon cancer with microsatellite stable profile of mutant genes may have nonsynonymous mutations in the range of 50 to 100 (Vogelstein et al. 2013). Common driver mutations in colon cancers include APC, rat sarcoma family of genes (RAS), phosphatidyl inositol-3-kinase (PI3K), and TGFβ; passenger mutations are much harder to identify and presumed to account for the vast majority of mutations in cancer, particularly those in the noncoding regions of DNA. Although alterations in protein coding DNA are believed to have a higher potential of being driver mutations, the majority of colon cancer mutations in the protein coding DNA are single base substitutions and these are believed to be observed more often in passenger genes. Driver genes are more often identified by their chromosome translocations, homozygous deletions, and gene amplifications (Vogelstein et al. 2013). While tremendous progress is being made in understanding the altered genes that are responsible for the development of colon cancer, each human colon tumor has a unique set of mutant genes and sorting out to specific contribution of each gene in this complexity continues to be challenging.

Complexity of Dietary Modulation of Colon Cancer

Considerable evidence points to colon cancer being among the cancers that are most modifiable by diet (World Cancer Research Fund/American Institute for Cancer Research 2007). Dietary factors for which there is convincing, probable, or suggestive evidence that they increase or decrease the risk of cancer were assessed by a systematic review process and the results are summarized in Table 1 (World Cancer Research Fund/American Institute for Cancer Research 2007). Dietary factors with convincing evidence for increasing colon cancer risk include red meat, processed meat, alcoholic beverages (in men), abdominal fatness, and adult attained height. Probable protective dietary factors include foods containing dietary fiber, garlic, milk and calcium, while alcoholic beverage consumption (in women) is probable for increasing risk. In agreement with this assessment, comprehensive dietary modification was recommended involving reducing red meat, processed meat, highly refined grains, and simple sugars. Recommended substitutions to fulfill protein needs were with poultry, fish, and plant sources; to fulfill fat needs were with primarily unsaturated fats; and to fulfill carbohydrate needs were with unrefined grains, legumes, and fruits (Chan and Giovannucci 2010). They urge that these primary prevention strategies will complement colon cancer screening in the prevention of colon cancer. They also summarize the potential impacts of these and other potential dietary mediators of colon cancer on inflammatory pathways involving arachidonic acid metabolism, and such pathways have been found to be important in colon carcinogenesis (Chan and Giovannucci 2010).

Table 1.

Diet and activity factors related to colorectal cancer prevention.

	Decreases risk	Increases risk
Convincing	Physical Activity	Red meat Processed meat Alcoholic drinks (men) Body/abdominal fatness Adult attained height
Probable	Foods containing dietary fiber Garlic Milk Calcium	Alcoholic drinks (women)
Suggestive	Nonstarchy vegetables Fruits Foods containing folate/Se or Vit D Fish Se	Cheese Foods containing Fe/animal fat or sugars

Source: Abstracted from World Cancer Research Fund/American Institute for Cancer Research (2007).

Epidemiological studies on immigrant populations have strongly supported a role of diet in CRC. For example, Japanese people in Japan have traditionally experienced much lower incidences of CRC than has been observed in people in Western countries. However, when Japanese migrate to the United States their rates of CRC increase to or above the CRC rates found in the U.S. population (Moore et al. 2005). Changes in cancer rates with migration have long been expected to reflect environmental factors, with diet being a prime mediator (Moore et al. 2005).

Probably the longest standing controversy in the area of diet and colon cancer has been the theory that diets high in fiber are protective against colon cancer. This important issue has continued to receive considerable attention and a meta-analysis of prospective studies revealed a dose–response protective relationship (Aune et al. 2011). Furthermore, more recently an 11-year follow-up of the ongoing European Prospective Investigation into Cancer and Nutrition (EPIC) study with 521,448 participants supported the hypothesis that fiber-rich diets protect against CRC. The results suggested a significant protection in both genders with colon and rectal cancers (Murphy et al. 2012). They observed a protective association that did not differ by age, gender, anthropometry, lifestyle, or diet for CRC. For colon cancers, fiber from cereals or from fruits and vegetables were similarly protective; but for rectal cancer, cereal fiber was the only form that was statistically significant (Murphy et al. 2012). In contrast, the controversy on the role of high-fat diets in the cause of CRC continues with recent meta-analysis of 13 prospective cohort studies finding no evidence for increases in CRC with high-fat intakes (Liu et al. 2011).

Complexity and Impacts of Diet on the Microbiome

Considerable advances have been made in DNA sequencing technologies and strategies to identify and characterize the human microbiome, which includes the genetic composition of the microbial communities that inhibit specific environments. Although most of the species of microorganism remain unculturable, metagenomic approaches are revolutionizing our understanding of the bacteria that live throughout the human body (Grice and Segre 2012). For example, the NIH Human Microbiome project revealed striking diversity in the bacteria colonizing human sites such as the mouth, skin, gut, and urogenital tract (Grice and Segre 2012). Each region of the body had signature taxa of microorganisms. Further, the metabolic and functional characteristics of the microorganisms in each body region were more stable than the particular taxa and, fortunately, pathogenic bacteria were rarely present in the normal microbiota (Grice and Segre 2012). In focusing on the gut microbiome, it has been widely proposed that a healthy gut microbial profile is important for health maintenance, nutrient metabolism, drug metabolism, and proper immune system regulation (Dutton and Turnbaugh 2012; Xu et al. 2013). Recent reviews have addressed dietary polysaccharides (Xu et al. 2013), fiber (Kuo 2013), polyphenols (Moco, Martin, and Rezzi 2012), and probiotics (Arora, Singh, and Sharma 2013) on human gut microbiome. Physiological conditions such as obesity (Tehrani et al. 2012) are associated with an altered microbiome. In general, abundance of Prevotella in the gastrointestinal (GI) tract compared to Bacteroides has been associated with diets rich in fruits and vegetables with modest meat intakes (Jeffery and O’Toole 2013). However, no one composition of the gutmicrobiome has been conclusively related to health promotion and disease prevention. Wu et al. (2011) studied the gut microbiome in 98 individuals using 16s RNA gene sequencing and assessed diet by recall using food frequency questionnaires. They found that Bacteroidetes and Actinobacteria were positively associated with dietary fat and negatively associated with dietary fiber intakes; Firmicutes and Proteobacteria were negatively associated with fat and positively with fiber. The relative abundance of Prevotella was greater in individuals with a carbohydrate-based diet, while Bacteroides was more abundant in those with high meat consumption. In 10 of the subjects, they conducted a controlled feeding trial comparing a high-fat/low-fiber with a low-fat/high-fiber diet and analyzed stool DNA on days 1 and 10 and observed modest changes in enterotype after 24 hr and no stable switch in enterotype after 10 days of controlled diet (Wu et al. 2011). The results from this and many other studies suggest that humans exhibit a stable gut microbiome that resists change in short-term studies and that returns to the individuals typical microbial profile when they go off of an experimental protocol or an intervention such as a probiotic designed to alter their gut microbiota.

The Gut Microbiome and CRC

Evidence is growing that the gut microbiome plays a role in the development of CRC. For example, a relative higher abundance of Fusobacterium was observed in the rectal mucosa of CRC patients and in individuals with colorectal adenomas than in healthy controls (McCoy et al. 2013). Further, these investigators observed positive correlations between local expression of cytokine genes associated with inflammation and the abundance of Fusobacterium, suggesting its possible contribution to mucosal inflammation. However, currently there is no consensus on particular microbial profiles that are characteristic of colon cancer.

The gut microorganisms play important roles in nutrient metabolism and absorption. For example, short chain fatty acids are the principal metabolites generated from colonic microbial metabolism of carbohydrate and protein (Macfarlane and Macfarlane 2012). Acetate, propionate, and butyrate are the most abundant short chain fatty acids generated, but there is considerable interest in the diversity of short chain fatty acids produced. The production of butyrate by gut microorganisms is believed to, at least in part, explain protection against colon cancer by butyrate-producing bacteria for several reasons: (1) butyrate serves as the principal energy source for the enterocyte, providing 60% to 70% of its energy requirement, (2) high butyrate production by gut bacteria has been associated with reduced colon cancer risk, and (3) butyrate treatment of cultured colon cancer cells blocks their proliferative, precancerous phenotype and promotes differentiation and apoptosis (Macfarlane and Macfarlane 2012). However, the optimal doses of butyrate for promotion of differentiation in cultured cells is much lower than the level of butyrate commonly found in the intestinal contents, so the congruence of the argument for butyrate being key is questionable (Macfarlane and Macfarlane 2012).

Furthermore, it is probably significant that the gut microorganisms also contribute to protein digestion yielding amines, phenols, indoles, thiols, carbon dioxide (CO₂), hydrogen (H₂), and hydrogen sulfide (H₂S), and many of these metabolites may have toxic properties (Macfarlane and Macfarlane 2012). It has been proposed that production of potential toxic metabolites needs to be balanced by antiproliferative–anti-inflammatory agents (Arthur and Jobin 2011). Finally, the relationship is likely more complex because of the role of the gut micro-biome in inflammation in the colon (Arthur and Jobin 2011) and the importance of inflammation in colon carcinogenesis.

Studies on Dietary-resistant Starch on Colon Cancer Pre-neoplasia and Colon Microbial Populations

Introduction

Digestion-resistant starch resists enzymatic hydrolysis in the upper intestine and is instead metabolized by the gut microbiota in the lower intestine (Englyst, Kingman, and Cummings 1992). Resistant starch is hypothesized to be useful in the prevention and control of diabetes because it is not rapidly hydrolyzed to simple sugars and absorbed, and it is hypothesized to protect against colon cancer because it is metabolized by the gut microbiota into short chain fatty acids such as butyrate.

Approach, Hypothesis, and Objectives

An integrated research team involving 16 research laboratories was assembled with expertise in food and nutrition science, starch biochemistry, microbiology, statistics, agronomy, sociology, and pathology to develop novel resistant starches, to produce acceptable food products with these starches, and to test these resistant starches in humans and laboratory animals. A naturally digestion-resistant starch was studied, high amylose starch (HA7), and a more resistant processed starch was developed by complexing steric acid (SA) with the high amylose starch (HA7-SA). The HA7-SA starch was shown in humans to reduce postmeal glucose and insulin elevations that were observed in humans, given a meal containing normal starch (Hasjim et al. 2010). This report will primarily focus on published studies conducted on the prevention of pre-neoplastic lesions in the colon of rats (Zhao et al. 2011). The hypothesis being tested in this experiment was that the colonic fermentation of the undigested portion of resistant starch to short chain fatty acids is central for colon cancer prevention. The fermentation products, especially butyrate, benefit colonic health, by regulating colonic enterocyte proliferation, differentiation, and apoptosis. This regulation leads to a less proliferative and more differentiated phenotype and results in the production of fewer pre-neoplastic lesions by the colon carcinogen azoxymethane (AOM). We further hypothesized that resistant starch diets have a significant impact on the composition of the bacterial community in the colon and that the change in gut microbiota provoked by resistant starch could favor microorganisms that produce butyrate.

Materials and Methods

Some of the materials and methods were previously described in detail (Zhao et al. 2011).

Diets

HA7-SA was prepared by our collaborators (Hasjim et al. 2010) and studied in the colon cancer prevention experiments described here. The HA7-SA starch was compared with normal corn starch and HA7 that is naturally more digestion-resistant than normal starch. These three starches were cooked in boiling water and cooled before incorporation at 55% into a semi-purified rodent diet (Zhao et al. 2011). The resistance of the starches and diets were determined as described (Zhao et al. 2011).

Animals

Fischer 344 rats were fed diet with control starch from 5 weeks of age through 8.5 weeks of age. At 7 and 8 weeks of age, 30 rats were treated with AOM at a dose of 20 mg/kg body weight each week and 15 rats were treated with saline. Groups of 10 AOM-treated and 5 saline-treated rats were fed each of the three diets from 8.5 weeks of age for 8 additional weeks until they were killed and the ceca and colons were collected for analysis. The ceca were weighed with and without contents and the pH of the contents was recorded. The colon was opened and washed with saline and the distal 75 mm was fixed and stained for assessment of aberrant crypt foci (ACF) and mucin-depleted foci (MDF; Femia et al. 2007). ACF are putative precursors of colon cancer identified in rodents in 1987 (Bird 1987). They are abnormal colonic crypts (a) of at least 2 to 3× larger than normal crypts, (b) having increased pericryptal space, (c) having a slit-like opening, and (d) having thicker epithelium that stains darker with methylene blue. MDF were proposed to be aberrant crypts with reduced mucin production that are identified by alcian blue, a cationic dye that stains acid mucins. They were proposed to highly correlate with tumor development in AOM-treated rats (Femia, Caderni, and Caderni 2004).

Bacterial Community Profiling

Total DNA was extracted from the distal colonic contents of individual rats using a commercially available kit (PowerSoil DNA Isolation kit; Mo-Bio Laboratories, Inc; Carlsbad, CA). The DNA was used as template for polymerase chain reaction (PCR) amplification of the V3 to V5 variable region. Protocols for PCR amplification, indexing (bar coding) of individual samples, and sequencing library preparation were done essentially as described for the Human Microbiome Project (hmpdacc.org/doc/16S_Sequencing_SOP_4.2.2.pdf). PCR products were quantified by a Quibit fluorometer (Life Technologies, Inc; Carlsbad, CA) and submitted to the Human Genome Sequencing Center at Baylor University School of Medicine for amplicon sequencing using Roche/454 pyrose-quencing technology.

DNA sequence data were filtered to remove data that were not classified as bacterial at 90% confidence. Data were processed using the QIIME qiime.org (Caporaso et al. 2010) and mothur.org (Schloss et al. 2009) programs, which included aligning sequences to the Ribosome Data Base and generating distance matrices from the aligned sequences. Sequences were further analyzed to identify the OTUs, that is, operational taxanomic units and consensus taxonomies.

Result

Some of the results were previously described in detail (Zhao et al. 2011).

Resistant Starch Content of Boiled Starches and Diets

Following cooking in boiling water the control, HA7, and HA7-SA starches had resistance on a dry starch basis of 4% ±1%, 39% ±2% and 71% ±2%, respectively. The prepared diets with control, HA7, and HA7-SA starches had 2% ± 2%, 14% ±3%, and 26% ±5% resistance, respectively. No differences were detected in food disappearance over the 8 weeks of experimental diet feeding; however, disappearance data are prone to many errors.

Body Weight, Cecal Content Weight, and pH

Body weights did not differ by AOM-saline treatment or by diet treatment and a gain of 101 ±15 was observed in the rats by the end of the experiment. Total cecal contents increased in weight threefold to fourfold between the rats fed control diet and those fed the HA7 and HA7-SA diets in both carcinogen and saline-treated rats (p <.05). Cecal pH decreased progressively from the control to the HA7 and the HA7-SA treatment groups, irrespective of carcinogen treatment (p <.05).

Pre-neoplastic Lesions

ACF were somewhat reduced between the control and the HA7 (~16% reduction) and HA7-SA treatment (~37% reduction) but the differences were not statistically significant. MDF were dramatically reduced (p <.05), with an approximate 50% reduction in the rats fed the HA7 diet and an approximate 90% reduction in rats fed the HA7-SA diet.

Impact of Resistant Starch on the Bacterial Microbiota

The composition of the bacterial communities at the phylum level in the lower GI tract of the animals is summarized in Figure 1. As anticipated, the Firmicutes dominate the GI microbiota in rats fed a conventional diet. However, both the HA7 and the HA7-SA diets shifted the distribution of bacteria such that the Bacteriodetes significantly increased in prevalence, while the Firmicutes were reduced. The Actinobacteria were also reduced, but this change was observed only in rats on theHA7-SAdiet. All of the resistant starch diets also correlated with a reduction in Proteobacteria, a group that includes bacteria associated with GI disease.

Figure 1.

Phylum-level distribution of the gut bacteria from rats fed different diets, including conventional, control diet (top), and resistant starches HA7 (middle) and HA7-SA (bottom). HA7 = high amylose starch; HA7-SA, high amylose starch complexed with steric acid.

Discussion

This research provided suggestive evidence for colon cancer prevention by dietary-resistant starch. However, subsequent research with rats treated with antibiotics revealed that the MDF may not be reliable predictors of colon cancer risk because MDF were observed in rats treated with antibiotics but not treated with carcinogens and fortunately there is no evidence that antibiotics heighten the risk for colon cancer. Ongoing studies are assessing the prevention of β-catenin expressing lesions as well as ACF in mice. The three major phyla that predominate in the mammalian gut, Firmicutes, Bacteroidetes, and Actinobacteria, all showed significant changes in relative abundance in animals fed resistant starch. While members of each of the major taxonomic units are known to contribute to starch fermentation in the gut, we observed that the numbers of Firmicutes were reduced, while the Bacteriodetes were significantly elevated in rats fed HA7 and especially HA7-SA diets. This result is consistent with a recent report that showed that a chemically modified resistant starch (RS4) likewise increased the Bacteroidetes and decreased Firmicutes in a human feeding study (Martıńnez et al. 2010). Interestingly, this pattern does not hold for all diets, however, as the type 2 resistant starch (RS2), a crystalline native starch, did not correlate with an increase in Bacteroidetes (Martıńez et al. 2010). Likewise, the abundance of Firmicutes/Proteobacteria positively correlated with dietary fiber intake in humans (Wu et al. 2011). We conclude from these collective studies that resistant starches and dietary fibers do not impact the gut microbiota in the same way, and significant changes can be brought about by different starch formulations. Other differences among the studies could also be attributed to differences between humans and rodents and the length of the dietary regimines. A better understanding of how the changes in bacterial composition of the colon are associated with different types of resistant starch should be informative regarding how dietary changes can help prevent CRC.

Overarching Discussion and Future Research Directions

There is considerable ongoing research into CRC genomics, diet impacts on CRC, the role of the gut microbiota in CRC, and the impact of diet on the gut microbiome. Research on all of these fronts will be required to turn the complexity into strategies for CRC prevention and control. Research by our team is currently evaluating the impact of resistant starches on alternative colon pre-neoplasia and CRC models and assessing the impacts of these modifiers on the colon microbiota. Other close collaborators are assessing gene expression in colon crypts (base to tip) and exploring the changes in small molecules in the colon contents.

Challenges to be addressed include identifying tight mechanistic links between the metabolic activities of the gut microbiota and its impact on host health. For example, are levels of butyrate, and other short chain fatty acids, sufficiently elevated to prevent the development of CRC? Also, are other microbial metabolites also beneficial to the host in preventing cancerous lesion development? Taxanomic profiling of microbial community composition in both the rat study reported here (data not shown) and human studies (Martıńez et al. 2010) revealed signficant variation in microbial composition among individuals in their response to resistant starch diets. Identifying the host factor that contributes to these interindividual differences will be important to develop new dietary interventions for prevention, control, and treatment of CRC. Identifying ways to stabilize a more beneficial microbiota in the host will also be important. Investigation into how dietary intervention in combination with administration with probiotics should be informative. Fortunately, the continued development of new technologies, including deep sequencing at reduced costs, improved methods for monitoring transcriptional changes in both the microbiota and the host (transcriptomics), more sensitive ways to measure protein (proteomics) and metabolite (metabolomics) abundance on a global scale, as well as detecting epigenetic modifications to the host (e.g., through ChIP-seq), will accelerate our understanding of cancer development, prevention, and control.

Abbreviations

AOM

azoxymethane

APC

adenomotosis polyposis coli

CRC

colorectal cancer

DNA

deoxyribonucleic acid

EPIC

European Prospective Investigation into Cancer and Nutrition

HA7

high amylose starch

HA7-SA

high amylose starch complexed with steric acid

ACF

aberrant crypt foci

MDF

mucin-depleted foci

CO₂

carbon dioxide

H₂

hydrogen

H₂S

hydrogen sulfide

PCR

polymerase chain reaction

OTUs

operational taxanomic units

RS2

type 2 resistant starch