Potential role of fructose on human colon DNA methylation in racial disparities observed for colorectal cancer risk

Introduction

The natural sweetness of dietary fructose makes it an appealing additive to processed foods and beverages, and most caloric sweeteners contain at least 40% fructose¹. High fructose corn syrup (HFCS) is the most common form of sweetener added to sugar sweetened beverages (SSB)^{2, 3}. Shifts in dietary patterns have seen dramatic increase of HFCS consumption in recent years, with research showing it to account for approximately 6-10% of caloric intake among US adults^4-6. Similarly, there was also a ∼63% increase in SSB consumption among US children of schooling age between the years 1989 to 2008⁷. Racial disparities in SSB consumption are also evident^8-10. For example, the odds of drinking SSBs one or more times per day is significantly higher among non-Hispanic black respondents (AOR = 1.65) versus non-Hispanic white respondents⁸. This shift in increased SSB consumption and documented racial disparities are gravely concerning given the increasingly recognized role of excess dietary fructose in various, negative health outcomes^11-13, including two recent landmark murine studies, which have begun to define mechanisms linking SSB/HFCS to colon carcinogenesis^{14, 15}. Despite these studies, the role of dietary fructose in healthy human colon, and as such, the mechanisms that may drive fructose-mediated CRC risk, remain poorly understood.

An increasing body of observational studies has directly linked intake of fructose in obesity, diabetes and CRC risk^{11, 16-18}. A small study in Iran recently found that increased fructose intake was associated with CRC after adjusting for multiple confounding factors including body mass index (BMI)¹⁹. This finding is in line with results from a larger study of two prospective US cohorts, which found that dietary fructose and SSB consumptions were both positively associated with incidence and mortality of proximal CRC¹⁸. Other research has shown that high SSB consumption was also associated with early-onset CRC in women¹⁶. Notably, African Americans (AAs) have higher incidence of CRC compared to any other racial/ethnic group²⁰. AAs are also more likely to be diagnosed at a younger age and with poorer survival outcomes than EAs^{20, 21}. Moreover, AAs are more likely to develop right-than left-sided colon cancer and more likely to develop right-sided colon cancer than EAs²². However, the mechanisms driving the observed racial disparities in CRC risk and outcomes and tumor sidedness are poorly understood.

DNA methylation (DNAm) represents one epigenetic mechanism through which gene expression patterns may be modulated to alter disease risk in response to environmental cues²³. Aberrant DNAm is a hallmark of CRC^{24, 25}, and differences in DNAm have also been found to be associated with fructose and HFCS in other tissues^26-28. We have established a repository of right and left normal colon biopsies from a cohort of well-characterized average-risk AA and EA individuals²⁹. In this cohort, we previously reported colon location and race specific DNAm differences, where we found that many of these differences occurred in regions known to be associated with CRC. Further analysis of this cohort has the potential to reveal novel insight into the biological mechanisms that drive CRC racial disparities and to help inform the development of tailored prevention strategies.

Here, we extend the results identified in our previous study to identify differences in DNAm associated with total dietary fructose in AA colon. We hypothesize that this will provide novel insight into the mode of action of fructose in the human colon, and that through the overlay of CRC-related publicly available data, we may contextualize these mechanisms in the framework of CRC risk.

Materials and Methods

Results

Dietary fructose consumption is associated with widespread DNAm differences in AA right colon

We used DMRcate³⁵ to identify significant regions of DNAm that correlated with total dietary fructose consumption. Given the relatively limited sample size of EA individuals who provided data on dietary fructose consumption (n=34), we restricted our analysis AAs (n=79). In right colon, 4,263 DMRs survived FDR corrections. This included a hypomethylated DMR at chr1:9130738-9131854 (FDR=4.09E^-03), a genomic region near the transcriptional start site of SLC2A5 (Figure 1A). DNA hypomethylation of chr11:18415641-18415964, which corresponds to lactate dehydrogenase A (LDHA, FDR=0.036)) was also identified. Increased LDHA expression was previously identified in the intestinal epithelium of HFCS-treated mice¹⁵. To better determine the coordinated effects of dietary fructose on DNAm, we performed pathway enrichment analysis using missMethyl⁴². Here, we identified enrichments for terms previously associated with fructose, such as ether lipid metabolism (P=9.61E^-05)⁴³, arachidonic acid metabolism (P=1.56E^-04)⁴⁴ and linoleic acid metabolism (P=4.75E^-04)⁴⁴ (Figure1B). Other notable enrichments include the glycolytic process through fructose-6-phosphate (P=7.44E^-03) and phosphofructokinase activity (P=9.53E^-03), developmental process (P=6.20E^-04), long-chain fatty acid metabolic process (P=5.45E^-03) and cell population proliferation (P=9.10E^-03).

• Download figure
• Open in new tab

Figure 1: Summary of DMR analysis of dietary fructose in right AA colon.

(A) SLC2A5 was significantly hypomethylated in subject biopsies who consumed increasingly high amounts of fructose. (B) Summary of the top 20 most significantly enriched terms within our DMR analysis. “Overlap” corresponds to the number of DMRs identified within our analysis divided by the total number of genes within that pathway. Red and blue dashed lines correspond to nominal and FDR cut-offs, respectively.

Defining DMR targets through analysis of TCGA-COAD

Given that some DMRs occur within intergenic regions or map near to multiple genes, we aimed to better define the likely targets of each fructose- and CRC-related DMR (fructose risk DMRs) by determining if their corresponding genes were found to be differentially expressed in CRC. Of the three cohorts considered in this analysis, only TCGA-COAD had matched RNA-seq data. A repeated measurement regression model was fitted for each gene to identify differences between matched tumor versus NAT across 41 pairs present within this dataset. Of the 1,581 fructose-related DMRs that were present in at least TCGA-COAD, 869 corresponded to differentially expressed genes (DEGs). Fisher’s test confirmed a highly significant enrichment for the differential expression of fructose risk DMR genes in TCGA-COAD (P=4.06E^-73). In left colon, seven of 16 fructose risk DMRs with gene annotations corresponded to CRC DEGs (Figure 2).

• Download figure
• Open in new tab

Figure 2: Volcano plot demonstrating the significance of genes corresponding to fructose-risk DMRs in TCGA-COAD expression data.

Genes were color coded depending upon whether they were identified as fructose-risk DMRs in right (blue), left (red) or purple (both) analyses. Positive log₂ fold changes correspond to DEGs overexpressed in CRC tumors versus matched NAT.

Discussion

Our study represents the first, relatively large-scale interrogation of the association between DNAm and dietary fructose intake in normal colon tissues of AA patients free of CRC. It presents novel targets through which increased fructose consumption may drive CRC-related signaling. In recent years, there has been a growing interest in the use of dietary interventions to support cancer therapeutics⁴⁵. Tumors frequently develop while retaining the metabolic pathways of their tissue of origin¹⁵. Cells that utilize the energy availability of their environment can display accelerated growth and outcompete their neighbors. Understanding the molecular drivers of these advantages in healthy colon tissue may therefore provide important mechanistic insight into the drivers of dietary risk genes. Such genes may not only provide biomarkers for risk, especially within individuals who consume high quantities of fructose, but may also provide novel targets for future therapeutics by focusing on genes driving preferential tumor-associated microenvironments.

Mouse studies have shown that intestinal absorption capacity is reached following moderate consumption of fructose (1g/kg), leading to unaltered fructose entering the colon⁴⁶. Two landmark mouse model studies demonstrated a role for HFCS in colon carcinogenesis that was determined to be independent of its effect on body weight or the development of metabolic syndrome^{14, 15}. They first show that HFCS enhances intestinal tumorigenesis through aberrant activation of glycolysis and promotion of fatty acid synthesis. Importantly, this study identified KHK as a key gene mediating this process. Later, Taylor et al., demonstrated that feeding mice a HFCS diet led to increased expression of hypoxia-inducible factor 1-alpha (HIF-1α), KHK and SLC2A5, which was accompanied by longer villi in both the proximal jejunum and the duodenum¹⁵. Further analysis in human CRC cell lines revealed that the metabolism of fructose by KHK led to the formation of fructose 1-phosphate, which promoted cell survival through the inhibition of another pyruvate kinase M2 (PKM2), thus contributing to cell survival under hypoxic conditions. These important studies were among the first to define a role for fructose/SSBs in CRC tumorigenesis. In the present study, our analysis of fructose risk DMRs did not identify any regions corresponding to KHK. However, we identified some overlap with the genes and pathways identified by both mouse studies. In this way, our human study provides partial support for previous findings in murine models, while also highlighting the imperative need for further mechanistic investigations into the role of CRC risk mediated by fructose in healthy, human colon epithelium. Several glycolysis and fatty acid terms were among the most significantly enriched in our analysis of fructose in right colon, and many of the DMRs that contributed to those terms, such as CBFA2T3, HDAC4, HTRA2, PGAM2 (glycolysis) and ACSF2, CYP2E1, PGIF, PLA2G4F, SLC27A5 and TNXB (fatty acid) overlapped regions identified in CRC tumors. Tumor cells frequently retain the metabolic pathways of their tissue of origin¹⁵ and the process of glycolysis provides an ample carbon source for fatty acid synthesis, which are required to house energy stores for the excessive energy demands of cancer proliferation¹⁴.

SSBs represent the leading source of added sugars in the US⁴⁷ and added sugars account for over 25% of total calories for one in six AAs, in contrast to one in eleven EAs⁴⁸, indicating that increased levels of dietary fructose consumption, particularly through SSB, may play a role. Importantly, AAs are 30% more likely to develop right-than left-sided colon cancer²¹, and are also more likely to develop right-sided colon cancer than EAs⁴⁹. Differences in health care and dietary patterns driven by poorer socioeconomic status have been considered as contributory factors to observed disparities^{22, 50}. While the role of SSB in racial disparities for CRC has yet to be fully explored, a previous multi-center study of 33,106 participants from the Nurses’ Health Study II found that SSB intake in adolescence was associated with increased adulthood adenoma incidence in proximal (right) colon and rectum. The same study also found that total fructose intake was only associated with increased risk for distal (left) and rectal adenomas¹⁷. However, a larger combined study of the Nurses’ Health Study and Health Professionals Follow-Up Study revealed that both SSB and total fructose were only associated with proximal/right-sided colon cancer¹⁸. Given that SSBs represent more of the daily caloric intake of AAs than EAs, they represent one potential driver of this racial disparities in CRC sidedness that warrants further investigation.

In line with findings showing a greater risk for dietary fructose in right than left colon cancer, our study identified a far greater number of differences in DNAm in right than matched left AA colon. Further, many of those DMRs identified exclusively in right colon (2,019 of 2,025) were also found in CRC tumors. Excess fructose enters the right colon following the saturation of SLC2A5 in the small intestine. Many of the DMRs identified in right colon were also differentially expressed in CRC tumors versus NAT and a subset (list subset) were also validated through a controlled dosing experiment of normal colon organoids. Our findings therefore imply a coordinated mechanism through which dietary fructose alters DNAm at regions relevant to CRC, leading to DEGs that modulate disease risk. The amount of fructose and its direct metabolites that are present in the left colon following further absorption to the hepatic portal system remains unclear. It is worth noting that while our study identified many more fructose-risk DMRs in AA right than left colon, a subset of fructose-risk DMRs exclusive to left colon were also found, including a hypermethylated DMR corresponding to the intestine-specific transcription factor, CDX2. A reduction in this tumor suppressor was also identified between CRC tumors and NATs in our analysis of TCGA-COAD. CDX2 plays important roles in the regulation of epithelial cell differentiation and the reduction of its expression has been previously associated with poorer survival outcomes in left-sided colon cancer⁵¹. While the data presented within this study provides potential mechanisms that may drive the observed increase in right-sided colon cancer risk, larger epidemiological studies are needed to confirm this through finer site-stratified analyses. Given the notable differences in consumption of SSBs between AA and EA, and the relatively higher risk of AAs to develop right-sided colon cancer, these studies should also consider the effects of racial ancestry and ethnicity.

To relate our findings to CRC risk, we made use of publicly available CRC data and overlaid findings with those found to occur in CRC tumors. However, there are a number of limitations to this approach. First, CRC is a highly heterogenous disease and as such, it stands to reason that CRC risk pathways are also diverse. Second, total dietary fructose was highly correlated with other dietary sugars. Despite this, our results replicated multiple previously identified pathways and targets that are specific to fructose, including SLC2A5. However, validation through the coupling of high-throughput omics readouts and controlled dosing within the colon organoid system may further reveal mechanistic insight into the role of fructose in CRC. Nonetheless, our findings highlight several plausible targets through which dietary fructose may influence DNAm to modulate CRC risk and some of these targets were verified through fructose exposure to human AA colon organoids.

Conclusion

There is increasing evidence for a role of dietary fructose and HFCS/SSB consumption in CRC risk. Here, we provide novel insight into colon location-specific mechanisms in AA subjects through which this risk may develop. That fructose consumption is associated with a far greater number of potential CRC risk DMRs in right than matched, left colon is intriguing. AAs are significantly more likely than EAs to consume greater quantities of fructose and to develop right-side colon cancer. Our findings highlight potential mechanisms through which increased dietary fructose consumption may differentially impact DNAm at regions associated with CRC to alter risk. Our findings provide some mechanistic support for reducing dietary fructose consumption.