Genomics: Experimental approaches to identify gene–nutrient interactions

The term nutrigenetics focuses on investigating the relationship between common genetic variations or polymorphisms (or epigenetic modifications) and the nutritional environment. Such gene–nutrient combinations may determine an individual’s nutrient requirements, their metabolic response and/or their responsiveness to a nutritional intervention, all of which may predispose an individual to a lesser or greater risk of developing a diet-related disease. Personalised nutrition approaches sometimes focus on the effect of genetic variation in response to dietary change, because some polymorphisms or epigenetic states may determine an individual’s response and/or the therapeutic efficacy of a dietary intervention, which may in turn determine the outcome of certain disease states.

The candidate gene approach has been traditional for identifying the genes involved in a diet-related condition. The candidate genes can be identified according to biological function and/or linkage studies. Each subject with and without the disease (case and control groups) donates a DNA sample, the sequence of which is determined using an appropriate sequencing or array platform. Then association tests are carried out for significant differences in the allele frequencies of the SNPs of interest. If one variant of the allele is more frequent in the ‘at-risk’ patient group compared to a healthy control population, then the SNP is ‘associated’ with the disease. This candidate gene approach can be used within a case-control and/or prospective study design. While there have been a number of studies published using this approach, and interesting research has shown different interactions between different-risk SNPs and nutritional status. Nevertheless, it is important to note that overall there has been limited success from candidate gene studies, in terms of defining the genetic determinants of diet-related diseases. This limitation reflects the fact that a number of metabolic pathways with several candidate genes are often involved in diet-related disease and metabolic responses.

Furthermore, it is important to be aware of the potential limitations of gene association studies. Firstly, it is essential to replicate any potential finding in an independent cohort, as too often positive results in one cohort have not been replicated in subsequent studies. The main reasons for this are inadequate statistical power, multiple hypothesis testing, population stratification, publication bias and phenotypic differences. It is becoming increasingly evident that the identification of true genetic association in common multifactorial conditions, such as obesity, T2D or cancers, requires large studies consisting of thousands rather than hundreds of subjects. In addition, it is desirable to demonstrate a functional effect of a given SNP. For example, if a polymorphism introduces a missense change in the coding region of the peroxisome proliferator activator receptor gamma (PPARγ) gene, it probably represents a functional variant. Alternatively, functional assays may show greater or lesser activity of the protein arising from the gene. As an example, it has been shown that the Pro12Ala polymorphism in the transcription factor PPARγ gene was associated with less DNA-binding affinity and reduced transcriptional activity. Lastly, the absence of large single-gene effects and the detection of multiple small effects accentuate the need for the study of larger populations in order to identify reliably the size of the effect that we now expect for complex diseases.

Positional cloning involves mapping the susceptibility/causative loci purely on their chromosomal location, using multigenerational pedigrees and/or a large number of sibling pairs. This allows identification of genes without any prior knowledge of biological function or the disease mechanism. Linkage and linkage disequilibrium (LD) analysis relies on the fact that genes with similar chromosome positions will only rarely be separated during genetic recombination, so that susceptibility to causative genes can be localised by search for genetic markers that co-segregate with disease. However, to date this approach has identified relatively few candidate genes relevant to diet-related diseases. One of the earliest significant linkage peaks was at chromosome 2q37.3, which led to the identification of Calpain 10 (CAPN10) as a new putative diabetes gene. The authors implicated an A to G polymorphism in intron 3 of the gene encoding CAPN10 with a greater risk of T2D. Subsequent to this initial report, several groups showed modest associations between CAPN10 polymorphisms and metabolic phenotypes associated with T2D. Nevertheless, other groups have failed to show a relationship between CAPN10 and metabolic traits indicative of T2D. The lack of a firm gene–phenotype relationship arising from a gene, identified using a positional cloning approach, may be due to a number of causes of inconsistency in association studies, including population-specific patterns of LD, population-specific environmental triggers and gene–gene or gene–nutrient interactions.

Genome-wide association studies (GWAS) focus on determining the prevalence of several (thousands of) SNPs simultaneously. The principle is similar for the candidate gene approach, with GWAS comparing the DNA of groups of cases versus control subjects. The DNA sequence of the cases and controls is determined using SNP arrays to determine whether the frequency of gene variant(s) is greater in the population with the condition versus controls. The associated SNPs are then considered to mark a region of the human genome that influences disease risk. GWAS goes beyond the candidate gene approach, which is limited to one or a few genetic regions based on prior knowledge and research bias. GWAS investigates the entire genome, thus it has the opportunity to identify novel genetic determinants. Again this approach has limitations, as while GWAS can identify new SNPs in DNA that are associated with a disease, it cannot specify which genes are causal. The potential limitations of GWAS relate to optimal study design. Sufficient numbers of accurately phenotyped cases and controls, adequate measures to control and correct for multiple testing to avoid the generation of false positive results and control for population stratification are important design issues that need to be addressed at the outset of a new study.

Next-generation sequencing (NGS) and high-throughput sequencing are technologies that have the capability greatly to enhance efficiency, producing thousands or millions of sequences at once. High-throughput sequencing technologies are intended to lower the cost of DNA sequencing compared to standard methods. NGS achieves high-throughput sequencing by completing the sequencing process of multiple DNA fragments that overlap in parallel or simultaneously. Then the final DNA sequence is reassembled by collating the DNA sequence information from the multiple overlapping DNA fragment sequences.