Metabolomics

Metabolomic profiling has the potential to identify biomarkers of food intake⁽ Reference Oresic ^{14 )}. Metabolomics is a new high-throughput technology that measures endogenous metabolites in cells, tissues and other biosamples such as blood, urine and saliva⁽ Reference Nicholson and Lindon ^{15 )}. These metabolites provide a direct signature of the biochemical activities of the individual at a particular time, reflecting the metabolic effects of nutritional intake, physical activities and environmental exposures as well as the biological pathways associated with diet-related diseases including obesity, CVD and type 2 diabetes⁽ Reference Suhre, Meisinger and Doring ^{16 –} Reference Menni, Mangino and Cecelja ^{19 )}. Metabolites have been proven to be of use in nutritional research and novel dietary biomarkers within the metabolome have been identified⁽ Reference Altmaier, Kastenmuller and Romisch-Margl ^{20 –} Reference Scalbert, Brennan and Manach ^{23 )}. However, as the individual genetic makeup influences the levels of metabolites, explaining up to 81 % of the variation in their blood levels⁽ Reference Illig, Gieger and Zhai ^{24 –} Reference Shin, Fauman and Petersen ^{27 )}, the twin model can be employed to estimate the nutritional impact on metabolites, segregating the genetic component.

To date there are relatively few studies that combine a twin study design and metabolomics in nutritional sciences. Studies from TwinsUK have successfully identified potential biomarkers of alcohol intake⁽ Reference Jaremek, Yu and Mangino ^{28 )} including lysophosphatidylcholines, diethyl ether lipids, diaclylphosphatidycholines and sphingolipids, as well as biomarkers of self-reported dietary patterns⁽ Reference Menni, Zhai and Macgregor ^{22 )}, food preference patterns⁽ Reference Pallister, Sharafi and Lachance ^{9 )} and self-reported food intakes⁽ Reference Pallister, Haller and Thorand ^{29 ,} Reference Pallister, Jennings and Mohney ^{30 )}.

Taking advantage of the twin nature of the TwinsUK data, we first looked for association in the larger twin population excluding MZ twins discordant for a nutritional intake. Then, for each significant metabolite–dietary variable we run the same analysis on the discordant MZ twin pairs to replicate results in this set.

Using a targeted metabolomics approach (Bioacrates platform; 163 metabolites), we identified forty-two dietary pattern–metabolite significant associations in 1003 female twins and successfully replicated eleven associations in a subcohort of MZ twin discordant for dietary intake (between twenty-eight and forty discordant twins independent from the first analysis)⁽ Reference Menni, Zhai and Macgregor ^{22 )}. The strongest associations were observed for fruit and vegetables intake with the glycerophospholipid phosphatidylcholine diaclyl C₃₈ : 6 and with the sphingolipid sphingomyeline C₂₆ : 1. Significant associations were also found with coffee and garlic intake and for hypoenergetic dieting.

In a more recent study from our group, we looked for association between 601 blood metabolites (456 measured with the untargeted Metabolon platform and 145 measured with Biocrates) and seventy-one reported food intakes from FFQ in 3500 female twins to look for novel biomarkers of nutrition intake. We identified 180 associations with thirty-nine food group after meta-analysing the discovery and replication cohort (MZ discordant), overall consisting of 106 different metabolites, mainly lipids, including seventy-three novel associations⁽ Reference Pallister, Jennings and Mohney ^{30 )}. In particular, we identified the amino-acid trans-4-hydroxyproline, a component of collagen, as a novel biomarker for red meat intake; ergothioneine as a potential marker of mushroom intake and two metabolites derived from the gut bacteria transformation of phenolic compounds as biomarkers of fruit intake. All the findings are compiled into the open access DietMetab database (www.twinsuk.ac.uk/dietmetab-data).

Using a similar design and replicating in two independent European populations, we have also identified four novel biomarkers of milk intake: trimethyl-N-aminovalerate, uridine, hydroxysphingomyelin C₁₄ : 1 and diacylphosphotidylcholine C₂₈ : 1⁽ Reference Pallister, Haller and Thorand ^{29 )}.

Microbiome

The study of the gut microbiome is an exciting area of medicine because of its immediate potential for therapeutic interventions⁽ Reference Marchesi, Adams and Fava ^{31 )}. The term microbiome describes the DNA material of microbial communities within an animal and recent advances in technology are revealing its complexity⁽ Reference Lepage, Leclerc and Joossens ^{32 )}. Human subjects have about 100 trillion gut microbes that outnumber their cells ten to one⁽ Reference Turnbaugh, Ley and Hamady ^{33 )}. Human subjects and their microbiome have co-evolved and live intimately in a symbiotic relationship⁽ Reference Ley, Peterson and Gordon ^{34 )} with the microbes producing a wide range of enzymes, chemicals, hormones and vitamins that can potentially interact with the host⁽ Reference Marchesi, Adams and Fava ^{31 )}. The composition of the microbiome varies by anatomical site, with the primary determinant of community composition being anatomical location⁽ Reference Ley, Peterson and Gordon ^{34 )}. Diversity measures the number of microbes present at a particular site^{( 35 )}. The α diversity refers to the number of different species at a particular site⁽ Reference Jurasinski, Retzer and Beierkuhnlein ^{36 )}, while β diversity considers how many taxa are shared between populations, acting as a similarity score⁽ Reference Leprieur, Albouy and De Bortoli ^{37 )}. A low diversity of gut microbes (sometimes called dysbiosis) has been implicated in many human diseases including CVD, obesity, inflammatory bowel disease, colitis and type 2 diabetes⁽ Reference Turnbaugh, Hamady and Yatsunenko ^{38 –} Reference Patterson, Ryan and Cryan ^{46 )}. Only recently, thanks to genetic sequencing have we been able to study microbes properly and realise that the vast majority are not harmful and many are beneficial⁽ Reference Marchesi, Adams and Fava ^{31 )}. Research has shown that the largest influence on the gut microbiome comes from diet⁽ Reference Wu, Chen and Hoffmann ^{47 )} and hence we could potentially alter many diseases with food and monitor the effects via microbes. For instance, dietary fibre is fermented in the intestine by the colonic microbiota resulting in increased production of SCFA⁽ Reference Kasubuchi, Hasegawa and Hiramatsu ^{48 )}. SCFA are involved in the energy metabolism as they regulate the balance between fatty acid synthesis, fatty acid oxidation, and lipolysis in the body. A reduction in the concentrations of free fatty acids in plasma results in a decrease in body weight⁽ Reference Kasubuchi, Hasegawa and Hiramatsu ^{48 )}.

Studies so far on food and the microbiome have mainly focused on meat-based/plant-based diet and highlighted differences in the microbiota of plant- and animal-based diets⁽ Reference Wu, Chen and Hoffmann ^{47 ,} Reference Buddington, Williams and Kostek ^{49 ,} Reference De Filippo, Cavalieri and Di Paola ^{50 –} Reference Hirakawa, Arase and Amakawa ^{52 )}. In particular, they have demonstrated that human gut responds rapidly to major dietary changes⁽ Reference Sonnenburg and Backhed ^{53 )}. A meat-based diet, rich in animal protein, fats, artificial additives, and lacking in fibre, has been shown to drive chronic conditions such as obesity, the metabolic syndrome and atherosclerosis by encouraging gut dysbiosis. Dietary interventional studies, however, have shown that the effects of meat-based diets on the gut microbiome could be rapidly reversed by adopting a plant-based, minimally processed diet⁽ Reference Ward, Ninonuevo and Mills ^{54 ,} Reference Sela and Mills ^{55 )}. Long-term dietary habits, however, tend to determine the composition of the gut microbiota with studies showing that long-term dietary habits associated with the individual microbiome signature⁽ Reference Wu, Chen and Hoffmann ^{47 )}. Also, changes in diet are dependent on the individual gut microbiota composition suggesting that not everyone reacts the same to the same dietary change⁽ Reference David, Maurice and Carmody ^{51 )}.

Moreover, the way in which individuals respond to dietary interventions may be complicated by host genetics and so the twin model could be a valuable tool to study the association of the gut microbiome with nutritional phenotypes. Novel twin data shows that although many microbes are driven by environment, a substantial proportion of microbiota with disease associations have a heritable component⁽ Reference Goodrich, Waters and Poole ^{56 ,} Reference Goodrich, Davenport and Beaumont ^{57 )}. The most heritable taxon, the Christensenella family was shown to be associated with leanness in human subjects and prevents weight gain under high-fat diet in mice⁽ Reference Goodrich, Waters and Poole ^{56 )}.

Using discordant twins, we have also successfully identified a striking negative association between frailty due to ageing and gut microbiota diversity⁽ Reference Jackson, Jeffery and Beaumont ^{58 )} as well as a significant alteration in the gut microbiota in proton pump inhibitors users⁽ Reference Jackson, Goodrich and Maxan ^{59 )}. In general, therefore, a more diverse gut microbiome signals a healthier state and this is modified by age, diet and use of medication.