Metabolomics (2015) 11:9–26 DOI 10.1007/s11306-014-0707-1 ORIGINAL ARTICLE Molecular phenotyping of a UK population: defining the human serum metabolome Warwick B. Dunn • Wanchang Lin • David Broadhurst • Paul Begley • Marie Brown • Eva Zelena • Andrew A. Vaughan • Antony Halsall • Nadine Harding • Joshua D. Knowles • Sue Francis-McIntyre • Andy Tseng • David I. Ellis • Steve O’Hagan • Gill Aarons • Boben Benjamin • Stephen Chew-Graham • Carly Moseley • Paula Potter • Catherine L. Winder • Catherine Potts • Paula Thornton • Catriona McWhirter • Mohammed Zubair • Martin Pan • Alistair Burns • J. Kennedy Cruickshank • Gordon C. Jayson • Nitin Purandare • Frederick C. W. Wu • Joe D. Finn • John N. Haselden • Andrew W. Nicholls • Ian D. Wilson • Royston Goodacre • Douglas B. Kell Received: 9 May 2014 / Accepted: 9 July 2014 / Published online: 25 July 2014 Ó The Author(s) 2014. This article is published with open access at Springerlink.com Abstract Phenotyping of 1,200 ‘healthy’ adults from the concentrations between the 1,200 subjects ranged from less UK has been performed through the investigation of than 5 % to more than 200 %. Variations in metabolites diverse classes of hydrophilic and lipophilic metabolites could be related to differences in gender, age, BMI, blood present in serum by applying a series of chromatography– pressure, and smoking. Investigations suggest that a sample mass spectrometry platforms. These data were made robust size of 600 subjects is both necessary and sufficient for to instrumental drift by numerical correction; this was robust analysis of these data. Overall, this is a large scale prerequisite to allow detection of subtle metabolic differ- and non-targeted chromatographic MS-based metabolo- ences. The variation in observed metabolite relative mics study, using samples from over 1,000 individuals, to provide a comprehensive measurement of their serum metabolomes. This work provides an important baseline or Nitin Purandare is now deceased. reference dataset for understanding the ‘normal’ relative Electronic supplementary material The online version of this concentrations and variation in the human serum metabo- article (doi:10.1007/s11306-014-0707-1) contains supplementary lome. These may be related to our increasing knowledge of material, which is available to authorized users. W. B. Dunn Á W. Lin Á D. Broadhurst Á P. Begley Á M. Brown Á W. B. Dunn Á W. Lin Á P. Begley Á M. Brown E. Zelena Á A. A. Vaughan Á A. Halsall Á N. Harding Á Centre for Advanced Discovery and Experimental Therapeutics S. Francis-McIntyre Á A. Tseng Á D. I. Ellis Á S. O’Hagan Á (CADET), Central Manchester University Hospitals NHS C. L. Winder Á R. Goodacre (&) Á D. B. Kell (&) Foundation Trust, Manchester Academic Health Sciences Faculty of Engineering and Physical Sciences, School Centre, Manchester M13 9WL, UK of Chemistry, Manchester Institute of Biotechnology, The University of Manchester, Manchester M1 7DN, UK W. B. Dunn (&) e-mail: [email protected] School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK D. B. Kell e-mail: [email protected] e-mail: [email protected] D. Broadhurst W. B. Dunn Á C. L. Winder Á R. Goodacre Á D. B. Kell Department of Medicine, Katz Group Centre for Pharmacy & Faculty of Engineering & Physical Sciences, Manchester Centre Health, University of Alberta, Edmonton, AB T6G 2R3, Canada for Integrative Systems Biology, Manchester Institute of Biotechnology, The University of Manchester, J. D. Knowles Manchester M1 7DN, UK Faculty of Engineering & Physical Sciences, School of Computer Science, The University of Manchester, W. B. Dunn Á W. Lin Á P. Begley Á M. Brown Manchester M13 9PL, UK Faculty of Medical and Human Sciences, Centre for Endocrinology and Diabetes, Institute of Human Development, The University of Manchester, Manchester, UK 123 10 W. B. Dunn et al. the human metabolic network map. Information on the variations in metabotype may be applied in disease risk Husermet study is available at http://www.husermet.org/. prediction and diagnosis, in understanding molecular Importantly, all of the data are made freely available at pathophysiology, in interpreting the influence of our MetaboLights (http://www.ebi.ac.uk/metabolights/). environment and lifestyle and in the development and assessment of drug efficacy, toxicity and adverse drug Keywords Human serum Á Metabolic phenotyping Á UK reactions. Metabolomics thus has an important role to play population Á Mass spectrometry Á Clinical biochemistry in personalized and stratified medicine (Nicholson et al. 2012; van der Greef et al. 2006). Both genetics and the environment contribute signifi- 1 Introduction cantly to human function and phenotype. Recent studies have sought to relate the influence of the genetic fingerprint The biochemical composition of human cells, tissues and on metabolism, including through the application of gen- biofluids is highly complex, and their integrative and ome-wide association (GWAS)-metabolomics studies dynamic interactions (termed the interactome) defines (Suhre and Gieger 2012; Suhre et al. 2011). These and function and phenotype (Vidal et al. 2011). Of these bio- other studies have shown the importance of applying chemicals, small molecule metabolites are involved in metabolomics, alone or as part of integrated multi-omic many important processes, from acting as the building studies to investigate human phenotypes. The use of 1H blocks for larger biochemicals and structures, in regulation NMR spectroscopy to analyse urine samples, collected in of biochemical processes, and within metabolism to gen- large scale epidemiological studies, has revealed interest- erate essential cellular components (Dunn et al. 2011). The ing trends between populations and provided new bio- quantitative collection of metabolites in a biological system markers, related for example to blood pressure differences is defined as the metabolome (Oliver et al. 1998), with between individuals and populations (Holmes et al. 2008; sample-specific metabolomes differing in composition both Yap et al. 2010). However, whilst robust and precise, 1H qualitatively and quantitatively. For fundamental reasons, NMR spectroscopy does not access the whole metabolome the metabolome is expected [e.g. Kell (2004, 2006a, b), and the use of other metabolite profiling technologies such Kell and Westerhoff (1986)] and is indeed found (Ra- as gas chromatography–mass spectrometry (GC–MS) and amsdonk et al. 2001), to amplify changes observed in the ultra performance liquid chromatography–mass spectrom- transcriptome and proteome. The holistic study of the etry (UPLC–MS) offer excellent opportunities for quantitative complement of metabolites in humans pro- expanding metabolome coverage due to the prior chro- vides a sensitive and dynamic snapshot of the human matographic separation of the many thousands of small metabolic phenotype (Dunn et al. 2011) [also referred to as molecules estimated via analysis of the human metabolic the metabotype (Gavaghan et al. 2000)]. Knowledge of network (Kell and Goodacre 2014; Thiele et al. 2013)tobe G. Aarons J. K. Cruickshank Á N. Purandare Faculty of Medical & Human Sciences, Institute of Inflammation Faculty of Medical & Human Sciences, Institute of and Repair, Salford Royal Dermatopharmacology, The Cardiovascular Sciences, The University of Manchester, University of Manchester, Manchester M6 8HD, UK Manchester M13 9PT, UK B. Benjamin Á S. Chew-Graham Á M. Zubair Á A. Burns J. K. Cruickshank Clinical & Cognitive Neurosciences, Faculty of Medical & St Thomas’ Hospital, King’s College, University of London & Human Sciences, Institute of Brain, Behaviour & Mental Health, King’s Health Partners, London SE1 9NH, UK The University of Manchester, Manchester M13 9PT, UK J. N. Haselden Á A. W. Nicholls C. Moseley Á P. Potter Á P. Thornton Á G. C. Jayson Investigative Preclinical Toxicology, GlaxoSmithKline, David Faculty of Medical and Human Sciences, Institute of Cancer Jack Centre for Research and Development, Park Road, Ware, Sciences, Christie Hospital, The University of Manchester, Hertfordshire SG12 0DP, UK Manchester M20 4BX, UK I. D. Wilson C. Potts Á C. McWhirter Á F. C. W. Wu Á J. D. Finn Faculty of Medicine, Department of Surgery & Cancer, Sir Faculty of Medical & Human Sciences, Institute of Human Alexander Fleming Building, Imperial College London, Development, The University of Manchester, London, SW7 2AZ, UK Manchester M13 9WL, UK I. D. Wilson M. Pan DMPK Innovative Medicines, AstraZeneca, Neurology and Gastrointestinal Centre of Excellence for Drug Cheshire SK10 4TF, UK Discovery, GlaxoSmithKline, Third Avenue, Harlow, Essex CM19 5AW, UK 123 Molecular phenotyping of a UK population 11 in the human metabolome, followed by sensitive MS-based composed of a single serum sample from 120 subjects and detection. A small scale study to characterize the human analysed across a five-day period. Each batch included the serum metabolome has been performed in \150 subjects periodic analysis of a pooled quality control (QC) sample) (including quantification of a subset of metabolites). This to allow analytical variation to be measured quantitatively study, which employed multiple analytical platforms within and between these analytical experiments (Dunn highlighted the importance of this strategy to broaden the et al. 2012). The same pooled QC sample was applied for coverage of the metabolome and provided the first exper- all analytical experimental runs. imentally-derived serum metabolome database (Psychogios