Human metabolomics: the contributing factors
The prized goal of personalised medicine, although distant, appears to be getting closer with a multitude of studies being published on the human metabolome. Several different techniques, such as NMR, GC/MS and LC/MS, have been able to provide targeted analysis of hundreds of biomolecules, forming a platform for a personalised approach.
This type of study must take into account the range of metabolome influences that might contribute to human health. So, it is not only endogenous metabolites that are of interest, but also external compounds derived from areas such as the diet and the environment. These compounds that might have an impact on the disease risk of an individual have been labelled the exposome.
This expanded range of compounds, endogenous and exogenous, sets a stiff challenge for the analyst. NMR has reasonable sample throughput but limited sensitivity. Mass spectrometry provides increased sensitivity but needs to be linked to a separation technique such as chromatography or capillary electrophoresis.
The most common technique is LC/MS/MS using ion fragmentation for compound identification and isotope dilution for measurement but it is not ideal. The sensitivity decreases as the number of ionised compounds increases and there can be complex mixtures of product ions to interpret. Isotope dilution requires isotope-labelled standards of all the analytes, which might not be readily available and is of no use for unexpected or uncharacterised compounds.
Increased power in metabolomics using high resolution mass spectrometry
One potential alternative to these techniques is high resolution mass spectrometry, which has been investigated by a team of scientists in the USA. Dean Jones, Jennifer Johnson, Tianwei Yu and Frederick Strobel from Emory University, Atlanta, GA, recognised that these instruments can predict the elemental composition of small molecules without the need for fragmentation. This has the added advantage that co-eluting compounds could be differentiated, thereby reducing the separation requirements.
With a Fourier transform ion cyclotron resonance (FTICR) mass spectrometer operating at high resolution, up to 20,000 distinct components can be identified without prior separation. Add a chromatograph to the front end and the power increases.
Jones and his team developed a process for examining the metabolome of human plasma which involved the minimum of sample preparation followed by automated analysis. Plasma from four volunteers was treated with acetonitrile to precipitate the proteins and the centrifuged supernatant was analysed without further treatment.
The extracts were analysed by LC/MS, with an anion exchange column and a rapid gradient of formic acid in aqueous acetonitrile. The column was fitted with a short precolumn for desalting and to optimise separation. The eluting compounds were fed to a hybrid linear ion trap-FTICR mass spectrometer operating in electrospray ionisation mode and at high resolution and data were collected over the whole 10 minutes of the chromatographic run.
The mass spectrometer is optimised to scan over a 10-fold range of m/z values. In this case, the team selected m/z 85-850, because there was significant interference from a solvent peak at m/z 82 which precluded use of the range m/z 75-750.
The chosen range omitted glycine and other small metabolites in the range m/z 75-84 but still included 91% of the metabolites compiled in a list of known human metabolites plus 1769 dipeptides and tripeptides.
Individual metabolomes revealed: extracting and comparing the data
A principal components analysis of the data from repeated runs revealed that 4-5 injections of sample were required before reproducibility was achieved. This was necessary to equilibrate the column with the plasma matrix and the chromatographic gradient conditions. After this preliminary period, the reproducibility held for up to 80 runs.
The data files were processed automatically using a software package developed in the authors' lab. It performed self-adjusting peak identification and quantification to extract the m/z peak value and intensity and the peak retention time.
Each 10-minute run revealed more than 2000 known and unknown compounds. The profiles of the four individuals were compared and there was a significant overlap in the common metabolites present, as expected.
About 62% of the features were common to all of the volunteers but 10% were unique to an individual. This latter figure is probably an overestimate because of the small sample size. Analysis of plasma from more people should produce more commonality.
Nevertheless, a statistical analysis of microarrays procedure revealed that 770 different compounds were significant for distinguishing the four people.
A comparison of the m/z values against the compiled list succeeded in identifying only 209 compounds (10%). This was improved to 445 compounds (22%) by using the Madison Metabolomics Consortium Database which contains plant metabolites, drugs and common environmental metabolites. So, a large number of compounds in human plasma remained unidentified.
This confirms the need to measure compounds in biological systems regardless of their origins, which would lead to new links between chemicals and disease.
The high-resolution mass spectrometric method described here has the capability to perform such studies by measuring the m/z values of more than 2000 species in a relatively short time with good reproducibility from a small volume of human plasma.
The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.
|
