Metabolomics, Metabonomics and Metabolite Profiling

The Royal Society of Chemistry

Contents

Mass Spectrometry for Metabolite Identification

YUQIN WANG AND WILLIAM J. GRIFFITHS

The School of Pharmacy, University of London, 29–39 Brunswick Square, London WC1N 1AX, UK

1.1 Introduction

Mass spectrometry (MS) and nuclear magnetic resonance (NMR) constitute the two major pillars upon which the disciplines of metabolomics and metabolite profiling are built. Both these techniques have their advantages and disadvantages, but the fundamental difference in the nature of their spectroscopy means that they provide complementary information to the analytical scientist. NMR spectroscopy is discussed in detail in Chapter 2, while this chapter will concentrate on mass spectrometry and associated methodologies appropriate in metabolomics research. The principles of mass spectrometry will be described and examples of metabolite analysis given. As the range of metabolite structures present in biology (almost) exceeds the imagination, we will take many examples from the class of biomolecules that we have been most intimately involved with, i.e. sterols and steroids. However, many of the references quoted are equally applicable to other area of metabolite profiling and metabolomics.

1.2 Mass Spectrometry

1.2.1 Principles

Simplistically, a mass spectrometer consists of an ion source, a mass analyser, a detector and a data system (Figure 1.1). Sample molecules are admitted to the ion source, where they become ionised. The ions, which are now in the gas phase, are separated according to their mass-to-charge ratio (m/z) in the mass analyser and are then detected. The resulting signals are transmitted to the data system and a plot of ion abundance against m/z corresponds to a mass spectrum. In many cases, a separating inlet device precedes the ion source, so that complex mixtures can be separated prior to admission to the mass spectrometer. Today, the separating inlet device is usually either a capillary gas chromatography (GC) column or a high-performance liquid chromatography (HPLC) column, although capillary electrophoresis and thin-layer chromatography can be interfaced with mass spectrometry.

For metabolite analysis a number of different types of ionisation methods are used to generate gas-phase ions and these include: electron ionisation (EI), chemical ionisation (CI), electrospray (ES), atmospheric pressure chemical ionisation (APCI), atmospheric pressure photoionisation (APPI), and the recently introduced, desorption electrospray ionisation (DESI) technique. Other ionisation techniques used, but to a lesser extent, are liquid secondary ion mass spectrometry (LSIMS) and fast atom bombardment (FAB), or for more specific applications, matrix-assisted laser desorption/ionisation (MALDI) (see Chapter 9) and desorption ionisation on silicon (DIOS). The most widely used ionisation modes are discussed below, as are the chromatographic devices to which they are interfaced.

1.2.2 Ionisation

1.2.2.1 Electron Ionisation (EI) and Chemical Ionisation (CI)

Historically, the most important method of ionisation of small biomolecules (<~500 Da) is EI. The effluent from a GC column is readily transferred to an EI source, thereby allowing the combination of the high separating power of a GC column with mass analysis. EI involves the bombardment of gas-phase sample molecules (M) with high-energy electrons (e-), usually of 70 eV energy; the result is the generation of [M]+• ions which are usually radical cations, and thermal energy free electrons (e-) (eqn 1).

[FORMULA NOT REPRODUCIBLE IN ASCII] (1)

In many cases the molecular ions, [M]+•, are unstable and fragment to generate more stable products (eqn 2).

[FORMULA NOT REPRODUCIBLE IN ASCII] (2)

Fragmentation upon EI can be seen as both advantageous and disadvantageous. On the plus side, the fragmentation pattern resulting from decomposition of a molecular ion can provide structural information, allowing its identification. On the negative side, however, fragmentation may be so extensive that the molecular ion may not be observed, and thus the molecular weight of the compound of interest not determined. EI can be used to generate either positive ions [M]+•or negative ions [M]+-. Negative ions are generated via an electron capture event, which involves the capture of secondary low-energy electrons generated by ionisation of a bath gas (e.g. Ar, N2) (eqn 3).

[FORMULA NOT REPRODUCIBLE IN ASCII] (3a)

[FORMULA NOT REPRODUCIBLE IN ASCII] (3b)

A prerequisite of EI is that the sample to be ionised must be in the gas phase; this is also true for GC and has led to the extensive development of derivatisation chemistry to allow the vaporisation of many small biomolecules without their decomposition.

CI is a close relative of EI. It differs in that analyte ionisation is achieved via proton attachment rather than electron ejection (positive ion). In CI the ion source contains a reagent gas, often methane, which becomes ionised by EI and acts as a proton donor to the analyte (eqn 4).

[FORMULA NOT REPRODUCIBLE IN ASCII] (4a)

[FORMULA NOT REPRODUCIBLE IN ASCII] (4b)

[FORMULA NOT REPRODUCIBLE IN ASCII] (4c)

The resulting ion, [M + H]+, is an even-electron protonated molecule, which is more stable than the equivalent odd-electron molecular ion, [M] +•, formed by EI, and thus fragments to only a minor extent.

Electron-capture negative ionisation (ECNI), also called electron-capture negative chemical ionisation (EC-NCI), exploits the electron capturing properties of groups with high electron affinities (eqn 5). The method often utilises fluori-nated agents in the preparation of volatile derivatives with high electron affinities. For example, trifluoroacetic, pentafluoropropionic or heptafluorobutyric anhydrides can be used to prepare acyl derivatives of amines and hydroxyl groups, perfluorinated alcohols can be used to generate esters of carboxylic acids, while carbonyl groups can be converted to oximes which can then be converted to pentafluorobenzyl oximes or pentafluorobenzylcarboxymethoximes, for example (Scheme 1.1). Ionisation proceeds with the capture of a secondary low-energy electron generated under CI conditions, by the high-electron affinity fluorinated groups. Ionisation may lead to the formation of stable [M]-• ions, or it may be dissociative, depending on the analyte and the derivative used. The major advantage of EC-NCI is that ionisation is specific to compounds containing the electron capturing tag and provides excellent sensitivity in terms of signal to noise ratio when either the stable [M]-• ion or a negatively charged fragment ion is monitored.

[FORMULA NOT REPRODUCIBLE IN ASCII] (5a)

[FORMULA NOT REPRODUCIBLE IN ASCII] (5b)

[FORMULA NOT REPRODUCIBLE IN ASCII] (5c)

It is of historical interest that HPLC has been combined with EI and CI, but as both of these ionisation modes require high vacuum, the necessary removal of HPLC solvent has made this combination difficult. The natural marriage for HPLC is with atmospheric ionisation (API) methods discussed later.

1.2.2.2 Liquid Secondary Ion Mass Spectrometry (LSIMS) and Fast Atom Bombardment (FAB) Ionisation

It can be argued that the introduction of the FAB method of ionisation by Barber and colleagues in 1981 initiated the revolution in biological mass spectrometry.Although rarely used today, FAB was widely used for metabolite analysis throughout the 1980's and into the early 1990's, for example in bile acid and steroid analysis, and protocols developed for metabolite analysis during this era are easily incorporated into analytical procedures using API methods.

FAB is most suitable for the ionisation of polar or ionic biomolecules. FAB ionisation is achieved by the generation of a fast atom beam of neutral atoms (6–8 keV kinetic energy, usually Ar or Xe atoms) in the FAB gun by a process of ionisation, acceleration and neutralisation, which impinges on a viscous solution of sample dissolved in a matrix, usually of glycerol. In the positive-ion mode proton transfer reactions result in the formation of protonated molecules, [M + H]+, while in the negative-ion mode deprotonated molecules are formed, [M - H]-. Usually, both protonated and deprotonated molecules are stable, and little fragmentation occurs in the ion source. LSIMS is very similar to FAB; however, a beam of Cs+ ions (20-30 keV), rather than a beam of neutral atoms, is used to bombard the matrix. LSIMS spectra are essentially identical to those generated by FAB, and in this chapter, for simplicity, both ionisation modes will be referred to as FAB.

Negative-ion FAB was found to be particularly suitable for the ionisation of bile acids and steroid sulphates, alleviating the need for hydrolysis, solvolysis and derivatisation reactions necessary for GC-MS analysis. Urine and plasma samples can be analysed by FAB following a simple C18 solid-phase extraction step, allowing the rapid diagnosis of certain liver diseases, e.g. cerebrotendinous xanthomatosis (CTX), and 7α-hydroxylase deficiency. FAB is a vacuum ionisation method and is not suitable for combination with regular HPLC. However, capillary column HPLC (250 µm i.d., 2µL/min flow-rate) has been successfully interfaced with FAB.

1.2.2.3 Atmospheric Pressure Ionisation (API)

(a) Electrospray (ES). Like FAB, ES is suitable for the analysis of polar and ionic biomolecules. ES occurs at atmospheric pressure and is readily coupled with HPLC. Fenn and colleagues were the first practitioners of ES mass spectrometry and by the turn of the last century, ES had almost completely replaced FAB for the analysis of polar and ionic metabolites.

In ES, the analyte is dissolved in a solvent (very often methanol, ethanol, aqueous methanol or ethanol, a mixture of acetonitrile and water, or of chloroform and alcohol), and sprayed from a metal or fused silica capillary (needle) of 20-100 µm i.d. at a flow rate of 1-500 µL/min (Figures 1.2 and 1.3). An electrospray is achieved by raising the potential on the spray capillary to ~4 kV (+ 4 kV in the positive-ion mode, and -4 kV in the negative-ion mode) and applying a back pressure to the contents of the capillary (e.g. via a syringe pump or HPLC pump). The resulting spray of charged droplets is directed toward a counter electrode which is at a lower electrical potential (Figure 1.2). As the spray of fine droplets travels towards the counter electrode, the droplets lose solvent, shrink and break up into smaller droplets. The small offspring droplets are derived from the surface of their predecessors, which contain the highest concentration of charge, and hence the offspring droplets are generated with an enhanced charge-to-mass ratio. Eventually, the droplets become so small that the charge density on the droplets exceeds the surface tension and gas-phase ions are desorbed (ion evaporation model), or alternatively very small droplets containing a single charged species completely lose solvent leaving the residual charged species free (charge residue model). Surface active compounds tend to be enhanced in the small droplets, and hence are preferentially brought into the gas phase. The counter electrode contains a circular orifice through which ions are transmitted into the vacuum chamber of the mass spectrometer. By traversing differentially pumped regions via skimmer lenses, the ions are transmitted to the high vacuum region of the mass spectrometer for subsequent analysis (Figure 1.2). Early ES experiments were performed at flow rates of 5-50µL/min, compatible with microbore HPLC columns, although today ES interfaces are compatible with 4.2 mm columns operating at 500 µL/min flow rates. It should, however, be remembered that ES is a concentration rather than a mass dependent process, which effectively means that maximum sensitivity is achieved with high concentration, low volume samples, analysed at low flow rates (e.g. 1 pmol/µL, 1 µL, 1 µL/min) rather with dilute, high volume samples, analysed at high flow rates (cf. 0.02 pmol/µL, 50 µL, 50 µL/min). So, theoretically, a combination of low flow rate HPLC with ES should provide better sensitivity than conventional flow HPLC with ES. This has been particularly exploited by the proteomics community, who combine low flow rate HPLC with micro-ES or nano-ES. For a more detailed discussion on the electrospray process the reader is directed to the excellent volume edited by Cole, and series of articles published in the Journal of Mass Spectrometry and Mass Spectrometry Reviews.

Today, the terms "micro-ES" and "nano-ES" are interchangeable, although the terms were originally coined to discriminate between two slightly different forms of ES. The term micro-ES was initially coined by Emmett and Caprioli and used to refer to a miniaturised form of pressure-driven ES (i.e. pumped flow) operated at sub µL/min flow rate. Alternatively, nano-ES was invented by Wilm and Mann and differs from micro-ES in that it is a pure form of ES, where sample flow is at nL/min rates and initiated by the electrical potential between the capillary tip and counter electrode (Figure 1.2), rather than being pressure driven. We prefer the term low flow rate ES to either micro-ES or nano-ES, and use it here to refer to ES operated at flow rates of 1 µL/min and below.

(b) Thermospray (TS). TS mass spectrometry predates ES. TS, like ES, is a technique which involves the spraying of analyte dissolved in solvent from a capillary into an ion source at atmospheric pressure. But TS differs from ES in that vaporisation and ionisattion are a result of thermall heating the spray, as opposed to raising the sprayer to a high potential. TS performs best with buffered aqueous mobile phase (0.1-0.01 M ammonium acetate), and droplets become charged by an uneven distribution of cations and anions between droplets, with the result that gaseous [M + H]+, [M + NH4] or [M + Na]+ ions are formed in the positive-ion mode in an ion evaporation process similar to that occurring in ES (ion evaporation). In the negative-ion mode [M - H]- ions are formed, and this is the ionisation mode of choice for the analysis of acidic metabolites. TS can be operated in two modes: "filament-on" mode or 'filament-off' mode (direct ion evaporation). By the incorporation of a filament in the ion source, analytes which are not readily ionised by direct ion evaporation can be ionised via a chemical ionisation process in the "filament-on" mode. Spectra of biomolecules, e.g. steroids and bile acids, generated by ES and TS are similar; however, dehydration of the protonated or deprotonated molecule tends to occur to a much greater extent in TS than in ES.

(c) Atmospheric Pressure Chemical Ionisation (APCI). APCI is a technique that has become popular for the ionisation of neutral biomolecules. It is very similar to TS described above. Analyte dissolved in solvent is sprayed into an atmospheric pressure ion source. Vaporisation of sample and solvent is achieved by the application of heat, and ionisation of analyte is achieved by a chemical ionisation event. The APCI source differs from the ES source in that it additionally contains a corona discharge needle. Analyte ionisation can be achieved by two processes.

(i) The first is a primary CI process. The nebulised spray results in small droplets of differing charge formed as a result of statistical random sampling of buffer ions (cf. TS). The charged droplets will shrink, as in the ES process, with the eventual formation of gas phase buffer ions, analyte molecules (M) and solvent molecules (Sv). If ammonium acetate [NH4]+ is the buffer gas phase, [NH4] and [CH3CO2]- ions will be generated. The buffer gas ions will be free to react with analyte molecules in a CI event to generate analyte ions (eqns 6 and 7).

[FORMULA NOT REPRODUCIBLE IN ASCII] (6a)

[FORMULA NOT REPRODUCIBLE IN ASCII] (6b)

[FORMULA NOT REPRODUCIBLE IN ASCII] (6c)

[FORMULA NOT REPRODUCIBLE IN ASCII] (6d)

In the positive-ion mode the exact products of the CI event will depend on the gas-phase basicity of the analyte, solvent and buffer. Adduct ions can also be formed (e.g. [M + NH4] + and [M + CH3CN + H]+ in aqueous acetonitrile buffered with ammonium acetate). When the ion source is operated in the negative-ion mode, the products of the CI event will depend on the gas-phase acidity of the sprayed components.

[FORMULA NOT REPRODUCIBLE IN ASCII] (7a)

[FORMULA NOT REPRODUCIBLE IN ASCII] (7b)

(ii) APCI can also be achieved in a secondary process, in which electrons from the corona discharge ionise nitrogen gas in the APCI source, leading to the eventual CI of the analyte. In the positive-ion mode, again the eventual products depend on the proton affinity of the components (eqn 8), while in the negative-ion mode the gas phase acidity of the components will define which deprotonated molecules are generated.