JXB Advance Access originally published online on December 23, 2004
Journal of Experimental Botany 2005 56(410):219-243; doi:10.1093/jxb/eri069
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
RESEARCH PAPER |
Chemical derivatization and mass spectral libraries in metabolic profiling by GC/MS and LC/MS/MS
1Bourne Laboratory, Centre for Chemical and Bioanalytical Sciences, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK
2Specialist Bioanalytical Services Limited, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK
* To whom correspondence should be addressed. E-mail: j.halket{at}rhul.ac.uk
Received 28 June 2004; Accepted 29 October 2004
 |
Abstract |
|---|
 Top Abstract IntroductionGas chromatography-mass...DiscussionOnline materialReferences |
|---|
An overview is presented of gas chromatography/mass spectrometry (GC/MS) and liquid chromatography/mass spectrometry (LC/MS), the two major hyphenated techniques employed in metabolic profiling that complement direct ‘fingerprinting’ methods such as atmospheric pressure ionization (API) quadrupole time-of-flight MS, API Fourier transform MS, and NMR. In GC/MS, the analytes are normally derivatized prior to analysis in order to reduce their polarity and facilitate chromatographic separation. The electron ionization mass spectra obtained are reproducible and suitable for library matching, mass spectral collections being readily available. In LC/MS, derivatization and library matching are at an early stage of development and mini-reviews are provided. Chemical derivatization can dramatically increase the sensitivity and specificity of LC/MS methods for less polar compounds and provides additional structural information. The potential of derivatization for metabolic profiling in LC/MS is demonstrated by the enhanced analysis of plant extracts, including the potential to measure volatile acids such as formic acid, difficult to achieve by GC/MS. The important role of mass spectral library creation and usage in these techniques is discussed and illustrated by examples.
Key words: Derivatization, electrospray ionization, food analysis, gas-liquid chromatography-mass spectrometry, ion trap, liquid chromatography-mass spectrometry, urine analysis, mass spectral library, metabolic profiling, tandem mass spectrometry, time-of-flight
|
Introduction |
|---|
|
TopAbstractIntroductionGas chromatography-mass...DiscussionOnline materialReferences |
|---|
Metabolic profiling and ‘fingerprinting’
The emerging field of metabolomics requires profiling and fingerprinting methods (Fiehn et al., 2000a
; Glassbrook and Ryals, 2001; Harrigan and Goodacre, 2003; Sumner et al., 2003) capable of measuring the absolute or relative amounts of all metabolites (the metabolome). The great diversity of chemical properties and wide concentration ranges of these compounds pose a significant challenge as the methods need to be robust and reproducible to enable samples to be reliably compared (Glassbrook et al., 2000). Profiling refers to the detailed analysis by hyphenated techniques such as gas chromatography-mass spectrometry (GC/MS), liquid chromatography-mass spectrometry (LC/MS) or capillary electrophoresis-mass spectrometry (CE/MS). Such techniques provide a detailed chromatographic profile of the sample and consequently measurements of the relative or absolute amounts of the components. The number of components measured will depend on the resolution of the chromatographic system and the specificity of the detection technique. A mass spectrometer can function as a highly specific chromatographic detector and a high resolution mass spectrometer even more so.
‘Fingerprinting’ refers to more rapid and general screening methods such as direct infusion atmospheric pressure ionization (API) MS (particularly at high mass resolution), NMR spectroscopy, and other methods such as Raman spectroscopy and Fourier transform infra-red spectroscopy (Fig. 1), all of which provide complementary information. Such techniques can be configured as ‘high-throughput’ and are suitable for determining differences and classifying samples. Generally, fingerprinting involves much larger numbers of measurements than profiling.
|
Targeted analyses
In cases where a relatively small number of analytes can be defined in advance, for example, 20 known amino acids, or a specific set of organic acids, so-called targeted GC/MS and LC/MS methods can be employed. Such methods can achieve great accuracy and precision particularly where stable isotope labelled internal standards are available. In such targeted analyses, signals from all other components are ignored. However, such targeted methods can only touch a tiny fraction of the metabolome. In addition, stable isotope-labelled standards are very expensive and are only available commercially for a relatively small number of common compounds.
Non-targeted analyses
GC/MS can also perform non-targeted analysis where all chromatographic peaks, or all peaks above a certain intensity, can be characterized by their mass spectral patterns and GC retention indices, thereby enabling their reliable recognition in further samples. The number of measurements can be increased by deconvolution of the spectra using numerical methods (Colby, 1992; Dromey et al., 1976; Holland et al., 1997; Stein, 1999). Several software programs are available for this purpose and have been applied to metabolic profiling (Halket et al., 1999; Fiehn, 2003; Waterman et al., 2004; Weckwerth et al., 2004).
LC/MS with atmospheric pressure ionization (API) is also being applied to metabolic profiling, but is again limited by the relatively small numbers of analytes amenable to the analysis. Of particular importance is the development of suitable HPLC columns (Tolstikov and Fiehn, 2002), peak detection software (Tolstikov et al., 2003), high mass resolution (van der Greef et al., 2004), the potential of ultra high resolution LC (Tolley et al., 2001; Legido-Quigley et al., 2002) and chemical derivatization (Gröger et al., 2003). The latter is covered in greater detail below and demonstrates that the approach can dramatically alter the ionization characteristics of different functional groups so that they may be detected more readily in a single analysis.
Capillary electrophoresis and related techniques when combined with mass spectrometry appear to have great potential in metabolic profiling (Soga et al., 2002; Guillo et al., 2004). However, robustness and sample loading problems remain to be overcome.
Profiling methods are more suited to the investigation of only small parts of the metabolome, perhaps the elucidation of a simple pathway. It is estimated that there are maybe 200 000 metabolites, present in plants and, of these, 10 000 are known compounds (Fiehn et al., 2000a). Considering that popular techniques such as GC/MS and LC/MS in their present form can analyse only up to several hundred components with structures assigned to fewer than a hundred (Fiehn, 2003), the so-called ‘fingerprinting’ techniques where the extract is spectroscopically analysed without chromatography may prove to be more useful. In this way, a much ‘bigger picture’ perhaps consisting of thousands of signals can be obtained very quickly albeit at the expense of specificity.
Fingerprinting
Fingerprinting techniques include nuclear magnetic resonance spectroscopy (NMR), Fourier transform infra-red spectroscopy (FTIR), and direct infusion atmospheric pressure ionization MS. The latter, particularly at high mass resolution, has been employed with success to classify a variety of samples (Vaidyanathan et al., 2001, 2002; Goodacre et al., 2002; Kaderbhai et al., 2003; Castrillo et al., 2003; Overy et al., 2004) and has potential for the initial detection of putative biomarkers. The employment of high mass resolution will provide possible molecular formulae for such biomarkers and attempts can then be made to elucidate their structures using more-specific techniques involving chromatography. Fourier transform mass spectrometry can provide even higher mass resolution and has been used to measure the relative amounts of over 6000 metabolites in ripening strawberries using electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) with recording of both positively and negatively-charged ions (Goodenowe, 2001; Aharoni et al., 2002). The development and application of such ultra high mass resolution instruments has a great role to play in the future of metabolic profiling, assuming that current limitations on data processing and interpretation software can be overcome.
An overview of the approximate range of applications of GC/MS (electron ionization) and LC/MS (ESI, APCI) for the analysis of selected compound classes over a range of polarity and relative molecular mass is given in Fig. 2. In GC/MS, the analytes are normally derivatized prior to analysis in order to reduce their polarity and facilitate chromatographic separation on a column of low polarity as usually employed in metabolic profiling. For example, fatty acids are commonly esterified by methylation before GC separation. In the case of ESI, polar and pre-ionized components are favoured making the technique highly complementary to GC/MS. By comparison, the related APCI technique covers lower polarity compounds and therefore has great potential in metabolic profiling by LC/MS. The harsher conditions employed in APCI prevents its use for larger molecules, but smaller analytes (generally <2000 Da) can be detected over a wider polarity range (Ardrey, 2003). The technique is also more tolerant to changes in experimental conditions and is commonly employed in quantitative analysis. Although APCI has a requirement for high flow rates, making it less compatible with microbore column technology, a post-column ‘make-up’ flow of mobile phase can be added. A newer ionization technique, atmospheric pressure photoionization (APPI; Robb et al., 2000) is not indicated, but has potential for the analysis of compounds of lower polarity and thus considerable potential for metabolic profiling by LC/MS. In conventional GC/MS, the electron ionization technique is employed and only the much more abundant positively charged ions are measured. In addition, the energy supplied to induce fragmentation of the parent ion in order to obtain a fragmentation pattern (mass spectrum, or plot of ion abundance versus mass-to-charge ratio) is kept constant (70 eV) thereby enabling reproducible mass spectra to be obtained. In this way, libraries of such spectra can be shared between investigators and several libraries are commercially available (see below). The LC/MS situation is more complex. The atmospheric pressure ionization techniques generally produce pseudo-molecular ions ([M+H]+ or [M-H]–) depending on a number of factors: the chemical properties of the analyte, the polarity of the electrospray voltage, the nature of the matrix and the solvent composition. It is not always a simple matter to predict whether positive or negative ions will be preferentially produced (Cech and Enkei, 2001). For example, the production of negative ions from carboxylic acids can be enhanced by the addition of weak acid, rather than the expected base (Wu et al., 2004). Matrix effects (Matuszewski et al., 2003) can include ionization suppression (King et al., 2000; Choi et al., 2001) and ionization enhancement (Mallet et al., 2004; Liang et al., 2003) caused by the presence of salts and other components being ionized at the same time. The probability of such effects is greater where no chromatographic steps are employed as in fingerprinting by direct infusion ESI (Goodacre et al., 2002) and the analyst must always be aware of the dangers.
|
Thus, there is a much more complex situation in LC/MS as many compounds in a plant extract will not ionize optimally, or at all, under the conditions employed in a single analytical run. Currently, metabolic profiling by LC/MS is more suited to groups of compounds, such as alkaloids, which ionize similarly under given conditions. In complex mixtures such as plant extracts, some compounds will preferentially form positive ions, others negative ions, and some will be difficult to ionize under fixed conditions. Chemical derivatization has the potential to alter the ionization properties of the analyte molecules favourably, as indicated by the arrows in Fig. 2: the polarities of organic acids can be reduced by esterification facilitating their analysis by GC/MS. Similarly, organic acids can be derivatized in such a way that their polarity is increased (arrowed, Fig. 2) making them more amenable to analysis by positive electrospray and examples of such a transformation are described below.
In addition, energy supply to such parent ions in order to induce fragmentation and obtain the product ion mass spectra, which can be used for library searching, is not standardized between instrument types and complicates the analysis. This situation is covered in the LC/MS library section, below.
GC/MS and LC/MS: peak tables
The data transformation required in profiling techniques such as GC/MS and LC/MS is illustrated in Fig. 3. Component peaks (a,b,c...n) located in the mass chromatograms are detected by the manufacturers' software and quantified by the integration of selected ion currents usually relative to one or more internal standards. The integrated values are entered into a peak table for each sample chromatogram (1,2,3...z). Mass spectrometry data systems are generally equipped for ‘target compound analysis’ where preselected components can be recognized by their mass spectral patterns or pattern of qualifier ions (see below) and chromatographic retention data (time, relative retention time or retention index; see below). In this case, the data matrix is easily constructed. The situation with non-targeted analysis is more complex. Components appearing or disappearing in some samples, for example, following genetic modification, or in a disease state, will require manual adjustment of the matrix. Commercially available software to assist with this step is only just becoming available. A selection of programs, some of which can read a variety of file formats and carry out mass spectral deconvolution, are listed in Table 1.
|
|
In comprehensive metabolic profiling of plant extracts by GC/MS, the majority of compounds measured have not yet been formally identified. Over 400 components were detected in Cucurbita maxima phloem, using automated mass spectral deconvolution GC/MS, but only 90 of them were tentatively identified (Fiehn, 2003
). However, the power of mass spectrometry enables relative measurements of compounds to be reliably made, even though their chemical structures are not known. When measuring the relative amounts of such compounds, usually against one or more stable isotope labelled internal standards, it is important to be sure that one is comparing like with like across multiple samples. Mass spectral libraries and GC retention data are, therefore, of paramount importance in such work. Some aspects of metabolic profiling by GC/MS and LC/MS will be considered below, focusing on the development and application of mass spectral libraries, chemical derivatization and chromatographic retention data, all important parameters when carrying out both targeted and non-targeted analyses. In addition, examples of chemical derivatization for LC/MS are shown and potential for metabolic profiling briefly discussed.
|
Gas chromatography-mass spectrometry (GC/MS) |
|---|
|
TopAbstractIntroductionGas chromatography-mass...DiscussionOnline materialReferences |
|---|
GC/MS is now a robust and very widely used technique and combines high sensitivity and specificity for suitable analyte classes (Fig. 2). Several introductory and advanced texts are available (Halket et al., 1990; Kitson et al., 1996
; McMaster and McMaster, 1998; Gerhards et al., 1998; Niessen, 2001). Prior to its application in the plant arena (Roessner et al., 2000), GC/MS has had a fairly long history in metabolic profiling (Horning and Hornung, 1971) through the detailed study of metabolic disorders in humans. Indeed, the technique was instrumental in the identification of a large number of hitherto unknown inborn errors (Witten et al., 1973; Eldjarn et al., 1974; Goodman, 1980; Tanaka et al., 1980; Chalmers et al., 1982). Such metabolite profiling techniques have since been routinely applied to the analysis of plant extracts (Roessner et al., 2000; Fiehn et al., 2000a; Waterman et al., 2004; Shepherd et al., 2004).
Sample extraction
Analytes present in wet or freeze-dried plant tissue are normally extracted with methanol or methanol-water prior to derivatization. Lipophilic components may be partitioned using chloroform (Roessner et al., 2000). Liquid–liquid extraction using ethyl acetate or ethyl acetate–diethyl ether is commonly employed for organic acids in acidified urine. Solid phase extraction (anion exchange) techniques have also been employed (Verhaeghe et al., 1988; Chalmers and Lawson, 1982). A more general profile of urine samples (organic acids, amino acids, sugars) has been carried out by the use of urease followed by trimethylsilylation of the residue after drying (Shoemaker and Elliott, 1991). The urease removes large amounts of urea which can obscure other components and reduce the chromatographic resolution.
GC/MS: chemical derivatization
Some plant metabolites such as those contained within essential oils are highly suited to GC/MS analysis without derivatization (Jennings and Shibamoto, 1980; Adams, 2001). However, chemical derivatization (Knapp, 1979; Blau and King, 1977; Blau and Halket, 1993) is usually required to reduce the polarities of the functional groups, facilitate their separation by GC and can influence their mass spectral properties (Halket, 1993). Organic acids were commonly esterified by reaction with diazomethane, but now silylation (Poole, 1977; Evershed, 1993; Halket and Zaiken, 2003), predominantly trimethylsilylation or tert-butyldimethylsilylation, is more popular. In addition, keto- (oxo-) groups are usually oximated in order to improve their GC properties and prevent enolization reactions which can introduce multiple products, thereby complicating the chromatograms. During high resolution chromatography, the syn- and anti- isomers of the oximes can sometimes partially separate giving rise to recognizable shoulders on the GC peaks. Table 2 shows the common silylation and oximation reactions: trimethylsilyl (TMS) and tert-butyldimethylsilyl (TBDMS) from hydroxyl and primary amine groups and methoximation (MO) of carbonyl groups, together with a selection of mass increments in each case. As the number of derivatized groups increases, there is a danger that the molecular mass of the derivative will be outside the mass range of the detector, typically m/z 650–1000, or be too high that the derivative will not pass through the GC column. In addition, the likelihood of sterically hindered groups can lead to the formation of multiple products, thereby complicating the chromatogram.
|
tert-Butyldimethylsilylation has been evaluated as being the most comprehensive for metabolite profiling applications (Birkemeyer et al., 2003
) and has a further advantage that the derivatives are less sensitive to the hydrolytic effects of moisture than the corresponding TMS derivatives. They often give mass spectra having abundant fragment ions (m/z M-57, where M is the molecular mass of the analyte) from facile loss of the tert-butyl moiety, convenient for structural assignments of plant metabolites (Fiehn et al., 2000b). A disadvantage of TBDMS derivatives lies in the significant increase in molecular mass (Table 2), particularly where multiple derivatizable groups are present in the molecule. In addition, problems of steric hindrance might lead to mixtures of fully and partially derivatized analytes.
The analyst should be aware of possible artefact formation during the trimethylsilylation process. A useful study has been published (Little, 1999) with an internet reference for updates (Little; http://users.chartertn.net/slittle/). In addition, treatment with silylation reagents, even under the mild conditions generally employed in plant metabolite profiling, can lead to conversion. For example, arginine is converted to ornithine by reaction with BSTFA (Leimer et al., 1977). Figure 4 illustrates the same conversion using MSTFA under conditions commonly employed for metabolic profiling of plant extracts (37 °C, 20 min). When studying metabolic processes in detail, particularly where the intermediate compounds may be reactive, or unstable, the analyst should always be aware of such possibilities when interpreting the results. Where there is any doubt, alternative derivatization procedures are generally available for specific functional groups (Blau and King, 1977; Blau and Halket, 1993; Knapp, 1979).
|
GC/MS: retention indices and mass spectral deconvolution
The trimethylsilyl derivatives of many plant components, including sugars such as glucose and fructose, have very similar EI mass spectra. In order to distinguish them, GC retention parameters are particularly important. Since retention times vary with the column length, type of stationary phase and temperature, suitable parameters for comparison include relative retention times and so-called retention indices (RI). Relative retention times are simply the ratios of analyte times to the time of a chosen standard compound. A more universal solution is based upon so-called Kovats indices (Kovats, 1958
) which relate the retention times of the components to the retention times of n-alkanes analysed under the same conditions, even by co-injection. For example, a sugar derivative having the same retention time as the n-alkane C20 would be assigned an index of 2000 or a ‘methylene unit’ value of 20.00. One appearing half-way between C20 and C21, would be assigned an index of 2050, and so on. Most manufacturers' data systems do not handle RI values well, if at all. However, the Automated Mass Spectral Deconvolution and Identification System (AMDIS) (Stein, 1999) from the National Institute of Standards and Technology (NIST) has excellent RI capabilities and is readily available for download (http://chemdata.nist.gov/mass-spc/amdis/). The software can read most manufacturers' data files and perform mass spectral deconvolution in order to ‘clean up’ the mass spectra prior to library searching. In addition, user library creation is simple and spectra can be searched against the NIST database (Table 3, www.hdscience.com). The AMDIS software has been applied to plant (Fiehn, 2003) and urinary metabolites (Halket et al., 1999).
|
GC/MS: mass spectral integrity
In metabolic profiling, mass spectra are scanned during the chromatographic peak elution. Modern quadrupole mass spectrometers can normally record about 1–6 spectra/s. If the scanning speed is too slow, peak ‘skewing’ can occur where the amount of sample from the GC peak changes significantly during the mass spectral scan. Software such as AMDIS can correct for such skewing, but the analyst has to exercise care and ensure that a sufficient number of scans are recorded across the GC peak to secure a sufficient number of samples, perhaps 10, to define the peak and enable measurements of adequate quality to be made. Since the quantitation depends on the areas or heights of selected ion chromatograms, an insufficient sampling rate will increase the error in measurement. Detectors based on time-of-flight MS can operate at much higher scan speeds, up to 500 scans s–1, so that skewing is effectively eliminated (Veriotti and Sacks, 2003
).
Qualifier ions: chromatographic peak integrity
In addition to a matching retention index between the analyte and the known compound, positive assignment will require a mass spectral library match, performed by some manufacturers' data systems (see section below), or at least agreement with the ratios of one or two qualifier ions. Figure 5 shows the GC/MS analysis of cholesterol TMS with three reconstructed ion chromatograms; (a) m/z 458, (b) m/z 443, and (c) m/z 368, together with a part of the mass spectrum recorded at the peak apex (28.0 min). The ratios of the peak areas (a), (b) and (c) should correspond to the ratios of abundances of the ions m/z 458, 443, and 368, respectively, in the library spectrum. The software checks these ratios and reports deviations outside a preset range. The operation is analogous to the wavelength absorbance ratio method in HPLC to detect impure peaks (Sievert and Drouen, 1993).
|
GC/MS: automatic interpretation
Since the start of GC/MS for the diagnosis of metabolic disorders, software approaches to automated data interpretation have been made in individual laboratories (Jellum et al., 1975
; Mizuno et al., 1981; Halket et al., 1999; Yamaguchi et al., 1999), but much remains to be achieved in this area.
GC/MS: mass spectral library searching
A useful introduction to library searching is given in a recent text (Smith and Busch, 1999). In addition, detailed studies of search algorithms have been carried out (Stein and Scott, 1994; McLafferty et al., 1998).
Table 3 lists some common commercially available libraries with an indication of the numbers of spectra available in each case. The mass spectra contained within the NIST library are studied in detail by professional evaluators before inclusion (Ausloos et al., 1999) ensuring high quality.
Of particular interest to metabolic profilers are the useful libraries downloadable from the Max-Planck Institute for Plant Physiology in Golm, Germany (http://www.mpimp-golm.mpg.de/mms-library/). These include TMS and TBDMS derivatives of known compounds and also an important library of unassigned spectra (TMS). Compound listings include retention times. Collaborations are invited for structure elucidation (mms-library{at}mpimp-golm.mpg.de).
The computerized matching of an unknown spectrum with a database is a very rapid and useful tool in metabolic profiling. Despite the advances made in algorithms and software, such automated matching cannot be entirely relied upon (Sparkman, 1996) and data must be checked by an experienced analyst in order to ensure the integrity of the results: a daunting task when one considers the amount of data produced in a short time by modern hyphenated techniques. The additional application of retention indices (see above) can improve the situation when trying to distinguish between compounds having similar mass spectra.
GC/MS: identification of unknowns
There is currently a great need for new methods to speed up the structural elucidation of metabolites. In addition to a library search, a number of steps can be taken to ‘interpret’ the mass spectrum, including accurate mass measurements by high resolution mass spectrometry, study of isotope ratios, study of the neutral losses, MS/MS, etc. Several texts are available outlining such procedures in some detail (McLafferty and Turecek, 1993; Smith and Busch, 1999). Progress has also been made by applying knowledge of isotopic ratios after derivatization and accurate mass measurement (Fiehn et al., 2000b).
GC/MS: the future
Great potential is offered by the high throughput capabilities of GC/TOF/MS. Currently, the available instruments are designed for high scanning speeds (up to 500 scan s–1) or high mass resolution. Higher scanning speeds are already being used for metabolic profiling with mass spectral deconvolution (Weckwerth et al., 2004; Waterman et al., 2003; Shepherd et al., 2004) and has great potential to solve the through-put problem in GC/MS. Also, the so-called ‘ultrafast’ GC, in which very high oven temperature programming rates are used for fast chromatographic separations, can be used with fast scanning TOF mass spectrometers. An example of a 2 min metabolic profile is given in Fig. 6: (a) shows the total ion current chromatograms of a C8–C28 hydrocarbon mixture used for retention index assignments via AMDIS software, (b) a derivatized tomato extract (Waterman et al., 2003), (c) the mass spectrum of the GC peak at 0.94 min, and (d) the match obtained by searching the NIST02 database (Table 3) indicating the identity as phosphoric acid TMS. In this case, a narrow bore (0.1 mm i.d.) column has been used. The column is resistively heated at rates of up to 1200 °C min–1.
|
GCxGC: a revolutionary development
A very simple and revolutionary innovation in GC is so-called ‘GCxGC’ or ‘comprehensive 2-dimensional GC’ in which a non-polar column can be coupled to a shorter polar column. The temperature of the junction between the columns is modulated by a moving heater (Kinghorn and Marriott, 1998a
) or jets of gas (Ledford and Billesbach, 2000; Beens et al., 2001) so that peaks from the first column are continually ‘frozen’ (modulated) and transferred to the second, faster-running column. If the first column can separate 300 peaks and the second column 15 peaks, the total resolving power of the system is 300x15=4500 peaks, representing a staggering increase in analytical power! In addition, the modulation focuses the peaks leading to a significant increase in sensitivity (Kinghorn and Marriott, 1998b). Such systems are available commercially: (pjm{at}rmit.edu.au; www.zoex.com). A desirable instrument would utilize fast GCxGC, high dynamic range and accurate mass measurements and it remains to be seen if such a combination becomes available. However, the data generated in a short time would be overwhelming, considering current computing limitations, except for targeted analyses of a set of analytes small enough to allow review by the analyst.
Liquid chromatography/mass spectrometry (LC/MS)
Many excellent sources of background information on the major ionization techniques, such as ESI and APCI, as well as the powerful hyphenated techniques, LC/MS, LC/MS/MS, and LC/MSn are readily available (Cole, 1997; Niessen, 1999; Pramanik et al., 2002; Willoughby et al., 2002; Ardrey, 2003).
LC/MS: chemical derivatization
Compared with GC/MS, a major advantage of LC/MS (API) has been the avoidance of a requirement to derivatize the samples and the combined technique is already seeing applications in metabolic profiling (Huhman and Sumner, 2002; Tolstikov and Fiehn, 2002; Tolstikov et al., 2003; Stobiecki et al., 2003) with appropriate choices of polarity and mobile phase compositions. As derivatization by definition induces chemical changes and is sometimes performed under harsh conditions with a consequent danger of artefact formation, it is best avoided if at all possible. However, the literature now contains many examples where derivatization can be advantageously employed to enhance the signals obtained from soft ionization MS techniques, such as fast atom bombardment (FAB), thermospray, ESI, and matrix-assisted laser desorption-ionization (MALDI).
The rationale behind such attempts to alter the chemical properties can be seen in Fig. 2. Analytes in the ESI region containing different functional groups can have diverse ionization properties, giving preferentially positive or negative ions depending on the mobile phase composition. For example, the ESI method in positive mode would work particularly well with positively charged (or easily chargeable) species (highly polar) so that a derivatization scheme which introduces such species to a wide variety of compounds would facilitate simultaneous measurement (profiling) of an increased number of compounds in one analysis. The same would apply to negatively charged (or easily chargeable) species.
The early introduction of soft ionization techniques such as fast atom bombardment (Barber et al., 1981) allowed peptides to be analysed directly without the need for the chemical derivatization, hitherto essential for direct probe EI mass spectrometry and GC/MS. Despite this, derivatives have been reported to improve the MS/MS fragmentation patterns and facilitate de novo sequencing (Vath and Biemann, 1990; Stults et al., 1993; Zaia and Biemann, 1995; Huang et al., 1997; Brancia et al., 2000; Lindh et al., 2000). Derivatives for FAB MS have been reviewed (Spreen, 1993). Similarly, the FAB/MS analysis of acylcarnitines and amino acids was improved by n-butylation of their carboxylic acid groups (Millington et al., 1991).
Urinary carboxylic acids have been analysed directly by LC/MS when the thermospray technique (Blakley and Vestal, 1983) was introduced, but the method was not powerful enough to achieve routine application. Alkylaminoethyl derivatization of the carboxylic acid groups improved the thermospray detection limits of prostaglandins and thromboxane B2 by an order of magnitude (Voyksner et al., 1987).
Although atmospheric pressure LC/MS/MS is eminently suited to the profiling of polar compounds without such derivatization (Tolstikov and Fiehn, 2002), applications of derivatization are increasing, usually for specific applications. For example, the clinically important group of acylcarnitines (Rashed et al., 1994) and amino acids (Casetta et al., 2000) are commonly n-butylated prior to direct ESI MS/MS analysis. The ion trap MS/MS spectra of the butyl esters of acylcarnitines yield more fragment ions than the underivatized compounds giving more detailed mass spectral ‘fingerprints’ and thereby increasing their utility for identification (Baumann et al., 2000). Interesting derivatives based on electrochemically-generated ions (Van Berkel et al., 1992) can enhance detection characteristics. Ferrocene-based electrochemically ionizable derivatives facilitate analysis of simple alcohols, sterols and phenols (Van Berkel et al., 1998; Diehl and Karst, 2002) and provide further structural information. Alcohols could be selectively detected by ESI at low levels in a complex fruit extract (Quirke et al., 2000). Fatty acids have been analysed by ESI using alkyldimethylaminoethyl derivatives (Johnson, 2000). Polar derivatives have been applied to steroids (Nakagawa and Hashimoto, 2002; Griffiths et al., 2003), fatty alcohols and alcohol ethoxylate surfactants in wastewater (Dunphy et al., 2001), carbonyl compounds (Brombacher et al., 2002), and drug impurities (Barry et al., 2003b) with improved detection. In MALDI MS, charged tris(2,4,6-trimethoxyphenyl)phosphonium (TMPP) derivatives were used for the derivatization of peptide amino groups (Huang et al., 1997), yielding improved sequence information. Such derivatives have since been applied to amines and carboxylic acids (Leavens et al., 2002), alcohols, aldehydes, and ketones (Barry et al., 2003a), oligosaccharides (Naven et al., 1996), and carboxylic acids (Gröger et al., 2003; Cartwright et al., 2004; Waterman et al., 2004; J Halket et al., unpublished data). The chromatographic properties of the TMPP derivatives of amines have also been investigated (Spikmans et al., 2002) and a preliminary study has been carried out on the utility of TMPP derivatives for metabolic profiling (Gröger et al., 2003; J Halket et al., unpublished data). Also, electron-capturing derivatives have been shown to facilitate greatly the analysis of small molecules at attomole levels (Singh et al., 2000) with application to lipidomics (Lee et al., 2003).
The application of stable-isotope-labelled derivatives (Gygi et al., 1999) is having an enormous impact on protein analysis (isotope coded affinity tags, ICAT), a clear indication of the potential utility of chemical derivatization (tagging) in API MS.
LC/MS: chromatographic retention data
The use of retention parameters in HPLC is complicated by the variety of column stationary phases available and the infinite number of mobile phase combinations which can be used to provide suitable separations. Despite such difficulties, retention parameters have been described and this area will have increasing importance in metabolic profiling by LC/MS. One study used the co-injection of a series of nitroalkanes (Bogusz and Wu, 1991) and another study involved the assignment of hydrophobicity indices to the analytes, based upon separation of a series of compounds of known indices across a water–acetonitrile gradient (Valko et al., 1997). It may be that a calculation approach will find application in this area (Palmblad et al., 2002).
The creation of peak tables (metabolomic data matrices) as illustrated in Fig. 3, and comparison of LC/MS data between laboratories would be greatly facilitated by the availability of a universal index. It is hoped that further research will be carried out in this important area.
LC/MS: can we use mass spectral libraries?
Computerized library searching of electron ionization (EI) mass spectra has revolutionalized the application of GC/MS, and much was expected of LC/MS using the particle beam interface (Willoughby and Browner, 1984) which gives conventional EI spectra. However, the technique was found to lack the sensitivity required for many applications (Wolff et al., 2001). Softer ionization techniques, including atmospheric pressure ionization (API), are much more sensitive than particle beam, but generally produce much simpler spectra than EI.
The EI mass spectrum of the turmeric pigment curcumin (Curcuma longa L.) is shown in Fig. 7A together with the ESI spectrum (Fig. 7B), consisting almost entirely of a simple ‘pseudo’ or ‘quasi’ molecular ion signal ([M+H]+). Such simple mass spectra are less suitable as mass spectral fingerprints for library searching. However, further fragmentation of the pseudo-molecular ion by a process known as collisionally induced dissociation (CID) or collisionally activated dissociation (CAD) usually produces a more detailed product ion mass spectrum, such as that shown in Fig. 7C.
|
Although API/MS/MS techniques have been used for many years, workers have been reluctant to create and search libraries of product ion mass spectra because the mass spectral patterns are less reproducible than electron impact ionization (EI) mass spectra between instruments from different manufacturers. The fragmentation patterns depend on a large number of factors, many of which are not properly understood: ion source designs, ion source potentials, mobile phase effects, etc. However, efforts are being made and several API/MS/MS libraries have been created and applied successfully.
Three common types of product ion mass spectra will be considered here: (i) source-CID or transport region CID in a single quadrupole mass spectrometer; (ii) CID in a triple quadrupole mass spectrometer; and (iii) CID in a quadrupole ion trap mass spectrometer
Source-CID or transport region CID in a single quadrupole mass spectrometer
In single quadrupole instruments, the ions can be energized in the transport region between the ion source and mass spectrometer by increasing the cone voltage and thereby promoting collisions with solvent and gas molecules. There is no ion selection as in tandem mass spectrometry (see below) so that this so-called ‘source CID’ or ‘transport region CID’ is applied to all molecules present in the system. Other names for the technique include orifice or nozzle-skimmer voltage CID. As in EI, the mass spectra produced will be overlapped so that chromatographic separation is desirable (as in GC/MS).
Such transport region CID spectral patterns have been shown to vary greatly with instrumental conditions (Bogusz et al., 1999). Corresponding libraries have only become available following the introduction of tuning compounds, enabling fairly reproducible spectra to be obtained, albeit on instrument types from the same manufacturer (Marquet et al., 2003; Hough et al., 2000; Weinmann et al., 2001). The equipment is relatively inexpensive.
Several libraries have been created and successfully applied, mainly in the forensic toxicology arena (Table 4).
|
MS/MS: CID in a triple quadrupole mass spectrometer
This sophisticated technique represents true tandem mass spectrometry where the different stages of the process are separated in space. Parent ions are selected in the first quadrupole, subjected to CID by collision with gas molecules (argon, nitrogen, air) in a second quadrupole, which functions as a collision cell, and the product ion mass spectrum scanned in the third. The instrumentation is still relatively expensive. Some libraries are listed in Table 5.
|
MS/MS: CID in a quadrupole ion trap mass spectrometer
In the quadrupole ion trap, different stages of MS are carried out in time (Bier and Schwartz, 1997
). In addition to an inherent high sensitivity, multiple levels of MS (MSn) can be carried out by sequentially isolating and fragmenting selected ions in the trap. The ability automatically to switch triggering of MS/MS according to parent ion intensities (data-dependent scanning, (Tiller et al., 1997)) and further automation facilities (Drexler et al., 1998) are powerful features. Longer residence times in the ion trap can lead to more rearrangement ions than with the other techniques described above. The instrumentation is very compact and relatively inexpensive. A selection of ion trap MS/MS libraries is listed in Table 6.
|
The library references made here are by no means complete. Many libraries are proprietary and details are not made public. In addition, the accuracy of the information given in Tables 4–6
could not be verified in each case prior to publication, library creation is hopefully not static! The references are given as a guide or starting point for the reader.
Some examples are now given of the ion trap library being created in the authors' laboratory (Baumann et al., 2000; Eichler et al., 2000; Cintora et al., 2001). The MS/MS library has been compiled using the powerful technique of normalized collision energy (Lopez et al., 1999), which automatically compensates for the mass-dependent energy deposition characteristics typical of ion trap instruments and makes the MS/MS spectra remarkably reproducible and relatively insensitive to instrumental settings.
Rapid confirmation of curcumin in turmeric (Curcuma longa L.) powder
A simple extract of turmeric (Curcuma longa L.) powder (approximately 0.5 mg mixed with 1 ml methanol–water, 1:1, v:v) was infused into an ion trap mass spectrometer at 5 µl min–1. A clear signal was obtained at m/z 369 (data not shown) which could correspond to the [M+H]+ ion of curcumin. The product ion mass spectrum obtained for m/z 369 is shown in Fig. 8A, together with the NIST library match for curcumin (B), a substance exhibiting beneficial antioxidant properties and approved as a colouring material in foodstuffs. The library spectrum had been made several years beforehand on a different LCQ instrument. The whole analysis took only a few minutes.
|
MSn spectra
A great advantage of ion trap mass spectrometers is the ability to carry out multiple levels of MS. In cases where the MS/MS spectrum is dominated by a single ion, this ion can be selected and fragmented further to give an MS/MS/MS (MS3) spectrum.
A simple example is shown in Fig. 9. A solution of the tomato alkaloid tomatine (5 µg ml–1 in methanol–water, 1:1, v:v, containing 0.1% formic acid, 5 µl min–1) was infused into the LCQ XP. The mass spectrum gave a single ion at m/z 1037.5 (data not shown). The MS/MS spectrum obtained by fragmentation of this ion (50% collision energy (Lopez et al., 1999) is shown in Fig. 9A in NIST user library format. In addition to the major fragment ion at m/z 1017, a smaller fragment ion at m/z 416, corresponding to the molecular mass of tomatidine, was selected for a further fragmentation step. The product ion mass spectrum (MS3) obtained is shown in Fig. 9B together with the best library match showing tomatidine contained in a 1000 compound ion trap library (Baumann et al., 2000). The spectra are not identical, giving a NIST match factor (Stein, 1994) of only 625, a consequence of the different energies used to obtain them. However, they are similar enough to indicate the usefulness of the procedure for rapid chemical analysis. An advantage of relatively low cost ion trap mass spectrometers is the capability of carrying out MSn experiments, useful in structure elucidation (Tolstikov and Fiehn, 2002). The m/z selection and ionization steps can be automated to give fragmentation maps and can be repeated up to 10 times, although the sensitivity of the technique reduces after each step. In many cases, MS4 or MS5 will be achievable with biological samples, depending on the analyte concentration.
|
Development of an ion trap library
Recently, spectra are being acquired with a variety of conditions and contributions of spectra are being received from collaborators. The library now contains approximately 1200 spectra, about half of which are drug-related. The library will be distributed free of charge to contributors and other Thermo LCQ users. The spectra will also be included in a large MS/MS library to be distributed by NIST (j.halket{at}rhul.ac.uk).
LC/MS/MS: towards a universal library?
Further to early work on the reproducibility of triple quadrupole CID spectra (Martinez, 1991), the factors affecting the API product ion mass spectra originating from a number of instruments are being actively studied (SE Stein, personal communication). Mass spectra from different instrument types and manufacturers are being compared. Although perfectly matchable spectra from different instruments seems unlikely, early indications are that useful universal libraries may be created (Bristow et al., 2002, 2004; Gergov et al., 2004).
LC/MS/MS of derivatized organic acids
As previously demonstrated, charged tris(2,4,6-trimethoxyphenyl)phosphonium (TMPP) derivatives of organic acids can be prepared using a TMPP propylamine reagent following activation of the carboxylic acid group with 2-chloro-1-methylpyridinium iodide (Leavens et al., 2002; J Halket et al., unpublished data).
In the first part of a comprehensive LC/MS/MS derivatization study, MS/MS spectra of a wide range of organic acid TMPP derivatives have been recorded and stored in a NIST format user library. The derivatization improves the detection characteristics of the carboxylic acids and femtogram sensitivity has been achieved with standards (J Halket et al., unpublished data). Utility of the derivative is illustrated by preliminary examples of metabolic profiling in extracts of tomato tissue. The work represents the first stage in the development of a sequential derivatization and multi-component analysis procedure applicable in the field of metabolic profiling.
The TMPP derivatization step is shown in Fig. 10A for gibberellic acid (Cat+=m/z 918) together with the corresponding MS/MS spectrum (NIST user library format) of the product in Fig. 10B. The mass spectrum has several high molecular mass fragment ions eminently suitable for the recognition and also for the quantitation of this compound.
|
Reconstructed mass chromatograms (C1–C6) for a mixture of volatile carboxylic acids following separation on a short HPLC column are shown in Fig. 11A together with the MS/MS spectrum of the formic acid TMPP derivative (NIST user library format) in Fig. 11B. Using this procedure, compounds covering a wide range of molecular mass can be determined in a single analysis. The determination of low molecular weight acids is difficult to carry out by GC/MS, particularly in the same separation as compounds having much greater molecular mass, such as oleic acid.
|
The degree of separation of the acid TMPP derivatives obtained using such a short HPLC column is remarkable. Formic and acetic acids are nearly resolved and acetic and propionic acids are resolved completely within about 8 min. The new technique should prove to be useful for the qualitative and quantitative analysis of acidic compounds present at very low levels in plant tissues. The ion trap TMPP derivative library now contains nearly 100 product ion mass spectra, including some MS/MS/MS (MS3) spectra.
Metabolic profiling of tomato extracts
A selection of the ion chromatograms present in a total ion current profile of TMPP derivatized tomato extract is shown in Fig. 12 and illustrates how relative integrated peak areas could be utilized in statistical comparisons of biological samples, such as the comparison of GM versus non-GM foodstuffs (Gröger et al., 2003; Waterman et al., 2004). The boxed peak indicated in Fig. 12A for m/z 764 is shown in expanded form in Fig. 12B. The product ion mass spectrum recorded for the small peak with a retention time of 16.2 min is shown in Fig. 12C together with the best library match,