Nuclear magnetic resonance spectroscopy
NMR spectroscopy is rapid and nondestructive, and it has the advantage of being highly reproducible and robust. It is based on the absorption and re-emission of energy by the atom nuclei due to variations in an external magnetic field. Different types of metabolomics data can be generated depending on the targeted atom nuclei. However, in the analysis of biological samples, hydrogen is the most commonly used type of nuclei (
1H–NMR) because of its naturally high abundance in these samples. Other nuclei, such as carbon-13 (
13C–NMR) and phosphorus-31 (
31P NMR), can also be used to provide additional information on specific metabolite types.
31P NMR is useful for studies of cellular energy states in vivo and ex vivo, but a limitation is the overlapping of
31P signals from phosphorylated compounds. NMR spectroscopy is a powerful technology that offers atom-centered information that is crucial for elucidating molecular structures (Emwas et al
2013). The resulting spectral data allow quantification and identification of the metabolites. Peak areas are used for quantification, whereas the spectral patterns permit metabolite identification. The spectral data generated by NMR techniques can be divided into two NMR strategies regarding the frequency axis used. Frequency axes are referenced by the chemical shift expressed in parts per million (ppm). The chemical shift is calculated as the difference between the metabolite resonance frequency and that of a reference substance (Nagana Gowda and Raftery
2017). One-dimensional NMR (1D–NMR) spectra are based on a single frequency axis, where the peaks of each molecule occur within the resonant frequencies of that axis. This method is the most used in high-throughput metabolomics. Two-dimensional NMR (2D–NMR), which is based on two frequency axes, can be used to complement 1D–NMR. Signals are either binned and then analyzed or fitted to patterns of signals corresponding to the metabolites expected to be present in the mixture.
13C NMR signals are better resolved, but they exhibit low sensitivity due to a low natural abundance of
13C (Markley et al
2017). In 2D–NMR, the second dimension allows separation of overlapping spectral peaks and therefore provides additional and orthogonal chemical information on the investigated metabolites within the analyzed matrix (Larive et al
2015). 2D–NMR methods include
1H-
1H COZY (correlated spectroscopy),
1H–
1H TOCSY (total correlation spectroscopy), and
1H–
13C HSQC (heteronuclear single-quantum correlation) (Emwas et al
2013). Of note, nuclei with low natural abundance, including
2H (deuteron),
13C, and
15N, may serve as excellent metabolic tracers (Fan et al
2016). Despite its relatively low sensitivity, often at the μM level, NMR spectroscopy offers many advantages because it allows rigorous quantification of highly abundant metabolites present in biological fluids, cell extracts, and tissues with minimal or no sample preparation (Fan and Lane
2016). NMR spectroscopy is useful for molecules that are difficult to ionize or require derivatization for MS analysis. NMR spectroscopy also allows the identification of isomeric molecules, and it is the gold standard for determining structures of unknown compounds. Using stable isotope labels, NMR spectroscopy can be used for dynamic assessment of compartmentalization of metabolic pathways, such as metabolite transformations and drug metabolism. Finally, intact tissue NMR imaging and spectroscopy are very appealing for in vivo metabolic investigations (Verma et al
2016). The main drawback of NMR methods is its low sensitivity and resolution compared with MS-based methods (Emwas
2015).
Mass spectrometry
Mass spectrometry is an analytical technique that retrieves chemical data from the gas-phase ions produced from a sample. The ions generate different peak patterns that define the fingerprint of the original molecule in the form of a mass-to-charge ratio (
m/z) and a relative intensity of the measured chemical features (e.g., metabolites). The sample is introduced into the mass spectrometer via the sample inlet, an ion source generates gas-phase ions, a mass analyzer separates the ions according to their
m/z, and a detector generates an electric current from the incident ions that is proportional to their abundances (Murray et al
2013). A sample can be directly injected into a mass spectrometer such as in direct infusion mass spectrometry (DIMS) (González-Domínguez et al
2017). The major drawback is the ion suppression effect, which leads to metabolite information loss and prohibits separation of isomers. Mass analyzers can be used alone or in combination with the same type of mass analyzer or with different mass analyzers (hybrid instruments). Such combinations are the foundation for the analytical mode of tandem mass spectrometry (MS/MS). In MS/MS, the ions that arrive at the first mass analyzer (precursor ions) are selected, then fragmented in a collision cell. The fragmented ions are separated according to their
m/z in a second mass analyzer and then detected. Different operation modes are possible, including data dependent analysis (DDA) and data independent analysis (DIA). In DDA, a fixed number of precursor ions whose
m/z values were recorded in a survey scan are selected using predetermined rules and are subjected to a second stage of
mass selection in an
MS/MS analysis (Mann et al
2001). Modes include single reaction monitoring (SRM) or multiple reaction monitoring (MRM), which is the application of SRM with parallel detection of all transitions in a single analysis. In DIA, all precursor ions within a defined
m/z window undergo fragmentation. The analysis is repeated as the high resolution mass spectrometer progresses through the full selected
m/z range (Plumb et al
2006). This process yields accurate metabolite quantification without being limited to profiling predefined metabolites of interest (Zhou et al
2017). One caveat of applying this to metabolomics, is that in a complex sample it may exhibit co-eluting compounds with similar fragments. Therefore, MS/MS acquisition on a wider range of masses may lead to specificity issues related to fragment ions from multiple parent ions. To handle this, the MS
E technique alternates between “high energy” and “low energy” scans on a Q-TOF instrument. MS
E has a fast duty cycle (~0.3 s) which makes the technique compatible with ultra-high performance liquid chromatography (UHPLC). Furthermore, the use of ion mobility separation prior to metabolite fragmentation may improve precursor selectivity. For some mass analyzers, such as quadrupole ion traps, several steps of MS/MS can be performed. For example, the fragmented ions can be further fragmented and detected. The experiment is called multiple-stage mass spectrometry (MS
n, n refers to the number of MS steps). MS/MS and MS
n improve structural identification, combining information from both molecular and fragmented ions generated from precursor ions. The main performance characteristics of mass analyzers are (1) mass accuracy, or mass resolving power, which is related to the ability of an MS analyzer to generate distinct signals for two ions with a small
m/z difference; (2) mass range, which is the range of
m/z over which a mass spectrometer can detect ions to record a mass spectrum; (3) sensitivity; (4) scan speed; and (5) duty cycle time, which is the fraction of ions that effectively reach the detector in the mass spectrometer. The mass analyzer choice is mainly based on the type of metabolomics approach to be carried out, targeted or untargeted. Single quadrupole (Q), triple quadrupole (QqQ), quadrupole ion trap (QIT), and Orbitrap (OT) are suitable for targeted metabolomics because of their sensitivity and duty cycle characteristics. In comparison, dynamic range, mass accuracy, and resolution power are the main characteristics of a mass analyzer to be used in untargeted metabolomics studies. Time of flight (TOF), quadrupole time of flight (QTOF), Fourier transform ion cyclotron resonance (FTICR), and OT are the most used mass analyzers for this purpose. The principle underlying TOF and QToF involves the time required for ions to travel a flight tube. Ions are accelerated in an electric field, reaching a linear velocity that depends on their
m/z ratio. The velocity can reach 10,000 per second scan speed, with a mass error of 5 ppm. A QTOF mass analyzer is a hybrid instrument that can generate high-resolution MS/MS spectra (Forcisi et al
2013). FTICR is an ultra-high-resolution (10
5–10
7 depending on the detection time and magnetic field) mass analyzer that uses cyclotron frequency in a fixed magnetic field to measure
m/z ions at the cost of relatively slow acquisition rates (typically 1 Hz). In the same way, the OT is also an FTMS instrument, which is based on harmonic ion oscillations in an electrostatic field. Ions are trapped around a central electrode, and ion oscillation frequencies are used to measure the
m/z values. The OT provides high mass resolution (>100,000 FWHM), high mass accuracy (2–5 ppm), and an acceptable dynamic range. However, the scan speed is inversely related to mass resolution. Recently, the high-field Orbitrap has provided a resolution above 1,000,000 at
m/z 300–400 with 3 s detection time, using an absorption mode (Denisov et al
2012). A wide range of instrumental and technical variants are currently available for MS spectrometry. These variants are mainly characterized by different ionization and mass selection methods (Glish and Vachet
2003).
Because of the matrix effect limit and potential isomers, MS is generally preceded by a separation step in metabolomics. This step reduces the complexity of a biological sample and allows sequential MS analysis of the different molecules. Different separation methods coupled to MS have been described, such as liquid chromatography (LC-MS) (Want et al
2010; Want et al
2013), gas chromatography (GC-MS) (Chan et al
2011), and capillary electrophoresis (CE-MS) (Ramautar et al
2015). Thus, metabolites with different chemical properties will spend different amounts of time (retention time, t
R) in the separation dimension. These different separation methods enhance the sensitivity and the dynamic range of MS and provide complementary and orthogonal molecular information.
LC-MS is widely used in metabolomics because of its analytical versatility, covering separation performance of different classes of molecules, from very polar to very lipophilic compounds. This high versatility is achieved through the variety of chromatographic columns along with stationary phases (Kuehnbaum and Britz-McKibbin
2013). The LC separation basics depend on physico-chemical properties, such as hydrophobicity, molecular size, and polarity. The separation of compounds occurs in a chromatographic column composed of a stationary phase with polar or lipophilic properties. When polar stationary phase columns are used, the method is referred to as normal-phase liquid chromatography (NPLC); when nonpolar stationary phase columns are used, the method is called reversed-phase liquid chromatography (RPLC). The choice of LC columns depends on the polarity of the metabolites and the analytical scope. To analyze nonpolar and/or weakly polar metabolites, nonpolar C18 and C8 columns are mostly used for untargeted metabolomics (Forcisi et al
2013). However, for hydrophilic, ionic, and polar compounds, hydrophilic interaction liquid chromatography (HILIC) is recommended. HILIC is similar to NPLC, but it differs because of the mobile phase, which is composed of a polar and/or aprotic organic solvent miscible in water that is easier to use with electrospray-mass spectrometry (Tang et al
2014). Recently, Prinsen et al reported a HILIC tandem MS-based method for the analysis of 36 underivatized plasma aminoacids in an 18 min run (Prinsen et al
2016). Sowell et al developed a HILIC tandem MS-based method for the quantification of free and total carnitine avoiding the derivatization step (Sowell et al
2011). For further details on HILIC-based metabolomic strategies, the reader may refer to a recent comprehensive review (Tang et al
2014). Multiple-column strategies could be used for more extensive metabolome coverage (Haggarty and Burgess
2017). Recently, RPLC and HILIC columns with a smaller internal diameter (e.g., 1 mm) and shorter length have drawn interest in metabolomics. These columns allow the use of regular LC flow rates with very high back pressure. Thus, instruments that can operate at very high pressure—ultra-performance liquid chromatography (UHPLC)—coupled to mass spectrometry have been introduced to improve metabolite coverage and detection. UHPLC methods allow increased resolution, better sensitivity, and lower ion suppression. As a result, better metabolome coverage is obtained in comparison with conventional HPLC. Moreover, lower solvent consumption is observed because of the low flow rate (150–250 μL/min) (Kaufmann
2014). It is to be noted that chromatographic conditions are crucial in regarding metabolome coverage and the unbiased proprieties of untargeted metabolomics studies (Boudah et al
2014).
GC-MS is often used for analysis of volatile compounds and molecules with low vapor pressure, such as lipids, long-chain alcohols, amides, alkaloids, sugar alcohols, and organic acids. In addition, using derivative techniques widens the coverage of GC-MS. GC-MS has been accepted as a robust metabolomics platform because of its selective separation, reproducibility, and robustness. The greatest advantage of GC-MS is that its ionization mode is highly reproducible and standardized (based on electron ionization at 70 eV) across GC-MS systems worldwide and across different vendors (Kopka et al
2005), which has allowed comprehensive GC-MS mass spectral libraries such as NIST and FiehnLab to be established (Vinaixa et al
2016). As a result, GC-MS has been a set and reliable platform for MS-based metabolomics. The main limitation of GC-MS is the necessary derivatization step for some metabolite classes. In metabolomics, derivatization usually uses oximation and a silylation/chloroformate reagent. This step is time consuming, hampers the throughput, and can introduce error by adding analytical variability (Moros et al
2017). Moreover, GC-MS metabolome coverage is limited by the stationary phase stability as well as the thermal stability of metabolites and their derivatives (Kaal and Janssen
2008).
Capillary electrophoresis (CE) offers an orthogonal separation mechanism. CE-specific characteristics, such as high efficiency and resolution, high throughput, and, importantly, the ability to assess the most polar compounds without derivatization, have made CE an attractive method for metabolomics (García et al
2016). CE-MS was the last pre-ionization separation technique to be paired with MS in metabolomics. Capillary zone electrophoresis (CZE) is the simplest and most commonly used CE mode because of its principle of separation and its broad application to the analysis of diverse samples, spanning small to large biomolecules. In CZE, analytes are separated according to their intrinsic differential electrophoretic mobility in a capillary filled with separation buffer under the influence of an electric field. The mobilities depend on the ion
m/z and the viscosity of the medium (García et al
2016). The main drawback of CZE is that neutral molecules are not separated. To overcome this disadvantage, other CE modes have been developed, such as micellar electrokinetic chromatography, capillary isotachophoresis, capillary isoelectric focusing based on pH gradient, capillary electrochromatography, capillary gel electrophoresis, and affinity capillary electrophoresis. Because of its simplicity, CZE is the preferred CE mode in metabolomics. Recently, DiBattista et al described an elegant high throughput multiplexed separation platform based on CE-MS combined with temporal signal pattern recognition for screening of different inherited metabolic diseases (IMD). Their result showed comparable performances with flow injection analysis. Furthermore, the authors described new biomarkers for galactosemia screening N-galactated amino acids (DiBattista et al
2017). Despite the recent technical advances of CE-MS, its use in metabolomics is still limited compared with NMR spectroscopy and chromatography-based methods. For more details about CE-MS applications in metabolomics, the reader may refer to a recent review (Rodrigues et al
2017).
Another gas phase separation, ion mobility spectrometry (IMS), (Hill et al
1990), is drawing interest in metabolomics (Dwivedi et al
2010; Wickramasekara et al
2013; Paglia et al
2014; Smolinska et al
2014; Hauschild et al
2015; Maldini et al
2015; Paglia et al
2015). In general, the multidimensional coupling of different separation techniques requires that the resolution obtained from each anterior separation must be largely preserved as the analytes pass to the following dimensions. This preservation is particularly difficult when all analytes travel along the same path during the analysis, as is the case for tempo-dispersive techniques. Thus, the solution is to incrementally increase the sampling frequency of each subsequent time dimension so that multiple measurements are obtained within a fixed time interval. In this way, the arrival time in each anterior dimension can be reassembled based on the integrated signal of subsequent dimensions. This strategy is commonly used when coupling condensed phase separations such as GC, LC, or CE to MS. IMS is an appealing post-ionization separation method that is based on molecular size, shape, and charge. It is typically performed on a millisecond timescale, which can be perfectly nested between chromatography (seconds) and high-resolution MS detection (microseconds) timescales. Hence, coupling IMS with high-resolution mass spectrometry and chromatography (LC-IMS-MS) provides additional analytic selectivity without significantly compromising the speed of MS-based measurements. As a result, the MS dimension affords accurate mass information, while the IMS dimension provides molecular, structural, and conformational information through the determination of the ion collision cross-section (CCS), which is a valuable and predictable chemical descriptor. Indeed, ion mobility spectrometry adds a separation dimension to the hybrid MS instruments, allowing a higher analytical coverage of complex biological mixtures (Fenn and McLean
2008; Fenn et al
2009; Kliman et al
2011; Paglia et al
2014; Tebani et al
2016a,
b). One important feature of IMS is its ability to separate isomers (Domalain et al
2013); the predictability of the CCS and peak width for one isomer mainly depend on ion diffusion (Jeanne Dit Fouque et al
2015; Harper et al
2016; Zhou et al
2016). Furthermore, exploring a multivectorial space containing retention time, accurate mass, and CCS obtained by the combination of multiple separation methods with MS allows valuable measurement integration, which enhances molecular identification and consequently biomarker discovery (May et al
2015; Sherrod and McLean
2015).
Recent introduction of ambient ionization sources has significantly increased the high throughput of global metabolic profiling analysis. These techniques permit direct sampling of complex matrices under ambient conditions, and they include atmospheric solids analysis probe (Twohig et al
2010), desorption electrospray ionization (Eberlin et al
2013; Ferreira et al
2015; Kerian et al
2015), and rapid evaporative ionization MS methods (Balog et al
2013; Balog et al
2015). These techniques can provide real-time, interpretable MS data on biofluids and tissues, in vivo and ex vivo, and they are reshaping high-throughput real-time metabolome analysis in different areas (Arentz et al
2017; Dunham et al
2017). For example, in many surgeries, visually distinguishing between healthy and diseased tissues is often difficult. It requires time-consuming biopsies and immuno-staining procedures to be performed by experienced trained histopathologists during surgery. By eliminating this need for external tissue histotyping, techniques such as the iKnife could open the way to true real-time precision surgery. For more details about the use of ambient MS in clinical diagnosis, refer to a recent and detailed review by Ifa and Eberlin (Ifa and Eberlin
2016). Table
1 presents a comparison between different analytical strategies used in metabolomics.
Table 1
Comparison of main analytical technologies in metabolomics
Nuclear magnetic resonance | 1 Dimension 2 Dimensions | Chemical shift Chemical shift × chemical shift | Uses interaction of spin active nuclei (1H, 13C, 31P) with electromagnetic fields, yielding structural, chemical, and molecular environment information | Nondestructive Highly reproducible Exact quantification possible Minimal sample preparation Molecular dynamic and compartmental information using diffusional methods Relatively high throughput Availability of databases for identification | High instrumentation cost Overlap of metabolites Low sensitivity |
Mass spectrometry | Direct injection (DI-MS) |
m/z
| Uses a nanospray source directly coupled to MS detector. It does not require chromatographic separation. | Very high throughput High sensitivity No cross-sample contamination No column carryover Low-cost analysis Automated analysis Low sample volume requirement Allows MS imaging | Samples not recoverable (destructive) No retention time information, which gives limited specificity Inability to separate isomers Subjected to significant ion suppression phenomenon High ionization discrimination (ESI) |
Liquid chromatography (LC-MS) | Time × m/z
| Uses chromatographic columns that enables liquid phase chromatographic separation of molecules followed by MS detection (suitable for polar to hydrophobic compounds) | Minimal sample preparation (protein precipitation or dilution of biological sample) High-throughput capability UHPLC can be coupled to any type of MS Flexibility in column chemistry widening the range of detectable compounds High sensitivity | Samples not recoverable (destructive) Very polar molecules need specific chromatographic conditions Retention times are highly dependent on exact chromatographic conditions Batch analysis Lack of large metabolite databases High ionization discrimination (ESI) |
Gas chromatography (GC-MS) | Time × m/z
| Uses chromatographic columns that enables gas phase chromatographic separation of molecules followed by MS detection (suited for apolar and volatile compounds) | Structure information obtained through in-source fragmentation Availability of universal databases for identification High sensitivity Reproducible | Samples not recoverable (destructive) Requires more extensive sample preparation Only volatile compounds are detected Polar compounds need derivatization Low ionization discrimination |
Capillary electrophoresis (CE-MS) | Time × m/z
| Uses electrokinetic separation of polar molecules paired with a mass spectrometry detector | Excellent for polar analysis in aqueous samples Measures inorganic and organic anions Low running costs | Samples not recoverable (destructive) Relatively low throughput profiling |
Ion mobility spectrometry (IMS-MS) | Time × m/z
(CCS × m/z)
| Uses a uniform or periodic electric field and a buffer gas to separate ions based on charge, size, and shape paired with mass spectrometry | Very robust and reproducible (ability to determine collision cross-section, which is a robust chemical descriptor) High peak capacity High selectivity Separation of isomeric and isobaric compounds Very high throughput | Samples not recoverable (destructive) CCS and mass are highly correlated parameters, which limits the orthogonality of the method |