Review
For reprint orders, please contact [email protected]
Multidimensional gas chromatography methods for bioanalytical research
Multidimensional gas chromatography (MDGC) methods are high-resolution volatile chemical separation techniques, and comprise classical heart-cutting MDGC and its more recent incarnation, comprehensive 2D GC. Although available for a long period, MDGC approaches are still not widely practiced in the field of bioanalysis, possibly reflecting the general preference for regular GC versus MDGC approaches. With the recent introduction of ‘-omic’ techniques that emphasize global nontargeted profiling of metabolites within living systems, it is evident that MDGC is gaining momentum as a separation tool, since it offers very high resolution. By untangling metabolites within highly complex biological matrices, and expanding the metabolic coverage, MDGC plays a frontline role in ‘-omics’ based studies. This review highlights state-ofthe-art MDGC approaches, and summarizes the recent developments in bioanalytics.
Why GC for bioanalysis? Bioanalysis, the characterization and/or quantitative measurement of small and large molecules (e.g., drugs, metabolites, proteins and biomarkers) in biological systems, is conceptually simple but technically challenging [1] . In bioanalysis, the first and most important task normally requires isolation of the ‘target analytes’ from the complex distribution of biological matrices, followed by the appropriate measurement of these entities. In many cases, conventional technologies (for instance, single-dimensional [1D] chromatography) pose limitations for untangling constituents in real-world biological samples, due to sample complexity. Gas chromatography (GC), high performance liquid chromatography (HPLC) and capillary electrophoresis (CE) are routinely used in a wide variety of biological analyses for molecular measurement. These methods tend to rely upon 1D separation, utilizing a single, uniform separation mechanism in a static or varying physical or chemical field. In reality, many biological samples are too complex for conventional 1D separations, and the available peak capacity of the separation system (essentially the number of components
10.4155/BIO.14.186 © 2014 Future Science Ltd
that can be separated by the system) is often far exceeded by the number of sample components. These results in peak overlap, which may not be a problem in cases where a target component can be reliably identified and quantified in presence of coeluting species. However, a decrease in the quality of analysis or a reduction in analytical precision may often produce uncertainty in the identity of a compound. The need for capacity increase in turn drives the development of more advanced analytical technologies in the field of instrumental analytical chemistry, aiming at expanding the separation space (i.e., by using multiple analytical dimensions of chromatography and/or MS). Without the ability to provide precise identification, biomarkers and biological profiles have limited meaning; in other words, they cannot be fully explored. This review focuses on GC with particular reference to multidimensional GC (MDGC) as an integrated part of a technological solution to study biological components derived from humans. The first part presents an introduction to multidimensionality in GC, with discussions on fundamentals, technical implementation, detection options, capabilities and limitations of these technologies.
Bioanalysis (2014) 6(18), 2461–2479
Yong Foo Wong1, Constanze Hartmann2,3 & Philip J Marriott*,1 Australian Centre for Research on Separation Science, School of Chemistry, Monash University, Victoria, Australia 2 Division of Food Chemistry, Department of Chemistry & Pharmacy, Friedrich-Alexander-University Erlangen-Nuernberg, Erlangen, Germany 3 Fraunhofer Institute for Process Engineering & Packaging (IVV), Freising, Germany *Author for correspondence: Tel.: +61 3 9905 9630 Fax: +61 3 9905 8501 [email protected] 1
part of
ISSN 1757-6180
2461
Review Wong, Hartmann & Marriott The second part provides a comprehensive review on bioanalysis applications using MDGC techniques. The potential of MDGC techniques for the analysis of metabolites derived from humans will be highlighted. Attempts to untangle the components of interest from highly complex biological matrices through the use of MDGC techniques will be discussed. Why MDGC for bioanalysis? An introduction While the principles of GC need not be described in detail here, a brief comment on the suitability of GC for bioanalysis is warranted. Although GC is regarded as one of the most powerful methods of separation, the first and most important requirement is that the compounds of interest should be volatile (derivatization where applicable may be used to increase the volatility of analytes) and thermally stable. In some cases, derivatization is mandatory to overcome adsorption of polar functional groups on the active surfaces of the injector and at column walls, which would result in distorted peak shape and/or chemical losses. It should be noted that derivatization (alkylation, acylation or silylation reactions) may result in more than one peak for a compound of interest, and derivatization time and temperature can affect quantitative formation of the derivatives. Hence, careful tuning or optimization of some of these parameters is required in order to obtain satisfactory derivatization of the target analytes. The selection of derivatization agents is highly dependent on the classes of analyte and it is users’ experience to opt for the best one that suits their case. A more comprehensive review on derivatization techniques can be obtained elsewhere and it is not the scope of this review to reiterate this further [2,3] . Generally, GC has been well recognized as capable of excellent reproducibility, sensitivity and resolution and is capable of analyzing gases, liquids and solids (after dissolution). The principal advantage of GC in terms of efficiency is undeniable; an outstanding (but unrealistic) plate number of 1.2 million using a 1300-m fused silica column has Key terms Biomarkers: Objectively measured characteristic compounds that might be associated with a specific biological condition or state. Heart-cutting multidimensional GC: The entire sample is separated in a first-dimension column, and one or more chosen elution regions of first-dimension effluent are transferred into a second-dimension column for further separation. Comprehensive 2D GC: Complete transfer of sample effluent from the first-dimension column to the seconddimension column via a modulation device and under a defined modulation period.
2462
Bioanalysis (2014) 6(18)
been reported [4] . A typical GC system is applicable to a molecular mass range from 2 to about 1500 Da [4] , which is well suited for the analysis of metabolites (lowmolecular-weight organic or inorganic compounds [5,6] with a mass of less than 1500 Da [7]). GC coupled to MS (GC–MS) is routine in forensic, toxicology and doping control, and specific approaches must be implemented to resolve overlapping components. The World AntiDoping Agency (WADA) has recommended GC–MS as the primary method for the preliminary screening of anabolic agents, stimulants and opiates [8] . However, a major concern is the incapability of GC–MS to distinguish coeluting components with similar structure and/or MS spectra, for instance positional isomers. Isomeric forms among various chemical classes of components within complex biological matrices means that peak capacity is often exceeded and peak overlap will be the general expectation. Poor separation of the analytes makes identification and quantification difficult or uncertain. For accurate and reliable identification, each peak should be well resolved and should appear as a single component. This would avoid ambiguities resulting from contributions of coeluting components to an unresolved peak. Heart-cut MDGC: targeting specific analytes The introduction of MDGC is a logical response to the need of increasing separation power – translated into the number of components resolved, and ideally an increase in number of components detected and hence a greater coverage of total components identified in a sample. Much of the early MDGC work involved heart cutting, which is not normally applied to all regions of a sample, so is not a truly ‘comprehensive’ multidimensional technique. In heartcutting MDGC (GC–GC), either a single or only a few selected regions (containing the target analytes) from the first dimension (1D) of separation are heartcut into the second-dimension (2D) column, which is of different selectivity to provide further separation. The 2D column, having different selectivity, is usually conducted under a separate temperature program. Column interfacing is achieved with either a switching valve, or using a pressure-driven switching device (Deans switch). Currently, the Deans switch is the most popular device for implementing GC–GC systems due to its simplicity and flexibility in controlling pressure and flow. The pneumatic device that controls the flow of the carrier gas automatically switches only the desired region into the 2D column based on reproducible relative pressures in the streams. It is possible to locate a cryotrap at the inlet section of the 2D column to trap and refocus the heart-cut fraction into a narrow band thus effectively minimizing the peak
future science group
Multidimensional gas chromatography in bioanalysis
dispersion from the first column, which maximizes the separation in the 2D column. More detailed information regarding history and operating principles of GC–GC can be obtained elsewhere [9–11] . Figure 1 depicts a schematic diagram of a typical GC–GC arrangement. This method serves to increase the separation power and peak capacity over traditional 1D GC, since the first column separation functions as a prefractionation step. Components not heart-cut to 2D are still measured at detector 1. The peak capacity (nc) of such a system is approximately the sum of that of 1D and that of the 2D, the latter multiplied by the number (x) of transferred fractions (n1+[n2 × x]) [11,12] . However, it should be noted that GC–GC is effective only in targeted analysis, where information regarding a limited number of individual fractions is required. If the total sample requires analysis in two different dimensions to enhance the separation for all components, then other alternative MDGC techniques (i.e., comprehensive 2D GC) must be considered. Comprehensive 2D GC: full profiling Giddings defined ‘comprehensiveness’ in 2D separations. A 2D separation can be called comprehensive if every fraction of the sample is subjected to two or more mutually independent separations in which the resolution obtained in 1D is essentially maintained until completion of the overall separation process [13] . Hence, GC–GC is not a comprehensive 2D separation, since only a part of the effluent containing the target analytes is subjected to separation on the 2D column. The first comprehensive 2D GC separation (GC × GC) was conducted by John Phillips in 1991 [14] and it was regarded as one of the most promising innovations in GC since the invention of the capillary column [15] . In contrast to GC–GC, GC × GC is a continuous sampling process in which sequential segments of effluent from the 1D column are introduced into the 2D column periodically, and their elution is usually faster than 1D peak widths, for further separation. Advantages of GC × GC over conventional 1D GC methods include superior separation power as a result of increase in peak capacity, sensitivity enhancement through band compression (or reconcentration) of the primary effluent fraction and its injection as a short chromatographic pulse into the 2D column and the benefits of a structured chromatogram (contour plot) in 2D space, which aids the identification of components and, potentially, unknowns. The two main approaches to a GC × GC method include thermal (usually cryogenic) modulation, and flow modulation.
future science group
Review
Concepts & instrumentation of GC × GC: doing the experiment In a typical GC × GC system (Figure 2), 1D column is normally a conventional capillary GC column (15–30 m, 0.25–0.32 mm I.D., and a film thickness, df, of 0.25–1.0 μm) coupled with a short narrow-bore 2 D column (typical dimensions of 0.5–2.5 m, 0.1 mm I.D., df of 0.1 μm). The key to the method lies in the modulator, a device that interfaces the effluent from the 1D to the 2D column. The modulator traps the primary column effluent fraction, refocuses it into a sharp band and re-injects it as a short chromatographic pulse into the secondary column for further separation. This process will continuously repeat until the end of the entire analysis. Modulators can basically be divided into two types: thermal-based modulators and flow-based modulators. Thermal based-modulators can be further divided into heated- and cryogenic-based modulators. Cryogenic modulators dominate, and function by trapping primary effluent below ambient temperatures using cryogens (liquid carbon dioxide or liquid nitrogen). The first of these was the longitudinally modulated cryogenic system (LMCS) developed by Marriott in the late 1990s [16] . The working principle of the LMCS is that the cryotrap oscillates back and forth along the capillary GC column to cryogenically trap the compounds of interest that are then released once the cold trapping region moves and the cold column zone is heated by the oven temperature. Commercially available cryogenic modulators include cryojet modulators, in which there are no mechanical moving parts. The first cryojet modulator was reported by Ledford, which is a dual stage, quad-jet (two liquid CO2 cold jets and two hot gas jets) thermal-based modulator [17] . The analytes elute out from the 1D column, to be effectively trapped when the first (upstream) cold jet was Inlet
DET 1
1
Dc
DET 2
2
DS
Dc
CT Oven 1
Oven 2
Figure 1. A typical GC–GC system. The second oven is optional. 1 Dc : First-dimension column; 2Dc: Second-dimension column; CT: Cryotrap; DET: Detector; DS: Deans switch.
www.future-science.com
2463
Review Wong, Hartmann & Marriott
Inlet
DET
M 1
Dc
Oven 1
Dc
2
Oven 2
Figure 2. A typical GC × GC system. The second oven system for the 2Dc is optional. 1 Dc : First-dimension column; 2Dc: Second-dimension column; DET: Detector; M: Modulator.
turned on. The first hot gas jet and second cold jet are then activated simultaneously, transferring the trapped analytes to the cold spot created by the second (downstream) cold jet, to be refocused. This is followed by re-injection into the 2D column when the second hot jet is engaged. On the other hand, flow-based modulators generate primary effluent pulses by using interfaces that collect packets of effluent into a collection column segment and then rapidly flush these into the 2 D column by switching the flow through the collecting column. A high flow rate is used, so the 2D column is relatively long, and of wide bore compared with those of a cryogenic system [18] . The main advantage of a flow-based modulator is its relatively simple design and that it does not utilize cryogens to modulate 1D peaks. However, flow modulation often may present lower resolution or lower sensitivity than thermal modulation [18] . A more comprehensive review on the development of GC × GC modulators can be obtained elsewhere [15,18] . In GC × GC, it is noteworthy that the 2D column operates as a fast GC column with a relatively short analysis time based on the chosen modulation period, PM, ranging from, for example, 2 to 6s, which ensures that the 2D separation finishes in a time shorter than PM to prevent the occurrence of wraparound, where 2D Key terms Modulation ratio: The peak width (1w b or 4σ) of a first-dimension peak divided by the second-dimension modulation period (PM ) defines the modulation ratio (MR), which approximate corresponds to the number of large peaks generated by the modulation process. Orthogonality: The measure of the different separation mechanisms of the chromatographic columns that are coupled in the GC × GC experiment. The higher the orthogonality, the higher the potential use of the 2D separation space.
2464
Bioanalysis (2014) 6(18)
peaks appear in subsequent elution sequences, caused by 2D retention times that are longer than the PM. If the 1D peak width of an analyte exceeds PM, then the peak entering the modulator will be sliced (or modulated) into several pulsed peaks on 2D column. Marriott defined the modulation ratio (MR), which is the ratio of four-times the first column peak SD (4σ or wb) divided by PM : MR = 4σ /PM = wb /PM [19] . He also demonstrated that MR of at least three should be used for the improved quantitative measurement of trace compounds, while an MR of approximately 1.5 is sufficient for the semiquantitative analysis of major components. The heating rate of the oven temperature program is also an important criteria for governing the MR (for instance, if the oven heating rate is increased, the primary peaks will become narrower, causing the MR value to decrease). A decrease in MR will reduce the observed resolution in the 1D separation by leading to neighboring 1D peaks being partially collected in the same modulation event. This must be taken into consideration when choosing the right MR for the desired analysis. The data obtained are converted into a GC × GC chromatogram (a 2D presentation contour plot, or a 3D plot with two retention times and signal intensity as the axes; Figure 3) [20] using appropriate computer software (e.g., Transform, GC Image, ChromaTOF, HyperChrom, and so on). The obtained 2D plot is highly effective in classification of structurally related compounds or classes (isomers, congeners, analogs) through the ordered structure of the 2D chromatogram, where position in 2D space is related to chemical property. Peak capacity: 1D expanded to 2D The main goal of coupling different chromatographic techniques (e.g., HPLC–GC) or columns is to enhance the nc of the analytical system. Giddings defined nc as the number of individual components that can be placed, side by side, as single entities, within the separation space [21] . Hence, the ability of GC to separate the individual constituents of a mixture is dependent on its nc [22] . In contrast to a MDGC system, nc of a GC × GC system is now equal to the product of the 1D column (1nc) and that of the 2D column (2nc); nc = 1nc × 2nc. For instance, in a temperature programed GC × GC experiment with a analysis time of 50 min, if 1D column has a capacity of 300 peaks (average peak wb of 10 s), and 2D column is capable of separating 15 peaks in 3 s (average peak wb of 200 ms), then the system should have a total capacity of 4500 peaks if the total available separation space is fully utilized. The ‘extra’ peak capacity should equate to greater separation of components. Although a conventional 1D GC system in the example above could never fully resolve
future science group
Multidimensional gas chromatography in bioanalysis
Review
nality in GC × GC. If the separation on 1D column is essentially based on boiling points of the analytes, and 2 D column will separate analytes that overlap (and so have similar boiling points) on the basis of polaritybased interactions (dipole–dipole, hydrogen bonding, etc.), then an acceptable orthogonality will arise. 2D retention of solutes will be in order of polarity. Figure 4 illustrates the retention position in 2D space on the basis of the physical and chemical nature of the analytes. It should be noted that orthogonality in MD separations is a very important measure for estimating the resolving power. However, it is important to note that ‘true orthogonality’ may not exist, limited by the available GC column stationary phases and the fact that volatility is a common property for all columns. For instance, commercially available open-tubular columns coated with different monomer chemistries or functional groups (e.g., methyl, phenyl, cyanopropyl, trifluoropropyl, and so on) are basically based on
300 components in a sample, a 2D column must offer greater resolution because it further separates components unresolved on 1D. This requires consideration of proper choice of column combination (set) to ensure some measure of ‘orthogonality,’ as discussed below. Orthogonality: maximizing component separation Both the 1D and 2D columns must have different selectivity to one another (ideally orthogonal, if Giddings’ vision was realizable) to attain the maximum peak capacity; thus the retention mechanism in the 1D and 2D must be independent in order to qualify as the best multidimensional separation [13] . A combination of column set that achieves acceptable orthogonality will normally correspond to best use of the available 2D separation space. For instance, a nonpolar (NP) 1 D/polar (P) 2D column combination is one of the best examples to illustrate an attempt to achieve orthogo-
1D chromatogram (at first column outlet)
3D plot 1. Modulation
Contour plot
di
dim
d
ens
ion
2n
1st
2. Transformation
m
en
sio
n
Raw 2D chromatogram (at second column outlet)
2D color plot
3. Visualization
on
si
1st
dim
ens ion
d 2n
di
m
en
Second dimension chromatograms stacked side by side
Figure 3. Schematic diagram for the generation of 2D and 3D GC × GC chromatograms. Reproduced with permission from [20] © Elsevier (2003).
future science group
www.future-science.com
2465
Review Wong, Hartmann & Marriott
5
2
D retention time (polar column)
High volatility High polarity
Low volatility High polarity B
4
D 3 S High volatility Low polarity
2
Low volatility Low polarity
A
1
C 0
0
1
2
3
4 5 Time (min) 1 D retention time (nonpolar column)
6
7
8
Figure 4. Schematic illusion for the retention mechanism of a nonpolar to polar column set in GC × GC. For example, compounds A and B have similar retention in the first-dimension column (similar boiling points), but different retention in the second-dimension column (different polarity). The same applies to compounds C and D.
a poly(siloxane) backbone. Hence, all these stationary phases have some similar properties in respect to nonselective intermolecular interactions (dispersive interactions) [23] . This indicates that volatility based effects will always be present as a separation force in both 1D and 2D columns. Hence, in current practice, ‘orthogonality’ in GC × GC might only correspond to the maximum possible usage of the available 2D separation space (which in turn increases nc), aiming for achieving resolution for compounds of interest. Up to now, the concept of ‘orthogonality’ in GC × GC has not been clearly defined, and it is still a subject that needs to be further investigated. Detection: spectroscopic versus nonspectroscopic The focusing effect of cryogenic modulation generates a sharp and focused zone which leads to an increase in solute mass-per-unit-time at the detector (i.e., increase in response heights or peak signals). However, the degree of signal-to-noise enhancement in GC × GC is a function of experimental conditions, although an increase factor in the range of 5–25 is a reasonable range characterized by differing performances [24] . Since GC × GC produces fast peaks (wb of 100 ms or smaller have been reported [25,26]), a fast data acquisition detector is critical to collect enough data points over a peak (≥10 points) to obtain a reliable and accurate peak reconstruction. If a detector
2466
Bioanalysis (2014) 6(18)
with an acquisition rate of 100 Hz is used for a 100 ms peak, about ten measurements over the peak (about three data points per peak standard deviation) will be acquired, which is barely sufficient for quantitative measurement. A flame ionization detector (FID) offers excellent GC × GC performance due to its highacquisition rate (up to 500 Hz), large response range, ease of operation and reproducible response [27] . However, FID lacks the ability to provide structural information for analytes, which reduces its applicability for the profiling of unknowns in untargeted analysis. Several element-selective detectors – micro-electron capture [28] , flame-photometric (FPD) [29] and nitrogen- and sulfur-chemiluminescence detectors [30,31]) have also been reported for GC × GC analysis. Engel et al. examined use of GC × GC with various types of detectors (micro-electron capture detector, FPD, nitrogen-phosphorus detector and FID) for the quantitative determination of organophosphorus pesticides, synthetic pyrethroids and fungicides [32] . However, these element-selective detectors do not offer MS information, which limits their applicability in screening for unknown compounds. For tentative identification of compounds, MS is still the detector of choice, since it adds another dimension to the analysis (mass-spectral dimension). Quadrupole MS (QMS) has been employed for detection in GC × GC due to its robustness, sensitivity and selectivity; yet, QMS is unable to meet the criteria of
future science group
Multidimensional gas chromatography in bioanalysis
‘fast’ detection for GC × GC due to its relatively low scanning or sampling rate in full mass scan mode, and spectral bias [33] . Hence, GC × GC users may opt to use a narrow (or reduced) mass scanning range to increase the scanning speed. Song et al. first described this approach for the analysis of 77 underivatized drugs with a reduced mass scan range to achieve the scanning frequency of 19.36 Hz [34] . Purcaro et al. explored the capabilities of GC × GC-QMS, with a rapid-scanning QMS for the full-scan quantification of perfume allergens. The detector operated at 50 Hz scan frequency, with a scan speed of 20,000 amu/s over a mass range of 40–330 m/z [35] . By comparison, GC × GC with TOF MS is capable of generating a complete spectrum for every pulse of ions from the ion source in a remarkably short time. The nonscanning nature of the TOF process leads to fast data acquisition rate, spectral continuity and good dynamic range [33] . The data acquisition rate of TOF MS (up to 500 spectra/s) [33] , corresponds to 50 data points per 100 ms peak. The spectral continuity of TOF MS permits more precise mathematical deconvolution of overlapped spectra, [33] and provides a higher mass accuracy. More recently, hybrid quadrupole-TOF MS (QTOF MS) technology has been tested with GC × GC. In QTOF MS, the Q serves to select target ion(s) and the TOF analyzer measures the collisioninduced dissociation mass spectrum of the ion. Definitive opportunities for QTOF MS with GC × GC are still to be explored. In recent years, isotope-ratio MS (IRMS) is gaining interest as an alternative mode of MS detection for MDGC. IRMS measures the differences in the ratio of natural isotopic abundance (for instance, 2H/1H, 13 C/12C,15N/14N and 18O/16O) in the sample relative to standard material. These differences can propose chemical, biological or geographical origin of the compounds (i.e., natural vs synthetic chemical compounds). The major application areas of this technique include forensic analysis [36] , doping analysis [37,38] and other fields where the origin or source of specific compounds are of interest. Application of MDGC in bioanalysis Introduction to metabolomics
The term ‘bioanalysis’ is linked to metabolomics or metabonomics, one of the most important functional genomics techniques in the study of life sciences [39] . In recent years, metabolite profiling (metabolomics or metabonomics) has become a major pillar for system biology studies [40] , covering the kingdoms of Animalia and Plantae. Metabolomics is defined as the global quantification of metabolites in cell, tissues and biological fluids of living systems at a given time [41,42] .
future science group
Review
Conversely, metabonomics involve the quantitative measurement of multiparametric metabolic responses of living systems to pathophysiological stimuli or genetic modification [43] . The two ‘-omic’ techniques are both involved with the analysis of metabolites; metabolomics is a global and unbiased quantification of all or a large number of metabolites (i.e., metabonomics is considered as a subset of metabolomics) in a biological system [43,44] . Metabolites are the intermediates, byproducts and end products (usually restricted to small molecules with molecular mass
For reprint orders, please contact [email protected]
Multidimensional gas chromatography methods for bioanalytical research
Multidimensional gas chromatography (MDGC) methods are high-resolution volatile chemical separation techniques, and comprise classical heart-cutting MDGC and its more recent incarnation, comprehensive 2D GC. Although available for a long period, MDGC approaches are still not widely practiced in the field of bioanalysis, possibly reflecting the general preference for regular GC versus MDGC approaches. With the recent introduction of ‘-omic’ techniques that emphasize global nontargeted profiling of metabolites within living systems, it is evident that MDGC is gaining momentum as a separation tool, since it offers very high resolution. By untangling metabolites within highly complex biological matrices, and expanding the metabolic coverage, MDGC plays a frontline role in ‘-omics’ based studies. This review highlights state-ofthe-art MDGC approaches, and summarizes the recent developments in bioanalytics.
Why GC for bioanalysis? Bioanalysis, the characterization and/or quantitative measurement of small and large molecules (e.g., drugs, metabolites, proteins and biomarkers) in biological systems, is conceptually simple but technically challenging [1] . In bioanalysis, the first and most important task normally requires isolation of the ‘target analytes’ from the complex distribution of biological matrices, followed by the appropriate measurement of these entities. In many cases, conventional technologies (for instance, single-dimensional [1D] chromatography) pose limitations for untangling constituents in real-world biological samples, due to sample complexity. Gas chromatography (GC), high performance liquid chromatography (HPLC) and capillary electrophoresis (CE) are routinely used in a wide variety of biological analyses for molecular measurement. These methods tend to rely upon 1D separation, utilizing a single, uniform separation mechanism in a static or varying physical or chemical field. In reality, many biological samples are too complex for conventional 1D separations, and the available peak capacity of the separation system (essentially the number of components
10.4155/BIO.14.186 © 2014 Future Science Ltd
that can be separated by the system) is often far exceeded by the number of sample components. These results in peak overlap, which may not be a problem in cases where a target component can be reliably identified and quantified in presence of coeluting species. However, a decrease in the quality of analysis or a reduction in analytical precision may often produce uncertainty in the identity of a compound. The need for capacity increase in turn drives the development of more advanced analytical technologies in the field of instrumental analytical chemistry, aiming at expanding the separation space (i.e., by using multiple analytical dimensions of chromatography and/or MS). Without the ability to provide precise identification, biomarkers and biological profiles have limited meaning; in other words, they cannot be fully explored. This review focuses on GC with particular reference to multidimensional GC (MDGC) as an integrated part of a technological solution to study biological components derived from humans. The first part presents an introduction to multidimensionality in GC, with discussions on fundamentals, technical implementation, detection options, capabilities and limitations of these technologies.
Bioanalysis (2014) 6(18), 2461–2479
Yong Foo Wong1, Constanze Hartmann2,3 & Philip J Marriott*,1 Australian Centre for Research on Separation Science, School of Chemistry, Monash University, Victoria, Australia 2 Division of Food Chemistry, Department of Chemistry & Pharmacy, Friedrich-Alexander-University Erlangen-Nuernberg, Erlangen, Germany 3 Fraunhofer Institute for Process Engineering & Packaging (IVV), Freising, Germany *Author for correspondence: Tel.: +61 3 9905 9630 Fax: +61 3 9905 8501 [email protected] 1
part of
ISSN 1757-6180
2461
Review Wong, Hartmann & Marriott The second part provides a comprehensive review on bioanalysis applications using MDGC techniques. The potential of MDGC techniques for the analysis of metabolites derived from humans will be highlighted. Attempts to untangle the components of interest from highly complex biological matrices through the use of MDGC techniques will be discussed. Why MDGC for bioanalysis? An introduction While the principles of GC need not be described in detail here, a brief comment on the suitability of GC for bioanalysis is warranted. Although GC is regarded as one of the most powerful methods of separation, the first and most important requirement is that the compounds of interest should be volatile (derivatization where applicable may be used to increase the volatility of analytes) and thermally stable. In some cases, derivatization is mandatory to overcome adsorption of polar functional groups on the active surfaces of the injector and at column walls, which would result in distorted peak shape and/or chemical losses. It should be noted that derivatization (alkylation, acylation or silylation reactions) may result in more than one peak for a compound of interest, and derivatization time and temperature can affect quantitative formation of the derivatives. Hence, careful tuning or optimization of some of these parameters is required in order to obtain satisfactory derivatization of the target analytes. The selection of derivatization agents is highly dependent on the classes of analyte and it is users’ experience to opt for the best one that suits their case. A more comprehensive review on derivatization techniques can be obtained elsewhere and it is not the scope of this review to reiterate this further [2,3] . Generally, GC has been well recognized as capable of excellent reproducibility, sensitivity and resolution and is capable of analyzing gases, liquids and solids (after dissolution). The principal advantage of GC in terms of efficiency is undeniable; an outstanding (but unrealistic) plate number of 1.2 million using a 1300-m fused silica column has Key terms Biomarkers: Objectively measured characteristic compounds that might be associated with a specific biological condition or state. Heart-cutting multidimensional GC: The entire sample is separated in a first-dimension column, and one or more chosen elution regions of first-dimension effluent are transferred into a second-dimension column for further separation. Comprehensive 2D GC: Complete transfer of sample effluent from the first-dimension column to the seconddimension column via a modulation device and under a defined modulation period.
2462
Bioanalysis (2014) 6(18)
been reported [4] . A typical GC system is applicable to a molecular mass range from 2 to about 1500 Da [4] , which is well suited for the analysis of metabolites (lowmolecular-weight organic or inorganic compounds [5,6] with a mass of less than 1500 Da [7]). GC coupled to MS (GC–MS) is routine in forensic, toxicology and doping control, and specific approaches must be implemented to resolve overlapping components. The World AntiDoping Agency (WADA) has recommended GC–MS as the primary method for the preliminary screening of anabolic agents, stimulants and opiates [8] . However, a major concern is the incapability of GC–MS to distinguish coeluting components with similar structure and/or MS spectra, for instance positional isomers. Isomeric forms among various chemical classes of components within complex biological matrices means that peak capacity is often exceeded and peak overlap will be the general expectation. Poor separation of the analytes makes identification and quantification difficult or uncertain. For accurate and reliable identification, each peak should be well resolved and should appear as a single component. This would avoid ambiguities resulting from contributions of coeluting components to an unresolved peak. Heart-cut MDGC: targeting specific analytes The introduction of MDGC is a logical response to the need of increasing separation power – translated into the number of components resolved, and ideally an increase in number of components detected and hence a greater coverage of total components identified in a sample. Much of the early MDGC work involved heart cutting, which is not normally applied to all regions of a sample, so is not a truly ‘comprehensive’ multidimensional technique. In heartcutting MDGC (GC–GC), either a single or only a few selected regions (containing the target analytes) from the first dimension (1D) of separation are heartcut into the second-dimension (2D) column, which is of different selectivity to provide further separation. The 2D column, having different selectivity, is usually conducted under a separate temperature program. Column interfacing is achieved with either a switching valve, or using a pressure-driven switching device (Deans switch). Currently, the Deans switch is the most popular device for implementing GC–GC systems due to its simplicity and flexibility in controlling pressure and flow. The pneumatic device that controls the flow of the carrier gas automatically switches only the desired region into the 2D column based on reproducible relative pressures in the streams. It is possible to locate a cryotrap at the inlet section of the 2D column to trap and refocus the heart-cut fraction into a narrow band thus effectively minimizing the peak
future science group
Multidimensional gas chromatography in bioanalysis
dispersion from the first column, which maximizes the separation in the 2D column. More detailed information regarding history and operating principles of GC–GC can be obtained elsewhere [9–11] . Figure 1 depicts a schematic diagram of a typical GC–GC arrangement. This method serves to increase the separation power and peak capacity over traditional 1D GC, since the first column separation functions as a prefractionation step. Components not heart-cut to 2D are still measured at detector 1. The peak capacity (nc) of such a system is approximately the sum of that of 1D and that of the 2D, the latter multiplied by the number (x) of transferred fractions (n1+[n2 × x]) [11,12] . However, it should be noted that GC–GC is effective only in targeted analysis, where information regarding a limited number of individual fractions is required. If the total sample requires analysis in two different dimensions to enhance the separation for all components, then other alternative MDGC techniques (i.e., comprehensive 2D GC) must be considered. Comprehensive 2D GC: full profiling Giddings defined ‘comprehensiveness’ in 2D separations. A 2D separation can be called comprehensive if every fraction of the sample is subjected to two or more mutually independent separations in which the resolution obtained in 1D is essentially maintained until completion of the overall separation process [13] . Hence, GC–GC is not a comprehensive 2D separation, since only a part of the effluent containing the target analytes is subjected to separation on the 2D column. The first comprehensive 2D GC separation (GC × GC) was conducted by John Phillips in 1991 [14] and it was regarded as one of the most promising innovations in GC since the invention of the capillary column [15] . In contrast to GC–GC, GC × GC is a continuous sampling process in which sequential segments of effluent from the 1D column are introduced into the 2D column periodically, and their elution is usually faster than 1D peak widths, for further separation. Advantages of GC × GC over conventional 1D GC methods include superior separation power as a result of increase in peak capacity, sensitivity enhancement through band compression (or reconcentration) of the primary effluent fraction and its injection as a short chromatographic pulse into the 2D column and the benefits of a structured chromatogram (contour plot) in 2D space, which aids the identification of components and, potentially, unknowns. The two main approaches to a GC × GC method include thermal (usually cryogenic) modulation, and flow modulation.
future science group
Review
Concepts & instrumentation of GC × GC: doing the experiment In a typical GC × GC system (Figure 2), 1D column is normally a conventional capillary GC column (15–30 m, 0.25–0.32 mm I.D., and a film thickness, df, of 0.25–1.0 μm) coupled with a short narrow-bore 2 D column (typical dimensions of 0.5–2.5 m, 0.1 mm I.D., df of 0.1 μm). The key to the method lies in the modulator, a device that interfaces the effluent from the 1D to the 2D column. The modulator traps the primary column effluent fraction, refocuses it into a sharp band and re-injects it as a short chromatographic pulse into the secondary column for further separation. This process will continuously repeat until the end of the entire analysis. Modulators can basically be divided into two types: thermal-based modulators and flow-based modulators. Thermal based-modulators can be further divided into heated- and cryogenic-based modulators. Cryogenic modulators dominate, and function by trapping primary effluent below ambient temperatures using cryogens (liquid carbon dioxide or liquid nitrogen). The first of these was the longitudinally modulated cryogenic system (LMCS) developed by Marriott in the late 1990s [16] . The working principle of the LMCS is that the cryotrap oscillates back and forth along the capillary GC column to cryogenically trap the compounds of interest that are then released once the cold trapping region moves and the cold column zone is heated by the oven temperature. Commercially available cryogenic modulators include cryojet modulators, in which there are no mechanical moving parts. The first cryojet modulator was reported by Ledford, which is a dual stage, quad-jet (two liquid CO2 cold jets and two hot gas jets) thermal-based modulator [17] . The analytes elute out from the 1D column, to be effectively trapped when the first (upstream) cold jet was Inlet
DET 1
1
Dc
DET 2
2
DS
Dc
CT Oven 1
Oven 2
Figure 1. A typical GC–GC system. The second oven is optional. 1 Dc : First-dimension column; 2Dc: Second-dimension column; CT: Cryotrap; DET: Detector; DS: Deans switch.
www.future-science.com
2463
Review Wong, Hartmann & Marriott
Inlet
DET
M 1
Dc
Oven 1
Dc
2
Oven 2
Figure 2. A typical GC × GC system. The second oven system for the 2Dc is optional. 1 Dc : First-dimension column; 2Dc: Second-dimension column; DET: Detector; M: Modulator.
turned on. The first hot gas jet and second cold jet are then activated simultaneously, transferring the trapped analytes to the cold spot created by the second (downstream) cold jet, to be refocused. This is followed by re-injection into the 2D column when the second hot jet is engaged. On the other hand, flow-based modulators generate primary effluent pulses by using interfaces that collect packets of effluent into a collection column segment and then rapidly flush these into the 2 D column by switching the flow through the collecting column. A high flow rate is used, so the 2D column is relatively long, and of wide bore compared with those of a cryogenic system [18] . The main advantage of a flow-based modulator is its relatively simple design and that it does not utilize cryogens to modulate 1D peaks. However, flow modulation often may present lower resolution or lower sensitivity than thermal modulation [18] . A more comprehensive review on the development of GC × GC modulators can be obtained elsewhere [15,18] . In GC × GC, it is noteworthy that the 2D column operates as a fast GC column with a relatively short analysis time based on the chosen modulation period, PM, ranging from, for example, 2 to 6s, which ensures that the 2D separation finishes in a time shorter than PM to prevent the occurrence of wraparound, where 2D Key terms Modulation ratio: The peak width (1w b or 4σ) of a first-dimension peak divided by the second-dimension modulation period (PM ) defines the modulation ratio (MR), which approximate corresponds to the number of large peaks generated by the modulation process. Orthogonality: The measure of the different separation mechanisms of the chromatographic columns that are coupled in the GC × GC experiment. The higher the orthogonality, the higher the potential use of the 2D separation space.
2464
Bioanalysis (2014) 6(18)
peaks appear in subsequent elution sequences, caused by 2D retention times that are longer than the PM. If the 1D peak width of an analyte exceeds PM, then the peak entering the modulator will be sliced (or modulated) into several pulsed peaks on 2D column. Marriott defined the modulation ratio (MR), which is the ratio of four-times the first column peak SD (4σ or wb) divided by PM : MR = 4σ /PM = wb /PM [19] . He also demonstrated that MR of at least three should be used for the improved quantitative measurement of trace compounds, while an MR of approximately 1.5 is sufficient for the semiquantitative analysis of major components. The heating rate of the oven temperature program is also an important criteria for governing the MR (for instance, if the oven heating rate is increased, the primary peaks will become narrower, causing the MR value to decrease). A decrease in MR will reduce the observed resolution in the 1D separation by leading to neighboring 1D peaks being partially collected in the same modulation event. This must be taken into consideration when choosing the right MR for the desired analysis. The data obtained are converted into a GC × GC chromatogram (a 2D presentation contour plot, or a 3D plot with two retention times and signal intensity as the axes; Figure 3) [20] using appropriate computer software (e.g., Transform, GC Image, ChromaTOF, HyperChrom, and so on). The obtained 2D plot is highly effective in classification of structurally related compounds or classes (isomers, congeners, analogs) through the ordered structure of the 2D chromatogram, where position in 2D space is related to chemical property. Peak capacity: 1D expanded to 2D The main goal of coupling different chromatographic techniques (e.g., HPLC–GC) or columns is to enhance the nc of the analytical system. Giddings defined nc as the number of individual components that can be placed, side by side, as single entities, within the separation space [21] . Hence, the ability of GC to separate the individual constituents of a mixture is dependent on its nc [22] . In contrast to a MDGC system, nc of a GC × GC system is now equal to the product of the 1D column (1nc) and that of the 2D column (2nc); nc = 1nc × 2nc. For instance, in a temperature programed GC × GC experiment with a analysis time of 50 min, if 1D column has a capacity of 300 peaks (average peak wb of 10 s), and 2D column is capable of separating 15 peaks in 3 s (average peak wb of 200 ms), then the system should have a total capacity of 4500 peaks if the total available separation space is fully utilized. The ‘extra’ peak capacity should equate to greater separation of components. Although a conventional 1D GC system in the example above could never fully resolve
future science group
Multidimensional gas chromatography in bioanalysis
Review
nality in GC × GC. If the separation on 1D column is essentially based on boiling points of the analytes, and 2 D column will separate analytes that overlap (and so have similar boiling points) on the basis of polaritybased interactions (dipole–dipole, hydrogen bonding, etc.), then an acceptable orthogonality will arise. 2D retention of solutes will be in order of polarity. Figure 4 illustrates the retention position in 2D space on the basis of the physical and chemical nature of the analytes. It should be noted that orthogonality in MD separations is a very important measure for estimating the resolving power. However, it is important to note that ‘true orthogonality’ may not exist, limited by the available GC column stationary phases and the fact that volatility is a common property for all columns. For instance, commercially available open-tubular columns coated with different monomer chemistries or functional groups (e.g., methyl, phenyl, cyanopropyl, trifluoropropyl, and so on) are basically based on
300 components in a sample, a 2D column must offer greater resolution because it further separates components unresolved on 1D. This requires consideration of proper choice of column combination (set) to ensure some measure of ‘orthogonality,’ as discussed below. Orthogonality: maximizing component separation Both the 1D and 2D columns must have different selectivity to one another (ideally orthogonal, if Giddings’ vision was realizable) to attain the maximum peak capacity; thus the retention mechanism in the 1D and 2D must be independent in order to qualify as the best multidimensional separation [13] . A combination of column set that achieves acceptable orthogonality will normally correspond to best use of the available 2D separation space. For instance, a nonpolar (NP) 1 D/polar (P) 2D column combination is one of the best examples to illustrate an attempt to achieve orthogo-
1D chromatogram (at first column outlet)
3D plot 1. Modulation
Contour plot
di
dim
d
ens
ion
2n
1st
2. Transformation
m
en
sio
n
Raw 2D chromatogram (at second column outlet)
2D color plot
3. Visualization
on
si
1st
dim
ens ion
d 2n
di
m
en
Second dimension chromatograms stacked side by side
Figure 3. Schematic diagram for the generation of 2D and 3D GC × GC chromatograms. Reproduced with permission from [20] © Elsevier (2003).
future science group
www.future-science.com
2465
Review Wong, Hartmann & Marriott
5
2
D retention time (polar column)
High volatility High polarity
Low volatility High polarity B
4
D 3 S High volatility Low polarity
2
Low volatility Low polarity
A
1
C 0
0
1
2
3
4 5 Time (min) 1 D retention time (nonpolar column)
6
7
8
Figure 4. Schematic illusion for the retention mechanism of a nonpolar to polar column set in GC × GC. For example, compounds A and B have similar retention in the first-dimension column (similar boiling points), but different retention in the second-dimension column (different polarity). The same applies to compounds C and D.
a poly(siloxane) backbone. Hence, all these stationary phases have some similar properties in respect to nonselective intermolecular interactions (dispersive interactions) [23] . This indicates that volatility based effects will always be present as a separation force in both 1D and 2D columns. Hence, in current practice, ‘orthogonality’ in GC × GC might only correspond to the maximum possible usage of the available 2D separation space (which in turn increases nc), aiming for achieving resolution for compounds of interest. Up to now, the concept of ‘orthogonality’ in GC × GC has not been clearly defined, and it is still a subject that needs to be further investigated. Detection: spectroscopic versus nonspectroscopic The focusing effect of cryogenic modulation generates a sharp and focused zone which leads to an increase in solute mass-per-unit-time at the detector (i.e., increase in response heights or peak signals). However, the degree of signal-to-noise enhancement in GC × GC is a function of experimental conditions, although an increase factor in the range of 5–25 is a reasonable range characterized by differing performances [24] . Since GC × GC produces fast peaks (wb of 100 ms or smaller have been reported [25,26]), a fast data acquisition detector is critical to collect enough data points over a peak (≥10 points) to obtain a reliable and accurate peak reconstruction. If a detector
2466
Bioanalysis (2014) 6(18)
with an acquisition rate of 100 Hz is used for a 100 ms peak, about ten measurements over the peak (about three data points per peak standard deviation) will be acquired, which is barely sufficient for quantitative measurement. A flame ionization detector (FID) offers excellent GC × GC performance due to its highacquisition rate (up to 500 Hz), large response range, ease of operation and reproducible response [27] . However, FID lacks the ability to provide structural information for analytes, which reduces its applicability for the profiling of unknowns in untargeted analysis. Several element-selective detectors – micro-electron capture [28] , flame-photometric (FPD) [29] and nitrogen- and sulfur-chemiluminescence detectors [30,31]) have also been reported for GC × GC analysis. Engel et al. examined use of GC × GC with various types of detectors (micro-electron capture detector, FPD, nitrogen-phosphorus detector and FID) for the quantitative determination of organophosphorus pesticides, synthetic pyrethroids and fungicides [32] . However, these element-selective detectors do not offer MS information, which limits their applicability in screening for unknown compounds. For tentative identification of compounds, MS is still the detector of choice, since it adds another dimension to the analysis (mass-spectral dimension). Quadrupole MS (QMS) has been employed for detection in GC × GC due to its robustness, sensitivity and selectivity; yet, QMS is unable to meet the criteria of
future science group
Multidimensional gas chromatography in bioanalysis
‘fast’ detection for GC × GC due to its relatively low scanning or sampling rate in full mass scan mode, and spectral bias [33] . Hence, GC × GC users may opt to use a narrow (or reduced) mass scanning range to increase the scanning speed. Song et al. first described this approach for the analysis of 77 underivatized drugs with a reduced mass scan range to achieve the scanning frequency of 19.36 Hz [34] . Purcaro et al. explored the capabilities of GC × GC-QMS, with a rapid-scanning QMS for the full-scan quantification of perfume allergens. The detector operated at 50 Hz scan frequency, with a scan speed of 20,000 amu/s over a mass range of 40–330 m/z [35] . By comparison, GC × GC with TOF MS is capable of generating a complete spectrum for every pulse of ions from the ion source in a remarkably short time. The nonscanning nature of the TOF process leads to fast data acquisition rate, spectral continuity and good dynamic range [33] . The data acquisition rate of TOF MS (up to 500 spectra/s) [33] , corresponds to 50 data points per 100 ms peak. The spectral continuity of TOF MS permits more precise mathematical deconvolution of overlapped spectra, [33] and provides a higher mass accuracy. More recently, hybrid quadrupole-TOF MS (QTOF MS) technology has been tested with GC × GC. In QTOF MS, the Q serves to select target ion(s) and the TOF analyzer measures the collisioninduced dissociation mass spectrum of the ion. Definitive opportunities for QTOF MS with GC × GC are still to be explored. In recent years, isotope-ratio MS (IRMS) is gaining interest as an alternative mode of MS detection for MDGC. IRMS measures the differences in the ratio of natural isotopic abundance (for instance, 2H/1H, 13 C/12C,15N/14N and 18O/16O) in the sample relative to standard material. These differences can propose chemical, biological or geographical origin of the compounds (i.e., natural vs synthetic chemical compounds). The major application areas of this technique include forensic analysis [36] , doping analysis [37,38] and other fields where the origin or source of specific compounds are of interest. Application of MDGC in bioanalysis Introduction to metabolomics
The term ‘bioanalysis’ is linked to metabolomics or metabonomics, one of the most important functional genomics techniques in the study of life sciences [39] . In recent years, metabolite profiling (metabolomics or metabonomics) has become a major pillar for system biology studies [40] , covering the kingdoms of Animalia and Plantae. Metabolomics is defined as the global quantification of metabolites in cell, tissues and biological fluids of living systems at a given time [41,42] .
future science group
Review
Conversely, metabonomics involve the quantitative measurement of multiparametric metabolic responses of living systems to pathophysiological stimuli or genetic modification [43] . The two ‘-omic’ techniques are both involved with the analysis of metabolites; metabolomics is a global and unbiased quantification of all or a large number of metabolites (i.e., metabonomics is considered as a subset of metabolomics) in a biological system [43,44] . Metabolites are the intermediates, byproducts and end products (usually restricted to small molecules with molecular mass
