Screening of Underivatized Oligosaccharides ... - ACS Publications

Sarah Robinson, Edmund Bergström, Mark Seymour, and Jane Thomas-Oates*. Department of Chemistry, University of York, Heslington, York, North Yorkshir...
0 downloads 0 Views 217KB Size
Download PDF
Anal. Chem. 2007, 79, 2437-2445

Screening of Underivatized Oligosaccharides Extracted from the Stems of Triticum aestivum Using Porous Graphitized Carbon Liquid Chromatography-Mass Spectrometry Sarah Robinson,† Edmund Bergstro 1 m,† Mark Seymour,‡ and Jane Thomas-Oates*,†

Department of Chemistry, University of York, Heslington, York, North Yorkshire, YO10 5DD, U.K., and Syngenta, Jealott’s Hill International Research Centre, Bracknell, Berkshire, RG42 6EY, U.K.

Highly polar oligosaccharide analytes are notoriously difficult to separate by HPLC without prior derivatization or the use of highly alkaline eluent systems. Using a porous graphitic carbon (PGC) HPLC column, we have studied a pool of endogenous underivatized water-soluble oligosaccharides that were extracted from the stems of a range of wheat cultivars. The aqueous/organic eluents that are used with this stationary phase are ideal electrospray solvents and hence facilitate the on-line coupling of the analysis to mass spectrometry. Our on-line PGC-LC-MS method has allowed the separation of native oligosaccharides, dp 2-20, in under 30 min. The method is robust and suitable for the separation of other complex oligosaccharide mixtures. We propose that isomers of fructan structures are separated and that the branching in these structures can affect their elution order. Further, our findings on the size and type of oligosaccharides extracted from wheat stems have been compared to grain yield data. Cultivars known to be high in stem carbohydrate content have been shown to contain larger oligosaccharide structures than cultivars classified as low in stem carbohydrate content. Interestingly, the largest oligosaccharides were present in the stems of wheat plants harvested 14 days after flowering, which correlates directly with the time that grain filling occurs. Triticum aestivum (bread wheat) is the most widely cultivated variety of all of the Triticum species, and more importantly, it is the world’s major crop in terms of food production. The world’s burgeoning population is placing increasing demands on our available agricultural land to produce greater yields from crops. Routes toward increasing crop yields may be found in traditional cross-breeding, genetic modification, crop protection (through the addition of pesticides), and farming cultivars appropriate to their local environment. The cultivars utilized in this study have been produced through traditional breeding methods, with an aim to exploit specific cultivar characteristics; in this instance, cultivars were classified according to their stem carbohydrate content. * To whom correspondence should be addressed. E-mail: [email protected]. Fax: +44 (0) 1904 432516. † University of York. ‡ Syngenta. 10.1021/ac0616714 CCC: $37.00 Published on Web 02/07/2007

© 2007 American Chemical Society

The stems of wheat contain pith, which varies in quantity between different cultivars and at different stages of the growth cycle. It is known that stored in the pith is a pool of endogenous water-soluble carbohydrate metabolites and that these carbohydrates are translocated into the plant’s grains.1 It has further been suggested that some wheat genotypes, which contain high concentrations of soluble carbohydrates in their stems, may be able to deposit more carbohydrates into the grains of the plant and significantly increase the grain yield.2 Traditionally, new crosses are grown at field scale and then the yields of the crop are established after the crop is harvested. This is a slow, costly, and labor-intensive method to determine each new cultivar’s grain yield. It would, thus, clearly be advantageous if a more rapid, less labor-intensive, and less expensive means of assessing grain yields of new crosses was available. The aim of this work was thus to develop an analytical method for the study of this pool of oligosaccharides found in wheat stems, in a rapid and efficient way so that many new crosses may be screened and, potentially, used to make predictions about the yield characteristics of new crosses. The separation and detection of such a pool of oligosaccharides is not easy owing to the fact that these analytes are very polar, potentially highly branched, isomeric structures, which contain no chromophore and which are poorly or completely unretained on reversed-phase HPLC columns. Many workers tackle detection issues via derivatization of the reducing terminal of their carbohydrate analytes;3 however, this presupposes that all carbohydrates are reducing. This additional step also demands extra time and sample handling and involves inevitable losses, as well as changing the original structure of the analyte. Direct infusion of native species into the mass spectrometer (MS) allows very fast analyses, and hyphenation of high-performance liquid chromatography (HPLC) with MS analysis not only allows the separation of oligosaccharide analytes, it increases the sensitivity of the analysis by significantly reducing ion suppression in the electrospray4 (ESI). Although ESI-MS analyses have been successfully (1) Wardlaw, I. F.; Porter, H. K. Aust. J. Biol. Sci. 1967, 20, 309-18. (2) Ford, M. A.; Blackwell, R. D.; Parker, M. L.; Austin, R. B. Ann. Bot. 1979, 44, 731-8. (3) Mechref, Y.; Novotny, M. V. Chem. Rev. 2002, 102, 321-69. (4) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-9.

Analytical Chemistry, Vol. 79, No. 6, March 15, 2007 2437

applied to the analysis of native carbohydrates,5 there are some difficulties associated with the ionization of these molecules; they have a low proton affinity, meaning they do not protonate efficiently, and hence, they often form cationized molecules.6 These yield a particularly low mass spectrometric response, especially when compared with, for example, peptides. High-performance liquid chromatography separations of carbohydrate mixtures have previously been carried out on many sorts of stationary phase, including normal phase (or hydrophilic interaction chromatography7), reversed phase (RP),8,9 and highperformance anion exchange chromatography (HPAEC).10,11 One of the first HPLC separations of an oligosaccharide mixture utilized a normal-phase type separation on an aminopropyl silica gel bonded stationary phase.12 These stationary phases suffer from the disadvantage that carbohydrate analytes are constantly lost to glycosylamine formation because of interactions between the reducing terminus of the glycan with the amino groups of the stationary phase.13 A further disadvantage is that, because of the polar nature of carbohydrates, these analytes interact very strongly with the column stationary phase; indeed, normal-phase separations of oligosaccharide mixtures have been reported in which oligosaccharides up to Hex17 have taken over 2 h to be eluted from the column.14 Reversed-phase HPLC separations have been applied for the analysis of carbohydrates,8,9 but as reversed-phase separations are based on hydrophobicity, only weak interactions are experienced between polar underivatized oligosaccharides and the hydrophobic stationary phase. As water is often the major eluent in RP separations, very polar oligosaccharides are often unretained and can, therefore, elute immediately. High-performance anion exchange chromatography has been highly successful in its application to the separation of carbohydrate mixtures.10,11,15 The technique utilizes a strongly alkaline mobile phase, which in turn forms negatively charged oxyanions of the carbohydrates. These anions interact with the ionic stationary phase. The alkaline mobile phase is also required for the detection system that is most commonly used with this technique, pulsed amperometric detection (PAD), as it enables the oxidation reaction to occur at the working electrode. HPAEC has real advantages over reversed- and normal-phase separations; it offers rapid analysis and separation of anomers and branched isomers, and it is reasonably sensitive without any pre- or postcolumn derivatization. One disadvantage is that the alkalinity can induce side reactions in the analytes.3 In addition, the high salt content in the mobile phase does not lend itself to a robust direct coupling of the eluent to MS. On-line membrane suppres-

sors that can remove some sodium from the eluent have been used with limited success for on-line HPAEC-ESI-MS analysis;16 but more commonly, off-line analysis without further sample cleanup after the membrane suppressor has been carried out using HPAEC-MALDI-MS,17-19 for example. HPAEC-ESI with subsequent peracetylation of collected fractions to enable easy desalting prior to MS analysis has also been carried out. Although this method provides good-quality MS data from the separated oligosaccharides, the extra derivatization step makes the method more labor intensive.19 Over 50 years ago, charcoal, as a column stationary phase, was shown to effectively separate oligosaccharides.20 PGC (or “porous glassy carbon” as it was originally known) was recognized as a highly selective stationary phase for carbohydrate separations,21 but it was also evident that the material was too fragile for use in HPLC. After detailed studies into the structure of PGC particles, changes were made to the preparation process.22 The PGC produced displayed mechanical strength that was comparable to that of silica, and thus, PGC has since been successfully used as an HPLC stationary phase. Koizumi et al.23 successfully separated mono- and disaccharides using the first commercially available PGC HPLC columns (Hypercarb), and since then, these columns have been used for HPLC separations of many types of carbohydrate compounds,24 including the separation of neutral branched isomeric mixtures.25 Further, the aqueous/organic eluents used with PGC are ideal for generating electrosprays in on-line LC-MS analyses.26 Here we present a rapid, robust, and efficient chromatographic separation method that utilizes the highly selective stationary phase of porous graphitic carbon in an on-line coupling of liquid chromatography with electrospray mass spectrometry (PGC-LCMS). Our method was applied to the separation and analysis of the native water-soluble oligosaccharides (dp 2-20) extracted from the pith of a range of wheat cultivars and their crosses, but it is generally applicable to the analysis of complex oligosaccharide mixtures that contain a range of carbohydrate structures. Additionally, we describe the development, characterization, and use of a nanoscale LC column coupled with a nanospray ionization source, which enabled us to miniaturize the system and obtain much improved sensitivity.

(5) Harvey, D. J.; Naven, T. J. P.; Kuster, B. Biochem. Soc. Trans. 1996, 24, 905-12. (6) Reinhold, V. N.; Reinhold, B. B.; Costello, C. E. Anal. Chem. 1995, 67, 1772-84. (7) Alpert, A. J. J. Chromatogr. 1990, 499, 177. (8) Sharp, J. K.; Valent, B.; Albersheim, P. J. Biol. Chem. 1984, 259, 1131220. (9) Rassi, Z. E. J. Chromatogr., A 1996, 720, 93-118. (10) Corradini, C.; Bianchi, F.; Matteuzzi, D.; Amoretti, A.; Rossi, M.; Zanoni, S. J. Chromatogr., A 2004, 1054, 165-73. (11) Lee, Y. C. J. Chromatogr., A 1996, 720, 137-49. (12) Linden, J. C.; Lawhead, C. L. J. Chromatogr. 1975, 105, 125-33. (13) Churms, S. C. J. Chromatogr., A 1996, 720, 75-91. (14) Rudd, P. M.; Dwek, R. A. Curr. Opin. Biotechnol. 1997, 8, 488-97. (15) Townsend, R. R.; Hardy, M. R.; Lee, Y. C. Methods Enzymol. 1989, 179, 65-76.

(16) Conboy, J. J.; Henion, J. Biol. Mass Spectrom. 1992, 21, 397-407. (17) Kabel, M. A.; Schols, H. A.; Voragen, A. G. J. Carbohydr. Polym. 2001, 44, 161-5. (18) van Alebeek, G. J. W. M.; Zabotina, O.; Beldman, G.; Schols, H. A.; Voragen, A. G. J. Carbohydr. Polym. 2000, 43, 39-46. (19) Brull, L.; Huisman, M.; Schols, H.; Voragen, F.; Crithley, G.; Thomas-Oates, J.; Haverkamp, J. J. Mass Spectrom. 1998, 33, 713-20. (20) Whistler, R. L.; Durso, D. F. J. Am. Chem. Soc. 1950, 72, 677-9. (21) Gilbert, M. T.; Knox, J. H.; Kaur, B. Chromatographia 1982, 16, 138-46. (22) Knox, J. H.; Kaur, B.; Millward, G. R. J. Chromatogr. 1986, 352, 3-25. (23) Koizumi, K.; Okada, Y.; Fukuda, M. Carbohydr. Res. 1991, 215, 67-80. (24) Koizumi, K. J. Chromatogr., A 1996, 720, 119-26. (25) Lipniunas, P. H.; Neville, D. C. A.; Trimble, R. B.; Townsend, R. R. Anal. Biochem. 1996, 243, 203-9. (26) Kawasaki, N.; Ohta, M.; Hyuga, S.; Hashimoto, O.; Hayakawa, T. Anal. Biochem. 1999, 269, 297-303.

2438

Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

EXPERIMENTAL SECTION Extraction of Oligosaccharides from Wheat Stems. The varieties of wheat used in this work were Rialto, Spark, and a range of crosses. After harvesting, wheat stems were immediately frozen in liquid nitrogen and stored at - 25 °C. The stems were cut into

∼2-cm pieces and boiled for 30 min in ethanol. The ethanol extract was dried, and the stem pieces were further boiled in water for 30 min. The ethanol and water extracts were mixed and washed with hexane until all coloration had been removed. Each extract was filtered through an inert 0.22-µm syringe filter (Cameo syringe filters, Sigma, Poole, UK) prior to LC analysis. Permethylation of Oligosaccharides. Oligosaccharides were permethylated according to the method of Ciucanu and Kerek.27 Briefly, permethylation experiments were carried out in anhydrous conditions. A 50-µL aliquot of 1 mg mL-1 carbohydrate analyte was dried in a 13 × 100 mm round-bottomed glass tube (with Teflon-coated screw cap) before being dissolved in 0.5 mL of DMSO. NaOH pellets were ground in a glass pestle and mortar, and 300 mg of the ground hydroxide was added as quickly as possible to the sample. At 0, 10, and 30 min, 250-µL aliquots of CH3I were added to the sample. After 20 min, the reaction was quenched with 1 mL of 100 mg mL-1 aqueous sodium thiosulfate solution (freshly made) to the sample. Each sample had 1 mL of dichloromethane added; samples were mixed well and centrifuged to break the emulsion. The upper (aqueous) layer was removed and discarded and the organic layer washed three times with water before being dried under a stream of N2. Monosaccharide Composition Analysis. Acid hydrolysis conditions were as follows: 50 µL of stem extract (the carbohydrate concentration of extracts from wheat stems was unknown) was placed in a glass 13 × 100 mm round-bottomed, screw-capped tube. A 100-µL aliquot of 1 M aqueous HCl was added to hydrolyze the oligosaccharides into their monosaccharide components. The tubes were placed in a heating block (Grant Instruments, Cambridge, UK) at 100 °C for 4 h. The tubes were allowed to cool before the acid was removed using a vacuum centrifuge (Savant Speedvac, Thermo Electron Corp., Runcorn, UK). The conditions for HPAEC-PAD separation of monosaccharides were as follows: monosaccharides were chromatographically separated using a Dionex (Sunnyvale, CA) GP50 gradient pump with an ED50 electrochemical detector. A CarboPac PA100 (Dionex) guard column and 250 × 4 mm i.d. CarboPac PA100 (Dionex) anion exchange column were used for the chromatographic separations. The system was operated via Chromeleon software, version 6.11. Eluent A was 500 mM NaOH, freshly prepared daily using 50% w/v NaOH solution. Eluent C was HPLC grade water. Both eluents were filtered through a 0.2-µm nylon membrane filter (Whatman) and degassed with helium for 15 min. The pump heads were primed with both eluents before allowing the system to equilibrate at 5.5% A and 94.5% C. A multistep gradient was programmed: t ) 0-20 min 5.5% A + 94.5% C, t ) 20-21 min 100% A, t ) 21-31 min 100% A, t ) 31-32 min 5.5% A + 94.5% C, and t ) 32-42 min 5.5% A + 94.5% C. The eluent flow rate was 1 mL min-1, and 20 µL of sample was injected onto a 50-µL loop. The waveform applied to the detection settings was as follows: t ) 0.0 (0.1 V), t ) 0.2 (0.1 V), t ) 0.4 (0.1 V), t ) 0.41 (-2.0 V), t ) 0.42 (-2.0 V), t ) 0.43 (0.6 V), t ) 0.44 (-0.10 V), and t ) 0.5 s (-0.10 V). The 0.1 mg mL-1 glucose and fructose solutions in water were prepared in 1.5-mL microcentrifuge tubes using dry compound and HPLC grade water as solvent. (27) Ciucanu, I.; Kerek, F. Carbohydr. Res. 1984, 131, 209-17.

Reduction of Oligosaccharides. A 250-µL aliquot of 10 mg mL-1 sodium borohydride in 0.5 M ammonium hydroxide solution was added to 50 µL of dried oligosaccharide extract. After 2 h, the reaction was quenched with glacial acetic acid. Each sample was washed three times with 10% acetic acid in methanol solution and then washed a further three times with methanol. The oligosaccharides analyzed as controls were linear maltose homologues, Glc4 to Glc10 (maltooligosaccharides from corn syrup, Sigma). Electrospray Quadrupole Orthogonal Time-of-Flight Mass Spectrometry (ESI-qoToF-MS). Analyses of permethylated and reduced-permethylated oligosaccharides were carried out on a hybrid qoToF Applied Biosystems API QSTAR Pulsar i (Foster City, CA). The data analysis software was Analyst QS (Applied Biosystems). Underivatized oligosaccharides were electrosprayed in 50:50 methanol/water; derivatized (permethylated) oligosaccharides were sprayed in methanol. Nitrogen was used as the nebulizing, curtain and collision gases. The samples were infused at 0.4 µL min-1, ion source gas was 6 (arbitrary units), and the ion spray voltage was 5300 V. Analyses were carried out in the positive ion mode. Liquid Chromatography ESI Ion Trap (IT) MS. Oligosaccharides extracted from the stems of wheat were separated on a Hypercarb (PGC) column (5 µm, 100 × 4.6 mm; Thermo Electron) using a Surveyor LC system (Thermo Finnigan, San Jose, CA). Solvent A was water, solvent B was acetonitrile, and solvent C was 2-propanol. The following gradient conditions were applied: t ) 0 min, 94% solvent A, 3% solvent B, and 3% solvent C; t ) 10 min, 92% solvent A, 4% solvent B, and 4% solvent C; t ) 20 min, 85% solvent A, 6% solvent B, and 9% solvent C; t ) 30 min, 82% solvent A, 8% solvent B, and 10% solvent C. The flow rate was 600 µL min-1. Oligosaccharides were injected in water, and the LC eluent was directly coupled to the electrospray source on an LCQ Deca XP plus IT mass spectrometer (Thermo Finnigan). The ionization conditions were as follows: spray voltage 6 kV, sheath gas 70 (arbitrary units), auxiliary gas 60 (arbitrary units), capillary temperature 350 °C, capillary voltage 12.0 V. Mass spectra over the range m/z 200-2000 were acquired in positive ion mode at a scan speed of 5500 amu/s. The software used for data analysis was Xcalibur version 1.3 (Thermo Finnigan). Nanoscale Liquid Chromatography NanoESI IT MS. Oligosaccharides extracted from the stems of wheat were separated on a 5-µm, 100 × 0.1 mm i.d. Hypercarb (PGC) column (Thermo Electron) using a Surveyor LC system (Thermo Finnigan) and an AC-400-VAR flow splitter (Acurate, LC Packings, San Francisco, CA). Eluents were pumped from the Surveyor LC system at 400 µL min-1 and split so that the flow through the column was 0.15 µL min-1. The gradient used was the same as that reported above in the microscale experiments. The 20-nL oligosaccharide solutions in water were injected using a Valco low-dispersion, fourport valve with an internal 20-nL loop (LC Packings). The LC eluent from the column was directly coupled to a nanoelectrospray source fitted with a front-coated PicoTip emitter (New Objective, USA) and positioned to spray directly into an LCQ Deca XP plus IT mass spectrometer (Thermo Finnigan). Helium was used as the damping gas in the trap. The ionization conditions were as follows: spray voltage 3.1 kV, sheath gas 0 (arbitrary units), auxiliary gas 0 (arbitrary units), capillary temperature 200 °C, Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

2439

Figure 1. MALDI-qoToF mass spectrum of permethylated oligosaccharides extracted from the stems of wheat cultivar Rialto (harvested 14 days postanthesis). All species correspond to [M + Na]+.

capillary voltage 14.0 V. Mass spectra over the range m/z 2002000 were acquired in positive ion mode at a scan speed of 5500 amu/s. The software used in data analysis was Xcalibur version 1.3 (Thermo Finnigan). A series of a 1-kestose solutions in water (Fluka Chemie, Buchs SG, Switzerland) were prepared for limit of detection studies; the concentrations of the solutions were as follows: 5 µM, 50 nM, 0.5 nM, 0.1 nM, and 50 pM. To calculate the peak height from the chromatograms, selected ion chromatograms (SICs) were boxcar smoothed 7 points and integrated using the default settings in Genesis (peak algorithm calculation in Xcalibur v. 1.3). RESULTS AND DISCUSSION The pool of endogenous water-soluble oligosaccharides present in the stems of wheat were extracted in boiling water according to Kerepesi et al.28 The extracts had nonpolar material removed via aqueous/organic partition and were filtered and diluted prior to LC analysis. The stems of Rialto, SR4 (31A), and SR24 (117A) are known to be high in stem carbohydrates and Spark, SR5 (38A), and SR11 (68B) are known to be low in stem carbohydrates. All cultivars are winter wheat cultivars (i.e., the seeds have been vernalized), and samples of each cultivar were harvested at anthesis (flowering) and at 14 and 22 days postanthesis. Derivatization prior to MS Analysis. In order to gain an insight into the size distribution and range of oligosaccharide (28) Kerepesi, I.; Toth, M.; Boross, L. J. Agric. Food Chem. 1996, 44, 3235-9.

2440 Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

species present in each wheat extract to enable development of an appropriate separation system, the native oligosaccharides were derivatized prior to MS analysis. Per-O-methylation was the derivatization method of choice, as it has been shown to increase the hydrophobicity of oligosaccharide molecules and increase their volatility.29 This, in turn, can significantly enhance the mass spectrometric intensity of carbohydrate-derived ions30 and, in particular, allow the ionization of larger oligosaccharides more readily.31 Upon MS analysis, a range of hexose oligomers was observed in each wheat extract. The largest oligosaccharide observed was Hex19, and this was found in the extract of Rialto, harvested 14 days postanthesis (Figure 1). Thus, we deduced that the oligosaccharides in the stem are highly water soluble and even large oligosaccharides are extracted. This information influenced our approaches to developing a chromatographic separation for these oligosaccharides as the chromatographic method must, therefore, be able to separate oligosaccharides up to at least Hex19, which was a broader range of structures than we had initially anticipated. Monosaccharide Composition Analysis. Aqueous solutions of authentic glucose and fructose standards were separated using (29) Antonopoulos, A.; Bonnet, P.; Botek, E.; Debrun, J. L.; Hakim, B.; Herbreteau, B.; Morin-Allory, L. Rapid Commun. Mass Spectrom. 2003, 17, 122-5. (30) Dell, A.; Oates, J. E.; Morris, H. R.; Egge, H. Int. J. Mass Spectrom. 1983, 46, 415-8. (31) Dell, A.; Morris, H. R.; Egge, H.; Vonnicolai, H.; Strecker, G. Carbohydr. Res. 1983, 115, 41-52.

Figure 2. HPAEC-PAD chromatograms of (i) 0.1 mg mL-1 glucose and (ii) fructose overlaid against (iii) hydrolyzed extract from the stem of wheat cultivar Rialto, harvested 14 days postanthesis.

HPAEC and detected by pulsed amperometric detection. The retention time for glucose on our gradient was ∼11 min, and the retention time of fructose on the same gradient was ∼15 min (Figure 2). Although the same amount and concentration of glucose and fructose standards were injected onto the anion exchange column, the amperometric response for each analyte was not the same. Although we may have expected the peak intensity of the glucose and fructose chromatographic peaks to be the same, other workers’ data also show that these two monosaccharides do not induce the same amperometric response.32 It has been reported that PAD can generate different responses that are characteristic of each individual compound, and thus, unless the instrument is calibrated individually for each peak detected, the results can only be qualitative.33 The unfractionated stem extract of the wheat cultivar Rialto (harvested at 14 days postanthesis) was subjected to hydrolyzing conditions. The dry hydrolyzed extracts were rehydrated with water and analyzed using the same HPAEC-PAD gradient as the glucose and fructose standards (Figure 2). The chromatogram of the HPAEC-PAD analysis of the hydrolyzed stem extract of Rialto displayed a very large chromatographic peak that eluted at the same retention time as the glucose standard and a smaller chromatographic peak that eluted at the same time as the fructose standard. We thus conclude from these analyses that the major monosaccharide constituents of the wheat stem extracts are glucose and fructose. Reduction, prior to MS Analysis, of Underivatized Oligosaccharides. Mass spectrometric analysis of our permethylated oligosaccharide species suggested that these are hexose oligomers owing to their masses and the 204 Th increment between signals. However, a limitation of mass spectrometric analysis is that isobaric ions are impossible to differentiate, and thus, we cannot determine which monosaccharide or how many different (potentially branched) isomers of these hexose oligomers there may be by mass spectrometry alone. There is evidence in the literature to suggest that highly water soluble polyfructosylsucroses (or fructans) are commonly found (32) Cheng, X.; Kaplan, L. A. J. Chromatogr. Sci. 2003, 41, 1-5. (33) Pavis, N.; Chatterton, N. J.; Harrison, P. A.; Baumgartner, S.; Praznik, W.; Boucaud, J.; Prud’homme, M. P. New Phytol. 2001, 150, 83-95.

as storage sugars in the stems of monocotyledons.34 Plant fructans are oligomers built up from the disaccharide sucrose (a nonreducing carbohydrate) by the addition of fructose moieties in β(2-1) and β(2-6) linkage to either the glucosyl or the fructosyl residue. Nonlinear fructan oligomers are known as gramminans, and the range of different chain lengths and branched structures that can occur is still not well understood.35 As fructans are nonreducing carbohydrates, we chose to test our oligosaccharides extracted from wheat stems for their behavior in reducing conditions. Figure 3A shows standard maltooligosaccharides (reducing Glc1 to Glc10 oligosaccharides) that were subjected to reducing and permethylating conditions prior to ES-MS analysis. Two series of ions that correspond to sodiated molecules of permethylated hexose oligomers and more intense sodiated molecules of permethylated Hexn-hexitol oligomers were observed. Although the presence of the permethylated hexose oligomers suggested that the reduction reaction had not gone to completion, the presence of the permethylated Hexn-hexitol oligomers demonstrated that the oligosaccharides were reducible under these conditions. When the same reducing and permethylating conditions were applied to our oligosaccharide extracts from wheat stems (Figure 3B), no signals for reduced and permethylated hexoses were observed; only signals for sodiated molecules of permethylated and unreduced hexoses were observed. These experiments, together with the composition analysis data, indicate that the structures of the aqueous soluble oligosaccharides extracted from the stems of wheat are consistent with the nonreducing oligosaccharides, the fructans (polyfructosylsucroses). PGC-LC-ESI-IT-MS of Native Oligosaccharides. It has recently been published that linear Glc2-Glc11 oligomers have been separated in 30 min using β-cyclodextrin as the HPLC stationary phase.36 We have investigated PGC as an alternative stationary phase for the separation of very polar analytes, such as native water-soluble oligosaccharides in wheat stem. Initially, two eluent systems based on acetonitrile and water and methanol and water were tested to try to separate the (34) Ritsema, T.; Smeekens, S. Curr. Opin. Plant Biol. 2003, 6, 223-30. (35) Pollock, C. J. New Phytol. 1986, 104, 1-24. (36) Liu, Y.; Urgaonkar, S.; Verkade, J. G.; Armstrong, D. W. J. Chromatogr., A 2005, 1079, 146-52.

Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

2441

Figure 3. (A) ESI-qoToF mass spectrum of maltooligosaccharides (Glc1-Glc10) that were subjected to reducing and permethylating conditions. (B) ESI-qoToF mass spectrum of oligosaccharides extracted from the stems of Rialto (harvested 14 days postanthesis) and subjected to reduction and permethylation conditions. All annotated peaks correspond to sodiated molecules.

underivatized oligosaccharide mixtures extracted from wheat stems. We found, like previous workers ,37,38 that acetonitrile had a higher eluotropic strength on the PGC stationary phase than methanol. In an isocratic elution using 95:5 (v/v) water/acetonitrile, after 10 min, only the monosaccharides in the solution of standards had eluted. It was, thus, apparent that a gradient was required to elute larger oligosaccharides in a reasonable analysis time. Binary gradients were evaluated, and it was found that monosaccharides could be removed much more quickly from the column using this approach. However, the binary solvent system provided only a partial separation of the oligosaccharide mixture; when washing the column at the end of gradients with 100% organic eluent, ions consistent with larger oligosaccharides were observed in the mass spectra and indicated that these oligosaccharides had not been eluted from the column during the gradient. An organic eluent, 2-propanol, is reported to be higher in the PGC eluotropic series than acetonitrile, although it can be difficult (37) Pereira, L.; Woodruff, M. Thermo Hypersil-Keystone brochure, 2003. (38) Deschamps, F. S.; Gaudin, K.; Baillet, A.; Chaminade, P. J. Sep. Sci. 2004, 27, 1313-22.

2442 Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

to use in LC systems, as the high viscosity of 2-propanol can create high back pressures. 2-Propanol was, therefore, mixed 50:50 (v/ v) with acetonitrile to reduce the viscosity. After mixing 2-propanol with acetonitrile, chromatographic analyses were found to elute mono-, di-, tri-, and oligosaccharides from the column, with a Hex20 species being observed after 15 min. However, the larger oligosaccharides were not chromatographically well resolved. The 2-propanol was, thus, used separately so that a ternary gradient could be applied. The chromatographic approach of the ternary gradient afforded better control of the separation of the oligosaccharide mixture. In the early part of the chromatographic separation, mono-, di-, tri-, and tetrasaccharides were removed in a relatively steep aqueous/acetonitrile gradient. After these analytes were eluted, the gradient was made gentler and the 2-propanol eluent was gradually increased in order to remove the larger and very strongly retained oligosaccharides from the column. This extensive method development ultimately allowed the successful separation of the complex mixtures of oligosaccharides that had been extracted from wheat stems and that ranged in size from dp 2 to 20, in under 30 min. A series of selected ion chromatograms of the oligosaccharides extracted from the stems of Rialto (harvested 22 days postanthesis) are presented in Figure 4. Although ions consistent with Hex18, Hex19, and sometimes Hex20 species (as sodiated molecules) were observed in the full scan mass spectra of PGC-LC-MS analyses of wheat stem extracts, their ion intensities were not significantly different from background, so that their ions were not distinguishable in the SIC. In the mass spectra, we observed sodiated singly charged, doubly charged, and dimeric (two analyte molecules with one chargebearing cation) species of the oligosaccharide analytes in the extracts from wheat stems (Figure 5). In addition to the separation of the range of oligosaccharide sizes present in the sample, it was noted that for some m/z values there were multiple peaks in the SIC. Identical m/z values were observed under each peak in the SIC, suggesting that isomeric oligosaccharides have been separated in this method. However, it should be noted that it is well-established that when oligosaccharide concentrations are high, nonspecific clustering occurs, with the possibility of multiple charging. We propose that ions arising due to this phenomonenon are observable in these data (Figure 4). For example, the peak for a Hex2 oligosaccharide, eluting at ∼6 min also gives rise to a peak in the Hex3 extracted ion trace; this can be rationalized as arising from the 13C isotopic ion from a [(Hex2)3 + Na + H]2+ cluster. The two peaks in the Hex6 trace eluting around 12-13 min can be explained similarly as deriving from Hex4 species clustering. It was noted that the elution order of the oligosaccharides was not necessarily in the order of increasing oligosaccharide size; sometimes larger oligosaccharides coeluted with a smaller oligosaccharide. For example in Figure 4, a Hex6 chromatographic peak (tR, 11 min) can be observed to coelute with a Hex4 chromatographic peak (tR, 11 min), while further Hex6 peaks elute later in the chromatographic run. This phenomenon can be explained from a consideration of the principles of how the stationary-phase PGC interacts with analytes; planar molecules are more strongly retained on PGC than less planar molecules, as they do not induce as strong dipole interactions with the PGC

Figure 4. PGC-LC-ESI-IT-MS selected ion chromatograms of underivatized oligosaccharides extracted from the stems of Rialto (harvested 22 days postanthesis). Chromatograms are Gaussian smoothed 15 points.

surface. We propose, therefore, that the elution order of differently sized oligosaccharides is further evidence that these analytes are isomeric structures and that some of these isomers have more planar structures than others. Comparison of Oligosaccharides Extracted from the Stems of Different Wheat Cultivars. The on-line PGC-LC-MS method developed was then used to analyze high and low stem carbohydrate wheat cultivars at anthesis and 14 and 22 days postanthesis. Using the chromatographic and mass spectral data from these analyses, the largest oligosaccharide structure detected in each cultivar was identified (Figure 6). It has previously been proposed that water-soluble carbohydrates in the stems of wheat can be directly linked to the grain yield of the plant. Yield data for these cultivars farmed at two different locations (data not shown) collected over a 3-year period indicate that the cultivars Rialto, SR4 (31A), and SR24 (117A), which are known to be high in stem

carbohydrates, do indeed have a higher yield based on the weight of 1000 grains. Upon comparison of the PGC-LC-MS data plotted for the high and low stem carbohydrate cultivars (Figure 6), we observe that, in general, the high stem carbohydrate cultivars produced longer oligosaccharide structures. A further conclusion that we draw from the data presented in Figure 6 is that it is noticeable that the size of the oligosaccharides in the stem increases between anthesis and 14 days postanthesis. After a further 8 days (22 days postanthesis), it is noticeable that the size of the oligosaccharides has begun to decrease. It is known that grain filling occurs approximately two weeks after anthesis. We, therefore, propose that the production of larger oligosaccharides in the stem two weeks postanthesis is in preparation for the oligosaccharides to translocate into and fill the plant’s grains. NanoPGC-LC-nanoESI-IT-MS of Native Oligosaccharides. Given the speed and robustness of our on-line LC-MS analyses, we speculated how far we could streamline the sample-handling aspects of the analysis by working with less biological material. In order to do this, we had to establish a miniaturized system. There are many reported benefits of using miniaturized systems; for example, the use of nanoscale electrospray can overcome sensitivity problems found when analyzing neutral oligosaccharides in conventional microscale electrospray.39 In addition to this increase in sensitivity, it has been reported that miniaturization of normal-phase HPLC systems coupled with nanoESI analyses of oligosaccharides can be performed with limits of detection down to low-femtomole levels.40 Other workers have compared capillary LC-ESI-MS analyses with nanoscale LC-nanoESI-MS analyses and shown a 10-fold sensitivity increase and 100-fold absolute signal increase using nanoscale systems in the analysis of glycoprotein oligosaccharides.41 It is proposed that the sensitivity increase is due to more efficient ion transfer into the ion source, rather than a lowering of the chemical background. In light of these observations, and more importantly the recent availability of nanoscale PGC columns, we decided to compare the limit of detection (LOD) of the on-line PGC-LC-ESI-MS separation that we have developed with the LOD obtainable using an on-line nanoscale system (Figure 7, panel A). Our on-line microbore method used a 4.6-mm-internal diameter PGC column; in our nanoscale experiments, we used a 0.1-mm-internal diameter PGC column. The flow rate used previously was 600 µL min-1; however, to achieve nanoflow rates, we used a flow splitter to obtain a flow rate of 0.15 µL min-1 through the nanoscale column. As the flow rate was dramatically decreased, we predicted that the elution time of the chromatographic run would increase. Based on the column dimensions, it could be predicted that the linear velocity of the mobile phase would be 1.9 times higher in the 4.6mm column. After optimization, we could reproduce a similar chromatographic separation of oligosaccharides extracted from wheat stems in 50 min (compared with 30 min using the 4.6-mm column) (Figure 7, panel B). The quantity of oligosaccharides extracted from wheat stems was unknown; therefore, our comparative study utilized a dilution series of a standard fructan trisaccharide compound, 1-kestose (39) Bahr, U.; Pfenninger, A.; Karas, M. Anal. Chem. 1997, 69, 4530-35. (40) Wuhrer, M.; Koeleman, C. A. M.; Deelder, A. M.; Hokke, C. N. Anal. Chem. 2004, 76, 833-8. (41) Karlsson, N. G.; Wilson, N. L.; Wirth, H. J.; Dawes, P.; Joshi, H.; Packer, N. H. Rapid Commun. Mass Spectrom. 2004, 18, 2282-92.

Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

2443

Figure 5. PGC-LC-ESI-IT-MS positive ion mass spectra of underivatized oligosaccharides extracted from the stems of Rialto (harvested 22 days postanthesis). Retention times are marked in parentheses; m/z value of sodiated or disodiated molecules follows.

Figure 6. Data plot of the largest hexose structures detected using PGC-LC-MS analysis of underivatized oligosaccharides extracted from the stems of different wheat cultivars (solid line, high stem carbohydrate cultivars; dashed line, low stem carbohydrate cultivars) that were harvested at anthesis and 14 and 22 days postanthesis.

(Figure 7, panel A). The solutions were analyzed using the PGCnanoLC-nanoESI-MS method, and samples were analyzed in the order of decreasing concentration. Water was injected onto the column between each sample analysis, to be certain that there was no analyte carryover. 1-Kestose solutions were analyzed until a concentration was injected that could no longer be detected. The same 1- kestose solutions were analyzed using the 4.6-mm column in the PGC-LC-ESI-MS method, again interspersed with blanks. From the SIC of each sample run in triplicate, the average peak height was calculated. An approximation of the noise level in the system was made by measuring the noise level in the baseline close to the analyte chromatographic peak. A signal three 2444

Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

times the noise was used as an approximation to the detection limit. The lowest concentration from the dilution series injected that could be detected in the PGC-nanoLC-nanoES-IT-MS analysis of 1-kestose was 50 pM. From the peak height and the noise measured in the single ion chromatogram of these analyses, the concentration LOD for this analyte in this system was calculated as 14 pM. In contrast, the concentration LOD calculated from the chromatogram from an injection of 5 µM 1-kestose observed with PGC-LC-ES-IT-MS using the 4.6-mm column was found to be 14 nM. Upon comparison of the concentration LODs in the two systems, it may be concluded that there is a 1000-fold improvement in LOD when using the 0.1-mm column compared with the 4.6-mm column. In addition to calculating the concentration LOD of each system, the amount LOD for the two systems was calculated by multiplying the concentration LODs by the injection volumes of 20 nL and 10 µL used for the 0.1-mm and 4.6-mm columns, respectively. The amount LOD obtained in the PGC-nanoLCnanoES-IT-MS analyses is 2.7 × 10-19 mol, and in the PGC-LCES-IT-MS analyses using the 4.6-mm column, it was determined to be 1.4 × 10-13 mol. Upon comparison of the amount LODs in the two systems, it can be concluded that there is a 500 000-fold increase in absolute amount sensitivity when using the 0.1-mm column as compared with the 4.6- mm column. CONCLUSIONS We have developed and optimized an on-line LC-MS method, using a PGC stationary phase, for the separation of a complex mixture of nonreducing oligosaccharides extracted from the stems of wheat. It has recently been published that linear Glc2-Glc11 oligomers have been separated in 30 min using β-cyclodextrin

These qualities are essential if this method is to be applied to the early analysis of a large number of samples. We have shown that significantly smaller quantities of analyte are required for detection when using a nanobore PGC LC column and nanoESI source compared with those required for use with the microbore PGC LC column and microESI source. The implications of these findings suggest that in the future it may be possible to obtain the same information about the oligosaccharides present in wheat stems from much smaller amounts of stem. The amount LODs calculated from the nanoscale experiments also suggest that analytes are detected at the zeptomole level. As ions are well separated on the PGC column, this low amount LOD may improve the dynamic range of the analysis. Nanoscale analyzers are also “greener” as they create significantly less organic solvent waste. This sensitive and rapid analysis of oligosaccharides from a biological system has provided useful data that have allowed comparisons to be made between the oligosaccharides extracted from the stems of a number of different wheat cultivars. We have noted that cultivars classified as high in stem carbohydrate contain larger oligosaccharides than cultivars that are classified as low in stem carbohydrate. We have also noted that the largest oligosaccharides were present in all cultivars when the wheat was harvested 14 days postanthesis. This corresponds to the time that grain filling occurs, thus suggesting that larger oligosaccharides are present in the stem, ready to translocate into and fill the grain. Our system could, therefore, be applied for the early, and potentially predictive, analysis of wheat cultivar crosses to study further the link between stem carbohydrate and grain yield, but also this system could be applied to the analysis of other highly polar analytes that have previously been difficult to analyze by other separation methods.

Figure 7. PGC-nanoLC-nanoESI-QIT-MS selected ion chromatograms of a 2-amol injection of the standard oligosaccharide 1-kestose (A) and underivatized oligosaccharides extracted from the stems of Rialto (harvested 22 days postanthesis) (B). Chromatograms are Gaussian smoothed 15 points.

bonded HPLC as stationary phase.36 We have significantly surpassed this performance by separating Hex2 to Hex20 in 30 min. Our system displays high chromatographic resolution and selectivity for a broad size range of isomeric oligosaccharides in a relatively short analysis time. In addition, the system is robust and automation has allowed very high throughput of samples.

ACKNOWLEDGMENT We gratefully acknowledge S.R.’s studentship, provided by Syngenta and the Engineering and Physical Sciences Research Council (EPSRC). J.T.-O. gratefully acknowledges Thermo Finnigan’s support and funding from the Analytical Chemistry Trust Fund, the RSC Analytical Division, and EPSRC. The authors gratefully acknowledge Professor John Snape at the John Innes Centre, Norwich, for provision of seeds of Rialto, Spark, and their crosses and also for providing crop yield data for these cultivars.

Received for review December 15, 2006.

September

5,

2006.

Accepted

AC0616714

Analytical Chemistry, Vol. 79, No. 6, March 15, 2007

2445