Anal. Chem. 2010, 82, 6660–6666
Evaluation of Sampling and Extraction Methodologies for the Global Metabolic Profiling of Saccharophagus degradans Min Hye Shin,† Do Yup Lee,‡ Kwang-Hyeon Liu,§ Oliver Fiehn,*,‡ and Kyoung Heon Kim*,† School of Life Sciences and Biotechnology, Korea University, Seoul 136-713, Republic of Korea, Genome Center, University of California, Davis, California 95616, and Department of Pharmacology, Inje University College of Medicine, Busan 614-735, Republic of Korea Metabolomics is based on the unbiased and global analysis of metabolites of organisms at specific time points. Therefore, the nonselective and reproducible recovery of all existing metabolites while preventing their transformation is the ideal criterion for metabolome sample preparation. We evaluated currently used sampling methods and extraction solvents for global metabolic profiling of a Gram-negative bacterium, Saccharophagus degradans, using gas chromatography-time-of-flight mass spectrometry (GC-TOF MS) with an emphasis on three steps: the sampling method, which consisted of cold methanol quenching or fast filtration; the subsequent washing step; and the extraction solvents. After cold methanol quenching with 70% (v/v) methanol at -70 °C, washing with 2.3% NaCl produced a serious loss of intracellular metabolites. In addition, when cold methanol quenching and fast filtration were compared, severe cell leakage caused by cold methanol quenching resulted in a significant loss of intracellular metabolites, which was confirmed by spectrometric analysis at 260 and 280 nm. Upon evaluation of extraction solvents such as pure methanol (MeOH), acetonitrile/water (50ACN; 1:1, v/v), acetonitrile/methanol/water mixture (AMW; 2:2:1), and water/isopropanol/ methanol (WiPM; 2:2:5). AMW and WiPM were found to be superior extraction solvents for S. degradans based on the total peak intensities of the metabolites, the total number of metabolite peaks, and the reproducibility of recovered metabolite quantities; however, the metabolite profiles differed significantly between AMW and WiPM. This is the first evaluation of each step of sample preparation involved in global scale metabolic analysis by GCTOF MS, which can be used as a model in the preparation of organism-specific samples for metabolome analysis. Metabolomics, which is the study of the change of metabolites in living systems, consists of sample preparation of metabolites, instrumental analysis of the metabolites, and statistical and * To whom correspondence should be addressed. K.H. Kim: phone, +82 2 3290 3028; fax, +82 2 925 1970; e-mail,
[email protected]. O. Fiehn: phone, +1 530 754 8258; fax, +1 530 754 9658; e-mail,
[email protected]. † Korea University. ‡ University of California, Davis. § Inje University.
6660
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
biological interpretation of the analytical results.1,2 Since metabolomic interpretation relies on the metabolites in samples, the accuracy and reliability of sample preparation are important. Although many studies have been conducted to develop techniques for metabolite analysis,3-8 the sample preparation, which generally consists of sampling, quenching, and metabolite extraction, has received little attention.9-11 The first step in sample preparation is quenching, and the most common quenching method used for microorganisms employs 60% (v/v) aqueous methanol at a temperature below -40 °C. This method was first developed for Saccharomyces cerevisiae4,12-14 and has been extensively adapted for other microorganisms without optimization or validation.3,15,16 Cold methanol quenching has been widely reported to cause serious cell leakage9,10,17,18 or cold shock,19 thus resulting in the loss of metabolites from both Grampositive and negative bacteria and yeast.11,20 (1) Ryan, D.; Robards, K. Anal. Chem. 2006, 78, 7954–7958. (2) Fiehn, O. Comp. Funct. Genomics 2001, 2, 155–168. (3) Buchholz, A.; Takors, R.; Wandrey, C. Anal. Biochem. 2001, 295, 129– 137. (4) Castrillo, J. I.; Hayes, A.; Mohammed, S.; Gaskell, S. J.; Oliver, S. G. Phytochemistry 2003, 62, 929–937. (5) Koek, M. M.; Muilwijk, B.; van der Werf, M. J.; Hankemeier, T. Anal. Chem. 2006, 78, 1272–1281. (6) Soga, T.; Ueno, Y.; Naraoka, H.; Ohashi, Y.; Tomita, M.; Nishioka, T. Anal. Chem. 2002, 74, 2233–2239. (7) Teleman, A.; Richard, P.; Toivari, M.; Penttila¨, M. Anal. Biochem. 1999, 272, 71–79. (8) Tweeddale, H.; Notley-McRobb, L.; Ferenci, T. J. Bacteriol. 1998, 180, 5109–5116. (9) Bolten, C. J.; Kiefer, P.; Letisse, F.; Portais, J. C.; Wittmann, C. Anal. Chem. 2007, 79, 3843–3849. (10) Faijes, M.; Mars, A. E.; Smid, E. J. Microb. Cell Fact. 2007, 6, 27. (11) Villas-Boas, S. G.; Hojer-Pedersen, J.; Akesson, M.; Smedsgaard, J.; Nielsen, J. Yeast 2005, 22, 1155–1169. (12) de Koning, W.; van Dam, K. Anal. Biochem. 1992, 204, 118–123. (13) van Dam, J. C.; Eman, M. R.; Frank, J.; Lange, H. C.; van Dedem, G. W. K.; Heijnen, S. J. Anal. Chim. Acta 2002, 460, 209–218. (14) Ewald, J. C.; Heux, S.; Zamboni, N. Anal. Chem. 2009, 81, 3623–3629. (15) Moritz, B.; Striegel, K.; de Graaf, A. A.; Sahm, H. Eur. J. Biochem. 2000, 267, 3442–3452. (16) Winder, C. L.; Dunn, W. B.; Schuler, S.; Broadhurst, D.; Jarvis, R.; Stephens, G. M.; Goodacre, R. Anal. Chem. 2008, 80, 2939–2948. (17) Jensen, N. B. S.; Jokumsen, K. V.; Villadsen, J. Biotechnol. Bioeng. 1999, 63, 356–362. (18) Wittmann, C.; Kromer, J. O.; Kiefer, P.; Binz, T.; Heinzle, E. Anal. Biochem. 2004, 327, 135–139. (19) Link, H.; Anselment, B.; Weuster-Botz, D. Metabolomics 2008, 4, 240– 247. (20) Canelas, A. B.; Ras, C.; ten Pierick, A.; van Dam, J. C.; Heijnen, J. J.; van Gulik, W. M. Metabolomics 2008, 4, 226–239. 10.1021/ac1012656 2010 American Chemical Society Published on Web 07/15/2010
As an alternative to cold methanol quenching, fast filtration with direct extraction has been developed to reduce metabolite loss.9,18 Washing to remove extracellular metabolites is not included in methanol quenching since the quenching solution is discarded. However, during fast filtration, a washing step follows fast filtration prior to the extraction step. The ionic strength of the washing solvent has also been found to be critical for minimization of the loss of metabolites.9,18 The next step after quenching or fast filtration with washing is the extraction of metabolites from the cell pellet using solvents. Because the basis of metabolomics is the unbiased analysis of metabolites, all metabolites need to be completely, nonselectively, and reproducibly extracted by avoiding degradation and conversion to other metabolites.10 Popular metabolite extraction solvents for microorganisms are boiling ethanol8,13,21,22 and perchloric acid.23-25 Recently, the extraction of metabolites with cold methanol has been proposed26 and tested.11 Solvent mixtures have also been applied to bacteria27 and other eukaryotic organisms28,29 to enable extraction of more diverse metabolites for global metabolic profiling than can be obtained using a single solvent. Several studies are available for validation of quenching9,18,20 or fast filtration18,30 and extraction16 for microbial metabolites. However, few such validation studies cover both the quenching or fast filtration steps and the extraction,16 and few studies provide direct comparisons of the quenching method with the fast filtration method.18 Moreover, most validation studies were conducted using specific metabolites instead of global metabolites.16,29 Therefore, direct comparison of the quenching and fast filtration methods and evaluation of the performance of extraction solvents based on the global metabolite profiling by gas chromatographytime-of-flight mass spectrometry (GC-TOF MS) will provide valuable information. In this study, the performance of the cold methanol quenching and fast filtration were compared and the effects of washing and the efficacies of metabolite extraction with solvent mixtures were investigated to enable the metabolic profiling of a Gram-negative Saccharophagus degradans 2-40. S. degradans has been shown to be capable of degrading a wide range of polysaccharides, and its full genome was recently sequenced.31 The unique features of this marine bacterium enable its application to the saccharification of lignocellulose32 and marine (21) Entian, K.-D.; Zimmermann, F. K.; Scheel, I. Mol. Gen. Genet. 1977, 156, 99–105. (22) Gonzalez, B.; Franc¸ois, J.; Renaud, M. Yeast 1997, 13, 1347–1356. (23) Theobald, U.; Mailinger, W.; Reuss, M.; Rizzi, M. Anal. Biochem. 1993, 214, 31–37. (24) Weuster-Botz, D. Anal. Biochem. 1997, 246, 225–233. (25) Buchholz, A.; Hurlebaus, J.; Wandrey, C.; Takors, R. Biomol. Eng. 2002, 19, 5–15. (26) Maharjan, R. P.; Ferenci, T. Anal. Biochem. 2003, 313, 145–154. (27) Rabinowitz, J. D.; Kimball, E. Anal. Chem. 2007, 79, 6167–6173. (28) Weckwerth, W.; Wenzel, K.; Fiehn, O. Proteomics 2004, 4, 78–83. (29) Lee, D. Y.; Fiehn, O. Plant Methods 2008, 4, 7. (30) Barsch, A.; Patschkowski, T.; Niehaus, K. Funct. Integr. Genomics 2004, 4, 219–230. (31) Weiner, R. M.; Taylor, L. E., II; Henrissat, B.; Hauser, L.; Land, M.; Coutinho, P. M.; Rancurel, C.; Saunders, E. H.; Longmire, A. G.; Zhang, H.; Bayer, E. A.; Gilbert, H. J.; Larimer, F.; Zhulin, I. B.; Ekborg, N. A.; Lamed, R.; Richardson, P. M.; Borovok, I.; Hutcheson, S. PLoS Genet. 2008, 4, e1000087. (32) Lynd, L. R.; Laser, M. S.; Bransby, D.; Dale, B. E.; Davison, B.; Hamilton, R.; Himmel, M.; Keller, M.; McMillan, J. D.; Sheehan, J.; Wyman, C. E. Nat. Biotechnol. 2008, 26, 169–172.
macroalgae33 for biofuels production, and it has even been considered for commercialization.34 Because only its genomic data is currently available, global metabolic profiling will be useful to have insight into its metabolic pathways and regulations such as the recent studies conducted to profile S. degradans under various carbon sources35,36 by employing cold methanol quenching. The systematic evaluation of metabolite sample preparation provided here will be an effective tool for the global metabolic analysis of S. degradans as well as an example of optimizing metabolite sampling methods. EXPERIMENTAL SECTION Strain, Media, and Culture Conditions. S. degradans 2-40 (ATCC 43961) was grown at 27 °C on half-strength marine agar (18.7 g L-1 Marine Broth (BD, Franklin Lakes, NJ) and 1.5% (w/v) agar (BD)) or in minimal broth (23 g L-1 Instant Ocean sea salt (Aquarium Systems, Mentor, OH), 1 g L-1 yeast extract, 50 mM Tris buffer (pH 7.4), and 0.05% (w/v) NH4Cl) while shaking at 200 rpm.31 To obtain metabolites samples, cells were grown in minimal media broth supplemented with 0.2% (w/v) glucose for 9 h until the midexponential phase. Metabolite Sample Preparation. Cold methanol quenching with washing or without washing was compared with fast filtration. The use of four different extraction solvents for metabolite recovery was also compared. Approximately 0.2 mg of cell mass mL-1, which was equivalent to 108 cells mL-1, and six replicate samples were taken for each analysis. Cold Methanol Quenching. Experiments were conducted as previously described.29 Briefly, l mL of cell culture was injected into 1 mL of 70% (v/v) aqueous methanol at -70 °C. After centrifugation at 16 100 relative centrifugal force (rcf) for 5 min, the cell pellet was collected. The cells were then disrupted in a mixer mill (Retsch, Haan, Germany) using steel balls. Alternatively, in case the effect of washing after cold methanol quenching was examined, the cell pellet was washed twice with 1 mL of 2.3% (w/v) NaCl at 4 °C and then vacuum-dried for 6 h prior to cell disruption. The disrupted cells were then extracted with 500 µL of extraction solvent such as pure methanol (MeOH), acetonitrile/ water mixture (50ACN; 1:1, v/v), acetonitrile/methanol/water (AMW; 2:2:1, v/v/v), or water/isopropanol/methanol (WiPM; 2:2: 5, v/v/v) at 0 °C. The extract was then concentrated to dryness in a vacuum concentrator. To fractionate out complex lipids and waxes, the residue was resuspended in 500 µL of fresh 50ACN and centrifuged at 16 100 rcf for 5 min. The supernatant was concentrated to dryness and kept at -80 °C prior to derivatization and analysis by GC-TOF MS. Fast Filtration. Fast filtration as an alternative to cold quenching was conducted using a slight modification of previous methods.18,27 Cells were harvested by vacuum filtration of 1 mL of cell culture using a nylon membrane filter (0.45 µm pore size, 30 mm diameter; Whatmann, Piscataway, NJ) and washed with 8, 12, 16, or 20 mL of 2.3% NaCl at room temperature. The entire (33) E.I. du Pont de Nemours and Company. http://arpa-e.energy. gov/ProgramsProjects/BroadFundingAnnouncement/BiomassEnergy/ MacroAlgaeButanol.aspx (accessed July 2010). (34) Zymetis. http://www.zymetis.com (accessed April 2010). (35) Shin, M. H.; Lee, D. Y.; Skogerson, K.; Wohlgemuth, G.; Choi, I.-G.; Fiehn, O.; Kim, K. H. Biotechnol. Bioeng. 2010, 105, 477–488. (36) Shin, M. H.; Lee, D. Y.; Wohlgemuth, G.; Choi, I.-G.; Fiehn, O.; Kim, K. H. New Biotechnol. 2010, 27, 156–168.
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
6661
procedure described above was completed in less than 30 s. The filter loaded with cells was then quenched in a six-well culture dish filled with 1 mL of extraction solvent at 0 °C. After stirring the culture dish containing cell-loaded filters for 15 min at 0 °C, the cell-solvent mixture was removed from the dish and set aside. An additional 1 mL of solvent was then used to wash the filter, after which the wash liquid was combined with the initial cell-solvent mixture. Next, the combined cell-solvent mixture and wash liquid was centrifuged at 16 100 rcf for 5 min and then concentrated to dryness in a vacuum concentrator, re-extracted with 500 µL of 50ACN to remove lipids and wax, and then concentrated to dryness. Metabolite Analysis. For derivatization of metabolites prior to GC-TOF MS analysis, the metabolite extracts were reconstituted with 5 µL of methoxyamine hydrochloride (Pierce, Rockford, IL) in pyridine at a concentration of 40 mg mL-1 at 30 °C for 90 min and then mixed with 45 µL of N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA; Pierce) at 37 °C for 30 min. A mixture of fatty acid methyl esters was then added to the extract as retention index markers. GC-TOF MS analysis was conducted using an Agilent 6890 GC (Hewlett-Packard, Atlanta, GA) coupled to a Pegasus III TOF mass spectrometer (Leco, St. Joseph, MI). An Rtx-5Sil MS column with a length of 30 m, an inner diameter of 0.25 mm, and a 0.25 µm film thickness (Restek, Bellefonte, PA) and an additional 10 m long integrated guard column were used. For the GC analysis, the metabolite samples were initially subjected to 50 °C, which was held for 1 min, after which it was ramped to 330 °C at 20 °C min-1, where it was finally held for 5 min. Mass spectra were acquired in a scan range of 85 to 500 m/z at an acquisition rate of 10 spectra s-1. The ionization mode was subjected to electron impact at 70 eV, and the ion source was set at 250 °C. Metabolite Identification and Statistical Analysis. The spectra over the range of 85-500 m/z were preprocessed by the Leco Chroma TOF software (version 2.32; St. Joseph, MI) using automated peak detection and mass spectral deconvolution. These spectra were further processed using the in-house programmed database, BinBase.29 The BinBase identified metabolite peaks by matching their mass spectra and retention indices with the customized reference mass spectral libraries that were acquired using authentic standards under identical data acquisition parameters of the Fiehn library, which hosts more than 1 000 unique metabolites and 2 212 unique spectra.37 Since the same culture was used for the entire experiment, the normalization of metabolite data by cell dry weight35 was not necessary. The resulting data sets were imported into Statistica (version 7.1, StatSoft, Tulsa, OK) for multivariate and univariate statistical analyses.29 Determination of Residual Extracellular Culture Components after Washing. To determine the amount of extracellular culture components carried over to the final intracellular metabolite samples after washing, three isotopes and xylitol were used as the internal standards (Method S-1 in the Supporting Information). Cell Leakage Test. To evaluate possible leakage of cellular materials following sample preparation, cells were treated differently for each sample preparation method and analyses included (37) Kind, T.; Wohlgemuth, G.; Lee, D. Y.; Lu, Y.; Palazoglu, M.; Shahbaz, S.; Fiehn, O. Anal. Chem. 2009, 81, 10038–10048.
6662
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
negative and positive controls (Method S-2 in the Supporting Information).38 RESULTS AND DISCUSSION Effect of Washing in Cold Methanol Quenching. Quenching a cell culture in cold methanol to stop further metabolic action is one of the most commonly used methods during the preparation of microbial metabolite samples.10,16,39,40 Because cold methanol quenching does not usually involve washing, contamination or interference of the culture medium components with intracellular metabolites may be possible. However, no studies have evaluated the incorporation of washing into the quenching method in detail, and only the loss of metabolites during washing has been investigated to date.9,41 In this study, 1 mL cell culture at 27 °C was immediately transferred into an equal volume of 70% methanol at -70 °C. A washing step was introduced after quenching and rapid centrifugation, during which the cell pellet was washed twice with 1 mL of 2.3% NaCl at 4 °C. The sample was then extracted with WiPM and analyzed by GC-TOF MS. As shown in Table S-1 in the Supporting Information, a total of 106 intracellular metabolites that were conserved in at least 80% of every biological design class in this study were identified using the BinBase algorithm.29,42 The intracellular metabolites of S. degradans included various chemical entities such as amino acids, polyamines, organic acids, phosphates, sugars, and fatty acids. To compare the metabolomic differences obtained using quenching only and quenching with washing, principal component analysis (PCA) and independent t-tests were conducted. The PCA model for the comparison of metabolite profiles of the cold methanol methods with and without washing exhibited a good fit of R2X, 0.74, and a prediction of Q2, 0.86, based on cumulative values up to PC 2. In the score plot, PC 1 (t1) explained most of the variations induced by the washing and nonwashing quenching methods (Figure S-1a in the Supporting Information). There was also a clear separation between the metabolite profiles of the samples prepared using both methods, which were analyzed based on the PCA of 106 identified metabolites from six biological replicates. While the scores reflect the contribution of each PC in the samples, the loadings reflect the importance and weight of the original variables (Figure S-1b in the Supporting Information). On the basis of the differential distribution on the loading plot, metabolites with high loadings in PC 1 (p1), which corresponded to samples that were quenched without washing, were more hydrophilic. Conversely, hydrophobic metabolites such as stearic acid, palmitic acid, and montanic acid were found to be relatively predominant in the washed samples. These results imply that a large amount of hydrophilic metabolites were washed out during the washing step. (38) Amor, K. B.; Breeuwer, P.; Verbaarschot, P.; Rombouts, F. M.; Akkermans, A. D. L.; De Vos, W. M.; Abee, T. Appl. Environ. Microbiol. 2002, 68, 5209–5216. (39) Wellerdiek, M.; Winterhoff, D.; Reule, W.; Brandner, J.; Oldiges, M. Bioprocess. Biosyst. Eng. 2009, 32, 581–592. (40) Villas-Boas, S. G.; Bruheim, P. Anal. Biochem. 2007, 370, 87–97. (41) Taymaz-Nikerel, H.; de Mey, M.; Ras, C.; ten Pierick, A.; Seifar, R. M.; van Dam, J. C.; Heijnen, J. J.; van Glilik, W. M. Anal. Biochem. 2009, 386, 9–19. (42) Fiehn, O.; Wohlgemuth, G.; Scholz, M.; Kind, T.; Lee, D. Y.; Lu, Y.; Moon, S.; Nikolau, B. Plant J. 2008, 53, 691–704.
Figure 1. Effect of washing after quenching on the recovery yield of selected metabolites from different chemical classes when intracellular metabolites were extracted with WiPM. Representative metabolites (11) were selected to cover a wide range of chemical classes including amino acids, polyamines, organic acids, sugars, phosphates, and fatty acids. To aid in visualization of the data, peak intensities of some metabolites were multiplied by 10-1 000, and t-tests were conducted to evaluate statistical significance between the two methods with and without washing at p < 0.05 (*) and p < 0.01 (**).
To evaluate the effect of washing after quenching on the recovery of intracellular metabolites, the intracellular metabolites, which were not detected as culture broth components or extracellular metabolites at all or were detected in only small abundance (e.g., less than 5 or 10% of intracellularly detected abundance), were selected for comparison. Figure 1 shows the results of t-tests conducted to evaluate the peak intensities of the selected metabolites. The subsequent washing following quenching led to decreased abundance of many metabolites including amino acids, sugars, sugar phosphates, and polyamines when compared to samples that were quenched without washing. In particular, spermidine, glutamic acid, and palmitoleic acid decreased significantly when washing was included in the process. However, some metabolites such as glycerol-R-phosphate, dihydroabietic acid, and linoleic acid were less affected by washing. We assumed that many metabolites, especially polar compounds, were removed by washing with NaCl. The significant loss of intracellular metabolites by washing in S. degradans could be related to the cell leakage that might have occurred during cold methanol quenching.9,10,17,18 In other words, the damaged membranes of cold methanol-quenched cells no longer had the capability to confine metabolites inside cells during washing. The original intention of including a washing step to prevent contamination of extracellular culture components was overwhelmed by the significant loss of metabolites (Figure 1), as well as by the shifted metabolite profiles of the washed and quenched samples when compared to those of unwashed samples (Figure S-1a,b in the Supporting Information). Effect of Washing Volume during Fast Filtration. The fast filtration has been used as an alternative to the quenching due to cold methanol quenching causing cold shock, especially to Gram-
negative bacteria such as Escherichia coli.9,18,19,30,39,43 Unlike the cold methanol quenching method, which relies on the extreme temperature to stop cellular metabolism, the metabolic arrest induced by the fast filtration depends on the rapid separation of cells and subsequent washing and extraction. Therefore, the possibility of cold shock and cell leakage could be lower than that of cold methanol quenching. However, the loss of metabolites by washing could be an issue since the results of the present study demonstrated that serious loss of metabolites occurred in response to washing in conjunction with cold methanol quenching. The use of different amounts of washing solvent (2.3% NaCl) was employed to investigate the effects of various volumes of washing solvent on the recovery of intracellular metabolites (Figure S-2a in the Supporting Information). For comparison of different volumes, intracellular metabolites were selected using the same criteria as in Figure 1. Although ribose-5-phosphate and linoleic acid shown in Figure 1 were not detected in the experiments for Figure S-2a in the Supporting Information, 2-monopalmitin was detected in the experiments for Figure S-2a in the Supporting Information. In general, the abundance of metabolites recovered by the fast filtration and subsequent extraction with WiPM was not significantly influenced by the amounts of washing solvent used. This low impact of washing volume on the metabolite recovery yield obtained using the fast filtration was probably due to the lower loss of intracellular metabolites through the lower cell leakage problem observed in response to the fast filtration when compared with cold methanol quenching. The main purpose of washing during fast filtration is to minimize contamination of intracellular metabolites with extracellular culture components.9 Alanine-d4, glutamic-acid-d5, asparagine-d8, and xylitol, which are not found in either culture broth or as intracellular metabolites, were added as the internal standards to cell culture samples prior to subjecting the samples to the fast filtration method, and the residual amount of each internal standard was then analyzed by washing with different volumes of washing solvent, extraction, and GC-TOF MS analysis. The general trend was that the amounts of residual standards decreased with increasing washing volume (Figure S-2b, in the Supporting Information). However, the differences in the residual amounts obtained using different washing volumes were not significant at the 5% significance level. The residual amounts of all of the standards were higher when the highest volume of solvent (20 mL) was used than when 16 mL was used. Therefore, 16 mL was selected as the optimal volume for washing during fast filtration. Comparison of Quenching and Fast Filtration in Metabolite Recovery. For the unbiased analysis of the metabolome,44,45 it is important to recover all the existing metabolites in high quantities when conducting metabolic analysis, including those present in relatively low concentrations. In this study, the peak intensities of representative intracellular metabolites from various chemical classes including amino acids, organic acids, amines, sugars, and fatty acids obtained using the two optimized quench(43) Bolten, C. J.; Wittmann, C. Biotechnol. Lett. 2008, 30, 1993–2000. (44) Oldiges, M.; Lutz, S.; Pflug, S.; Schroer, K.; Stein, N.; Wiendahl, C. Appl. Microbiol. Biotechnol. 2007, 76, 495–511. (45) van der Werf, M. J.; Overkamp, K. M.; Muilwijk, B.; Coulier, L.; Hankemeier, T. Anal. Biochem. 2007, 370, 17–25.
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
6663
ing and fast filtration methods in this study were determined. Figure 2 shows the effects of the two methods on the recovery yields of diverse representative metabolites that cover wide ranges of chemical classes.17 All of the amino acids listed in Figure 2a were present in lower abundance with the cold methanol quenching method than with the fast filtration method. Specifically, aspartic acid, glutamine, and tyrosine were present in significantly lower quantities in samples obtained using the cold methanol quenching. However, as the polarity of amino acid metabolites including alanine, isoleucine, and proline decreased, the difference in the peak intensity of samples obtained by the cold methanol quenching and the fast filtration became less noticeable. Similarly, most metabolites belonging to organic acids, amines, sugars, and sugar derivatives (Figure 2b) exhibited significantly lower peak intensities when the cold methanol quenching method was conducted than when the fast filtration method was conducted, except for fructose, adenosine, and gluconic acid. Comparison of the recovery yields of fatty acids obtained using the two methods (Figure 2c) revealed that only palmitic acid, pelargonic acid, and myristic acid were significantly higher when the fast filtration method was used. Conversely, the levels of monopalmitin-1glyceride and 2-monopalmitin were slightly higher when the cold methanol quenching method was used. Even when washing was eliminated in the cold methanol quenching, a significant loss of metabolites occurred during the use of this method when compared to the fast filtration. When using cold methanol during quenching, the possible effects of a fluidity increase or thickness decrease in cell membranes in response to alcohol46,47 and an increased membrane permeability caused by cold or osmotic shock19,48 were considered to be the main causes of the low metabolite recovery. Specifically, the lower peak intensities of metabolites in samples obtained using the cold methanol quenching method, which were predominantly composed of hydrophilic compounds such as amino acids, organic acids, amines, and sugars, can be partly attributed to the preferential extraction of hydrophilic metabolites by methanol through permeabilized membranes during the cold quenching. In addition, the higher loss of hydrophilic metabolites in this study may have been enabled through hydrophilic channels that were found to be formed by the crystallization of liquid-like lipids within cell membranes as a result of cold shock in E. coli.49 Cell Leakage Test. Earlier in the study, a lower metabolite recovery was observed in response to cold methanol quenching with washing compared to quenching without washing and in response to quenching without washing when compared to the fast filtration. The overall low recovery yields of intracellular metabolites in response to cold methanol quenching when compared to the fast filtration method was likely due to the increased release of intracellular materials through the permeabilized or disrupted cytoplasmic membrane and outer membrane of cold methanol-quenched Gram-negative S. degradans. Therefore, the release of nucleic acids and proteins was quantified by measuring the absorbance of the supernatant of cell suspensions (46) Chin, J. H.; Goldstein, D. B. Mol. Pharmacol. 1977, 13, 435–441. (47) Ly, H. V.; Longo, M. L. Biophys. J. 2004, 87, 1013–1033. (48) Canelas, A. B.; ten Pierick, A.; Ras, C.; Seifar, R. M.; van Dam, J. C.; van Gulik, W. M.; Heijnen, J. J. Anal. Chem. 2009, 81, 7379–7389. (49) Leder, I. G. J. Bacteriol. 1972, 111, 211–219.
6664
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
Figure 2. Comparison of the effects of quenching and fast filtration methods on the recovery yields of (a) amino acids; (b) organic acids, amines, and sugars; and (c) fatty acids from S. degradans when intracellular metabolites were extracted using WiPM. To aid in visualization of the data, peak intensities of some metabolites were multiplied by 0.1-10.
treated by cold quenching or fast filtration at 260 and 280 nm, respectively.
As shown in Figure S-3 in the Supporting Information, the supernatant obtained using both the cold methanol quenching and the fast filtration showed higher values than the negative control that was the supernatant of a cell suspension in 2.3% NaCl. These findings indicated that both the fast filtration and the cold methanol quenching caused the release of cellular materials through cell membranes and cell walls. The absorbance of the supernatant from the fast filtration or the cold methanol quenching was significantly lower than that of the positive control (i.e., the supernatant of a heated cell suspension). In comparison of the fast filtration and cold methanol quenching methods, the absorbance of the supernatant obtained using the fast filtration was significantly lower than that of the supernatant obtained using the cold methanol quenching. This indicated that higher cell leakage was induced by the cold methanol quenching. To minimize the loss of intracellular metabolites that occurred as a result of cell leakage by S. degradans, the fast filtration method would be better than the cold methanol quenching method in terms of metabolite recovery yield. However, since the fast filtration method is slower (i.e., < 30 s in the current study) than the immediate cold quenching, fast filtration needs to be further improved to be expedited either for rapid sampling in dynamic experiments or for steady-state measurements with short turnover times.10,41 Also, care should be taken in the selection of sampling methods depending on the primary focus of the metabolic profiling. Optimization of Extraction Solvent. Ideally, an extraction solvent covering a broad range of chemical properties of metabolites should be used to enable extraction of all metabolites in high yield regardless of their chemical classes, thus ensuring that the metabolites will be successfully and reproducibly released from the cell interior and from the macromolecular cell structures.16,48 However, such extraction solvents for global metabolite profiling are not easy to generate since each extraction solvent has its own preference for certain chemical classes. To acquire an optimal solvent system for the global metabolite profiling of S. degradans, we tested four different extraction solvent systems known to have high extracting efficiency for microbial metabolome analysis,10,16,27,29,42 MeOH, 50ACN, AMW, and WiPM. We then evaluated the performance of these extraction solvents based on three evaluation criteria, the number of peaks detected for each extraction solvent, the peak intensity of structurally identified compounds, and the reproducibility of metabolite quantification. A total of 371 peaks (88 knowns and 283 unknowns) were detected by GC-TOF MS across all four extraction solvents (Table S-2 in the Supporting Information). The number of peaks detected was counted in the raw data before replacing missing values using a postprocessing module.42 Among these solvents, AMW and WiPM showed similar, but the highest and the second highest numbers of peaks, for both identified and unidentified peaks, respectively. On the basis of the use of extracting diverse metabolites as the most important criterion for the extraction solvent, AMW and WiPM were better extraction solvents than MeOH and 50ACN since they were capable of extracting more metabolites than the other two solvents. As the second criterion for extraction solvents, when metabolite peak intensities that were summed based on their dependence on their chemical classes, in which the peak intensity of each
metabolite was subtracted by its mean and then divided by its standard deviation for the normalization by unit variance scaling,50 were compared between the extraction solvents. WiPM showed the highest peak intensity for the total identified metabolites, with a mean of the summation value, 29.9. The mean summed value for AMW, MeOH, and 50ACN were -2.3, -12.0, and -15.6, respectively. Each extraction solvent had preferentially higher extraction efficiency on a certain individual metabolite; however, no specific trend of peak intensity differences was shown in any chemical class of metabolites between the extraction solvents except for WiPM showing the highest peak intensities in almost every class (Figure S-4 in the Supporting Information). For example, inosine was most efficiently extracted with 50ACN, while lignoceric acid was most efficiently extracted with AMW, stearic acid was most efficiently extracted with MeOH, and gluconic acid was most efficiently estimated with WiPM. The third criterion for the selection of optimal extraction solvents is the reproducibility of the quantification of each metabolite. The extraction solvents were assessed based on the distribution frequency, of which identified metabolites were quantified in 10% coefficient of variation (CV) precision intervals. The median %CV values of the identified metabolites for AMW, WiPM, MeOH, and 50ACN were 27.8, 33.8, 34.4, and 42.1%, respectively (Figure S-5 in the Supporting Information). Although AMW showed the highest reproducibility with respect to extraction yields for the identified compounds, with the exception of samples obtained using 50ACN, no significant differences were observed in response to the different extraction solvents. Overall, AMW extracted the highest number of metabolites, followed by WiPM. In addition, WiPM showed the highest peak intensity among the extraction solvents, followed by AMW. Moreover, AMW had the lowest median %CV, indicating the highest reproducibility with respect to the extraction yield of metabolites. This was followed by WiPM. Therefore, AMW and WiPM were both considered to be superior solvents for the extraction of metabolites from S. degradans when compared to MeOH and 50ACN. Among the well-known solvents used in microbial metabolite extraction such as perchloric acid, alkaline, boiling ethanol, chloroform/methanol, and cold methanol, most recent studies have considered cold methanol to be the best extraction solvent.10,11,26,48 However, only a few studies have provided a direct comparison of extraction efficiency between various organic solvents including methanol, acetonitrile, and isopropanol,27,29,51 especially when using the fast filtration method.27 In this study, AMW and WiPM were both found to be superior extraction solvents for S. degradans, while AMW was shown to be advantageous for extraction of metabolites related to the central carbon metabolism of E. coli when compared to other solvents such as chloroform/methanol, methanol/water, and ethyl acetate.27 However, in yeast quenched with pure methanol, boiling ethanol or chloroform/methanol was a better extraction solvent, while acidic AMW performed poorly.48 Therefore, extraction solvents need to be optimized specifically for the type of microorganism and the sample preparation method being employed. (50) van den Berg, R. A.; Hoefsloot, H. C. J.; Westerhuis, J. A.; Smilde, A. K.; van der Werf, M. J. BMC Genomics 2006, 7, 142. (51) Want, E. J.; O’Maille, G.; Smith, C. A.; Brandon, T. R.; Uritboonthai, W.; Qin, C.; Trauger, S. A.; Siuzdak, G. Anal. Chem. 2006, 78, 743–752.
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
6665
had the poorest metabolite recovery yield. As evaluated herein, although AMW and WiPM showed almost equally superior performance with respect to the three criteria considered, the hierarchical clustering of the identified intracellular metabolites showed significantly different profiles between AMW and WiPM. The metabolite profiles obtained with 50ACN and AMW shared more common features, and the extraction solvents, MeOH and WiPM, showed similar metabolite profiles with respect to the hierarchical clustering. Therefore, these results indicate that, in addition to the overall quantitative aspect of metabolite recovery such as the recovery yield of metabolites, reproducibility, and the number of detected metabolites, the metabolite profile can also be significantly influenced by the extraction solvent. CONCLUSIONS This is the first study conducted to evaluate and optimize each step of sample preparation for the microbial metabolome based on global scale metabolic analysis by GC-TOF MS. For the global metabolite profiling of a Gram negative S. degradans, washing with 2.3% NaCl should be avoided after cold methanol quenching due to the serious loss of metabolites during washing. Although the washing step was skipped during cold methanol quenching, the metabolite recovery yield obtained using cold methanol quenching was significantly lower than that obtained using fast filtration due to serious cell leakage caused by the cold methanol quenching. On the basis of evaluation criteria of the number of detected peaks, the peak intensity of metabolites, and the reproducibility of the extraction recovery yield of metabolites, AMW and WiPM were found to be superior extraction solvents for S. degradans. However, since the metabolite profiles obtained by AMW and WiPM differed, the extraction solvent needs to be carefully selected depending on the primary purpose of the metabolic profiling.
Figure 3. Clustered heat map of intracellular metabolites of S. degradans extracted using MeOH, 50ACN, AMW, and WiPM. Symbols, * and **, represent significant differences in the metabolites as determined by analysis of variance (ANOVA) of the samples obtained using different extraction solvents at the 5% and 1% significance level, respectively. 51
Figure 3 shows the hierarchical clustering of identified intracellular metabolites for the present study. As shown in the sums of peak intensities normalized by employing unit variance scaling as the second criterion for evaluation of the extraction solvent, the hierarchical clustering also clearly indicated that WiPM gave the highest metabolite recovery yield, while 50ACN
6666
Analytical Chemistry, Vol. 82, No. 15, August 1, 2010
ACKNOWLEDGMENT This work was supported by a grant (Grant 2008-N-B108-P01-3-050) from the New and Renewable Energy Technology Development Project from the Korea Energy Management Corporation and also a grant from Ministry for Food, Agriculture, Forestry and Fisheries, Korea. The authors are grateful to Gert Wohlgemuth at UC Davis Genome Center for establishing the metabolite database. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review May 14, 2010. Accepted June 29, 2010. AC1012656
