Min-Kyu Oh, Department of Chemical and Biological Engineering, Korea University, Seoul 136-713, Korea. E-mail: firstname.lastname@example.org
The analytical study of intracellular (IC) metabolites has developed with advances in chromatography-linked mass spectrometry and fast sampling procedures. We applied the IC metabolite analysis to characterize the role of GCY1 in the glycerol (GLY) catabolic pathway in Saccharomyces cerevisiae.
Methods and Results
Strains with disrupted or overexpressing GLY catabolic genes such as GCY1, DAK1 and DAK2 were constructed. The strains were cultivated under different aeration conditions and quickly quenched using a novel rapid sampling port. IC concentrations of GLY, dihydroxyacetone (DHA), glycerol 3-phosphate and dihydroxyacetone phosphate were analysed in the strains by gas chromatography/mass spectrometry. DHA was not detected in the gcy1 gene-disrupted strain but accumulated 225·91 μmol g DCW−1 in a DHA kinase gene-deficient strain under micro-aerobic conditions. Additionally, a 16·1% increase in DHA occurred by overexpressing GCY1 in the DHA kinase-deficient strain.
Metabolic profiling showed that the GCY1 gene product functions as a GLY dehydrogenase in S. cerevisiae, particularly under micro-aerobic conditions.
Significance and Impact of the Study
Metabolic profiling of the GLY dissimilation pathway was successfully demonstrated in S. cerevisiae, and the function of GCY1 was explained by the results.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Microbial metabolomics has received much attention recently because it provides important information on a wide range of microbial topics from medical applications such as discovering biomarkers to industrial strain development. Metabolomics, or metabolic profiling, refers to the comprehensive (quantitative and qualitative) analysis of a set of low molecular weight metabolites that exist inside and outside of growing cells at a certain time during culture. The information about metabolites itself or combined with other omics data can play critical roles in the understanding of microbial metabolism and its regulation (Theobald et al. 1993; Kresnowati et al. 2006). Quantitative analysis of metabolites has been traditionally carried out using enzyme-based assays (Bergmeyer 1984; Theobald et al. 1997; Hajjaj et al. 1998). However, there are several limitations to these assays, including the large volume of sample required, only one or few metabolites are detected per experiment, and the complex cellular matrix could interfere with detecting an accurate metabolite level. To overcome these problems, metabolic profiling based on a fast sampling procedure has been introduced, which instantly deactivates endogenous enzymes and extracts and quantifies metabolites with analytical tools (Weibel et al. 1974). Several manually operated devices for fast sampling from a bioreactor have been developed (de Koning and van Dam 1992; Theobald et al. 1993). But, manual operation is laborious, and results vary depending on operator skill. Electrically operated sampling has been developed but is still limited because of the many manual steps (Larsson and Törnkvist 1996; Lange et al. 2001). Then, fully automated sampling systems were introduced using a sampling nozzle attached to a bioreactor. Fully automated sampling devices enable rapid perturbation experiments outside of a bioreactor and subsequent rapid sampling (Visser et al. 2002; Mashego et al. 2006).
As sampling devices have been developed, analytical tools have been developed as well. Mass spectrometry (MS) is well established as an analytical instrument for diverse metabolites based on the ion molecular mass-to-charge (m/z) ratio. Various mass analyser configurations such as quadrupole, time-of-flight (TOF) and Fourier transform ion cyclotron resonance combined with chromatographic technique are now available for metabolic profiling. A quadrupole mass analyser enables more sensitive analyses of a few designated metabolites, whereas a TOF analyser can analyse a wide range of metabolites (Dunn and Ellis 2005; Dunn et al. 2005). The sensitive and robust technique of chromatography coupled with MS, such as liquid chromatography/mass spectrometry (LC/MS), gas chromatography/mass spectrometry (GC/MS) and capillary electrophoresis/mass spectrometry (CE/MS), allow simultaneous quantification of many different metabolites and require small sample volumes (<10 μl) with detection limits in the pmol range (Cech and Enke 2001; Buchholz et al. 2002; van Dam et al. 2002; Villas-Bôas et al. 2005; Wu et al. 2005). Among them, GC/MS is commonly used for sensitive measurements of volatile metabolites, but nonvolatile metabolites can also be measured via the derivatization method (Koek et al. 2006; Cipollina et al. 2009). Therefore, efficient and reproducible derivatization methods have also been developed for successful GC/MS-based metabolic profiling (Zamboni et al. 2009). Metabolic profiling using GC/MS has been successfully used to reveal metabolite levels in vivo and then applied to estimate fluxes when combined with a computational approach in the metabolic engineering field (Koek et al. 2006; Zamboni et al. 2009).
Glycerol (GLY) is a by-product of the biodiesel industry, and converting GLY to useful chemicals has been attempted in various micro-organisms (Lee et al. 2000; Yazdani and Gonzalez 2008; Blankschien et al. 2010; Jung et al. 2011). However, utilizing GLY in Saccharomyces cerevisiae as a carbon source is relatively inefficient and has not received much attention. Two GLY dissimilation pathways exist in S. cerevisiae (Fig. 1); GLY can be converted to glycerol 3-phosphate (G3P) and then to dihydroxyacetone phosphate (DHAP) by Gut1p (GLY kinase) and Gut2p (G3P dehydrogenase) under aerobic condition (Pavlik et al. 1993; Ronnow and Kielland-Brandt 1993). Rsf1p and Rsf2p are responsible regulators for the transcriptional induction of GUT1 and GUT2 genes, which allows respiratory growth of S. cerevisiae in GLY medium (Lu et al. 2003, 2005). Alternatively, GLY is converted to dihydroxyacetone (DHA) by GLY dehydrogenase and to DHAP by Dak1p and Dak2p (DAK kinases) under anaerobic conditions (Molin et al. 2003). Little has been known about the regulation of this pathway in S. cerevisiae, except that the genes involved in this pathway are activated under stress conditions or in low-oxygen condition (Norbeck and Blomberg 1997; Costenoble et al. 2000). Among these GLY catabolic enzymes, GLY dehydrogenase has not been experimentally confirmed in S. cerevisiae. So far, GCY1 has been assigned as a gene encoding a putative GLY dehydrogenase in S. cerevisiae because of its 65% sequence identity to GLY dehydrogenase of Aspergillus niger species and a similar mRNA expression level change with DHA kinase by osmotic stress in S. cerevisiae (Norbeck and Blomberg 1997). The GCY1 gene has been characterized as a gene encoding an aldo-keto reductase that reacts with various aldehyde substrates (Chang et al. 2007).
In this study, metabolic profiling was established with fast sampling (<0·5 s) including cold methanol quenching, ethanol extraction, derivatization and quantitative analysis by GC/MS. Three recombinant strains were constructed by deletion and/or overexpression of the GCY1, DAK1 and DAK2 genes in S. cerevisiae to identify the role of the GCY1 gene. We measured the metabolite levels in the GLY pathway at the exponential growth phase under aerobic and micro-aerobic conditions. We concluded from the metabolic profiling data that the GCY1 gene product functions as a GLY dehydrogenase in S. cerevisiae under micro-aerobic conditions.
Materials and methods
Strain development and plasmid construction
Saccharomyces cerevisiae YPH499 (MATa ura3-52 lys2-801_amber ade2-10_orche trp1-Δ63 his3-Δ200 leu2-Δ1) was genetically modified to identify the role of GCY1 (Table 1). Escherichia coli strain DH5α was used for plasmid construction.
Table 1. Saccharomyces cerevisiae strains and primer sequences used in this study
The underlined sequences represent the introduced restriction enzyme sites for DNA recombination.
499 wild type
MATa ura3-52 lys2-801amberade2-101orchetrp1-Δ63 his3-Δ200 leu2-Δ1
499 with gcy1Δ
499 with dak1Δ and dak2Δ
In this study
499 with dak1Δ and dak2Δ, pRS423-HIS-GCY1
5′-CTC GAG ATG CCT GCT ACT TTA CAT-3′
5′-CCT AGG TTA CTT GAA TAC TTC GAA-3′
5′ CAT CAA AGA ATA AGA TTA CAT TCT ATA TCT AAG ACT AAA TTT TAA ATG CGT ACG CTG CAG GTC GAC 3′
5′ TAT ATA TAT CAT AAG TAT CTT GAT ATG TAT TCG TGA GCC AAG TAC TTA ACT AGT GGA TCT GAT ATC 3′
5′ TAT TTG TTA CTG TCA ATT GTC TGG C 3′
5′ ATC GGT CTT TCT GAA GAG AAT TTT T 3′
5′ GGT TAC ACT TTA GTG GCA GGA GTT A 3′
5′ TGT TAA CGG TAT TTC CTT CTT GTT C 3′
5′ TAC CTC AAA ATC TGA CAA AAC CCA ACT ACA ATT GAC TAA ATA ATC ATG CGT ACG CTG CAG GTC GAC 3′
5′ GCC TTT AAA GCT GTT ATG TTT GGC TTC TAG TGT GTA CGA GCA ATT CTA ACT AGT GGA TCT GAT ATC 3′
5′ AAT TTA TCT ACA TAT TAC AAT CAT ACG AG AAA CAC GCA A AA ACA ATT AAC TAG TGG ATC TGA TAT C 3′
5′ CTT CTC AGA AAG GGG TAG AAT CAA T 3′
5′ TTT TAC CAG TCT TTG GTA GTT CCT G 3′
5′ CTT TGT CTA CTG AGG ACT TTG AAG C 3′
5′ AAC CGT TAT TAT TGC TCA TTA TCC A 3′
5′ AAG TCA CCA ATG CAC TCA 3′
5′ GGC ACT TTC CAG AGC GGT 3′
A set of primers (gcy1f and gcy1r in Table 1) were utilized for polymerase chain reaction (PCR) amplification with pUC27-HIS as a template to develop a gcy1-deleted strain. pUC27-HIS was constructed by replacing the KanMX6 cassette with the S. cerevisiae HIS3 gene including 317 nucleotides upstream and +201 nucleotides downstream of the HIS3 gene. The PCR product was transformed into S. cerevisiae using the lithium acetate method, and the deletion mutant was selected on solid SD medium-HIS (Ito et al. 1983). Replacement of the GCY1 gene with HIS3 was verified by the appearance of PCR products with expected sizes using primers that span the left and right junctions of the deletion module within the genome. Four open reading frame (ORF)-specific confirmation primers (gcy1_A, HIS_B, HIS_C and gcy1_D in Table 1) were used to verify deletion of gcy1 (Jung et al. 2011). After confirmation of gcy1 deletion, the selection marker was removed for reuse with subsequent gene manipulation as described previously (Gueldener et al. 2002).
dak1 and dak2 were deleted sequentially to develop a dak1 and dak2 double-deleted strain. The procedure was the same as gcy1 deletion, but with different primer sets (dak1f, dak1r, dak2f and dak2r in Table 1). Two sets of four ORF-specific confirmation primers (dak1_A, HIS_B, HIS_C and dak1_D, and dak2_A, HIS_B, HIS_C, and dak2_D in Table 1) were used to verify the dak1 and dak2 double deletion, respectively.
To overexpress the putative GLY dehydrogenase (Gcy1p), a set of primers listed in Table 1 were used for PCR amplification using S. cerevisiae sKk12 genomic DNA as the template. The PCR product was cloned into the pRS423 vector using XhoI/BamHI under the GPD promoter (ATCC, Manassas, VA, USA). The constructed expression vector was transformed into the S. cerevisiae sKk12 strain using the lithium acetate method.
LB media was used for E. coli, and ampicillin (50 μg ml−1) was added for selection. SD medium [1% GLY, 0·67% yeast nitrogen base without amino acid, yeast synthetic drop-out medium supplement (amino acid supplement mixture without histidine, leucine, tryptophan and uracil; Sigma-Aldrich, St Louis, MO, USA) 0·074%] was used for S. cerevisiae. Wild-type YPH499, sKg and sKk12 strains were grown on SD medium-com (SD medium with added leucine, tryptophan, histidine and uracil). The sKGk12 strain was grown on SD medium-HIS (SD medium with added leucine, tryptophan and uracil). Solid media were made by adding 2% (w/v) agar.
Saccharomyces cerevisiae YPH499 and its derived mutant strains were fermented under aerobic and micro-aerobic conditions in a 1-l bioreactor (BioCNS Co, Daejeon, South Korea) in a 700 ml working volume. GLY aerobic batch cultures were conducted for 72 h at an agitation speed at 200 rev min−1, pH 5·0, 30°C and air supply of 1·0 l min−1. Agitation speed and air supply were adjusted to 100 rev min−1 and 0·2 l min−1, respectively, for micro-aerobic batch cultures.
The following chemicals were purchased from Sigma-Aldrich: pyridine (≥99·8%) was used to maintain the anhydrous condition of the derivatives, erythritol (≥99%) was used as an internal standard to check GC/MS performance, methoxyamine-HCL was used for oximation and N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA) was used for silylation. GLY (≥99%), G3P (≥85%), DHA (≥97%) and DHAP (≥93%) were used as standards in the qualitative and quantitative GC/MS analyses.
Four kinds of samples such as intracellular (IC) metabolite, quenching solution (QS), whole broth (T) and extracellular (EX) metabolite are needed to quantify interesting metabolites and for leakage checks (Fig. 2). The sampling device consisted of a sampling port pipe and a bullet-shaped rod with a solenoid coil and an air pipe (Fig. 3). All components such as the pipe and rod were made of STX316L (stainless steel) to prevent corrosion and to maintain durability during sterilization. Programmable logic control (PLC) was used to control sampling and injection time of air before and after sampling (BioCNS Co, Daejeon, South Korea).
Intracellular metabolite sample and quenching solution sample
Cell broths were collected from a designed sampling port. Approximately 1 g (±0·05) of broth was withdrawn and injected (<0·5 s) into a tube containing 5 ml of 60% (v/v) methanol containing aqueous solution precooled to −40°C. The contents in the tube were vortexed quickly (<1 s), and the tube was placed in a cryostat (Daihan Scientific Co., Seoul, South Korea) at −40°C. The tubes were centrifuged at 16 600 g for 10 min at −20°C (Gyrozen, Daejeon South Korea). After centrifugation, the tube contents were divided into a cell pellet for IC metabolite sample and a supernatant for the QS sample. Intracellular metabolites were extracted from the cell pellet by boiling with ethanol (Lange et al. 2001). Briefly, 5 ml of 75% (v/v) ethanol in water was added to the tubes, followed by vortexing. The tubes were placed in a water bath at 95°C for 3 min and then placed back in the cryostat. The QS was vortexed quickly with methanol solution, and a 300–500 μl aliquot of the solution was transferred to an empty precooled tube at −40°C. The boiling ethanol extraction was performed to minimize sample matrix effects.
Whole broth sample
The sampling procedure was the same as described previously without centrifugation. After vortexing, a300–500 μl aliquot of the suspension was transferred to an empty precooled tube at −40°C, and the boiling ethanol extraction was performed as described above to minimize sample matrix effects and to ensure cell disruption.
Extracellular metabolite sample
Cell broth was collected in a syringe from the sampling port containing cold beads at a temperature <−40°C (Mashego et al. 2005). The broth was quickly filtered using 0·20-μm disc filter (Albet, Barcelona, Spain) and directly placed into 5 ml of 60% (v/v) methanol containing aqueous solution precooled at −40°C. The filtered solution was vortexed quickly and transferred to an empty tube at −40°C. Then, the boiling ethanol extraction was performed as described above.
Derivatization of samples for GC/MS analysis
Erythritol at a final concentration of 40 μmol l−1 was added to all samples as a standard for GC/MS. The samples were dried by SpeedVac centrifuge (Hanil Science Industrial, Incheon, South Korea), resuspended in 50 μl of 240 mmol l−1 methoxyamine-HCl in pyridine solution and incubated for 50 min at 60°C. Then, the samples were silyated for 50 min at 60°C after adding 80 μl MSTFA. Then, the methoxime-trimethylsilyl derivatives were stored at −80°C for GC/MS analysis.
GLY, G3P, DHA and DHAP derivatives were analysed by GC/MS with a Varian 400-GC instrument coupled with Varian 300-MS single quadruple mass spectrometer (Varian Inc., Palo Alto, CA, USA). A 1 μl sample aliquot was injected onto a VF-17ms column (30 m × 250 μm internal diameter, 0·25 μm film thickness) in splitless mode. Helium gas flow rate as a carrier gas was set at 1·5 ml min−1. The GC temperature gradient was set at 70°C for 1 min, 1°C min−1 up to 76°C, 10°C min−1 up to 150°C and 30°C min−1 up to 320°C. The temperature of the transfer line to the mass device was set to 250°C, and the ion source was set to 280°C. The electron ionization (EI) method was used at 70 eV. The selected ion monitoring (SIM) mode was used for quantitative measurements.
GLY was measured with a high performance liquid chromatography (HPLC) LC-20H (Dong-il Shimadzu Corp., Seoul, Korea) instrument using an RI detector RID-10A model and HPX-87X anion exchange column (Bio-Rad, Hercules, CA, USA) at 75°C with 10 mmol l−1 sulfuric acid as the mobile phase at a flow rate of 0·5 ml min−1. DHA was measured by an enzymatic method (Bergmeyer 1984). Briefly, DHA was phosphorylated with ATP and GLY kinase to DHAP. DHAP was then reduced with NADH and glycerol phosphate dehydrogenase to l-(-)-glycerol 3-phosphate. The decrease in NADH concentration was measured by the change in absorbance at 339, 334 and 365 nm by spectrometry (Shimadzu, Tokyo, Japan). The result was proportional to the DHA concentration.
Development of a rapid sampling device
As emphasized in the study by Lange et al. (2001), rapid sampling is important during metabolic profiling, because it can minimize residence time of samples under undefined conditions. A novel sampling device consisting of a sampling port pipe, bullet-shaped rod with solenoid coil and an air-feeding pipe was manufactured for rapid sampling. The cell broth was collected via the sampling port pipe connected to the bioreactor. Sample volume and collection time were controlled by two components. The first one was the bullet-shaped rod with a solenoid coil that plays a role as automatic switchgear and is controlled by the PLC up to a 0·1-s scale. The other was a pressure regulator connected to the bioreactor vessel to control sample volume (Fig. 3). An air-feeding system was developed to minimize the remaining culture broth after sampling. Filtered air was fed to the air pipe before and after sampling to push the remaining culture broth into the space between the bullet-shaped rod and the wall of the bioreactor. The test sampling procedure was conducted to validate sample volume reproducibility. Six samplings were performed during the 72-h culture with fermentation broth using the YPH499 wild-type strain, and the volumes were 0·976 (±0·02) ml and 1·02 (±0·02) ml under aerobic and micro-aerobic conditions, respectively. The same sampling procedure was used for metabolic profiling of the wild-type and recombinant strains.
Validation of metabolite recovery and leakage
Quenching techniques are essential for accurate metabolic profiling by inactivating IC enzymes during the sampling procedure (Bolten et al. 2007). Quenching in 60% (v/v) methanol at −40°C is commonly used for S. cerevisiae, but there is still a concern for metabolite leakage (Canelas et al. 2008). Therefore, leakage levels of IC metabolites were measured by preparing and quantifying whole broth (T), EX metabolite, IC metabolite and QS samples (Fig. 2). The metabolite concentrations in the whole broth sample should be similar to the sum of those in the IC metabolite and QS samples if there is no metabolic loss during analysis. Additionally, the concentration of whole broth metabolites subtracted from those in EX metabolites should be comparable with those in IC metabolites if there was no leakage of IC metabolite into the QS during the sampling procedure (Canelas et al. 2008).
In this study, the four different samples were prepared from all strains under aerobic and micro-aerobic conditions to validate metabolite recovery and leakage during fast sampling. For example, the skK1 strain was cultured, and four types of samples were collected at 36 h in the mid-log phase. The data showed that 93·9–101·7% of GLY, DHA, G3P and DHAP were recovered in the IC metabolite or QS samples when compared with that of whole broth (Table S1). High recovery of metabolites in the IC metabolite and QS samples showed that the sequential sampling steps such as quenching, derivatization and concentration were reasonably well established for accurate metabolic profiling. Additionally, relatively low standard deviations (<10%) of metabolites (Table S1) suggested that the data were reasonably reproducible compared with other metabolic profiling data (Koek et al. 2006).
Then, the concentration in the whole broth sample was subtracted from that in EX metabolite and compared with that in IC metabolite. Among them, the GLY data were not useful to estimate leakage, because EX GLY level was almost 102-fold higher than the IC level. In this case, small measurement errors in whole broth and EX metabolite can cause a huge difference between concentrations of whole broth subtracted by EX metabolite and IC metabolite. Among other metabolites, 24·3, 1·3 and 6·54% of IC DHA, G3P and DHAP were estimated to leak into the QS, respectively (Fig. 4). The concentrations in whole broth subtracted from those in EX metabolite provided more accurate IC metabolite levels than those in IC metabolite, because the EX metabolite sample was exposed to extreme cold conditions for only a few seconds and is free from metabolite leakage (Canelas et al. 2008). Therefore, we used the concentration in whole broth subtracted by that in EX metabolite as the IC metabolite levels in the following study.
Confirmation of metabolite concentrations
A cross-comparison of quantification methods was performed to verify concentrations of metabolites measured by GC/MS. GLY and DHA were analysed by HPLC and an enzymatic method, respectively. The sample preparation procedure was exactly the same as that for the GC/MS analysis except that the derivatization step was omitted. No significant difference was observed between the HPLC and GC/MS results for detecting GLY. Less than 9·4% difference was shown for the four different samples (Fig. 5a). But, quantification by the enzymatic method resulted in 39 and 66% higher concentrations of DHA in some samples than those analysed by GC/MS, and the relative standard deviation of the enzymatic method results was much larger than that of the GC/MS results (Fig. 5b).
Metabolite concentrations in the different strains
The metabolic enzymes in pathway II (Fig. 1) were perturbed to identify the role of the GCY1 gene (encoding putative NADP(+)-coupled GLY dehydrogenase) in S. cerevisiae. A GCY1 deletion mutant (sKg) and a DAK1 and DAK2 (encoding DHA kinase isoenzymes) double-deletion mutant were constructed (sKk12). Additionally, GCY1p was overexpressed in sKk12 (sKGk12). Then, the wild type and mutant were cultured under different aeration conditions. However, the growth of the yeast strains was hampered under the anaerobic condition, and meaningful data were not collected (data not shown).
The growth pattern and GLY uptake pattern of the mutant strains were very similar to those of the wild type under aerobic culture (Fig. 6a), meaning that perturbation of GLY dehydrogenase and DHA kinase did not change cell metabolism under the aerobic condition. The DHA concentration was 6·23 μmol g DCW−1 in the wild-type strain at 36 h, whereas it was not detected in the sKg strain. DHA levels increased slightly in the sKk12 and sKGk12 strains compared with those in the wild type. However, the DHA levels were much lower than the G3P levels in all strains (Fig. 7). Again, these results support that pathway II was not very functional under the aerobic condition.
The specific growth rates of the sKg, sKk12 and sKGk12 strains under micro-aerobic culture decreased 12·9, 17·7 and 12·9% compared with those in the wild-type strain, as their GLY uptake rates decreased 24·0, 39·7 and 37·3%, respectively (Fig. 6b). Metabolic profiling under the micro-aerobic condition resulted in a much higher concentration of DHA (60·38 μmol g DCW−1) in the wild type. The GCY1 gene-deficient strain, sKg, did not produce DHA under either aerobic or micro-aerobic conditions (Table S1). The DHA level in sKk12 increased 374% compared with that in the wild type. The DHA level further increased in the sKGk12 strain by overexpressing GCY1. Meanwhile, G3P levels in the wild-type and mutant strains decreased significantly under the micro-aerobic condition. The amount of G3P decreased most significantly in the sKGk12 strain under the micro-aerobic condition, which was only 28·6% of the G3P level under the aerobic condition (Fig. 7). This result suggested that overexpressing GCY1 distributed more carbon flux into pathway II.
A fast sampling device can minimize the time cell broth is left in an undefined condition such as unknown dissolved oxygen, temperature and pH, which can affect metabolite conversion in vivo before the quenching procedure. Although significant improvements were achieved, the remaining culture broth was considered a serious problem, because it can contaminate the sample pipe line and disturb accurate metabolite quantification (Schadel and Franco-Lara 2009). Therefore, we developed a completely automated sampling device that is free from remaining culture broth by injecting filtered air into the sampling device. Air pressure and time of air injection were controlled by a PLC, and all sampling procedures were performed with a single pedal. In this study, the use of the developed sampling device was limited to the batch fermentation experiment, but it could also be used for a rapid perturbation experiment with time difference equipment (Visser et al. 2002; Mashego et al. 2006).
Among several quenching methods for micro-organisms, cold methanol quenching is commonly used for S. cerevisiae (Canelas et al. 2008). The extremely cold conditions during quenching can cause leakage of target metabolites into the QS. Although the mechanism for metabolite leakage was not identified, rapid cell chilling was suspected to cause changes in membrane fluidity, which allowed metabolites to leak from the IC to the EX space (Cao-Hoang et al. 2008). Metabolite leakage was estimated by detecting four different samples (whole broth, EX metabolite, IC metabolite and QS) in this study. A significant amount of metabolite leakage was noticed with DHA but not with G3P or DHAP. The negative phosphoryl group on the metabolites may have contributed to less leakage in G3P and DHAP, but that was not consistent with another report (Canelas et al. 2008). Up to now, no optimized quenching method is available for every metabolite, so the concentrations in the whole broth sample subtracted from those in the EX metabolite are considered the most reasonable method to estimate IC metabolite concentrations.
We performed a cross-platform comparison of the target metabolite quantification analysis such as GLY and DHA using another analytical platform to verify the derivatization efficiency of the target metabolites. Analysis of GLY using GC/MS in all samples (whole broth, EX metabolite, IC metabolite and QS) showed very similar results to the data obtained by HPLC with slightly lower standard deviations (Fig. 5a). This means that GC/MS produced data as reproducible as HPLC. In the quantitative analysis, we considered matrix effects caused by constituents in the analytical sample that could corrupt the quantification of an analyte of interest. The only difference between the GC/MS and HPLC samples was the derivatization step. Therefore, this suggests that no significant matrix effect was observed in the derivatization step at least with GLY. The DHA analysis using the enzymatic method resulted in much poorer reproducibility than that of GC/MS results. In particular, the error range of the IC DHA level was much larger than that of the EX level. The principles of the enzymatic method usually rely on measuring absorbance at different wavelengths depending on a cofactor such as NAD(P)H and NAD(P). In the case of IC metabolites, more nonspecific reactions of these cofactors might occur, so measuring IC metabolites using enzymatic methods was less reproducible than analysis using GC/MS.
Based on our fast sampling system and a GC/MS analytical platform, we obtained reproducible metabolic profiling data for GLY, G3P, DHA and DHAP. The technique was useful for identifying the function of the GCY1 gene product. All metabolites showed higher IC levels under aerobic condition compared with those under micro-aerobic condition except DHA, which supported higher growth and GLY consumption rates under the aerobic condition (Table S1). In particular, the G3P level under the aerobic condition was much higher than that under the micro-aerobic condition and was not much affected by GLY dehydrogenase and DHA kinase gene manipulations (Fig. 7b). This demonstrated that those enzymes are not the main GLY dissimilation pathway under aerobic condition. In contrast, the DHA level sensitively responded to GLY dehydrogenase and DHA kinase gene manipulations under the micro-aerobic condition (Fig. 7a). Deletion of the GCY1 gene abolished DHA production from GLY under the aerobic and micro-aerobic conditions. Overexpression of the putative GLY dehydrogenase in the DHA kinase-deleted strain improved DHA production. These results clearly demonstrate that the function of the GCY1 gene product is a GLY dehydrogenase. Our study is a clear example that metabolic profiling with gene manipulated strains can reveal the role of a specific gene and pathway in vivo.
In conclusion, a novel automated sampling device was developed for accurate and reproducible IC metabolic profiling by rapidly collecting culture broth (<0·5 s) and removing the culture broth from the sampling pipe line. The effects of metabolite derivatization and leakage during cold methanol quenching were tested, and the results ensured a reasonably accurate quantitative metabolite analysis. The concentrations of GLY and its catabolized metabolites such as G3P, DHA and DHAP were measured under different aeration conditions and demonstrated the effects of aeration and gene manipulation in yeast strains. The function of the GCY1 gene product under micro-aerobic conditions was clearly demonstrated to be a GLY dehydrogenase in S. cerevisiae.
This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (nos. 2012-0005359 and 2012M1A2A2026560).