Metabolic profiling of the fission yeast S. pombe: quantification of compounds under different temperatures and genetic perturbation

Abstract

Metabolomics is a rapidly growing branch of post-genomic chemical biology. The fission yeast Schizosaccharomyces pombe is an excellent eukaryotic model organism. Although the entire S. pombegenome has been sequenced and detailed transcriptomic analyses were performed, little metabolic profiling has been done. Here we report the first global semi-quantitative analysis of the S. pombe metabolome using liquid chromatographyhigh-resolution mass spectrometry. Procedures to obtain metabolic compounds from S. pombe extracts were established. One hundred and twenty-three distinct metabolites were identified while approximately 1900 peaks from the ∼6000 observed were assigned. A software system (MZviewer) was developed to visualize semi-quantitative metabolome data using a dynamically generated scatter plot. We examined the metabolome of S. pombe cells exponentially grown in synthetic culture medium (EMM2) at two different temperatures, 26 °C and 36 °C. The profiles were similar except for varying amounts of certain amino acids and a significant increase in several compounds at 36 °C, such as trehalose (200-fold), glycerophosphoethanolamine (50-fold), arabitol (16-fold), ribulose (8-fold), and ophthalmic acid (5-fold). Reproducibility was demonstrated using a deletion mutant sib1Δ that lacked ferrichrome synthetase and showed no significant metabolic effects except the disappearance of the hexapeptide ferrichrome and the appearance of a putative dipeptide precursor. Taking advantage of the metabolic profile similarity at 26 °C and 36 °C, we analyzed the metabolome of a temperature-sensitive hcs1-143 mutant defective in the HMG-CoA synthase. As expected, HMG-CoA was decreased. In addition, extensive secondary metabolic effects, including a decrease in urea cycle intermediates and an increase in acetylated compounds, were observed. These findings confirm that S. pombe can be applied as an appropriate model to monitor metabolic responses to environmental conditions as well as genetic perturbations.

---

Introduction

Metabolomics, which aims to profile all of the small molecules present within a cell, tissue, organ, or organism, is a dynamically developing branch of post-genomic chemical biology. Together with transcriptomic and proteomic approaches, it is an important tool for studying biological regulation in a systematic and integrative way by supplying a comprehensive data set on cellular metabolites. Nuclear magnetic resonance (NMR), which is a powerful method for structural identification, and mass spectrometry (MS), which offers high sensitivity, are the most widely applied metabolitedetection methods. Information obtained by metabolomics may be essential to understanding the metabolic aspects of biological and medical phenomena that include nutritional regulation or pharmacologic effects of drugs, and may also facilitate the discovery of biomarker metabolites for diagnosing a wide range of diseases, such as cancer, diabetes, and obesity.^1–4

The actual implementation of metabolomics is still formidable, as a large number of cellular metabolites must be monitored. A genome -scale reconstructed metabolic network containing 1175 metabolic reactions has been reported for the budding yeast Saccharomyces cerevisiae, which has a small genome and is also presumed to have a comparatively simple metabolome.⁵ Current estimates set the total number of eukaryotic intracellularmetabolites at 4000 to 20 000,¹ although many compounds are not commercially available for use as standards. Actual experimental outcomes, even in the most thorough metabolomic reports, however, have led to the identification of no more then 200 ‘named’ compounds.^3,6,7 This large difference in the number of predicted and actually identified metabolic compounds is partially due to the high diversity both in the chemical properties and cellular concentrations of metabolites, making detection of whole complex mixtures difficult, as well as to the lack of a comprehensive unified methodology for identifying metabolites using MS, tandem MS, and/or NMR data.

The fission yeast Schizosaccharomyces pombe has proved to be an excellent model for studying cell-cycle control, chromosome segregation, gene silencing, mitochondrial biogenesis, and meiosis. S. pombe is easily amenable to genetic manipulation and its biology resembles that of mammals in many ways.^8–15 The whole S. pombegenome has been sequenced¹⁶ and a convenient genomic database exists.^17,18 According to the KEGG database,¹⁹ 1140 products among the ∼5000 S. pombegenes have a known enzymatic activity. Comprehensive analyses using high-throughput transcriptomic and proteomic approaches have been intensively applied to meiosis and environmental stress, as well as the regular cell cycle of S. pombe.^20–24 In the field of metabolomics, however, the number of reports is limited.^25,26

Here we report the global metabolomic analysis of the fission yeast S. pombe, the first attempt for this organism to our knowledge, using an MS method combined with the software tools MZmine 2 and MZviewer for data analysis and presentation, respectively. The procedures developed here allowed us to obtain highly reproducible results on the levels of various classes of metabolites within cells. An extraction procedure suitable for S. pombeintracellularmetabolites was established, utilizing cold methanol for quenching²⁷ and glass beads for cell disruption. To separate and detect metabolites, liquid chromatography-mass spectrometry (LC-MS) was employed using an LTQ Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific) whose high precision m/z values²⁸ enabled us to assign metabolite candidates even without prior use of a pure standard, and also revealed certain modified (e.g., methylated or acetylated) compounds. We successfully identified and characterized the relative changes of 123 named metabolic compounds present in proliferating wild-type S. pombe cells cultivated at 26 °C or 36 °C. Taking advantage of the wide availability of genetically modified S. pombe strains, we further analyzed two mutants, an sib1Δ deletion mutant lacking the synthetic enzyme for ferrichrome and hcs1-143, a conditional lethal mutant defective in 3-hydroxy-3-methylglutaryl (HMG)-CoA synthase.²⁹

Results

Preparation of metabolome samples and detection of thousands of peaks

Haploid wild-type S. pombe 972 h⁻ cells were cultivated in a synthetic minimal medium (EMM2; Supplementary Table 1† ) at either 26 °C or 36 °C until they reached a concentration of 5 × 10⁶ cells ml⁻¹. Rod-like cells grown at 26 °C or 36 °C were indistinguishable in their appearance under a light microscope (Fig. 1A).

Fig. 1 LC-MS analysis of metabolic compounds in the fission yeast Schizosaccharomyces pombe. A. Micrographs of S. pombe cells grown exponentially at 26 °C and 36 °C in a liquid EMM2 medium, which are indistinguishable. B. Schematized procedures to obtain metabolic compounds. Metabolites were extracted in 50% methanol from bead-disrupted S. pombe cells and isolated by a 10 kDa cut-off filter (Materials and methods). C. A 3D profile of raw LC-MS data in negative ESI mode; X-axis: retention time (min), Y-axis: m/z; Z-axis: peak height (signal intensity). Twenty-three identified compounds are shown as examples. The inset indicates an enlarged portion of the profile.

Cells were harvested by vacuum filtration (Materials and methods). We applied 100% methanol at −40 °C for quenching (Fig. 1B), a method recommended to prevent metabolite leakage.²⁷ Cells were then disrupted with glass beads in 50% methanol at 0 °C using a Multi-beads shocker machine (Yasui-Kikai). The resulting extracts were centrifugally filtered on a 10 kDa cut-off filter (Millipore) at 4 °C. The filtrates were concentrated (∼10-fold) on a rotary evaporator. After evaporation, each sample was re-suspended in 50% acetonitrile, and separated on an LC ZIC-pHILIC column (Merck SeQuant), followed by MS analysis using an LTQ Orbitrap (Thermo Fisher Scientific) operated in full-scan mode (100–1000 m/z).

For each sample, we obtained two sets of raw data with the MS ionization source set to either the negative or positive ESI (electrospray ionization) mode. This LC-MS data contained information about thousands of individual peaks (Fig. 1C). LC provided the retention time (min in X-axis) of the peaks, whereas MS produced the peak m/z (mass/charge) values on the Y-axis and the peak height (signal intensity) on the Z-axis. Twenty-three of the identified compounds are shown as examples.

LC-MS data were analyzed using the current development version (1.95) of the MZmine 2 software (http://mzmine.sourceforge.net), originally introduced by Katajamaa et al.³⁰ After performing peak detection, isotopic peaks were removed and peak areas were normalized by a weighted contribution of internal standard peak areas. Two metabolically inert compounds, PIPES and HEPES (MW 302.4 and 238.3, respectively), were used as internal standards, and spiked into the samples in fixed amounts. Peaks 21 and 22 in Fig. 1C represent HEPES and PIPES, respectively.

After removing the isotopes, a total of 3480 peaks were detected in the positive ionization mode and 2723 peaks in the negative ionization mode (Table 1). Fig. 2A shows a histogram of the number of detected peaks versus their m/z (negative ESI mode). As the majority of the ions had a single charge, the m/z values mostly represented molecular mass. Over 70% of the peaks were smaller than 500 m/z. Fig. 2B shows a histogram of number of detected peaks versus their normalized peak areas in log₁₀-scale (negative ESI mode). Most of the peaks (74%) were small (normalized peak area between 0.001 and 1), therefore barely visible in Fig. 1C, where the highest peak (UDP-glucose) had an area of 413, roughly 10⁶-fold larger than the smallest peak area.

Table 1 Summary of analysis of peaks detected by LC-MS^a

Total detected peaks	Negative ESI mode	Positive ESI mode
	2723	3480
Non-metabolite peaks	781	946
Na adducts	172	207
K adducts	12	55
Mg adducts	4	6
NH3 adducts	4	54
H3PO4 adducts	16	4
H2SO4 adducts	17	4
H2CO3 adducts	39	57
Peak complexes	255	214
In-source fragments	262	345

---

Potential metabolite peaks	1942	2534
	(335 in both modes)

---

a Thousands of peaks were detected in the negative and positive ionization modes. Many of the peaks, however, represented adducts , fragments, and complexes. These non-metabolite peaks were determined using the automatic detection method provided by the MZmine 2 software (Materials and methods). Among more than 4000 potential metabolite peaks, 123 (98 + 102 − 77) actual metabolites were identified (see text).
Identified metabolites	98	102
	(77 in both modes)

---

Fig. 2 Distribution of the m/z values and peak areas of detected peaks and identified metabolites. A. Histogram of the number of detected peaks at 26 °C (negative ESI mode) versus their m/z value. B. Histogram of the number of detected peaks at 26 °C (negative ESI mode) versus their normalized peak area in log₁₀-scale. C. Histogram of the number of identified metabolites at 26 °C (negative ESI mode) versus their m/z value. D. Histogram of the number of identified metabolites at 26 °C (negative ESI mode) versus their normalized peak area in log₁₀-scale.

Assignment of detected peaks to authentic compounds

Because identification of actual metabolites among detected peaks requires considerable manual effort, we focused on identifying the biggest peaks and also peaks showing significant changes between sample conditions. The search for candidate compounds was mainly based on the high precision of molecular mass detected by LTQ Orbitrap.²⁸ The PubChem compound database (NCBI) was searched for monoisotopic masses and, for plausible candidates, pure standards were obtained and examined for verification. As even a 1-ppm mass accuracy is not sufficient to assign an exact chemical compound,³¹ the retention times and MS/MS fragmentation patterns obtained by LC and Orbitrap, respectively, were essential for distinguishing compounds with identical mass. For example, by comparing the retention time, an unknown compound with 282.120 m/z was concluded to be 1-methyladenosine rather than two other methyladenosine candidates that have identical mass but different retention times (Fig. 3A). 2′-O-Methyladenosine was distinguished from N₆-methyladenosine by comparing the MS/MS fragmentation patterns; despite the fact that their retention times were very similar. 1-Methyladenosine is a minor nucleoside of tRNA, produced from adenosine by a methyltransferase enzyme. 1-Methyladenosine is found in human urine and is thought to be a potential tumor marker.³²

Fig. 3 Identification procedures of metabolite peaks. A. An unidentified peak with a value of 282.120 m/z (positive ESI mode) precisely fits to methyladenosine. Three candidate isomers of methyladenosine (1-methyladenosine, 2′-O-methyladenosine, and N₆-methyladenosine) with the same m/z value were compared. By matching the retention time and MS/MS fragmentation patterns, the unknown compound was determined to be 1-methyladenosine (see text). B. In the case of ferrichrome, a siderophore,⁵⁰ its unique isotopic pattern was used for identification. Isotope patterns of ferrichrome predicted and actually detected in negative ESI mode are shown. C. At least 13 derivatives of trehalose were generated during MS analysis as displayed by 2D plot (X axis: retention time, Y axis: m/z) that shows derivative peaks detected within a narrow retention time range of 10–15 min. Within this retention time range, trehalose, heptose-phosphate, F6P (fructose-6-phosphate), R5P (ribose-5-phosphate), citrulline, and C₃H₇O₆P (possible glycerone-phosphate) are present and shown in red. Trehalose dimer and various complexes of trehalose with other peaks are marked in blue and several fragments of trehalose are marked in green.

The isotope pattern predicted from the compound formula greatly facilitated the process of candidate selection, particularly if the candidate compound formula contained specific elements with an unusual isotope distribution, like Fe, B, Br, or Cl. Fig. 3B shows an isotope pattern of the iron-binding siderophoreferrichrome calculated using Xcalibur software (Thermo Fisher Scientific), compared to the pattern of the actually detected peak from cell extracts. In this case, just comparing the isotope patterns was sufficient to confirm the identity of the compound. Ferrichrome is a cyclic hexapeptide compound, synthesized and secreted by microorganisms, that specifically binds to ferric iron with exceptionally high affinity. S. pombe contains a non-ribosomal peptide-synthesizing enzyme Sib1 that synthesizes ferrichrome.³³

In a similar isotope analysis, we found a ribulose–borate complex (Supplementary Fig. 1A† ) with a 307.084 m/z value in negative ionization mode. Formation of the pentose–borate complexes was previously reported,³⁴ yet we observed such complexes even when injecting pure pentose compounds into LC-MS, suggesting that the source of the boron was non-cellular. By comparing the retention time of various pentoses (Supplementary Fig. 1B† ), we identified the 307.084 m/z peak as ribulose. This was verified by mixing pure ribulose with the cell extract, which formed a single peak (Supplementary Fig. 1C† ). Ribulose is a monosaccharide involved in the production of many bioactive substances and, in a phosphorylated form, is an intermediate of the pentose–phosphate pathway.

A single compound analyzed by LC-MS often produced multiple peaks, particularly for highly abundant compounds, due to fragmentation or adduct formation induced during the ionization process. Importantly, these multiple peaks had the same elution profile (retention time) as the original compound in LC. For example, trehalose, which was abundant in cells cultivated at 36 °C, produced at least 13 different peaks within a narrow band of retention time (Fig. 3C). These included a trehalose dimer, various fragments, ion adducts , and complexes formed with other co-eluting compounds. The MS/MS fragmentation pattern was useful for assigning fragments produced by degradation of the main peak compound. It is important to stress that such additional peaks did not represent actual compounds present in the analyzed sample; instead they were produced during the ionization process. By using the software to automatically detect adducts , fragments, and complexes (Materials and methods), 781 peaks in the negative ESI mode and 946 peaks in the positive ESI mode, respectively, were determined to represent non-metabolite ions, leaving a total of 4476 (1942 + 2534) potential metabolite peaks, only 335 of which were detected in both negative and positive ESI modes (Table 1). We defined the peaks detected in both modes as peaks eluting at the same retention time and having a mass difference of 2.016 m/z (two protons).

Through examining peaks by the above procedures employing standard compounds, we successfully identified 123 distinct metabolites (108 verified by pure standards), which are listed in Table 2. Their detailed properties in LC-MS (m/z values, retention times, and peak areas) are shown in Supplementary Table 2.† They were physiologically diverse compounds, nucleotides, amino acids, sugars, vitamins, organic acids, redox compounds, etc. The majority (77/123 = 63%) of these assigned metabolites were detectable in both negative and positive ESI modes (Table 1), in contrast to only 7.5% (335/4476 = 7.5%) of the total LC-MS peaks detected in both modes. Fig. 2C shows a histogram of the number of assigned metabolitesversus their m/z values (negative ESI mode). Forty-two percent of the compounds were in the range of 100 to 200 m/z. Metabolites over 500 m/z were quite rare and no metabolite was identified in the range of 900 to 1000 m/z. Fig. 2D shows a histogram of the number of assigned metabolitesversus their normalized peak areas in log₁₀-scale (negative ESI mode).

Table 2 List of 123 identified metabolites^a

	Name	Ionization modes	WT 972 26 °C	WT 972 36 °C
a One hundred twenty-three metabolic compounds identified in extracts from wild-type S. pombe growing cells, detected by LC-MS. The mass spectrometer ionization mode is specified as negative (neg), positive (pos), or both (neg pos) for each detected compound. The bold ionization mode provided a bigger peak, which was used for quantification. Normalized peak areas in 26 and 36 °C extracts are shown in log2 scale using dots (actual numerical values including standard deviations are listed in Supplementary table 2^1 ). Each additional dot represents a two-fold difference in the normalized peak area. Peaks increased over four-fold in each condition are marked with an arrow (↑). Compounds in italics were not verified using a pure standard so that their assignment was based only on molecular mass.
Nucleotides	AMP	neg pos	••••••	••••••
	ADP	neg pos	••••••	••••••
	ATP	neg pos	••••••••	••••••••
	3′-5′-cAMP	neg pos	•	•
	CMP	neg pos	•	•
	CDP	neg pos	••	•••
	CTP	neg pos	•••••	••••••
	GMP	neg pos	•	•
	GDP	neg pos	••••	••••
	GTP	neg pos	••••••	•••••••
	UMP	neg pos	••••	••••
	UDP	neg pos	•••••	••••••
	UTP	neg pos	••••••	•••••••
	FAD	neg pos	•••	••••
	IMP	neg pos	•••	•••
	NAD+	neg pos	•••••••••	•••••••••
	NADH	neg pos	••••••••	••••••••
	NADP+	neg pos	•••	••••
	NADPH	neg pos	•••	••••
	Nicotinate D-ribonucleotide	pos	•	•
Nucleosides, nucleobases	Adenine	pos	•••••	••••••
	Adenosine	pos	•••• ↑	•
	Guanosine	neg pos	•••	•••
	Inosine	neg pos	•••••	•••••
	Uracil	neg	•	•
	Cytidine	pos	••	••
	Xanthine	neg	••	••
	Xanthosine	neg pos	•••	•••
	Nicotinate D-ribonucleoside	pos	•••••••	••••••••
Methylated nucleosides	1-Methyladenosine	pos	•••••	••••••
	1-Methylguanosine	pos	•	•
	Dimethyl-guanosine	pos	•	•
Coenzymes	Coenzyme A	neg pos	•	•
	Coenzyme B	neg pos	•••••	•••••
	Acetyl-CoA	neg pos	•••	•••
Amino acids	Arginine	neg pos	•••••••••	••••••••
	Asparagine	neg	•	•
	Aspartate	neg pos	••••	•••••• ↑
	Citrulline	neg pos	••••••	••••••
	Ectoine	neg	•	•
	Glutamate	neg pos	••••••••	•••••••••
	Glutamine	neg pos	•••••••	•••••••
	Histidine	neg pos	••••••	•••••
	Isoleucine	neg pos	•	••• ↑
	Leucine	neg pos	•	••
	Lysine	neg pos	•••••	•••••
	Methionine	neg pos	••	•••
	Ornithine	neg pos	••••	••••
	Phenylalanine	neg pos	••••••	••••••
	Proline	neg pos	•••	••••• ↑
	Pyroglutamic acid	neg	•	•••• ↑
	Saccharopine	neg pos	••••	•••••
	Serine	pos	••	•••
	Threonine	neg pos	••••	•••••
	Tryptophan	neg pos	•••	••••
	Tyrosine	neg pos	•••••	••••
	Valine	neg pos	••	•••• ↑
	5-Aminovalerate	neg	•	•
	S-Adenosyl-homocysteine	pos	•••••••	•••••••
	S-Adenosyl-methionine	neg pos	•••••	••••••• ↑
Methylated amino acids	Betaine(trimethyl-glycine)	pos	•	•
	Dimethyl-glutamate	pos	••••	••••••• ↑
	Dimethyl-arginine	pos	••	••
	Dimethyl-lysine	pos	•	•
	Trimethyl lysine	pos	•	•
	Trimethyl histidine	pos	••	••
	Leucine methylester	pos	•	•
	Glutamate methyl ester	pos	••	•••• ↑
Acetylated amino acids	N 2-Acetyl-lysine	pos	•••	••••
	N 6-Acetyl-lysine	neg pos	•••	••••• ↑
	N-Acetyl-ornithine	neg pos	•••	••••• ↑
	N-Acetyl-arginine	neg pos	••••••	•••••
	N-Acetyl-glutamate	neg pos	••••	•••••
	N-Acetyl-histidine	neg pos	••	••
	Acetyl-glutamine	neg pos	••	••
Redox compounds	Ergothioneine	pos	•	•
	Glutathione	neg pos	•••••••••	•••••••••
	Glutathione (oxid.)	neg pos	•••••••••	•••••••••
Sugars	2-Keto-3-deoxyoctonate	neg	••••	••••
	Ribulose (boron complex)	neg	••••	••••••• ↑
	Trehalose	neg pos	•••	••••••••••• ↑
Sugar phosphates	Ribose-5-phosphate	neg pos	•••••••	•••••••
	5-phosphoribose-1-diphosphate	neg pos	•	•
	Glucose-6-phosphate	neg pos	••••••	••••••
	Fructose-6-phosphate	neg	••••••	•••••••
	Fructose-1-6-diphosphate	neg pos	••••••••	••••••••
	Heptose-phosphate	neg pos	••••••	••••••
	Heptose-diphosphate	neg pos	••••••	••••••
	Octose-phosphate	neg pos	••••	••••
	Octose-diphosphate	neg pos	•••••	•••••
	Disaccharide-phosphate	neg	•••	•••••• ↑
	FGAR (Phosphoribosylformylglycinamide)	neg pos	••••••	•••••••
Nucleotide-sugars	UDP-acetyl-glucosamine	neg pos	•••••	••••••
	UDP-glucose	neg pos	•••••••••	•••••••••
	GDP-glucose	neg pos	•••••	•••••
Ferrichrome	Deferriferrichrome	neg	•	•
	Ferrichrome	neg pos	••••	••••
Vitamins	Biotin	neg pos	••••	•••
	Pantothenate	neg pos	•••	•••
	Pyridoxine	pos	•	•
	Riboflavin	pos	•	••
TCA cycle	2-Oxoglutarate	neg	•	••
	Citrate	neg pos	•••••	•••••
	Fumarate	neg	•	•
	Malate	neg	•	••
Others	2-Aminoadipate	pos	•	••• ↑
	4-Guanidinobutyrate	pos	•	•
	6-Phospho-gluconate	neg pos	•••••	•••••
	Arabitol	neg	•	••• ↑
	Citramalate	neg	••	••
	Gluconate	neg	••	••
	Glutaric acid	neg	•	•
	Glycerol-phosphate	neg pos	•••••	•••••
	Glycerophosphocholine	neg pos	•••••••••	•••••••••
	Glycerophosphoethanolamine	neg pos	•	•••• ↑
	Nicotinamide	pos	•	•
	Ophthalmic acid	neg pos	•••••••	••••••••• ↑
	Phosphoenolpyruvate	neg pos	•••••	•••••
	Phospho-glyceric acid	neg pos	••••••••	••••••••
	Phosphopantothenate	neg	•	•
	Phthalic acid	neg pos	••••	•••••
	Propylmalate	neg	•••••	•••••
	Quinic acid	neg	••••	••••

---

Quantification of compound peaks

We investigated the quantification aspects of the obtained LC-MS data. First, authentic samples (containing 0, 1, 2, and 4 pmol each of AMP, ADP, ATP, tyrosine [tyr] and tryptophan [trp]) were injected with constant amounts (10 pmol) of HEPES and PIPES as the internal standards. Their peak profiles are shown in Fig. 4A. Although AMP and ATP displayed distinct peak heights (∼6-fold difference), probably due to different ionization efficiencies, we were able to quantify a 2-fold difference within each of the injected pure compounds. The size of the peaks of the HEPES and PIPES controls were constant (Fig. 4A bottom). The peak areas integrated by the MZmine 2 software were used as the basis for quantification.

Fig. 4 Quantification of compounds by LC-MS using pure standards. A. Four mixture samples containing 0, 1, 2, and 4 pmol of AMP (346.06 m/z), ADP (426.02 m/z), ATP (505.99 m/z), tyrosine (tyr; 180.07 m/z), and tryptophan (trp; 203.08 m/z) were injected (negative ESI mode), and their extracted ion chromatograms obtained (upper panel) using a constant amount (10 pmol) of standard PIPES (301.05 m/z) and HEPES (237.09 m/z) as internal controls (lower panel). B. The amount of pure compounds and resulting size of peak area in LC-MS were plotted, using six representative compounds, GDP-glucose, ATP, trehalose, citrate, arginine, and glutamate. Each data point represents peak area (Y-axis) and injected amount (X-axis). Trend lines were constructed by applying quadratic regression on the log₁₀-scale data. C. In cell extracts, the peak of trehalose increased about 200-fold after the temperature shift from 26 °C (upper panel) to 36 °C (middle panel). The amount of trehalose in extracts from cells grown at 36 °C was estimated by spiking a known amount (7.5 nmol in 1 μl injection volume) into the extract from cells grown at 26 °C to obtain peaks of equal size (lower panel). D. The peak of trehalose increased rapidly and became saturated within 10 min after a temperature shift from 20 °C to 36 °C. E. After decreasing the cultivation temperature from 36 °C to 26 °C, the trehalose level was restored to the original within 30 to 45 min.

To further examine the validity of the quantification described above, we tested pure compounds injected in a wide range (10⁴-fold) of amounts from 1 pmol to 10 nmol and obtained log₁₀-based curves for the integrated peak areas versus the injected amounts (Fig. 4B). There was a good correlation between the peak areas and the injected amounts, although the shape of the calibration curves depended on the individual compounds. For example, the peak area made by 1 pmol arginine was more than 100-fold smaller than those produced by 1 pmol of other compounds. A higher concentration (10 nmol) of arginine, however, produced a peak area comparable to that of other compounds. These results strongly suggested that quantification should be restricted to relative changes of the peak area of the same compound and a calibration curve made with a wide range of concentrations should be used. Five of six compounds (trehalose, ATP, glutamate, citrate, UDP-glucose, and arginine), however, showed a roughly similar (<10-fold difference) order for the peak areas versus compound amount.

Trehalose was the largest peak in the extracts from cells cultivated at 36 °C. To estimate its amount, we took advantage of the fact that the peak area of trehalose at 26 °C was only 0.5% (7.3/1487 × 100) of the area at 36 °C (Supplementary Table 2† ). Pure trehalose (7.5 nmol) was spiked into 1 μl (final injection volume) of collected cells cultured at 26 °C before starting the extraction procedure, and the resulting peak size (Fig. 4C bottom) was similar to that obtained in cells cultured at 36 °C (Fig. 4C middle). By knowing the approximate amount of trehalose detected, we could estimate the amount per cell as (7.5 nmol/1 μl injection volume × 80 μl final sample volume)/(2.5 × 10⁸ cells) = 2.4 fmol per cell at 36 °C, equivalent to ∼1.5 × 10⁹ molecules per cell. Assuming an average intracellular volume of 148.5 μm³,³⁵ we obtained a 16.2 mM (2.4 fmol/148.5 μm³) intracellular concentration. This value was similar to that previously reported (∼21 mM) for cells grown at 37 °C.³⁶

Time-course experiments revealed that the high amount of trehalose was produced within 10 min after the temperature shift to 36 °C (Fig. 4D), suggesting that enzymatic activation of trehalose synthesis was swift. Following a temperature decrease to 26 °C, the concentration of trehalose was reduced back to the original level within 30 to 45 min (Fig. 4E). Such time length is much shorter than the generation time, therefore trehalose was actively metabolically converted to other compounds such as glucose, rather then being diluted by cell division.

Comparison of cells grown at 26 °C and 36 °C

Metabolic profiling of cells grown at 26 °C and 36 °C was performed using three independently prepared cell cultures at each temperature. First, the top 15 compounds with the biggest peak areas (not the actual amounts in extracts, as explained above) were compared (Supplementary Tables 3 and 4† ). Among them, 13 were common at 26 °C and 36 °C: free amino acids (arginine, glutamate), glutathione, nucleotides, and nucleosides (NAD, ATP, UDP-glucose, nicotinateD-ribonucleoside) and other phosphorylated compounds (fructose-1,6-diphosphate, ribose-5-phosphate, phospho-glyceric acid, glycerophosphocholine). Trehalose and ophthalmic acid became abundant (among the top 15 peaks) at 36 °C.

A software system called MZviewer was developed using the MZmine 2 platform to visualize the semi-quantitative metabolome data in a scatter plot (Materials and methods). First, we examined the global metabolic profiles (all detected peaks). Fig. 5A shows the comparison of two independent experiments with cells grown at the same temperature, 26 °C. A part of the scatter plot data is enlarged in the insets (indicated by the arrows in the left panel) so that individual peaks can be distinguished. The yellow spots [∼900 in negative (−) ESI mode and ∼1000 in positive (+) ESI mode, respectively] represent peaks with an assigned identity, including all identified fragments, adducts , and complexes. Eighty-five percent of all the peaks were within the 2-fold change markers indicated by the red lines. In contrast, when comparing samples from different cultures grown at 26 °C and 36 °C (Fig. 5B), only approximately 60% of the peaks changed 2-fold or less. The temperature shift, therefore, significantly altered the overall structure of the metabolome.

Fig. 5 All LC-MS peaks detected in S. pombe extracts compared using a scatter plot. A. Scatter plot of normalized peak areas of all peaks detected in two independent extracts from cells grown at the same culture condition at 26 °C, from both negative and positive ionization modes. A part of the scatter plot data is enlarged to reveal individual peaks. Approximately 85% of the peaks were within the 2-fold difference lines and mostly small peaks showed larger variation. Peaks marked yellow (∼900 in negative mode, ∼1000 in positive mode) were assigned. Apart from the 123 identified metabolites, these include fragments, adducts , and complexes. B. Scatter plot of normalized peak areas of all peaks detected in extracts from cells grown at different temperatures (26 °C vs. 36 °C), from both negative and positive ionization modes. Approximately 60% of the peaks changed 2-fold or less. The peak of trehalose, the most increased metabolite at 36 °C, as well as peaks of the trehalose dimer and various complexes of trehalose and other compounds, are marked in the plot (see text).

We then compared the peak areas of the 123 identified metabolites. As shown in the scatter plot in Fig. 6A, 96% of the metabolites in cells independently cultured at the same temperature (26 °C) changed 2-fold or less; therefore, the reproducibility of the extraction and detection for these identified compounds was very good. Only 69% of the peaks changed 2-fold or less when comparing the metabolites obtained from cells cultured at 26 °C and 36 °C (Fig. 6B, left and right panels). Among the compounds showing the most significant increase at 36 °C were trehalose (peak 1, ∼200-fold increase, Supplementary Table 2† ), glycerophosphoethanolamine (peak 6, 50-fold), arabitol (peak 7, 16-fold), ribulose (peak 3, 8-fold), ophthalmic acid (peak 2, 5-fold), and various amino acids (peak 5, citruline; peak 8, valine; peak 9, proline; etc). Compounds that decreased at 36 °C were arginine (peak 13), histidine (peak 14), N-acetyl-arginine (peak 15), adenosine (peak 22), heptose-phosphate (peak 23), and ergothioneine (peak 24).

Fig. 6 Comparison of 123 identified metabolites. A. Two scatter plots showing normalized peak areas (in negative and positive ionization modes, respectively) of the 123 identified metabolites are compared in two independently obtained extracts from cells grown under the identical culture conditions (EMM2 at 26 °C). Approximately 96% of the peaks changed 2-fold or less (see Supplementary Table 2 for the standard deviations† ). B. Scatter plots of normalized peak areas of the 123 identified metabolites in extracts from cells grown in the same medium but at two different temperatures (26 °C vs. 36 °C). Approximately 69% of the compounds changed 2-fold or less. The most changed peaks are annotated according to the table.

Nucleotides and amino acids were then highlighted among the 123 detected metabolites, with the number indicating each nucleotide and amino acid, respectively (Supplementary Fig. 2 A, B† ). None of the nucleotides had a greater than 2-fold change at 36 °C. On the contrary, peak areas of amino acids changed significantly at 36 °C. Fifty percent of the 30 amino acids increased more than 2-fold at 36 °C, and only 3 amino acids decreased.

Ferrichrome, non-essential for the cell division cycle, is absent in sib1Δ

Ferrichrome is a cyclic hexapeptide siderophore composed of three glycine and three modified ornithine residues produced by various microorganisms. In S. pombe, ferrichrome synthesis requires a very large enzyme, Sib1 (4924 amino acids), that catalyzes the synthetic reactions.^33,37 It is non-essential and the deletion mutant sib1Δ grew normally in EMM2 medium (Fig. 7A). We compared the metabolome of wild-type and sib1Δ cells grown at 26 °C. Results from the negative ESI mode are shown in the scatter plot in Fig. 7B. Ninety-seven percent of identified metabolites changed 2-fold or less, with only two of them, ferrichrome and ornithine-glycinedipeptide (assigned by its mass), showing striking changes in the deletion mutant. The ferrichrome peak was decreased to the level of noise in sib1Δ, confirming that ferrichrome was not produced in this mutant. In contrast, ornithine-glycinedipeptide was abundant in sib1Δ (45-fold increase in peak area), suggesting that this may be a precursor peptide for ferrichrome produced by another enzyme; three of the Orn-Gly peptides might be polymerized and circularized by the Sib1 enzyme.

Fig. 7 Metabolomic analysis of sib1Δ deletion and hcs1-143 ts mutants. A. The growth of sib1Δ was not affected in EMM2 medium at 26 °C. B. Scatter plot of normalized peak areas of all peaks detected in extracts from the wild type 972 vs. sib1Δ deletion mutant grown at 26 °C in negative ESI mode. Eighty-six identified metabolites are highlighted in yellow. Sib1 encodes an enzyme that synthesizes ferrichrome. Ferrichrome was absent in the deletion mutant, while ornithine-glycine (orn-gly) dipeptide was abundant. C. Chemical structure of HMG-CoA (3-hydroxy-3-methylglutaryl-CoA). D. HMG-CoA (912.16 m/z) could be detected only after restricting the scan range of the mass spectrometer to 900–950 m/z. HMG-CoA present in wild-type extract (top panel) was absent in hcs1-143 ts mutant (middle panel) at the restrictive temperature (36 °C). Pure HMG-CoA standard peak with the same retention time is shown in the lower panel. E. Scatter plot of normalized peak areas of all peaks detected in extracts from the wild-type 972 strain vs. hcs1-143 ts mutant both cultured at 36 °C for 6 h, in positive ESI mode. Ninety identified metabolites are highlighted in yellow. Forty-three percent of them changed over 2-fold in the mutant. Most changed peaks are annotated according to the table. Compounds in italics were not verified using a pure standard, therefore only assigned by their mass.

Complex changes occurred in the metabolome of the hcs1-143 mutant

To examine the effect of a temperature-sensitive essential gene mutation on metabolome data, we took advantage of a large collection of S. pombe temperature-sensitive (ts) mutants previously generated by random mutagenesis³⁸ and selected the HMG-CoA synthase mutant hcs1-143²⁹ for an initial analysis, as the Hcs1 enzyme is a ‘super-housekeeping’ gene that is required for both proliferation and quiescence, and essential for lipid metabolism. HMG-CoA (Fig. 7C) is an important intermediate of a highly conserved mevalonate-dependent route, a pathway essential for a large number of diverse cellular processes. The mevalonate pathway is also a possible target of anti-cancer therapy for humans.³⁹

We compared the metabolome of wild type and hcs1-143 cells grown initially at 26 °C (permissive temperature) and then shifted to 36 °C (restrictive temperature) for 6 h. Because HMG-CoA was not detectable in the previous metabolomic experiments, we performed an additional injection of each sample with the mass spectrometer operated in a narrow scan range of 900 to 950 m/z, targeting the HMG-CoA mass of 911 Da. Using this strategy, we successfully detected the HMG-CoA peak in wild-type cells and confirmed its absence in the hcs1-143 mutant (Fig. 7D).

The data from the positive ESI mode are shown in a scatter plot (Fig. 7E); comparison of wild-type with the hcs1-143 mutant shows the occurrence of complex changes in the metabolic profile. Of a total of 90 assigned metabolites in this experiment (marked yellow in the scatter plot), the peak area changed more than 2-fold for 39 compounds and more than 4-fold for 17 compounds. Notably, urea cycle intermediates, arginine, citrulline, and arginino-succinate, decreased over 4-fold in mutant cells. In contrast, ornithine increased 3-fold, possibly in relation with the increase of acetyl-ornithine (see below). Several unknown peaks were not detected in the hcs1-143 mutant. One of these was identified as arginine-glutamatedipeptide by its mass, yet the interpretation of this change is unclear. Among the compounds that increased in hcs1-143 were acetyl-ornithine, acetyl-lysine, acetyl-glutamine, acetyl-glutamate, and acetyl-CoA (acetyl group donor). Ergothioneine, a protective compound with antioxidative properties, and its precursor trimethyl-histidine, also increased in mutant cells. In contrast, glutathione decreased, suggesting that oxidative stress might not be the cause of the increase in ergothioneine. Unexpectedly, the ferrichrome peak increased over 3-fold (including the deferri- form lacking an iron atom) in the hcs1-143 mutant, indicating that HMG-CoA synthase might have an important role in regulating the levels of these modified and unmodified amino acids and non-ribosomal peptide synthesis.

Discussion

In the present study, we performed the first systematic metabolic profiling of the fission yeast S. pombe. Initially focusing only on the effect of cultivation temperature and basic genetic perturbation, we developed procedures for the extraction and detection of metabolic compounds from S. pombe cultures and outlined an identification strategy for unknown peaks. We also performed a number of experiments to ensure proper interpretation of the quantitative aspects of the LC-MS data. Although we could only identify 123 compounds, presumably a small fraction of the whole intracellular metabolome, our highly reproducible semi-quantitative data demonstrated both large and subtle metabolic changes. Taking advantage of powerful yeast genetics, we examined the metabolome of cells cultured at 26 °C and 36 °C for the use of ts mutants and proceeded to compare wild-type and genetically altered strains.

The stability of the overall metabolome data despite the deletion of the ferrichrome synthetase gene in sib1Δ is an important asset for the reproducibility and reliability of the metabolomic results we obtained. It is consistent with the observation that the sib1⁺gene was not essential for a normal cell division cycle. Metabolic profiling of sib1Δ supported the identification of ferrichrome as its peak was missing, and suggested that the ornithine-glycinedipeptide accumulated in sib1Δ might be a precursor metabolite for ferrichrome, hinting at the mechanism for ferrichrome synthesis.

The peak area of HMG-CoA was rather small and not detectable by routine LC-MS analysis so that narrowing of the mass spectrometer scanning range was required to observe its peak. The identified peak was confirmed, as it was absent in hcs1-143 mutant cells defective in HMG-CoA synthase. HMG-CoA is the substrate for HMG-CoA reductase, the key enzyme for the mevalonate pathway. Despite the apparently low abundance of free HMG-CoA, an Hcs1 synthase defect caused extensive secondary changes in the metabolic profile; a large increase in acetylated amino acids, and a decrease of urea cycle components, S-adenosyl-homocysteine and Arg-Glu dipeptide. These changes might reflect the importance of HMG-CoA’s role in amino acid cellular regulation. The physiologic significance of such a wide range of alterations in metabolic profiles, however, remains to be investigated.

Whereas more than 6000 peaks were detected in the negative and positive ionization modes by the LC-MS procedures used in the present study, some of them (28%) were identified as broken fragments, adducts , or other type of derivatives produced during the ionization process within the mass spectrometer. The remaining ∼4000 peaks may still include a large amount of unidentified chemical ‘noise’. Sixty-three percent (77/123) of the identified metabolites were detected in both ionization modes, whereas only 7.5% (335/4476) of all peaks could be detected in both modes. Assuming that the 63% ratio is empirically valid for the whole metabolome, we speculate that ∼530 (335/63%) total metabolites were present among the ∼4000 candidate peaks. The number of actual unique compounds detectable within the routine LC-MS data using the samples prepared according to the procedures used in this study may therefore be on the order of several hundreds, rather than thousands. This kind of problem for identifying substances in metabolomics does not appear to exist in the case of proteome or transcriptome analyses, as transcripts and proteins can be rigorously and unambiguously determined. Metabolites are chemically diverse, often unstable and highly reactive, so that they are susceptible to various structural changes. Moreover, not all metabolites are detectable using the LC-ESI-MS method. Application of other detection techniques such as gas-chromatography mass-spectrometry (GC-MS) or nuclear magnetic resonance (NMR) may be necessary to cover a larger portion of the whole metabolome. GC-MS provides powerful libraries of thousands of already identified spectra, however detection of highly polar metabolites such as nucleotides or sugar-phosphates, as well as isolation of novel compounds not yet present in the databases, are rather difficult. NMR can further be applied to determine the structural information of unknown metabolites, if these can be purified from cell extracts in sufficient amount.

The normalized peak areas of the 123 metabolites identified in wild-type S. pombe cells differed on the order of 10⁴-fold. Trehalose detected at 36 °C had the largest peak area (1487 ± 246), while 3′5′-cyclic AMP, CMP, or coenzyme A belonged to a group having the smallest peaks (0.1–0.4). The semi-quantitative metabolic profiles compared at 26 °C and 36 °C were highly reproducible and similar but not identical. The normalized peak areas of approximately 31% of the identified metabolites had a greater than 2-fold change, whereas in metabolic profiles of two independently obtained samples at 26 °C, the normalized peak areas of only 4% of the compounds exceeded a 2-fold change. Certain compounds such as trehalose, glycerophosphoethanolamine, arabitol, ribulose, and ophthalmic acid had significantly increased peak areas at 36 °C. Trehalose, a disaccharide, is a well known stabilizer of proteins and membranes, and is produced in response to heat, cold, and dehydration.⁴⁰Glycerophosphoethanolamine is a product of phospholipiddeacylation, which reportedly acts as a growth stimulator for hepatocytes.⁴¹ The reason for the increase in this compound at 36 °C is unknown. Arabitol is a pentose-alcohol produced from D-ribulose in yeast cells,⁴² therefore the increase in both arabitol and ribulose might be consistent. Arabitol is used as a non-cariogenic substitute sweetener. Ophthalmic acid, initially discovered in calf lens, seems to be a ubiquitous tripeptide analogue of glutathione in which the cysteine group is replaced with L-2-aminobutyrate. Its level increases when the level of glutathione decreases in mouse liver so that the compound may be used as a marker for oxidative stress in the liver.⁴³ In the case of S. pombe, however, the peak area of glutathione was large (338–384) and was not changed by the shift to 36 °C despite the observed increase in ophthalmic acid; therefore, the level of ophthalmic acid in S. pombe was apparently not linked to that of glutathione. The role of ophthalmic acid in fungi is unknown. In S. pombe, it may be possible to investigate the role of ophthalmic acid by genetically modifying the enzymes required for its synthesis.

The peak areas of the amino acids generally increased at 36 °C. This might be due to more active protein synthesis at 36 °C to accelerate the cell division time (∼3.8 h at 26 °C versus ∼2.5 h at 36 °C). Arginine dramatically decreased, whereas dimethyl-arginine increased. The peak area of lysine did not change at 36 °C, but the peak areas of dimethyl- and trimethyl-lysine significantly increased at 36 °C. The peak area of methyl-adenosine also increased. These results suggested that the methylation reaction became abundant at 36 °C, consistent with the peak area increase for the methyl–donor compound S-adenosyl methionine (from 18.3 at 26 °C to 77.8 at 36 °C).

We showed that peak areas could not be easily converted to actual (molar) amounts, but a rough approximation may be obtained by injecting a pure compound with the same amount of spiked internal standard (125 pmol of PIPES and HEPES) and comparing the normalized peak areas with the obtained data. We used this approach to estimate the amount of ATP and ADP in wild-type S. pombe cells grown at 26 °C. The normalized peak area of ATP in the cell extract was 160.5, and that of ADP was 48.2. The normalized peak area of a pure 100-pmol injection was 40.8 for ATP and 46.2 for ADP. The normalized peak area of a pure 1000-pmol injection was 373.9 for ATP and 307.5 for ADP. Simple linear approximation leads to an amount of 423.4 pmol of ATP and 107 pmol of ADP within the injected cell extract sample (1 μl volume). The total sample volume was 80 μl and 2.5 × 10⁸ cells were used to prepare the sample, therefore we can estimate the number of molecules per cell as:

ATP: (424.4 pmol μl⁻¹ × 80 μl)/(2.5 × 10⁸ cells) = 135.5 amol per cell (∼8 × 10⁷ molecules per cell)

ADP: (107 pmol μl⁻¹ × 80 μl)/(2.5 × 10⁸ cells) = 34.2 amol per cell (∼2 × 10⁷ molecules per cell)

Assuming an average intracellular volume of 148.5 μm³,³⁵ we obtained the following intracellular concentrations:

ATP: (135.5 amol/148.5 μm³) = 0.91 mM

ADP: (34.2 amol/148.5 μm³) = 0.23 mM

These results are in good agreement with previously reported values of 0.94 mM ATP and 0.22 mM ADP.⁴⁴ However, this approach may not always be reliable, as ionization efficiency of the same compound in cell extracts may differ unpredictably from that of the pure compound.⁴⁵

The 123 identified metabolites in this study may represent a small fraction of the whole metabolome, but provide a solid base for various targeted analyses, as the list covers a broad range of physiologically important metabolites and the magnitude of observable change in the peak areas was as much as 10⁴-fold. S. pombe thus proves to be an appropriate organism for monitoring both large and subtle metabolic responses to cellular states and environmental conditions. As shown by the use of sib1Δ and hcs1-143 mutants, the approaches of genetic alteration are the most promising. The LC-MS procedures combined with the use of mutants indicate a systematically dissecting way for understanding the cellular metabolism under genetic perturbations. Additionally, mutant analysis provides the most convincing evidence for identifying compounds. Furthermore, we are currently undertaking analyses of metabolic states caused by nutritional changes such as nitrogen deficiency that promotes profound cellular state and morphology.²⁹

Materials and methods

Chemicals and reagents

Approximately 200 pure standards (listed in Supplementary Table 5† ) were purchased and analyzed to verify the identity of LC-MS peaks. HPLC-grade acetonitrile, methanol, and distilled water were obtained from Wako.

Strains and growth conditions

The following S. pombe strains were used in this study: a heterothallic haploid 972 h⁻ wild-type strain;⁴⁶ haploid h⁻sib1Δ (sib1::kanMX4) deletion strain produced by back-crossing an h⁺ ade6-M210(216) ura4-D18 leu1-32 sib1::kanMX4 strain obtained from a purchased deletion mutant library (Bioneer) with the 972 h⁻ wild-type strain; a temperature-sensitive (ts) hcs1-143 strain previously isolated.²⁹ Liquid synthetic EMM2 medium^47,48 was used for cell cultivation (the exact composition is specified in Supplementary Table 1† ). Wild-type cells were grown exponentially in EMM2 at 26 °C until the cell concentration reached 1.6 × 10⁶ cells ml⁻¹. One-half of the cultures was then diluted to 1 × 10⁶ cells ml⁻¹ and the culture temperature was shifted up to 36 °C for 6 h. The deletion mutant sib1Δ cells were grown at 26 °C only. The ts mutant hcs1-143 cells were grown at 26 °C and then shifted to 36 °C for 6 h. All cultures were harvested at a final concentration of 5 × 10⁶ cells ml⁻¹.

Cell harvesting and quenching

Fifty millilitres of cell culture were used for each sample. Cells were harvested by vacuum filtration using an Omnipore membrane filter (Millipore, 47 mm diameter and 1 μm pore size). The filter was immediately dropped into a 50 ml Falcon tube containing 25 ml methanol previously chilled to −40 °C. Tubes were kept at −40 °C for approximately 3 min, mixed with a vortexer for 10 s and the filter was removed using forceps. Cells were collected by 3000 rpm centrifugation at −20 °C for 3 min and the supernatant was discarded. The pellet was re-suspended in 1 ml 50% methanol (−40 °C) and 10 μl of 1 mM PIPES + HEPES mixture was added as an internal standard.

Metabolite extraction

Harvested S. pombe cells were disrupted by shaking with glass beads (Sigma-Aldrich, catalog number G8772) on a Multi-Beads Shocker (Yasui Kikai) 10 times with 1-min cycles, 2700 rpm at 0 °C. Proteins were removed by filtering the extracts on an Amicon Ultra 10-kDa cut-off filter (Millipore) for 60 min at 4 °C. The resulting filtrate volume was approximately 800 μl per sample.

Sample concentration

Sample solvent was evaporated by centrifugation on a vacuum concentrator (TOMY CC-105) for 60 min and the remaining liquid was re-suspended into 40 μl 50% acetonitrile by gentle pipetting. The final sample volume was approximately 80 μl, therefore the amount of spiked internal standards (PIPES + HEPES) in a 1-μl LC-MS injection volume was 125 pmol (10 μl × 1 mM/80 μl = 125 pmol).

LC-MS conditions

The hydrophilic interaction liquid chromatography method, which provides good separation of polar compounds as well as improved ionization efficiency at the mass spectrometer interface,⁴⁹ was applied. LC-MS data were obtained using a Paradigm MS4 HPLC system (Michrom Bioresources) coupled to an LTQ Orbitrap hybrid ion-trap/Fourier transform mass spectrometer (Thermo Fisher Scientific). LC separation was performed on a ZIC-pHILIC column (Merck SeQuant; 150 × 2.1 mm, 5 μm particle size). Acetonitrile (A) and 10 mM ammonium carbonatebuffer, pH 9.3 (B) were used as the mobile phase, with gradient elution from 80% A (20% B) to 20% A (80% B) in 30 min and 100 μl min⁻¹ flow rate. For MS detection, an electrospray ionization (ESI) source was used. Each sample was injected twice (1 μl volume/injection), once with the ESI operated in negative ionization mode and once in positive ionization mode. Spray voltage was set to 2.8 kV (negative ESI) or 4 kV (positive ESI) and capillary temperature to 350 °C or 300 °C, respectively. Nitrogen was used as the carrier gas. The mass spectrometer was operated in a full scan mode with a 100–1000 m/z scan range and automatic data-dependent MS/MS fragmentation scans.

Data processing

Raw data were analyzed using the current development version (1.95) of the MZmine 2 software (http://mzmine.sourceforge.net). The exact parameter values used for each step of the data analysis are specified in Supplementary Table 6.† After performing peak detection, isotopic peaks were removed and peak lists of individual samples were aligned using their corresponding m/z and retention time values. Finally, peak areas were normalized by a weighted contribution of the internal standards (PIPES and HEPES). Each sample was prepared three times. Mean and standard deviations of normalized peak areas were calculated (Supplementary Table 2† ).

Identification of peaks

The error of the m/z values provided by the LTQ Orbitrap is within 5 ppm at 400 m/z,²⁸ therefore molecular mass can be used as a key factor for metabolite identification. By observing the distance between isotopes of individual peaks, we determined that most of the metabolite ions were single-charged (protonated in case of positive ESI or deprotonated in case of negative ESI). We used the MZmine 2 built-in methods to recognize fragments, adducts , and complexes. Fragment peaks were identified using two criteria. First, they had the same retention time (<10 s difference) with another (parent) peak of higher mass. Second, their m/z value was present in the parent peak’s MS/MS pattern. Adduct peaks were identified by comparing retention times (<10 s difference) and m/z difference (21.983 for Na adduct , 37.956 for K adduct , 21.969 for Mg adduct , 17.027 for NH₃adduct , 97.977 for H₃PO₄adduct , 97.967 for H₂SO₄adduct , and 62 for H₂CO₃adduct ). Peak complexes were identified as peaks co-eluting (<10 s difference) with a pair of peaks of smaller mass, whose sum of m/z values was equal to the m/z of the complex peak (plus or minus mass of single proton, depending on the ionization mode). The remaining unidentified peaks were treated as potential metabolites. We attempted to identify the largest peaks and peaks showing significant alteration among sample groups. Metabolite candidates were assigned using the following approach. First, neutral mass was calculated by removing/adding the mass of a proton (mass = m/z ± 1.0078), depending on the ionization mode. Second, the PubChem compound database (NCBI) was searched for the calculated monoisotopic mass within the 5 ppm error limit. For meaningful candidates, the theoretical isotope pattern was predicted using the Xcalibur software (Thermo Fisher Scientific) and compared to the actual isotope distribution of the raw data. When a good match was found, a pure standard was purchased (if available) to verify its retention time and MS/MS fragmentation pattern.

Scatter plot visualization

A software system called MZviewer was developed using the platform of MZmine 2 to visualize the results of the LC-MS experiments. The software allows presentation of LC-MS data in a table form (including m/z, retention time, peak shapes and areas for each peak) or in a form of a dynamically generated scatter plot (Fig. 5 and 6). The complete software package including all source code and experimental data is available for download.‡

Acknowledgements

We thank Dr Koji Nagao for his contribution at the early stage of this investigation and Dr Yoshiya Oda from Eisai Co., Ltd. for his kind assistance with HPLC method development. We also thank the Okinawa Institute of Science and Technology Promotion Corporation for providing the funding of our research.

References

1. A. R. Fernie, R. N. Trethewey, A. J. Krotzky and L. Willmitzer, Metabolite profiling: from diagnostics to systems biology, Nat. Rev. Mol. Cell Biol., 2004, 5, 763–769.
2. J. L. Griffin, The Cinderella story of metabolic profiling: does metabolomics get to go to the functional genomics ball?, Philos. Trans. R. Soc. London, Ser. B, 2006, 361, 147–161.
3. A. Sreekumar, L. M. Poisson, T. M. Rajendiran, A. P. Khan, Q. Cao, J. Yu, B. Laxman, R. Mehra, R. J. Lonigro, Y. Li, M. K. Nyati, A. Ahsan, S. Kalyana-Sundaram, B. Han and X. Cao, et al., Metabolomic profiles delineate potential role for sarcosine in prostate cancer progression, Nature, 2009, 457, 910–914.
4. P. D. Whitfield, A. J. German and P. J. Noble, Metabolomics: an emerging post-genomic tool for nutrition, Br. J. Nutr., 2004, 92, 549–555.
5. J. Förster, I. Famili, P. Fu, B. Ø. Palsson and J. Nielsen, Genome-scale reconstruction of the Saccharomyces cerevisiae metabolic network, Genome Res., 2003, 13, 244–253.
6. K. Urano, K. Maruyama, Y. Ogata, Y. Morishita, M. Takeda, N. Sakurai, H. Suzuki, K. Saito, D. Shibata, M. Kobayashi, K. Yamaguchi-Shinozaki and K. Shinozaki, Characterization of the ABA-regulated global responses to dehydration in Arabidopsis by metabolomics, Plant J., 2009, 57, 1065–1078.
7. M. J. van der Werf, K. M. Overkamp, B. Muilwijk, L. Coulier and T. Hankemeier, Microbial metabolomics: toward a platform with full metabolome coverage, Anal. Biochem., 2007, 370, 17–25.
8. Y. Chikashige, C. Tsutsumi, M. Yamane, K. Okamasa, T. Haraguchi and Y. Hiraoka, Meiotic proteins bqt1 and bqt2 tether telomeres to form the bouquet arrangement of chromosomes, Cell, 2006, 125, 59–69.
9. Y. Harigaya, H. Tanaka, S. Yamanaka, K. Tanaka, Y. Watanabe, C. Tsutsumi, Y. Chikashige, Y. Hiraoka, A. Yamashita and M. Yamamoto, Selective elimination of messenger RNA prevents an incidence of untimely meiosis, Nature, 2006, 442, 45–50.
10. N. F. Kaufer, V. Simanis and P. Nurse, Fission yeast Schizosaccharomyces pombe correctly excises a mammalian RNA transcript intervening sequence, Nature, 1985, 318, 78–80.
11. P. Nurse, Universal control mechanism regulating onset of M-phase, Nature, 1990, 344, 503–508.
12. B. Schafer, Genetic conservation versus variability in mitochondria: the architecture of the mitochondrial genome in the petite-negative yeast Schizosaccharomyces pombe, Curr. Genet., 2003, 43, 311–326.
13. W. M. Toone and N. Jones, Stress-activated signalling pathways in yeast, Genes Cells, 1998, 3, 485–498.
14. T. A. Volpe, C. Kidner, I. M. Hall, G. Teng, S. I. Grewal and R. A. Martienssen, Regulation of heterochromatic silencing and histone H3 lysine-9 methylation by RNAi, Science, 2002, 297, 1833–1837.
15. M. Yanagida, Basic mechanism of eukaryotic chromosome segregation, Philos. Trans. R. Soc. London, Ser. B, 2005, 360, 609–621.
16. V. Wood, R. Gwilliam, M. A. Rajandream, M. Lyne, R. Lyne, A. Stewart, J. Sgouros, N. Peat, J. Hayles, S. Baker, D. Basham, S. Bowman, K. Brooks, D. Brown and S. Brown, et al., The genome sequence of Schizosaccharomyces pombe, Nature, 2002, 415, 871–880.
17. M. Aslett and V. Wood, Gene Ontology annotation status of the fission yeast genome: preliminary coverage approaches 100%, Yeast, 2006, 23, 913–919.
18. C. Hertz-Fowler, C. S. Peacock, V. Wood, M. Aslett, A. Kerhornou, P. Mooney, A. Tivey, M. Berriman, N. Hall, K. Rutherford, J. Parkhill, A. C. Ivens, M. A. Rajandream and B. Barrell, GeneDB: a resource for prokaryotic and eukaryotic organisms, Nucleic Acids Res., 2004, 32, 339D–343.
19. M. Kanehisa and S. Goto, KEGG: kyoto encyclopedia of genes and genomesm, Nucleic Acids Res., 2000, 28, 27–30.
20. D. Chen, W. M. Toone, J. Mata, R. Lyne, G. Burns, K. Kivinen, A. Brazma, N. Jones and J. Bahler, Global transcriptional responses of fission yeast to environmental stress, Mol. Biol. Cell, 2003, 14, 214–229.
21. J. Mata, R. Lyne, G. Burns and J. Bahler, The transcriptional program of meiosis and sporulation in fission yeast, Nat. Genet., 2002, 32, 143–147.
22. G. Rustici, J. Mata, K. Kivinen, P. Lio, C. J. Penkett, G. Burns, J. Hayles, A. Brazma, P. Nurse and J. Bähler, Periodic gene expression program of the fission yeast cell cycle, Nat. Genet., 2004, 36, 809–817.
23. M. Shimanuki, S. Y. Chung, Y. Chikashige, Y. Kawasaki, L. Uehara, C. Tsutsumi, M. Hatanaka, Y. Hiraoka, K. Nagao and M. Yanagida, Two-step, extensive alterations in the transcriptome from G0 arrest to cell division in Schizosaccharomyces pombe, Genes Cells, 2007, 12, 677–692.
24. B. T. Wilhelm, S. Marguerat, S. Watt, F. Schubert, V. Wood, I. Goodhead, C. J. Penkett, J. Rogers and J. Bahler, Dynamic repertoire of a eukaryotic transcriptome surveyed at single-nucleotide resolution, Nature, 2008, 453, 1239–1243.
25. C. J. Bolten and C. Wittmann, Appropriate sampling for intracellular amino acid analysis in five phylogenetically different yeasts, Biotechnol. Lett., 2008, 30, 1993–2000.
26. M. E. Weeks, J. Sinclair, A. Butt, Y. L. Chung, J. L. Worthington, C. R. Wilkinson, J. Griffiths, N. Jones, M. D. Waterfield and J. F. Timms, A parallel proteomic and metabolomic analysis of the hydrogen peroxide- and Sty1p-dependent stress response in Schizosaccharomyces pombe, Proteomics, 2006, 6, 2772–2796.
27. A. B. Canelas, C. Ras, A. t. Pierick, J. C. v. Dam, J. J. Heijnen and W. M. v. Gulik, Leakage-free rapid quenching technique for yeast metabolomics, Metabolomics, 2008, 4, 226–239.
28. A. Makarov, E. Denisov, A. Kholomeev, W. Balschun, O. Lange, K. Strupat and S. Horning, Performance evaluation of a hybrid linear ion trap/orbitrap mass spectrometer, Anal. Chem., 2006, 78, 2113–2120.
29. K. Sajiki, M. Hatanaka, T. Nakamura, K. Takeda, M. Shimanuki, T. Yoshida, Y. Hanyu, T. Hayashi, Y. Nakaseko and M. Yanagida, Genetic control of cellular quiescence in S. pombe, J. Cell Sci., 2009, 122, 1418–1429,  DOI:10.1242/jcs.046466.
30. M. Katajamaa, J. Miettinen and M. Oresic, MZmine: toolbox for processing and visualization of mass spectrometry based molecular profile data, Bioinformatics, 2006, 22, 634–636.
31. T. Kind and O. Fiehn, Metabolomic database annotations via query of elemental compositions: mass accuracy is insufficient even at less than 1 ppm, BMC Bioinformatics, 2006, 7, 234.
32. A. Seidel, S. Brunner, P. Seidel, G. I. Fritz and O. Herbarth, Modified nucleosides: an accurate tumour marker for clinical diagnosis of cancer, early detection and therapy control, Br. J. Cancer, 2006, 94, 1726–1733.
33. T. Schwecke, K. Gottling, P. Durek, I. Duenas, N. F. Kaufer, S. Zock-Emmenthal, E. Staub, T. Neuhof, R. Dieckmann and H. von Dohren, Nonribosomal peptide synthesis in Schizosaccharomyces pombe and the architectures of ferrichrome-type siderophore synthetases in fungi, ChemBioChem, 2006, 7, 612–622.
34. Q. Li, A. Ricardo, S. A. Benner, J. D. Winefordner and D. H. Powell, Desorption/ionization on porous silicon mass spectrometry studies on pentose-borate complexes, Anal. Chem., 2005, 77, 4503–4508.
35. J. M. Mitchison, The growth of single cells. I. Schizosaccharomyces pombe, Exp. Cell Res., 1957, 13, 244–262.
36. J. Fernández, T. Soto, J. Vicente-Soler, J. Cansado and M. Gacto, Heat-shock response in Schizosaccharomyces pombe cells lacking cyclic AMP-dependent phosphorylation, Curr. Genet., 1997, 31, 112–118.
37. S. Labbé, B. Pelletier and A. Mercier, Iron homeostasis in the fission yeast Schizosaccharomyces pombe, BioMetals, 2007, 20, 523–537.
38. T. Hayashi, Y. Fujita, O. Iwasaki, Y. Adachi, K. Takahashi and M. Yanagida, Mis16 and Mis18 are required for CENP-A loading and histone deacetylation at centromeres, Cell, 2004, 118, 715–729.
39. J. L. Goldstein and M. S. Brown, Regulation of the mevalonate pathway, Nature, 1990, 343, 425–430.
40. A. D. Elbein, Y. T. Pan, I. Pastuszak and D. Carroll, New insights on trehalose: a multifunctional molecule, Glycobiology, 2003, 13, 17R–27R.
41. C. Nelson, B. Moffat, N. Jacobsen, W. J. Henzel, J. T. Stults, K. L. King, A. McMurtrey, R. Vandlen and S. A. Spencer, Glycerophosphorylethanolamine (GPEA) identified as an hepatocyte growth stimulator in liver extracts, Exp. Cell Res., 1996, 229, 20–26.
42. T. Suzuki and H. Onishi, Aerobic dissimilation of D-ribulose by yeasts, Antonie van Leeuwenhoek, 1970, 36, 227–232.
43. T. Soga, R. Baran, M. Suematsu, Y. Ueno, S. Ikeda, T. Sakurakawa, Y. Kakazu, T. Ishikawa, M. Robert, T. Nishioka and M. Tomita, Differential metabolomics reveals ophthalmic acid as an oxidative stress biomarker indicating hepatic glutathione consumption, J. Biol. Chem., 2006, 281, 16768–16776.
44. S. W. Ingram and L. D. Barnes, Disruption and overexpression of the Schizosaccharomyces pombe aph1 gene and the effects on intracellular diadenosine 5′,5‴-P1,P4-tetraphosphate (Ap4A), ATP and ADP concentrations, Biochem. J., 2000, 350(3), 663–669.
45. A. Kloepfer, J. B. Quintana and T. Reemtsma, Operational options to reduce matrix effects in liquid chromatography-electrospray ionization-mass spectrometry analysis of aqueous environmental samples, J. Chromatogr., A, 2005, 1067, 153–160.
46. H. Gutz, H. Heslot, U. Leupold and N. Loprieno, Schizosaccharomyces pombe, Handb. Genet., 1974, 1, 395–446.
47. J. M. Mitchison, Physiological and cytological methods for Schizosaccharomyces pombe, Methods Cell Biol., 1970, 4, 131–165.
48. P. Nurse, Genetic control of cell size at cell division in yeast, Nature, 1975, 256, 547–551.
49. H. P. Nguyen and K. A. Schug, The advantages of ESI-MS detection in conjunction with HILIC mode separations: Fundamentals and applications, J. Sep. Sci., 2008, 31, 1465–1480.
50. M. Schrettl, G. Winkelmann and H. Haas, Ferrichrome in Schizosaccharomyces pombe—an iron transport and iron storage compound, BioMetals, 2004, 17, 647–654.

---

Footnotes

† Electronic supplementary information (ESI) available: Supplementary figures and tables. See DOI: 10.1039/b908784b

‡ The raw data in mzML 1.1 format (4 GB), the MZmine 2 software (51 MB), and the MZviewer software including all processed peak lists and complete source codes (62 MB) can be downloaded from ProteomeCommons.org Tranche, https://proteomecommons.org/tranche/, using the following hash:

1JRSjOJBW1QWZgnRYyr4QdUATFrXDb27udKAsvymLrLcnB8GEl/q/lNwCPXY4krH+GQpNc9WGu5tZDa+VEKyIr9eWt4AAAAAAA6AXQ==

The hash may be used to determine the exact files that were published as part of this manuscript's data set, and the hash may also be used to check that the data has not changed since publication.

---