Depiction of metabolome changes in histidine-starved Escherichia coli by CE-TOFMS

Abstract

Metabolic changes in response to histidine starvation were observed in histidine-auxotrophic Escherichia coli using a capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS)-based metabolomics technique. Prior to the analysis, we prepared an E. coli metabolome list of 727 metabolites reported in the literature. An improved method for metabolite extraction was developed, which resulted in higher extraction efficiency in phosphate-rich metabolites, e.g., ATP and GTP. Based on the results, 375 charged, hydrophilic intermediates in primary metabolisms were analysed simultaneously, providing quantitative data of 198 metabolites. We confirmed that the intracellular levels of intermediates in histidine biosynthesis are rapidly accumulated in response to a drop in histidine level under histidine-starved conditions. Simultaneously, disciplined responses were observed in the glycolysis, tricarboxylic acid cycle, and amino acid and nucleotide biosynthesis pathways as regulated by amino acid starvation.

---

Introduction

Metabolomics, the unbiased determination of metabolite levels, is expected to be a valuable approach for the characterisation of bioprocesses in combination with genomics, transcriptomics, and proteomics. Different from the other “omics”, metabolomics includes methodological problems derived from heterogeneity in chemical properties, that is, it involves the development of extraction methods of metabolites that can be applied to a broad range of chemical species, and analytical methods that are achieved by as few processes as possible. However, it is impossible for all metabolites to be detected by single analytical method at present.

The chemical properties of metabolites are widely diverse, but the majority of them have a few common properties in terms of physico-chemical traits: molecular weight less than 500, high hydrophilicity,¹ and charge in aqueous solution.² On the basis of these properties, we have developed a metabolome analysis method using capillary electrophoresis electrospray ionization mass spectrometry (CE-MS) to determine the majority of metabolic intermediates.^3–7 We found 150 metabolites in sporulating Bacillus subtilis, and analysed the changes in profiles in major energy metabolisms.⁶ However, it is unclear how many metabolites in bacterial cells can be determined by our analytical system. Extraction of metabolites from cells is also one of the most dominant factors in metabolomics. Extraction must be adapted to the target metabolites, organisms, and analytical methods employed. To date, many extraction methods have been comparatively studied, with cold methanol extraction,^8,9 and acidic acetonitrile extraction¹⁰ as the recommended techniques. We have employed a quenching method by pure methanol followed by methanol-water-chloroform extraction to remove hydrophobic metabolites.⁶ While this method satisfies particular constraints in CE-MS analysis, the efficiency and data reproducibility is not sufficient especially in phosphate-rich metabolites, e.g., ATP. In this study, we modified the method to improve these problems, and applied it to determine the intermediates of histidine biosynthesis in Escherichia colihistidine-auxotroph.

The stringent response is known to be a starvation-stimulated adaptation mechanism which includes functional arrests of chromosome replication, cell division, transcription, translation, and metabolisms.¹¹ Translation arrest induced by amino acid starvation on ribosomes activates RelA protein , resulting in temporal accumulation of a bacterial alarmone, guanosine 3′-diphosphate 5′-diphosphate (ppGpp). Recently, transcriptome and proteome analyses have been demonstrated in bacterial stringent responses, indicating the possibility that metabolisms are dramatically influenced by amino acid starvation.^12–14 However, no metabolomics approach has been used, raising a question of how to change the metabolism in the starved, auxotrophic situation in bacteria.

In this study, we prepared a complete metabolite list for E. coli by searching databases and the literature. The method of metabolome extraction was also improved, realising good metabolome analyses using capillary electrophoresis electrospray ionization time-of-flight mass spectrometry (CE-TOFMS). We applied the method to the analysis of the E. colihistidine-auxotroph and demonstrated through the analysis of 375 metabolites including primary metabolism intermediates that the metabolome profile of E. coli is dramatically changed during amino acid starvation. The quantitative data obtained for 198 hydrophilic, charged metabolites provides new insights in bacterial starvation.

Results and discussion

Target metabolites of CE-MS based metabolomics

The definition of the metabolome that is widely accepted in various fields is the entire set of low-molecular weight intermediates in metabolisms including amino acids, amines , nucleotides , sugars , lipids , and other substances. Generally, DNA, RNA , and protein are excluded from the definition, but their digests are occasionally recognised to be the targets of metabolomics. These metabolites are widely diverse in their chemical properties, i.e., polarity that is a determinant of solubility, electronic charge of ions, volatility, and molecular weight. Metabolites found in E. coli cells are summarised in the EcoCyc database,^15,16 and their physico-chemical properties have been previously observed.¹ Prior to the metabolome analysis, we prepared a list of E. coli metabolites to understand the target metabolites in CE-TOFMS based metabolomics. Our survey of the known E. coli metabolites, carried out through a database and literature search (see Experimental), resulted in a list of metabolome including 727 metabolites (supplementary Table 1† ). Of these, 453 (62%) belong to primary metabolites and their degradation intermediates (classes I and II), which are expected to be found in cells grown in minimal medium (Fig. 1A). Since most of the primary metabolites, for which the number is calculated as 92% (hydrophilic in Fig. 1B) × 96% (anion, cation, and nucleotide in Fig. 1C) = 88% (402 metabolites), are hydrophilic and charged, it is expected that CE-TOFMS, which principally targets water-soluble, charged metabolites is an extremely promising analysis method for metabolomics in primary pathways. Since some of the metabolites (e.g., acetic acid) cannot be detected rationally, the number of the target metabolites in CE-TOFMS analysis was 375, which corresponds to 83% of the primary metabolites in E. coli.

Fig. 1 Classification of E. coli metabolites. A. Distribution of metabolites by metabolic pathways. The classes are described in the Experimental section. B. Hydrophobicity of metabolites in classes I and II. H₂O was excluded from the calculation. C. Distribution of analytical modes in CE-TOFMS analysis for hydrophilic metabolites in classes I and II. The neutral metabolites cannot be analysed by CE-MS.

Optimisation of metabolome extraction procedure

A problem was inherent in our previous studies regarding bacterial metabolome analyses.^5,6 The extraction efficiencies of phosphate-rich metabolites such as ATP were often relatively low, affecting reproducibility of the analyses. We considered that the problem may be due to the interaction between phosphate groups of metabolites and phospholipids in the cell membrane. To improve this problem, we introduced sonication treatment during methanol extraction of metabolites from the cells (Fig. 2 and Experimental section). By this additional process, cells are peeled from the filter and completely suspended in methanol. Here, an ultrasonic syringe is suitable rather than a cell disruptor, since the methanol is transpired by excessively strong ultrasonic of cell disruptor, resulting in fluctuation and enhanced error of the metabolomics data (data not shown). To compare the extraction efficiencies, we used four methods for the methanol extraction process; (i) with sonication (improved method), (ii) without sonication (conventional method), (iii) with sonication, but cells are removed by brief centrifugation before chloroform addition (sonicated cell removal method), and (iv) with incubation at −80 °C for 12 h, but cells are removed (cold method).

Fig. 2 Metabolome extraction procedure modified in this study. A. Schematic of the metabolome extraction procedure. Modified point is indicated in the green box. B. Cell suspension prepared by sonication of methanol.

To examine the reproducibility of the improved method, we repeatedly collected the data obtained in the improved method (the data are distinguished as “1st” and “2nd”). Fig. 3A shows the reproducibility and improvement of extraction efficiency by sonication in nucleotide metabolites containing phosphate groups. The metabolome source was E. coli W3110 grown in rich medium at 37 °C until the middle-logarithmic growth phase. Strikingly, extraction efficiency of GTP was improved 130-fold relative to the conventional method reported previously.⁶ For adenosine-phosphates, ATP, ADP, and AMP, the signal intensities were enhanced 39-, 17-, and 1.6-fold, respectively, by the improved method. Enhancement of detected signal intensities depended on the number of phosphate groups in the metabolites, indicating that efficient extraction of phosphate-rich metabolites is difficult by methanol extraction alone. It has been assumed that the phenomenon is due primarily to interaction of the metabolites with cell membranes by the intense negative charge of phosphate, but further investigations are needed. The improvement of extraction efficiency was not achieved by the sonicated cell removal method or cold method (Fig. 3A). This result suggests that the cells (and enzymes in the cells) fixed in methanol are damaged or inverted by contact with chloroform, and release the metabolites including phosphate groups into the solvent. The extraction efficiency of most amino acids was independent of the extraction methods except for lysine and arginine (Fig. 3B). This suggests that extraction of basic metabolites is also conditional in methanol extraction. Another basic amino acid, histidine, was efficiently extracted by all methods, possibly due to its more feeble imidazole group charge. As the isoelectric points of arginine, lysine, and histidine are 10.76, 9.74, and 7.59, respectively, the extremely intense positive charges of arginine and lysine most likely interact with the phosphate group in the phospholipids of the cell membrane. On the other hand, a problem was raised in the improved method. In some metabolites, NADH, NADPH, and coenzyme A (CoA), the extraction efficiency was worsened relative to that of the conventional method (Fig. 3A). Levels of oxidised forms of these metabolites, NAD⁺, NADP⁺, and CoA dimer (CoA–CoA) were increased in the improved method. This was not found in the other methods. Total amounts of these oxidised and reduced forms of metabolites were enhanced in the improved method (data not shown), suggesting that the extraction of them is basically improved. However, decreased levels of the oxidised forms most likely suggest that the reduced forms of the metabolites are oxidised during preparation procedures. Although it is not clear why oxidation of the metabolites is accelerated in the improved method, enhanced levels of some oxidants, e.g., flavin nucleotides may stimulate the oxidation.

Fig. 3 Improved extraction efficiency of nucleotide metabolites by the improved method developed in this study. A. Comparison of the detected levels of nucleotide metabolites. The number of phosphate groups that are contained in the metabolite is shown in parentheses. B. Comparison of the detected levels of amino acids. ●, area ratio of [improved method (1st)] to [improved method (2nd)], that is, the reproducibility of the improved method; △, area ratio of [improved method (1st)] to [conventional method]; □, area ratio of [improved method (1st)] to [sonicated cell removal method]; ○, area ratio of [improved method (1st)] to [cold method]. See text for details of the extraction procedures.

Determination of histidine-biosynthesis intermediates using improved methods

Histidine is synthesised via ten steps which are evolutionarily conserved in all organisms that synthesise histidine.¹⁷ However, standard chemicals of most of the intermediates are not commercially available, making pathway analysis difficult (Fig. 4A). Providing the migration times of the intermediates in CE-TOFMS analysis enables the analysis of histidine biosynthesis as a part of metabolome analysis. Since, most intermediates in histidine biosynthesis are phosphate-including complex compounds which are detected by CE-TOFMS in nucleotide mode, we performed CE-TOFMS analyses of the metabolites extracted by the improved preparation method presented here. To accumulate the histidine intermediates, we employed an E. coli mutant (JW2002) carrying a null mutation in the hisDallele that codes for histidine dehydrogenase catalysing the conversion of L-histidinol to L-histidine viaL-histidinal.¹⁸

Fig. 4 Signal identification of histidine biosynthesis intermediates. A. Histidine biosynthesis pathway in E. coli. Metabolites that are detected in nucleotide mode by CE-TOFMS analysis are indicated in blue. Red circles show the metabolites for which standard chemicals are not commercially available. L-Histidinal coloured gray cannot be detected because it is not released during the overall reaction catalysed by histidine dehydrogenase.³¹ B. Accumulation of intermediates of histidine biosynthesis after histidine downshift. The data are indicated as the area values relative to that of internal standard. C. Electropherograms of the intermediates obtained by CE-TOFMS analyses. The metabolites except for L-histidinol and L-histidine were detected with nucleotide mode. The data are indicated as the signal intensities (arbitrary unit).

The histidine biosynthesis pathway is repressed, as a posttranslational regulation, by histidine at the first step of the pathway,¹⁹ and is down-regulated by attenuation of the gene expression under conditions of excess histidine.²⁰ The expression of hisoperon is also induced by amino acid starvation in a ppGpp-mediated manner.²¹ Hence, it is expected that amino acid starvation stimulates the accumulation of histidine biosynthesis intermediates. E. coli JW2002 was grown in a minimal medium supplemented with histidine and then resuspended in a minimal medium without histidine at the middle-to-late logarithmic growth phase. Accumulation of histidine biosynthesis intermediates are shown in Fig. 4B. The intermediates for which standard chemicals are unavailable were identified based on their calculated m/z values (Fig. 4C). A downshift of histidine in the medium activates histidine biosynthesis provoking accumulation of its intermediates. Actually, the intracellular level of histidine dropped rapidly within 30 min of the downshift, resulting in the activation of the histidine biosynthesis pathway. Though phosphoribosyl-ATP was not detected here, chronological accumulation profiles of the intermediates were observed. The phosphoribosylpyrophosphate (PRPP) level was increased at the onset of starvation (0–15 min), and then slightly decreased in concert with the drop of histidine concentrations (30–60 min). These results are consistent with the previous observations,¹⁹ and clearly indicate the accumulation of intermediates in histidine biosynthesis. Additionally, among the intermediates only PRPP and L-histidinol are commercially available at present, and thus the results yield important information of CE-TOFMS analysis in histidine biosynthesis.

Metabolome profiles under histidine starvation conditions

Amino acid starvation induces the accumulation of ppGpp, stimulating stringent response in cells.¹¹ Actually, ppGpp levels were significantly increased after the histidine downshift (supplementary Table 2† ). This observation led us to view all the metabolite changes during histidine starvation as the stringent response. We determined the metabolome profiles including 375 metabolites belonging to classes I and II (supplementary Table 2). Of the metabolites, a total of 198 were successfully detected in cells during histidine starvation, and are depicted in the E. coli metabolic pathway map shown in Fig. 5 (high resolution, printable version is available as supplementary Fig. 1† ). Of the metabolites detected in cells, 45 were determined based on their m/z values because the standard chemicals are not available. At a glance, dramatic changes were found in regional primary metabolisms, but most of the metabolites were not changed. Since the metabolic network in E. coli is generally robust against environmental perturbations,²² the metabolic changes found here seem to involve homeostasis of metabolic processes.

Fig. 5 Metabolomic profile of E. coli after histidine downshift. The data of each metabolite level are summarized in supplementary Table 2. Enlarged, printable version is available as supplementary Figure 1.†

To identify the metabolic changes, we performed clustering analysis of the metabolome data using hierarchical clustering analysis (HCA), generating five significant clusters. Clustering analysis is applicable and valuable in transcriptomic analysis, since the accumulation of a specific mRNA directly shows the activation of the gene expression.²³ However, it should be noted in metabolomics that the accumulation of a metabolite does not indicate activation of the metabolic pathway, it should be estimated in combination with the change of its down- and upstream metabolites. The clusters generated here are shown in Fig. 6 and listed in Table 1. The metabolites for which levels were unchanged were excluded from HCA. Clusters 1, 2, and 3 resembled each other in the profiles and tend to be accumulated after histidine downshift (Fig. 6B). Cluster 1 metabolites were temporary accumulated during 15 to 30 min after the histidine downshift, and then decreased (Fig. 6B). Cluster 2 metabolites were rapidly accumulated at 15 min and maintained thereafter (Fig. 6B). Accumulations of cluster 3 metabolites were delayed to 60 min (Fig. 6B). Clusters 4 and 5 tended to be decreased gradually and rapidly, respectively (Fig. 6B).

Table 1 List of metabolites clustered by HCA

Cluster No.	Pathway	Class^a	Metabolite	Abbreviation	KEGG ID	Analytical mode
a Classes: I, primary matabolism; II, degradation of primary metabolites; III, Degradation of environmental compounds; IV, secondary or unconventional metabolism; V, pathway unknown; VI, intermediates in putative in vitro reaction.
1	Arginine biosynthesis	I	N-Acetyl-L-glutamic acid	N-AcGlu	C00624	Cation
1	Cysteine biosynthesis	I	O-Acetyl-L-serine	AcSer	C00979	Cation
1	Leucine biosynthesis	I	3-Isopropylmalic acid	3IPMA	C04411	Anion
1	Methionine biosynthesis	I	Homocysteine	HomoCys	C05330	Cation
1	Pyruvic acid oxidation	I	Acetyl-CoA	AcCoA	C00024	Nucleotide
1	Salvage of Ade, HypXan	I	Adenine	—	C00147	Cation
1	Serine biosynthesis	I	3-Phosphoserine	3PSer	C01005	Anion
2	Arginine biosynthesis	I	L-Citrulline	Citrulline	C00327	Cation
2	Arginine biosynthesis	I	L-Arginosuccinic acid	ArgSuccinate	C03406	Cation
2	Biotin biosynthesis	I	7,8-Diaminononanoic acid	7,8 DANA	C01037	Cation
2	Cholanic acid biosynthesis	I	α-D-Galactose-1-phosphate	Gal1P	C00446	Anion
2	Cholanic acid biosynthesis	I	UDP-galactose	UDP-Gal	C00052	Nucleotide
2	Chorismic acid biosynthesis	I	3-Deoxy-D-arabinoheptulosonate-7-phosphate	3DAH7P	C04691	Anion
2	Chorismic acid biosynthesis	I	Shikimate-3-phosphate	S3P	C03175	Anion
2	dTDP-rhamnose biosynthesis	I	dTDP-D-glucose	dTDP-Glc	C00842	Nucleotide
2	Fravin biosynthesis	I	5-Amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione	5A6RbA2,4PD	C04732	Cation
2	Glutamic acid biosynthesis	I	L-Glutamic acid	Glu	C00025	Cation
2	Glutathione biosynthesis	I	L- γ-Glutamylcysteine	GluCys	C00669	Cation
2	Glycolysis(Embden-Meyerhof-Parnas)	I	α-D-Glucose-6-phosphate	G6P	C00092	Anion
2	Glycolysis (Embden-Meyerhof-Parnas)	I	D-Fructose-1,6-diphosphate	F1,6P	C00354	Anion
2	Glycolysis (Embden-Meyerhof-Parnas)	I	D-Glyceraldehyde-3-phosphate	G3P	C00661	Anion
2	Glycolysis (Embden-Meyerhof-Parnas)	I	Dihydroxyacetonephosphate	DHAP	C00111	Anion
2	Inorganic metabolite	I	Pyrophosphate	Ppi	C00013	Anion
2	Isopentenyldiphosphate biosynthesis	I	1-Hydroxy-2-methyl-2(E)-butenyl-4-diphosphate	HMeBuPP	C11811	Anion
2	Leucine biosynthesis	I	2-Ketoisocaproic acid	2KICA	C00233	Anion
2	Leucine biosynthesis	I	L-Leucine	Leu	C00123	Cation
2	Lysine biosynthesis	I	L,L-Diaminopimelic acid	DPA	C00666	Anion
2	Lysine biosynthesis	I	L-Lysine	Lys	C00047	Cation
2	Methionine biosynthesis	I	O-Succinyl-L-homoserine	SucHomoSer	C01118	Cation
2	Methionine biosynthesis	I	L-Methionine	Met	C00073	Cation
2	Methionine oxidation	I	L-Methionine sulfoxide	MetSFX	C02989	Cation
2	NAD biosynthesis	I	Deamido-NAD	—	C00857	Nucleotide
2	NAD biosynthesis	I	Nicotinamideadenine dinucleotidephosphate (reduced form)	NADP+	C00006	Nucleotide
2	NAD biosynthesis	I	Nicotinamideadenine dinucleotidephosphate (oxidized form)	NADPH	C00005	Nucleotide
2	Pentose phosphate pathway (non-oxidative)	I	D-Sedoheptulose-7-phosphate	S7P	C05382	Anion
2	Pentose phosphate pathway (non-oxidative)	I	D-Erythrose-4-phosphate	E4P	C00279	Anion
2	Pentose phosphate pathway (non-oxidative)	I	D-Ribose-5-phosphate	R5P	C00117	Anion
2	Pentose phosphate pathway (oxidative)	I	6-Phospho-D-gluconic acid	6-PG	C00345	Anion
2	Phenylalanine biosynthesis	I	Phenylpyruvic acid	—	C00166	Anion
2	Polyamine biosynthesis	I	Agmatine	—	C00179	Cation
2	Polyamine biosynthesis	I	S-Adenosylmethioninamine	dAdoMet	C01137	Cation
2	ppGpp biosynthesis	I	Guanosine-5′-diphosphate 3′-diphosphate	ppGpp	C01228	Nucleotide
2	Proline biosynthesis	I	L-Proline	Pro	C00148	Cation
2	PRPP biosynthesis	I	5-Phosphoribosyldiphosphate	PRPP	C00119	Nucleotide
2	PRPP biosynthesis	I	D-Ribose-1,5-diphosphate	R1.5P	C01151	Nucleotide
2	Purine biosynthesis	I	5′-Phosphoribosyl-N-formylglycineamide	FGAR	C04376	Nucleotide
2	Purine biosynthesis	I	Adenosine-5′-diphosphate	ADP	C00008	Nucleotide
2	Purine biosynthesis	I	Adenosine-5′-triphosphate	ATP	C00002	Nucleotide
2	Purine biosynthesis	I	Xanthosine-5-phosphate	XMP	C00655	Anion
2	Purine biosynthesis	I	Deoxyadenine-5′-triphosphate	dATP	C00131	Nucleotide
2	Purine biosynthesis	I	Guanosine-5′-diphosphate	GDP	C00035	Nucleotide
2	Purine biosynthesis	I	Guanosine-5′-triphosphate	GTP	C00044	Nucleotide
2	Pyridine nucleotide cycling	I	Nicotinamide	—	C00153	Cation
2	Pyridoxal-10′-phosphate biosynthesis	I	Pyridoxal-5′-phosphate	Pyr5′P	C00018	Anion
2	Pyrimidine biosynthesis	I	Carbamoyl-L-aspartic acid	Carbamoyl-Asp	C00438	Cation
2	Pyrimidine biosynthesis	I	Dihydroorotic acid	—	C00337	Anion
2	Pyrimidine biosynthesis	I	Deoxycytidine-5′-triphosphate	dCTP	C00458	Nucleotide
2	Salvage of Ade, HypXan	I	Adenosine	—	C00212	Cation
2	Salvage of Gua, Xan	I	Deoxyguanosine	dGuanosine	C00330	Cation
2	Salvage of pyridoxal-6′-phosphate	I	Pyridoxamine	—	C00534	Cation
2	Serine biosynthesis	I	L-Serine	Ser	C00065	Cation
2	Sulfuric acid assimilation	I	Adenosine-5′-phosphosulfate	APS	C00224	Nucleotide
2	Sulfuric acid assimilation	I	Adenosine-3′,5′-diphosphate	pAp	C00054	Nucleotide
2	TCA cycle	I	Citric acid	—	C00417	Anion
2	TCA cycle	I	cis-Aconitic acid	—	C00417	Anion
2	TCA cycle	I	2-Oxoglutaric acid	2-OG	C00026	Anion
2	TCA cycle	I	Succinyl-CoA	SucCoA	C00091	Nucleotide
2	TCA cycle	I	Succinic acid	—	C00042	Anion
2	TCA cycle	I	Fumaric acid	—	C00122	Anion
2	TCA cycle	I	Malic acid	—	C00497	Anion
2	Thiamine biosynthesis	I	Thiaminephosphate	ThiamineP	C010181	Cation
2	Threonine biosynthesis	I	O-Phospho-L-homoserine	P-HomoSer	C01102	Anion
2	Tyronine biosynthesis	I	p-Hydroxyphenylpyruvic acid	HPP	C01179	Anion
2	UDP-N-acetylglucosamine	I	UDP-N-acetyl-D-glucosamine	UDP-GlcNAc	C00043	Nucleotide
2	Valine biosynthesis	I	2-Acetolactic acid	2-Ac	C06010	Anion
2	Valine biosynthesis	I	2-Ketoisovaleric Acid	2-KIV	C00141	Anion
2	Valine biosynthesis	I	2,3-Dihydroxyisovaleric acid	2,3-DIV	C04039	Anion
2	Valine biosythesis	I	L-Valine	Val	C00183	Cation
2	Acetylglucosamine dissimilation	II	N-Acetylneuraminic acid	NeuNAc	C00270	Anion
2	Arginine degradation	II	N ^2-Succinylglutamic acid	N2SucGLu	C05931	Cation
2	Lysine degradation	II	Cadaverine	—	C01672	Cation
2	Methylmalonyl pathway	II	D-Methylmalonyl-CoA	—	C02557	Nucleotide
3	Alanine biosynthesis	I	L-Alanine	Ala	C00041	Cation
3	Arginine biosynthesis	I	L-Arginine	Arg	C00062	Cation
3	Asparagine biosynthesis	I	L-Asparagine	Asn	C00152	Cation
3	Aspartic acid biosynthesis	I	L-Aspartic acid	Asp	C00049	Cation
3	Chorismic acid biosynthesis	I	5-Enolpyruvylshikimate-3-phosphate	5EPS3P	C01269	Anion
3	Common antigen biosynthesis	I	dTDP-4-acetamido-4,6-dideoxy-D-galactose	dTDP-AADDGal	-	Nucleotide
3	Glutamine biosynthesis	I	L-Glutamine	Gln	C00064	Cation
3	Glutathione biosynthesis	I	Glutathione (oxidized form)	GSSG	C00127	Cation
3	Glycine biosynthesis	I	Glycine	Gly	C00037	Cation
3	Histidine biosynthesis	I	Phosphoribosyl-AMP	PR-AMP	C02741	Nucleotide
3	Histidine biosynthesis	I	Phosphoribosylformimino-AICAR-phosphate	PRFAP	C04896	Nucleotide
3	Histidine biosynthesis	I	D-Erythroimidazoleglycerolphosphate	EIGP	C04666	Nucleotide
3	Histidine biosynthesis	I	Imidazoleacetolphosphate	IAP	C01267	Nucleotide
3	Histidine biosynthesis	I	L-Histidinol phosphosphate	Histidinol-P	C01100	Nucleotide
3	Histidine biosynthesis	I	L-Histidinol	Histidinol	C00860	Cation
3	Isopentenyldiphosphate biosynthesis	I	2-C-Methyl-D-erythritol-2,4-cyclodiphosphate	MeEry2, 4CP	C11453	Anion
3	Methylglyoxal pathway	I	S-Lactoylglutathione	SL-GSH	C03451	Cation
3	Peptideglycan biosynthesis	I	D-Alanyl-D-alanine	AlaAla	C00993	Cation
3	Peptideglycan biosynthesis	I	L-Alanyl-D-glutamyl-meso-A2pm	AlaGluA2pm	—	Cation
3	Phenylalanine biosynthesis	I	L-Phenylalanine	Phe	C00079	Cation
3	Polyamine biosynthesis	I	5′-Methylthioadeosine	MTA	C00170	Cation
3	Polyamine biosynthesis	I	Spermidine	—	C00315	Cation
3	Purine biosynthesis	I	Aminoimidazole carboxamide ribonucleotide	AICAR	C04677	Nucleotide
3	Pyrimidine biosynthesis	I	Uridine-5′-phosphate	UMP	C00105	Nucleotide
3	Pyrimidine biosynthesis	I	Cytidine-5′-triphosphate	CTP	C00063	Nucleotide
3	Salvage of pyridoxal-7′-phosphate	I	Pyridoxamine phosphate	PX5′P	C00647	Anion
3	Salvage of pyrimidine	I	Cytidine-5′-phosphate	CMP	C05822	Nucleotide
3	SAM biosynthesis	I	S-Adenosyl-L-methionine	SAM	C00019	Cation
3	Tryptophan biosynthesis	I	L-Tryptophan	Trp	C00078	Cation
3	Tryrosine biosynthesis	I	L-Tyrosine	Tyr	C00082	Cation
3	Ubiquinone biosynthesis	I	p-Hydroxybenzoic acid	PHBA	C00156	Antion
3	Arginine degradation	II	N ^2-Succinyl-L-arginine	N2SucArg	C03296	Cation
4	AppppA biosynthesis	I	P(1), P(4)-Bis(5′-adenosyl)tetraphosphate	AppppA	C01260	Nucleotide
4	Arginine biosynthesis	I	N-α-Acetylornithine	N-AcOrn	C00437	Cation
4	Arginine biosynthesis	I	L-Ornithine	Ornthine	C00077	Cation
4	Biotin biosynthesis	I	Dethiobiotin	—	C01909	Cation
4	Fravin biosynthesis	I	Flavinmononucleotide (oxidized form)	FMN	C00061	Nucleotide
4	Fravin biosynthesis	I	Flavinadeninedinucleotide (oxidized form)	FAD	C00016	Nucleotide
4	Glycolysis (Embden-Meyerhof-Parnas)	I	3-Phosphoglyceric acid	3-PG	C00597	Anion
4	Glycolysis (Embden-Meyerhof-Parnas)	I	Phosphoenolpyruvic acid	PEP	C00074	Anion
4	Homoserine biosynthesis	I	L-Homoserine	HomoSer	C00263	Cation
4	Lactic acid fermentation	I	D-Lactic acid	Lactic acid	C01432	Anion
4	Lysine biosynthesis	I	N-Succinyl-2-amino-6-ketopimelic acid	SAKPA	C04462	Anion
4	Lysine biosynthesis	I	N-Succinyl-L,L-2,6-diaminopimelic acid	SDPA	C04421	Anion
4	Methionine biosynthesis	I	Cystathionine	—	C00542	Cation
4	NAD biosynthesis	I	Nicotinamideadeninedinucleotide (oxidized form)	NADH	C00004	Nucleotide
4	Peptideglycan biosynthesis	I	UDP-N-acetylmuramic acid	UDP-MurNaAcA	C01050	Nucleotide
4	Purine biosynthesis	I	5-Phosphoribosyl-N-formylglycineamidine	FGAM	C04640	Nucleotide
4	Pyrimidine biosynthesis	I	Uridine-5′-triphosphate	UTP	C00075	Nucleotide
4	Tetrahydrofolic acid biosynthesis	I	2-Amino-4-hydroxy-6-hydroxymethyl-7,8-dihydropteridinediphosphate	AHHMeDHPDP	C04807	Nucleotide
4	Arginine degradation	II	N ^2-Succinylornithine	N2SucOrn	C03415	Cation
4	S-Adenosylhomocysteine catabolism	II	S-D-Ribosyl-L-homocysteine	RibHomoCys	C03539	Cation
4	Threonine degradation	II	Propionyl-CoA	—	C00100	Nucleotide
5	CoA biosynthesis	I	Dephospho-CoA	dePCoA	C00882	Nucleotide
5	CoA biosynthesis	I	Coenzyme A	CoA	C00010	Nucleotide
5	Glutathione biosynthesis	I	Glutathione (reduced form)	GSH	C00051	Cation
5	Glutathionylspermidine metabolism	I	Glutathionylspermidine	GSH-spermidine	C05730	Cation
5	Glycolysis (Embden-Meyerhof-Parnas)	I	Pyruvic acid	—	C00022	Anion
5	Histidine biosynthesis	I	L-Histidine	His	C00135	Cation
5	Isoleucine biosynthesis	I	L-Isoleucine	Ile	C00407	Cation
5	KDO biosynthesis	I	D-Arabinose-5-phosphate	Ara5P	C01112	Anion
5	Leucine biosynthesis	I	2-Isopropylmalic acid	2IPMA	C02504	Anion
5	Pantothenic acid biosynthesis	I	2-Dehydropantoic acid	2DHPA	C00966	Anion
5	Pantothenic acid biosynthesis	I	Pantothenic acid	—	C00864	Anion
5	Pentose phosphate pathway (oxidative)	I	D-Ribulose-5-phosphate	Ru5P	C00199	Anion
5	Polyamine acetylation	I	N ^1-Acetylspermidine	N1AcSpermidine	C00612	Cation
5	Proto-and siroheme biosynthesis	I	Protoheme IX	—	C00032	Anion
5	Pyrimidine biosynthesis	I	Uridine-5′-diphosphate	UDP	C00015	Nucleotide
5	Pyrimidine biosynthesis	I	Cytidine-5′-diphosphate	CDP	C00112	Nucleotide
5	Pyrimidine biosynthesis	I	Deoxythymidine-5′-diphosphate	dTDP	C00363	Nucleotide
5	Salvage of Ade, HypXan	I	Hypoxanthine	—	C00262	Cation
5	Salvage of Ade, HypXan	I	Inosine	—	C00294	Cation
5	Salvage of Ade, HypXan	I	Xanthine	—	C00385	Cation
5	Salvage of Gua, Xan	I	Guanine	—	C00242	Cation
5	Salvage of Gua, Xan	I	Guanosine	—	C00387	Cation
5	Salvage of pyrimidine	I	Cytosine	—	C00380	Cation
5	Salvage of pyrimidine	I	Thymine	—	C00178	Cation
5	Threonine biosynthesis	I	L-Threonine	Thr	C00188	Cation
5	UDP-N-acetylglucosamine biosynthesis	I	N-Acetylglucosamine-1-phosphate	GlcNAc-P	C04256	Anion
5	Histidine degradation	II	Imidazole lactic acid	Im-lactate	C05568	Cation
5	Threonine degradation	II	Aminoacetone	—	C01888	Cation

---

Fig. 6 Clustering analysis of metabolites for which intracellular levels were changed during histidine starvation. A. Tree view of distances between the clusters generated by HCA. B. Average view of the clusters. Signal intensities relative to that of internal standard were standardised.

During stringent responses, DNA replication, transcription, translation, and cell division are arrested in a ppGpp-dependent manner.¹¹ Actually, growth was seen to be arrested by monitoring optical density at 600 nm at the onset of histidine starvation (data not shown). High-energy carrier molecules, ATP and GTP, were accumulated slightly (cluster 2), suggesting that DNA replication, and transcription of ribosomal RNAs are stopped, as described elsewhere.¹¹ This is considered to be a reservation for further growth when de novohistidine biosynthesis is activated (growth cannot occur because of hisD disruption in the cells). The building blocks of DNA, dNTP, were maintained after histidine downshift. Amino acid levels were significantly accumulated after downshift: most are distributed to clusters 2 and 3. In particular, accumulation of lysine, leucine, arginine, and phenylalanine was prominent. Of amino acids, threonine, except histidine, was assigned to cluster 5, indicating threonine was consumed after histidine-downshift. These results suggest that threonine biosynthesis from aspartic acid was shut down during starvation, resulting in the drop of isoleucine and following stimulation of valine and leucine biosynthesis.²⁴

Next, we focused on the behavior of cluster 2 metabolites which were clustered with ppGpp in order to discuss the candidate pathways under stringent control (Table 1). The first half of glycolysis, from glucose-6-phosphate (G6P) to glyceraldehyde-3-phosphate (G3P) and dihydroacetonephosphate (DHAP), was accumulated in concert with the intermediates in the pentose phosphate pathway (Fig. 5, positions B-2 and C-2). This effect seems to involve increased histidine biosynthesisviaPRPP. However, the middle of glycolysis, from 3- or 2-phosphoglyceric acid (3- or 2-PG) to phosphoenolpyruvic acid (PEP), was unchanged (Fig. 5, position C-3), while late step of glycolysis, pyruvic acid, dropped remarkably (Fig. 5, position C-3). These results suggest that the behavior of glycolysis is complicated during histidine starvation: that is, the first half is activated with the pentose phosphate pathway, but the flux is not extended to pyruvic acid synthesis. Moreover, the entire tricarboxylic acid (TCA) cycle intermediates were distributed into cluster 2, indicating that the TCA cycle was activated to yield precursors of downstream pathways and consumes intracellularpyruvic acid. Although we cannot exclude the possibility that the aforementioned pathways are regulated in a ppGpp-independent manner, the results provide new insights for understanding the stringent response in bacteria. Metabolomics observations in the relA mutant are needed to unveil the mechanism of stringent response in metabolism.

Experimental

Preparation of E. colimetabolite list

Metabolites in E. coli were selected from the EcoCyc^15,16,25 and KEGG^26,27 databases. The metabolites listed in the databases were identified one by one in the literature. Finally, 727 metabolites were selected as E. coli metabolites (supplementary Table 1† ) and then classified based on the metabolism pathways as follows. Class I metabolites belong to primary metabolisms including biosynthesis of building block metabolites of cells, energy metabolisms such as the glycolysis and TCA cycle, and other biosynthesis pathways. Class II includes the intermediates of degradation pathways of primary metabolites, e.g., amino acid degradation. The metabolites in classes I and II are expected to be found in cells grown in general minimal medium. Class III involves the degradation pathways of environmental compounds which are not included in general minimal medium. Class IV includes secondary metabolism or unconventional metabolisms. Class V includes the metabolites found in E. coli cells previously, but the pathways are unknown. Class VI metabolites are intermediates and products of in vitro enzymatic reactions using putative gene products.

Bacterial strains and growth conditions

E. coli W3110 (laboratory stock) grown in Luria-Bertani (LB) liquid medium²⁸ was used for optimisation of metabolome extraction procedure. The hisD deletion mutant, JW2002, was distributed from the Keio collection.²⁹ The cells were pre-grown on LB plates at 37 °C for 12 h. The fresh colonies that appeared were inoculated into M9 liquid medium²⁸ containing 50 µg ml⁻¹ of L-histidine, and this was followed by incubation at 37 °C with shaking. When the cells reached the middle-to-late logarithmic phase (optical density at 600 nm = 0.8), they were harvested by brief centrifugation, washed with M9 medium without histidine, and then suspended in the M9 medium. The culture was continuously shaken at 37 °C until the sampling times.

Metabolite extraction procedure

Intracellular metabolites were extracted as described previously with some modifications.⁶ Schematic of the metabolome extraction procedure is shown in Fig. 2. Cells grown as described above were collected for metabolome analysis at indicated times. Culture including approximately 10⁹ cells (calculated as optical density at 600 nm × sampling volume of culture (ml) = 20) was filtered by a vacuum filtration system using a 0.4 µm pore size filter. The residual cells on the filter were washed twice with 5 ml of Milli-Q water. The filter was immersed in 2 ml of methanol including 5 µM each of internal standards, methioninesulfone and D-camphor-10-sulfonic acid (CSA). The dish was sonicated for 30 s using an Elma Transsonic T460/H ultrasonic syringe (not an ultrasonic cell disrupter) (Elma Hans Schmidbauer GmbH & Co., Singen, Germany) to suspend the cells completely. Effects of this process on metabolite extraction efficiency are discussed in the results and discussion section. A 1.6 ml portion of the methanol cell suspension was transferred to a Falcon Blue Max Jr., 352097 centrifugal tube (15 ml) (Becton Dickinson & Co., New Jersey, USA), and mixed with 1.6 ml of chloroform and 640 µl of Milli-Q water. After vortexing well, the mixture was centrifuged at 4600 × g and 4 °C for 5 min. The aqueous layer (750 µl) was distributed to three Amicon Ultrafree-MC ultrafilter tips (Millipore Co., Massachusetts, USA) and centrifuged at 9100 × g and 4 °C for approximately 2 h. The filtrate was dried and preserved at −80 °C until CE-MS analysis. Prior to analysis, the sample was dissolved in 25 µl of Milli-Q water.

Instrumentation

CE-TOFMS was carried out using an Agilent CE Capillary Electrophoresis System equipped with an Agilent 6210 Time-of-Flight mass spectrometer, Agilent 1100 isocratic HPLC pump, Agilent G1603A CE-MS adapter kit, and Agilent G1607A CE-ESI-MS sprayer kit (Agilent Technologies, Waldbronn, Germany). The system was controlled by Agilent G2201AA ChemStation software version B.03.01 for CE (Agilent Technologies, Waldbronn, Germany). Data acquisition was performed by Analyst QS Build: 7222 software for Agilent TOF (Applied Biosystems, California, USA/MDS Sciex, Ontario, Canada).

CE-TOFMS conditions

Separations and detections of metabolites were basically performed as described previously for cationic metabolites,^3,6 anionic metabolites,^4,6 and nucleotides .³⁰ For cationic metabolites, capillary electrophoreses were performed using a fused silica capillary. The electrolyte was 1 M formic acid. Methanol-water (50% v/v) containing 0.5 µM reserpine (the lock mass for exact mass measurements) was delivered as the sheath liquid at 10 µl min⁻¹. For anionic metabolites, a polymer coated SMILE(+) capillary (Nacalai tesque, Kyoto, Japan) was used. The electrolyte was 50 mM ammonium acetate (pH 8.5). Ammonium acetate (5 mM) in 50% (v/v) methanol-water containing 1 µM reserpine was delivered as the sheath liquid at 10 µl min⁻¹. For nucleotides , separations were performed using a fused silica capillary. The electrolyte was 50 mM ammonium acetate (pH 7.5). The sheath liquid for anionic metabolites was used as the electrolyte. The capillary was pretreated with preconditioning buffer including 25 mM ammonium acetate and 75 mM sodium phosphate at pH 7.5. Pressure of 50 mbar was applied to inlet capillary during run to reduce the analysis time. For all analytical modes, inner diameter and total length of capillary are 50 µm and 100 cm, respectively. The applied voltage was set at +30 kV and −30 kV for cation and anion modes and nucleotide mode, respectively.

Electrosplay ionisation-TOFMS was operated in the positive ion mode (4 kV), the negative ion mode (3.5 kV), and the negative ion mode (3.5 kV) for cationic metabolites, anionic metabolites, and nucleotides , respectively. A flow rate of heated dry nitrogen gas (heater temperature 300 °C) was maintained at 10 psig. Exact mass data were acquired over a 50–1000 m/z range.

Cluster analysis

The metabolome data set was prepared by selection of metabolites based on a coefficient of variation of more than 0.2 to exclude the metabolites for which levels were unchanged. HCA was performed on the data set using MATLAB 2007a (The Math Works, Massachusetts, USA). Distances between the metabolites were calculated by the equation, correlation coefficient + 1.

Conclusion

In this study, we produced a list of previously reported E. coli metabolites together with analytical information for CE-TOFMS-based metabolome analysis. This list is also useful for other analytical methods using MS as a detector. The metabolome extraction method was also improved, realising quantitative analyses of phosphate-rich metabolites and other hydrophilic, charged metabolites. However, this method cannot be applied to the analysis of metabolites oxidised in the metabolome pool. Extraction methods inhibiting oxidation of the metabolites are needed to determine them. We successfully determined 198 metabolites in primary metabolisms of E. colihistidine-auxotroph. The metabolome profiles were analysed by mapping data to metabolic pathways and HCA, suggesting separated activation of amino acid biosynthesis pathways, glycolysis, and the TCA cycle. Although more detailed studies using the relA mutant is required to unveil the entire view of stringent responses mediated by ppGpp, we believe that metabolome profiling should provide new insights into bacterial adaptation to environmental changes.

Acknowledgements

We are grateful to Hirotada Mori, Tomoya Baba, and Kenji Nakahigashi for providing the E. coli mutant strain and to Masatomo Hirabayashi and Yuji Sakakibara for providing excellent technical expertise in CE-TOFMS analysis. We also thank Mineo Morohashi for helpful discussions.

References

1. I. Nobeli, H. Ponstingl, E. B. Krissinel and J. M. Thornton, J. Mol. Biol., 2003, 334, 697–719.
2. M. R. Monton and T. Soga, J. Chromatogr., A, 2007, 1168, 237–246.
3. T. Soga and D. N. Heiger, Anal. Chem., 2000, 72, 1236–1241.
4. T. Soga, Y. Ueno, H. Naraoka, Y. Ohashi, M. Tomita and T. Nishioka, Anal. Chem., 2002, 74, 2233–2239.
5. T. Soga, Y. Ueno, H. Naraoka, K. Matsuda, M. Tomita and T. Nishioka, Anal. Chem., 2002, 74, 6224–6229.
6. T. Soga, Y. Ohashi, Y. Ueno, H. Naraoka, M. Tomita and T. Nishioka, J. Proteome Res., 2003, 2, 488–494.
7. T. Soga, R. Baran, M. Suematsu, Y. Ueno, S. Ikeda, T. Sakurakawa, Y. Kakazu, T. Ishikawa, M. Robert, T. Nishioka and M. Tomita, J. Biol. Chem., 2006, 281, 16768–16776.
8. S. G. Villas-Bôas, J. Højer-Pedersen, M. Åkesson, J. Smedsgaard and J. Nielsen, Yeast, 2005, 22, 1155–1169.
9. R. P. Maharjan and T. Ferenci, Anal. Biochem., 2003, 313, 145–154.
10. J. D. Rabinowitz and E. Kimball, Anal. Chem., 2007, 79, 6167–6173.
11. M. Cashel, D. R. Gentry, V. J. Hernandez and D. Vinella, in Escherichia coli and Salmonella: cellular and molecular biology, American Society for Microbiology Press, Washington DC, USA, 1992, 2nd edn.
12. C. Eymann, G. Homuth, C. Scharf and M. Hecker, J. Bacteriol., 2002, 184, 2500–2520.
13. O. Brockmann-Gretza and J. Kalinowski, BMC Genomics, 2006, 7, 230.
14. A. Hesketh, J. Chen, J. Ryding, S. Chang and M. Bibb, Genome Biol., 2007, 8, R161.
15. I. M. Keseler, J. Collado-Vides, S. Gama-Castro, J. Ingraham, S. Paley, I. T. Paulsen, M. Peralta-Gil and P. D. Karp, Nucleic Acids Res., 2005, 33, D334–337.
16. http://ecocyc.org/.
17. A. Nagai, E. Ward, J. Beck, S. Tada, J. Y. Chang, A. Scheidegger and J. Ryals, Proc. Natl. Acad. Sci. U. S. A., 1991, 88, 4133–4137.
18. J. C. Loper and E. Adams, J. Biol. Chem., 1965, 240, 257–268.
19. H. S. Moyed and M. Friedman, Science, 1959, 129, 968–969.
20. M. E. Winkler, D. J. Roth and P. E. Hartman, J. Bacteriol., 1978, 133, 830–843.
21. J. C. Stephens, S. W. Artz and B. N. Ames, Proc. Natl. Acad. Sci. U. S. A., 1975, 72, 4389–4393.
22. N. Ishii, K. Nakahigashi, T. Baba, M. Robert, T. Soga, A. Kanai, T. Hirasawa, M. Naba, K. Hirai, A. Hoque, P. Y. Ho, Y. Kakazu, K. Sugawara, S. Igarashi, S. Harada, T. Masuda, N. Sugiyama, T. Togashi, M. Hasegawa, Y. Takai, K. Yugi, K. Arakawa, N. Iwata, Y. Toya, Y. Nakayama, T. Nishioka, K. Shimizu, H. Mori and M. Tomita, Science, 2007, 316, 593–597.
23. J. Quackenbush, Nat. Rev. Genet., 2001, 2, 418–427.
24. H. E. Umbarger, in Escherichia coli and Salmonella: cellular and molecular biology, American Society for Microbiology Press, Washington DC, USA, 1992, 2nd edn.
25. C. A. Ouzounis and P. D. Karp, Genome Res., 2000, 10, 568–576.
26. H. Ogata, S. Goto, K. Sato, W. Fujibuchi, H. Bono and M. Kanehisa, Nucleic Acids Res., 1999, 27, 29–31.
27. http://www.genome.jp/kegg/.
28. J. Sambrook, E. F. Fritsch and T. Maniatis, Molecular cloning: a laboratory manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. USA, 1989, 2nd edn.
29. T. Baba, T. Ara, M. Hasegawa, Y. Takai, Y. Okumura, M. Baba, K. A. Datsenko, M. Tomita, B. L. Wanner and H. Mori, Mol. Syst. Biol., 2006, 2 2006.0008.
30. T. Soga, T. Ishikawa, S. Igarashi, K. Sugawara, Y. Kakazu and M. Tomita, J. Chromatogr., A, 2007, 1159, 125–133.
31. A. Kheirolomoom, J. Mano, A. Nagai, A. Ogawa, G. Iwasaki and D. Ohta, Arch. Biochem. Biophys., 1994, 312, 493–500.

---

Footnote

† Electronic supplementary information (ESI) available: Supplementary tables and high quality figures. See DOI: 10.1039/b714176a

---