Analytical Biochemistry 475 (2015) 4–13

Contents lists available at ScienceDirect

Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio

Alkaline conditions in hydrophilic interaction liquid chromatography for intracellular metabolite quantification using tandem mass spectrometry Attila Teleki 1, Andrés Sánchez-Kopper 1, Ralf Takors ⇑ Institute of Biochemical Engineering, University of Stuttgart, 70569 Stuttgart, Germany

a r t i c l e

i n f o

Article history: Received 17 September 2014 Received in revised form 8 December 2014 Accepted 5 January 2015 Available online 16 January 2015 Keywords: Metabolic profiling HILIC LC–MS/MS

a b s t r a c t Modeling of metabolic networks as part of systems metabolic engineering requires reliable quantitative experimental data of intracellular concentrations. The hydrophilic interaction liquid chromatography–e lectrospray ionization–tandem mass spectrometry (HILIC–ESI–MS/MS) method was used for quantitative profiling of more than 50 hydrophilic key metabolites of cellular metabolism. Without prior derivatization, sugar phosphates, organic acids, nucleotides, and amino acids were measured under alkaline and acidic mobile phase conditions with pre-optimized multiple reaction monitoring (MRM) transitions. Irrespective of the polarity mode of the acquisition method used, alkaline conditions achieved the best quantification limits and linear dynamic ranges. Fully 90% of the analyzed metabolites presented detection limits better than 0.5 pmol (on column), and 70% presented 1.5-fold higher signal intensities under alkaline mobile phase conditions. The quality of the method was further demonstrated by absolute quantification of selected metabolites in intracellular extracts of Escherichia coli. In addition, quantification bias caused by matrix effects was investigated by comparison of calibration strategies: standard-based external calibration, isotope dilution, and standard addition with internal standards. Here, we recommend the use of alkaline mobile phase with polymer-based zwitterionic hydrophilic interaction chromatography (ZIC–pHILIC) as the most sensitive scenario for absolute quantification for a broad range of metabolites. Ó 2015 Elsevier Inc. All rights reserved.

Metabolomics addresses the identification and quantification of small molecule metabolites, in essence the reactants of biological systems, and enables the comprehensive analysis of complex biochemical networks systemic studies. Whereas the genome and proteome represent upstream biochemical events, the metabolome reveals underlying regulatory mechanisms and therefore reflects, most closely, the actual cellular physiological state [1–3]. As such, metabolome analysis can be subdivided into targeted and nontargeted approaches. Nontargeted high-throughput applications such as metabolic fingerprinting aim at the rapid and qualitative or semiquantitative analysis of whole-cell metabolic patterns, reducing the analytical effort considerably [4–6]. On the contrary, targeted applications such as quantitative metabolic profiling are focused on the nonbiased absolute quantification of a subset of metabolites of predefined metabolic pathways or classes of compounds [2,4,6]. In this context, monitoring of intracellular ⇑ Corresponding author. Fax: +49 0711 685 65164. 1

E-mail address: [email protected] (R. Takors). Contributed equally to this manuscript.

http://dx.doi.org/10.1016/j.ab.2015.01.002 0003-2697/Ó 2015 Elsevier Inc. All rights reserved.

metabolism dynamics as a result of rapid changes in the extracellular environment within stimulus–response experiments can be used to investigate in vivo enzyme kinetics offering insights into underlying regulatory mechanisms [7–10]. Considering that the enormous number of intracellular metabolites within various biological matrices and concentrations range over several orders of magnitude (pM–mM), a variety of different analytical methods have been investigated [2,6]. Used analytical methods vary widely, from enzymatic assays [11] for targeted measurements to modern and more powerful soft ionization mass spectrometry (SIMS)2 coupled to liquid chromatography (LC) [12–19], gas chromatography

2 Abbreviations used: SIMS, soft ionization mass spectrometry; LC, liquid chromatography; GC, gas chromatography; CE, capillary electrophoresis; NMR, nuclear magnetic resonance; LC–MS, liquid chromatography combined with mass spectrometry; HILIC, hydrophilic interaction liquid chromatography; LC–ESI–MS/MS, liquid chromatography–electrospray ionization–tandem mass spectrometry; MRM, multiple reaction monitoring; ZIC–pHILIC, polymer-based zwitterionic hydrophilic interaction chromatography; LB, Luria–Bertani; MDL, method detection limit; ANOVA, analysis of variance; RPLC, reversed phase liquid chromatography; WWR, wrong-way-round.

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

(GC) [20–25], or capillary electrophoresis (CE) [26–28] as well as nuclear magnetic resonance (NMR) [29,30]. For comprehensive metabolite quantification studies, a combination of multiple analytical methods is usually indispensable, resulting in data sets with varying precision and accuracy [1,2,6,31]. However, for dynamic modeling of cellular metabolism, in connection with stimulus–response studies, reliable quantifications of absolute intracellular concentrations in defined cellular states and time points are of crucial importance [7,10,25]. Compared with GC, CE, and NMR, liquid chromatography combined with mass spectrometry (LC–MS) offers inherent advantages; LC–MS does not require further derivatization like GC, is able to separate polar and nonpolar compounds unlike CE with selectivity depending on column material, and offers a higher sensibility compared with NMR. Consequently, LC–MS approaches are becoming highly accepted as the most universal analytical platform for reliable quantification of intracellular metabolites [2,3,25]. Recently, hydrophilic interaction liquid chromatography (HILIC) has become increasingly popular because it enables the separation of charged and polar analytes and exhibits excellent compatibility with MS detection [32–34]. Related studies focused mainly on silica-based stationary HILIC phases, modified by various functional groups, enabling comprehensive quantitative analysis of metabolites under acidic and neutral mobile phase conditions [12,13,15,16]. Motivated by these encouraging studies, we developed the approach further for exploiting its potential, concentrating on so far nonaddressed modes of operation. In this study, we present an analytical platform based on two liquid chromatography–electrospray ionization–tandem mass spectrometry (LC–ESI–MS/MS) methods for quantitative profiling of more than 50 key metabolites of the cellular metabolism, comprising amino acids, sugar phosphates, organic acids, nucleotides, and coenzymes. Metabolites were detected by MS/MS on a triple quadrupole instrument in the multiple reaction monitoring (MRM) mode with high selectivity based on pre-optimized MRM transitions. Chromatographic separation was performed by polymer-based zwitterionic hydrophilic interaction chromatography (ZIC–pHILIC), and the mobile phase pH effect with respect to analyte sensibility was evaluated. Special focus placed on method quantification limits, linear dynamic range, and repeatability for a representative cross section of 56 common intracellular metabolites in acidic and alkaline mobile phase conditions. Finally, the applicability of the method was demonstrated by absolute quantification of selected metabolites in intracellular extracts of Escherichia coli showing consistency with previously published endogenous steady-state concentrations. In addition, occurrence of quantification bias caused by matrix effects is presented, comparing quantitative results obtained by standards-based external calibration, isotope dilution, and standard addition as calibration strategies. Materials and methods Chemicals Metabolite standards, reagents, and U-13C-labeled lyophilized algal cells were supplied by Sigma–Aldrich (Taufkirchen, Germany). MS-grade water was purchased from VWR (Darmstadt, Germany). MS-grade acetonitrile was purchased from Carl Roth (Essen, Germany). Standard stock solutions were prepared in LC–MS water and stored at 70 °C.

5

(NH4)2HPO4, 20.3 mM (NH4)2SO4, 6.2 mM Na2SO4, 1.0 mM MgSO4, 0.1 mM CaCl2, and 0.01 mM thiamine hydrochloride. Overnight precultures (12 h, 37 °C, 120 rpm) were inoculated from a cryoculture (50% glycerol and Luria–Bertani [LB] medium) and were grown in 100-ml baffled shake flasks with 20 ml of minimal medium (12 h, 37 °C, 120 rpm). Main cultures (37 °C, 150 rpm) were inoculated in 1:100 dilution and grown in triplicates in 500-ml baffled shake flasks with 60 ml of minimal medium (10 h, 37 °C, 150 rpm). Sampling, quenching, and metabolite extraction Cells were sampled during the exponential growth phase at a biomass concentration of approximately 2 g L 1. Amounts of approximately 4 mg biomass were sampled by fast centrifugation (20,000g, 20 s) and washed with 2 ml of isotonic 0.9% (w/v) sodium chloride solution (20,000g, 20 s). Biomasses were quenched by liquid nitrogen ( 196 °C) and temporarily stored at 70 °C. Subsequently, a defined amount of water (MS grade) was added to obtain an extraction concentration of 15 g L 1. Resulting suspensions were immediately pre-incubated for 1 min at 100 °C in a water bath for enzymatic inactivation and resuspended by short-time vortexing. Subsequently, samples were incubated for 5 min at 100 °C in a water bath and afterward chilled on ice water. Metabolite extracts were separated from cell debris by centrifugation (20,000g, 10 min) and stored at 70 °C [35]. Preparation of U-13C-labeled internal standard Commercially available U-13C-labeled lyophilized algal cells (>99 atom% 13C, lot no. 487945, Sigma–Aldrich) were weighed into 2-ml reaction vessels in small amounts. Preheated water (100 °C) was added aiming for an extraction concentration of 90 g L 1. The resulting suspensions were incubated at 100 °C for 2 min in a water bath and resuspended by short-time vortexing. The procedure was repeated two more times, and the resulting samples were chilled on ice water. 13C-labeled metabolite extracts were separated from algal cells by centrifugation (20,000g, 10 min) and stored at 70 °C [25]. Optimization of chromatographic and source conditions Development of chromatography was performed on a Sequant ZIC–pHILIC column (150  2.1 mm, 5 lm) with guard column (Sequant ZIC–pHILIC, 20  2.1 mm, 5 lm). The optimization of chromatographic conditions was based on a selection of highly polar intracellular metabolites within the central metabolism (6-phophogluconate, fructose 1,6-bisphosphate, glucose 6-phosphate, phosphoenolpyruvate, malate, succinate, and citrate). Bicratic chromatographic runs were evaluated in regard to selectivity and sensitivity of applied standard mixtures. Optimization was focused on pH of mobile phases (5.0, 7.0, and 9.0), buffer concentration (2–20 mM ammonium acetate), flow rate (0.05–0.20 ml/min), column temperature (20–60 °C), and gradient slope of polar eluent B (2,5–5% B/min) within a bicratic elution mode. In addition, source parameters (ESI) were optimized with respect to maximal signal intensities, focusing on the nitrogen gas flow (7–13 L/min), nebulizer pressure (15–60 psi), and capillary voltage (±3000–4000 V). An overview of optimized chromatographic and source parameters is shown further below.

Strains and growth condition Sample processing and optimized chromatographic conditions E. coli K-12 MG1655 (purchased from DSMZ, Leipzig, Germany) was grown in minimal medium with 0.75% (w/v) D-glucose as sole carbon source. The minimal medium contained the following components: 63.1 mM NaH2PO4, 26.9 mM K2HPO4, 10.0 mM

Standards and endogenous cellular extracts were analyzed on an Agilent 1200 high-performance liquid chromatography (HPLC) system (Agilent, Waldbronn, Germany) consisting of a degasser, a

6

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

binary pump, a thermostated column compartment, and a bio-inert thermostated autosampler (Agilent 1260 Infinity series) maintained at 5 °C. Samples and standards were prepared in 60% (v/v) acetonitrile and 10 mM ammonium acetate buffer (adjusted to pH 5.6 or 9.2). An injection of 5 ll was separated on a Sequant ZIC–pHILIC column (150  2.1 mm, 5 lm) with guard column (Sequant ZIC–pHILIC, 20  2.1 mm, 5 lm) kept at 40 °C and with a flow rate of 0.2 ml/min. Mobile phases were composed of 10% (v/v) aqueous buffer solution (10 mM ammonium acetate) and 90% (v/v) acetonitrile for eluent A and 90% (v/v) aqueous buffer solution and 10% (v/v) acetonitrile for eluent B, both adjusted to pH 5.6 (with acetic acid) or pH 9.2 (with 25% [v/v] ammonium hydroxide). Gradient elution was carried out by the following program: isocratic hold 0% B for 1 min, next a linear gradient from 0% B to 75% B for 30 min, followed by a linear gradient from 75% B to 100% B for 4 min. At the end of the run, the column was washed with 100% B for 5 min and afterward equilibrated to starting conditions by a linear gradient from 100% B to 0% B for 10 min and 0% B for 15 min. Instrumentation and ESI–MS/MS Data were acquired with an Agilent 6410B Triple Quad mass spectrometer (Agilent) with ESI ion source. Negative (ESI ) and positive (ESI+) ionization modes were performed in separated runs in MRM mode depending on metabolite group. Source conditions were as follows: nitrogen gas flow rate of 10 L min 1 with a temperature of 350 °C, a nebulizer pressure of 30 psi, and a capillary voltage of 4000 V. Fragmentor voltages and collision energies were optimized in continuous direct injection mode for each metabolite precursor-to-product ion transition. Dwell time was adjusted to 100 ms for each MRM transition. System control, acquisition, and analysis of data were performed by use of commercial MassHunter B.04.00 software. Linearity range and sensibility tests Linearity ranges of both method configurations (acidic and alkaline mobile phase conditions) were evaluated with regard to the amount of respective injected analytes. Multicomponent standard solutions were prepared from properly adjusted metabolite mixtures and were assigned to three different groups depending on ionization mode and retention time. Focusing on small concentrations (65 lM) and depending on sensitivities of measured metabolite groups, a calibration range of 1 nM to 5 lM based on 12 levels was applied. Additional linearities of several selected metabolites were evaluated by use of an extended calibration range (0.5– 200 lM) based on 12 levels. Sensibility tests were implemented by measuring 5-lM standards for each metabolite under acidic and alkaline mobile phase conditions. Absolute quantification in bacterial metabolite extracts (quantification strategies) The suitability of the method was demonstrated by quantification of absolute concentrations of selected metabolites in intracellular extracts of wild-type E. coli biomasses. In addition, quantification bias caused by matrix effects was evaluated by comparison of quantitative results by external calibration, isotope dilution, and standard addition. Multicomponent standard solutions for external calibration of defined nonlabeled metabolite mixtures were obtained. Depending on previously estimated concentrations (obtained by spiking experiments), metabolite-specific calibration ranges with seven levels were adjusted. Quantifications based on isotope dilution were performed by constant addition of U-13C-labeled algal extracts as internal standard to samples and

external calibration standards. External calibration curves were prepared by linear regression of ratios of analyte peak areas and internal standard areas against the concentration levels of the respective analyte. Standard addition was performed by addition of defined amounts of respective standard directly to sample aliquots. Applied multicomponent metabolite mixtures were adjusted based on previously estimated concentrations of selected metabolites. Internal calibration curves were prepared by linear regression of nonspiked and incrementally increased analyte peak areas against concentration levels of the respective added standards. Results The objective of the current work was to develop a robust LC–MS/MS method for quantitative analysis of a broad range of analytes with high selectivity and sensitivity in complex matrix samples such as intracellular metabolite extracts. To overcome matrix effects and coelution of isomeric compounds, frequently present in intracellular matrices, a suitable chromatographic method is required. Recent developments in polymer-based HILIC stationary phases enable chromatographic separation of complex samples under alkaline conditions, improving sensitivity and chromatographic performance. Optimization of MS/MS parameters Measurements were performed on an HPLC–ESI–QQQ (triple quadrupole) instrument in MRM mode with a metal-free bio-inert HPLC system. Precursor-to-product ion transitions were optimized using continuous direct injection of pure standard solutions in positive and negative ionization modes. Suitable product ions, fragmentor voltages, and collision energies were selected in regard to maximal signal intensities. Specific MRM transitions for a representative cross section of 56 common intracellular metabolites, including amino acids, carboxylic acids, sugar phosphates, nucleotides, and coenzymes as well as their U-13C-labeled analogues for isotope dilution, are provided in the online Supplementary material. Chromatographic performance under alkaline and acidic mobile phases Chromatographic performance was evaluated on retention time reproducibility, peak width, and peak shape. To avoid potential column effects due to pH switches, two different columns from the same sorbent lot were used for acidic and alkaline conditions. Even when similar chromatographic quality was obtained for organic acids and amino acids, the peak shape was significantly improved for nucleotides under basic conditions. Only citrate and isocitrate presented better resolution in higher concentration levels under acidic conditions. A selection of extracted ion chromatograms of isobaric metabolites is shown in Fig. 1. Retention time reproducibility was fairly high: Analyzing individual series, 98% of the 5-lM standards showed absolute standard deviations less than 0.03 min under alkaline conditions, whereas 88% of the metabolites fulfill the likewise criterion under acidic conditions. The highest variability was observed for cytidine triphosphate (CTP, 0.17 min) and guanosine triphosphate (GTP, 0.12 min) under acidic conditions. The comparison of multiple measurement series confirmed this finding, with 98% of the metabolites showing absolute time shifts less than 0.04 min under alkaline conditions and 80% under acidic conditions. An overview of analyte retention times is given in Table 1. A compilation of MRM chromatograms of all 56 metabolites measured under acidic and alkaline mobile phase conditions is included in the Supplementary material.

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

7

Fig.1. Extracted ion chromatograms obtained with optimized HILIC method under alkaline (pH 9.2) and acidic (pH 5.6) mobile phase conditions focusing on isobaric compounds frequently present in intracellular matrices. Citrate/Isocitrate mixtures of 5 and 50 lM were injected at pHs 9.2 and 5.6, respectively.

Evaluation of linearity and method detection limits for different mobile phase pHs The linearity range for each metabolite was determined under acidic (pH 5.6) and alkaline (pH 9.2) mobile phase conditions. A range of metabolite concentrations was considered linear when the squared correlation coefficient (R2) was better than 0.98 for the average of three calibration curves. To increase the predictive significance of the measurements, internal standards were considered for monitoring of instrumental fluctuations but not for normalization of obtained peak areas. Extending the calibration range from 0.5 to 200 lM, the linearity of instrument response with respect to the amount of the injected analyte was revealed to be specific for each metabolite, sample context, and mobile phase condition. Fig. 2 shows two different metabolite standards with exemplary deviations from linearity of applied calibration lines. Obtained metabolic-specific linearity ranges differ significantly at lower concentrations (1 nM to 5 lM) between both conditions. The median for the lower linearity limits was approximately 30 nM for alkaline and 120 nM for acidic mobile phase conditions. Within alkaline conditions, 40% of the metabolites show a linearity range between 5 and 20 nM as lower boundary and 5 lM as upper limit. In contrast to this, only 12% of the metabolites fulfilled the same criterion under acidic conditions. An overview of metabolic-specific linearity ranges for acidic and alkaline mobile phase conditions is shown in Table 1. The method detection limit (MDL) was calculated as the amount of metabolite that results at statistically significant peak areas above background noise. MDL was based on the standard deviation of triplicates at the lower linearity boundary (lowest concentration of the linearity range) multiplied by the expansion coefficient (t = 6.96) for defining the 99% confidence level [36]. In column MDLs were 0.01 to 3.7 pmol for alkaline conditions and 0.02 to 14.7 pmol for acidic conditions. Fully 89% of the analyzed metabolites showed MDLs lower than 100 nM (0.5 pmol on column) under alkaline conditions, whereas just 46% of the metabolites fulfilled this criterion under acidic conditions (Fig. 3). The lowest detection limit was achieved for malic acid in positive ionization mode (2.23 nM, 0.011 pmol on column) using basic conditions. An overview of the metabolite-specific MDLs for acidic and alkaline mobile phase conditions is shown in Table 1. Comparing both conditions, the most significant differences were observed for the di- and triphosphate nucleotides and NADP as well as for

fumaric acid, isocitrate, aspartic acid, and glutamine, showing approximately 20 to 50 times lower detection limits under alkaline conditions. ESI–MS responsivity under alkaline and acidic mobile phases Next, the impact of pH of mobile phases on ESI–MS responsivity was studied based on comparison of peak areas obtained by injection of 5-lM metabolite standards under both conditions. Special care was given to the preparation of 10-mM ammonium acetate buffer solutions to afford comparable ion strength under acidic and alkaline conditions. Either acetic acid or ammonium hydroxide was added in equimolar amounts to adjust the buffer system pH to 5.6 or 9.2. To compare both scenarios, averaged peak areas obtained by alkaline conditions were normalized by those obtained by acidic conditions. Strikingly, 70% of the metabolites presented more than 1.5-fold higher signal intensities under alkaline mobile phase conditions (Fig. 4). Absolute quantification in bacterial metabolite extracts For challenging the applicability of the approach, absolute concentrations of selected metabolites were quantified in intracellular extracts of E. coli biomasses. Because matrix effects are suspected to be a major source for inaccuracies and unreliability in ESI– MS-based quantitative metabolomics [6,16,18], related impacts were evaluated as well. Quantification strategies were adjusted based on previously estimated concentrations of the targeted intracellular metabolites. Intracellular pool sizes were detected by standard-based external calibration, isotope dilution, and standard addition for subsequent comparison. Standard-based external calibration and isotope dilution were applied by metabolic-specific calibration ranges in multicomponent mixtures. Standard addition was performed for each metabolite at concentrations corresponding to approximately 2-, 3-, and 4-fold the estimated values. Results obtained from exemplary measurements of six intracellular metabolites with three different approaches are summarized in Fig. 5. Corresponding correlation coefficients are shown in Table 2. Good (R2 P 0.99) to excellent (R2 P 0.999) linearity coefficients were found for all three calibration strategies. Four replicates were made for sample analysis, where calculated relative standard deviations of the measured concentrations were less than 9% for each metabolite. As depicted in Fig. 5, individual sample analysis

Metabolite

Polarity mode

Linearity range

Method detection limit

Alkaline

Acid

Alkaline

Acid

(min)

(min)

4.2 606.0 138.0 8.5 59.6 513.3 454.1 41.6 181.9 1131.1 1713.6 67.5 67.5 53.2 79.3 2949.3 423.1 532.7 148.4 148.4 13.1 1495.0 144.1 1428.0 1130.4 198.5 2274.7 1525.2 160.8 1030.1 2004.1 123.4 1094.0 1711.8 375.8 3.6 18.2 22.0

0.10 0.17 0.59 0.01 0.05 2.20 0.11 0.02 0.15 3.72 – 0.77 0.77 0.18 0.17 1.80 0.27 0.43 0.44 0.44 0.04 0.16 0.07 0.20 0.30 0.25 0.35 0.51 0.25 0.28 0.26 0.02 0.20 0.36 0.30 0.21 0.06 0.05

0.02 3.03 0.69 0.04 0.30 2.57 2.27 0.21 0.91 5.66 8.57 0.34 0.34 0.27 0.40 14.75 2.12 2.66 0.74 0.74 0.07 7.47 0.72 7.14 5.65 0.99 11.37 7.63 0.80 5.15 10.02 0.62 5.47 8.56 1.88 0.02 0.09 0.11

8.02 ± 0.04 22.07 ± 0.02 20.59 ± 0.02 21.83 ± 0.01 21.42 ± 0.02 24.15 ± 0.01 24.83 ± 0.01 23.08 ± 0.01 21.15 ± 0.01 24.03 ± 0.01 – 21.11 ± 0.02 21.27 ± 0.01 21.63 ± 0.02 22.49 ± 0.02 24.99 ± 0.01 20.77 ± 0.01 23.61 ± 0.02 22.86 ± 0.02 23.11 ± 0.02 19.64 ± 0.01 23.15 ± 0.01 19.41 ± 0.01 21.31 ± 0.01 22.80 ± 0.02 22.07 ± 0.01 23.95 ± 0.02 25.45 ± 0.01 21.38 ± 0.01 23.21 ± 0.01 24.66 ± 0.01 20.54 ± 0.01 22.57 ± 0.01 24.11 ± 0.01 15.35 ± 0.01 15.86 ± 0.01 17.64 ± 0.01 17.51 ± 0.01

8.63 ± 0.07 22.57 ± 0.01 20.12 ± 0.01 22.14 ± 0.01 21.77 ± 0.01 23.97 ± 0.02 24.75 ± 0.02 22.76 ± 0.01 21.14 ± 0.01 24.53 ± 0.01 21.73 ± 0.11 – – 21.81 ± 0.01 22.72 ± 0.01 25.93 ± 0.07 20.78 ± 0.01 24.07 ± 0.02 23.32c ± 0.02 23.61c ± 0.02 20.03 ± 0.01 23.51 ± 0.01 19.33 ± 0.01 22.08 ± 0.03 23.75 ± 0.05 22.11 ± 0.01 24.91 ± 0.03 26.73c ± 0.12 21.34 ± 0.01 24.02 ± 0.01 25.90 ± 0.17 20.40 ± 0.01 23.33 ± 0.02 25.24 ± 0.01 15.65 ± 0.01 16.26 ± 0.01 18.02 ± 0.01 17.92 ± 0.01

33.3 14.0 4.9 28.5 41.0

286.0 38.3 22.6 99.2 167.5

0.17 0.07 0.02 0.14 0.20

1.43 0.19 0.11 0.50 0.84

15.76 ± 0.01 18.69 ± 0.01 21.45 ± 0.01 17.96 ± 0.01 16.91 ± 0.01

15.00 ± 0.01 19.08 ± 0.01 22.10 ± 0.01 18.37 ± 0.01 17.32 ± 0.01

0.9964

9.9

140.9

0.05

0.70

18.45 ± 0.01

18.90 ± 0.01

0.9968

9.1

66.8

0.05

0.33

14.37 ± 0.01

14.76 ± 0.01

5000

0.9964

26.4

49.7

0.13

0.25

14.54 ± 0.01

14.94 ± 0.01

5000

0.9967

43.1

97.6

0.22

0.49

17.14 ± 0.01

17.58 ± 0.02

20

5000

0.9965

7.9

35.9

0.04

0.18

12.74 ± 0.01

13.15 ± 0.01

0.9990

20

5000

0.9966

17.2

22.2

0.09

0.11

13.22 ± 0.01

13.64 ± 0.01

0.9991

100

5000

0.9964

15.0

100.6

0.08

0.50

18.35 ± 0.01

18.79 ± 0.01

R2

Min (nM)

Max (nM)

R2

(nM)

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (+) (+) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (+) (+) (+)

50 50 500 20 50 50 20 20 50 1000 – 20 20 50 50 500 200 50 100 100 10 20 50 100 100 100 100 500 100 50 200 100 50 100 100 20 10 20

5000 5000 5000 5000 5000 5000 5000 1000 5000 5000 – 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 2000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000

1.0000 0.9999 0.9999 0.9998 0.9974 0.9938 0.9953 0.9899 0.9884 0.9903 – 0.9973 0.9973 0.9954 0.9972 0.9998 0.9990 0.9904 0.9874 0.9874 0.9982 0.9901 0.9995 0.9928 0.9925 0.9990 0.9844 0.9915 0.9990 0.9905 0.9915 0.9991 0.9858 0.9968 0.9970 0.9996 0.9994 0.9990

50 500 500 20 100 1000 500 50 200 1000 2000 50 50 100 100 2000 500 500 500 500 20 500 100 1000 1000 100 1000 2000 100 1000 1000 100 500 2000 100 50 20 50

5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000 5000

0.9998 0.9951 0.9976 0.9998 0.9916 0.9838 0.9945 0.9854 0.9953 0.9950 0.9972 0.9956 0.9956 0.9990 0.9992 0.9980 0.9985 0.9994 0.9984 0.9984 0.9963 0.9847 0.9987 0.9832 0.9836 0.9960 0.9975 0.9971 0.9974 0.9938 0.9842 0.9985 0.9840 0.9850 0.9988 0.9938 0.9920 0.9963

20.7 33.8 117.1 2.2 9.4 439.6 21.5 3.9 30.1 744.3 – 154.5 154.5 35.1 33.5 360.8 53.9 86.5 87.8 87.8 7.6 31.4 13.4 40.1 59.6 49.9 70.9 102.4 49.9 56.3 51.9 4.7 39.2 72.8 59.3 41.7 12.7 10.2

L-Ala

(+) (+) (+) (+) (+)

20 10 10 200 100

5000 5000 5000 5000 5000

0.9996 0.9991 0.9953 0.9992 0.9990

200 50 50 500 200

5000 5000 5000 5000 5000

0.9986 0.9956 0.9938 0.9963 0.9956

L-Ser

(+)

100

5000

0.9988

200

5000

L-Pro

(+)

10

5000

0.9992

50

5000

L-Val

(+)

20

5000

0.9991

100

L-Thr

(+)

50

5000

0.9993

100

L-Leu

(+)

20

5000

0.9993

L-Ile

(+)

20

5000

L-Asn

(+)

50

5000

L-Hom-Ser

O-Ac-Ser OSHS lL-Cysta Gly

Alkaline

Acid

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

Alkaline

Max (nM)

aKB Fum Suc Mal aKG Cit IsoCit cis-Aco Xu5P 6PG E4Pb R5Pa Ru5Pa F6P G6P FbP DHAP PEP 2PGa 3PGa NAD NADP AMP ADP ATP GMP GDP GTP CMP CDP CTP UMP UDP UTP cAMP AIBA GABA

Retention time

(pmol, on column)

Min (nM)

Acid

8

Table 1 Overview of obtained linearity ranges (<5 lM) with related regression coefficients and method detection limits of investigated low-molecular-weight metabolites measured under alkaline and acid mobile phase conditions.

18.35 ± 0.02 – – 106.3

Fig.2. Standard calibration curves of malonic acid (A) and adenosine triphosphate (B) in the range of 0.5 to 200 lM under alkaline mobile phase conditions. Shown are typical source-dependent deviations at the upper limit of calibration curves, occurring as concentration-dependent signal suppression (A) or enhancement (B). Error indicators display standard deviations (n = 3).

Note. Abbreviations are listed in the online Supplementary material. a Peaks were not completely separated. R(u)5P, ribose/ribulose-5-phosphate; 2(3)PG, 2- and 3-phosphoglycerate. b Insufficient stability under alkaline conditions.

200 – (+) b L-His

20 (+)

–

–

5000

0.9860

–

0.33 67.9 0.9983 5000 50 0.9991 5000

20 (+)

L-Trp

9

0.53

25.01 ± 0.01

14.27 ± 0.01 13.84 ± 0.01

13.14 ± 0.01

65.5

0.34

24.65 ± 0.01

12.69 ± 0.01 0.02 0.02

49.9 6.1

0.9965

0.9869 3000 100 0.9936 3000

20 (+)

L-Arg

L-Phe

20 (+)

20 0.9989 5000

5000

0.03

4.9 4.1

0.25

20.18 ± 0.01

14.43 ± 0.01 14.01 ± 0.01

100 (+)

L-Met

20 (+)

L-Glu

L-Lys

20

20 0.9990 5000

5000

0.9966

0.08 21.9 16.8

0.11

24.78 ± 0.01 24.40 ± 0.01

19.75 ± 0.01 0.25

0.18

49.6 37.9

0.10 36.3 20.7 0.9933

0.9965 5000

5000

0.9991 5000

100

0.9987 5000

100

0.19

20.52 ± 0.01

18.37 ± 0.01 17.95 ± 0.01

20.05 ± 0.01 0.96

0.27

0.16

3.4

0.02

192.0

54.6

31.7

0.9959

0.9972 5000

0.9987 5000

50

(+) L-Gln

L-Asp

(+)

50

0.9989 5000

200

R2 Max (nM) Min (nM) R2 Max (nM) Min (nM)

Acid Alkaline

Linearity range Polarity mode Metabolite

Table 1 (continued)

5000

(min) (min)

Alkaline

(pmol, on column)

Alkaline Acid

(nM)

Retention time

Alkaline

Acid Method detection limit

Acid

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

Fig.3. Metabolite distribution in different MDLs under alkaline and acidic mobile phase conditions.

revealed a high similarity between quantitative results within the different calibration methods, as confirmed by one-way analysis of variance (ANOVA). With the exception of NAD (<10%), the concentrations obtained by the three different quantification strategies deviated by less than 6%. This finding is taken as evidence for the successful reduction or even elimination of matrix effects due to coeluting matrix ions thanks to chromatographic separation [12,13], whereas nonwanted coeluting matrix ions are known to strongly affect ionization yields of targeted analytes [13,16]. Although the set of investigated metabolites turned out to be straightforward quantifiable by external calibration with satisfying accuracy, other metabolites could require more elaborative calibration strategies. Even though sample preparation and extraction protocols were not the focus of this optimization, obtained results were

10

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

Fig.4. Overview of metabolic-specific responsivities depending on mobile phase pH. Shown are relative peak areas obtained by measurement of 5-lM standards under basic conditions normalized to those at acidic conditions.

consistent with previously published endogenous pool concentrations in E. coli under similar cultivation conditions. Notably, intracellular pool sizes are known to be significant depending on strain context, cultivation strategy, sample preparation, and assumed or investigated aqueous intracellular volumes [37–40]. Nevertheless, the intracellular metabolite concentrations of L-alanine

and L-lysine are in good agreement with the reported measurements of Bennett and coworkers and Yuan and coworkers [38,39]. Measured NAD and L-valine pools are in the same concentration range but differ significantly [37,40]. Steady-state concentrations of glycine and L-leucine have not yet been published for E. coli under comparable cultivation conditions.

Discussion From the analytical perspective, the ultimate goal of metabolomics is the reliable identification and quantification of all metabolites present in biological samples taken at distinct time points, thereby representing individual cellular states. Considering the complexity of metabolic networks, comprising a vast number of compounds covering a wide chemical diversity, quantitative analyses are still a great challenge. In this work, we have presented a robust HILIC–MS/MS method for quantitative analysis of a broad range of common intracellular metabolites with high selectivity and sensitivity. So far, the majority of related LC–MS studies use reversed phase [41–48] or hydrophilic interaction stationary phases [12,13,15,16,49–54]. Despite multiple advantages such as robustness and straightforward implementation, reversed phase liquid chromatography (RPLC) typically lacks adequate retention for nonderivatized hydrophilic analytes on silica-based reversed phase stationary phases [33]. For example, RPLC approaches offer only poor retention performance for small molecules with acidic functional groups in their

deprotonated form, and lowering the mobile phase pH to improve compound retention usually leads to unacceptable peak shapes of polar basic or zwitterionic compounds [55]. Consequently, complex multicomponent samples of polar analytes are analyzed either by two separated methods or using ion pairing agents such as trifluoroacetic acid (TFA) and tributylamine (TBA) [41–48]. The latter may cause handling and performance issues for coupled ESI–MS detection with respect to chromatographic reproducibility of polar compounds, ion suppression due to interfering ion pair agents, or contamination of the mass spectrometer. In this context, HILIC is increasingly preferred because it enables the separation of charged and polar analytes and exhibits excellent compatibility with ESI–MS detection [12,13]. HILIC is characterized by the analyte partitioning between a water-enriched layer, partially immobilized on the stationary phase, and a predominantly polar organic mobile phase [32–34]. On the other side, polymeric material-based columns support alkaline mobile phase conditions and acidic pH elution buffers for enhancing retention and chromatographic quality of polar analytes within multifunctional metabolite extracts. Compared with RPLC, HILIC methods are suspected to show extended reequilibration times and retention time drifts that cannot (yet) be modeled adequately [34,56]. We have observed that reequilibration of 8 to 10 column volumes between sample injections is sufficient to achieve adequate analytical reproducibility provided that appropriate initial column equilibration was performed (50 column volumes). This procedure resulted in retention time standard deviations of less than 0.06 min under alkaline conditions even for complex intracellular matrices. It is evident that metabolite retention times are more or less dependent on column sorbent lots, mobile phase preparation, and matrix context of the samples. To consider potential variations of sample composition, we recommend test runs for adapting MRM detection windows.

11

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

pairs such as fructose/glucose 6-phosphate, citrate/isocitrate, and

Fig.5. Absolute intracellular concentrations of selected metabolites in exponentially growing E. coli K-12 MG1655 quantified by HILIC–MS/MS under alkaline conditions (pH 9.2) Measurements are based on standard-based external calibration (white), isotope dilution (gray), and standard addition (black). Error bars describe the standard deviations of four replicate measurements.

An overview of analyte retention in intracellular extracts is shown in Table 2. Although alkaline mobile phase conditions can distinctly enhance chromatographic quality of multifunctional metabolite extracts, so far most chromatographic methods focus on acidic or neutral mobile phase conditions regardless of applied stationary phases. Regardless of which approach is followed, it needs to fulfill the requirements for dealing with complex metabolic matrices. To be precise, the quality of the chromatographic method to separate isobaric compounds that lack specific MRM transitions is of crucial importance for achieving distinct and accurate quantification of intracellular metabolites. The complete separations of isobaric

L-leucine/L-isoleucine (Fig. 1) are likewise examples frequently present in endogenous extracts that confirm the quality of the presented method. In this study, we clearly showed that strong alkaline mobile phase conditions in HILIC achieve acidic performance data at least, mostly being even superior with respect to peak shapes and resolutions. An overview of extracted ion chromatograms of all 56 metabolites measured under acidic and alkaline mobile phase conditions is included in the Supplementary material. Regarding the ionization efficiency in ESI–MS, it has been assumed that signal intensity of measured analytes strongly depends on bulk solution pH, resulting in a direct relationship between ESI–MS response and equilibrium concentrations of ionizable species in the mobile phase. For example, amino acids are still preferably detected in positive electrospray mode under acidic conditions to promote ionization and, hence, increase analyte responsivity [41]. Likewise, in negative ionization mode, alkaline conditions are preferred for quantification of phosphorylated compounds such as phosphopeptides to enhance the formation of phosphorylation-specific marker ions [57]. On this point, related reports have demonstrated ‘‘unexpected’’ improved method sensitivity of various substances under alkaline conditions. Peptides measured in positive ionization mode presented higher signal-to-noise ratios when using ammonium hydroxide (pH 10.0) mobile phase [58]. Besides, using high-pH resistant reversed phase C18 columns, it was possible to investigate the influence of mobile phase pH on ESI–MS responsivities of a broad range of chemical entities of pharmaceutical interest, where analyte signals were significantly increased under basic pH conditions [55,59]. Studying ESI of amino acids, Mansoori and coworkers defined this phenomenon as ‘‘wrong-way-round’’ (WWR) ionization and outlined the absence of a simple and/or predictive dependency of analyte ESI–MS response on eluent pH [60]. During recent years, different mechanisms were controversially discussed to explain this ‘‘anomalous’’ ionization behavior – electrolytic production of protons, proton emission from electrospray droplet surface layer, gas phase proton transfers by strong gas phase acids such as ammonia, and gas phase chemical ionization by corona discharge induction or by precursors in the same solution [59,61]. Independent of the highly valuable discussion about feasible molecular interactions, our approach achieved approximately 1.5-fold higher signal intensities under alkaline conditions (vs. acidic conditions) for 70% of the investigated metabolites. Furthermore, responsivities of phosphorylated compounds were enhanced in negative ionization mode, as expected. The remaining 30% exhibited similar or even slightly improved signal intensity, with the only exceptions being L-arginine and cyclic adenosine monophosphate (cAMP). Furthermore, 90% of evaluated metabolites show improved MDLs lower than 100 nM (0.5 pmol on column) with significantly reduced lower linearity limits at alkaline mobile phase conditions.

Table 2 Calibration functions and corresponding correlation coefficients of investigated metabolites. Metabolite

Retention time (min)

External calibration

Isotope dilution

Standard addition

One-way ANOVA

Slope

Intercept

R2

Slope

Intercept

R2

Slope

Intercept

R2

F

P value

L-Leu

13.17 ± 0.04

3.20E+05

3.80E+04

0.9910

4.01E 01

4.50E 02

0.9913

3.28E+05

2.55E+05

0.9986

3.560

0.0726

L-Lys

24.94 ± 0.05

2.41E+05

2.90E+04

0.9975

5.08E 02

4.10E 03

0.9969

2.35E+05

4.09E+05

0.9995

1.622

0.2503

L-Val

14.98 ± 0.04

1.42E+05

4.23E+03

0.9969

2.59E 01

3.40E 03

0.9975

1.45E+05

2.99E+05

0.9989

1.379

0.3002

NAD Gly

20.06 ± 0.05 18.42 ± 0.05 17.36 ± 0.05

4.25E+04 1.91E+04 6.93E+04

2.74E+03 7.34E+03 5.56E+04

0.9978 0.9994 0.9977

1.31E+00 7.77E 02 5.03E 02

0.9982 0.9991 0.9984

4.40E+04 1.95E+04 7.06E+04

1.36E+05 2.60E+05 8.90E+05

0.9994 0.9999 0.9997

9.193 0.121 1.778

0.0067 0.8876 0.2235

L-Ala

Note. F(crit) = 4.256.

8.93E 02 1.96E 02 3.82E 02

12

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13

Approximately 66% of the metabolites reveal linearity ranges of 5 to 50 nM as lower boundary, whereas only 30% of the metabolites meet this target at acidic conditions. The increase in signal intensities under alkaline conditions in negative mode, especially for phosphorylated compounds, agrees with the expected analyte behavior. On the other hand, metabolites measured in positive mode do not show the classical performance, which could be explained by more efficient ionization through the aforementioned WWR mechanism. Summarizing, these results confirm that ESI–MS response depends on interfering effects, including chemical nature of targeted analytes, pH, buffer types, and organic modifier in the mobile phase. Consequently, more comprehensive and metabolite-specific studies are necessary to unravel underlying mechanisms for identifying the most prominent parameters. The quality of the method was demonstrated by absolute quantification of selected metabolites in intracellular extracts of E. coli biomasses using standard-based external calibration, isotope dilution, and standard addition as calibration strategies. To verify metabolite stability under alkaline conditions, direct injection measurements of standard solutions were made lasting for 14 h. Phosphorylated compounds and organic acids presented almost no signal decrease, and amino acids presented less than 20% signal decrease. Exceptions with higher degradation such as L-glutamine and erythrose-4-phosphate should be analyzed by isotope dilution as the calibration strategy to compensate analyte loss. Detailed data of stability evaluation of investigated metabolites under alkaline conditions are shown in the Supplementary material. Sample analysis of six metabolites revealed consistent results with published intracellular steady-state concentrations and exhibits a high similarity between calibration methods, emphasizing a sufficient reduction of matrix effect due to suitable chromatographic separation. Nevertheless, the authors suggest carefully evaluating specific matrix effects for preventing nonwanted bias on metabolite concentrations derived from straightforward quantification by standard-based external calibration. In summary, using alkaline mobile phases with ZIC–pHILIC emerged as a powerful approach offering highly selective and sensitive separation conditions for absolute metabolite quantification. As such, the technique should be applicable for a broad range of analytes in comprehensive metabolic studies, enabling simultaneous quantitation of dozens of metabolites in a single run. Acknowledgments The authors thank Oliver Vielhauer, Olga Bungart, Benjamin Gann, Andrea Seipel, Mira Lenfers-Lücker, Lisa Junghans, and Michael Kraml (Institute of Biochemical Engineering, University of Stuttgart) for experimental support. The work was funded by the Bundesministerium für Bildung und Forschung (Grant 0315867, BMBF, Berlin, Germany) in cooperation with Evonik Industries and was cosupported by the German Academic Exchange Service (DAAD, Bonn, Germany) and the Instituto Tecnológico de Costa Rica (ITCR, Cartago, Costa Rica). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ab.2015.01.002. References [1] O. Fiehn, W. Weckwerth, Deciphering metabolic networks, Eur. J. Biochem. 270 (2003) 579–588. [2] M. Oldiges, S. Lütz, S. Pflug, K. Schroer, N. Stein, C. Wiendahl, Metabolomics: current state and evolving methodologies and tools, Appl. Microbiol. Biotechnol. 76 (2007) 495–511.

[3] F. Fei, D.M. Bowdish, B.E. McCarry, Comprehensive and simultaneous coverage of lipid and polar metabolites for endogenous cellular metabolomics using HILIC–TOF-MS, Anal. Bioanal. Chem. 406 (2014) 3723–3733. [4] O. Fiehn, Combining genomics, metabolome analysis, and biochemical modelling to understand metabolic networks, Comp. Funct. Genomics 2 (2001) 155–168. [5] J. Allen, H.M. Davey, D. Broadhurst, J.K. Heald, J.J. Rowland, S.G. Oliver, D.B. Kell, High-throughput classification of yeast mutants for functional genomics using metabolic footprinting, Nat. Biotechnol. 21 (2003) 692–696. [6] W.B. Dunn, D.I. Ellis, Metabolomics: current analytical platforms and methodologies, Trends Anal. Chem. 24 (2005) 285–294. [7] M. Oldiges, M. Kunze, D. Degenring, G.A. Sprenger, R. Takors, Stimulation, monitoring, and analysis of pathway dynamics by metabolic profiling in the aromatic amino acid pathway, Biotechnol. Prog. 20 (2004) 1623–1633. [8] J.B. Magnus, D. Hollwedel, M. Oldiges, R. Takors, Monitoring and modeling of the reaction dynamics in the valine/leucine synthesis pathway in Corynebacterium glutamicum, Biotechnol. Prog. 22 (2006) 1071–1083. [9] J.B. Magnus, M. Oldiges, R. Takors, The identification of enzyme targets for the optimization of a valine producing Corynebacterium glutamicum strain using a kinetic model, Biotechnol. Prog. 25 (2009) 754–762. [10] M. Oldiges, R. Takors, Applying metabolic profiling techniques for stimulus– response experiments: chances and pitfalls, Adv. Biochem. Eng. Biotechnol. 92 (2005) 173–196. [11] H.C. Bergmeyer, Methods of Enzymatic Analysis, Verlag Chemie, Weinheim, Germany, 1985. [12] B. Preinerstorfer, S. Schiesel, M. Lämmerhofer, W. Lindner, Metabolic profiling of intracellular metabolites in fermentation broths from b-lactam antibiotics production by liquid chromatography–tandem mass spectrometry methods, J. Chromatogr. A 1217 (2010) 312–328. [13] S. Schiesel, M. Lämmerhofer, W. Lindner, Multitarget quantitative metabolic profiling of hydrophilic metabolites in fermentation broths of b-lactam antibiotics production by HILIC–ESI-MS/MS, Anal. Bioanal. Chem. 396 (2010) 1655–1679. [14] A. Buchholz, J. Hurlebaus, C. Wandrey, R. Takors, Metabolomics: quantification of intracellular metabolite dynamics, Biomol. Eng. 1 (2002) 5–15. [15] M. Dell’mour, L. Jaitz, E. Oburger, M. Puschenreiter, G. Koellensperger, S. Hann, Hydrophilic interaction LC combined with electrospray MS for highly sensitive analysis of underivatized amino acids in rhizosphere research, J. Sep. Sci. 33 (2010) 911–922. [16] P. Kiefer, J.C. Portais, J.A. Vorholt, Quantitative metabolome analysis using liquid chromatography–high-resolution mass spectrometry, Anal. Biochem. 382 (2008) 94–100. [17] J.M. Buescher, S. Moco, U. Sauer, N. Zamboni, Ultrahigh performance liquid chromatography–tandem mass spectrometry method for fast and robust quantification of anionic and aromatic metabolites, Anal. Chem. 82 (2010) 4403–4412. [18] L. Wu, M.R. Mashego, J.C. van Dam, A.M. Proell, J.L. Vinke, C. Ras, W.A. van Winden, W.M. van Gulik, J.J. Heijnen, Quantitative analysis of the microbial metabolome by isotope dilution mass spectrometry using uniformly 13 C-labeled cell extracts as internal standards, Anal. Biochem. 336 (2005) 164–171. [19] M.R. Mashego, L. Wu, J.C. van Dam, C. Ras, J.L. Vinke, W.A. van Winden, W.M. van Gulik, J.J. Heijnen, MIRACLE: mass isotopomer ratio analysis of U-13C-labeled extracts – A new method for accurate quantification of changes in concentrations of intracellular metabolites, Biotechnol. Bioeng. 20 (2004) 620–628. [20] U. Roessner, C. Wagner, J. Kopka, R.N. Trethewey, L. Willmitzer, Simultaneous analysis of metabolites in potato tuber by gas chromatography–mass spectrometry, Plant J. 23 (2000) 131–142. [21] S.G. Villas-Bôas, S. Mas, M. Akesson, J. Smedsgaard, J. Nielsen, Mass spectrometry in metabolome analysis, Mass Spectrom. Rev. 24 (2005) 613– 646. [22] J.M. Büscher, D. Czernik, J.C. Ewald, U. Sauer, N. Zamboni, Cross-platform comparison of methods for quantitative metabolomics of primary metabolism, Anal. Chem. 81 (2009) 2135–2143. [23] H.F. Kvitvang, T. Andreassen, T. Adam, S.G. Villas-Bôas, P. Bruheim, Highly sensitive GC/MS/MS method for quantitation of amino and nonamino organic acids, Anal. Chem. 83 (2011) 2705–2711. [24] M.M. Koek, B. Muilwijk, M.J. van der Werf, T. Hankemeier, Microbial metabolomics with gas chromatography/mass spectrometry, Anal. Chem. 78 (2006) 1272–1281. [25] O. Vielhauer, M. Zakhartsev, T. Horn, R. Takors, M. Reuss, Simplified absolute metabolite quantification by gas chromatography–isotope dilution mass spectrometry on the basis of commercially available source material, J. Chromatogr. B 879 (2011) 3859–3870. [26] J.L. Edwards, C.N. Chisolm, J.G. Shackman, R.T. Kennedy, Negative mode sheathless capillary electrophoresis electrospray ionization–mass spectrometry for metabolite analysis of prokaryotes, J. Chromatogr. A 1106 (2006) 80–88. [27] R. Ramautar, G.W. Somsen, G.J. de Jong, CE–MS for metabolomics: developments and applications in the period 2010–2012, Electrophoresis 34 (2013) 86–98. [28] T. Soga, Y. Kakazu, M. Robert, M. Tomita, T. Nishioka, Qualitative and quantitative analysis of amino acids by capillary electrophoresis–electrospray ionization–tandem mass spectrometry, Electrophoresis 25 (2004) 1964– 1972.

Alkaline HILIC MS metabolite quantification / A. Teleki et al. / Anal. Biochem. 475 (2015) 4–13 [29] A.A. de Graaf, K. Striegel, R.M. Wittig, B. Laufer, G. Schmitz, W. Wiechert, G.A. Sprenger, H. Sahm, Metabolic state of Zymomonas mobilis in glucose-, fructose-, and xylose-fed continuous cultures as analysed by 13C- and 31 P-NMR spectroscopy, Arch. Microbiol. 171 (1999) 371–385. [30] A. Teleman, P. Richard, M. Toivari, M. Penttilä, Identification and quantitation of phosphorus metabolites in yeast neutral pH extracts by nuclear magnetic resonance spectroscopy, Anal. Biochem. 272 (1999) 71–79. [31] Q.Z. Wang, C.Y. Wu, T. Chen, X. Chen, X.M. Zhao, Integrating metabolomics into a systems biology framework to exploit metabolic complexity: strategies and applications in microorganisms, Appl. Microbiol. Biotechnol. 70 (2006) 151– 161. [32] K. Spagou, H. Tsoukali, N. Raikos, H. Gika, I.D. Wilson, G. Theodoridis, Hydrophilic interaction chromatography coupled to MS for metabonomic/metabolomic studies, J. Sep. Sci. 33 (2010) 716–727. [33] H.P. Nguyen, K.A. Schug, The advantages of ESI-MS detection in conjunction with HILIC mode separations: fundamentals and applications, J. Sep. Sci. 31 (2008) 1465–1480. [34] B. Buszewski, S. Noga, Hydrophilic interaction liquid chromatography (HILIC) – a powerful separation technique, Anal. Bioanal. Chem. 402 (2012) 231–247. [35] C.J. Bolten, P. Kiefer, F. Letisse, J.C. Portais, C. Wittmann, Sampling for metabolome analysis of microorganisms, Anal. Chem. 79 (2007) 3843–3849. [36] G. Wells, H. Prest, C.W. Russ IV, Signal, noise, and detection limits in mass spectrometry, Agilent Technologies, 2011. http://www.chem.agilent.com/ Library/technicaloverviews/Public/5990-7651EN.pdf. [37] J. Hiller, E. Franco-Lara, D. Weuster-Botz, Metabolic profiling of Escherichia coli cultivations: evaluation of extraction and metabolite analysis procedures, Biotechnol. Lett. 29 (2007) 1169–1178. [38] J. Yuan, W.U. Fowler, E. Kimball, W. Lu, J.D. Rabinowitz, Kinetic flux profiling of nitrogen assimilation in Escherichia coli, Nat. Chem. Biol. 2 (2006) 529–530. [39] B.D. Bennett, E.H. Kimball, M. Gao, R. Osterhout, S.J. van Dien, J.D. Rabinowitz, Absolute metabolite concentrations and implied enzyme active site occupancy in Escherichia coli, Nat. Chem. Biol. 5 (2009) 593–599. [40] A. Buchholz, R. Takors, C. Wandrey, Quantification of intracellular metabolites in Escherichia coli K12 using liquid chromatographic–electrospray ionization tandem mass spectrometric techniques, Anal. Biochem. 295 (2001) 129–137. [41] M. Piraud, C. Vianey-Saban, K. Petritis, C. Elfakir, J.P. Steghens, A. Morla, D. Bouchu, ESI-MS/MS analysis of underivatised amino acids: a new tool for the diagnosis of inherited disorders of amino acid metabolism – Fragmentation study of 79 molecules of biological interest in positive and negative ionisation mode, Rapid Commun. Mass Spectrom. 17 (2003) 1297–1311. [42] M. Zoppa, L. Gallo, F. Zacchello, G. Giordano, Method for the quantification of underivatized amino acids on dry blood spots from newborn screening by HPLC–ESI-MS/MS, J. Chromatogr. B 831 (2006) 267–273. [43] M. Armstrong, M.K. Jonscher, M.A. Reisdorph, Analysis of 25 underivatized amino acids in human plasma using ion-pairing reversed-phase liquid chromatography/time-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 21 (2007) 2717–2726. [44] M. de Person, P. Chaimbault, C. Elfakir, Analysis of native amino acids by liquid chromatography/electrospray ionization mass spectrometry: comparative study between two sources and interfaces, J. Mass Spectrom. 43 (2008) 204– 215. [45] F. Michopoulos, N. Whalley, G. Theodoridis, I.D. Wilson, T.P. Dunkley, S.E. Critchlow, Targeted profiling of polar intracellular metabolites using ion-pair high performance liquid chromatography and ultra-high performance liquid chromatography coupled to tandem mass spectrometry: applications to serum, urine, and tissue extracts, J. Chromatogr. A 1349 (2014) 60–68.

13

[46] R.M. Seifar, C. Ras, A.T. Deshmukh, K.M. Bekers, C.A. Suarez-Mendez, A.L. da Cruz, W.M. van Gulik, J.J. Heijnen, Quantitative analysis of intracellular coenzymes in Saccharomyces cerevisiae using ion pair reversed phase ultra high performance liquid chromatography tandem mass spectrometry, J. Chromatogr. A 1311 (2013) 115–120. [47] P. Kiefer, N. Delmotte, J.A. Vorholt, Nanoscale ion-pair reversed-phase HPLC– MS for sensitive metabolome analysis, Anal. Chem. 83 (2011) 850–855. [48] B. Luo, K. Groenke, R. Takors, C. Wandrey, M. Oldiges, Simultaneous determination of multiple intracellular metabolites in glycolysis, pentose phosphate pathway, and tricarboxylic acid cycle by liquid chromatography– mass spectrometry, J. Chromatogr. A 1147 (2007) 153–164. [49] H. Schlichtherle-Cerny, M. Affolter, C. Cerny, Hydrophilic interaction liquid chromatography coupled to electrospray mass spectrometry of small polar compounds in food analysis, Anal. Chem. 75 (2003) 2349–2354. [50] T. Langrock, P. Czihal, R. Hoffmann, Amino acid analysis by hydrophilic interaction chromatography coupled on-line to electrospray ionization mass spectrometry, Amino Acids 30 (2006) 291–297. [51] M. Kato, H. Kato, S. Eyama, A. Takatsu, Application of amino acid analysis using hydrophilic interaction liquid chromatography coupled with isotope dilution mass spectrometry for peptide and protein quantification, J. Chromatogr. B 877 (2009) 3059–3064. [52] H.G. Gika, G.A. Theodoridis, U. Vrhovsek, F. Mattivi, Quantitative profiling of polar primary metabolites using hydrophilic interaction ultrahigh performance liquid chromatography–tandem mass spectrometry, J. Chromatogr. A 1259 (2012) 121–127. [53] G. Paglia, M. Magnúsdóttir, S. Thorlacius, O.E. Sigurjónsson, S. Guðmundsson, B.Ø. Palsson, I. Thiele, Intracellular metabolite profiling of platelets: evaluation of extraction processes and chromatographic strategies, J. Chromatogr. B 898 (2012) 111–120. [54] B.A. Rappold, R.P. Grant, HILIC–MS/MS method development for targeted quantitation of metabolites: practical considerations from a clinical diagnostic perspective, J. Sep. Sci. 34 (2011) 3527–3537. [55] L. Peng, T. Farkas, Analysis of basic compounds by reversed-phase liquid chr omatography–electrospray mass spectrometry in high-pH mobile phases, J. Chromatogr. A 1179 (2008) 131–144. [56] D.Q. Tang, L. Zou, X.X. Yin, C.N. Ong, HILIC–MS for metabolomics: an attractive and complementary approach to RPLC–MS, Mass Spectrom. Rev., in press, 2014. http://dx.doi.org/ 10.1002/mas.21445. [57] A. Beck, M. Deeg, K. Moeschel, E.K. Schmidt, E.D. Schleicher, W. Voelter, H.U. Häring, R. Lehmann, Alkaline liquid chromatography/electrospray ionization skimmer collision-induced dissociation mass spectrometry for phosphopeptide screening, Rapid Commun. Mass Spectrom. 15 (2001) 2324– 2333. [58] A.J. Tomlinson, R.M. Chicz, Microcapillary liquid chromatography/tandem mass spectrometry using alkaline pH mobile phases and positive ion detection, Rapid Commun. Mass Spectrom. 17 (2003) 909–916. [59] J. Tso, D.S. Aga, Wrong-way-round ionization of sulfonamides and tetracyclines enables simultaneous analysis with free and conjugated estrogens by liquid chromatography tandem mass spectrometry, Anal. Chem. 83 (2011) 269–277. [60] B.A. Mansoori, D.A. Volmer, R.K. Boyd, ‘‘Wrong-way-round’’ electrospray ionization of amino acids, Rapid Commun. Mass Spectrom. 11 (1997) 1120– 1130. [61] Y. Cheng, Z. Lu, U. Neue, Ultrafast liquid chromatography/ultraviolet and liquid chromatography/tandem mass spectrometric analysis, Rapid Commun. Mass Spectrom. 15 (2001) 141–151.