Isotopic Studies of Metabolic Systems by Mass ... - ACS Publications

Sep 16, 2014 - Figure 1. Representation of Pascal's triangle containing the binomial coefficient used to predict CIDs as a function of the abundance o...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/ac

Isotopic Studies of Metabolic Systems by Mass Spectrometry: Using Pascal’s Triangle To Produce Biological Standards with Fully Controlled Labeling Patterns Pierre Millard,†,‡,§ Stéphane Massou,†,‡,§ Jean-Charles Portais,†,‡,§ and Fabien Létisse*,†,‡,§ †

Université de Toulouse; INSA, UPS, INP; LISBP, 135 Avenue de Rangueil, 31077 Toulouse, France INRA, UMR792, Ingénierie des Systèmes Biologiques et des Procédés, 31400 Toulouse, France § CNRS, UMR5504, 31400 Toulouse, France ‡

S Supporting Information *

ABSTRACT: Mass spectrometry (MS) is widely used for isotopic studies of metabolism in which detailed information about biochemical processes is obtained from the analysis of isotope incorporation into metabolites. The biological value of such experiments is dependent on the accuracy of the isotopic measurements. Using MS, isotopologue distributions are measured from the quantitative analysis of isotopic clusters. These measurements are prone to various biases, which can occur during the experimental workflow and/or MS analysis. The lack of relevant standards limits investigations of the quality of the measured isotopologue distributions. To meet that need, we developed a complete theoretical and experimental framework for the biological production of metabolites with fully controlled and predictable labeling patterns. This strategy is valid for different isotopes and different types of metabolisms and organisms, and was applied to two model microorganisms, Pichia augusta and Escherichia coli, cultivated on 13C-labeled methanol and acetate as sole carbon source, respectively. The isotopic composition of the substrates was designed to obtain samples in which the isotopologue distribution of all the metabolites should give the binomial coefficients found in Pascal’s triangle. The strategy was validated on a liquid chromatography−tandem mass spectrometry (LC-MS/MS) platform by quantifying the complete isotopologue distributions of different intracellular metabolites, which were in close agreement with predictions. This strategy can be used to evaluate entire experimental workflows (from sampling to data processing) or different analytical platforms in the context of isotope labeling experiments.

M

The above strategies can be used to assess the quality of data of specific compounds or specific labeled forms, or to assess the consistency of measurement between different analytical platforms rather than the accuracy of the data per se. A major issue is that the MS measurement of isotopologue distributions is not only compound-dependent but also isotopologue-dependent. This is because isotopic effects can occur at all stages of the analytical procedures and may lead to significant differences in the behavior of metabolites, depending on their isotopic content. This means that the reliability of MS measurements needs to be evaluated for each individual isotopologue of all the compounds of interest. Hence, an ideal standard should contain all the isotopologues of interest in known (or predictable) and sufficient amounts. Biologically produced standards at natural abundance contain all compounds of interest but only their lightest isotopologues can be detected.20,21 Similarly, only the heaviest isotopologues can be detected in biological samples containing

ass spectrometry (MS) is extensively used for isotopic studies of metabolism, either to investigate the nature and rate of biochemical reactions or to identify or quantify metabolites.1−10 For a given molecule, MS distinguishes molecular entities according to the number of label atoms incorporated (i.e., by distinguishing between isotopologues), from which the labeling patterns of each metabolite can be determined in terms of isotopologue distribution.11 The reliability of isotopic measurements determines the quality of the isotopic information and, hence, the biological value of the labeling experiments.12−14 The MS measurement of isotopic patterns is prone to diverse errors that have been detailed in various publications and can occur at all levels of the analytical workflow.9,15−19 The reliability of these measurements is assessed using standards or biological samples at natural isotopic abundance or using mixtures of commercially available labeled compounds.15,16,20,21 Another strategy consists of comparing isotopic data obtained from the analyses of the same labeled biological samples across different MS methods or platforms.13,20,22 © 2014 American Chemical Society

Received: July 7, 2014 Accepted: September 16, 2014 Published: September 16, 2014 10288

dx.doi.org/10.1021/ac502490g | Anal. Chem. 2014, 86, 10288−10295

Analytical Chemistry

Article

metabolites of P. augusta, 1 mL of broth was filtered through 0.2 μm Sartolon polyamide (Sartorius, Goettingen, Germany). Cells were washed with 1 mL of fresh culture medium containing the same mix of labeled substrate as the one used for the 13Clabeling experiment. The filters were rapidly plunged into liquid nitrogen and then stored at −80 °C. Intracellular metabolites were extracted by incubating filters in closed glass tubes containing 5 mL of an ethanol/H20 (75/25) mixture at 95 °C for 5 min. To sample intracellular metabolites of E. coli, 120 μL of broth were rapidly sprayed into precooled centrifuged tubes maintained at −80 °C containing 500 μL of ethanol, homogenized using a vortex, and centrifuged at 12 000 g, at −20 °C for 5 min (Sigma 3-18K, Sigma, Osterode am Harz, Germany). Extraction was performed by pouring 5 mL of an ethanol/H20 (75/25) solution at 95 °C on the pellets, which were then incubated in closed tubes for 2 min. Cellular extracts were cooled on ice and cell debris was removed by centrifugation at 6500 g, at 4 °C for 5 min.26 Supernatants were evaporated for 4 h (SC110A SpeedVac Plus, ThermoFisher, Waltham, MA, USA). The remaining aqueous extract was freeze-dried, resuspended in 200 μL of Milli-Q water, and stored at −80 °C.7 Determination of Carbon Isotopologue Distributions by LC-MS/MS Analysis. Intracellular metabolites were analyzed by ion chromatography (ICS 2000 system, Dionex, CA, USA) coupled with a 4000QTrap triple quadrupole mass spectrometer (ABSciex, Framingham, MA, USA) equipped with a TurboV source (ABSciex) for electrospray ionization, as described elsewhere.20 Isotopic clusters of molecular ions [M -H]− were quantified in the multiple reaction monitoring (MRM) mode (with a mass resolution of 0.5 amu at half peak height) for organic acids (citrate, succinate, fumarate, malate) and phosphorylated compounds (AMP, ADP, CDP, G6P, F1P, F6P, M6P, FBP, 6PG, R5P, PEP, and 2PG/3PG). The daughter ion was a phosphate group (PO3−, m/z = 79, or H2PO4−, m/z = 97) for phosphorylated metabolites, and fragments with loss of a carboxylic group ([M−H−12CO2]− or [M−H−13CO2]−) for organic acids. Carbon isotopologues distributions (CIDs) were calculated from isotopic clusters after correction for naturally occurring isotopes of elements other than carbon using IsoCor27 and considering the natural isotopic abundances given in ref 28. The 13C-enrichment (p) of each metabolite was determined from its CID, assuming fully random distribution of 13C atoms in the molecule. For each compound, p was estimated by iteratively minimizing the sum of the squared weighted differences between measured CIDs and CIDs predicted from eq 2. Optimization was performed using Brent’s method, as implemented in the optimize.f minbound function provided in the SciPy package of Python (http://python.org). For each organism, standard deviations of measured CIDs and of estimated 13C-enrichments were determined from the analysis of three independent biological samples. Sensitivity Analysis, Accuracy, and Precision. The proposed strategy for producing metabolites with fully controlled labeling patterns was evaluated by comparing measured CIDs with those predicted from eq 2, according to the actual composition of labeled substrates measured by NMR. To account for errors in the preparation of labeled substrates and for the precision of NMR measurements, we estimated confidence intervals on predicted CIDs by a Monte Carlo analysis. CIDs were calculated from isotopic compositions of the labeled substrate randomly generated within the range of the NMR

fully labeled metabolites, which are now routinely produced for the purpose of metabolite quantification.23,24 In specific cases, biological samples containing known amounts of all isotopologues can be obtained, and have been used to increase the accuracy of MS analysis through the construction of calibration curves.25 However, in most cases, the isotopologue distribution is dependent on the metabolic state of the organism (i.e., on the distribution of metabolic fluxes) and cannot be known in advance,13,22 which prevents their use for assessment of the quality of MS measurement. In this paper, we provide a complete conceptual and practical framework for the biological production of metabolites with fully controlled labeling patterns that can be used as quality control standards. We first show that the isotopic content of metabolites can be predicted with high precision and fully controlled if the isotopic composition of the labeled substrate is defined appropriately. This strategy is valid for different isotopic tracers and can be used to produce labeled standards for different types of metabolism and different organisms. The proposed strategy was successfully applied to produce 13C-labeled intracellular metabolites with fully predetermined isotopic content by Pichia augusta, which is a methylotrophic organism, and by Escherichia coli, which is a multicarbon utilizer.



EXPERIMENTAL SECTION Organisms and Cultures. Pichia augusta was grown on YNB Difco medium without amino acids (BD, Franklin Lakes, NJ, USA) supplemented with 15 g L−1 of methanol as a unique carbon source. Methanol (HPLC grade) was obtained from Sigma−Aldrich (St. Louis, MO, USA) and 13C-methanol from Eurisotop (St. Aubin, France). Batch cultivations were carried out at 40 °C in a 500 mL Multifors bioreactor (Infors HT, Bottmingen-Basel, Switzerland). The pH was maintained at 4.0 with NaOH (1 M) and dissolved oxygen content was maintained above 20% saturation by aeration with synthetic gas (80% N2/ 20% O2, containing