Explorative Analysis of Urine by Capillary Electrophoresis-Mass Spectrometry in Chronic Patients with Complex Regional Pain Syndrome

Capillary electrophoresis-mass spectrometry for metabolic profiling of body fluids

Capillaire electroforese-massaspectrometrie voor het profileren van metabolieten in lichaamsvloeistoffen

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht

op gezag van de rector magnificus, prof.dr. J.C. Stoof,

ingevolge het besluit van het college voor promoties

in het openbaar te verdedigen

op woensdag 17 februari 2010

des middags te 2.30 uur

door

Rawi Ramautar

geboren op 22 juli 1979 te Paramaribo, Suriname

Promotor: Prof. dr. G.J. de Jong Co-promotoren: Dr. G.W. Somsen Dr. O.A. Mayboroda

This thesis was financially supported by Beckman Coulter, Inc., USA, de Nederlandse Vereniging van Posttraumatische Dystrofie Patiënten, afdeling Neurologie van het Leids Universitair Medisch Centrum, de Jurriaanse Stichting, Spark Holland en afdeling Infectieziekten van het Leids Universitair Medisch Centrum. This Thesis was printed by Ridderprint Offsetdrukkerij BV (Ridderkerk, the Netherlands). ISBN/EAN: 978-90-5335-252-6 Copyright: © 2010 by Rawi Ramautar Niets uit deze uitgave mag verveelvoudigd en/of openbaar gemaakt worden zonder voorafgaande schriftelijke toestemming van de auteur. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means without written permission of the author and the publisher holding the copyrights of the published articles. Cover: An adapted version of the urine wheel by Ullrich Pinder, which was published in 1506 in his book Epiphanie Medicorum. In this figure, the possible colors, smells and tastes of urine are described and these properties were used to diagnose disease.

Voor mijn ouders

Table of Contents Chapter 1 General Introduction 7 Chapter 2 Capillary Electrophoresis-Mass Spectrometry 15

in Metabolomics

Chapter 3 Direct Sample Injection for Capillary 47 Electrophoretic Determination of Organic Acids in Cerebrospinal Fluid

Chapter 4 Metabolic Analysis of Body Fluids by Capillary 65

Electrophoresis using Noncovalently Coated Capillaries

Chapter 5 Evaluation of Capillary Electrophoretic 81

Methods for Global Metabolic Profiling of Urine Chapter 6 Capillary Electrophoresis-Time of Flight- Mass 101

Spectrometry using Noncovalently Bilayer- Coated Capillaries for the Analysis of Amino Acids in Human Urine

4

5

Chapter 7 Explorative Analysis of Urine by Capillary 121

Electrophoresis-Mass Spectrometry in Chronic Patients with Complex Regional Pain Syndrome

Chapter 8 Metabolic Profiling of Human Urine by 145 Capillary Electrophoresis-Mass Spectrometry using

a Positively Charged Capillary Coating Chapter 9 Conclusions and Perspectives 157 Nederlandse Samenvatting (Summary in Dutch) 167 List of abbreviations 173 List of publications 174 Curriculum Vitae 176 Dankwoord 177

Chapter 1

General Introduction


1. Introduction The metabolome is the complete set of endogenous low-molecular-weight metabolites in cells, tissues, and body fluids. The comprehensive analysis of low-molecular-weight metabolites in biological samples is known as metabolomics, which is one of the “omics” sciences. Metabolomics is particularly useful for gaining insight into the (patho)physiological state of an individual as endogenous metabolites represent real end-points of the biochemical pathways that may be perturbed by other 'omes' such as the genome, transcriptome and proteome [1-3]. Therefore, metabolomics has become an essential part of systems biology. Given that the overall status of an individual is reflected by his or her metabolic state, metabolomics has the potential to have a great impact upon medical practice by providing a wealth of relevant biochemical data [4]. For example, by using a metabolomics approach, sarcosine, an N-methyl derivative of the amino acid glycine, was identified as a metabolite that was highly increased in urine during prostate cancer progression [5]. At present, the possibility to use sarcosine as a biomarker for early disease detection and aggressivity prediction of prostate cancer is investigated. The study of the complete metabolome is highly challenging due to its complexity and size. Instead, non-targeted and targeted metabolic profiling approaches are often used [6]. In non-targeted analyses, the aim is to analyze a wide range of endogenous metabolites simultaneously within a single analysis, whereas in targeted analyses the focus is on one or a few classes of metabolites. By analyzing differences between metabolic profiles using multivariate data analysis techniques, metabolites which are relevant to a specific phenotype can be identified. Currently, there are numerous definitions of metabolomics and metabolic profiling, but regardless of terminology, any definition implies an enormous analytical challenge, i.e., to cover a wide range of polarities, concentrations, and sizes of chemical entities composing the metabolome. The overriding need is, therefore, for analytical methods that can produce global metabolic profiles of biological samples, such as urine, in a reproducible way [7]. The development of analytical methods fulfilling these requirements is not an easy task [8]. At present, the analytical methods most often used for metabolic profiling of body fluids are nuclear magnetic resonance spectroscopy (NMR) and gas chromatography-mass spectrometry (GC-MS) [9, 10]. NMR has been used extensively for multicomponent analysis of body fluids and the concept of metabonomics, i.e., the quantitative measurement of the dynamic multi parametric metabolic response of living systems to physiological stimuli or genetic modification, has arisen from NMR-based profiling studies [11]. GC-MS is a routine technique for the screening of inborn errors of metabolism for more than two decades now, and today it is also used for

9

Chapter 1

plant metabolomics studies [10, 12]. Both NMR and GC-MS are considered robust analytical techniques for metabolic profiling studies and they are used in a routine way. More recently, liquid chromatography-mass spectrometry (LC-MS) has shown a great potential for metabolic profiling studies [6]. In contrast to GC-MS, with LC-MS polar compounds can be determined without the requirement for derivatization, and aqueous samples can be injected avoiding the need for solvent transfer. Detection limits in LC-MS for many compounds are significantly lower than in NMR, providing the possibility to monitor metabolic changes that remained unnoticed with NMR. LC-MS allows analysis of a broad range of metabolite classes. The widespread availability of robust LC-MS methods in many research laboratories has induced an enormously increased use of LC-MS for metabolic profiling over the past few years. Recent LC-MS studies in metabolomics have focused on the evaluation of new methodological developments. For instance, the application of LC columns with sub 2-μm particles as packing materials combined with high pressures has shown increased peak capacities, resolution, sensitivity and fast separations. Several recent applications have revealed the great potential of these new LC systems for metabolic profiling [13, 14]. In general, reversed-phase gradient LC-MS methods and electrospray ionization (ESI) in positive and negative mode have been used to obtain metabolic profiles. Reversed-phase LC is very suitable for the analysis of compounds of medium and low polarity, but less suited for highly polar and charged compounds which are poorly retained. Therefore, several researchers have studied the use of hydrophilic interaction liquid chromatography (HILIC) which can provide separations of polar compounds, and if it is used together with RPLC-MS an increased coverage of the metabolome can be obtained [15]. Capillary electrophoresis (CE) is particularly suited for the separation of polar ionogenic and charged substances, as compounds are separated on the basis of their charge-to-size ratio. Since many components in body fluids display high polarity and water-solubility, CE is an attractive technique for metabolic profiling due to its ability to separate ionogenic compounds in an aqueous medium with high efficiency and speed [16]. With regard to metabolic profiling of body fluids, CE has been most often used for targeted metabolite analysis in urine with applications mainly related to the screening of inborn errors of metabolism [17-19]. The potential of CE-MS for the analysis of a small number of endogenous metabolites in urine samples has been shown for patients with various metabolic disorders [20]. The first application of CE-MS for non-targeted metabolic profiling was demonstrated for bacterial extracts, in which more than 1600 metabolites were analyzed [21]. At the start of our project described in this thesis, the utility of CE-MS for metabolic profiling of body fluids (urine, plasma, cerebrospinal fluid) had been hardly explored.

10


The use of CE-MS for the analysis of body fluids appeared to be hindered by reproducibility problems, as indicated by the occurrence of migration-time shifts between analyses and by the loss of separation efficiency in comparison with CE with conventional UV absorbance detection. This means that in order to establish the role of CE-MS in a new field as metabolomics, system stability should be improved, especially for applications where separation profiles have to be compared. The main objective of the studies described in this thesis was to develop reproducible, efficient and possibly fast CE-MS methods for the global metabolic profiling of body fluids with minimal sample pretreatment. 2. Scope and outline of the thesis Using CE with bare fused-silica capillaries, the analysis of body fluids is often difficult due to adsorption of proteins or other matrix components to the capillary wall causing varying electro-osmotic flows and migration times. As reproducible migration times are of utmost importance for the reliable comparison of metabolic profiles and for the observation of small changes in sample composition, the potential of capillaries noncovalently coated with charged polymers, was studied. In this thesis, specifically the performances of coatings consisting of a triple layer of polybrene-dextran sulfate-polybrene (PB-DS-PB) or a bilayer of polybrene-poly(vinylsulfonate) (PB-PVS) were evaluated for metabolic profiling of body fluids. This evaluation was performed by considering aspects as migration-time repeatability, separation efficiency and separation window using real samples, i.e., urine, plasma and CSF. Moreover, the influence of the sample matrix (proteins and salts) on the overall performance of the CE methods using test compounds was studied. In order to evaluate the potential of CE-MS methods based on noncovalently coated capillaries for global metabolic profiling, low and high pH separation conditions were tested to achieve separations of both basic and acidic metabolites. The use of CE in combination with time-of-flight mass spectrometry (TOF-MS) had hardly been explored. Therefore, the coupling of CE with TOF-MS has been optimized and MS-related aspects important for reliable metabolic profiling, such as mass accuracy and ion suppression were evaluated. To improve the concentration sensitivity of CE-MS for metabolic profiling of body fluids, in-capillary preconcentration techniques, such as pH-mediated stacking, have been applied. The applicability of the CE-MS methods has been tested by analyzing clinically relevant samples representing diseases as urinary tract infection, complex regional pain syndrome and bacterial meningitis. Multivariate data analysis was applied for comparison of metabolic profiles obtained with CE-MS.

11

Chapter 1

An overview of the role of CE-MS in the field of metabolic profiling is presented in Chapter 2. This chapter covers various CE separation modes, capillary coatings, MS analyzers, sample preparation techniques, and data analysis methods used in CE-MS for metabolomics studies. The applicability of CE-MS in metabolic profiling is illustrated by examples of the analysis of bacterial extracts, plant extracts, and human urine, plasma, and cerebrospinal fluid. In Chapter 3 a study on the evaluation of a capillary coating consisting of a triple layer of charged polymers (Polybrene-dextran sulfate-Polybrene) for the analysis of organic acids in CSF is presented. The influence of albumin and sodium chloride concentration on the migration times and plate numbers of test compounds was systematically studied. The potential of the PB-DS-PB CE method for the analysis of organic acids in CSF samples of patients with bacterial meningitis was demonstrated. Chapter 4 reports on the potential of capillaries noncovalently coated with charged polymers for the metabolic analysis of body fluids by CE. Direct injections of CSF, plasma and urine samples are applied. The PB-PVS CE system was combined with time-of-flight mass spectrometry (TOF-MS) and evaluated for the fast and reproducible analysis of amino acids in CSF and urine with minimal sample pretreatment. The usefulness of PB-PVS and PB-DS-PB coated capillaries for recording global electrophoretic profiles of urinary metabolites covering a broad range of different compound classes is investigated in Chapter 5. The PB-PVS and PB-DS-PB capillary coatings were studied at acidic (pH 2.0) and alkaline (pH 9.0) separation conditions, thereby providing separation conditions for basic and acidic compounds. Migration-time repeatability and plate numbers were determined for a test mixture representing various metabolite classes. The CE methods were coupled to TOF-MS using ESI in positive and negative mode, and the number of molecular features detected was determined for rat urine samples. The influence of ion suppression on test compounds spiked to rat urine was also evaluated. The development of a CE-TOF-MS method using PB-PVS coated capillaries for the profiling of amino acids in human urine is presented in Chapter 6. The relatively low sample loading capacity of CE was circumvented by using an in-capillary preconcentration step based on pH-mediated stacking. Special attention was paid to the influence of matrix effects on the quantification of amino acids in urine. The applicability of the PB-PVS CE-MS method is demonstrated for the analysis of amino acids in urine samples from patients with urinary tract infection and healthy controls. Multivariate data analysis was used to find differential metabolites. Chapter 7 describes the applicability of the PB-PVS CE-TOF-MS method for metabolic profiling of human urine. It is shown that the CE-TOF-MS method is

12


capable of separating different classes of metabolites simultaneously in human urine. The stability of the TOF-MS instrument was studied by comparing measured masses of test compounds in urine samples with the theoretical masses. The PB-PVS CE-TOF-MS was applied to urine samples of patients with complex regional pain syndrome (CRPS) in order to find metabolic indicators potentially characteristic for CRPS. Multivariate data analysis was used to find differential metabolites that discriminate between urine samples from CRPS patients and controls. The biological relevance of these compounds with respect to CRPS is discussed. In Chapter 8 the applicability of the PB-DS-PB CE-MS method for the metabolic profiling of cationic compounds in human urine is studied. It is shown that various metabolite classes can be analyzed and stable metabolic profiles of urine samples are obtained over time. The PB-DS-PB CE-MS method was applied to the analysis of urine samples from 30 males and 30 females, which previously had been analyzed by reversed-phase UPLC-MS. Multivariate data analysis was used to find metabolites that discriminate between urine samples from males and females. The results were compared with the data obtained by UPLC-MS in order to demonstrate the complementary character of both techniques. Chapter 9 provides some general conclusions and comments on the developed CE-MS methods for metabolic profiling of body fluids. Also perspectives and recommendations are presented. References [1] Van der Greef, J., Stroobant, P., Van der Heijden, R., Curr Opin Chem Biol

2004, 8, 559-565. [2] Nicholson, J. K., Connelly, J., Lindon, J. C., Holmes, E., Nat Rev Drug

Discov 2002, 1, 153-161. [3] Kim, K., Aronov, P., Zakharkin, S. O., Anderson, D., et al., Mol Cell

Proteomics 2009, 8, 558-570. [4] Kaddurah-Daouk, R., Krishnan, K. R., Neuropsychopharmacology 2009, 34,

173-186. [5] Sreekumar, A., Poisson, L. M., Rajendiran, T. M., Khan, A. P., et al., Nature

2009, 457, 910-914. [6] Lenz, E. M., Wilson, I. D., J Proteome Res 2007, 6, 443-458. [7] Theodoridis, G., Wilson, I. D., J Chromatogr B Analyt Technol Biomed Life

Sci 2008, 871, 141-142. [8] Dunn, W. B., Bailey, N. J., Johnson, H. E., Analyst 2005, 130, 606-625. [9] Fiehn, O., Plant Mol Biol 2002, 48, 155-171.

13

Chapter 1

14

[10] Beckonert, O., Keun, H. C., Ebbels, T. M., Bundy, J., et al., Nat Protoc 2007, 2, 2692-2703.

[11] Nicholson, J. K., Lindon, J. C., Holmes, E., Xenobiotica 1999, 29, 1181-1189. [12] Kopka, J., J Biotechnol 2006, 124, 312-322. [13] Gika, H. G., Theodoridis, G. A., Wilson, I. D., J Sep Sci 2008, 31, 1598-1608. [14] Gika, H. G., Macpherson, E., Theodoridis, G. A., Wilson, I. D.,

J Chromatogr B Analyt Technol Biomed Life Sci 2008, 871, 299-305. [15] Cubbon, S., Antonio, C., Wilson, J., Thomas-Oates, J., Mass Spectrom Rev

2009. [16] Ramautar, R., Demirci, A., De Jong, G. J., Trends Anal Chem 2006, 25, 455-

466. [17] Presto Elgstoen, K. B., Jellum, E., Electrophoresis 1997, 18, 1857-1860. [18] Garcia, A., Barbas, C., Aguilar, R., Castro, M., Clin Chem 1998, 44, 1905-

1911. [19] Barbas, C., Adeva, N., Aguilar, R., Rosillo, M., et al., Clin Chem 1998, 44,

1340-1342. [20] Elgstoen, K. B., Zhao, J. Y., Anacleto, J. F., Jellum, E., J Chromatogr A 2001,

914, 265-275. [21] Soga, T., Ohashi, Y., Ueno, Y., Naraoka, H., et al., J Proteome Res 2003, 2,

488-494.

Chapter 2

Capillary Electrophoresis-Mass Spectrometry in Metabolomics

R. Ramautar, G.W. Somsen, G.J. de Jong,

Electrophoresis 2009, 30, 276-291

Capillary Eectrophoresis –Mass Spectrometry in Metabolomics

Abstract An overview of the use of capillary electrophoresis-mass spectrometry (CE-MS) in the field of metabolomics is provided. Metabolomics is concerned with the comprehensive analysis of endogenous low-molecular weight compounds in biological samples. CE-MS has demonstrated to be a powerful technique for the profiling of polar metabolites in biological samples. This review covers the use of various CE separation modes, capillary coatings, MS analyzers, sample preparation techniques and data analysis methods used in CE-MS for metabolomics. The applicability of CE-MS in metabolomics research is illustrated by giving examples of the analysis of bacterial extracts, plant extracts, urine, plasma and cerebrospinal fluid (CSF) samples. The relevant CE-MS metabolomics studies published between 2000 and 2008 are presented in tabular form, including information on sample type and pretreatment and MS detection mode. Future developments with regard to the use of alternative ionization techniques, the use of coupled separation systems and the potential of microchip CE systems for metabolomics are discussed.

17

Chapter 2

1. Introduction Metabolomics is concerned with the comprehensive analysis of endogenous low-molecular weight metabolites in a biological system [1, 2]. Metabonomics is the evaluation of biological systems for changes in endogenous metabolite levels that result from disease or therapeutic treatments [3]. The terms metabolomics and metabonomics are often used interchangeably, and they include approaches as metabolite profiling and metabolite target analysis. Metabolite profiling involves the comparison of metabolite profiles of various biological samples (also called metabolic fingerprinting) and the final goal is often the identification and quantification of a group of metabolites belonging to particular metabolic pathways. Metabolite target analysis is focused on the analysis of a few known metabolites in a certain biological system. Knowledge about metabolite profiles of body fluids such as urine and cerebrospinal fluid (CSF) provides an insight into the metabolic state of an organism and specific biochemical processes. Therefore, metabolomics has become an important approach for the screening of potential diagnostic markers of various diseases and for the detection of pharmacological/toxicological effects obtained following dosing of compounds [3]. This approach has the potential to replace more classical methods in clinical laboratories which are often focused on the measurement of a few specific endogenous metabolites [4]. The major goal in metabolomics is to measure as many endogenous metabolites as possible in a biological sample, often a body fluid, within a single analysis. However, no analytical technique has the potential to give a fully comprehensive metabolite profile of a biological sample. Moreover, it is also not possible to extract the whole metabolome from cells or body fluids without losing some compounds. Instead, individual analytical techniques often provide information on a – preferably predefined – subset of metabolites only, thereby representing a more targeted approach. To date, most metabolomic analyses are performed with gas chromatography-mass spectrometry (GC-MS), liquid chromatography-MS (LC-MS), and nuclear magnetic resonance spectroscopy (NMR) [5]. Applications of GC-MS for large-scale metabolite analysis are mainly found in the field of plant metabolomics [6]. GC-MS is not suitable for non-volatile, thermolabile and /or highly polar compounds and, therefore, derivatization of metabolites is often needed to yield volatile and thermostable analytes. NMR spectroscopy has been used for the metabolic profiling of body fluids such as serum and urine [7]. NMR is rapid, non-destructive and requires minimal sample preparation. Nevertheless, the sensitivity of NMR is limited and analyte amounts of several micrograms are often required. The widespread use of LC–MS in metabolomics is relatively new [8]. LC-MS can supply information on the

18


chemical structure and the quantity of low-abundance metabolites without the need for derivatization. More advanced LC systems, i.e. monolithic capillary LC or UPLC offering improved separation efficiencies, are gaining attention in metabolomics studies [9]. For instance, a monolithic silica-based capillary reversed-phase LC-MS system has been used for the profiling of metabolites in a plant extract [10]. However, many compounds in body fluids are highly polar and ionic, and separation of these compounds using common reversed-phase LC (RPLC) can be problematic [11]. For this reason, novel RPLC stationary phases, such as for example Atlantis dC18 and Alltima HP C18, have recently been developed. These new stationary phases show increased retention for polar compounds [12]. Another approach to increase the retention of ionic compounds in RPLC is the use of ion-pair agents. For instance, a RPLC-MS method using hexylamine as an ion-pair agent has been developed for the simultaneous analysis of nucleotides, sugar nucleotides and sugar phosphates in bacterial extracts [13]. The disadvantage of ion-pair agents, however, is that they can cause significant ionization suppression of analytes in MS. Hydrophilic interaction chromatography (HILIC) is also a promising separation technique for polar compounds. In HILIC compounds are separated using a hydrophilic stationary phase and retention is dominated by polar interactions providing a fully different selectivity than RPLC. HILIC-MS has been used for the analysis of highly polar metabolites in plant extracts [14]. Capillary electrophoresis (CE) in essence is particularly suited for the separation of polar and charged compounds [15], as compounds are separated on the basis of their charge-to-mass ratio. The separation mechanism fundamentally differs from RPLC and, therefore, CE can provide complementary or additional information on the composition of a biological sample. CE separations can be achieved in a fast and highly efficient way without the need for extensive sample pretreatment. Other advantages of CE include the very low - or even absence of - organic solvent consumption, the small amount of other reagents needed, and the use of simple fused-silica capillaries instead of expensive LC columns. A drawback of CE is its poor concentration sensitivity due to the limited sample volume (nanoliters) that can be introduced into the capillary (nanoliters) and the low optical path length when UV absorbance detection is used. However, the sensitivity can be improved by combining CE with MS. Moreover, MS can provide selective detection and structural information of (unknown) metabolites. The use of in-capillary preconcentration techniques can give further gain of sensitivity. Therefore, CE-MS has been recognized recently as an attractive complementary technique for metabolomic studies. Over the last years, quite a number of papers has been published on the use of CE-MS for metabolomics purposes. Recently, Soga et al. reviewed the potential of CE-MS

19

Chapter 2

in metabolomics [16]. Technical aspects of coupling CE to MS were discussed and selected examples from different fields of application (medical science, toxicology, nutrition and agriculture science) were described. The present review discusses general aspects of sample pretreatment for CE-MS in metabolomics, different CE modes, coated capillaries and MS analyzers are outlined, and applications for bacterial and plant extracts, urine, plasma and CSF analysis are shown. CE-MS metabolomic studies reported between 2000 and 2008 are summarized in a table format giving information on sample type, analytes, sample pretreatment, MS detection mode, and limits of detection. Future developments with regard to the use of different mass analyzers, the use of hyphenated systems and the potential of microchip CE systems for metabolomics are discussed. 2. CE-MS methodology 2.1 CE separation modes Capillary zone electrophoresis (CZE) is the main CE separation mode used in CE-MS for metabolomics [16]. The possibility to use volatile background electrolytes (BGEs) provide a high compatibility of CZE with MS. CZE has been used for the targeted profiling of amino acids and organic acids in body fluids in order to screen for metabolic disorders [15, 17, 18]. The potential of CZE-MS for metabolomics was clearly demonstrated by a large-scale metabolite analysis of Bacillus subtilis extracts by Soga et al. [19]. Distinct CZE-ESI-MS methods were described for anionic and cationic metabolites. CZE-ESI-MS analysis of the cationic metabolites was performed with a bare fused-silica capillary using 1 M formic acid (pH 1.8) as BGE. The analysis of the anionic metabolites was performed on a cationic polymer-coated capillary using 50 mM ammonium acetate (pH 8.5) as BGE. A total of 1692 metabolite features were analyzed in a Bacillus subtilis extract by these CE-MS methods of which 150 were identified. Recently, CZE and micellar electrokinetic chromatography (MEKC), which employs micelles as pseudo-stationary phases in the BGE, were used for metabolic fingerprinting of urine samples from diabetic rats [20]. CZE using a polyacrylamide-coated capillary and reverse polarity was used for the profiling of anionic metabolites, and MEKC was used for the profiling of neutral and cationic metabolites. As MEKC offers extra selectivity (it allows the simultaneous separation of neutral and charged metabolites) an extended metabolomic profile of the urine sample was obtained by combining the data obtained with both CE systems. A sulphated β-cyclodextrin-modified MEKC method has also been used for metabolic profiling of human urine, allowing the separation of 80 compounds within 25 min [21]. Although an improved

20


separation and extra selectivity can be obtained using MEKC methods, the coupling of MEKC to MS is problematic and often provides limited sensitivity. Recently, the potential of pressure assisted CEC, an analytical separation technique which combines the high peak efficiency of CE with the stationary phase selectivity of LC, has been evaluated for metabolic profiling of rat urine [22]. Figure 1 clearly shows that CEC is able to measure more urinary metabolites with good resolution than a standard RPLC or capillary LC method. As UV absorbance was used for detection, only eight compounds could be identified in the urine sample using comparison with standards. CEC-MS for metabolomics has not been described yet. However, it would have a good potential as indicated by the recent use of CEC-MS for the screening of protein and peptide biomarkers in plasma from patients with gangrenous and phlegmonous appendicitis [23]. Figure 1. Comparison of capLC (A), HPLC (B), and pCEC (C) analysis of rat urine. CapLC conditions: column, 150 m i.d., 50 cm total length, packed with C18, 5m particle size. HPLC conditions: column, 4.6 mm 250 mm, packed with C18, 5 m particle size; Gradient of water (0.02% TFA) and 95% methanol (0.02%TFA) used for LC separation. CEC conditions: column, 150 m i.d., 50 cm total length, packed with C18, 5m particle size.; Gradient of water (0.01% TFA) and acetonitrile (0.01%TFA) used for separation. Injection volume, 5 L; detection, UV at 214 nm. Reproduced from Ref. [22] with permission.

21

Chapter 2

2.2 Capillary coatings Variability of migration time is an important issue in CE. Procedures to correct for migration time shifts and for aligning electropherograms have been developed [11, 21]. Still, for metabolic profiling studies of biological samples, reproducibility of migration time and peak areas is of utmost importance for a reliable comparison of profiles and to observe small differences in sample composition. Using CE with bare fused-silica capillaries, the analysis of biological samples with minimal sample pretreatment is often not possible due to adsorption of proteins or other matrix components to the capillary wall causing irreproducible electro-osmotic flows and migration times. To minimize these problems, fused-silica capillaries coated with polymers may be used, as has been demonstrated in several CE-based metabolomics studies. For metabolic profiling of human urine by CE-MS a bare fused-silica capillary was dynamically coated with a cationic polymer PolyE-323 to achieve fast analysis times at low pH [11]. The urine samples were analyzed with minimal sample pretreatment. A CE-MS method using a capillary coated with polyvinyl alcohol (PVA) was developed for the simultaneous analysis of catecholamines and metanephrines in human urine. Figure 2 shows the separation of the compounds without and with the capillary coating. A baseline separation of the compounds was obtained with the PVA coated capillary due to the suppressed electro-osmotic flow (EOF) [24]. Figure 2. CE separation of catecholamine and metaphrines (1 M) on a (A) uncoated and (B) PVA-coated capillary. Peaks: 1, dihydroxybenzylamine; 2, 4-hydroxy-3-methoxybenzylamine ; 3, dopamine ; 4, 3-methoxytyramine ; 5, noradrenaline ; 6, normetanephrine ; 7, adrenaline; 8, metanephrine. Reproduced from Ref. [24] with permission.

22


A CE-MS method using capillaries noncovalently coated with a bilayer of Polybrene (PB) and poly(vinyl sulfonate) (PVS) was used for the fast and direct analysis of amino acids in human urine at low pH [25]. The PB-PVS coating proved to be very consistent yielding stable CE-MS patterns of amino acids in urine with favourable migration time repeatability (RSDs <1.5%). In general, these coating procedures have shown to be a simple and fast approach to improve the reproducibility of CE-MS analyses and allow direct injection of biological samples. However, capillary coatings may also have some limitations, such as bleeding, which can influence the MS signals and lead to migration time shifts. Using multiple layers of polymers, stable and consistent coatings can be made which provide an electro-osmotic flow (EOF) which is largely independent from the pH of the BGE. Furthermore, the use of coatings of negative or positive charge allows the direction of the EOF to be adjusted using normal or reversed CE polarity, providing CE systems for anions and cations [26]. 2.3 CE-MS coupling 2.3.1 Interfacing Until now, ESI is the main ionization technique used for CE–MS in metabolomics [16]. The coupling of CE to ESI-MS can be performed via a sheath-liquid interface or a sheathless interface [27]. The sheath-liquid interface is most widely used for CE–MS in metabolomics. In this configuration, the separation capillary is inserted in a tube of larger diameter in a coaxial setting. The conductive sheath liquid, to which the CE terminating voltage is applied, is administered via this outer tube and merges with the CE effluent at the capillary outlet. Usually, a gas flow is applied via a third coaxial capillary in order to facilitate spray formation in the ESI source [27]. The sheath liquid can be used to optimize the ESI process and, therefore, the composition and flow rate of the sheath liquid is very important. For instance, for the CE-MS analysis of catecholamines a number of organic solvent mixtures (methanol/water, acetonitrile/water, 2-propanol/water) were tested as sheath liquid [28]. The most optimal signal-to-noise ratio for the catecholamines were obtained with a sheath liquid of methanol/water (80:20 v/v) containing 0.5% acetic acid using a flow rate of 6 μL/min. The stability of CE-MS methods using sheath-liquid interfaces is often determined on the basis of intraday and interday variability of peak areas of compounds spiked into a body fluid. For instance, the relative standard deviation of peak areas of amino acids in urine was below the 10% for ten consecutive analyses [25].

23

Chapter 2

In the sheathless interface configuration, the CE voltage is directly applied to the CE buffer at the capillary outlet. This can be achieved by applying a metal coating to the end of a tapered separation capillary or by connecting a metal-coated, full metal or conductive polymeric sprayer tip to the CE outlet. Another way to make a closed circuit is by insertion of a metal micro-electrode through the capillary wall into the CE buffer end or by direct introduction of a micro-electrode into the end of a CE capillary [27]. CE-MS methods using a sheathless interface can offer an improved sensitivity, however, their potential in metabolomics have not been demonstrated yet [29]. The constituents of the BGE are an important aspect in CE–ESI-MS. Volatile buffers of low concentrations are preferred for sensitive ESI-MS, but these BGEs may not give optimum CE separations. However, it has been shown that satisfactory results can be obtained using volatile BGEs such as ammonium acetate or formic acid [30, 31]. The addition of organic modifiers, like methanol, to the BGE can improve the separation and MS detection performance of metabolites. For instance, in a CE-MS method for amino acids, a baseline separation of the isomers leucine and isoleucine was obtained using a BGE of 2 M formic acid (pH 1.8) with 20% methanol [32]. 2.3.2 Mass analyzers So far, the triple quadrupole (TQ) and ion trap (IT) have been the most commonly used mass analyzers in CE-MS for the analysis of low-molecular weight metabolites in biological samples [33]. These MS instruments, especially TQ, provide high sensitivity with the capability to obtain structural information on unknown compounds. However, a disadvantage of these mass analyzers, especially with respect to fast and highly efficient CE separations, is the relatively slow scanning process and low duty cycle. These MS instruments are not able to obtain sufficient data points across a very narrow CE peak to accurately define it. In addition, the mass resolution of these instruments is limited providing a selectivity that may not be sufficient to distinguish hundreds of metabolites in a single CE-MS run. Using a quadrupole mass analyzer,

Soga et al. limited the scan range to a window of 30 m/z for the detection of

metabolites in a bacterial extract [19]. To cover a range of m/z 70–1027, 33 runs were required resulting in an analysis time of approximately 16 hours. In general, TQ and IT mass analyzers are more suited for targeted metabolomic studies by CE-MS [15, 16]. TOF mass analyzers have an inherent ability for high spectral acquisition rates allowing a high number of data points (e.g. 10 spectra s-1) to be collected across a narrow CE peak. TOF-MS also provides a high mass resolution and high mass

24


accuracy with errors below 5 ppm [34, 35]. The high speed and high mass resolution of TOF-MS makes this instrument very suitable for full scan analyses in metabolomics. Indeed, when the same analysis of metabolites in a bacterial extract as mentioned above was performed with a CE-TOF-MS method, three runs were

sufficient to measure the metabolites in the m/z 70-1027 region [19]. TOF-MS instruments are increasingly being used in nontargeted metabolomics studies. The accurate mass obtained for (unknown) metabolites also can strongly aid in their provisional identification. Despite the increased resolution of TOF analyzers, potential interferences from solvent ions, adducts, environmental contaminants and compounds with the same nominal mass as the metabolites in the biological sample can still disturb the analysis. Therefore, efficient separations prior to MS analysis are of pivotal importance. These problems could also be circumvented by using FT-ICR MS which has a mass resolution of >500,000 and provides very accurate mass measurements with sub-ppm errors. However, accurate mass measurements with FT-ICR MS are relatively slow which can compromise the analysis of narrow CE peaks, and faster analysis may also lead to reduced sensitivity in FT-ICR MS [36]. 3. Sample pretreatment and matrix effects Sample pretreatment for metabolomic analysis depends on the goal of the study. For nontargeted metabolomic analysis, it is desirable that the biological sample is analyzed with minimal pretreatment to prevent the loss of metabolites. In order to allow proper analysis, it is essential that the low-molecular weight metabolites are separated from large molecules (proteins, lipids and large peptides) and salts. For nontargeted extraction of metabolites from bacteria several procedures have been investigated, such as extraction with cold methanol, hot methanol, ethanol, lysis with chloroform-methanol, and extractions with strong acids [37]. Extraction with cold methanol showed the highest recovery for the polar metabolites and more compounds were extracted compared to the other methods [37]. Extraction with cold methanol was applied to the comprehensive analysis of anionic metabolites from Bacillus subtilis cells by CE-MS [19]. For the extraction of metabolites from other matrices, such as plant tissue, various extraction procedures are often evaluated in order to find out which one would lead to the highest number of metabolites extracted [38]. For example, an extraction procedure of metabolites from maize flour was evaluated using CE-UV [39]. Extraction solvents covering a wide range of polarity were investigated for their potential to extract the relatively highest number of compounds in a reproducible way. Figure 3 shows CE-UV electropherograms obtained using water, ethanol/water (50:50, v/v) and methanol/water (50:50, v/v) as extractant. Best

25

Chapter 2

extraction was obtained with methanol/water (50:50, v/v) which was further used for the metabolomics study. For nontargeted metabolomic analysis of body fluids, such as plasma and serum, containing relatively large amounts of proteins deproteinization with an organic solvent is often applied to prevent the adsorption of proteins to the capillary wall during CE analysis [40].

A

B C Figure 3. CE-UV electropherograms of metabolites from Aristis maize obtained after extraction with (A) water, (B) ethanol/water (1:1, v/v), and (C) methanol/water (1:1, v/v). Letters (a, b, c, etc.) correspond to reproducible peaks observed in two different plant extracts obtained and analyzed under identical conditions. Reproduced from Ref. [39] with permission. Body fluids such as urine and CSF have often been analyzed with minimal sample pretreatment. For instance, CE-TOF-MS has been used for the analysis of amino acids in human urine with direct sample injection [32]. Amino acids and basic compounds were preconcentrated within the capillary using pH-mediated stacking. More recently, a CE-TOF-MS method using a capillary noncovalently coated with a bilayer of oppositely charged polymers has been developed for the fast metabolic

26


profiling of human CSF samples without any sample pretreatment [26]. A metabolic profile of positively charged compounds was obtained in less than 20 min. Figure 4 shows the overall results obtained for CSF from a patient with complex regional pain syndrome (CRPS) in comparison with the profile obtained for CSF from a healthy individual.

Figure 4. Metabolic profiling of CSF by CE–TOF-MS. Base peak electropherogram of (A) a CSF sample from a healthy individual and (B) a CSF sample from a patient with complex regional pain syndrome. Experimental conditions: BGE, 1M Formic acid (pH 1.8); sample injection, 90 mbar for 90 s; pre-injection, ammonium hydroxide (12.5%) at 50 mbar for 9 s; capillary, PB–PVS coated; scan range, 50–450 m/z. Reproduced from Ref. [26] with permission. Ion suppression (or enhancement) is an important aspect that should be considered when analyzing biological samples with MS detection. Ion suppression effects can be corrected for using isotopically labeled internal standards, however, within metabolomics this approach is less practical as the identity of many metabolites is unknown [41, 42]. Moreover, isotopically labeled compounds are quite expensive. The selection of a proper internal standard in order to improve the reliability of the CE-MS method is not trivial in metabolomics [15]. Endogenous metabolites stem from a wide variety of metabolic pathways and differ widely in structures and concentrations. Hence, the selection of one internal standard is not an option for nontargeted

27

Chapter 2

metabolomic analysis. The selection of several internal standards each representing a class of compounds (e.g. carboxylic acids) would be required. However, the inclusion of internal standards should not lead to co-migration with analytes or to (extra) ion-suppression. Another way to account for ion suppression, is to perform quantitative analyses using calibration curves of target metabolites constructed in the sample matrix. Still, the degree of ionization suppression may vary from sample to sample as the exact composition of the matrix may be subject and/or time dependent. Standard additions to each sample would solve this problem, but would lead to a very large number of analyses. In practice, pooled samples from healthy subjects, which provide a representative matrix, are often used to construct external calibration curves of target metabolites which are then used to calculate the concentrations of these metabolites in the biological samples [25]. For targeted metabolomic analysis, deproteinization of the biological sample is often followed by off-line solid-phase extraction (SPE) which is used for sample desalting and (selective) preconcentration of the target metabolites from the sample matrix [15]. Highly polar metabolites, however, do not show retention on commonly used reversed-phase SPE columns and elute simultaneously with the salts. There are only a limited number of SPE columns capable of effectively extracting (highly) polar metabolites. Generally, mixed-mode SPE sorbents, such as the Oasis HLB from Waters, are used for the simultaneous extraction of acids, neutrals and bases [43]. Extraction techniques for metabolomic analyses should ideally be applicable to various classes of metabolites ranging from anionic (e.g. organic acids), zwitter-ionic (e.g. amino acids), to cationic (e.g. amines) and neutral (e.g. steroids) analytes [44]. However, most SPE columns are quite selective and therefore the choice of an appropriate sample pretreatment technique is very critical in metabolomics. On the other hand, for targeted metabolomics, the use of off-line SPE for sample pretreatment can be very advantageous for the isolation and enrichment of the target metabolites from the biological sample. For instance, phenylboronic acid based SPE columns have been used for the selective preconcentration of nucleosides from human urine and subsequent nucleoside profiling by CE [45]. 4. Data treatment 4.1 Peak alignment and chemometrics Small migration time changes can lead to misinterpretation of data. In CE-MS analysis, migration time variation may result from adsorption of components to the capillary wall, thereby affecting the EOF. Migration time shifts can pose a greater

28


challenge when a large number of samples have to be analyzed and when the electrophoretic resolution of metabolite peaks is poor. For this reason, the development and the use of software tools for the alignment of successively recorded electropherograms has become important in CE-based metabolomics studies [11, 21, 46]. For instance, it was not possible to directly compare MEKC results obtained for rat urine samples due to shifts in migration times which occurred randomly across the obtained electropherograms [46]. Figure 5 shows that the comparison of electropherograms was possible after baseline correction, migration time alignment and normalization. Eventually, these data pretreatment tools combined with chemometrics allowed the distinction of urine from control rats and from rats with diabetics on the basis of MEKC profiles [46].

Figure 5. Comparison of profiles for control rat urines obtained by MEKC. (A) Before migration-time alignment (B) After migration-time alignment using a pair wise alignment procedure and normalization. Reproduced from Ref. [46] with permission. In CE-MS-based metabolomics studies software tools such as MetAlign, XCMS or MZMine have been explored for noise reduction, baseline correction and migration time alignment [25, 47-50]. For example, in a CE-MS study focused on the quantitative analysis of negatively charged metabolites in E. coli extracts the migration

29

Chapter 2

times of the metabolites varied from run to run and MZMine was used to correct for the migration time shifts [50]. Two internal standards and PIPES were added to every sample at a final concentration of 100 μM as reference compounds to correct for migration time shifts of the other compounds. This correction method was evaluated by analyzing 45 times a mixture of model compounds and 5 times two different biological samples. RSDs for migration times and peak areas were ca. 10% and 10–15%, respectively, for the model compounds. They were reduced to 0.07–0.13% for migration time and 5–7% for peak area after applying the correction procedure. Metabolomic analysis generates large and complex datasets. Therefore, chemometrics has become an essential part in CE-based metabolomics as it can provide interpretable models for complex data sets [23]. Multivariate data analysis such as principal component analysis (PCA) is often used to bring out hidden data structures. The main application of PCA is to reduce the dimensionality of data. In metabolomics studies, where there are typically thousands of variables, it can be extremely useful to present the variables into a smaller set of principal components. This makes most types of plots easier to interpret and can help to visualize structure in the data. PCA is often followed by a supervised analysis technique such as Partial Least Squares Discriminant Analysis (PLS-DA) or that can aid in obtaining a list of potential biomarkers which are statistically significant and which separate one class from another [51]. 4.2 Metabolite identification Following CE-MS analysis, metabolite identification or confirmation is based on

migration time and m/z value using standards and databases, such as the Human Metabolome Database, for comparison [52]. In a first stage, molecular weight databases can be used for the provisional identification of unknown compounds. For this purpose, a mass spectrometer with a high mass resolution such as TOF-MS is very practical as the accurate mass measurements obtained for unknown compounds considerably reduces the list of possible candidates delivered by a database. For example, CE-TOF-MS has been used for the metabolic profiling of transgenic versus conventional soybean extracts [38]. The identification of metabolites in the soybean

extracts was performed by using m/z values combined with the migration times of the compounds. The assignment was confirmed by expected soybean composition and metabolic pathways described in the literature. As TOF-MS was used for detection, a number of possible elemental compositions was obtained from the accurate mass of the metabolite peaks. These elemental compositions were matched against available databases using the deduced molecular formula as a search criterion [53]. Additional information about physico-chemical properties of the compounds was provided by

30


the migration times. For instance, in the CE-TOF-MS method used for this study the migration time of highly charged small molecules was longer than that for molecules of smaller charge and greater size. Identification of unknown compounds can be achieved by performing further CE-MS/MS experiments using an IT, TQ or quadrupole TOF-MS to provide fragment spectra [54]. In addition, these experiments can be performed to confirm the identification of results obtained by databases. For example, to increase the confidence in identification of analytes from E. coli extracts, MS/MS experiments on an IT was performed to collect fragment spectra [55]. These fragmentation patterns allowed the assignment of peaks with multiple possible candidates as derived from the database based on molecular mass only. The procedures for metabolite identification outlined above have also been used in other metabolomics studies performed by CE-MS [25, 32, 39, 50]. 5. Applications An overview of metabolomics studies performed by CE-MS and published from January 2000 to July 2008 is given in Table 1. The table provides information about the type of biological sample and compounds analyzed, the BGE, sample pretreatment procedure, the MS analyzer employed, limit of detection (LOD, when provided by the authors), and type of capillary coating used. Representative metabolomic applications classified according to the type of sample matrix are discussed below. 5.1 Bacterial extracts Metabolomic analysis of bacterial extracts can provide insight into processes as energy uptake and bacterial growth. Metabolic pathways, such as the central carbon metabolism, play an important role in these processes and the metabolites involved in these pathways are often highly polar and charged compounds. Hence, CE-MS is a well suited method for the analysis of these compounds. Indeed, over the last few years, CE-MS has been used for the metabolic profiling of various bacterial extracts. For instance, CE coupled to an ion trap MS was used for the analysis of amino acids in broth solutions from E. coli to study the nutrient uptake behavior [56]. The broth solution was diluted 20-fold in ammonium acetate (pH 7.0) prior to CE-MS analysis in order to ensure that metabolic concentrations were within the linear range for quantification, which was between the 0.1 and 100 μM for most amino acids. A 50-fold improvement in concentration sensitivity for cationic metabolites was obtained using online sample preconcentration based on tITP and pH-mediated stacking.

31

Chapter 2

32

Table 1 Overview of CE-MS applications in metabolomics Sample Matrix

Compounds BGE Sample pretreatment

MS Analyzer

LODa) Remarks Ref

Bacterial extract

Amino acids 1.6 M formic acid in MeOH/H2O (2:8, v/v)

Cold methanol extraction of metabolites. Salts removed by off-line SPE

FT-ICR 0.1 μM pH-mediated stacking and tITP for preconcentration

[36]

Bacterial extract

Anionic metabolites

50 mM ammonium acetate (pH 8.7)

Extraction with 80% methanol in water

TOF 0.2-2 μM

Cationic polymer PolyE-323 for EOF reversal

[50]

Bacterial extract

Cationic metabolites

1 M formic acid (pH 1.8)

20-fold dilution with ammonium acetate (pH 7.0)

IT Low nM range

tITP combined with pH-mediated stacking for preconcentration

[56]

Bacterial extract

Anionic metabolites

20 mM ammonium acetate/2-propanol (8:2, v/v)

Cold methanol extraction

IT 0.02-2.5 μM

Sheathless interface [55]

Bacterial extract

Sugar nucleotides

30 mM morpholine/formate (pH 9.0)

Ice-cold ethanol extraction and off-line anion-exchange SPE

TQ 0.2-3.8 nM

On-line sample stacking by tITP

[57]

Bacterial extract

Sugar phosphates

30 mM morpholine/formate (pH 9.0)

Ice-cold ethanol extraction

TQ-linear IT

2.5-10 μM

[69]

Bacterial extract

Anionic metabolites


Methanol extration IT 0.3-6.7 μM

Polybrene-dextran sulfate-polybrene coated capillary for EOF reversal

[30]

Bacterial extract

Cationic and anionic metabolites

50 mM ammonium acetate (pH 9.0); 1 M formic acid (pH 1.8)

Methanol extraction TQ Low μM range

Polybrene-dextran sulfate-polybrene coated capillary for anionic metabolites

[19]

Bacterial extract

Citrate isomers and (di)nucleotides


Extraction with chloroform and water

IT 0.4-3.7 μM

Capillary coated with poly(dimethyl)siloxane

[70]

Bacterial extract

Intracellular metabolites


Extraction with water, chloroform and methanol. Methanol extract filtered with 5-kDa filter and lyophilized

TOF ns [71]


Sample Matrix


MS Analyzer

LODa) Remarks Ref

Bacterial extract

Nucleotides 50 mM ammonium acetate (pH 7.5)

Extraction with water, chloroform and methanol. Methanol extract filtered with 5-kDa filter and lyophilized

IT 0.5-1.7 μM

Capillary treated with phosphate to prevent adsorption of multi-phosphorylated compounds

[58]

Bacterial extract

Glycolytic metabolites


Methanol extraction IT 0.3-6.5 μM


[72]

In vitro biochemical assay



Ultrafiltration to remove proteins

Quad- rupole

0.3-11 μM


[79]

Yeast extract

Sulfur-related metabolites


Extraction with cold chloroform and cold methanol

IT ns Matrix effects studied [73]

Yeast extract

Anionic metabolites

150 mM ammonium hydrogen carbonate/formate (pH 6.0)

Extraction with cold chloroform and cold methanol

IT ns Pressure-assisted CE using PEEK

[80]

Plant extract


50 mM ammonium acetate (pH 9.0); 1 M formic acid (pH 1.8)

Extraction with methanol, chloroform and water

IT Low μM range


[59]

Plant extract

Anionic metabolites


Extraction with methanol, chloroform and water

TQ-linear IT

0.1-8.8 μM

Sulfonated coated capillary

[74]

Plant extract

Flavonoids and phenolic compounds

10 mM ammonium carbonate (pH 9.25)/2-propanol (95:5, v/v)

Extraction with ethyl acetate and butanol and off-line SPE

IT ns [60]

Plant extract


50 mM ammonium hydrogen carbonate (pH 9.0)

Extraction with methanol/water (8:2, v/v)

TOF ns [38]

Plant extract


5% formic acid (pH 1.9)

Extraction with methanol/water (1:1, v/v)

TOF ns [39]

33

Chapter 2

Sample Matrix


MS Analyzer

LODa) Remarks Ref

Human urine

Organic acids, small peptides, amino acids and nucleosides

50 mM acetic acid + 50 mM formic acid (pH 2.5)

Lyophilization IT 1-10 μg/mL

[61]

Human urine

Steroids 20 mM ammonium acetate with 36.8 mM taurocholate and 29.5 mM SDS (pH 9.68)

Off-line mixed-mode SPE

TQ 0.5-1 μg/mL

Partial filling MEKC [75]

Human urine

Catecholamines and metanephrines

30 mM ammonium acetate (pH 4.2-4.5)

Off-line cation-exchange SPE

TOF 0.1-0.3 μM

Polyvinylalcohol coated capillary

[24]

Human urine

Anionic metabolites


Direct sample injection

TQ ns [76]

Human urine

Amino acids 1 M formic acid (pH 1.8)

Dilution with BGE (1:1, v/v)

TOF 85-280 nM

pH-mediated stacking for preconcentration; Polybrene-poly(vinyl-sulfonate) coated capillary

[25]

Human urine

Amino acids 2 M formic acid and 20% methanol (pH 1.8)


TOF Mid nM range

pH-mediated stacking for preconcentration

[32]

Human urine


Urine diluted with water (1:5, v/v)

TQ 0.1-14 μM

[77]

Human urine

Positively and negatively charged metabolites

20 mM formic acid/ammonium formate (pH 3.0); 20 mM ammonium acetate (pH 9.0)


TQ ns Cationic polymer PolyE-323 for anions

[11]

Human CSF

Amino acids 5 mM ammonium acetate (pH 9.7) + 5% acetonitrile

CSF diluted with water (1:5, v/v)

TOF 20-67 nM

Capillary coated with 1-(4-iodobutyl)4-aza-1-azoniabicyclo[2,2,2]octane iodide

[63]

Human CSF


Dilution with BGE (1:1)

TOF 20-215 nM

Polybrene-poly(vinyl-sulfonate) coated capillary

[26]

34


Sample Matrix


MS Analyzer

LODa) Remarks Ref

Human plasma

Short-chain carnitines

200 mM ammonium formate (pH 2.5)

Deproteinization with cold acetonitrile

IT 0.25-1 μM

[40]

Human hepatoma cell line Hep G2

Nucleosides and nucleotides

25 mM ammonium acetate (pH 10)

Extraction with 60% methanol

Quad- rupole

100-200 nM

[78]

Human erythro-cytes

Intracellular nucleotides

20 mM acetate buffer (pH 4.4)

Erythrocytes incubated with 5-amino-4-imidazole-carboxamide riboside

Quad- rupole

ns [62]

Mouse liver extracts


50 mM ammonium acetate (pH 9.0) and 1 M formic acid (pH 1.8)

Extraction with water and chloroform

TOF 0.1-1.7 μM

Polybrene-dextran sulfate-polybrene coated capillary

[54]

aLOD = limit of detection (S/N = 3); ns, not specified in paper. Nanomolar concentrations of amino acids could be detected with this approach. A computer-simulated algorithm was developed to predict the CE migration profile of charged analytes. Physico-chemical properties such as electrophoretic mobility and acid dissociation constants were used for the simulation. Good agreement was obtained between experimental and simulated relative migration times, with an average error of <2%. A negative mode sheathless CE-MS method was used for the analysis of anionic metabolites in extracts of E. coli [55]. The use of a sheathless interface resulted in a 10-fold improvement of concentration sensitivity compared to a sheath-flow CE-MS method, with LODs in the low nM range. A number of 118 metabolites were detected in full scan mode illustrating the potential of this method for metabolic profiling of biological samples. CE-MS has been used for the identification of sugar nucleotides in wild type and isogenic mutants from the bacterial pathogen Campylobacter jejuni [57]. A precursor ion scanning method was developed for the selective detection of sugar nucleotides. The concentration sensitivity was improved by on-line sample stacking using morpholine (35 mM, pH 9.0) as BGE and HEPES (5 mM; pH 9.0) as sample solvent. Sample-injection volumes were increased from ca. 70 nL to 200 nL. Stacking of sugar nucleotides was achieved as the mobilities in HEPES are higher than in morpholine, resulting in improved LODs ranging from 0.2 to 3.8 nM. The method was used to screen differences between the intracellular pool of sugar nucleotides of parent and isogenic mutants of Campylobacter jejuni.

35

Chapter 2

CE-MS has also been used for large-scale metabolite analysis of Bacillus subtilis cells for changes during sporulation [19]. Distinct CZE-ESI-MS methods have been described for anionic and cationic metabolites. Separation of cationic metabolites was performed with a bare fused silica capillary using 1 M formic acid as BGE (pH 1.8). Separation of anionic metabolites was performed on a cationic polymer-coated capillary using 50 mM ammonium acetate (pH 8.5) as BGE. A number of 1692 metabolites were measured of which 150 were identified. Many metabolites in the glycolytic, pentose phospate, and tricarboxylic acid pathway appeared to be significantly decreased in the early stage of sporulation (Figure 6). The levels of cis-aconitate and isocitrate were increased during the late growth phase. These findings were in good agreement with previously reported studies demonstrating the potential of CE-MS for the analysis of metabolites in bacterial cells that are involved in biological processes such as bacterial growth.

Figure 6. Metabolic maps of Bacillus subtilis cells during sporulation. Metabolites were measured by CE-MS. (A) Changes in metabolite levels during the late growth phase. (B) Changes in metabolite levels in the early growth phase. Magenta and red boxes indicate metabolites whose levels increased 2 to 10-fold and more than 10-fold, respectively. Blue and indigo boxes indicate metabolites whose level was decreased to 0.1-05 and less than 0.1 of the original level, respectively. White boxes indicate metabolites whose levels were unchanged. Reproduced from Ref. [19] with permission.

36


A reproducible and quantitative CE-MS method for the analysis of nucleotides was developed using phosphate ions to mask the silanol groups of bare fused silica capillaries before the separation [58]. The masking of the silanol groups was necessary to prevent the interaction of multivalent nucleotides with the capillary wall. Air pressure was applied to the inlet to increase the flow towards the cathode. The CE-MS was used to analyze nucleotides, nicotinamide adenine dinucleotides, and CoA compounds simultaneously in E. coli extracts. 5.2 Plant extracts CE-MS was applied to the analysis of metabolites in the rice plant Oryza sativa [59]. A number of 88 metabolites involved in the glycolysis pathway have been determined in the rice leaf extracts. For the metabolomic study, the metabolites of different energy pathways were grouped in neutral (sugars) and charged compounds. The charged compounds were analyzed with CE-MS and the sugars were analyzed with CZE using indirect UV detection. Figure 7 shows the result of a CZE-ESI-MS analysis of amino acids, amines and purine bases in the rice-leaf extract.

Figure 7. Extracted ion electropherograms of metabolites in rice leaf extracts. Peak identification: 1, glycine; 2, 1,4-butanediamine; 3, alanine: 4, -aminobutyric acid; 5, serine; 6, proline; 7, valine; 8, threonine; 9, iso-leucine; 10, leucine; 11, ornithine; 12, asparagine; 13, aspartic acid; 14, tyramine; 15, anthranilic acid; 16, spermidine; 17, lysine; 18, glutamine; 19, glutamic acid; 20, histidine; 21, phenylalanine; 22, arginine; 23, citrulline; 24, tyrosine; 25, methionine sulphone (IS); 26, adenosine; 27, glutathione. Reproduced from Ref. [59] with permission.

37

Chapter 2

For the separation, a bare fused-silica capillary was used with 1 M formic acid (pH 1.8) as BGE. The LODs obtained with the CE-MS method were 0.1–1 μM. A CE-MS method has been developed for the analysis of flavonoids and phenolic compounds in an extract of the Genista tenera plant [60]. 26 plant metabolites were separated within an analysis time of 10 min. This was faster than an LC-MS method, which required an analysis time of 100 min for only 5 compounds. 5.3 Urine CE-MS has repeatedly been used for the analysis of a subset of endogenous metabolites in human urine. There are only few reported CE-MS studies which focused on global metabolite profiling of urine. A CE-MS method for the profiling of human urine was developed using low pH conditions [61]. Positive and negative ionization modes were used in order to collect complementary information. While positive ionization showed better reproducibility, negative mode showed metabolites that could not be detected in positive ionization. A CE-TOF-MS method has been developed for the analysis of catecholamines and metanephrines in human urine [24]. The inner capillary wall was coated with PVA to suppress the EOF and to improve the separation of the compounds. Catecholamines and metanephrines were extracted and preconcentrated from urine with offline SPE using cation-exchange sorbents. Using electrokinetic injection, LODs ranged from 0.1 to 0.3 μM for the compounds. Amino acid profiles in human urine have been determined by CE-TOF-MS without any sample pretreatment [36]. LODs were in the low nM-range through the use of pH-mediated stacking. The method was used for the metabolic profiling of urine samples from healthy subjects and from patients with osteoarthritis. PCA analysis showed that urine samples from healthy subjects fall in a distinct group and urine samples from patients with osteoarthritis grouped into another. Examination of the loadings plot revealed that one of the compounds responsible for grouping of the

samples was histidine (m/z 156.09). A CE-TOF-MS method using capillaries noncovalently coated with a bilayer of Polybrene (PB) and poly(vinyl sulfonate) (PVS) was used for the of amino acids in human urine [25]. The PB-PVS coating provided a considerable EOF at low pH, thus facilitating the fast separation of amino acids using a BGE of 1 M formic acid (pH 1.8). The PB–PVS coating proved to be very consistent yielding stable CE-MS patterns of amino acids in urine with favourable migration time repeatability (RSDs <1.5%). The use of pH-mediated stacking resulted in LODs between the 20 and 300 nM for amino acids. Special attention was paid to the influence of matrix effects on the quantification of amino acids. The magnitude of ion suppression by the matrix

38


was similar for different urine samples. The CE-TOF-MS method was used for the analysis of urine samples of patients with urinary tract infection (UTI). Concentrations of amino acids were determined and compared with concentrations in urine of healthy

controls. PLS–DA of the CE-TOF-MS dataset in the 50–450 m/z region showed a distinctive grouping of the UTI samples and the control samples (Figure 8). Examination of the loadings plot revealed a number of compounds, including phenylalanine, to be responsible for grouping of the samples. This study demonstrated that metabolic profiling of urine is possible with minimal sample pretreatment, i.e. with minimal loss of analytes as required for metabolic profiling studies. In addition, the stability of migration times was guaranteed by the use of the capillary coating, which is important for the direct comparison of electropherograms of different urine samples.

Figure 8. PLS–DA score plot of urine samples analyzed by CE-MS from healthy controls (diamonds) and patients with urinary tract infection (circles) using a two-class model. Reproduced from Ref. [25] with permission.

5.4 Plasma and serum A CE-MS method has been developed for the fast analysis of short-chain carnitines in human plasma [40]. Plasma samples were deproteinized with cold acetonitrile, and despite the resulting sample dilution, the peak intensities of the analytes were not

39

Chapter 2

significantly decreased due to an acetonitrile-induced stacking effect. The CE-MS analysis of the short-chain carnitines in the deproteinized plasma samples took less than 10 min, with high sensitivity and specificity, as the mass spectrometer was used in product ion scan mode. LODs varied from 0.25 to 1 μM. The within-day and between-day repeatability and accuracy were within 15% of relative standard deviation (RSD) for low and high concentrations of short-chain carnitines in plasma. A CE-TOF-MS method was used for the metabolic profiling of serum and liver extracts of mice before and after acetaminophen-induced hepatotoxicity [54]. Changes in metabolite profiles and levels were measured (1859 peaks detected) and it was found that serum ophthalmic acid was an indicator of hepatic glutathione depletion, and therefore a biomarker for oxidative stress. Recently, organic acids have been analyzed in human plasma without any sample pretreatment by CE-UV using a triple layer capillary coating [26]. RSD values for migration time and peak area of organic acids was <2.5% and <4.5%, respectively. The good stability of this system provides high potential for the metabolic profiling of body fluids, especially, when the CE method will be coupled to MS. 5.5 CSF A CE-TOF-MS method has been used for the separation and detection of tryptophan metabolites of the kynurenic pathway [63]. The inner capillary wall was modified with 1-(4-iodobutyl)4-aza-1-azoniabicyclo[2,2,2]octane iodide, also named M7C4I, to deactivate the fused silica wall and to generate a stable reversed electro-osmotic flow for reproducible migration times. Using the M7C4I coated capillary, the analysis of tryptophan, kynurenine, kynurenic acid and metabolites spiked into 5-fold diluted CSF was performed in less than 5 min. 6. Conclusions and future outlook The use of CE-MS for metabolomics has increased considerably over the last few years. The studies reported so far in the literature demonstrate that CE combined with MS has the potential to provide informative metabolite profiles of biological samples. As CE is particularly useful for the separation of highly polar and charged metabolites, CE-MS offers complementary information on metabolic compositions of biological samples with respect to LC-MS. The use of both techniques would provide a broad coverage of the metabolome. ESI is the commonly used ionization technique in CE-MS, and this method is especially suitable for the analysis of (highly) polar compounds. Recently, CE has been

40


coupled to APPI-MS and APCI-MS for the detection of relatively non-polar compounds [81, 82]. Therefore, the applicability of CE-MS in metabolomics may be further broadened using APPI and APCI as additional ionization techniques. The lower concentration sensitivity of CE-MS compared to LC-MS, can be partly compensated for by preconcentration of biological samples within the capillary prior to separation, using techniques as tITP and pH-mediated stacking. Various examples have been described in this review indicating a trend of the use of in-capillary preconcentration strategies in CE-based metabolomics. Nonetheless, these in-capillary preconcentration techniques require careful optimization experiments in order to improve the concentration sensitivity with several orders of magnitude. For instance, in pH-mediated stacking, the pH of the sample and the BGE should be carefully selected to take full advantage of mobility changes of analytes, which depends on the pKa values for weakly acidic or basic compounds. Also the length of the sample injection plug has an influence on pH-mediated stacking, as demonstrated for the analysis of amino acids in human urine [25]. In general, the use of in-capillary preconcentration techniques is very attractive for CE-MS-based metabolomics studies, but careful optimization is required to use these techniques for the preconcentration of biological samples. TOF-MS has the most favourable characteristics for metabolomic studies with CE [64]. The fast scan rate and high mass resolution and accuracy facilitates selective detection and identification of metabolites, although, a general database for metabolite identification using CE-MS methods is not yet available. The large number of endogenous metabolites in a biological sample makes the use of a multidimensional separation technique very attractive to separate as many metabolites as possible [65]. A two-dimensional LC-CE system has been recently used for the analysis of metabolites in an extract of Bacillus Subtilis [66]. A monolithic micro-LC column was used as the first dimension, from which the eluted fractions were analyzed by CE in the second dimension. In-capillary preconcentration based on pH-mediated stacking and sweeping were selectively used to interface the two dimensions. Metabolites were separated according to their hydrophobicity by micro-LC with gradient mode. The early-eluting fractions were separated by CZE-UV and the late-eluting fractions were separated by MEKC-UV using sweeping. The middle fractions were analyzed by both CE modes. With this multidimensional system some important metabolites were identified which were not detected with the use of one separation technique. The coupling of this system to MS has great potential to enable the analysis and identification of more metabolites. Another promising development for the analysis of biological samples is CE on a microchip [67]. The small injection volumes, high electric fields, and short separation

41

Chapter 2

lengths can produce analysis times of seconds. CE microchips using dynamic or permanent coatings gain more interest because this can reduce analyte-wall interactions and it can give a more rapidly and efficient separation due to adjustment of the electro-osmotic flow. Microchip CE with laser induced fluorescence detection was recently used for the measurement of multiple inflammatory biomarkers in human patient samples using microdissected human skin biopsies as a model [68]. Electrophoretic separation of 12 peptides was obtained in less than 2.2 min with absolute LOD values around the 1 pg. The microchip CE method showed different peptide profiles for microdissected tissue samples from patients with mild and chronic skin lesions (Figure 9). Microchip CE systems provide very fast analysis times, and therefore, such systems have good potential for targeted metabolomic studies as the separation efficiency may not be high enough to separate many components in a complex mixture. Figure 9. Electropherograms obtained by microchip CE showing peptide profiles of a patient with chronic skin lesion (A) and a patient with a mild skin lesion. Peaks: 1, transforming growth factor-; 2, interleukin-6; 3, interleukin-1; 4, interferon-; 5, macrophage inflammatory protein-1; 6, macrophage chemoattractant protein-1; 7, tumor necrosis factor-; 8, calcitonin gene-related peptide; 9, neuropeptide Y; 10, interleukin-8; 11, vasoactive intestinal peptide; 12, substance P. Reproduced from Ref. [68] with permission.

42


References [1] Fiehn, O., Plant. Mol. Biol. 2002, 48, 155-171. [2] Van der Greef, J., Stroobant, P., Van der Heijden, R., Curr. Opin. Chem. Biol.

2004, 8, 559-565. [3] Nicholson, J.K., Wilson, I.D., Nature Rev. Drug Disc. 2003, 2, 668-676. [4] Wikoff, W.R., Gangoiti, J.A., Barshop, B.A., Siuzdak, G., Clin. Chem. 2007,

53, 2169-2176. [5] Fiehn, O., Kind, T., Methods Mol. Biol. 2007, 358, 3-17. [6] Kopka, J., J. Biotechnol. 2006, 124, 312-322. [7] Beckonert, O., Keun, H.C., Ebbels, T.M., Bundy, J. et al., Nat. Protoc. 2007,

2, 2692-2703. [8] Wilson, I.D., Plumb, R., Granger, J., Major, H., Williams, R., Lenz, E.M.,

J. Chromatogr. B 2005, 817, 67-76. [9] Lenz, E.M., Wilson, I.D., J. Proteome Res. 2007, 6, 443-458. [10] Tolstikov, V.V., Lommen, A., Nakanishi, K., Tanaka, N., Fiehn, O.,

Anal. Chem. 2003, 75, 6737-6740. [11] Ullsten, S., Danielsson, R., Bäckström, D., Sjöberg, P., Bergquist, J.,

J. Chromatogr. A 2006, 1117, 87-93. [12] t’Kindt, R., Alaerts, G., Vander Heyden, Y., Deforce, D., Van Bocxlaer, J.,

J. Sep. Sci. 2007, 30, 2002-2011. [13] Coulier, L. Bas, R., Jespersen, S., Verheij, E. et al., Anal. Chem. 2006, 78,

6573-6582. [14] Tolstikov, V.V., Fiehn, O., Anal. Biochem. 2002, 301, 298-307. [15] Ramautar, R., Demirci, A., De Jong, G.J., Trends Anal. Chem. 2006, 25, 455-

466. [16] Monton, M.R.N., Soga, T., J. Chromatogr. A 2007, 1168, 237-246. [17] Baena, B., Garcia-Martinez, D., Barbas, C., J. Chromatogr. A 2004, 1051, 199-

205. [18] Garcia, A., Barbas, C., Aguilar, R., Castro, M., Clin. Chem. 1998, 44, 1905-

1911. [19] Soga, T., Ohashi, Y., Ueno, Y., Naraoka, H., Tomita, M., Nishioka, T.,

J. Proteome Res. 2003, 2, 488-494. [20] Vallejo, M., Angulo, S., García-Martínez, D., García, A., Barbas, C.,

J. Chromatogr. A. 2008, 1187, 267-274. [21] Guillo, Ch., Barlow, D., Perrett, D., Hanna-Brown, M., J. Chromatogr. A

2004, 1027, 203-212. [22] Xie, G., Su, M., Li, P., Gu, X. et al., Electrophoresis 2007, 28, 4459-4468.

43

Chapter 2

[23] Johannesson, N., Olsson, L., Bäckström, D., Wetterhall, M. et al., Electrophoresis 2007, 28, 1435-1443.

[24] Peterson, Z.D., Collins, D.C., Bowerbank, C.R., Lee, M.L., Graves, S.W., J. Chromatogr. B 2002, 776, 221-229.

[25] Ramautar, R., Mayboroda, O.A., Derks, R.J.E., Van Nieuwkoop, C. et al., Electrophoresis 2008, 29, 2714-2722.

[26] Ramautar, R., Mayboroda, O.A., Deelder, A.M. Somsen, G.W., De Jong, G.J., J. Chromatogr. B 2008, 871, 370-374.

[27] Simpson, D.C., Smith, R.D., Electrophoresis 2005, 26, 1291-1305. [28] Vuorensola, K., Kokkonen, J., Sirén, H., Ketola, R.A., Electrophoresis 2001,

22, 4347-4354. [29] Schultz, C. L., Moini, M., Anal. Chem. 2003, 75, 1508-1513. [30] Soga, T., Ueno, Y., Naraoka, H., Ohashi, Y., Tomita, M., Nishioka, T., Anal.

Chem. 2002, 74, 2233-2239. [31] Neusüss, C., Pelzing, M., Macht, M., Electrophoresis 2002, 18, 3149-3159. [32] Mayboroda, O.A., Neusüß, C., Pelzing, M., Zurek, G. et al., J. Chromatogr. A

2007, 1159, 149-153. [33] Soga, T., Methods Mol. Biol. 2007, 358, 129-37. [34] Bajad, S., Shulaev, V., Trends Anal. Chem. 2007, 26, 625-636. [35] Lacorte, S., Fernandez-Albaz, A.R., Mass Spectrom. Rev. 2006, 25, 866-880. [36] Baidoo, E.E.K, Benke, P.I., Neusüss, C., Pelzing, M. et al., Anal. Chem. 2008,

80, 3112-3122. [37] Maharjan, R.P., Ferenci, T., Anal. Biochem. 2003, 313, 145-154. [38] Garcia-Villalba, R., Leon, C., Dinelli, G., Segurra-Carretero, A. et al.,

J. Chromatogr. A 2008, 1195, 164-173. [39] Levandi, T., Leon, C., Kaljurand, M., Garcia-Canas, V., Cifuentes, A.,

Anal. Chem. 2008, 80, 6329-6335. [40] Desiderio, C., De Rossi, A., Inzitari, R., Mancinelli, A., et al.,

Anal. Bioanal. Chem. 2008, 390, 1637-1644. [41] Martin-Girardeau, A., Renou-Gonnord, M.F., J. Chromatogr. B 2000, 742,

163-171. [42] Peterson, Z.D., Lee, M.L., Graves, S.W., J. Chromatogr. B 2004, 810, 101-

110. [43] Weigel, S., Kallenborn, R., Hühnerfuss, H., J. Chromatogr. A 2004, 1023,

183-195. [44] Simonet, B.M., Rios, A., Valcarel, M., Trends Anal. Chem. 2003, 22, 605-614. [45] La, S., Cho, j., Kim, j., Kim, K., Anal. Chim. Acta 2003, 486, 171-182.

44


[46] Barbas, C., Vallejo, M., Garcia, A., Barlow, D., Hanna-Brown, M., J. Pharm. Biomed. Anal. 2008, 47, 388-98.

[47] Smith, C.A., Want, E.J., O’Maille, G., Abagyan, R., Siuzdak, G., Anal. Chem. 2006, 78, 779-787.

[48] De Vos, R.C.H., Moco, S., Lommen, A., Keurentjes, J.J.B. et al., Nat. Protoc. 2007, 2, 778-791.

[49] Katajamaa, M., Orešič, M., J. Chromatogr. A 2007, 1158, 318-328. [50] Timischl, B., Dettmer, K., Kaspar, H., Thieme, M., et al., Electrophoresis

2008, 29, 2203-2214. [51] Westerhuis, J.A., Hoefsloot, H.C.J., Smit, S., Vis, D.J., et al., Metabolomics

2008, 4, 81-89. [52] Wishart, D.S., Tzur, D., Knox, C. Eisner, R., et al. Nucleic Acids Res. 2007,

35, D521-526. [53] Ibáñez, M., Sancho, J.V., Pozo, O.J., Niessen, W., Hernández, F.,

Rapid Commun. Mass Spectrom. 2005, 19, 169-178. [54] Soga, T., Baran. R., Suematsu, M., Ueno, Y., et al., J. Biol. Chem. 2006, 281,

16768- 16776. [55] Edwards, J.L., Chisolm, C.N., Shackman, J.G., Kennedy, R.T.,

J. Chromatogr. A 2006, 1106, 80-88. [56] Lee, R., Ptolemy, A.S., Niewczas, L., Britz-McKibbin, P., Anal. Chem. 2007,

79, 403-415. [57] Soo, E.C., Aubry, A.J., Logan, S.M., Guerry, P. et al., Anal. Chem. 2004, 76,

619-626. [58] Soga, T., Ishikawa, T., Igarashi, S., Sugawara, K. et al., J. Chromatogr. A 2007,

1159, 125-133. [59] Sato, S., Soga, T., Nishioka, T., Tomita, M., Plant J. 2004, 40, 151-163. [60] Edwards, E.L., Rodrigues, J.A., Ferreira, J., Goodall, D.M. et al.,

Electrophoresis 2006, 27, 2164-2170. [61] Benavente, F., Van der Heiden, R., Tjaden, U.R., Van der Greef, J.,

Hankemeier, T., Electrophoresis 2006, 27, 4570-4584. [62] Friedecky, D., Bednar, P., Prochazka, M., Adam, T.,

Nucleosides Nucleotides Nucleic acids 2006, 25, 1233-1236. [63] Arvidsson, B., Johannesson, N., Citterio, A., Righetti, P.G., Bergquist, J.,

J. Chromatogr. A 2007, 1159, 154-158. [64] Kolch, W., Neusüß, C., Pelzing, M., Mischak, H., Mass Spectrom. Rev. 2005,

24, 959-977. [65] Giddings, J.C.J., HRC & CC J. High Resolut. Chromatogr. Chromatogr.

Commun. 1987, 10, 319-323.

45

Chapter 2

46

[66] Jia, L., Liu, B., Terabe, S., Nishioka, T., Anal. Chem. 2004, 76, 1419-1428. [67] Dolnik, V., Liu, S., Jovanovich, S., Electrophoresis 2000, 21, 41-54. [68] Philips, T.M., Wellner, E.F., Electrophoresis 2007, 28, 3041-3048. [69] Hui, J.P.M., Yang, J., Thorson, J.S., Soo, E.C., Chembiochem 2007, 8, 1180-

1188. [70] Soga, T., Ueno, Y., Naraoka, H., Matsuda, K. et al., Anal. Chem. 2002, 74,

6224-6229. [71] Toya, Y., Ishii, N., Hirasawa, T., Naba, M. et al., J. Chromatogr. A 2007,

1159, 134-141. [72] Itoh, A., Ohashi, Y., Soga, T., Mori, H. et al., Electrophoresis 2004, 25, 1996-

2002. [73] Tanaka, Y., Higashi, T., Rakwal, R., Wakida, S. et al., J. Pharm. Biomed. Anal.

2007,44, 608-613. [74] Harada, K., Ohyama, Y., Tabushi, T., Kobayashi, A. et al., J. Biosci. Bioeng.

2008, 105, 249-260. [75] Sirén, H., Seppänen-Laakso, T., Orešič, M., J. Chromatogr. B 2008, in press. [76] Elgstoen, K.B.P., Zhao, J.Y., Anacleto, J.F., Jellum, E., J. Chromatogr. A

2001, 914, 265-275. [77] Soga, T., Kakazu, Y., Robert, M., Tomita, M., Nishiokoa, T., Electrophoresis

2004, 25, 1964-1972. [78] Liu, C.C., Huang, J.S., Tyrrell, D.L.J., Dovichi, N.J., Electrophoresis 2005, 26,

1424-1431. [79] Saito, N., Robert, M., Kitamura, S., Baran. R. et al., J. Proteome Res. 2006, 5,

1979-1987. [80] Tanaka, Y., Higashi, T., Rakwal, R., Wakida, S. Iwahashi, H., Electrophoresis

2008, 29, 2016-2023. [81] Mol, R., De Jong, G.J., Somsen, G.W., Electrophoresis 2005, 26, 146-154. [82] Tanaka, Y., Otsuka, K., Terabe, S., J. Pharm. Biomed. Anal. 2003, 30, 1889-

1895.

Chapter 3

Direct Sample Injection for Capillary Electrophoretic

Determination of Organic Acids in Cerebrospinal Fluid

R. Ramautar, G.W. Somsen, G.J. de Jong,

Analytical and Bioanalytical Chemistry 2007, 387, 293-301

Capillary Electrophoretic Determination of Organic Acids in Cerebrospinal Fluid

Abstract Organic acids in cerebrospinal fluid (CSF) are potential diagnostic markers for neurological diseases and metabolic disorders. A capillary electrophoretic (CE) method for the direct analysis, i.e., without any sample preparation, of organic acids in CSF was developed. A capillary coating consisting of a triple layer of charged polymers (polybrene-dextran sulfate-polybrene) was used in combination with a negative separation voltage providing fast and efficient analysis of acidic compounds. Separation conditions, such as background electrolyte (BGE) concentration and pH were optimized, and the influence of sample constituents like proteins and salts was systematically studied using a set of test compounds. With injection volumes of 44 nL plate numbers of up to ca. 150,000 were obtained with a BGE of 200 mM sodium phosphate (pH 6.0). It appeared that high sodium chloride concentrations in the sample hardly affected the peak width and shape of the organic acids, most probably due to transient-isotachophoresis effects occurring in the sample zone. Adverse effects of CSF proteins, which frequently compromise the CE performance, could be effectively minimized by the triple layer coating in combination with rinses of 0.1 M hydrochloric acid. Overall, the developed CE system allowed direct injections of CSF samples yielding good separation efficiencies and stable migration times (RSDs < 2%) for organic acids. Validation of the method with artificial and real CSF samples showed good linear responses (r>0.99), and LODs for the organic acids were in the

range of 2-8 g/ml when applying UV detection. RSDs for migration times and peak areas were <2% and <7%, respectively. The applicability of the CE system is shown for the determination of organic acids in CSF samples.

49

Chapter 3

1. Introduction Organic acids are important metabolites in major metabolic pathways, such as the Krebs Cycle or the Pentose Phosphate Pathway. The accumulation of organic acids in biological fluids, such as urine and cerebrospinal fluid (CSF), can provide useful information about metabolic diseases. This accumulation is often the consequence of genetic malfunctions causing enzymatic failure or decreased enzyme activities [1]. The determination of organic acids in CSF is important for early diagnosis of metabolic disorders and neurological diseases. For instance, lactic acid concentrations in CSF have been reported to be increased in patients with several inborn errors of metabolism, such as pyruvate dehydrogenase deficiency and respiratory chain disorders [2]. Relatively high concentrations of lactic acid in CSF have also been described in patients with bacterial and fungal meningitis [3]. Reduced levels of both citric acid and lactic acid have been found in CSF from patients with Huntington chorea, an inherited, degenerative disorder of the central nervous system [4]. Over the past decades, gas chromatography-mass spectrometry (GC-MS) was the predominant technique for the analysis of organic acids in CSF [5,6]. GC has the advantage that it can separate a large number of low-molecular weight metabolites with high resolution. However, two main drawbacks of GC for the analysis of polar compounds in biological samples are the relatively long time for sample preparation and analysis and the limitations caused by the requirement of sample volatility. Liquid chromatographic (LC) methods, such as reversed-phase and ion-exchange LC, have also been used for the determination of organic acids in biological fluids [7,8]. However, separation times in LC may be quite long and sample preparation is often required prior to analysis. Furthermore, the resolving power of LC methods is often not sufficient for the separation of structurally very similar organic acids, such as succinic and malic acid [9]. Capillary electrophoresis (CE) is an analytical separation technique capable of high-resolution separation of a diverse range of chemical compounds, and it is particularly suitable for highly polar and charged compounds. It provides considerably higher separation efficiencies (plate numbers) than LC, and the amount of sample required for analysis is very small. CE has demonstrated to be a powerful tool for the analysis of organic acids in biological fluids [10,11]. Frequently, a background electrolyte (BGE) of high pH is used with bare fused silica capillaries in order to obtain a high electro-osmotic flow (EOF) [12]. However, high mobility compounds, such as dicarboxylic acids, may exhibit an electrophoretic mobility that is larger, but oppositely directed, than that of the EOF, which can hinder their detection when a standard CE configuration (cathode at outlet) is used. Moreover, at high pH, the CE resolution of

50


several organic acids is not optimal. Therefore, suppression or reversal of the EOF combined with reversed voltage polarity and a BGE of neutral pH is often applied for the CE analysis of organic acids. For instance, organic acids have been analyzed in deproteinized rat plasma and diluted human urine using polyacrylamide-coated capillaries with sodium phosphate BGEs of about pH 6 [13,14]. Unfortunately, the preparation of polyacrylamide coated capillaries is laborious, time consuming, and the coating stability may be limited. The addition of a cationic surfactant, such as tetradecyltrimethylammonium bromide to the BGE, can be an effective tool to reverse the EOF, and has been applied for organic acid analysis in deproteinized CSF samples from patients with central nervous system diseases [15]. However, the stability of such a dynamic coating and the control of the EOF is often critical. Furthermore, cationic surfactants may seriously suppress analyte ionization when CE is combined with electrospray ionization mass spectrometry (ESI-MS) [16]. Recently, Soga et al. have used capillaries coated with a triple layer of charged polymers to achieve an anodic EOF for CE-MS of anionic metabolites in bacterial extracts [17]. The coating procedure was first described by Katayama et al. [18] and comprises rinses of the capillary with solutions of polybrene (PB) and dextran sulfate (DS) which are physically adsorbed to the capillary wall. Katayama et al. used the PB-DS-PB coating for the analysis of basic proteins, and showed that it provided good stability against various solvents such as 1 M NaOH and 0.1 M HCl. CE of CSF samples is preferably carried out without sample pre-treatment. However, direct injections of these samples can be difficult, as matrix compounds, such as salts and proteins, may seriously interfere with the CE performance. So far, studies regarding CE analyses of CSF without sample pre-treatment were mainly focused on the determination of derivatized amino acids using laser induced fluorescence detection [19]. The CSF samples were often diluted, sometimes at least a thousand fold prior to injection and therefore, the influence of the sample matrix on the CE performance was negligible [20,21]. More recently, micellar electrokinetic chromatography has been used for the analysis of drugs in CSF using direct injections [22]. Interactions of CSF proteins with the capillary wall were minimized due to complexation of the proteins with sodium dodecyl sulfate (SDS). Moreover, the SDS-protein complexes had relatively long migration times and, therefore, they did not interfere with the peaks of the small drugs. In the present study, the direct CE analysis of organic acids in CSF samples using a PB-DS-PB capillary coating is examined. Various separation parameters, such as type and concentration of BGE, have been optimized. The influence of matrix constituents, such as salt and proteins, on the separation efficiency and reproducibility

51

Chapter 3

of the migration time of organic acids has been studied. Capillary washing steps between the runs have been investigated with respect to efficient albumin removal. Finally, the applicability of the developed CE system has been tested by the determination of organic acids in CSF samples. 2. Experimental 2.1 Materials Polybrene (hexadimethrine bromide) and dextran sulfate sodium salt, and bovine albumin were purchased from Sigma–Aldrich (Steinheim, Germany). The BGE constituents boric acid, sodium dihydrogen phosphate and disodium hydrogen phosphate were from Merck (Darmstadt, Germany). Sodium or potassium salts of lactic, oxalic, citric, 2-hydroxybutyric, 3-hydroxybutyric, and glycolic acids were of analytical standard-grade and purchased from Sigma–Aldrich. Organic acid stock solutions (10 mg/mL) in water were weekly made. Test mixtures of organic acids

(133.3 g/mL each, unless otherwise stated) were made by adding an aliquot of the organic acid stock solutions to the appropriate volume of water or artificial CSF samples. For EOF determinations, formamide was added to test samples to final concentrations of 0.033–0.33% (v/v). Phosphate BGEs (50-200 mM) were made by preparing a sodium dihydrogen phosphate of the desired concentration and titrated the solution to pH values in the range of 5.0-8.0 with a disodium hydrogen phosphate solution of the same concentration. Boric acid BGEs (50-200 mM) were titrated to the desired pH (8.0-9.0) with sodium hydroxide (1 M). Human CSF samples were kindly provided by the Utrecht Medical Centre (Utrecht, The Netherlands) and the Academic Medical Centre (Amsterdam, The Netherlands),

and stored at -80C until analysis. No sample preparation was carried out prior to

injection. Artificial CSF was prepared by dissolving NaCl (8.66 g/L), MgCl26H2O

(0.163 g/L), KCl (0.224 g/L), CaCl22H2O (0.206 g/L), and bovine albumin (0.2 g/L) in 2 mM phosphate buffer (pH 7.0). The final salt and protein content closely resembles the composition of endogenous CSF [23,24]. 2.2 CE system Capillary zone electrophoresis was performed on a P/ACE ProteomeLab PA 800 (Beckman Coulter, Fullerton, CA, USA) equipped with a diode-array detector. Electropherograms of organic acids were monitored at 200 nm. Fused-silica capillaries

52


with an internal diameter of 75 m (Composite Metal Services Ltd., U.K.) had a total length of 60 cm and an effective length of 50 cm. Injections were performed hydrodynamically for 5 or 10 s at 0.2 or 0.5 psi, respectively. The separation voltage was -10 kV and the capillary was thermostated at 25 °C. Electropherograms were analyzed using 32 Karat Software, version 7.0 (Beckman Coulter). Calculation of plate numbers was based on peak-widths-at-half-maximum. 2.3 Capillary coating and flushing procedures New fused-silica capillaries were rinsed with deionized water for 5 min at 1380 mbar followed by 1 M NaOH for 30 min at 1380 mbar, and deionized water for 15 min at 1380 mbar. The triple layer coating was prepared by subsequently rinsing the capillary with 10% (m/v) PB solution for 15 min at 345 mbar, deionized water for 5 min at 1380 mbar, 3% (m/v) DS solution for 15 min at 345 mbar, deionized water for 5 min at 1380 mbar, and, finally with a 10% (m/v) PB solution for 15 min at 345 mbar. The capillary was then ready for use with the BGE of choice. Before analysis, capillaries were flushed with BGE for 5 min at 1380 mbar. Between runs, the coated capillaries were flushed with a 0.1% (m/v) PB solution and BGE, each for 2 min at 1380 mbar. During the analyses of artificial and real CSF samples, an additional rinse with 100 mM hydrochloric acid (3 min at 1380 mbar) was incorporated between the runs, unless otherwise stated. Throughout this study, the integrity and stability of the triple layer coating was examined by analyzing an aqueous solution of formamide (0.033-0.33%, v/v) in triplicate and calculating the RSD of migration time. The coating was considered well when the RSD was within 1%. Adsorption of or damage to the coating by sample matrix constituents will be revealed by changes in the migration time of formamide. 2.4 Method validation Linearity of response for organic acid standards in water was tested by duplicate

measurement of seven levels of concentrations ranging from 1 to 1000 g/ml. The lower concentration level was chosen around the limit of quantitation of the method. Linearity of response for the organic acids was also measured in artificial CSF. The limit of detection (LOD) was calculated as the lowest concentration producing a signal-to-noise ratio of 3. The repeatability of the method was determined by five

consecutive analyses of real CSF and artificial CSF, which contained 67 g/ml oxalic

acid, 133 g/ml lactic acid, and 267 g/ml citric acid. Accuracy was determined by

53

Chapter 3

the addition of known amounts of organic acids to CSF. The recovery of the added amount was calculated using the calibration curves obtained for the organic acids in artificial CSF, taking into account the endogenous organic acid concentrations. 3. Results and discussion 3.1 Optimization and evaluation of the CE system The triple layer capillary coating was prepared by successively rinsing a bare fused-silica capillary with solutions of PB, DS and PB, yielding a positively charged coating with an EOF directed towards the anode. To verify the performance of the produced coating, the mobility of formamide (EOF marker) was determined using three BGEs, viz. 150 mM sodium phosphate (pH 3.0), 150 mM sodium phosphate (pH 6.5), and 150 mM sodium borate (pH 8.5). Indeed, as described by Katayama et al. [18], for all BGEs an anodic and virtually pH independent EOF was found with a mobility of

about 2.0-2.310-8 m2V-1s-1. The determination of the electrophoretic mobility of formamide was used in further experiments to test the integrity and stability of the coating. Optimization of the CE system started with selecting the BGE for the analysis of organic acids. For that purpose several BGEs containing sodium borate or sodium phosphate of different concentrations were examined. A test mixture of oxalic, citric, glycolic, lactic, 2-hydroxybutyric and 3-hydroxybutyric acid in water was used. With 150 mM sodium phosphate (pH 7.0) the plate numbers of the organic acids were in the range of 100,000-200,000, whereas they were around 100,000 with 150 mM sodium borate (pH 8.5). Better resolution of the organic acids was obtained using sodium phosphate. Therefore, the optimization of the CE system was continued with sodium phosphate as BGE. The influence of the pH (5-8) on the separation of the organic acids was determined. Optimum resolution of the organic acids, including baseline separation of the isomers 2-hydroxybutyric and 3-hydroxybutyric acid, was obtained with a BGE of pH 6.0. Next, the influence of the BGE concentration in the range 50 to 200 mM on the separation of the organic acids was studied. At BGE concentrations above 200 mM unacceptably high currents were observed. For an increase of the BGE concentration the plate numbers of the doubly negatively charged organic acids (oxalic and citric acid) showed a modest increase reaching values of 120,000 and 200,000, respectively, at 200 mM sodium phosphate. The plate numbers of the singly negatively charged organic acids remained virtually the same (approximately 150,000) when the BGE concentration was varied. The increase of the plate number observed for the doubly

54


charged acids is probably due to reduced interaction with the positively charged capillary coating and not to stacking. This can be concluded from the fact that the same gain in plate numbers is found when the organic acids were dissolved in BGE instead of water. Using an injection volume of 9 nL (0.4% of capillary volume), as in the experiments

described above, the limit of detection (S/N=3) was in the 10-20 g/mL range for the organic acids. However, several organic acids are present in CSF at concentrations

below the 10 g/mL. In order to improve the sensitivity, the injection volume was increased to 44 nL (2% of capillary volume). Obviously, such a large injection volume could easily cause extra band broadening, however, with a BGE of 200 mM plate numbers of the organic acids were still 100,000-150,000. The only modest decrease in efficiency is most probably the result of stacking as can be concluded from Figure 1, which depicts the plate numbers of the organic acids as a function of the BGE concentration. The organic acids were dissolved in water and optimum plate numbers were found when the BGE concentration was high. Stacking is most apparent when the difference in conductivity between BGE and injection zone is very large. Indeed, very low plate numbers were obtained if the organic acids were dissolved in BGE (200 mM) instead of water.

0

50000

100000

150000

200000

50 100 150 200

BGE (mM)

pla

te n

um

ber

oxalic acid

citric acid

lactic acid

2-hydroxybutyric acid

Figure 1. Influence of BGE concentration on the plate numbers of organic acids dissolved in water. Conditions: BGE, sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. Mean and standard deviation of three experiments are depicted. The plate number of citric acid could not be determined at a BGE concentration of 100 mM due to co-migration with a system peak.

55

Chapter 3

For example, for citric and lactic acid dissolved in BGE, plate numbers were about 28,000 and 23,000, respectively. Using the large injection volume in combination with a BGE of 200 mM sodium phosphate (pH 6.0) for the test organic acids dissolved in water, LODs could be improved with a factor 4-5, while the separation efficiency remained largely unchanged. 3.2 Influence of the sample matrix The determination of organic acids in CSF by CE using direct sample injection may be challenging due to the high and variable concentrations of salts and proteins in CSF. The NaCl concentration in CSF is about 150 mM. In CE, samples of high ionic strength may strongly decrease the plate number and deteriorate the separation of analytes [25]. Therefore, the effect of NaCl (50-200 mM) in the sample on the plate numbers of the organic acids was evaluated using a BGE of 200 mM. Somewhat surprisingly, no adverse effects of NaCl on the plate numbers of the organic acids was observed, i.e., the plate numbers were similar to those obtained for the organic acids in deionized water. Figure 2 depicts the results for samples dissolved in 150 mM NaCl with respect to 0 mM (no salt added), and even shows a gain in plate number for the high-ionic-strength sample.

0

50000

100000

150000

200000

0 150

NaCl concentration (mM) in sample

pla

te n

um

ber

oxalic acid

citric acid

lactic acid

2-hydroxybutyric acid

Figure 2. Influence of NaCl concentration in sample solvent on the plate number of organic acids. Conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. Mean and standard deviation of three experiments are depicted.

56


Most probably, band broadening caused by the high sample conductivity is counteracted by a sample-induced transient isotachophoresis (tITP) process. This effect has been observed before during CE analyses of biological fluids containing significant NaCl concentrations [26-28]. The high concentration of chloride in the sample may function as leading electrolyte due to its much larger mobility compared to the organic acids, whereas the BGE can act as terminating electrolyte due to its high concentration and relative low mobility. The average protein (largely albumin) concentration in CSF is about 200 mg/L. In CE, proteins can adsorb to the capillary wall causing alterations of the electro-osmotic flow (EOF) and band broadening. Especially with bare fused silica capillaries the adsorption of proteins to the capillary wall can lead to very poor migration time repeatability and low plate numbers. One strategy to reduce protein adsorption is coating of the capillary wall. In the present study, we used a PB-DS-PB triple layer coating to generate a significant pH-independent anodic EOF. In order to test the ability of this coating to prevent protein adsorption, organic acid solutions containing 200 mg/L bovine albumin were analyzed. However, upon repeated injection, increasing analyte migration times were observed, which probably can be attributed to albumin adsorption to the capillary wall. Extensive washing with 1 M NaOH between runs did not solve the problem, as can be observed for oxalic acid in Figure 3.

0

4

8

12

16

0 100 200 300 400

concentration bovine albumin (mg/L) in sample

mig

rati

on

tim

e (m

in)

oxalic acid

citric acid

lactic acid

oxalic acid(NaOHwashing)

Figure 3. Influence of albumin concentration in sample solvent on migration times of organic acids. Conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. The washing step is rinsing with 100 mM hydrochloric acid between runs, unless otherwise stated. Migration time stability of oxalic acid was only studied in 200 mg/L albumin.

57

Chapter 3

NaOH is not suitable to remove albumin, as this protein is overall negatively charged at high pH, and therefore, may strongly interact with the positively-charged capillary wall. Indeed, rinsing with 0.1 M hydrochloric acid (HCl) proved to be very suitable for removal of albumin from the capillary wall. At a low pH, albumin is positively charged and electrostatically repelled from the positively-charged capillary wall. Using rinses of 0.1 M HCl between runs, the effect of albumin (50-400 mg/L) in the sample on the plate number of the organic acids became negligible. Figure 3 shows that when using HCl rinses the migration times of the organic acids was virtually constant for all albumin concentrations. This stability of the migration times of the organic acids can be primarily attributed to the triple layer coating, which appeared to be resistant against washing procedures with HCl as the EOF remained constant in time. Overall, it can be concluded that high salt concentrations and albumin in the sample can be handled using a PB-DS-PB coated capillary in combination with a HCl rinsing procedure. In other words, the CE system should allow direct CSF injections. Figure 4 shows the CE analysis of a mixture of six organic acids dissolved in artificial CSF (see Experimental). Baseline separation, also for the hydroxybutyric acid isomers, was obtained with plate numbers between 100,000 and 200,000. Albumin shows a discrete peak which does not interfere with the organic acids.

0

5

10

15

20

7 11 15 19

time (min)

Ab

sorb

ance

(m

AU

)

1

2 5 7

3 4 6

Figure 4. CE of artificial CSF spiked with six organic acids. Conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. Peaks: 1, oxalic acid (67 g/ml); 2, citric acid (267 g/ml); 3, glycolic acid (133 g/ml); 4, lactic acid (133 g/ml); 5, 2-hydroxybutyric acid (133 g/ml); 6, 3-hydroxybutyric acid (133 g/ml); 7, bovine albumin (200 g/ml).

58


3.3 Validation and application From the study described above, it was concluded that the PB-DS-PB coated capillary in combination with a BGE of 200 mM sodium phosphate (pH 6.0) is suitable for the direct analysis of organic acids in CSF. In order to apply the developed CE method for the quantitative determination of organic acids in CSF, several parameters such as linearity, limit of detection (LOD), recovery and repeatability were determined. Table 1 shows the calibration curves of organic acids in artificial CSF. All organic acids show a correlation coefficient (r) above 0.99. The same curve slopes were obtained for the organic acids in water in the specified concentration intervals. The intercepts obtained for the organic acids in artificial CSF were slightly different from the intercepts obtained for the organic acids in water. For glycolic acid a significantly higher intercept was obtained with the consequence that for low concentrations the determination is not very reliable. The limits of detection (S/N = 3) were between 2

g/ml for oxalic acid and 8 g/ml for citric acid. Table 1. Linearity and LODs for organic acids in artificial CSFa.

a CE conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds.

Acid Range (g/ml) Calibration curveb r LOD (g/ml) Oxalic 16-1000 y = 79.411x – 146.29 0.998 1.9 Citric 32-1000 y = 14.096x – 97.162 0.999 7.8 Lactic 16-1000 y = 12.974x – 11.850 0.999 6.3

Glycolic 32-1000 y = 21.293x + 570.57 0.998 5.9 2-hydroxybutyric 16-1000 y = 15.142x + 0.0175 0.998 6.0 3-hydroxybutyric 16-500 y = 10.893x – 0.0228 0.999 7.6

b Calibration curves are expressed as regression lines (y = ax + b), where y is the peak area and x is the concentration organic acid in g/ml. Each calibration point has been measured in duplicate. The repeatability of the method has been examined for five consecutive injections of three organic acids in artificial CSF and in real CSF (Table 2). Acceptable RSDs for migration times were obtained and RSDs obtained for peak areas were satisfactory. The very good repeatability for migration time can primarily be attributed to the triple-layer capillary coating. Recovery was evaluated by spiking CSF samples with three organic acids at three different concentration levels: normal (endogenous) CSF values, 50% below and 50% above normal CSF values. The recoveries of the added amounts were calculated using the calibration curves constructed in artificial CSF. For all tested organic acids, the recoveries were higher than 94% (Table 2), indicating that virtually no irreversible adsorption to the capillary wall occurs.

59

Chapter 3

Table 2. Accuracy and repeatability for organic acidsa. Acid Oxalic Citric Lactic

RSD for migration time Artificial CSF (n=5)b <1% <1% <1%

CSF (n=5) <2% <2% <3%

RSD for peak area

Artificial CSF (n=5)b <5% <5% <5%

CSF (n=5) <5% <5% <7%

Recovery (%) in CSF

Low/Normal/High-levelc >95% >94% >95%

a CE conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. b Artificial CSF spiked with 67 g/ml oxalic acid, 267 g/ml citric acid and 133 g/ml lactic acid. c Low and high level indicate spiking with 50% below and 50% above normal (endogenous) CSF values. In order to test the applicability of the method, CSF samples were analyzed with the developed CE system using direct sample injection. Figure 5 shows the electropherograms of CSF from a healthy individual and from a patient with bacterial meningitis. 4

B

6

5 1 2 3

A 4 6 3 7 5 1 2 Figure 5. CE analysis of bacterial meningitis CSF (upper trace, B) and healthy CSF (lower trace, A). Conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. Peaks: 1, oxalic acid; 2, citric acid; 3, glycolic acid; 4, lactic acid; 5, 2-hydroxybutyric acid; 6, 3-hydroxybutyric acid; 7, albumin.

60


Peaks were first provisionally identified by migration time, and their identity was subsequently confirmed by standard addition experiments. The concentrations of the organic acids in both CSF samples as determined by CE-UV are listed in Table 3. The concentrations found in healthy CSF were in correspondence with normal levels [29,30]. However, an increased concentration of lactic acid was found in the bacterial meningitis CSF sample. Indeed, patients suffering from bacterial infections often

exhibit a CSF lactic acid concentration that exceeds 300 g/ml [31]. Table 3. Concentration of organic acids (g/ml) in two CSF samplesa.

Acid Healthy CSF Bacterial meningitis CSF

Oxalic 9.9 8.4 Citric 112 112

Lactic 222 384 a CE conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 0.5 psi for 10 seconds. The hydroxybutyric acids levels in the bacterial meningitis CSF sample were below the LOD. The concentrations of oxalic, citric and glycolic acid were comparable with the concentrations found in healthy CSF. Overall, it can be concluded that the developed CE method allows fast quantitative analysis of organic acids in CSF without any sample pretreatment. This is advantageous in comparison to other techniques, such as GC-MS. 4. Concluding remarks A CE method for the determination of organic acids in CSF using direct sample injection was developed. In contrast to other analytical techniques, such as GC-MS, the organic acids can be analysed rapidly without sample preparation, offering high efficiency and good repeatability of migration times and peak areas. These merits can be primarily attributed to the use of a triple-layer capillary coating in combination with HCl washing procedures. The method has shown to be feasible for the fast and reliable quantitative analysis of organic acids in CSF samples with LODs in the 2-8

g/ml range. Although more laborious, GC-MS still provides a wider applicability owing to a higher sensitivity for organic acids. The stability and separation power of the present system seems to provide high potential for the profiling of endogenous metabolites in biological samples without any sample pre-treatment. Currently, we are investigating the possibility to use the CE system in combination with mass spectrometric detection for the sensitive and selective characterization of endogenous compounds in biological samples using direct

61

Chapter 3

sample injection. In this respect, it will be essential to study whether the same high performance can be achieved using volatile BGEs, such as ammonium acetate. References [1] K.B. Elgstoen, J.Y. Zhao, J.F. Anacleto and E. Jellum, J. Chromatogr. A 914

(2001) 265. [2] J.F. Benoist, C. Alberti, S. Leclercq, O. Rigal, R. Jean-Louis, H. Ogier de

Baulny, D. Porquet, D. Biou, Clin. Chem. 49 (2003) 487. [3] A. Hutchesson, M.A. Preece, G. Gray, A. Green, Clin. Chem. 43 (1997) 158. [4] M. Garseth, U. Sonnewald, L.R. White, M. Rod, J.A. Zwart, O. Nygaard, J.

Aasly, J. Neurosci. Res. 60 (2000) 779. [5] E. Jellum, A.K. Thorsrud, E. Time, J. Chromatogr. 559 (1991) 455. [6] T. Kuhara, Mass Spectrom. Rev. 24 (2005) 814. [7] R. Vonach, B. Lendl, R. Kellner, J. Chromatogr. A 824 (1998) 159.. [8] S.P. Wang, C.S. Liao, J. Chromatogr. A 1051 (2004) 213. [9] M. Bhattacharya, L. Fuhrman, A. Ingram, K.W. Nickerson, T. Conway, Anal.

Biochem. 232 (1995) 98. [10] T. Soga, M. Imaizumi, Electrophoresis 22 (2001) 3418. [11] B. Baena, A. Cifuentes, C. Barbas, Electrophoresis 26 (2005) 2622. [12] E. Jellum, H. Dollekamp, C. Blessum, J. Chromatogr. B 683 (1996) 55. [13] B. Baena, D. García-Martínez, C. Barbas, J. Chromatogr. A 1051 (2004) 199. [14] A. García, C. Barbas, R. Aguilar, M. Castro, Clin. Chem. 44 (1998) 1905. [15] A. Hiraoka, J. Akai, I. Tominaga, M. Hattori, H. Sasaki, T. Arato,

J. Chromatogr. A 680 (1994) 243. [16] J. Hagberg, J. Chromatogr. A 988 (2003) 127. [17] T. Soga, Y. Ueno, H. Naraoka, Y. Ohashi, M. Tomita, T. Nishioka, Anal.

Chem. 74 (2002) 2233. [18] H. Katayama, Y. Ishihama, N. Asakawa, Anal. Chem. 70 (1998) 5272. [19] J. Bergquist, S. Douglass Gilman, A. G. Ewing, R. Ekman, Anal. Chem. 66

(1994) 3512. [20] G. Nouadje, H. Rubie, E. Chatelut, P. Canal, M. Nertz, Ph. Puig, F. Couderc,

J. Chromatogr. A 717 (1995) 293. [21] S. Tucci, C. Pinto, J. Goyo, P. Rada, L. Hernández, Clin. Biochem. 31 (1998)

143. [22] H. Yeh, Y. Yang, S. Chen, Electrophoresis 27 (2006) 819. [23] H. Davson, Human Physiology of the Cerebrospinal Fluid, J. & A. Churchill,

Ltd., London, 1967.

62


63

[24] M.A. Watson, M.G. Scott, Clin. Chem. 41 (1995) 343. [25] J.R. Veraart, H. Lingeman, U. A. Th. Brinkman, J. Chromatogr. A 856 (1999)

483. [26] J. Boden, K. Bächmann, J. Chromatogr. A 734 (1996) 319. [27] L. Krivankova, P. Pantuckova, P. Gebauer, P. Bocek, J. Caslavska, W.

Thormann, Electrophoresis 24 (2003) 505. [28] A.R. Timerbaev, T. Hirokawa, Electrophoresis 27 (2006) 323. [29] J.F. Benoist, C. Albertini, S. Leclerq, O. Rigal, R. Jean-Louis, H.O. de Baulny,

D. Porquet, D. Biou, Clin. Chem. 49 (2003) 487. [30] G.F. Hoffmann, C.K. Seppel, B. Holmes, F. Hanefeld, D. Rating, W.L.

Nyhan, J. Chromatogr. B 617 (1993) 1. [31] J.A. Knight, S.M. Dudek, R.E. Haymond, Clin. Chem. 27 (1981) 1431.

Chapter 4

Metabolic Analysis of Body Fluids by Capillary Electrophoresis Using Noncovalently Coated Capillaries

R. Ramautar, O.A. Mayboroda, A.M. Deelder, G.W. Somsen, G.J. de Jong,

Journal of Chromatography B 2008, 871, 370-374

Metabolic Analysis of Body Fluids by Capillary Electrophoresis

Abstract The potential of capillaries noncovalently coated with charged polymers for the metabolic analysis of body fluids by CE is illustrated. Firstly, the usefulness of a coating consisting of a triple layer of polybrene-dextran sulfate-polybrene for the fast analysis of organic acids is described. The CE system allowed direct injections of CSF, plasma and urine samples, yielding good separation efficiencies. RSDs for migration times and peak areas of organic acids in plasma were <3% and <5%, respectively. The usefulness of the system is illustrated by the profiling of organic acids in plasma and urine samples. Secondly, a CE system comprising a bilayer coating of polybrene-poly(vinylsulfonate), which provides a considerable EOF at low pH is described. This system was combined with TOF-MS and used for the fast analysis of amino acids in cerebrospinal fluid (CSF) and urine with minimal sample pretreatment. RSDs for migration times and peak areas of amino acids in CSF and urine were <2% and <10%, respectively. The applicability of the system is demonstrated by the profiling of endogenous low-molecular weight metabolites in CSF from a healthy individual and a patient with complex regional pain syndrome.

67

Chapter 4

1. Introduction Metabolic profiling of body fluids, such as urine and cerebrospinal fluid (CSF), provides an insight into the metabolic state of an organism and specific biochemical processes. Hence, metabolic profiling has become important for the screening of potential diagnostic and prognostic markers of various diseases or for the detection of pharmacological or toxicological effects obtained following dosing of compounds [1]. This novel approach has a potential to replace more classical methods in clinical laboratories which are often focused on the measurement of a specific endogenous metabolite or a class of endogenous metabolites. It provides more information about the physiological state of an organism [2]. To date, metabolic profiling has been mostly performed with GC–MS and NMR spectroscopy [3-5]. Applications of GC-MS for large-scale metabolite analysis are mainly found in the field of plant metabolomics [6]. GC-MS is limited to volatile metabolites and derivatization is often needed to yield volatile and thermostable compounds. NMR spectroscopy has commonly been used for the metabolic profiling of body fluids, such as serum and urine [7]. NMR is rapid, non-destructive and requires minimal sample preparation. Nevertheless, the sensitivity of NMR is limited and analyte amounts of several micrograms are required. The widespread use of LC–MS for global metabolic profiling is relatively new [8]. LC-MS can supply information on the chemical structure and the quantity of low-abundance metabolites. More advanced LC systems, i.e. monolithic capillary LC or UPLC offering improved separation efficiencies, are gaining more attention for the metabolic profiling of body fluids [9]. However, LC-MS also has some limitations, especially when applied to highly polar compounds. As many components in body fluids are highly polar and ionic, separation of these components using common reversed-phase LC can be problematic [10]. Capillary electrophoresis (CE) is an analytical technique capable of high-resolution separation of a diverse range of chemical compounds. It is particularly suitable for the separation of polar and charged compounds [11]. Therefore, CE is very attractive for the separation of endogenous metabolites in body fluids. Furthermore, the costs for accessories are less than in LC, due to very low organic solvent consumption, the small amount of reagents and the use of inexpensive fused-silica capillaries. Another attractive feature of CE is its small sample requirement, making it particularly well-suited for samples, such as CSF, that are volume-limited. Capillary zone electrophoresis (CZE) has commonly been used for the analysis of a subset of endogenous metabolites in urine in order to study metabolic disorders [10-14]. In order to improve selectivity, micellar electrokinetic chromatography (MEKC),

68


which employs micelles as pseudo-stationary phases in the BGE, has been used for metabolic profiling of urine [15,16]. However, the coupling of MEKC to MS is problematic and often provides limited sensitivity. Variability of migration time is an important issue in CE. Procedures to correct for migration time shifts and for aligning electropherograms have been developed [10,16]. Still, for metabolic profiling studies of body fluids, reproducibility of migration times and peak areas is of utmost importance for a reliable comparison of profiles and to observe small changes in sample composition. Using CE with bare fused silica capillaries, the analysis of body fluids with minimal sample pretreatment is often not possible due to adsorption of proteins or other matrix components to the capillary wall causing irreproducible electro-osmotic flows and migration times. A promising approach to minimize this problem is the use of noncovalently coated capillaries, i.e. dynamically coating of the bare fused silica capillaries with charged polymers. Recently, we have described the methodological aspects of noncovalent coatings in CE-UV and CE-MS for metabolite profiling [17,18]. In the present study we describe the applicability of two noncovalently coated capillaries in CE-based metabolic analysis of body fluids with minimal or no sample pretreatment. 2. Experimental 2.1 Chemicals The background electrolyte (BGE) constituents sodium dihydrogen phosphate and disodium hydrogen phosphate were from Merck (Darmstadt, Germany). Sodium or potassium salts of lactic, citric, 3-hydroxybutyric, and pyroglutamic acids were of analytical standard-grade and purchased from Sigma-Aldrich (Steinheim, Germany). Organic acid stock solutions (10 mg/mL) in water were prepared weekly. Test mixtures of organic acids (133.3 μg/mL each, unless otherwise stated) were made by adding an aliquot of the organic acid stock solutions to the appropriate volume of water or plasma, CSF or urine samples. For electro-osmotic flow (EOF) determinations, formamide was added to test samples to a final concentration of 0.033% (v/v). Polybrene (hexadimethrine bromide), dextran sulfate sodium salt, and 25% m/v aqueous solution of poly(vinylsulfonate) (PVS) sodium salt were purchased from Sigma-Aldrich (Steinheim, Germany). Ammonium hydroxide (25%) was from Merck (Darmstadt, Germany). Amino acids (l-alanine, l-arginine, l-glutamic acid, l-isoleucine, l-leucine, l-lysine, l-methionine, l-phenylalanine, l-tyrosine, l-valine) with a concentration of 1 mM of each in 0.1 M hydrochloric acid were purchased from Sigma-Aldrich. HPLC-grade methanol was supplied by Biosolve BV (Valkenswaard,

69

Chapter 4

the Netherlands). Standard solutions of amino acids were prepared with water taken from a Milli-Q water purification system (Millipore, Bedford, MA, USA). BGE solution for CE-MS was prepared by dissolving formic acid in Milli-Q water (1 M, pH 1.8). Human CSF and plasma samples were provided by the Leiden University Medical Centre (Leiden, The Netherlands) and stored at −80 °C until analysis. Citric acid was used as an anticoagulant agent for human plasma samples. For CSF and plasma samples, no sample preparation was carried out prior to injection. Urine samples from a healthy control group were obtained from the department of Infectious Diseases of the Leiden University Medical Centre (Leiden, The Netherlands). After collection, urine samples were stored immediately at -80 ºC. Prior to CE-MS analysis, the urine samples were thawed to room temperature, mixed with BGE (1:1, v/v) and centrifuged (13,200 rpm) for 5 min. 2.2 Capillary coating procedures 2.2.1 PB-DS-PB coating New fused-silica capillaries were rinsed with deionized water for 5 min at 1380 mbar followed by 1 M NaOH for 30 min at 1380 mbar, and deionized water for 15 min at 1380 mbar. The triple layer coating was prepared by subsequently rinsing the capillary with 10% (m/v) PB solution for 15 min at 350 mbar, deionized water for 5 min at 1380 mbar, 3% (m/v) DS solution for 15 min at 350 mbar, deionized water for 5 min at 1380 mbar, and, finally with a 10% (m/v) PB solution for 15 min at 350 mbar. The capillary was then ready for use with the BGE of choice. Before analysis, capillaries were flushed with BGE for 5 min at 1380 mbar. Between runs, the coated capillaries were flushed with a 0.1% (m/v) PB solution and BGE, each for 2 min at 1380 mbar. During the analyses of CSF or plasma samples, an additional rinse with 100 mM hydrochloric acid (10 min at 1380 mbar) was incorporated between the runs, unless otherwise stated. 2.2.2 PB-PVS coating New bare fused-silica capillaries were rinsed with deionized water for 5 min at 1380 mbar followed by 1 M sodium hydroxide for 15 min at 1380 mbar, and deionized water for 5 min at 1380 mbar. Coating was performed by rinsing for 30 min at 350 mbar with 10% (m/v) PB and successively with water for 5 min at 1380 mbar. Subsequently, the capillary was flushed with 5% (v/v) PVS for 30 min at 350 mbar, and again water for 5 min at 1380 mbar. At the start of the day, coated capillaries were flushed with deionized water for 1 min at 1380 mbar and with BGE for 2 min at 1380

70


mbar. Between runs, the coated capillaries were flushed with 5% (v/v) PVS for 2 min at 1380 mbar and with BGE for 1 min at 1380 mbar. 2.3 Instrumentation and CE conditions CE was performed on a P/ACE ProteomeLab PA 800 (Beckman Coulter, Fullerton, CA, USA) equipped with a diode-array detector. Electropherograms of organic acids were monitored at 200 nm. Fused-silica capillaries with an internal diameter of 75 μm (Composite Metal Services Ltd., UK) had a total length of 60 cm and an effective length of 50 cm. Injections were performed hydrodynamically for 10 s at 35 mbar. The separation voltage was −10 kV and the capillary was thermostated at 25 °C. Sodium phosphate BGE (200 mM) was made by mixing equimolar solutions of sodium dihydrogen phosphate and disodium hydrogen phosphate in the appropriate ratio to reach a pH of 6.0. Electropherograms were analyzed using 32 Karat Software, version 7.0 (Beckman Coulter). For CE-MS, the CE experiments were also carried out on a P/ACE ProteomeLab PA 800 instrument. Capillaries had a total length of 130 cm and an internal diameter of 50 μm. Formic acid (1 M, pH 1.8) was used as BGE. Sample injections were performed hydrodynamically for 90 s at 90 mbar (1.3 psi). Prior to sample injection a small plug (50 mbar, 9 s) of 12.5% ammonium hydroxide was injected for pH-mediated stacking. The separation voltage was 30 kV and the capillary temperature was 25 °C. MS was performed using a micrOTOF-model orthogonal-accelerated time-of-flight (TOF) mass spectrometer (Bruker Daltonics, Bremen, Germany). Transfer parameters were optimized by direct infusion of an ESI tuning mix (Agilent Technologies, Waldbronn, Germany). Spectra were collected with a time resolution of 1 s. Post-run internal mass calibration was performed using sodium formate cluster ions Na+(HCOONa)1-9 ranging from 90.9766 to 430.9137

m/z, which are detected in the first part of the electropherogram after urine sample injection. CE-MS coupling was realized by a co-axial sheath liquid interface (Agilent Technologies, Waldbronn, Germany) with methanol-water-formic acid (50:50:0.1, v/v/v) as sheath liquid. The following spray conditions were used: sheath liquid flow, 4 μL/min; dry gas temperature, 180°C; nitrogen flow, 4 L/min; nebulizer pressure, 0.5 bar. Electrospray in positive ionization mode was achieved and ESI voltage was -4.5 kV. 2.4 Quantification studies Peak areas, corrected for migration times, of organic acids in plasma, urine and CSF samples were used for determination of the concentration using calibration curves

71

Chapter 4

constructed in 150 mM NaCl solutions. The migration time and peak area repeatability of the CE-UV method with the PB-DS-PB coating was determined for the compounds citric acid, lactic acid and pyroglutamic acid using their endogenous concentration levels by ten consecutive analyses of plasma, urine and CSF. The identity of organic acids was verified by standard addition experiments. The migration time and peak area repeatability of the CE-MS method with the PB-PVS coating was determined by 10 consecutive analyses of a pooled human urine sample spiked with 50 μM of eight amino acids. 3. Results and discussion 3.1 CE-UV system for the analysis of organic acids in body fluids The analysis of organic acids in body fluids plays an important role in the screening, diagnosis and monitoring of a variety of metabolic disorders. Using CE, reversed polarity is required for the fast analysis of organic acids as their electrophoretic mobility towards the anode is usually higher than the electro-osmotic flow (EOF) towards the cathode. Recently, we have developed a CE-UV method based on a triple layer capillary coating for the quantitative determination of organic acids in CSF [17]. The triple layer capillary coating was prepared by successively rinsing a bare fused-silica capillary with solutions of PB, DS, and PB, yielding a positively charged coating with an pH-independent EOF directed towards the anode [19]. As the EOF was reversed, the organic acids migrated before the EOF enabling relatively fast analysis times. The repeatability of the triple layer capillary coating was evaluated on the basis of the RSD values for migration times of two organic acids. The results are summarized in Table 1. Good repeatability of migration times was obtained for intraday, interday and capillary-to-capillary analyses indicating a high coating stability. Table 1. Migration time repeatability data (RSD, %) using noncovalently coated capillaries (n=5).

Coating PB-DS-PB PB-PVS

Citric acid1 Lactic acid1 Valine2 Phenylalanine2

Intraday 0.3 0.4 0.4 0.4 Interday 0.7 0.6 0.8 1.0

capillary-to-capillary 1.0 1.0 1.7 2.2 1Compound (100 M) dissolved in deionized water; data obtained by CE-UV. 2Compound (50 M) dissolved in 0.5 M formic acid (pH 1.8); data obtained by CE-MS.

72


It was possible to separate organic acids in CSF without any sample pretreatment as adverse effects of proteins could be effectively minimized by the triple layer coating in combination with rinses of 0.1 M hydrochloric acid. The organic acids were analyzed with high separation efficiency, i.e. plate numbers were above the 100,000 and albumin did not interfere with the analysis of the organic acids. The CE-UV method based on the PB-DS-PB coating was used for the profiling of organic acids in CSF samples from patients with bacterial meningitis. In the present study, we have evaluated the potential of this method for the profiling of organic acids in human plasma, which is also clinically very important. For instance, the simultaneous analysis of lactic acid and 3-hydroxybutyric acid is used for the monitoring of lactic acidosis in human plasma [20]. However, plasma is a much more complex matrix than CSF containing more proteins at higher levels. For example, the average concentration of albumin in plasma is ca. 60 mg/ml, while in CSF this is approximately 0.5 mg/ml. Nonetheless, Table 2 shows high repeatability for migration times and peak areas of three organic acids after direct injection of plasma. The constant migration times of the analytes demonstrate that the EOF is stable, indicating that no or minimal protein adsorption to the capillary wall occurs. It shows the high stability of the CE-UV method with the PB-DS-PB coating. Table 2. Precision data (RSD, %) for migration times and peak areas of a few organic acids in different body fluids obtained by PB-DS-PB CE-UV (n=10).

Organic acid Plasma Cerebrospinal fluid Urine

Citric acid

Migration time 1.8 1.6 1.9 Peak area 3.9 4.6 3.6 Lactic acid

Migration time 2.5 2.9 1.5 Peak area 3.7 6.8 2.9 Pyroglutamic acid

Migration time 2.4 1.7 1.8 Peak area 4.5 3.8 4.8

Figure 1 shows that the organic acids and other endogenous metabolites in plasma can be analyzed within 15 min and without interference from albumin. Plate numbers for most organic acids were above the 100,000 which is comparable with plate numbers for organic acids obtained in CSF. The electropherogram in Figure 1 shows a limited number of compounds, while plasma contains more organic acids, such as acetoacetic acid and pyruvic acid. These compounds co-migrated with a system band which appears between peak 1 and peak 2 (Figure 1). Other organic acids, such as fumaric acid and glutaric acid, were not detected in plasma which was in agreement with the

73

Chapter 4

results obtained by a GC-MS study [21]. Another reason for the limited number of compounds shown in Figure 1 might be the relatively poor concentration sensitivity of CE-UV. The concentrations found for lactic acid, 3-hydroxybutyric acid and pyroglutamic acid were 1540, 165, 180 μM, respectively (for details on quantification studies see section 2.4). These values fall within the range of reference values of organic acids in human plasma determined by GC-MS including derivatization. Citric acid was used as an anticoagulant agent and, therefore, it was not relevant to determine the concentration of this compound.

0

10000

20000

30000

6 9 12 15 18

Time (min.)

UV

ab

sorb

ance

(m

AU

)

1

54

23

Figure 1. CE-UV of a human plasma sample. Experimental conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 35 mbar for 10 s.; detection, UV at 200 nm.; capillary, PB-DS-PB coated. Peaks: 1, citric acid; 2, lactic acid; 3, 3-hydroxybutyric acid; 4, pyroglutamic acid; 5, albumin. The potential of the CE-UV system with the PB-DS-PB coating has also been investigated for the direct analysis of human urine. An electropherogram of a urine sample of a healthy human subject is shown in Figure 2. Many negatively charged compounds can be observed within 20 min (EOFmarker ca. 20 min) requiring no sample pretreatment. The separation could be improved by slowing down the EOF via the addition of organic solvents, like methanol, to the BGE but at the expense of analysis time [19]. Improved selectivity can also be obtained using MEKC methods [16], but the coupling of MEKC to MS is not straightforward. The concentrations found for citric acid, lactic acid and pyroglutamic acid were 211 μM, 37 μM and 39 μM,

74


respectively, and they correspond with normal values [21]. The average concentration of 3-hydroxybutyric acid is below the 1 μM in urine and, therefore, this compound could not be detected as the concentration was below the detection limit of the CE-UV method [21]. An overview of repeatability data for migration times and peak areas of organic acids in different body fluids is given in Table 2. In summary, the CE-UV method with a triple-layer coated capillary shows high potential for the fast and repeatable profiling of organic acids in plasma, CSF and urine without or minimal sample pretreatment.

10000

30000

50000

70000

6 9 12 15 18

Time (min.)

UV

ab

sorb

ance

(m

AU

)

3

1 2

4

Figure 2. CE-UV of a human urine sample. Experimental conditions: BGE, 200 mM sodium phosphate (pH 6.0); injection, 35 mbar for 10 s.; detection, UV at 200 nm.; capillary, PB-DS-PB coated. Peaks: 1, citric acid; 2, lactic acid; 3, pyroglutamic acid; 4, hippuric acid. 3.2 CE-MS system for metabolic profiling of body fluids Profiling of endogenous low-molecular weight metabolites in body fluids is important for the screening of deficiencies in amino acid metabolism in routine clinical analysis [22]. So far, CE-MS methods mostly used bare fused-silica capillaries for amino acids analysis in body fluids [23,24]. Analysis times were relatively long due to the slow EOF resulting from the low-pH background electrolyte (BGE) in combination with bare fused-silica capillaries. Obviously, this is not very suitable for clinical studies where high-throughput analyses are required. Moreover, using bare fused-silica

75

Chapter 4

capillaries, separation efficiency and reproducibility may be compromised. Changes in the surface of the capillary wall due to adsorption of matrix components may cause irreproducible EOFs leading to a poor migration-time repeatability. Recently, capillaries noncovalently coated with a bilayer of oppositely charged polymers have been used for the fast and highly reproducible analysis of peptides [25]. These coated capillaries were produced by first rinsing the capillary with a solution of the positively charged polymer PB and subsequently with a solution of the negatively charged polymer PVS. The resulting PB-PVS coating provides a considerable EOF at low pH, thereby facilitating the fast separation of positively charged peptides using a BGE of formic acid (pH 2.5). In addition, the PB-PVS coating is fully compatible with MS detection causing no ionization suppression [25]. The repeatability of the PB-PVS coating was evaluated on the basis of the RSD values for migration times of two amino acids. The results are summarized in Table 1. The good repeatability of migration times for intraday, interday and capillary-to-capillary analyses indicates a high coating stability. Using a CE-MS system with a PB-PVS coated capillary, all amino acids were analyzed within 15 min [18]. Plate numbers for the amino acids in CSF and urine varied from 50,000 to 300,000. The limits of detection for the amino acids were improved by in-capillary preconcentration based on pH-mediated stacking allowing 100-nL sample injection (i.e., ca. 4% of capillary volume). As a result, detection limits for amino acids were down to 20 nM. An important advantage of this method is the online regeneration of the capillary coating by flushing with a solution of PVS between the runs. During regeneration the ion source settings of the MS were adjusted (e.g., capillary voltage was set a 0 kV) to prevent contamination with PVS. Table 3 shows that the migration times of several amino acids were quite constant for ten successive analyses of the matrix spiked with 50 μM of each analyte. The RSD values for peak areas were below the 10%, which is acceptable for biological samples using ESI-MS. The potential of the PB-PVS coated capillary for the profiling of low-molecular weight metabolites in human CSF is shown in Figure 3 (upper trace). A metabolic profile of positively charged compounds was obtained in less than 20 min. The lower trace of Figure 3 shows that the metabolic profile of CSF from a patient with complex regional pain syndrome (CRPS) is different from the metabolic profile of CSF from a healthy individual (upper trace). However, a large set of CSF samples from both groups should be analyzed to compare metabolic profiles of the two groups using biostatistical techniques.

76


Table 3. Precision data (RSD, %) for migration times and peak areas of some amino acids in CSF and urine obtained by PB-PVS CE-MS (n=10).

Cerebrospinal fluid1 Urine1

Amino acid Migration time Peak area Migration time Peak area Lysine 1.4 7.7 1.2 8.6

Arginine 1.7 6.8 1.5 5.9 Alanine 1.5 7.8 1.2 8.9 Valine 0.8 8.4 1.1 8.9

Methionine 1.5 9.8 1.1 7.5 Glutamic acid 1.4 8.5 1.2 7.9 Phenylalanine 1.6 7.6 0.8 7.9

Tyrosine 1.6 7.4 0.9 8.3 1Spiked with 50 M of each amino acid.

Figure 3. Metabolic profiling of CSF by CE-TOF-MS. Base peak electropherogram of (A) a CSF sample from a healthy individual and (B) a CSF sample from a patient with CRPS. Experimental conditions: BGE, 1 M Formic acid (pH 1.8); sample injection, 90 mbar for 90 s; pre-injection, ammonium hydroxide (12.5%) at 50 mbar for 9 s.; capillary, PB-PVS coated; scan range, 50-450 m/z.

77

Chapter 4

The analytical separation window for low-molecular weight metabolites, e.g. amino acids and small peptides, is larger using bare fused-silica capillaries, however, the PB-PVS CE system is combined with a time-of-flight (TOF) MS and, therefore, the high mass accuracy provides the required extra selectivity. Therefore, the PB-PVS CE-TOF-MS method provides a very interesting tool for the fast and highly repeatable profiling of low-molecular weight metabolites in CSF and urine. 4. Conclusion In the present study, the potential of two noncovalently coated capillaries for the metabolic analysis of body fluids by CE(-MS) was illustrated. The high stability of the present systems provides good potential for the profiling of endogenous metabolites in biological samples with minimal sample pretreatment. Currently, we are investigating the possibility to use the CE system with the triple layer coated capillary in combination with mass spectrometric detection for the sensitive and selective profiling of negatively charged compounds in biological samples using direct sample injection. The CE system with the bilayer coated capillary will be used for the metabolic profiling of urine and CSF samples from patients with CRPS. Global metabolic profiling of body fluids can be achieved by combining the results obtained with CE-MS systems with double and triple layer capillary coatings for the analysis of positively charged and negatively charged compounds. Moreover, the selectivity of MS is very helpful for the characterization of co-migrating compounds and for the ultimate identification of compounds that are responsible for the differences between control and patient samples.

78


References [1] J.K. Nicholson, J. Connelly, J.C. Lindon, E. Holmes, Nat. Rev. Drug Discov.

1 (2002) 153. [2] W.R. Wikoff, J.A. Gangoiti, B.A. Barshop, G. Siuzdak, Clin. Chem. 53 (2007)

2169. [3] O. Fiehn, T. Kind, Methods Mol. Biol. 358 (2007) 3. [4] M.E. Bollard, E.G. Stanley, J.C. Lindon, J.K. Nicholson, E. Holmes, NMR

Biomed. 18 (2005) 143. [5] E. Jellum, J. Chromatogr. 452 (1988) 435. [6] J. Kopka, J. Biotechnol. 124 (2006) 312. [7] O. Beckonert, H.C. Keun, T.M. Ebbels, J. Bundy, E. Holmes, J.C. Lindon,

J.K. Nicholson, Nat. Protoc. 2 (2007) 2692. [8] I.D. Wilson, R. Plumb, J. Granger, H. Major, R. Williams, E.M. Lenz,

J. Chromatogr. B 817 (2005) 67. [9] E.M. Lenz, I.D. Wilson, J. Proteome Res. 6 (2007) 443. [10] S. Ullsten, R. Danielsson, D. Bäckström, P. Sjöberg, J. Bergquist,

J. Chromatogr. A 1117 (2006) 87. [11] R. Ramautar, A. Demirci, G.J. de Jong, Trends Anal. Chem. 25 (2006) 455. [12] K.B. Elgstoen, J.Y. Zhao, J.F. Anacleto, E. Jellum, J. Chromatogr. A 914

(2001) 265. [13] B. Baena, D. Garcia-Martinez, C. Barbas, J. Chromatogr. A 1051 (2004) 199. [14] A. Garcia, C. Barbas, R. Aguilar, Clin. Chem. 44 (1998) 1905. [15] Ch. Guillo, D. Perrett, M. Hanna-Brown, Chromatographia 59 (2004) S157. [16] Ch. Guillo, D. Barlow, D. Perrett, M. Hanna-Brown, J. Chromatogr. A 1027

(2004) 203. [17] R. Ramautar, G.W. Somsen, G.J. de Jong, Anal. Bioanal. Chem. 387 (2007)

293. [18] R. Ramautar, O.A. Mayboroda, R.J.E. Derks, C. van Nieuwkoop, J.T. van

Dissel, G.W. Somsen, A.M. Deelder, G.J. de Jong, Electrophoresis (2008), in press.

[19] H. Katayama, Y. Ishihama, N. Asakawa, Anal. Chem. 70 (1998) 5272. [20] M.J. Paik, E.Y. Cho, H. Kim, K.R. Kim, Y.H. Ahn, G. Lee,

Biom. Chromatogr. (2008), in press. [21] G. Hoffmann, S. Aramaki, E. Blum-Hoffmann, W.L. Nyhan, L. Sweetman,

Clin. Chem. 35 (1989) 587. [22] R. Venta, Clin. Chem. 47 (2001) 575.

79

Chapter 4

80

[23] O. A. Mayboroda, C. Neusüß, M. Pelzing, G. Zurek, R.J.E. Derks, I. Meulenbelt, M. Kloppenburg, E. P. Slagboom, A. M. Deelder, J. Chromatogr. A 1159 (2007) 149.

[24] T. Soga, Y. Kakazu, M. Robert, M. Tomita, T. Nishioka, Electrophoresis 25 (2004) 1964.

[25] J.R. Catai, J.S. Torano, G.J. de Jong, G.W. Somsen, Electrophoresis 27 (2006) 2091.

Chapter 5

Evaluation of Capillary

Electrophoretic Methods for Global Metabolic Profiling of Urine

R. Ramautar, J.S. Toraño, G.W. Somsen, G.J. de Jong

Submitted

Capillary Electrophoretic Methods for Global Metabolic Profiling of Urine

Abstract In this study the usefulness of noncovalently coated capillaries with layers of charged polymers is investigated in order to obtain global electrophoretic profiles of urinary metabolites covering a broad range of different compound classes in a highly repeatable way. Capillaries were coated with a bilayer of polybrene (PB) and poly(vinyl sulfonate) (PVS), or with a triple layer of PB, dextran sulfate (DS) and PB. The bilayer and triple layer coatings were evaluated at acidic (pH 2.0) and alkaline (pH 9.0) separation conditions, thereby providing separation conditions for basic and acidic compounds. A representative metabolite mixture and spiked urine samples were used for the evaluation of the four CE methods. Migration time repeatability (RSDs <2%) and plate numbers (N, 100,000-400,000) were similar for the test compounds in all CE methods, except for some multivalent ions that may exhibit adsorption to oppositely charged coatings. The analysis of cationic compounds with the PB-DS-PB CE method at low pH (i.e., after the EOF time) provided a larger separation window and number of separated peaks in urine compared to the analysis with the PB-PVS CE method at low pH (i.e., before the EOF time). Approximately 600 molecular features were detected in rat urine by the PB-DS-PB CE-MS method whereas about 300 features were found with the PP-PVS CE-MS method. This difference can be attributed to reduced co-migration of compounds with the PB-DS-PB CE-MS method and a related decrease of ion suppression. With regard to the analysis of anionic compounds by CE-MS, in general analyte responses were significantly lower than that for cationic compounds, most probably due to less efficient ionization and to ion suppression effects caused by the background electrolyte. Hence, further optimization is required for the sensitive CE-MS analysis of anionic compounds in body fluids. It is concluded that the selection of a CE method for profiling of cationic metabolites in urine depends on the purpose of the study. For high throughput analyses the PB-PVS CE-MS method is favored whereas the PB-DS-PB CE-MS method provides a more information-rich metabolic profile, but at the cost of prolonged analysis time.

83

Chapter 5

1. Introduction In recent years human metabolomics has acquired increasing importance in biomedical studies [1-4]. Metabolomics concerns the comprehensive analysis of low-molecular-weight endogenous metabolites in a biological system. Metabolomics strategies attempt to evaluate the metabolic profile of a given biological sample in an untargeted way, so that the number of observed chemical compounds is as high as possible. No single analytical technique is capable of providing a comprehensive coverage of the metabolites present in a biological sample [5, 6]. Hence, one of the major challenges facing the rapidly developing field of metabolomics is the simultaneous analysis of a broad array of metabolite classes within a single run. Another major challenge is to perform such analysis in a reproducible way. In order to increase the coverage of the metabolome, several analytical techniques have been used in conjunction enabling the detection of metabolites which could not be detected by the use of only one method [7]. For instance, hydrophilic interaction chromatography (HILIC) has been used as a complementary method to a reversed-phase LC (RPLC) method in order to improve the coverage of (highly) polar metabolites in Zucker rat urine [8]. A large part of endogenous metabolites present in biological samples is highly polar and ionic and, therefore, capillary electrophoresis (CE) is a very attractive separation technique for metabolic profiling of biological samples. CE separates compounds on the basis of their charge-to-size ratio [9, 10]. The separation mechanism fundamentally differs from RPLC and, therefore, CE can provide complementary or additional information on the composition of a biological sample. Other features of CE include the relatively fast and highly efficient separations without the need for extensive sample pretreatment, and the small sample requirement which makes CE in particular suitable for the analysis of biological samples that are volume-limited [11]. The utility of CE in conjunction with mass spectrometry (MS) for metabolic profiling was recently demonstrated for bacterial extracts, plant extracts, urine, plasma and cerebrospinal fluid (CSF) [12-19]. The first global CE-MS method for metabolic profiling was shown for Bacillus subtilis extracts by Soga et al. [20]. Distinct CE-MS methods were used for the analysis of anionic and cationic metabolites. Separation of cationic metabolites was performed with a bare fused-silica capillary using 1 M formic acid (pH 1.8) as background electrolyte (BGE). Capillaries coated with a triple layer of polybrene-dextran sulfate-polybrene (PB-DS-PB) were used for the analysis of anionic metabolites using 50 mM ammonium acetate (pH 8.5) as BGE. The analysis of bacterial extracts by two CE-MS methods resulted in the detection of more than 1600 compounds. This is very important from a metabolomics viewpoint: using different

84


CE methods for the analysis of the same biological sample provides the opportunity to obtain a global metabolic profile. A similar approach was used by Barbas et al. for metabolic profiling of urine samples [21]. In this case, urine samples of mice were analyzed with a micellar electrokinetic CE method using bare fused-silica capillaries for the separation of cationic and neutral compounds, and with a CE method using polyacrylamide-coated capillaries for the separation of anionic compounds. Distinctive urine profiles were obtained by combining the two methods providing useful information. However, UV absorbance detection was used which obviously limits the resolving power and the reliability of metabolite assignment and/or identification, which is essential in metabolic profiling studies. Recently, we have described two CE methods using non-covalently coated capillaries for the analysis of endogenous metabolites in body fluids [22, 23]. In the first CE method, capillaries non-covalently coated with a bilayer of Polybrene (PB) and poly(vinyl sulfonate) (PVS) were used for the fast separation of amino acids using a BGE of 1 M formic acid (pH 1.8). The PB-PVS coating proved to be very consistent yielding stable CE-MS profiles of urine with favorable migration time repeatability (RSDs <2%). The method was used for metabolic profiling of urine samples from patients with urinary tract infection and complex regional pain syndrome [22, 24]. In the second CE method, PB-DS-PB coated capillaries were used in combination with a negative separation voltage for the targeted analysis of organic acids in cerebrospinal fluid (CSF), plasma and urine using a BGE of 200 mM sodium phosphate (pH 6.0) [23, 25]. Based on these non-covalently coated capillaries, a global approach for metabolic profiling of body fluids can be envisioned. The use of the PB-PVS and PB-DS-PB CE method at both low and high pH separation conditions would enable the analysis of both basic and acidic compounds. By combining the results obtained with each CE method, a broader coverage of endogenous metabolites in body fluids would be achieved, which is essential for wide-ranging endogenous metabolite profiling. Moreover, both coatings produce a rather strong and constant electro-osmotic flow (EOF) virtually independent of the pH of the BGE. The EOF can be directed towards the capillary outlet providing a stable liquid flow favorable for achieving adequate and reproducible electrospray conditions and, thus, effective analyte ionization and MS detection. In order to evaluate the potential of the described approach for global metabolic profiling of body fluids several analytical aspects were examined using a representative metabolite mixture and rat urine samples. The coverage aspect was evaluated by analyzing a test mixture representing various metabolite classes by each CE method using low and high pH separation conditions. The separation power of each CE method was investigated by considering aspects such as plate numbers and separation

85

Chapter 5

window. The migration time repeatability for test compounds spiked in rat urine and the number of molecular features detected by each CE-MS method in urine was determined and ion suppression effects were assessed. 2. Materials and methods 2.1 Chemicals and reagents All the employed chemicals were of analytical grade or higher purity. Uric acid, L-tyrosine, L-phenylalanine, guanosine, formic acid, ammonium hydroxide and formamide were purchased from Fluka (Buchs, Switzerland). Creatinine, glutathione, dopamine, adrenaline, folic acid, orotic acid, flavin adenine dinucleotide (FAD), nicotinamide adenine dinucleotide phosphate (NADPH), polybrene (PB), dextran sulfate (DS) and (poly)vinylsulfonate (PVS) were from Sigma-Aldrich (Steinheim, Germany). Methanol was obtained from ChromasolvTM. Ammonium acetate was purchased from Merck (Darmstadt, Germany). 2.2 Coating procedures Two types of multiple layer coatings were used: PB-PVS and PB-DS-PB. The PB-PVS coating was prepared by rinsing for 30 min at 350 mbar with 10% (m/v) PB solution and successively with water for 5 min at 1380 mbar. Subsequently, the capillary was flushed with 5% (v/v) PVS solution for 30 min at 350 mbar, and again water for 5 min at 1380 mbar. At the start of the day, coated capillaries were flushed with deionized water for 1 min at 1380 mbar and with BGE for 2 min at 1380 mbar. Between runs, the coated capillaries were flushed with 5% (v/v) PVS for 5 min at 1380 mbar and with BGE for 3 min at 1380 mbar. The PB-DS-PB coating was prepared by subsequently rinsing the capillary with 10% (m/v) PB solution for 15 min at 350 mbar, deionized water for 5 min at 1380 mbar, 3% (m/v) DS solution for 15 min at 350 mbar, deionized water for 5 min at 1380 mbar, and, finally with a 10% (m/v) PB solution for 15 min at 350 mbar. The capillary was then ready for use with the BGE of choice. Before analysis, capillaries were flushed with BGE for 5 min at 1380 mbar. Between runs, the coated capillaries were flushed with a 1% (m/v) PB solution and BGE, each for 5 min at 1380 mbar.

86


2.3 CE-UV CE-UV experiments were performed on a Beckman Coulter PA 800 instrument (Beckman Coulter, Fullerton, CA, USA) equipped with a photo diode array detector. Fused-silica capillaries were from Composite Metal Services (The Chase, Hallow, UK) having an internal diameter of 50 μm. Separations were performed on a 130-cm long fused-silica capillary (length to detector was 120 cm). Capillaries were dynamically coated either with a double or a triple layer of charged polymers, as described in Section 2.2. The BGEs were 1 M formic acid (pH 2.0) or 25 mM ammonium acetate (pH 9.0). Samples were hydrodynamically injected at 0.5 psi for 30 s. The separation voltage was 30 kV. 2.4 CE-TOF-MS CE-MS experiments were performed on a Beckman Coulter PA 800 instrument (Beckman Coulter, Fullerton, CA, USA) coupled to a micrOTOF orthogonal-accelerated TOF mass spectrometer (Bruker Daltonics, Bremen, Germany) via a sheath-liquid electrospray interface from Agilent Technologies (Waldbronn, Germany). Separations occurred in an 100-cm long fused-silica capillary. A BGE of 1 M formic acid (pH 2.0) was used for the separation of cationic compounds. The sheath liquid was methanol/water (1/1, v/v) containing 0.1% FA and was infused using a 2.5 mL Hamilton syringe at 4 μL/min. A BGE of 25 mM ammonium acetate (pH 9.0) was used for the analysis of anionic compounds. The sheath liquid was methanol/water (1:1, v/v) containing 0.1% concentrated ammonium hydroxide and also infused at 4 μL/min. For all CE-MS experiments, samples were hydrodynamically injected at 0.5 psi for 30 s and the separation voltage was 30 kV. The electrospray voltage was -4.5 kV or + 4.0 kV. 2.5 Test mixture and urine samples A test mixture, comprising dopamine, adrenaline, creatinine, hippuric acid, orotic acid, uric acid, tryptophan, phenylalanine, glutathione, NADPH, FAD, guanosine and folic acid, was prepared and employed to evaluate the CE methods. Stock solutions of 1 mg/mL of each analyte were prepared by dissolving appropriate amounts in water. Aliquots of stock solutions were diluted with water in a 1.5 mL glass vial in order to obtain a working solution in which each analyte was present at 20 μg/mL. Stock solutions were kept at –20 ºC until usage. Rat urine samples were kindly provided by AstraZeneca (Dept. of Drug Metabolism and Pharmacokinetics, Macclesfield, United

87

Chapter 5

Kingdom). Urine samples were stored at -80°C prior to usage. Before analysis, urine samples were mixed with BGE (1:1, v/v) and centrifuged at 13,200 rpm for 5 min. Ion suppression effects were studied by the addition of test compounds to untreated and pretreated urine samples. The obtained peak areas for the test compounds were compared with the peak areas of test compounds dissolved in a mixture of BGE and water (1:1, v/v). 3. Results and discussion 3.1 CE-UV The performances of CE methods employing PB-PVS and PB-DS-PB coated capillaries were evaluated both at acidic and basic pH. In order to study the coverage and applicability of each method, first a mixture of metabolites was analyzed in a CE-UV set-up using volatile BGEs as required for CE-MS. The compounds used for the test mixture were selected with the aim to cover a range of polarities and molecular weights of the metabolites typically reported as components of body fluids. Table 1 shows the physicochemical properties of the compounds in the test mixture grouped in different chemical families, such as amino acids, catecholamines, organic acids, purines, nucleosides, nucleotides and small peptides. It is evident that the test compounds have a highly polar character. Table 1. Physico-chemical properties of the test mixture metabolites.

Classification Compound pKa1 acid pKa1 base Log P2 Amino acids L-Tyrosine 2.24 9.04 -2.39 L-Phenylalanine 2.18 9.09 -1.35 Catecholamines Dopamine 8.90 10.6 -0.40 Adrenaline 8.66 10.6 -0.82 Amine Creatinine - 9.20 -1.65 Organic acids Hippuric acid 3.60 - 0.24 Orotic acid 2.40 9.50 -0.89 Purine Uric acid 5.40 - -1.12 Nucleotides NADPH 6.05 3.50 -1.13 FAD 3.63 - -0.78 Vitamin Folic acid 3.50 8.38 -2.50 Nucleoside Guanosine 1.60 9.21 -2.06 Small peptide Glutathione 2.12 9.65 -2.74

1Data taken from CRC Handbook of Chemistry & Physics 89th Edition 2Predicted by ALOGPS

88


3.1.1 Cationic compounds When analyzing the test mixture with the PB-PVS CE-UV method using 1 M formic acid (pH 2.0) as BGE, compounds that are positively charged at pH 2.0 like creatinine, dopamine, adrenaline and tyrosine migrated before the EOF time (Figure 1A). Most acids are neutral at pH 2.0 and, therefore, co-migrate with the EOF. Hippuric acid, which is slightly negatively charged at pH 2.0, migrated directly after the EOF time. The EOF time was approximately 17.5 min, indicating that the effective separation window for the analysis of cationic compounds is ca. 8 min (starting from the migration time of creatinine). A

B Figure 1. CE-UV analysis of a test mixture of metabolites using (A) a PB-PVS coated capillary and (B) a PB-DS-PB coated capillary. Experimental conditions: BGE, 1 M Formic acid (pH 2.0); injection, 35 mbar for 10 s; detection wavelength, 200 nm.

89

Chapter 5

In principle, this CE method provides a comprehensive analysis of basic compounds as all compounds carrying an overall positive charge at low pH, such as amino acids, amines, nucleosides and small peptides will migrate before the EOF time. Comparable results were obtained by Soga et al. [20], who used a CE-MS method based on bare fused-silica capillaries for the separation of cationic metabolites at pH 1.8. In this method, the effective separation window for cationic compounds was ca. 15 min (due to the very low EOF). However, with bare fused-silica capillaries, we found considerable migration time instability for amino acids spiked in urine (RSDs of 5-15%, n=10) [22]. Table 2 gives the average migration times and corresponding RSD values for the test compounds in our method. RSDs were ≤1% for five repeated injections of the test mixture. Plate numbers ranged from 100,000 to 300,000 for the test compounds, except for folic acid and guanosine. Table 2. Migration times and RSDs (n=5) of the test compounds obtained with CE-UV using PB-PVS and PB-DS-PB coatings for cationic compounds using 1 M formic acid (pH 2.0) as BGE.

Compound PB-PVS coating PB-DS-PB coating MT (min.)1 RSDs (%) MT (min.)1 RSDs (%) Creatinine 10.3 0.8 55.9 0.9 Dopamine 11.5 0.5 38.4 0.7 Adrenaline 11.9 0.6 34.8 0.7 Phenylalanine 13.0 0.9 28.7 0.8 Tyrosine 13.2 0.9 28.2 0.9 Glutathione 13.7 1.0 26.6 0.8 Guanosine 14.3 0.9 nd2 nd2

1MT = migration time. 2nd = not determined due to co-migration.

With the PB-DS-PB CE-UV method using 1 M formic acid (pH 2.0) as BGE and reverse polarity, the positively charged compounds are expected to migrate after the EOF time and in a reversed migration order in comparison to what is observed in Figure 1A. The electropherogram shown in Figure 1B confirms these considerations. The obtained separation window was relatively large, for example, the migration time of creatinine was ca. 56 min whereas it was ca. 10 min in the PB-PVS CE-UV method (Table 2). With the PB-DS-PB CE method cationic compounds are analyzed on a capillary with a positively charged inner wall, thereby preventing potential electrostatic adsorption of analytes to the capillary wall. This is typically illustrated for guanosine and folic acid which injected separately provided plate numbers above 100,000. This was a significant improvement compared to the width of the peaks obtained for these compounds in the PB-PVS CE method (Figure 1A). In this system band broadening

90


occurred for these compounds which was due to adsorption to the negatively charged capillary wall, although this adsorption also contributed to their partial separation. A range of different metabolite classes can be covered with the PB-DS-PB CE method (i.e. nucleosides, amino acids, catecholamines, small peptides, amines), however, small and multivalent cationic compounds will not be covered due to their strong electrophoretic mobility towards the cathode (i.e., opposite to the direction of the EOF). RSDs for migration times of the test compounds were <1% (Table 2) and plate numbers ranged from 100,000 to 400,000 except for creatinine which showed a broad triangular shaped peak. 3.1.2 Anionic compounds When analyzing the test mixture with the PB-DS-PB CE-UV method using 25 mM ammonium acetate (pH 9.0) as BGE and reverse CE polarity, compounds that are negatively charged, like orotic acid, uric acid, hippuric acid and tyrosine migrated before the EOF time (Figure 2A). Most basic compounds are neutral at pH 9.0 and migrate with the EOF. The broad peak for FAD can be explained by electrostatic interaction with the positively charged capillary wall. NADPH could not be detected, most probably due to strong adsorption to the positively charged capillary wall. The same phenomenon has also been observed by Soga et al. [20]. The EOF time was ca. 25 min and somewhat longer than the EOF time observed with the PB-DS-PB CE method at pH 2.0. This can be explained by the difference in the ionic strength of the BGEs used for the low and high pH separation conditions. In principle, the PB-DS-PB CE method at high pH provides a comprehensive analysis of acidic compounds as all compounds carrying an acidic functionality, such as carboxylic acids, phosphorylated carboxylic acids, small peptides and nucleotides will migrate before the EOF time. However, anionic compounds are separated on a positively charged capillary wall which can lead to peak broadening (see above). RSDs for migration time of the test compounds were ≤1.6% (Table 3) and plate numbers ranged from 30,000 to 450,000. The adsorption of multivalent anionic compounds to the positively charged capillary wall can be prevented by using the PB-PVS CE-UV method with 25 mM ammonium acetate (pH 9.0) as BGE, as shown in Figure 2B. Compounds like FAD and NAPH appeared now as sharp peaks. An improved separation was also obtained for uric acid and glutathione which was probably due to a reduced interaction with the capillary wall. As organic acids, nucleotides, nucleosides, amino acids and small peptides were detected after the EOF time in the PB-PVS CE method, an increased separation window was obtained for anionic compounds. For example, the migration time of

91

Chapter 5

orotic acid is 50 min in the PB-PVS CE method while it is 13.8 min in the PB-DS-PB CE method (Table 3). A

B Figure 2. CE-UV analysis of a test mixture of metabolites on (A) PB-DS-PB coated capillary and (B) PB-PVS coated capillary. Experimental conditions: BGE, 25 mM ammonium acetate (pH 9.0); injection, 35 mbar for 10 s; detection wavelength, 200 nm. However, small and multiple charged anionic compounds (e.g., citric acid) will not be covered in the PB-PVS CE method due to their strong electrophoretic mobility towards the anode (i.e., opposite to the direction of the EOF). RSDs for migration times of the test compounds were <1.0% (Table 3) and the obtained plate numbers were comparable to those obtained with PB-DS-PB CE method at pH 9.0, except for multivalent compounds for which the plate numbers improved significantly.

92


Table 3. Migration times and RSDs (n=5) of the test compounds obtained with CE-UV using PB-PVS and PB-DS-PB coatings for anionic compounds using 25 mM ammonium acetate (pH 9.0) as BGE.

Compound PB-PVS coating PB-DS-PB coating MT (min.)1 RSDs (%) MT (min.)1 RSDs (%) Guanosine 20.8 0.9 20.9 0.9 Tyrosine 22.0 0.8 19.8 0.8 FAD 31.8 0.8 16.5 1.6 Hippuric acid 33.4 0.9 15.8 1.1 Uric acid 36.7 0.8 15.2 1.3 Orotic acid 50.0 0.7 13.8 0.8 Glutathione 37.2 0.9 nd2 nd2

NADPH 63.4 0.9 nd2 nd2

1MT = migration time. 2nd = not determined due to co-migration.

3.2 CE-MS We also evaluated the CE methods in combination with time-of-flight mass spectrometry (TOF-MS) using ESI and sheath-liquid interfacing. TOF-MS provides a high speed, high mass resolution and high mass accuracy with errors below 10 ppm, making it very suitable for analyses in metabolic profiling studies. Coupling of the PB-PVS and PB-DS-PB coated capillaries to ESI-TOF-MS did not affect the good migration time stability. RSDs for both cationic and anionic compounds at low and high pH were always below 2%. Plate numbers obtained with CE-MS for test compounds dissolved in water ranged from 50,000 to 400,000. In other words, transfer of the CE-UV method to CE-MS did not affect the separation performance of the CE systems. However, with the PB-DS-PB CE-MS method applying 25 mM ammonium acetate (pH 9.0) as BGE, we observed relatively low responses for the anionic test compounds. Therefore, the influence of the alkaline BGE on the MS signal of hippuric acid was examined by infusion experiments. Compared to the signal obtained when dissolved in water, the MS signal intensity of hippuric acid in BGE was clearly reduced, i.e., ca. 25% of the hippuric acid signal was left in 25 mM ammonium acetate (pH 9.0) and ca. 10% in 50 mM ammonium acetate (pH 9.0). This signal reduction can most probably be attributed to ion suppression caused by acetate, which was also observed by Yang et al. [26]. When we applied the PB-PVS CE-MS method for the analysis of anionic compounds, we also found relatively low analyte responses. Clearly, the CE-MS methods for the analysis of anionic compounds require further optimization from the viewpoint of ionization efficiency.

93

Chapter 5

For cationic compounds in positive ESI-MS mode, favorable analyte intensities were obtained and the BGE caused no ion suppression. Figure 3A shows a typical base peak electropherogram (BPE) of a rat urine sample obtained by the PB-PVS CE-MS method. The BPE shows that most compounds migrate between 10 and 20 min. The EOF time was 17.5 min providing a separation window for cationic compounds of about 8 min. The number of molecular features detected was approximately 300. With the PB-DS-PB CE-MS method an improved separation of compounds in rat urine was obtained as shown in Figure 3B.

A

B

Figure 3. Base peak electropherograms obtained during CE-MS analysis of rat urine using (A) a PB-PVS coated capillary and (B) a PB-DS-PB coated capillary. Conditions: BGE, 1 M Formic acid (pH 2.0); sample injection, 35 mbar for 60 s. Further conditions, see Experimental section.

94


Figure 4 shows an ion map from the analysis of the rat urine sample with the PB-DS-PB CE-MS method, giving an indication of the distribution of masses and migration times of the molecular features detected in this sample. This figure clearly reveals an

even distribution of molecular features throughout the whole separation time and m/z range. Using the same injection volume and capillary dimensions as in the PB-PVS CE-MS method, the number of molecular features detected was ca. 600. So, a larger number of molecular features was observed in a longer separation time.

Figure 4. An ion map of rat urine analyzed by CE-MS using a PB-DS-PB coated capillary. For experimental conditions, see Figure 3B. The influence of ion suppression on the MS responses of test compounds was also investigated. Ion suppression is generally caused by co-migration of sample (matrix) components as the simultaneous presence of multiple compounds in the ion source can result in competition in the ionization process and a subsequent reduction in the MS signals. The magnitude of ion suppression was determined by comparing peak areas for metabolite standards with different migration times with peak areas of metabolites spiked into urine samples. The test compounds were dissolved in the BGE and spiked into a pooled urine sample before sample pretreatment and after sample pretreatment. Subsequently, these samples were analyzed and the peak areas of the test compounds were determined. The peak areas were corrected for endogenous levels of the test compounds as determined by analyzing non-spiked pooled urine samples. The peak areas measured for the test compounds spiked into pretreated urine, i.e. urine mixed with BGE (1:1, v/v) and centrifuged, were similar to the peak areas found for the test compounds spiked into urine before pretreatment, thereby

95

Chapter 5

ruling out analyte losses during sample pretreatment. With the PB-PVS CE-MS method, the peak areas of the test compounds spiked into urine were significantly lower than the peak areas measured for the test compounds in BGE. A signal decrease of 30 to 60% depending on the test compound was observed (Figure 5). With the PB-DS-PB CE-MS method, the peak areas of the test compounds spiked into urine were also lower than the peak areas measured for the test compounds in BGE, but not as severe as with the PB-PVS CE-MS method. In this case, a signal decrease of 10–30%, depending on the test compound, was observed (Figure 5). This indicated that less ion suppression occurs in the CE-MS method with the longer separation time, which is most probably due to less co-migration of compounds in this method. This lower degree of ion suppression in the PB-DS-PB CE-MS method most probably also explains the higher number of molecular features observed.

0

20

40

60

80

100

Dopamine Tyrosine Hippuric acid

Mea

n s

ign

al (

%)

Figure 5. Relative mean signal intensity (n=3) of test compounds spiked in pooled urine obtained during CE-MS using a PB-PVS coated capillary (black bars) and a PB-DS-PB coated capillary (grey bars). The signal intensity was compared to the signal intensity obtained for the test compounds in BGE, which was set to 100%. As ion suppression may be different between urine samples, we have also analysed urine obtained from five different rats. Each urine sample was spiked with the test compounds (50 μM), and the peak areas of the test compounds were determined. For each test compound the peak area remained rather constant among the various urine samples with differences in peak area of not more than 10-20% and, therefore, comparable to the analytical variability of the CE-MS methods (which was about 10%). This rather constant matrix effect (i.e. similar ion suppression) is quite important in metabolic profiling studies where samples derived from many subjects have to be compared.

96


4. Concluding remarks CE-UV and CE-MS methods based on noncovalently coated capillaries have been evaluated for their potential to analyze a broad range of metabolites within a single run. The CE method based on a PB-PVS capillary coating and a low-pH BGE allowed the comprehensive analysis of basic compounds. Therefore, it can be used for the profiling of a broad array of metabolites, which is highly favorable in metabolic profiling studies. High plate numbers and good migration-time repeatability were obtained for test compounds spiked into urine and with minimal sample pretreatment. Approximately 300 molecular features were recorded in rat urine. The separation window was relatively small due to the fast analysis. The CE method based on a PB-DS-PB capillary coating and a low-pH BGE also provided high plate numbers and good migration-time repeatability for test compounds spiked into urine. An increased separation window was obtained as the cationic compounds migrated after the EOF time, resulting in the detection of ca. 600 molecular features in rat urine. Hence, this CE method can provide a more information-rich metabolic profile of urine compared to the PB-PVS CE-MS method, but at the cost of analysis time. Both CE methods were also used at high pH for the analysis of anionic compounds. It was found that the use of ammonium acetate as BGE caused ion suppression, thereby significantly reducing the signal of the anionic test compounds. Therefore, alternative strategies should be investigated in order to improve the detection sensitivity for these type of compounds. For instance, the use of a platinum spray needle instead of a stainless steel spray needle significantly improved the concentration sensitivity of CE-TOF-MS for anionic compounds and, therefore, it is very interesting to evaluate this approach for the profiling of anionic metabolites in body fluids [27]. It would be also interesting to study in-capillary preconcentration techniques, such as pH-mediated stacking, for anionic compounds [28]. Another way to improve the concentration sensitivity for anionic compounds is by using solid-phase extraction (SPE) columns. In this case, mixed-mode sorbents have favorable characteristics for the preconcentration of these type of compounds. In general, the four CE methods presented in this study provide the required migration time repeatability for metabolic profiling studies. The use of the same coated capillary for both the analysis of cationic and anionic compounds in urine can provide a flexible CE-MS approach to increase the coverage of metabolites. So far, the utility of the PB-PVS CE-MS method has been demonstrated for cationic profiling in a metabolomics context [22, 24]. However, such a study still needs to be performed for the PB-DS-PB CE-MS method regarding the analysis of clinical samples. Further research should also focus on the application of the CE-MS methods for the long-

97

Chapter 5

term metabolic profiling studies, i.e. to evaluate the performance of the four CE-MS methods for the analysis of a large cohort of clinically relevant samples. References [1] Van der Greef, J., Stroobant, P., Van der Heijden, R., Curr Opin Chem Biol

2004, 8, 559-565. [2] Nicholson, J. K., Lindon, J. C., Holmes, E., Xenobiotica 1999, 29, 1181-1189. [3] Nicholson, J. K., Connelly, J., Lindon, J. C., Holmes, E., Nat Rev Drug Discov 2002, 1, 153-161. [4] Fiehn, O., Kind, T., Methods Mol Biol 2007, 358, 3-17. [5] Ramautar, R., Somsen, G. W., de Jong, G. J., Electrophoresis 2009, 30, 276-

291. [6] Lenz, E. M., Wilson, I. D., J Proteome Res 2007, 6, 443-458. [7] T'Kindt, R., Storme, M., Deforce, D., Van Bocxlaer, J., J Sep Sci 2008, 31,

1609-1614. [8] Gika, H. G., Theodoridis, G. A., Wilson, I. D., J Sep Sci 2008, 31, 1598-1608. [9] Ullsten, S., Danielsson, R., Backstrom, D., Sjoberg, P., Bergquist, J.,

J Chromatogr A 2006, 1117, 87-93. [10] Monton, M. R., Soga, T., J Chromatogr A 2007, 1168, 237-246; discussion

236. [11] Ramautar, R., Demirci, A., de Jong, G. J., Trac-Trends in Analytical

Chemistry 2006, 25, 455-466. [12] Sato, S., Soga, T., Nishioka, T., Tomita, M., Plant J 2004, 40, 151-163. [13] Soga, T., Kakazu, Y., Robert, M., Tomita, M., Nishioka, T., Electrophoresis

2004, 25, 1964-1972. [14] Soga, T., Ueno, Y., Naraoka, H., Ohashi, Y., et al., Anal Chem 2002, 74,

2233-2239. [15] Soo, E. C., Aubry, A. J., Logan, S. M., Guerry, P., et al., Anal Chem 2004, 76,

619-626. [16] Mayboroda, O. A., Neususs, C., Pelzing, M., Zurek, G., et al., J Chromatogr A

2007, 1159, 149-153. [17] Lee, R., Ptolemy, A. S., Niewczas, L., Britz-McKibbin, P., Anal Chem 2007,

79, 403-415. [18] Desiderio, C., De Rossi, A., Inzitari, R., Mancinelli, A., et al., Anal Bioanal

Chem 2008, 390, 1637-1644. [19] Arvidsson, B., Johannesson, N., Citterio, A., Righetti, P. G., Bergquist, J.,

J Chromatogr A 2007, 1159, 154-158.

98


99

[20] Soga, T., Ohashi, Y., Ueno, Y., Naraoka, H., et al., J Proteome Res 2003, 2, 488-494.

[21] Garcia-Perez, I., Whitfield, P., Bartlett, A., Angulo, S., et al., Electrophoresis 2008, 29, 3201-3206.

[22] Ramautar, R., Mayboroda, O. A., Derks, R. J., van Nieuwkoop, C., et al., Electrophoresis 2008, 29, 2714-2722.

[23] Ramautar, R., Somsen, G. W., de Jong, G. J., Anal Bioanal Chem 2007, 387, 293-301.

[24] Ramautar, R., van der Plas, A. A., Nevedomskaya, E., Derks, R., et al., J Proteome Res 2009.

[25] Ramautar, R., Mayboroda, O. A., Deelder, A. M., Somsen, G. W., de Jong, G. J., J Chromatogr B Analyt Technol Biomed Life Sci 2008, 871, 370-374.

[26] Yang, W. C., Regnier, F. E., Adamec, J., Electrophoresis 2008, 29, 4549-4560. [27] Soga, T., Igarashi, K., Ito, C., Mizobuchi, K., et al., Anal Chem 2009. [28] Ptolemy, A. S., Britz-McKibbin, P., Analyst 2008, 133, 1643-1648.

Chapter 6

CE-TOF-MS using Noncovalently Bilayer-Coated Capillaries for the

Analysis of Amino Acids in Human Urine

R. Ramautar, O.A. Mayboroda, R.J.E. Derks, C. van Nieuwkoop, J.T. van Dissel,

G.W. Somsen, A.M. Deelder, G.J. de Jong

Electrophoresis 2008, 29, 2714-2722

CE-TOF-MS for the Analysis of Amino Acids in Human Urine

Abstract A capillary electrophoresis time-of-flight mass spectrometry (CE-TOF-MS) method for the analysis of amino acids in human urine was developed. Capillaries noncovalently coated with a bilayer of Polybrene (PB) and poly(vinyl sulfonate) (PVS) provided a considerable electro-osmotic flow at low pH, thus facilitating the fast separation of amino acids using a background electrolyte of 1 M formic acid (pH 1.8). The PB-PVS coating proved to be very consistent yielding stable CE-MS patterns of amino acids in urine with favourable migration time repeatability (RSDs <2%). The relatively low sample loading capacity of CE was circumvented by an in-capillary preconcentration step based on pH-mediated stacking allowing 100-nL sample injection (i.e., ca. 4% of capillary volume). As a result, detection limits for amino acids were down to 20 nM while achieving satisfactory separation efficiencies. Preliminary validation of the method with urine samples showed good linear responses for the amino acids (R2 >0.99), and RSDs for peak areas were <10%. Special attention was paid to the influence of matrix effects on the quantification of amino acids. The magnitude of ion suppression by the matrix was similar for different urine samples. The CE-TOF-MS method was used for the analysis of urine samples of patients with urinary tract infection (UTI). Concentrations of a subset of amino acids were determined and compared with concentrations in urine of healthy controls. Furthermore, partial least squares-discriminant analysis (PLS-DA) of the CE-TOF-MS

dataset in the 50-450 m/z region showed a distinctive grouping of the UTI samples and the control samples. Examination of score and loadings plot revealed a number of compounds, including phenylalanine, to be responsible for grouping of the samples. Thus, the CE-TOF-MS method shows good potential for the screening of body fluids based on the analysis of endogenous low-molecular weight metabolites such as amino acids and related compounds.

103

Chapter 6

1. Introduction The determination of free amino acids in human urine is important for the screening of deficiencies in amino acid metabolism in routine clinical analysis [1, 2]. Amino acid profiles are useful not only for the detection of relatively simple metabolic disorders, such as phenylketonuria, but could contribute significantly to diagnosis of malignancies, nutritional disorders or chronic fatigue syndrome. The analysis of amino acids is also important in the field of metabolomics where metabolic profiles of body fluids are measured with the aim to find potential diagnostic markers for diseases [3]. To date, the most commonly used techniques for the profiling of amino acids in body fluids, such as urine and cerebrospinal fluid, are ion-exchange liquid chromatography (IE-LC) and reversed-phase liquid chromatography (RPLC) [4]. Using IE-LC, post-column derivatization of the amino acids is normally required for detection, which is a complex procedure. Moreover, analysis times are relatively long and the applicability of this method is restricted to the amino acids. RPLC combined with mass spectrometry (MS) has also been used for amino acid analysis [5, 6]. However, pre-column derivatization or the use of volatile ion-pairing agents is necessary to prevent amino acids to co-elute with the void volume in RPLC [6, 7]. Capillary electrophoresis (CE) is a powerful alternative for the analysis of amino acids exhibiting a selectivity that is based on charge-to-mass ratio and, therefore, complementary to RPLC. Furthermore, CE offers high separation efficiencies and speed. Obviously, the coupling of CE with MS is very attractive for amino acid analysis, combining high-resolution separation with high detection selectivity. In addition, amino acids can be analysed without any derivatization using CE-MS. So far, several groups have demonstrated the potential of CE-MS for the analysis of amino acids [8-14]. For instance, CE coupled to a single quadrupole mass analyzer in SIM mode was developed for the determination of amino acids in soy sauce [11]. The influence of background electrolyte (BGE) composition was studied on resolution and peak shapes of amino acids were optimized. Detection limits for standard solutions of amino acids were between the 0.3 and 6.0 μM and the method was applied to the quantification of amino acids in deproteinized and 100 times diluted soy sauce. CE coupled to a triple quadrupole mass analyzer applying multiple reaction monitoring resulted in amino acid detection limits of 0.1-14 μM for standard solutions [12]. The method was used for the quantification of amino acids in 5 times diluted human urine. Migration-time repeatability and the influence of urine matrix components on the analysis of amino acids were not reported. Using a sheathless ESI interface for CE coupled to an ion trap MS, detection limits in the nM range were obtained for standard solutions of amino acids [13]. The analysis time for the amino acids was

104


approximately 40 min and the robustness of this method was not demonstrated. More recently, CE has been coupled to an orthogonal time-of-flight (TOF) instrument for amino acids analysis [14]. The high mass resolution of a TOF-MS instrument provides a better selectivity compared to quadrupole mass analyzers. The system has been used for the profiling of amino acids in human urine with minimal sample pretreatment. Detection limits in the mid nM-range for amino acids were achieved through the use of pH-mediated stacking. However, as with the aforementioned studies, analysis times were relatively long due to the slow electro-osmotic flow (EOF) resulting from the low-pH background electrolyte (BGE) in combination with bare fused silica capillaries. Obviously, this is not very suitable for clinical studies where high-throughput analyses are required [15]. Moreover, using bare fused silica capillaries, separation efficiency and reproducibility may be compromised. Changes in the surface of the capillary wall due to adsorption of matrix components may cause irreproducible EOFs leading to a poor migration-time repeatability [16]. Recently, capillaries noncovalently coated with a bilayer of oppositely charged polymers have been used for the fast and highly reproducible analysis of peptides [17]. The coated capillaries were produced by first rinsing the capillary with a solution of the positively charged polymer Polybrene (PB) and subsequently with a solution of the negatively charged polymer poly(vinylsulfonate) (PVS). The resulting PB-PVS coating provided a considerable EOF at low pH, thereby facilitating the fast separation of positively charged peptides using a BGE of formic acid (pH 2.5). The coating prevents protein adsorption and is fully compatible with MS detection causing no ionization suppression [17]. Hence, the use of noncovalently coated capillaries in combination with TOF-MS may be a very attractive approach for the fast analysis of amino acids. The aim of the present study is to develop a fast, highly repeatable and sensitive method for the analysis of amino acids in human urine using a PB-PVS CE-TOF-MS system. An in-capillary preconcentration strategy is evaluated for improvement of the concentration sensitivity for the amino acids. Special attention is given to the influence of matrix effects (ion suppression) on the quantification of amino acids and on the analytical validation of the system. To demonstrate the potential of the PB-PVS CE-TOF-MS system for the analysis of amino acids, urine samples from patients with urinary tract infection are analysed. Moreover, multivariate data analysis is applied to the CE-TOF-MS data over the entire mass range to illustrate the potential of the method for nontargeted metabolic profiling of urine.

105

Chapter 6

2. Materials and methods 2.1 Chemicals PB (hexadimethrine bromide) and a 25% m/v aqueous solution of PVS sodium salt were from Sigma-Aldrich (Steinheim, Germany). A 10% m/v PB solution was prepared by diluting PB with Milli-Q water. A 5% PVS solution was made by diluting the purchased PVS solution with Milli-Q water. The polymer solutions were passed through a 0.22 µm filter and stored at 4ºC. Formic acid was from Riedel de Haën (Seelze, Germany). Ammonium hydroxide 25% was from Merck (Darmstadt, Germany). Amino acids (L-Alanine, L-Arginine, L-Glutamic acid, L-Isoleucine, L-Leucine, L-Lysine, L-Methionine, L-Phenylalanine, L-Tyrosine, L-Valine) with a concentration of 1 mM of each in 0.1 M hydrochloric acid were purchased from Sigma-Aldrich. HPLC-grade methanol was supplied by Biosolve BV (Valkenswaard, the Netherlands). Standard solutions were prepared with water taken from a Milli-Q water purification system (Millipore, Bedford, MA, USA). BGE solution was prepared by dissolving formic acid in Milli-Q water (1 M, pH 1.8). 2.2 Urine samples Urine samples of patients suffering from urinary tract infections and urine samples from a healthy control group were obtained from the department of Infectious Diseases of the Leiden University Medical Centre. After informed consent was obtained (Institutional Review Board protocol no. P06.061), patients with a fever of 38.5 ºC or higher, dysuria and costovertebral tenderness (i.e., those with urinary tract infection and a clinical diagnosis of pyelonephritis) were enrolled into the study and samples of urine obtained on presentation. The diagnosis was confirmed by positive urine and/or blood culture. Moreover, urine samples were obtained from healthy controls without a medical history or symptoms of a urinary tract infection; moreover, controls did not have asymptomatic bacteriuria. After collection, the urine samples were stored immediately at -80 ºC. Prior to CE-MS analysis, the urine samples were thawed to room temperature, mixed with BGE (1:1, v/v) and centrifuged (13,200 rpm) for 5 min. 2.3 Instrumentation and procedures The CE experiments were carried out on a Beckman Coulter PA 800 instrument (Beckman Coulter, Fullerton, CA, USA). Fused-silica capillaries were from Composite

106


Metal Services (The Chase, Hallow, UK) having a total length of 130 cm and an internal diameter of 50 μm. Formic acid (1 M, pH 1.8) was used as BGE. Next, the optimized experimental conditions are described. Sample injections were performed hydrodynamically for 90 s at 90 mbar (1.3 psi). Prior to sample injection a small plug (50 mbar, 9 s) of 12.5% ammonium hydroxide was injected for pH-mediated stacking. Injection volumes were determined using the CE Expert program from Beckman Coulter. The separation voltage was 30 kV and the capillary temperature was 25 °C. New bare fused-silica capillaries were rinsed with deionized water for 5 min at 1380 mbar followed by 1 M sodium hydroxide for 15 min at 1380 mbar, and deionized water for 5 min at 1380 mbar. Coating was performed by rinsing for 30 min at 350 mbar with 10% (m/v) PB and successively with water for 5 min at 1380 mbar. Subsequently, the capillary was flushed with 5% (v/v) PVS for 30 min at 350 mbar, and again water for 5 min at 1380 mbar. At the start of the day, coated capillaries (see below) were flushed with deionized water for 1 min at 1380 mbar and with BGE for 2 min at 1380 mbar. Between runs, the coated capillaries were flushed with a solution of PVS for 2 min at 1380 mbar and with BGE for 1 min at 1380 mbar. During the flushing with PVS, MS source settings were adjusted (ESI voltage was 0 kV, nebulizer pressure was 0 bar, and the end plate offset was 0 volt) in the sequence method to prevent PVS to accumulate in the source. MS was performed using a micrOTOF orthogonal-accelerated time-of-flight (TOF) mass spectrometer (Bruker Daltonics, Bremen, Germany). Transfer parameters were optimized by direct infusion of an ESI tuning mix (Agilent Technologies, Waldbronn, Germany). Spectra were collected with a time resolution of 1 s. Post-run internal mass calibration was performed using sodium formate cluster ions Na+(HCOONa)1-9

ranging from 90.9766 to 430.9137 m/z, which are detected in the first part of the electropherogram after urine sample injection. CE-MS coupling was realized by a co-axial sheath liquid interface (Agilent Technologies, Waldbronn, Germany) with methanol-water-formic acid (50:50:0.1, v/v/v) as sheath liquid. The following spray conditions were used: sheath liquid flow, 4 μL/min; dry gas temperature, 180°C; nitrogen flow, 4 L/min; nebulizer pressure, 0.5 bar. Electrospray in positive ionization mode was achieved and ESI voltage was -4.5 kV. 2.4 Preliminary method validation Linearity of response for amino acids in 1:1 diluted BGE was evaluated by measurement of five analyte concentrations ranging from 0.5 (or 5) to 250 μM. Linearity of response for the amino acids was also measured in pooled human urine in the same concentration range. The limit of detection (LOD) for amino acids in BGE

107

Chapter 6

was calculated as the concentration producing a signal-to-noise ratio (S/N) of 3. The repeatability of the method was determined by 10 consecutive analyses of a pooled human urine sample spiked with 50 μM of each of the tested amino acids. Recovery was determined by the addition of known amounts of amino acids to urine. The recovery of the added amount was calculated using the calibration curves obtained for the amino acids in urine, and corrected for endogenous amino acid levels. Potential ion suppression effects were studied by the addition of amino acids to crude and pretreated urine samples. The obtained signal intensities for the amino acids were compared with the signal intensities of amino acids in 1:1 diluted BGE. 2.5 Data analysis CE-MS data were analysed using Bruker Daltonics Data Analysis software. Determination of amino acid concentrations was based on peak areas of amino acids from extracted ion electropherograms of urine samples. Amino acids concentrations in urine from healthy controls and urine from UTI patients were statistically compared at a confidence level of 95% with ANOVA. For multivariate data analysis, Bruker baf files were converted to NetCDF format. Each file was subsequently processed using MetAlign software (RIKILT, Wageningen, the Netherlands) for baseline correction. The output data were organized as Comma Separated Values (CSV) files in the form of arbitrary peak index (migration time – m/z pair), sample names and peak intensity information and analysed with partial least squares-discriminant analysis (PLS-DA) using SIMCA-P software (Umetrics, Umea, Sweden). Compounds that were significant for the classification were selected on the basis of the VIP (variable importance in the projection, more than 1.5) parameter and P-value (<0.05) in t-test. 3. Results and discussion 3.1 CE-MS method development The focus in this study was on the development of a fast, highly repeatable and sensitive CE-TOF-MS method for the analysis of amino acids in human urine. For method development ten representative amino acids (L-Alanine, L-Arginine, L-Glutamic acid, L-Isoleucine, L-Leucine, L-Lysine, L-Methionine, L-Phenylalanine, L-Tyrosine, L-Valine) were selected with different physico-chemical properties, i.e. amino acids with acidic, basic, aliphatic and aromatic side groups. To ensure that all amino acids are positively charged, the pH of the BGE should be strongly acidic as the isoelectric points of amino acids range from 2.77 to 10.76 [10]. At a pH lower

108


than 2.8 all amino acids will migrate faster than the EOF towards the capillary outlet. Using bare fused-silica capillaries in combination with a low-pH BGE, separation of amino acids would lead to long analysis times due to the very low EOF, and the resulting migration time repeatability might be poor. Therefore, we used PB-PVS-coated capillaries to speed up the analysis in a highly repeatable way. Formic acid was selected as volatile acid for the BGE as it has shown useful in other CE-MS studies regarding amino acid analysis [10, 11]. The formic acid concentration was optimized between 0.1 and 1 M with respect to separation efficiency. Most favorable plate numbers for the amino acids were obtained at a concentration of 1 M formic acid (pH 1.8) with values between 40,000 and 50,000 for singly charged amino acids and slightly lower for the basic (doubly charged) amino acids (lysine and arginine). The latter might be due to electrostatic interactions with the capillary coating. The ten amino acids were analysed within 15 min using the PB-PVS coated capillary. The EOF was

about 310-4 cm2/Vs under these conditions. Most amino acids showed different migration times, but were not completely baseline separated, whereas leucine and isoleucine fully co-migrated. Still, the amino acids could be selectively detected with a high mass accuracy (better than 5 mDa) using the TOF-MS instrument (except for leucine/isoleucine). This high mass accuracy is very important as it largely avoids the false annotation of target amino acids in the presence of co-migrating matrix components of similar molecular weight. Amino acid separations can in principle be improved by addition of organic solvents, like methanol, to the BGE [14]. However, this slows down the EOF, and thus increased analysis times. Furthermore, 10% methanol in the BGE affected the stability of the PB-PVS coating as migration times of the amino acids increased during subsequent runs (using flushing with PVS between runs). Using an injection volume of approximately 25 nL (~1% of the capillary volume), the limit of detection (S/N=3) was around the 1 μM for most amino acids. This may not be sufficient for the detection of amino acids in some clinical samples. In order to improve the concentration sensitivity, we studied the possibility of pH-mediated stacking, which is an in-capillary preconcentration technique [18, 19]. In the present study, stacking was achieved by the pre-injection of a short plug of ammonium hydroxide followed by injection of the sample acidified with BGE (1:1, v/v). Briefly, upon applying a high voltage, positively charged amino acids will migrate into the basic plug and become neutral or negatively charged. As a result, the amino acids will stack at the boundary of the sample and ammonium hydroxide plug until the basic plug is neutralized and acidified by the BGE. To optimize the pH-mediated stacking procedure for amino acids, various ammonium hydroxide concentrations and sample injection volumes were examined. Figure 1 shows the influence of the ammonium

109

Chapter 6

hydroxide concentration of a ca. 9 nL plug on the plate number of valine and phenylalanine. Raising the ammonium hydroxide concentration caused an increase of the plate numbers of the amino acids, which indicates that a peak sharpening takes place.

0

100000

200000

300000

400000

500000

0 6,25 12,5 25

ammonium hydroxide concentration (%)

Pla

te n

um

ber

Valine

Phenylalanine

Figure 1. Influence of the ammonium hydroxide concentration in the pre-injection plug on the plate number of valine (grey) and phenylalanine (black) obtained with CE-TOF-MS. Conditions: BGE, 1 M Formic acid (pH 1.8); sample injection, 35 mbar for 60 s; pre-injection, ammonium hydroxide at 50 mbar for 9 s. Further conditions, see Materials and methods section. Note: 0% ammonium hydroxide reflects no pre-injection. As migration times also increased, an ammonium hydroxide concentration of the pre-injection plug of 12.5% was selected for pH-mediated stacking. Figure 2 shows the influence of the injection volume on the signal intensities of valine and phenylalanine when applying a pre-injection of ca. 9 nL of 12.5% ammonium hydroxide. Upon increasing the injection volume, initially a considerable increase of the analyte signals is observed. From an injection volume of about 100 nL the gain in signal starts to be more gradual. This could be attributed to peak broadening as the plate numbers of the amino acids started to decrease with injection volumes above 100 nL. On the basis of these results, an injection volume of 100 nL (~4% of capillary volume) was selected for further experiments. Figure 3 shows the extracted-ion electropherograms of a test mixture of ten amino acids obtained with CE-TOF-MS using the optimized pH-mediated stacking procedure. Compared with an injection volume of 25 nL without

110


pH-mediated stacking, the signal intensities of the amino acids were more than a factor of ten higher, yielding detection limits of 20 - 215 nM for the amino acids, which is sufficient for their analysis in clinical samples.

0

200000

400000

600000

800000

0 50 100 150 200

Injection volume (nL)

sig

nal

in

ten

sity

(C

ou

nts

)

Valine

Phenylalanine

Figure 2. Influence of injection volume on the signal intensity of valine (grey triangles) and phenylalanine (black squares) obtained with CE-TOF-MS. Conditions: BGE, 1 M Formic acid (pH 1.8); Injection, 35 mbar for 60, 120, 240, 480 s; pre-injection, ammonium hydroxide (12.5%) at 50 mbar for 9 s. Further conditions, see Materials and methods section.

Using the PB-PVS CE-TOF-MS method for the analysis of urine spiked with ten amino acids (50 μM each), migration times and plate numbers of the amino acids were similar to the migration times and plate numbers of amino acids obtained for standard experiments. The amino acids in urine could be selectively detected with the TOF-MS instrument. Furthermore, signal intensities of the amino acids spiked in urine were also increased more than ten times using pH-mediated stacking. This indicates that the sample matrix does not have an influence on pH-mediated stacking. The stability of the PB-PVS capillary coating was examined by determining the migration time repeatability of amino acids which were added to a pooled human urine sample at a concentration of 50 μM for each. For a reliable comparison of profiles and to be able to observe small changes in sample composition, migration time stability is of crucial importance.

111

Chapter 6

Figure 3. Extracted ion electropherograms obtained during PB-PVS CE-TOF-MS analysis of an amino acid test mixture (Concentration, 50 μM each). Peaks: 1, Lysine; 2, Arginine; 3, Alanine; 4, Valine; 5 and 6, Isoleucine and Leucine; 7, Methionine; 8, Glutamic acid; 9, Phenylalanine; 10, Tyrosine. Conditions: BGE, 1 M Formic acid (pH 1.8); sample injection, 90 mbar for 90 s; pre-injection, ammonium hydroxide (12.5%) at 50 mbar for 9 s. Figure 4 shows electropherograms of the repeated analysis of the spiked urine sample. Very stable profiles were obtained, which is confirmed by the constant migration times of the amino acids as shown in Table 1. The good migration-time repeatability for amino acids in urine samples can primarily be attributed to the PB-PVS capillary coating, as with bare fused-silica capillaries considerable migration time shifts were obtained for amino acids spiked in urine (RSDs for migration-time repeatability of the amino acids were between the 5 and 15%, n=10). In summary, the developed PB-PVS CE-TOF-MS method allows the fast, sensitive and repeatable analysis of a subset of amino acids in urine without derivatization and with minimal sample pretreatment.

112


Figure 4. Repeated CE-TOF-MS analysis (n=10) of a pooled human urine sample spiked with 10 amino acids (50 M of each amino acid). Base peak electropherograms (m/z 50-450) of spiked urine. Conditions: BGE, 1 M Formic acid (pH 1.8); sample injection, 90 mbar for 90 s; pre-injection, ammonium hydroxide (12.5%) at 50 mbar for 9 s. 3.2 Matrix effects As we wanted to use the CE-TOF-MS method for the quantitative determination of amino acids in urine, we studied the potential effect of urine matrix components on the peak areas of amino acids. For that purpose, amino acids were spiked into a pooled urine sample before sample pretreatment and after sample pretreatment, and also into BGE. Subsequently, these samples were analysed and the peak areas of the amino acids were determined (Figure 5). The peak areas were corrected for endogenous amino acid levels as determined by analyzing blank urine as well. The peak areas measured for the amino acids spiked into pretreated urine, i.e. urine mixed

113

Chapter 6

with BGE (1:1) and centrifuged, were very similar to the peak areas found for the amino acids spiked into the urine before pretreatment, thereby ruling out analyte losses during sample pretreatment. This could be expected as the pretreatment is very simple comprising just a 1:1 dilution with BGE and a centrifugation. However, the peak areas of the amino acids spiked into urine are significantly lower than the peak areas measured for the amino acids spiked into BGE. This decrease must be caused by ionization suppression by co-migrating urine matrix components. Ionization suppression was observed for all tested amino acids leading to a signal decrease of 30 – 60% depending on the specific amino acid.

0

1000000

2000000

3000000

4000000

5000000

Lysine Arginine Valine Phenylalanine Tyrosine

Pea

k ar

ea (

Co

un

ts)

Figure 5. Peak areas obtained for selected amino acids by CE-TOF-MS after their spiking (50 μM each) into BGE (grey), in urine before pretreatment (black) and in urine after pretreatment (white). Conditions: BGE, 1 M Formic acid (pH 1.8); sample injection, 90 mbar for 90 s; pre-injection, ammonium hydroxide (12.5%) at 50 mbar for 9 s. Mean and standard errors of three experiments are depicted. As ion suppression may be different between urine samples, we analysed urine obtained from five healthy individuals. Each of the five urine samples was spiked with amino acids (50 μM each), and the peak areas of the amino acids were determined. For each individual amino acid the obtained peak area remained quite constant among the various urine samples with differences in peak area of not more than 10%.

114


Actually, the variability caused by ionization suppression differences between urine samples is very small since the analytical variability of the method is about 10% (as determined by repeated analysis of the same spiked urine sample; see next section). Therefore, we measured calibration curves for the amino acids in pooled human urine samples and used these for the quantification of amino acids in the individual urine samples. 3.3 Preliminary method validation In order to apply the developed PB-PVS CE-TOF-MS method for the quantitative determination of amino acids in urine, repeatability, linearity, limits of detection (LODs) and recovery were determined. The repeatability of the method was examined by ten consecutive injections of urine spiked with amino acids (50 μM each). Favourable RSDs for migration times (<1.5%) of the amino acids were obtained and RSDs for peak areas (<10%) were satisfactory. Calibration plots of amino acids in urine showed correlation coefficients (R2) above 0.99 in the concentration range of 0.5 (or 5) – 250 μM (Table 1). Table 1. Mean migration times (MT), RSDs for migration times, linearity and LODs of amino acids in pooled human urine.

Amino acid MT(min.)1 RSDs (min.)1

Linear range (μM) R2 LOD (nM)

Lysine 10.8 1.2 5 - 250 0.991 210 Arginine 11.0 1.5 5 - 250 0.991 280 Alanine 11.2 1.2 0.5 - 250 0.992 185 Valine 11.4 1.1 0.5 - 250 0.995 225 Methionine 12.1 1.1 0.5 - 250 0.992 180 Glutamic acid 12.4 1.2 0.5 - 250 0.991 200 Phenylalanine 12.7 0.8 0.5 - 250 0.994 85 Tyrosine 12.9 0.9 0.5 - 250 0.991 150

1for 10 consecutive analyses of urine spiked with amino acids (50 M). LODs for amino acids in BGE were calculated as the concentration yielding a signal-to-noise ratio (S/N) of 3 via extrapolation of the S/N-ratio produced by the lowest concentration used for the construction of calibration plots of amino acids in BGE. The LODs for amino acids in BGE varied from 20 to 215 nM (S/N=3). LODs of amino acids in urine were determined by measuring signal intensities of each amino acid spiked into urine at a concentration of 50 μM. These signal intensities were corrected for the signal intensities of the endogenous amino acid concentrations in

115

Chapter 6

urine. The corrected signal intensities were used for calculating the LODs (S/N=3), which varied from 85 to 280 nM for the amino acids in urine (Table 1). The recovery of the amino acids from urine was determined by adding known amounts of each amino acid to a pooled human urine sample at three concentration levels, namely 5, 25

and 100 M. These concentrations were chosen as the average concentrations of

amino acids in human urine vary from 1 to 1000 M. The recoveries of the added amounts were calculated using the calibration curves constructed in pooled human urine, and corrected for the endogenous concentrations. The data in Table 2 show good recoveries for all amino acids. Table 2. Recovery (%) of amino acids in spiked urine samples at three concentration levels (5, 25 and 100 M). For the recovery, mean and standard deviation of five experiments are shown.

Amino acid 5 M (n=5) 25 M (n=5) 100 M (n=5) Lysine 81 ± 11 80 ± 9 92 ± 8 Arginine 75 ± 9 78 ± 11 90 ± 9 Valine 89 ± 10 93 ± 9 99 ± 7 Phenylalanine 84 ± 9 88 ± 10 88 ± 9 Tyrosine 89 ± 8 95 ± 7 95 ± 8 Methionine 87 ± 11 89 ± 9 90 ± 8 Alanine 87 ± 8 88 ± 9 92 ± 8 Glutamic acid 81 ± 10 85 ± 10 91 ± 7

3.4 Applicability In order to test the applicability of the developed PB-PVS CE-TOF-MS method, we analysed 9 urine samples of patients suffering from urinary tract infection (UTI) and 9 urine samples from a control group. Urinary tract infection is a heterogeneous disease, which can be grouped into acute uncomplicated UTI (women with signs and symptoms of UTI with no evidence of urological abnormalities), complicated UTI (sexual and/or urological abnormalities), and those with asymptomatic bacteriuria (pregnant women, diabetics) [20, 21]. The pathogens usually associated with UTI are Escherichia coli, Staphylococcus saprophyticus, Klebsiella and Proteus sp. Complicated urinary tract infection like pyelonephritis is the most dangerous form of UTI. If therapeutic measures are not taken promptly such an infection may develop rapidly into urosepsis and next progress into a septic shock and multiple organ failure. The mortality of the latter condition is high (i.e., about 50%). Urosepsis as a form of systemic inflammation affects general metabolism and leads to redistribution of the free amino acids pool,

116


affecting their rate of secretion. Therefore, we initially focused on the determination of the amino acids concentrations in the two sets of urine samples. The profiles of the total ion electropherograms (TIE) of the urine samples from the patients were quite different from the TIE profiles of the urine samples from the control group. The migration times of the individual amino acids in the different urine samples were constant with RSDs below 2%. The concentrations of eight amino acids in the urine samples from the control group and the patients are presented in Table 3 (Isoleucine and leucine were not separated). The concentrations of the eight amino acids in the control group correspond with normal values given in the literature [22, 23]. The huge standard deviations for the amino acid concentrations are due to the large biological variability of amino acid concentrations among individuals [4]. The concentrations of most amino acids in the patient and control samples were similar. Table 3. Concentrations of amino acids in healthy controls (μM) and patients suffering from urinary tract infections (μM). Mean values and standard deviations are shown.

Amino acid Healthy controls (n=9) Patients (n=9) Lysine 53 ± 45 48 ± 46 Arginine 15 ± 13 24 ± 22 Valine 18 ± 15 10 ± 10 Phenylalanine* 14 ± 13 41 ± 23 Tyrosine 13 ± 10 29 ± 25 Methionine 1.3 ± 1.2 1.4 ± 1.0 Alanine 53 ± 35 99 ± 74 Glutamic acid* 4.0 ± 2.7 12 ± 8.6

* Significantly different in patients and healthy controls (P < 0.05). The concentrations of phenylalanine and glutamic acid, however, were significantly increased in the patient samples. The concentrations of alanine and tyrosine in urine samples of the patients with UTI were higher than the concentrations for healthy controls, but this increase was not significant. As the excretion rates of amino acids may vary independently of creatinine over a 24-h period, especially for patients with complicated urinary tract infections, normalization to creatinine has no practical meaning [24]. Therefore, the concentrations of the amino acids were not normalized for creatinine to account for variations in urinary output volumes. Metabolic abnormalities associated with complicated urinary tract infection are not restricted to the changes in amino acids urinary excretion only. Since our CE-TOF-MS method provides much more information, we subjected the CE-TOF-MS data,

which were acquired over the m/z region from 50 to 450, to partial least squares- discriminant analysis (PLS-DA). PLS-DA is a regression extension of PCA that takes

117

Chapter 6

advantage of class information to maximize the separation between groups of observations. The PLS-DA score plot demonstrates an obvious visual clustering obtained on the basis of a two class model (Figure 6). The PLS-DA loadings plot

revealed that several compounds with various m/z values were responsible for this classification. The influence of each of these compounds was analysed by VIP (variable importance in projection, which reflects the importance of the compound variables in the projection). The VIP analysis indicated that 9 compounds were

important for the classification and their m/z values are 105.00, 133.03, 146.08, 153.07, 166.07, 168.07, 215.02, 218.99 and 414.89. Based on the exact mass results of the

TOF-MS determination and the migration time information, the compound with m/z 166.07 could be identified as phenylalanine. The identification of the other compounds requires MS/MS experiments combined with database searching (Human

Metabolome and METLIN Database). For compounds with m/z values 133.03, 146.08, and 218.99 only one possibility was provided by these databases. Therefore,

the compounds with m/z 133.03, 146.08 and 218.99 may be L-asparagine (amino acid), allysine (a derivative of lysine), and glutamylalanine (a dipeptide), respectively. The interpretation of the biological significance of the results is outside the scope of the current study.

Figure 6. PLS-DA score plot of urine samples from healthy controls (diamonds) and patients with UTI (circles) using a two class model.

118


4. Concluding remarks A CE-TOF-MS method using PB-PVS coated capillaries was developed for the analysis of amino acids in human urine. The PB-PVS capillary coating generated a relatively high and pH-independent EOF enabling the fast analysis of amino acids by CE-TOF-MS at low pH. High plate numbers and favorable migration-time repeatability were obtained for the amino acids in urine, without derivatization and with minimal sample preparation. These merits can be primarily attributed to the use of a PB-PVS coated capillary. The rather low resolution of the method is due to the fast analysis of the amino acids, but this was compensated by the high mass accuracy of the TOF-MS instrument, allowing the selective detection of amino acids in urine, except for the isobaric amino acids such as leucine and isoleucine. The sensitivity of the method was improved by the optimization of a pH-mediated stacking procedure resulting in LODs for amino acids in urine in the nM range. The method has shown to be feasible for the fast and quantitative determination of eight amino acids in urine samples from patients with UTI. An important characteristic of the PB-PVS CE-TOF-MS method is that it provides very stable and highly repeatable separations of urine samples in time. This is very important when various samples have to be profiled and compared for clinical studies. The potential for such a goal was illustrated by the metabolic profiling of urine samples of patients with UTI. With our procedure it was possible to differentiate

control samples from UTI samples and to identify the m/z values of the compounds which are responsible for the classification using PLS-DA. PB-PVS CE-TOF-MS is also suitable for the metabolic profiling of other complex sample matrices, in particular cerebrospinal fluid, for which only limited sample amounts are available. References [1] Blom, W., Huijmans, J.G.M., Amino Acids 1992, 2, 25-67. [2] Prata, C., Bonnafous, P., Fraysse, N., Treilhou, M. et al., Electrophoresis 2000, 22, 4129-4138. [3] Ellis, D.I., Dunn, W.B., Griffin, J.L., Allwood, J.W. et al., Pharmacogenomics

2007, 8, 1243-1266. [4] Venta, R., Clin. Chem. 2001, 47, 575-583. [5] Chaimbault, P., Petritis, K.N., Elfakir, C., Dreux, M., J. Chromatogr. A 1999, 855, 191-202. [6] Sarwar, G., Botting, H.G., J. Chromatogr. 1993, 615, 1-22. [7] Petritis, K.N., Chaimbault, P., Elfakir, C., Dreux, M., J. Chromatogr. A 1999,

119

Chapter 6

120

833, 147-155. [8] Lu, W., Yang, G., Cole, R.B., Electrophoresis 1995, 16, 487-492. [9] Martin-Girardeau, A., Renou Gonnord, M.F., J. Chromatogr. A 2000, 742, 163-

171. [10] Klampfl, C.W., Ahrer, W., Electrophoresis 2001, 22, 1579-1584. [11] Soga, T., Heiger, D.N., Anal. Chem. 2000, 72, 1236-1241. [12] Soga, T., Kakazu, Y., Robert, M., Tomita, M. et al., Electrophoresis 2004, 25, 1964-1972. [13] Schultz, C.L., Moini, M., Anal. Chem. 2003, 75, 1508-1513. [14] Mayboroda, O.A., Neusüss, C., Pelzing, M., Zurek, G. et al., J. Chromatogr. A 2007, 1159, 149-153. [15] Petersen, J.R., Okorodudu, A.O., Mohammad, A., Payne, D.A., Clin. Chim.

Acta 2003, 330, 1-30. [16] Ramautar, R., Somsen, G.W., de Jong, G.J., Anal. Bioanal. Chem. 2007, 387, 293-301. [17] Catai, J.R., Torano, J.S., de Jong, G.J., Somsen G.W., Electrophoresis 2006, 27, 2091-2099. [18] Imami, K., Monton, M.R., Ishihama, Y., Terabe, S., J. Chromatogr. A 2007, 1168, 250-255. [19] Neusüss, C., Pelzing, M., Macht, M., Electrophoresis 2002, 18, 3149-3159. [20] Lee, J.B.L., Neild, G.H., Medicine 2007, 35, 423-428. [21] Fünfstück, R., Ott, U., Naber, K.G., Int J Antimicrob Agents 2006, 28S, S72-

S77. [22] Fekkes, D., Voskuilen-Kooyman, A., Jankie, R., Huijmans, J., J. Chromatogr. B Biomed Sci Appl. 2000, 744, 183-188. [23] Carducci, C., Birarelli, M., Leuzzi, V., Santagata, G. et al., J. Chromatogr. A 1996, 729, 173-180. [24] Slupsky, C.M., Rankin, K.N., Wagner, J., Fu, H. et al., Anal. Chem. 2007, 79, 6995-7004.

Chapter 7

Explorative Analysis of Urine by Capillary Electrophoresis-Mass

Spectrometry in Chronic Patients with Complex Regional Pain

Syndrome

R. Ramautar, A.A. van der Plas, E. Nevedomskaya, R.J.E. Derks, G.W. Somsen, G.J.

de Jong, J.J. van Hilten, A.M. Deelder, O.A. Mayboroda

Journal of Proteome Research 2009, 8, 5559-5567

Explorative Analysis of Urine by CE-MS in Patients with CRPS

Abstract Complex Regional Pain Syndrome (CRPS) is characterized by various combinations of sensory, autonomic and motor disturbances. Pain disproportionate to the severity and duration of the inciting event is the most devastating symptom. Diagnosis of CRPS is difficult as the underlying mechanisms remain unclear. In order to try to derive metabolic indicators potentially characteristic for CRPS we applied capillary electrophoresis time-of-flight mass spectrometry (CE-ToF-MS) to the explorative analysis of urine. The CE-ToF-MS method provided fast and stable metabolic profiles of urine samples. The mean intraday and interday CVs were <2% and <9% for migration times and peak areas, respectively, demonstrating robustness of the method. Using multivariate chemometric analysis, discrimination between urine samples from CRPS patients and controls was obtained, emphasizing differences in metabolic signatures between CRPS-diseased patients and controls. Several compounds, such as 3-methylhistidine, were responsible for discriminating the samples. The biological relevance of these compounds with regard to CRPS is discussed. Thus, CE-ToF-MS-based metabolic profiling of urine from CRPS patients and controls revealed metabolites that differentiate between diseased and control, illustrating the usefulness of this approach to get more insight into the pathology underlying CRPS.

123

Chapter 7

1. Introduction Complex Regional Pain Syndrome (CRPS) is a chronic pain condition characterized by a combination of sensory, vasomotor, and trophic changes. About 25% of the patients with CRPS develop movement disorders, like tremor, myoclonia and dystonia (1,2). The syndrome is more common in women and usually is preceded by tissue injury like a fracture or surgery. Following tissue injury, the body may respond with a series of specific reactions aiming to repair the damage, promote wound healing and recruit host defence mechanisms. Indeed, there is some evidence to suggest that multiple biological mechanisms responsible for inflammation, vasomotor dysfunction and pain may become involved in CRPS (3-10). Collectively, these facts indicate that in the acute phase of CRPS the body response to trauma may turn aberrant. Still, in a substantial number of patients the disease course becomes chronic. In the majority of the patients, CRPS remains restricted to one extremity, however the syndrome may spread to other extremities without a new trauma (11-13) These more affected chronic phenotypes have a much younger age at onset, which might suggest a genetic predisposition (13). Mechanisms that play a role in this phase of the syndrome are largely unknown and even if such mechanisms are suggested, they are seldom based on solid biochemical data. Consequently, explorative “omics” based approaches, such as metabolomics, could be a good starting point for further delineation of mechanisms underlying CRPS. At first glance, metabolic profiling is quite similar to other “omics” approaches but unlike genomics and proteomics which often only indicate a potential for physiological (pathophysiological) effects, metabolomics data provide insight into the actual biochemical status of an organism (14-16), i.e. it provides real endpoints for the processes that are occurring within the body. The output of a metabolomic experiment could either be treated as “a metabolic signature” or “phenotype” and used for building up diagnostic routine, or extrapolated to other “omics” data, i.e. combining metabolomics data with genomics and proteomics data, in order to reveal systemic mechanisms of pathology. Metabolic profiling of body fluids has a broad analytical basis: nuclear magnetic resonance spectroscopy (NMR), gas chromatography-mass spectrometry (GC-MS), liquid chromatography-mass spectrometry (LC-MS) and capillary electrophoresis-mass spectrometry (CE-MS) (17). This diversity of analytical methods, in a way, reflects the diversity of chemico-physical properties of compounds composing the human metabolome. At the end, the final selection of a method depends entirely on the type of sample and the “target” group of metabolites. For example, the use of metabolic profiling through evaluation of patient urine is an effective approach to study the biochemistry of diseases as many

124


low-molecular-weight metabolites are excreted into urine. Moreover, to access the inflammatory axis of CRPS, one might consider urine as sample of choice. Furthermore, this body fluid can be obtained quickly, easily, and in a non-invasive manner in the clinic. Thus, metabolic profiling of urine has potential utility for biomarker discovery. The urinary metabolome consists mainly of (highly) polar and ionic compounds (18), and therefore, CE is a very attractive separation technique as it provides a high-resolution separation for highly polar and ionic compounds (19,20). Consequently, CE can provide information-rich metabolic profiles of urine and as a result potential insight into the biochemistry of CRPS. Recently, we have demonstrated the analytical quality and feasibility of a CE-MS method for the profiling of amino acids and related compounds in urine from patients with urinary tract infection (21). In the present study, we show that this CE-MS method can be used for metabolic profiling of human urine and we have applied this method for the first time to the explorative analysis of urine samples to investigate changes in metabolite patterns between patients with CRPS and control subjects. 2. Materials and Methods 2.1 Chemicals PB (hexadimethrine bromide) and a 25% m/v aqueous solution of poly(vinyl sulfonate) (PVS) sodium salt were from Sigma-Aldrich (Steinheim, Germany). A 10% m/v PB solution was prepared by diluting PB with Milli-Q water. A 5% PVS solution was made by diluting the purchased PVS solution with Milli-Q water. The polymer solutions were passed through a 0.22 µm filter and stored at 4ºC. Formic acid was from Riedel de Haën (Seelze, Germany). Ammonium hydroxide 25% was from Merck (Darmstadt, Germany). All L-Amino acids with a concentration of 1 mM of each in 0.1 M hydrochloric acid were purchased from Sigma-Aldrich. HPLC-grade methanol was supplied by Biosolve BV (Valkenswaard, the Netherlands). Hippuric acid, 3-methylhistidine, xanthine, guanosine, L-carnitine, phenylalanine-glycine, N-methylnicotinamide were purchased from Sigma-Aldrich. Standard solutions were prepared with water taken from a Milli-Q water purification system or with background electrolyte (BGE) (Millipore, Bedford, MA, USA). BGE solution was prepared by dissolving formic acid in Milli-Q water (1 M, pH 1.8).

125

Chapter 7

2.2 Patients and urine samples Subjects and CRPS diagnosis Thirty-four consecutive patients (Table 1) from the department of Neurology of the Leiden University Medical Center fulfilled the CRPS criteria of the International Association for the Study of Pain (IASP). Table 1. Demographic, CRPS characteristics and treatment (n=36).

Demographic and injury CRPS characteristics Previous treatments Current treatment Age, median (range): 41 (17 – 60)a

Genderb

Male 6 Female 30 Affected limb Upper 30 Lower 26 Type of injury Contusion 15 Fracture 8 Surgery 7 Other 6 Symptoms duration, median (range): 72 (1 – 340)c

Spontaneous pain 36 Mechanical allodynia 33 Color changes 30 Edema 29 Hyperhidrosis 20 Temperature changes 31 Motor changesd 34 1 limb 13 2 limbs 7 3 limbs 5 4 limbs 9 Trophic changes 26 NRS pain e median (range): 7.5 (5 – 10)

For CRPS NSAIDsf 15 Opioids/tramadol 19 Antidepressants 21 Anticonvulsants 16 Baclofen 22 Benzodiazepines 17 Epidural/blocks 6 Topical agents 5 P.T/O.Tg 27 Acupuncture 1 TENSh 6 SCSi 2 Ketamine i.v. 3 Mannitol i.v. 8 Scavengers 8 For other conditions Antihypertensives 10 Other 31

For CRPS NSAID’s 10 Opioids/tramadol 19 Antidepressants 14 Anticonvulsants 10 Baclofen 14 Benzodiazepines 11 P.T/O.T 26 TENS 1 SCS 1 Scavengers 4 For other conditions Antihypertensives 9 Other 29

a Median (range) age of controls = 27 (21 – 41) years b Gender of all controls was female c Duration in months d Dystonia was present in 31 patients (86%) e Numeric rating scale for pain f Non-steroidal anti-inflammatory drugs g Physiotherapy (P.T) and Occupational therapy (O.T) h Transcutaneous electrical nerve stimulation i Spinal cord stimulation

Thirty-one patients also had developed fixed dystonia of one or more extremities. Other causes of dystonia had been excluded using appropriate blood and imaging studies (computed tomography or magnetic resonance imaging) of the brain. All patients were also evaluated concerning the interval that had elapsed between onset of CRPS symptoms and time of sampling, medication used at the time of sampling, and previous therapy. The pain intensity was evaluated using a numeric rating scale. Patients with renal or bladder pathology were ruled out in order to limit confounding conditions. All patients were examined at our outpatient clinic for movement disorders. None of the patients had a family history of dystonia and other causes of dystonia, including birth injury, head trauma, neuroleptic treatments were not reported

126


or identified in the medical history or correspondence. Laboratory tests, including serum copper and ceruloplasmin and computed tomography or magnetic resonance imaging of the brain were normal. Control subjects were recruited from volunteers working in our hospital and selected to match the overall age and sex distribution of the patients, and had no history of disease. Only females, who reported no significant medical history and used no medication, were included to optimize the homogeneity of the group. This comprised the number of control subjects, who were eligible for participating in the study. Urine collection and analysis All samples were collected early in the morning between 4.00 and 7.00 a.m. to minimize potential confounding influences of circadian rhythms of constituents excreted in the urine. Patients used standard 50 mL non-sterile containers to store urine temporarily. The same day patients visited the hospital and within 10 hours after micturition (i.e. the act of urinating) samples were transferred to non-sterile polypropylene micro tubes of 0.5 mL and then stored at -80ºC. 100 μL of each urine sample from controls and patients were pooled and used as quality control (QC) samples. Prior to CE-MS analysis, the urine samples were thawed to room temperature, mixed with BGE (1:1, v/v) and centrifuged (13,200 rpm) for 5 min. The sample sequence started with five QC samples spiked with amino acids at different concentration levels for the measurement of a calibration plot, which was followed by a non-spiked urine sample. Next, urine samples from 18 control subjects and 36 patients were analyzed in an alternating way. This sample sequence was randomised with QC samples to evaluate the performance of the CE-MS method during the analysis of the sequence. 2.3 Instrumentation and procedures The CE experiments were carried out on a Beckman Coulter PA 800 instrument (Beckman Coulter, Fullerton, CA, USA). Fused-silica capillaries were from Composite Metal Services (The Chase, Hallow, UK) with a total length of 130 cm and an internal diameter of 50 μm. Capillaries were noncovalently coated with a bilayer of Polybrene (PB) and poly(vinyl sulfonate) (PVS). Formic acid (1 M, pH 1.8) was used as BGE. Sample injections were performed hydrodynamically for 90 s at 90 mbar (1.3 psi). Prior to sample injection a small plug (50 mbar, 9 s) of 12.5% ammonium hydroxide was injected for pH-mediated stacking. Electrophoretic separation was performed at

+30 kV and the capillary temperature was set at 25 C. For more experimental details we refer to reference (21).

127

Chapter 7

MS was performed using a micrOTOF orthogonal-accelerated time-of-flight (TOF) mass spectrometer (Bruker Daltonics, Bremen, Germany). Spectra were collected with a time resolution of 1 s. Post-run internal mass calibration was performed using

sodium formate cluster ions Na+(HCOONa)1-9 ranging from 90.9766 to 430.9137 m/z. CE-MS coupling was realized by a co-axial sheath liquid interface (Agilent Technologies, Waldbronn, Germany) with methanol-water-formic acid (50:50:0.1, v/v/v) as sheath liquid. The following spray conditions were used: ESI voltage was -4.5 kV; sheath liquid flow, 4 μL/min; dry gas temperature, 180°C; nitrogen flow, 4 L/min; nebulizer pressure, 0.5 bar. Micellar electrokinetic chromatography (MEKC) combined with UV detection was used for the determination of 3-methylhistidine in urine samples. The employed experimental procedure has been described in detail elsewhere (22). 2.4 Analytical validation The procedures for the determination of linearity of response for amino acids in BGE and in QC samples were described in a previous paper (21) The limit of detection (LOD) for amino acids in BGE and QC samples were calculated as the concentration yielding a S/N of 3 via extrapolation of the S/N-ratio produced by the lowest concentration used for the construction of calibration plots of amino acids in BGE and QC samples, respectively. Intra- and interday variation of migration time and peak area were determined for twelve endogenous urinary metabolites (L-arginine, L-valine, L-glutamic acid, L-phenylalanine, L-histidine, hippuric acid, N-methylnicotinamide, creatine, xanthine, L-carnitine, guanosine and phenylalanine-glycine) with different physico-chemical properties. The intraday variation for migration time and peak area was determined by analyzing ten replicates of a pooled urine QC sample. The interday variation was determined by analyzing a QC sample on 5 different days. Recovery was evaluated in a previous study (21). 2.5 Data analysis The raw data files were converted to netCDF format. For non-linear alignment and feature detection of the data we used an open source XCMS software (23). CE-MS data were processed using default settings (see documentation for XCMS, http://metlin.scripps.edu/download/) except for parameters: “steps” of xcmsSet method set to 3, and “bw” of group method set to 20. The normalization of the data was performed using nonparametric algorithm (24). The table of ion features and normalized intensities was imported into SIMCA-P+ (version 12.0; Umetrics)

128

http://metlin.scripps.edu/download/


software for further multivariate analysis. Principal component analysis (PCA) and partial least squares - discriminant analysis (PLS-DA) were performed both on original data and on data after orthogonal signal correction (OSC). PLS-DA models were constructed after removal of extreme outliers and validated using permutation test with 100 permutations. To identify metabolites of interest a rational chemical formula, i.e. incorporating unlimited C, H, N and O, and a restricted number of S, P, Na, and K per chemical formula, were generated based on internally-calibrated mono-isotopic masses within ≤10 mDa error, using the Generate Molecular Formula tool within the DataAnalysis (DA) software package (Bruker Daltonics). The chemically-reasonable formulas were submitted to metabolome databases: Kyoto Encyclopedia of Genes and Genomes (KEGG) ligand database (25), the Human Metabolome Database (HMDB) (26), and the METLIN database (27). The isotopic distribution patterns of the matched metabolite candidates were simulated with the Simulate Pattern tool of DA and compared with observed mass spectra to reduce further the number of potential elemental compositions (28;29). 3. Results 3.1 Evaluation of CE-ToF-MS method for metabolic profiling Recently, we have developed a CE-ToF-MS method for the analysis of amino acids in human urine using capillaries noncovalently coated with a bilayer of polybrene (PB) and poly(vinyl sulfonate) (PVS) (20;21). The PB–PVS coating proved to be very reproducible yielding stable CE-MS patterns of amino acids in human urine with favourable migration time repeatability (RSDs <2%). The CE-ToF-MS method showed linear responses between 0.5 or 5 μM and 250 μM for most amino acids (R2 >0.99), indicating a good dynamic range. Here, we demonstrate that our method can be used for the profiling of ionogenic compounds with electrophoretic properties similar to amino acids (e.g. amines, small peptides and related metabolites). In CE, all cationic compounds migrate toward the cathode (in our set-up towards the MS). Therefore, in principle all cationic compounds can be analyzed using the PB-PVS CE−MS configuration. To analyze all cations simultaneously, a very low pH electrolyte, 1 M formic acid (pH 1.8), was used to confer a positive charge on all cationic metabolites, making them amenable to MS analysis. Figure 1 shows a typical total ion electropherogram of a urinary profile of a control subject obtained by CE-

ToF-MS analysis. A metabolic profile of positively charged compounds in the m/z region from 50 to 450 was obtained in less than 18 min. Approximately 300 compounds were detected within 24 min (on the basis of an injection volume of ca.

129

Chapter 7

100 nL). The legend below Figure 1 shows the type of compounds detected in the given migration time intervals (indicated by a different colour). At the low pH conditions used for separation, all compounds with a basic functionality are positively charged, allowing their migration toward the MS detector. Compounds with a higher number of amine groups in their structure, such as lysine or arginine, migrated first (11-12 min). Next a group of metabolites comprising amino acids (e.g., valine, leucine, threonine, proline), vitamins (e.g. carnitine) and amines (e.g. creatine and N-methylnicotinamide) migrated together with small peptides like phenylalanine-glycine (12-14 min). Subsequently, the slower migrating amino acids comprising phenyl moieties (such as phenylalanine and tyrosine) migrated and then nucleosides like guanosine (14-17 min). Acidic metabolites, such as hippuric acid, are slightly anionic at pH 1.8, and therefore, migrate just after the EOF. Hence, the CE-ToF-MS method is capable of separating different classes of metabolites simultaneously in human urine, as required for metabolic profiling.

Figure 1. Total ion electropherogram of a urine sample from a control subject obtained by CE-ToF-MS analysis. The legend below the electropherogram shows the type of compounds migrating in the specified coloured time regions.

130


The analytical quality of the data derived from the CE-ToF-MS analysis of QC samples was evaluated. The intra- and interday variation of migration time and peak area were determined for a number of endogenous urinary metabolites with different physicochemical properties (Table 2) and are stated in Table 3. Table 2. Names and physico-chemical properties of the test mixture metabolites.

Compound MW1 pKa2

acid pKa2

base Log P3

L-Arginine 174.11168 2.01 9.04 -3.86 L-Valine 117.07898 2.29 9.74 -2.28 L-Glutamic acid 147.05316 2.10 9.47 -3.54 L-Phenylalanine 165.07898 2.58 9.24 -1.35 L-Histidine 155.0695 2.30 9.60 -2.66 Hippuric acid 179.05824 3.60 - 0.24 Creatine 131.06947 2.67 11.02 -1.58 N-Methylnicotinamide 136.06366 3.50 - -0.21 Xanthine 150.01778 0.80 7.44 -1.04 L-Carnitine 161.10519 3.80 - -3.90 Guanosine 283.09167 1.60 9.21 -2.06 Phe-Gly 222.1005 2.34 9.09 0.03

1Monoisotopic mass 2Data taken from CRC Handbook of Chemistry & Physics 89th Edition 3Predicted by ALOGPS

Table 3. Precision data (CV, %) for migration time and peak area of endogenous compounds in human urine analyzed by CE-ToF-MS. Compound Intraday (n=10) Interday (n=5) Migration time Peak area Migration time Peak area L-Arginine 1.5 5.9 1.4 6.4 L-Valine 1.1 8.9 1.2 8.5 L-Glutamic acid 1.2 7.9 1.2 8.2 L-Phenylalanine 0.8 7.9 1.0 7.5 L-Histidine 1.2 6.8 1.2 7.3 Hippuric acid 1.1 6.5 1.3 6.9 Creatine 0.9 8.6 1.1 7.8 N-Methylnicotinamide 1.1 5.5 1.0 7.0 Xanthine 1.2 7.0 1.2 7.6 L-Carnitine 1.2 7.7 1.2 7.8 Guanosine 1.3 6.4 1.2 7.3 Phe-Gly 1.4 7.6 1.5 7.5

131

Chapter 7

The intraday and interday CVs were <2% and <9% for migration times and peak areas, respectively, demonstrating robustness of the method. Analytical recovery was determined in a previous study (21). In brief, five QC samples before and after addition of known metabolite concentrations (5, 25 and 100 μM) were analyzed. Recoveries ranged from 75% to 99% and the CVs were 7%-11% for all metabolites. The LODs for amino acids in human urine varied from 85 to 280 nM (S/N = 3).We also evaluated the performance of the MS detector with respect to mass accuracy. For

this, the m/z value of histidine was determined in ten urine samples of control

subjects. The observed mass deviation from the theoretical m/z value of histidine was less than 2 mDa (Figure 2), indicating a stable performance of the ToF-MS instrument.

Figure 2. Instrument performance with respect to mass accuracy. The observed mass deviation from the theoretical m/z value of histidine less than 2 mDa as measured in 10 different urine samples. 3.2 Metabolic profiling and multivariate statistics We have used the CE-ToF-MS method for the metabolic profiling of urine samples from 36 patients with CRPS and from 18 control subjects (for the analytical workflow, see Figure 3). Data sets were subjected to a pre-processing routine – migration time alignment and normalization - and then, to an explorative analysis using a set of chemometrical methods. The PCA scores plot revealed only weak tendency for grouping of control samples and patient samples. At the same time, there were no indications of grouping based on acquisition time, sampling time or gender of subjects (Figure 4). To enhance class separating variance and at the same time to get an

132


overview of metabolites contributing to it we used a supervised data analysis method: PLS-DA.

Figure 3. Analytical work flow used for metabolic profiling of urine samples.

Figure 4. PCA score plot coloured according to observation ID; Unsupervised model shows no strong clustering of patients and controls for the first two principal components. Total variance covered by the PC1 and PC2 is about 10% and 9% respectively.

133

Chapter 7

As data were acquired in two large acquisition blocks over two days, we split our data in two parts: data acquired on day one (18 patient samples and 9 controls) was used as a training set (to build a model) and data acquired on day two (18 patient samples and 9 controls) was used as a prediction set (to test the model). The PLS–DA score plot demonstrates an obvious visual clustering based on a two-class model (Figure 5).

Figure 5. PLS-DA model based on CE-ToF-MS data of urine from control subjects and patients with CRPS. (A), PLS-DA model based on CE-ToF-MS data of urine samples measured on day 1. (B), Validation of the model using 100 permutations.

134


The quality of the PLS-DA model was verified by a correlation coefficient R2 (goodness to fit) and a cross-validated correlation coefficient Q2 (goodness of prediction), respectively. A good prediction model is achieved when Q2 > 0.5. The PLS-DA model showed a R2 of 0.96 and a Q2 of 0.855 indicating a good fitting and prediction ability. Then we used data acquired on the second day (second acquisition block) to test the predictive power of our two-class model. Figure 6 shows that the majority of the samples were predicted correctly. Thus, the CE-ToF-MS method is capable of dissecting the metabolic signatures discriminating cases and controls. The strongest identified loadings from our model are listed in Table 4. The provisional identification of compounds was based on accurate mass measurements, analysis of isotopic profile and migration time information. The molecular formula obtained was subsequently introduced in, for example, The Human Metabolome Database and METLIN database to obtain the identification of the metabolites. 4. Discussion Hitherto, the diagnosis of CRPS is fully based on clinical symptoms and signs and potential mechanisms of disease have only partly been elucidated. This situation reflects most likely the complex origin of CRPS and the fact that, so far, hypothesis-driven strategies have provided a limited contribution to our understanding of the syndrome. Here we present an attempt to use an alternative approach, namely explorative analysis of the metabolome of urine – a body fluid, which so far never has been analyzed systematically in CRPS. The urinary metabolome consists mainly of polar and ionic compounds and the choice of CE as a separation technique is, therefore, a logical one as in CE compounds are separated on the basis of their charge-to-size ratio. CE is especially suited for the high-resolution separation of highly polar and charged compounds. Other important advantages of CE are cost efficiency (fused-silica capillaries instead of expensive LC columns and low consumption of chemicals) and little or no need for sample preparation. The only drawback of CE is its limited concentration sensitivity due to the nanoliter-scale sample volume loaded into the capillary. The sensitivity, however, can be improved by the use of in-capillary pre-concentration techniques, such as pH-mediated stacking. The pre-concentration technique used here is based on the injection of a short plug of alkaline solution prior to a long hydrodynamic injection of an acidified urine sample. Using this approach, basic and zwitter-ionic compounds such as amino acids were pre-concentrated within the capillary resulting in detection limits at the low nanomolar range. This approach also exemplifies the compatibility of pH-mediated stacking for the direct analysis of complex samples while minimizing off-line sample handling.

135

Chapter 7

Figure 6. Predictive ability of PLS-DA model based on urine samples measured on day 1. (A), PLS-DA scatter plot showing the prediction of urine samples from control subjects and patients with CRPS measured on day 2 (B), Column plot displaying the predicted response value for all responses from the last component.

136


Table 4. Summary of metabolites identified that showed most significant differences between the two sample sets.

Name Molecular formula

m/z observed

m/z calculated

Error (mDa)

Increased/decreasedA

3-Methylhistidine C7H11N3O2 170.09185 170.09240 0.6 2-aminobenzoic acid C7H7NO2 138.05429 138.05495 0.7 3-amino- octanoic acid C8H17NO2 160.13341 160.13321 -0.2 Hippuric acid C9H9NO3 180.06658 180.06552 -1.1 Histidine C6H9N3O2 156.07529 156.07675 1.5 Proline betaine C7H13NO2 144.10248 144.10191 -0.6 Creatine C4H9N3O2 132.07590 132.07675 0.9 AConcentration increased or decreased compared to controls.

Advantages of using a mass spectrometer as a detector, and especially the favourable characteristics of a ToF mass analyzer have been discussed in literature before (30). Moreover, we have recently shown the applicability of our CE-ToF-MS method for urine analysis. In our previous publication (21) we have put main emphasis on the analysis of amino acids. Here, we show that our method covers a range of classes of metabolites: amino acids, small peptides, nucleosides and even organic acids (Figure 1, Table 3). Some organic acids, such as hippuric acid, for example, can be (partly) negatively charged at pH 1.8 and, although behaving as an anion during the CE separation, they can be detected as protonated molecules in positive ion mode MS. To sum up, all small compounds carrying a positive charge at pH 1.8 can be analyzed within 17.5 min (before electro-osmotic flow) with this method. Although, the CE-ToF-MS method is able to simultaneously analyze charged, low–molecular weight compounds comprising various metabolite classes, it is not effective for the separation of neutral compounds and large biomolecules such as fatty acids, sugars, steroids and larger peptides. Therefore, the analysis of urine samples of CRPS patients and controls by UPLC-ToF-MS and GC-ToF-MS methods has the potential to greatly expand the coverage of the urinary metabolome, thereby increasing the chance of finding more metabolic indicators potentially related to CRPS. The CE-ToF-MS method has been used for metabolic profiling of urine and so far, such an approach has not been carried out in CRPS. We have shown that metabolic profiles of urine of CRPS diseased-patients show a composition which is substantially different (Table 4). Considering the relatively small group size and the potential heterogeneity among subjects, it is not surprising that differences between patients and controls are not reflected in the main axis of a variance represented by PCA plot. Hence, a supervised method of analysis is more obvious choice. Indeed, the PLS-DA

137

Chapter 7

model not only revealed class dependent variance but predicted almost correctly an external hold-out set. Another important output of PLS-DA is the possibility to estimate and rank the influence of individual features on our model with VIP (variable influence on the projection). In theory all features above threshold (α ≥ 1) are considered to be significant for the given model, but in practice the threshold depends on the size of the data set. A limited number of samples may result in overfitting of the model and therefore, only molecules with α ≥ 3 and standard deviation significantly lower than the ranking factor were selected for molecule assignment and identification. Structural assignment of molecules, based on mass position, isotopic pattern, migration time and extensive use of open source metabolome databases (METLIN, KEGG, HMDB) provided a realistic chemical identity of all candidates. However, if for compounds like histidine, hippuric acid and creatine this assignment hardly could be questioned, for n-amino-octanoic acid and proline betaine it remains the best guess possible of the available data. Methylhistidine also belongs to a special case. Although the calculated elementary composition and migration time both point unequivocally to a methylated form of histidine, it is impossible to distinguish between 1- or 3-methylhisitidine. As correct assignment of methylhistidine isomer was essential for later data interpretation (see below), we used a MEKC method to confirm that the isomer was indeed 3-methylhistidine (Figure 7).

Figure 7. MEKC-UV electropherograms of 3-methylhistidine (3-MT). Four overlaid electropherograms are shown: (1) 3-MT dissolved in water (standard), (2) patient urine, (3) patient urine spiked with 3-MT at a concentration of 100 μM, (3) and (4) control urine.

138


The power of metabolomics lies not in the ability to build allusive pathways starting from one or few detected metabolites but in its multiplex characterization (27). This is underscored in this study – each single metabolite by itself failed to show a significant difference between both groups, whereas taken together a distinct multivariate metabolic signature of CRPS emerged. Although our approach revealed a rather solid model, the interpretation of its biological significance remains a challenging task. The characteristics of the patients with CRPS, as presented in Table 1, show a great heterogeneity, however we found no relation between possible breakdown products from previous or present medication and the urinary metabolic profile. All metabolites that contributed significantly to a class separation in our model belong to the category of so-called “usual suspects” – in other words reflect metabolites which are frequently reported as discriminators in clinical and toxicological studies (31) Thus, at the current stage we have not made an attempt to allocate them into pathways and to build around those pathways a mechanistic hypothesis. For example, it is impossible to explain whether a positive correlation of amino-benzoic acid results from changes in the tryptophan metabolism or whether it reflects a change of gut micro-flora. The same is true for the interpretation of changes of hippuric acid and proline betaine concentration. However, our analysis did reveal a conspicuous increased urine clearance of 3-methylhistidine in CRPS patients. 3-methylhistidine is formed as a result of post-translational methylation of actin and myosin. Since this amino acid is not re-used in protein synthesis, its elevated clearance in urine is a rather specific indication of catabolism of muscle tissue. The activation of proteolytic pathways may also be triggered by pro-inflammatory cytokines, which are assumed to play a major role in the pathogenesis of CRPS. However, further research is necessary to clarify the contribution of muscle protein catabolism and influence of inflammatory processes. This finding has also been associated with conditions like chronic physical stress, starvation, injury of muscle tissue, primary muscle diseases and malignancy (32). Since many of the latter causes of muscle catabolism have been excluded, this finding may indicate that in chronic CRPS factors like physical stress and dietary intake need to be further explored. 5. Conclusion In this work, it is shown that CE-ToF-MS is a very attractive method for metabolic profiling of human urine, especially applicable in the case of high risk explorative studies, which have to be performed with minimal costs and consumption of patient material. Using this method we identified a urinary metabolomic signature of chronic CRPS patients. Although our findings have to be validated in a further study with

139

Chapter 7

larger numbers of patients and age- and sex-matched controls, the results of this study unexpectedly highlight a conspicuous increased muscle catabolism. Since this condition may further compromise the physical health state of patients, this explorative study identified a potentially important issue for future research in CRPS.

References 1. Veldman PHJM, Reynen HM, Arntz IE, Goris RJA. Signs and symptoms of

reflex sympathetic dystrophy: prospective study of 829 patients. The Lancet 1993;342:1012-6.

2. Alexander GM, van Rijn MA, van Hilten JJ, Perreault MJ, Schwartzman RJ. Changes in cerebrospinal fluid levels of pro-inflammatory cytokines in CRPS. Pain 2005;116:213-9.

3. Birklein F, Handwerker HO. Complex regional pain syndrome: how to resolve the complexity? Pain 2001;94:1-6.

4. Huygen FJPM, de Bruijn AGJ, de Bruin MT, Groeneweg JG, Klein J, Zijlstra FJ. Evidence for local inflammation in complex regional pain syndrome type 1. Mediators of Inflammation 2002;11:47-51.

5. Leis S, Weber M, Isselmann A, Schmelz M, Birklein F. Substance-P-induced protein extravasation is bilaterally increased in complex regional pain syndrome. Exp Neurol 2003;183:197-204.

6. Wasner G, Schattschneider J, Heckmann K, Maier C, Baron R. Vascular abnormalities in reflex sympathetic dystrophy (CRPS I): mechanisms and diagnostic value. Brain 2001;124:587-99.

7. Ribbers GM, Mulder T, Geurts AC, den Otter RA. Reflex sympathetic dystrophy of the left hand and motor impairments of the unaffected right hand: impaired central motor processing? Arch Phys Med Rehabil 2002;83:81-5.

8. Maihofner C, Handwerker HO, Neundorfer B, Birklein F. Patterns of cortical reorganization in complex regional pain syndrome. Neurology 2003;61:1707-15.

9. Groeneweg JG, Huygen FJ, Heijmans-Antonissen C, Niehof S, Zijlstra FJ. Increased endothelin-1 and diminished nitric oxide levels in blister fluids of

140


patients with intermediate cold type complex regional pain syndrome type 1. BMC Musculoskelet Disord 2006;7:91.

10. Huge V, Lauchart M, Forderreuther S, Kaufhold W, Valet M, Azad SC et al. Interaction of hyperalgesia and sensory loss in complex regional pain syndrome type I (CRPS I). PLoS ONE 2008;3:e2742.

11. Veldman PH, Goris RJ. Multiple reflex sympathetic dystrophy. Which patients are at risk for developing a recurrence of reflex sympathetic dystrophy in the same or another limb. Pain 1996;64:463-6.

12. Maleki J, LeBel AA, Bennett GJ, Schwartzman RJ. Patterns of spread in complex regional pain syndrome, type I (reflex sympathetic dystrophy). Pain 2000;88:259-66.

13. van Rijn MA, Marinus J, Putter H, van Hilten JJ. Onset and progression of dystonia in Complex Regional Pain Syndrome. Pain 2007;130:287-93.

14. Nicholson JK, Wilson ID. UNDERSTANDING 'GLOBAL' SYSTEMS BIOLOGY: METABONOMICS AND THE CONTINUUM OF METABOLISM. Nat Rev Drug Discov 2003;2:668-76.

15. Nicholson JK, Lindon JC, Holmes E. 'Metabonomics': understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica 1999;29:1181-9.

16. Greef Jvd, Stroobant P, Heijden Rvd. The role of analytical sciences in medical systems biology. Current Opinion in Chemical Biology 2004;8:559-65.

17. Lenz EM, Wilson ID. Analytical strategies in metabonomics. J Proteome Res 2007;6:443-58.

18. Ullsten S, Danielsson R, Backstrom D, Sjoberg P, Bergquist J. Urine profiling using capillary electrophoresis-mass spectrometry and multivariate data analysis. J Chromatogr A 2006;1117:87-93.

19. Ramautar R, Somsen GW, de Jong GJ. CE-MS in metabolomics. Electrophoresis 2009;30:276-91.

141

Chapter 7

20. Ramautar R, Mayboroda OA, Deelder AM, Somsen GW, de Jong GJ. Metabolic analysis of body fluids by capillary electrophoresis using noncovalently coated capillaries. J Chromatogr B Analyt Technol Biomed Life Sci 2008;871:370-4.

21. Ramautar R, Mayboroda OA, Derks RJ, van NC, van Dissel JT, Somsen GW et al. Capillary electrophoresis-time of flight-mass spectrometry using noncovalently bilayer-coated capillaries for the analysis of amino acids in human urine. Electrophoresis 2008;29:2714-22.

22. Ferrari F, Fumagalli M, Viglio S, Aquilani R, Pasini E, Iadarola P. A rapid MEKC method for the simultaneous determination of creatinine, 1- and 3-methylhistidine in human urine. Electrophoresis 2009.

23. Smith CA, Want EJ, O'Maille G, Abagyan R, Siuzdak G. XCMS: Processing Mass Spectrometry Data for Metabolite Profiling Using Nonlinear Peak Alignment, Matching, and Identification. Anal Chem 2006;78:779-87.

24. Sidorov I.A., Hosack D.A., Gee D., Yang J., Cam M.C., Lempicki R.A., Dimitrov D.S. Oligonucleotide microarray data distribution and normalization. Information Sciences 2002;146:67-73.

25. Ogata H, Goto S, Sato K, Fujibuchi W, Bono H, Kanehisa M. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res 1999;27:29-34.

26. Wishart DS, Tzur D, Knox C, Eisner R, Guo AC, Young N et al. HMDB: the Human Metabolome Database. Nucleic Acids Res 2007;35:D521-D526.

27. Smith CA, O'Maille G, Want EJ, Qin C, Trauger SA, Brandon TR et al. METLIN: a metabolite mass spectral database. Ther Drug Monit 2005;27:747-51.

28. Kind T, Fiehn O. Metabolomic database annotations via query of elemental compositions: Mass accuracy is insufficient even at less than 1 ppm. Bmc Bioinformatics 2006;7.

29. Kind T, Fiehn O. Seven Golden Rules for heuristic filtering of molecular formulas obtained by accurate mass spectrometry. Bmc Bioinformatics 2007;8:105.

142


143

30. Ojanpera S, Pelander A, Pelzing M, Krebs I, Vuori E, Ojanpera I. Isotopic pattern and accurate mass determination in urine drug screening by liquid chromatography/time-of-flight mass spectrometry. Rapid Commun Mass Spectrom 2006;20:1161-7.

31. Robertson DG, Reily MD, Cantor GH. Metabonomics in Preclinical Pharmaceutical Discovery and Development. In: John CL, Jeremy KN, Elaine H, eds. The Handbook of Metabonomics and Metabolomics. Amsterdam: Elsevier Science B.V., 2007:241-77.

32. Elia M, Carter A, Bacon S, Winearls CG, Smith R. Clinical usefulness of urinary 3-methylhistidine excretion in indicating muscle protein breakdown. Br Med J (Clin Res Ed) 1981;282:351-4.

Chapter 8

Metabolic Profiling of Human

Urine by Capillary Electrophoresis-Mass Spectrometry using a

Positively Charged Capillary Coating

R. Ramautar, E. Nevedomskaya, O.A. Mayboroda, A.M. Deelder, I.D. Wilson, H.G.

Gika, G.A. Theodoridis, G.W. Somsen, G.J. de Jong

Manuscript in preparation

Metabolic Profiling of Urine by CE-MS using a Positively Charged Capillary Coating

Abstract In this study the potential of a CE-TOF-MS method using capillaries coated with a triple layer of polybrene-dextran sulfate-polybrene (PB-DS-PB) was evaluated for metabolic profiling of human urine. The method covered various metabolite classes and stable metabolic profiles of urine samples were obtained with favourable migration time repeatability (RSDs <1%). The PB-DS-PB CE-MS method was used for the analysis of human urine samples from 30 males and 30 females, which previously had been analyzed by a reversed-phase UPLC-TOF-MS method. Using multivariate analysis of the obtained data, discrimination between urine samples from males and females was found, emphasizing differences in metabolic signatures between male and female. The compounds responsible for male-female classification in CE-MS were different from the classifying compounds in UPLC-MS. Almost all compounds causing classification in the CE-MS study were highly polar and did not exhibit retention on the reversed-phase UPLC column. In addition, the CE-MS

classifiers had an m/z value in the range of 50-150, whereas 90% of the classifying

features found with UPLC-MS had an m/z value above 150. So, the CE-TOF-MS method seems to be highly complementary with respect to the UPLC-TOF-MS method providing information on highly polar and small metabolites.

147

Chapter 8

1. Introduction The aim of metabolic profiling is to analyze a wide range of endogenous metabolites in a biological sample within a single run [1-4]. No single analytical technique is capable of providing a comprehensive view of the endogenous metabolites present in a biological sample. In order to increase the coverage of the metabolome, several analytical techniques have been used in conjunction enabling the detection of metabolites which can not be detected by the use of only one method [5, 6]. For instance, hydrophilic interaction chromatography (HILIC) has been used as a complementary method to a reversed-phase LC (RPLC) method in order to improve the coverage of (highly) polar metabolites in rat urine and human urine [7, 8]. Recently, it has been demonstrated that the retention of highly polar analytes in RPLC can be improved by using modified C18 stationary phases such as mixed functional RP phases and Synergi C18 materials embedded with polar groups [9, 10]. It is becoming evident that for the satisfactory coverage of the metabolome, data from multiple separation techniques should be used [11]. Such a strategy, comprising six different analytical methods, has been recently developed for the comprehensive analysis of the microbial metabolome [12]. A large part of endogenous metabolites present in biological samples is highly polar and ionic and, therefore, capillary electrophoresis (CE) is a very attractive separation technique for metabolic profiling of biological samples, as compounds are separated on the basis of their charge-to-size ratio [13, 14]. Other features of CE include the relatively fast and highly efficient separations with minimal sample pretreatment. These together with the small sample requirement makes CE particularly suitable for the analysis of biological samples that are volume-limited [15]. Recently, we have developed a CE-TOF-MS method for the analysis of amino acids and related compounds in human urine [16]. In this method, CE capillaries were noncovalently coated with a bilayer of Polybrene (PB) and poly(vinyl sulfonate) (PVS) providing a considerable EOF at low pH, thus facilitating the fast separation of amino acids using a BGE of 1 M formic acid (pH 1.8). Although this CE-TOF-MS method can be used for the fast and reproducible profiling of amino acids in human urine samples, the separation window for cationic compounds was limited (ca. 10 min). A longer separation window for cationic compounds can be obtained by using bare fused-silica capillaries at low pH. However, using bare fused-silica capillaries we found considerable migration time variation for amino acids spiked in urine (RSDs of 5-15%) [16]. Another approach to increase the separation window for cationic compounds is by using a CE-TOF-MS method based on PB-dextran sulfate (DS)-PB coated capillaries. In this set-up, cationic compounds will migrate after the electro-

148


osmotic flow (EOF) time at low pH conditions and as a result a larger separation window is obtained. In the present study we have evaluated the potential of this PB-DS-PB CE-TOF-MS method for metabolic profiling of human urine by first examining the type of metabolite classes that can be analyzed. Subsequently, this method was applied to the analysis of urine samples from 30 male and 30 female-subjects. These samples had been analyzed previously with UPLC-TOF-MS in a follow-up of RPLC-MS-based metabolic profiling studies [17]. Multivariate chemometric analysis of the CE-MS data was carried out in order to find out if male urine samples could be discriminated from female urine samples on the basis of their metabolic profiles. The results obtained were compared with the data obtained by UPLC-TOF-MS. 2. Materials and methods 2.1 Chemicals and reagents All the employed chemicals were of analytical grade or higher purity. Uric acid, L-tyrosine, L-phenylalanine, guanosine, formic acid, ammonium hydroxide and formamide were purchased from Fluka (Buchs, Switzerland). Creatinine, glutathione, dopamine, adrenaline, folic acid, orotic acid, flavin adenine dinucleotide (FAD), nicotinamide adenine dinucleotide phosphate (NADPH), polybrene (PB), dextran sulfate (DS) and (poly)vinylsulfonate (PVS) were from Sigma-Aldrich (Steinheim, Germany). Methanol was obtained from ChromasolvTM. Ammonium acetate was purchased from Merck (Darmstadt, Germany). 2.2 Coating procedure The PB-DS-PB coating was prepared by subsequently rinsing the capillary with 10% (m/v) PB solution for 15 min at 350 mbar, deionized water for 5 min at 1380 mbar, 3% (m/v) DS solution for 15 min at 350 mbar, deionized water for 5 min at 1380 mbar, and, finally with a 10% (m/v) PB solution for 15 min at 350 mbar. The capillary was then ready for use with the BGE of choice. Before analysis, capillaries were flushed with BGE for 5 min at 1380 mbar. Between runs, the coated capillaries were flushed with a 1% (m/v) PB solution and BGE, each for 5 min at 1380 mbar.

149

Chapter 8

2.3 CE-TOF-MS and UPLC-TOF-MS CE-MS experiments were performed on a Beckman Coulter PA 800 instrument (Beckman Coulter, Fullerton, CA, USA), coupled to a micrOTOF mass spectrometer from Bruker (Bruker Daltonics, Bremen, Germany) via an ESI electrospray interface from Agilent (Waldbronn, Germany). Separations occurred in an 100 cm long fused-silica capillary with an internal diameter of 50 μm. A BGE of 1 M FA (pH 2.0) was used for the separation of cationic compounds. A sheath liquid composed of methanol/water (1:1, v/v) containing 0.1% FA was infused via a 2.5 mL Hamilton syringe at 4 μL/min. Samples were hydrodynamically injected by applying a positive pressure of 0.5 psi for 30 s. A 30 kV voltage using reversed polarity was applied during separation. The electrospray voltage was +4.0 kV. The experimental conditions for UPLC-TOF-MS have been described previously [17]. Briefly, chromatography was performed on a Acquity UPLC system (Waters) with an Acquity C18 BEH column 100x2.1 mm, 1.7 μm. The injection volume was 10 μL. A gradient program was applied for analyte elution starting from 100% A (0-2 min) and then switching to 90:10 A:B (2.01 min) followed by linear gradient to 100% B at 6.5 min where it stayed isocratic for column clean-up till 7 min. Flow rate was 0.2 ml/min in the first two min and 0.4 ml/min in the rest of the run and solvents were: 0.1% formic acid in LC-MS quality water (A) and 0.1% formic acid in acetonitrile (B). MS was performed using a Waters Micromass Q-TOF Micro (Milford, MA, USA) operating in positive ion electrospray mode. 2.4 Test mixture and urine samples A test mixture of standard substances was prepared and used to evaluate the CE-MS method. It comprised of dopamine, adrenaline, creatinine, hippuric acid, orotic acid, uric acid, glutathione, NADPH, FAD, guanosine and folic acid and amino acids (L-alanine, L-arginine, L-glutamic acid, L-isoleucine, L-leucine, L-lysine, L-methionine, L-phenylalanine, L-tyrosine, L-valine). Initially, 1 mg/mL stock solutions of each analyte were prepared by dissolving appropriate amounts in water. Subsequently, aliquots of stock solutions were diluted with water in a 1.5 mL glass vial in order to obtain a working solution in which each analyte was present at a 20 μg/mL concentration. Stock solutions were kept at –20 ºC until usage. Human urine samples were obtained from AstraZeneca (Dept. of Drug Metabolism and Pharmacokinetics, Cheshire, United Kingdom). Urine samples were collected and stored at -80°C prior to usage. Before analysis, urine samples were mixed with BGE (1:1, v/v) and centrifuged at 13,200 rpm for 5 min.

150


2.5 Data analysis CE-TOF-MS data were exported to mzXML format using Compass Export (Bruker Daltonics). For the alignment and peak picking XCMS software has been used (Scripps Center, CA, USA) with default parameters except for “missing” and “extra” both set to 15. The resulting table contained areas for the found peaks across all the samples, these values were normalized to total areas of the corresponding samples to compensate the difference in concentrations between the subjects. Subsequently the table was imported into SIMCA-P+ software for multivariate analysis. Principal component analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA) were performed using Pareto scaling. UPLC-TOF-MS data were processed by MarkerLynx software with the following track peak parameters were: peak width at 5% height: 15 s, peak-to-peak baseline noise: 80, intensity threshold: 100, mass window: 0.05 amu, retention time window: 0.2 min, noise elimination level: 6 and mass tolerance: 0.50 amu. Peak list data obtained by MarkerLynx were further processed by SIMCA P version 11 from Umetrics (Windsor, UK) for multivariate data analysis. 3. Results and discussion 3.1 PB-DS-PB CE-MS for metabolic profiling Metabolic profiling of urine can be quite difficult as many chemical classes should be analyzed simultaneously within a single run. A test mixture comprising amino acids, organic acids, nucleosides, nucleotides, catecholamines and small peptides was analyzed to determine the type of compound classes that can be covered with the PB-DS-PB CE-TOF-MS method. The compounds used for the evaluation cover a wide range of polarities and molecular weights and, therefore, representative for the large array of metabolite classes present in urine. When analyzing the test mixture with the PB-DS-PB CE-MS method using 1 M formic acid (pH 2.0) as BGE and reverse polarity, most organic acids co-migrated with the EOF as they are neutral at pH 2.0. Hippuric acid, which is slightly negatively charged at pH 2.0, migrated directly before the EOF time. Compounds that are positively charged at pH 2.0 like amino acids, catecholamines and small peptides migrated after the EOF time. The EOF time was ca. 8 min and the effective separation window for cationic compounds is ca. 20 min if we allow for a maximum separation time of 30 min, which is two times larger than the separation window obtained with the PB-PVS CE-TOF-MS method at low pH [16, 18]. Moreover, cationic compounds are separated on a positively charged coated

151

Chapter 8

capillary, thereby preventing potential electrostatic adsorption to the capillary wall. Plate numbers ranged from 100,000 to 300,000 for the test compounds. A range of different metabolite classes can be covered with this method (i.e. nucleosides, amino acids, catecholamines, small peptides and amines), however, small and multivalent cationic compounds will not be covered due to their strong electrophoretic mobility towards the cathode (i.e., opposite to the direction of the EOF). With the PB-DS-PB CE-TOF-MS method, about 500 molecular features (i.e. the number of compounds detected above a certain intensity threshold within the CE run time) were detected, which was twice as much as for the PB-PVS CE-TOF-MS method [18]. This is most probably due to the larger separation window resulting in less co-migration and, therefore, less ion suppression. For a reliable comparison of metabolic profiles and to be able to observe small changes in sample composition, migration time stability is of crucial importance. Figure 1 shows base peak electropherograms of the repeated analysis (n=10) of a spiked human urine sample. The RSDs for migration times of the test compounds were always smaller than 1%, indicating that stable profiles were obtained. Figure 1. Base peak electropherograms (50-1000 m/z) obtained during repeated CE-TOF-MS analysis (n=10) of a human urine sample spiked with test compounds (see experimental). Conditions: BGE, 1 M formic acid (pH 2.0); sample injection, 0.5 psi for 30 s. Separation voltage, -30 kV.

152


The good migration-time repeatability for test compounds in urine samples can be primarily attributed to the use of the PB–DS-PB capillary coating. Overall, the PB-DS-PB CE-TOF-MS method appears suitable for the profiling of (highly) polar metabolite classes in human urine within a single run in a repeatable way. 3.2 Metabolic profiling of human urine and multivariate data analysis The PB-DS-PB CE-TOF-MS method was applied to the analysis of urine samples from 30 male and 30 female human subjects, and PCA and PLS-DA of the CE-MS data was carried out. The PCA scores plot revealed a tendency for grouping the female and male urine samples, respectively. Subsequently, PLS-DA, which turns the projection so that it reflects not only the variation present in the data, but also maximizes the difference between the sample groups, was applied. The PLS-DA scores plot shows a separation of female and male urine samples (Figure 2A). The distinction between female and male urine samples is significant in the first PC, reflecting differences in metabolic profiles between female and male urine. The model parameters for the explained variance and predictive ability were R2Y=0.45 and Q2=0.376 indicating a satisfactory model. To ensure that discrimination in the PLS-DA model was not the result of data overfitting, a validation of the model was performed using permutation testing. Validation of the model showed a positive slope between the randomly permuted and original classified data indicating a valid model (Figure 2B).

Figure 2. (A) PLS-DA scores plot of urine samples analyzed by CE-TOF-MS (■ females; ∆ males), R2Y=0.445, Q2=0.376. (B) Cross validation using random permutations, ▲ stand for R2 values; ■ stand for Q2 values.

153

Chapter 8

An important output of PLS-DA is the possibility to estimate and rank the influence of individual features on our model with VIP (variable influence on the projection). In theory, all features above threshold (α ≥1) are considered to be significant for the given model, but in practice, the threshold depends on the size of the data set. A limited number of samples may result in overfitting of the model and, therefore, only features with α ≥1.5 and standard deviation significantly lower than the ranking factor were selected for molecule assignment and identification. The provisional identification of these compounds was based on accurate mass measurements, analysis of isotopic profile and migration time information. The molecular formula obtained was subsequently introduced in the Human Metabolome database or Metlin database to obtain the tentative identification of the metabolites. In this way 8 out of the 27 compounds with a VIP ≥1.5 could be provisionally identified and they are listed in Table 1. This table shows that the compounds responsible for the classification of female and male urine samples are highly polar and charged compounds. It was not possible to distinguish between 1- or 3-methylhistidine, as the PB-DS-PB CE-TOF-MS method could not separate these compounds. Table 1. Summary of the metabolites identified that showed most significant differences between female and male urine samples. Concentration of these metabolites was decreased in female urine samples.

Name Molecular formula

m/z observed

m/z calculated

Error (mDa)

Methylhistidine C7H11N3O2 170.1080 170.0924 15.6 Glutamic acid C5H9NO4 147.0850 147.0532 31.8 Pyroglutamic acid C5H7NO3 130.0630 130.0426 20.4 Hypotaurine C2H7NO2S 110.0802 110.0196 60.7 Threonine C4H9NO3 120.0912 120.0582 33.0 Methionine C5H11NO2S 150.0920 150.0511 40.9 Methylnicotinamide C7H9N2O 138.0702 138.0715 -1.3 Proline betaine C7H13NO2 144.1160 144.1019 14.1

The same set of urine samples was previously analyzed by a UPLC-TOF-MS method using an Acquity C18 column (1.7 μm; 2.1×100 mm). With the UPLC-TOF-MS study, PCA and PLS-DA also revealed a classification of male and female urine samples. The PLS-DA model parameters for the explained variance and predictive ability were R2Y=0.954 and Q2=0.795 indicating a good model. About 300 features with a VIP ≥1.5 were responsible for the separation of female and male urine in the UPLC-TOF-MS study, which is about ten times more than the number of classifying features obtained with CE-TOF-MS. This is most probably related to the fact that the concentration sensitivity of UPLC-TOF-MS is considerably larger than that of CE-

154


TOF-MS as much larger injection volumes (10 L vs. 30 nL for CE) can be applied. But most interestingly, the compounds responsible for male-female classification in CE-MS were different from the classifying compounds in UPLC-MS. Closer inspection of the provisionally identified compounds responsible for the separation of female and male urine samples in the CE-MS study showed that these compounds (except for glutamic acid) exhibited no retention on the reversed-phase UPLC column, i.e., they eluted with the column dead time. In addition, almost all features

causing classification in the CE-MS study have an m/z value between the 50-150,

whereas 90% of the classifying features found with UPLC-MS have an m/z value above 150. So, the CE-TOF-MS method seems to be highly complementary with respect to the UPLC-TOF-MS method providing information on highly polar and small metabolites. 4. Concluding remarks In the present study, the suitability of a PB-DS-PB CE-TOF-MS method for the metabolic profiling of human urine samples was demonstrated. Profiles covering several metabolite classes in human urine could be recorded in a repeatable way. On the basis of CE-TOF-MS data, female urine could be separated from male urine and the metabolites responsible for this separation reflected the relatively polar fraction of the urinary metabolome. The same set of samples was also measured by a UPLC-TOF-MS method and female and male samples were also separated here. However, the metabolites causing the classification in the CE-MS study were different from the compounds causing the classification in the UPLC-TOF-MS study. Hence, the combination of CE-MS and UPLC-MS may therefore be seen as complementary as the analysis of highly polar and charged compounds is better attainable with CE-MS. A comparison of the results obtained with the CE-MS method with literature NMR data on female and male urine samples reveals that NMR distinguishes mainly on metabolites involved in the tricarboxylic acid pathway [19-21]. In particular, carboxylic acids such as citric acid, lactic acid and fumaric acid, were responsible for this differentiation. In principle, these compounds can be analyzed by CE, however, at low pH separation conditions most carboxylic acids are neutral and, therefore, they cannot be separated and detected by MS using positive ionization mode. In order to analyze these type of compounds, a CE-MS method using alkaline separation conditions should be employed. In that case, it would be really interesting to compare the findings with the NMR studies.

155

Chapter 8

156

References [1] Van der Greef, J., Stroobant, P., Van der Heijden, R., Curr Opin Chem Biol

2004, 8, 559-565. [2] Nicholson, J. K., Lindon, J. C., Holmes, E., Xenobiotica 1999, 29, 1181-1189. [3] Nicholson, J. K., Connelly, J., Lindon, J. C., Holmes, E., Nat Rev Drug Discov

2002, 1, 153-161. [4] Fiehn, O., Plant Mol Biol 2002, 48, 155-171. [5] Lenz, E. M., Wilson, I. D., J Proteome Res 2007, 6, 443-458. [6] Theodoridis, G., Wilson, I. D., J Chromatogr B Analyt Technol Biomed Life Sci

2008, 871, 141-142. [7] Gika, H. G., Theodoridis, G. A., Wilson, I. D., J Sep Sci 2008, 31, 1598-1608. [8] Mohamed, R., Varesio, E., Ivosev, G., Burton, L., et al., Anal Chem 2009, 81,

7677-7694. [9] Myint, K. T., Aoshima, K., Tanaka, S., Nakamura, T., Oda, Y., Anal Chem

2009, 81, 1121-1129. [10] Ding, J., Sorensen, C. M., Zhang, Q., Jiang, H., et al., Anal Chem 2007, 79,

6081-6093. [11] Buscher, J. M., Czernik, D., Ewald, J. C., Sauer, U., Zamboni, N., Anal Chem

2009, 81, 2135-2143. [12] Van der Werf, M. J., Overkamp, K. M., Muilwijk, B., Coulier, L., Hankemeier,

T., Anal Biochem 2007, 370, 17-25. [13] Ullsten, S., Danielsson, R., Backstrom, D., Sjoberg, P., Bergquist, J.,

J Chromatogr A 2006, 1117, 87-93. [14] Ramautar, R., Somsen, G. W., de Jong, G. J., Electrophoresis 2009, 30, 276-291. [15] Monton, M. R., Soga, T., J Chromatogr A 2007, 1168, 237-246; discussion 236. [16] Ramautar, R., Mayboroda, O. A., Derks, R. J., van Nieuwkoop, C., et al.,

Electrophoresis 2008, 29, 2714-2722. [17] Gika, H. G., Theodoridis, G. A., Wingate, J. E., Wilson, I. D., J Proteome Res

2007, 6, 3291-3303. [18] Ramautar, R., van der Plas, A. A., Nevedomskaya, E., Derks, R., et al.,

J Proteome Res 2009. [19] Ramadan, Z., Jacobs, D., Grigorov, M., Kochhar, S., Talanta 2006, 68, 1683-

1691. [20] Slupsky, C. M., Rankin, K. N., Wagner, J., Fu, H., et al., Anal Chem 2007, 79,

6995-7004. [21] Psihogios, N. G., Gazi, I. F., Elisaf, M. S., Seferiadis, K. I., Bairaktari, E. T.,

NMR Biomed 2008, 21, 195-207.

Chapter 9

Conclusions and Perspectives


Over the last two decades, coupled capillary electrophoresis (CE)-mass spectrometry (MS) has been applied for drug analysis (including impurity profiling and chiral separations), forensic studies and for protein and peptide analysis. Still, the broad application of CE-MS seems to be hindered by robustness problems, such as the occurrence of migration-time shifts between analyses and loss of separation efficiency when coupled to MS. This means that in order to further establish the role of CE-MS in relatively new fields as proteomics and metabolomics, stable systems with a broad separation window are needed, especially in applications where separation profiles have to be compared. Moreover, for metabolic profiling studies the CE-MS methods should be able to produce profiles of biological samples covering a wide range of compound classes. In this thesis, novel approaches for reproducible and global metabolic profiling of body fluids by CE-MS were described and the utility of these methods have been demonstrated for clinically relevant problems. In the following sections aspects as migration-time reproducibility, coverage of metabolites and application to clinically relevant problems are discussed in more detail. 1. Reproducibility of CE systems In this thesis, the potential of PB-PVS and PB-DS-PB coated capillaries for the reproducible and efficient analysis of endogenous metabolites in body fluids was evaluated. It was shown that with the use of these coated systems, reproducible analysis of urine, plasma and CSF samples using minimal or no sample pretreatment is achievable without compromising the separation efficiency. Migration-time reproducibilities with PB-PVS and PB-DS-PB coated capillaries were in general within 2% RSD for all the test compounds in urine, CSF and plasma, and plate numbers ranged from 70,000 to 400,000. The capillaries can be rapidly coated with a good inter-capillary reproducibility as demonstrated in Chapter 4 of this thesis. The PB-PVS and PB-DS-PB coatings produce a rather strong and constant EOF virtually independent of the pH of the BGE. The strong EOF results in some loss of resolution for the separation of cationic compounds at low pH with the PB-PVS coating and for the anionic compounds at high pH with the PB-DS-PB coating. In such situations, high plate numbers are crucial in order to maintain sufficient resolution in the relatively short time available for separation. Extra selectivity can be obtained by coupling fast CE methods to a high-resolution MS instrument. For example, the rather low resolution of the PB-PVS CE-MS method for the analysis of amino acids in urine was in part compensated by the high mass accuracy of the TOF-MS instrument, allowing the selective detection of amino acids in urine. In summary, the PB–PVS and PB-DS-PB CE methods provide

159

Chapter 9

very stable and highly repeatable separations of endogenous metabolites in body fluids using minimal sample pretreatment. 2. Coverage of metabolites The simultaneous analysis of a wide range of different metabolite classes in body fluids is a prerequisite for global metabolic profiling studies. This condition was achieved by using both the PB-PVS and PB-DS-PB CE methods at low and high pH separation conditions in order to cover as many basic and acidic metabolites as possible. Moreover, sample pretreatment was minimal (comprising a 1:1 dilution only), which is very advantageous for metabolic profiling as in this way the loss of metabolites is prevented. In all the CE methods, there was a strong EOF towards the MS, which is essential for adequate and reproducible spray conditions and, thus, efficient analyte ionization. In this thesis it was shown that the PB-PVS CE-MS method can provide a comprehensive analysis of basic compounds as all compounds carrying a basic functionality, such as amino acids, amines, nucleosides and small peptides will migrate before the EOF time. Even though the effective separation window was relatively small, more than 300 compounds could be detected using pH-mediated stacking, which technique is very effective for the preconcentration of basic and zwitter-ionic compounds. With the PB-DS-PB CE-MS method at low pH, most basic and zwitter-ionic compounds migrate after the EOF time resulting in an increased separation window. Due to the longer separation window, less ion suppression was observed for test compounds spiked in urine samples compared to the ion suppression observed with the PB-PVS CE-MS method and more molecular features were observed. Although a range of different metabolite classes can be covered with the PB-DS-PB CE-MS method at low pH (i.e. nucleosides, amino acids, catecholamines, small peptides and amines), small and multivalent cationic compounds were not measured due to their strong electrophoretic mobility towards the cathode (i.e., opposite to the direction of the EOF). At high pH conditions, the PB-DS-PB CE-UV method can provide a comprehensive analysis of acidic compounds as all compounds carrying an acidic functionality, such as carboxylic acids, phosphorylated carboxylic acids, small peptides and nucleotides will migrate before the EOF time. In the PB-PVS CE-UV method, most of these compounds migrate after the EOF time, resulting in an increased separation window. Moreover, compounds like FAD and NAPH could be separated as sharp peaks in this method as there was no electrostatic interaction with the capillary wall. When these

160


methods were coupled to MS significantly lower analyte responses were observed compared to that for cationic compounds (see previous section). In general, the type of analytes that can be covered with the CE-MS methods have a highly polar and ionogenic character. These type of compounds constitute a large part of the urinary metabolome and, therefore, the availability of these CE-MS methods is very important for metabolic profiling studies as they can provide complementary information compared to the analytical methods generally used for metabolic profiling of body fluids at this moment, i.e. 1H-NMR, GC-MS and LC-MS. A broad range of metabolite classes can be analyzed by reversed-phase LC-MS including compound classes as sterols, steroids and phospholipids which are difficult to analyze by CE-MS. On the other hand, highly polar and charged compounds are not easy to retain on standard reversed-phase LC columns and, in that respect, the added value of CE-MS for the profiling of these type of compounds was demonstrated in this thesis. NMR is a robust technique for metabolic profiling of body fluids, however, the major disadvantage of NMR is its limited concentration sensitivity and as a result only information on the relatively high abundant endogenous metabolites in body fluids is provided. Generally, detection limits in GC-MS are two orders of magnitude lower than in NMR. Nonetheless, highly polar and non-volatile metabolites are difficult to analyze by GC-MS. Derivatization can be used to increase volatility but many compounds are not susceptible to oximation and silylation reactions. Hence, the analytical methods currently used in metabolic profiling can provide only a partial view of all the endogenous metabolites present in body fluids. In order to obtain a global metabolic profile of a biological sample, multiple techniques should be used in conjunction, i.e. using a metabolomics toolbox consisting of NMR, GC-MS, LC-MS and CE-MS. CE-MS is especially useful for providing a global metabolic profile of the highly polar and charged compounds in body fluids. 3. MS detection The performance of the TOF-MS instrument with respect to mass accuracy and reproducibility of peak areas was also evaluated. Besides reproducible migration times, a good mass accuracy is of pivotal importance for the identification of compounds in metabolic profiling studies. Throughout the studies mass accuracy and its precision

was determined by comparing measured m/z values of test compounds with the

theoretical m/z values. For example, for CE-MS studies in the positive ionization

mode the m/z value of histidine was determined in ten different urine samples of

control subjects. The mass deviation from the theoretical m/z value of histidine was less than 2 mDa indicating a stable performance of the ToF-MS instrument over time.

161

Chapter 9

Evaluation of peak area reproducibility and ion suppression is also a very important aspect in MS-based metabolic profiling as different sets of samples have to be compared. Using MS in positive ionization mode, no signal suppression for test compounds dissolved in the BGE (1 M formic acid) was observed. For test compounds spiked in urine, a signal decrease of 30 to 60% was observed with the PB-PVS CE-MS method. With the PB-DS-PB CE-MS method, a signal decrease of 10–30% was observed. Hence, less ion suppression occurs in the CE-MS method with the longer separation time, which is most probably due to less co-migration of compounds in this method. Additionally, the effect of ion suppression on test compounds was observed to be quite constant among the various urine samples analyzed with both methods. This rather constant ion suppression is very important in metabolic profiling where samples from many subjects have to be compared. In general, peak area reproducibility for a broad range of test compounds was within 10% RSD which is acceptable for MS-based bio-analytical studies. Using MS in negative ionization mode, analyte responses for test compounds were significantly lower. The lower analyte responses were due to ion suppression caused by the BGE (ammonium acetate, pH 9.0). Thus, CE-MS methods for the analysis of anionic compounds require further optimization. 4. Applicability CE-MS is gradually fulfilling the requirements for broad application in routine clinical practice, as demonstrated by the group of Mischak et al. for the validation of multiple urinary polypeptide markers for IgA nephropathy in a cohort of more than 400 patients and 200 healthy controls. Moreover, CE-MS-based polypeptide profiling of urine enabled the detection of diabetes and diabetic nephropathy, and predicted the development of diabetic nephropathy in a blinded, prospectively collected population. These results obtained by CE-MS are very promising, however, this level still has to be reached for profiling studies of small endogenous metabolites, although, first attempts indicate that CE is in the good direction. For instance, the group of Barbas et al. applied CE to the metabolic profiling of urine samples of mice infected with Schistosoma mansoni and this study revealed characteristic metabolic signatures for this infection compared to controls. Such a study was also performed on patients with urinary traction infection (UTI) using the PB-PVS CE-MS method. In this thesis it was shown that with the PB-PVS CE-MS method it was possible to find metabolites discriminating urine samples of healthy individuals and patients with UTI. The clinical utility of the PB-PVS CE-MS was also shown for complex regional pain syndrome, for which disease the biochemical mechanisms are still unclear. CE-MS-based

162


metabolic profiling was used as a starting point for the delineation of mechanisms underlying CRPS. It was possible to measure a characteristic metabolic signature in urine samples of chronic CRPS patients. Although these findings have to be validated in a further study with larger numbers of patients and age- and sex-matched controls, the results of this study highlighted an obvious increased muscle catabolism. Since this condition may further compromise the physical health state of patients with CRPS, this metabolic profiling study identified a potentially important issue for future research in CRPS. The group of Soga et al. used the PB-DS-PB CE-MS method for the global profiling of metabolites in bacterial extracts. Using multiple CE-MS methods in conjunction, more than 1600 compounds were analyzed, which revealed information on bacterial growth and sporulation. The first clinical utility of the PB-DS-PB CE-MS method was illustrated by the metabolic profiling of urine samples from female and male human subjects. With this methodology it was possible to differentiate urine samples from male and female on the basis of differences in metabolic profiles and to provisionally identify the metabolites responsible for this discrimination. Such a distinction was also found with a reversed-phase LC-MS study of these samples, however, it was demonstrated that a large part of the metabolites responsible for the classification of the samples in CE-MS were different from the metabolites responsible for the classification in the LC-MS study, thereby emphasizing the complementary nature of both techniques. The applicability of the PB-DS-PB CE method was also shown for CSF samples of patients with bacterial meningitis. These samples were analyzed without any sample preparation, which is highly advantageous for maintenance of sample integrity. An increased concentration of lactic acid was found in the bacterial meningitis CSF samples, which was in agreement with the literature. In summary, important and relevant clinical information can be obtained by using the PB-PVS and PB-DS-PB CE-MS methods for metabolic profiling of body fluids. 5. Perspectives The CE-MS methods described in this thesis are all set to be used for large-scale metabolic profiling studies, i.e. for a big cohort of urine, plasma, and CSF samples. This is very important from a clinical point of view, as the analysis of hundreds of samples is a routine procedure in this field. Therefore, the clinical utility of the CE-MS methods should be proven for such large studies. In targeted CE-MS-based metabolic profiling studies, internal standards can be used to correct for changes in sensitivity. However, this is hardly feasible in global metabolic profiling studies, because not all metabolites are known or identified. It is more important to demonstrate that the

163

Chapter 9

sensitivity of the method does not change dramatically over a long period of time (i.e. days or weeks). This can be evaluated by the use of quality control (QC) samples. The implementation of such samples for large-scale metabolic profiling studies is essential in order to monitor the stability of the method with respect to reproducibility of migration time, peak area and mass accuracy. Another aim of metabolic profiling is to translate the differences in the metabolic signatures into the phenotypic differences from which these samples were obtained. Therefore, it is crucial that the concentrations of metabolites can be determined reliably and with high precision at very different dynamic ranges. A complicating factor in this respect is that reference compounds are not available for many metabolites, thereby making it impossible to generate quantitative data. In these cases, the concentration of a compound can be estimated by constructing standard-addition curves in the sample matrix, although this is a time-consuming effort. For quantitative analysis, the peak areas for the individual metabolites should show a linear dependence on the metabolite concentration. Moreover, the reproducibility with respect to variations in the analytical method should generally be better than 10% and should at least be lower than the biological variation. The concentration sensitivity of CE-MS is in general suitable for metabolic profiling studies as the concentration of many endogenous metabolites in body fluids is in the micromolar-range. For the analysis of low abundant metabolites, the use of preconcentration techniques is required. This can be done by using in-capillary preconcentration techniques such as pH-mediated stacking or transient isotachophoresis. These strategies are very advantageous for metabolic profiling as there is no loss of metabolites during preconcentration. However, these in-capillary preconcentration techniques require careful optimization studies, as demonstrated for the analysis of amino acids in human urine in this thesis. It would be of high importance to study such approaches also for anionic compounds in order to improve the concentration sensitivity of CE-MS for metabolic profiling of anionic compounds. Another way to improve the concentration sensitivity of CE-MS is by using solid-phase extraction (SPE). In this case, mixed-mode sorbents have the most favorable characteristics for the preconcentration of a wide range of compounds in body fluids. However, it is still a challenge to extract highly polar and ionic compounds on these type of columns. Moreover, the loss of compounds during sample loading or washing is inevitable and, therefore, the use of SPE is more suited for targeted metabolic profiling studies. ESI is the most commonly used ionization technique in CE-MS, and this method is especially suitable for the analysis of (highly) polar compounds. In general, a sheath-liquid interface is used for the coupling of CE to MS. The high and stable velocity of

164


the EOF provided by the PB-PVS and PB-DS-PB coating could also be favorable for CE-MS using sheathless interfacing, which is very interesting in order to improve the concentration sensitivity. In such a method, effective ESI greatly depends on a stable and reproducible spray, which is most easily achieved when the capillary provides a constant and significant flow independent of the pH of the BGE. The applicability of CE-MS for metabolic profiling may be further broadened using APPI and APCI as additional ionization techniques as they are suited for the detection of relatively non-polar compounds. The studies described in this thesis have been performed with a TOF mass analyzer as it allows the simultaneous detection of all ions with high speed and high mass resolution. Therefore, TOF-MS is very suitable for full scan metabolic profiling studies of complex samples. Moreover, TOF-MS also allows a route to metabolite identification via the application of accurate mass measurements of the molecular ion, with typical mass accuracies within the 1 to 100 mDa, which is suitable for provisional identification. As stated, the identification of metabolites is essential in metabolic profiling studies as without any identification the biological significance of metabolite changes is not understandable. Besides, providing accurate mass, TOF-MS can also provide the elemental composition of the molecular ion. However, in most cases, various metabolite structures are possible for a specific elemental composition, often tenths or up to a hundred. In this respect, public reference databases, such as the Human Metabolome database, can be very important. However, currently these metabolome databases are not complete yet. Moreover, although knowing the identity of all endogenous metabolites in the metabolome is the ultimate goal in metabolic profiling, in view of the anticipated efforts required to establish the exact chemical identity of a metabolite in these highly complex samples, another strategy seems to be much more reasonable, namely, to first find the compounds that are important for the biological question under study (e.g., by applying multivariate data analysis) and to focus the identification study on the selected compounds. The identification of metabolites on the basis of accurate mass can proceed faster using MS instruments as the Orbitrap or Fourier transform ion cyclotron resonance (FT-ICR) MS. With these instruments a mass resolution of >500.000 can be obtained and mass measurements with sub-ppm errors. However, measurements with FT-ICR MS are relatively slow which can compromise the analysis of narrow CE peaks, and faster analysis may also lead to reduced sensitivity in FT-ICR MS. At present, the identification of unknown compounds is often performed by MS/MS experiments using an ion trap, triple quadrupole, or quadrupole TOF-MS providing fragment spectra, which can facilitate the process of compound identification.

165

Chapter 9

166

Another promising development for the analysis of body fluids is CE on a microchip. The high electric fields, and short separation lengths can produce analysis times of seconds. CE microchips using dynamic or permanent coatings gain more interest because this can reduce analyte-wall interactions and it can give a more rapid and efficient separation due to adjustment of the EOF. Microchip CE with laser-induced fluorescence detection has been used for the measurement of multiple inflammatory biomarkers in human patient samples using microdissected human skin biopsies as a model. The microchip CE method showed different peptide profiles for microdissected tissue samples from patients with mild and chronic skin lesions in less than a few min. Microchip CE systems provide very fast analysis times, and therefore such systems have high potential for targeted metabolomic studies as the separation efficiency may not be high enough to separate many components in a complex mixture. CE is also very attractive for single-cell measurements as the dimensions of CE make it amenable to small-volume sampling. Moreover, it is possible to simultaneously concentrate and separate analytes by CE and, therefore, CE is very promising for single-cell metabolomics.

Nederlandse Samenvatting


Metabolisme of stofwisseling is het geheel van biochemische processen die plaatsvinden in cellen en organismen. De producten van deze stofwisselingsprocessen zijn de metabolieten. Het metaboloom is de verzameling van alle metabolieten in een cel, orgaan, lichaamsvloeistof of organisme. Hoewel grote moleculen, zoals eiwitten, ook producten zijn van het metabolisme, worden in principe alleen kleine moleculen tot het metaboloom gerekend. De bestudering van het metaboloom wordt metabolomics genoemd. Metabolomics heeft als doel om alle metabolieten in cellen, lichaamsvloeistoffen (bijv. urine, plasma en cerebrospinale vloeistof) of weefsels te meten en te kwantificeren. Aangezien metabole processen de fysiologische toestand bepalen van een cel of organisme, is metabolomics uitermate geschikt om onderscheid te maken tussen ziek en gezond. Een voorbeeld hiervan is de bepaling van de concentraties van een reeks metabolieten in babybloed (dat met de hielprik wordt afgenomen) voor het opsporen van erfelijke stofwisselingsziekten. Binnen het metabolomics onderzoek wordt onderscheid gemaakt in “targeted metabolic profiling” (analyse van specifieke metabolieten zoals aminozuren) en “non-targeted metabolic profiling” (analyse van zoveel mogelijk endogene metabolieten in een biologisch systeem). Metabolietenonderzoek is een analytisch-chemische uitdaging omdat de eigenschappen van metabolieten erg van elkaar kunnen verschillen met grote variatie in bijvoorbeeld moleculaire polariteit, basiciteit/aciditeit en molecuulgewicht. Bovendien zijn er grote verschillen in concentraties. Daarom is het niet mogelijk alle metabolieten met behulp van één analytische techniek te analyseren; er zijn complementaire technieken nodig. Op dit moment worden met name kernspinresonantie (NMR), gaschromatografie (GC) en vloeistofchromatografie (LC) gekoppeld aan massaspectrometrie (MS) gebruikt voor de profilering van metabolieten in biologische monsters. In dit proefschrift zijn de mogelijkheden van capillaire electroforese (CE) gekoppeld aan MS geëvalueerd voor het meten van metabole profielen in lichaamsvloeistoffen (urine, plasma en cerebrospinale vloeistof). CE is een zeer efficiënte micro-scheidingstechniek die met name geschikt is voor het analyseren van polaire en ionogene verbindingen. De scheiding van verbindingen in CE is gebaseerd op verschillen in electroforetische mobiliteit die in essentie een functie is van de lading-massa verhouding van de geanalyseerde moleculen. CE wordt meestal uitgevoerd in fused-silica capillairen, gevuld met een waterige buffer (het achtergrond-electroliet) waarover een hoge elektrische spanning wordt aangelegd. Veel endogene metabolieten zijn oplosbaar in water en ionogeen (doordat ze zure en/of basische groepen bevatten) en kunnen worden gescheiden met CE. Andere voordelen van CE zijn de relatief korte analysetijden, de zeer kleine monstervolumina die nodig zijn en het lage

169


verbruik van oplosmiddelen en chemicaliën. Bij aanvang van dit promotieonderzoek was CE bijna alleen gebruikt voor het gericht analyseren van specifieke metabolieten in lichaamsvloeistoffen, en de mogelijkheden van CE en CE-MS voor targeted en non-targeted metabolic profiling van lichaamsvloeistoffen waren nauwelijks onderzocht. Het hoofddoel van de studies beschreven in dit proefschrift was de ontwikkeling van CE-MS systemen voor het profileren van zoveel mogelijk endogene metabolieten in lichaamsvloeistoffen op een efficiënte en reproduceerbare wijze. Reproduceerbare scheidingsprofielen zijn essentieel om consistente piektoewijzing en valide vergelijkingen mogelijk te maken. Echter, conventionele CE-systemen worden vaak gekenmerkt door een slechte reproduceerbaarheid van migratietijden en/of verminderde scheidingsefficiëntie, met name wanneer het de analyse van lichaamsvloeistoffen betreft. Deze problemen hangen ook samen met ongewenste interacties van matrix componenten met de binnenwand van het scheidingscapillair. In principe is het mogelijk om deze interacties te verminderen of te elimineren en de reproduceerbaarheid te verhogen door de capillair wand te voorzien van een geschikte coating. In dit proefschrift zijn de mogelijkheden bestudeerd om met geladen polymeren op eenvoudige en snelle wijze geadsorbeerde coatings te produceren die geschikt is voor de CE-analyse van zoveel mogelijk endogene metabolieten in lichaamsvloeistoffen met minimale monstervoorbewerking. In de onderhavige studie zijn de polymeren polybreen (PB), dextraan sulfaat (DS) en poly(vinylsulfonaat) (PVS) gebruikt om een geadsorbeerde tweevoudige laag (PB-PVS) of drievoudige laag (PB-DS-PB) te vormen op de capillair wand. Het gebruik van CE in combinatie met geavanceerde massaspectrometers, zoals time-of-flight massaspectrometrie (TOF-MS), voor metabolomics doeleinden stond bij aanvang van dit project nog in de kinderschoenen. Daarom is ook de koppeling van CE aan TOF-MS geoptimaliseerd en aspecten die relevant zijn voor het betrouwbaar profileren van metabolieten, zoals massanauwkeurigheid en ionsuppressie, zijn bestudeerd. De klinische toepasbaarheid van de ontwikkelde CE-MS technieken is gedemonstreerd voor aandoeningen als de urineweg-infectie, bacteriële meningitis en complex regionale pijnsyndroom (CRPS). Multivariate data analyse is gebruikt voor het vergelijken van metabole CE-MS profielen van patiënten en gezonde mensen. In Hoofdstuk 1 van dit proefschrift worden de uitgangspunten en het doel van het uitgevoerde onderzoek beschreven. Een overzicht van het gebruik van CE-MS voor metabolomics wordt gepresenteerd in Hoofdstuk 2. In Hoofdstuk 3 worden de mogelijkheden van een CE methode gebruikmakend van een PB-DS-PB-coating voor de analyse van organische zuren in cerebrospinale vloeistof (CSF) beschreven. De invloed van albumine en natriumchloride op de migratietijden en schotelgetallen van

170


organische zuren zijn onderzocht. De toepasbaarheid van de methode is gedemonstreerd voor de analyse van organische zuren in CSF van patiënten met bacteriële meningitis zonder gebruik te maken van monstervoorbewerking. Organische zuren kunnen in CSF gemeten worden met hoge schotelgetallen (100,000-150,000) en acceptabele reproduceerbaarheid voor migratietijden (RSDs <2%). De geschiktheid van capillairen gecoat met PB-PVS en PB-DS-PB voor de CE analyse van metabolieten in lichaamsvloeistoffen is onderzocht in Hoofdstuk 4. Directe injecties van urine, plasma en CSF zijn mogelijk. De PB-PVS CE-MS methode is geëvalueerd voor het snel en reproduceerbaar profileren van aminozuren in CSF en urine met minimale monstervoorbewerking. De toepasbaarheid van PB-PVS en PB-DS-PB CE-MS voor het profileren van zoveel mogelijk endogene metabolieten in urine is onderzocht in Hoofdstuk 5. Hiertoe zijn de twee systemen geëvalueerd met lage en hoge pH scheidingscondities om de analyse van zowel basische als zure verbindingen mogelijk te maken. Basische verbindingen kunnen relatief snel geanalyseerd worden bij lage pH met het PB-PVS CE-MS systeem aangezien alle positief geladen verbindingen voor de electro-osmotische flow (EOF) tijd migreren. In het PB-DS-PB CE-MS systeem migreren de basische verbindingen na de EOF tijd, resulterend in een groter scheidend vermogen. Voor de analyse van negatief geladen verbindingen bij hoge pH wordt een snelle analyse verkregen op het PB-DS-PB CE-MS systeem, terwijl het PB-PVS CE-MS systeem een groter scheidend vermogen oplevert. Acceptabele migratietijd-reproduceerbaarheden (RSDs <1.6%) en schotelgetallen (70,000-400,000) zijn verkregen voor testmetabolieten in ratten-urine. De invloed van ionsuppressie is systematisch onderzocht. De ontwikkeling van een CE-TOF-MS methode gebruikmakend van een PB-PVS coating voor het profileren van aminozuren in humane urine wordt beschreven in Hoofdstuk 6. Snelle (<15 min) en reproduceerbare analyses van aminozuren (RSD <2%) in urine zijn uitgevoerd en lage detectielimieten (20-300 nM) zijn verkregen met behulp van in-capillaire preconcentratie. De bruikbaarheid van de ontwikkelde methode is geïllustreerd aan de hand van het profileren van aminozuren in urinemonsters van patiënten met urineweginfecties en een controlegroep. De gevonden aminozuurconcentraties in de controlegroep correspondeerden met waarden in de literatuur. Multivariate data analyse van CE-MS profielen gaf aan dat de concentraties van enkele verbindingen, o.a. phenylalanine, significant hoger zijn in de urine van patiënten. In Hoofdstuk 7 is de ontwikkelde PB-PVS CE-TOF-MS methode voor aminozuren gebruikt voor het profileren van metabolieten in urine van patiënten met CRPS. Met multivariate data analyse is vastgesteld dat metabole profielen in urine van patiënten

171


172

significant verschillend zijn t.o.v. de metabole profielen in de controlegroep. Op basis van nauwkeurige massa en metaboloom databases kon de identiteit van enkele verbindingen, die het verschil tussen de twee groepen veroorzaken, vastgesteld worden. Een verbinding met een verhoogde concentratie in de patiëntengroep is 3-methylhistidine. Een verhoogde concentratie van dit metaboliet in urine kan wijzen op spierafbraak en de samenhang hiervan met CRPS dient nader onderzocht te worden. De bruikbaarheid van de PB-DS-PB CE-TOF-MS methode voor het profileren van metabolieten in humane urine wordt beschreven in Hoofdstuk 8. De PB-DS-PB CE-TOF-MS methode kan gebruikt worden voor het simultaan analyseren van verschillende klassen van metabolieten. De methode is toegepast voor het profileren van metabolieten in urinemonsters van 30 mannen en 30 vrouwen. Met multivariate data analyse konden mannen en vrouwen onderscheiden worden op grond van de gemeten metabole profielen en een deel van de metabolieten die verantwoordelijk is voor deze scheiding kon worden geïdentificeerd. Een overeenkomstige studie was ook uitgevoerd met LC-MS en daaruit bleek dat een andere set verbindingen verantwoordelijk is voor de man/vrouw classificatie. De verbindingen die de man/vrouw classificatie in de CE-MS studie veroorzaken, hebben geen retentie in het gebruikte reversed-phase LC systeem. Kortom, deze studie benadrukt het complementaire karakter van beide technieken voor het profileren van metabolieten in urine. Hoofdstuk 9 geeft algemene conclusies en opmerkingen over de bruikbaarheid van de PB-PVS en PB-DS-PB CE-MS methodes voor het profileren van metabolieten in lichaamsvloeistoffen, en enkele toekomstperspectieven worden bediscussieerd.

List of Abbreviations

List of Abbreviations APCI atmospheric pressure chemical ionization APPI atmospheric pressure photo-ionization BGE background electrolyte BPE base peak electropherogram CE capillary electrophoresis CRPS complex regional pain syndrome CZE capillary zone electrophoresis DA data analysis ESI electrospray ionization FT-ICR fourier transform ion cyclotron resonance GC-MS gas chromatography-mass spectrometry HILIC hydrophilic interaction liquid chromatography IT ion trap LC liquid chromatography LOD limit of detection MALDI matrix-assisted laser desorption ionization MEKC micellar electrokinetic chromatography MS mass spectrometry MS/MS tandem MS m/z mass-to-charge ratio NMR nuclear magnetic resonance spectroscopy PB polybrene PB-DS-PB polybrene-dextran sulfate-polybrene PB-PVS polybrene-poly(vinyl sulfonate) PC principal component PCA principal component analysis PLS-DA partial least squares discriminant analysis PVS poly(vinyl sulfonate) RPLC reversed-phase liquid chromatography RSD relative standard deviation TIE total ion electropherogram TOF time-of-flight TQ triple quadrupole UTI urinary tract infection

173

List of Publications

List of Publications 1. Ramautar R, Nevedomskaya E, Mayboroda OA, Deelder AM, Wilson ID, Gika

HG, Theodoridis GA, Somsen GW, De Jong GJ, Metabolic profiling of human urine by CE-MS using a positively charged capillary coating. To be submitted (Chapter 8).

2. Ramautar R, Toraño JS, Somsen GW, De Jong GJ, Evaluation of CE methods for

global metabolic profiling of urine. Submitted (Chapter 5). 3. Nevedomskaya E, Ramautar R, Derks RJ, Westbroek I, Zondag G, Van der

Pluijm I, Deelder AM, Mayboroda OA, Metabolic profiling of mouse urine by CE-MS: dissection of osteoporosis biomarkers from accelerated aging TTD mice. Submitted.

4. Ramautar R, Somsen GW, De Jong GJ, Recent developments in coupled solid-

phase extraction – capillary electrophoresis. Electrophoresis 2010, 31, 44-54. 5. Huhn C, Ramautar R, Wuhrer M, Somsen GW, Relevance and use of capillary

coatings in capillary electrophoresis-mass spectrometry. Analytical and Bioanalytical Chemistry 2010, 396, 297-314.

6. Ramautar R, Van der Plas AA, Nevedomskaya E, Derks RJ, Somsen GW, De

Jong GJ, Van Hilten JJ, Deelder AM, Mayboroda OA, Explorative analysis of urine by capillary electrophoresis-mass spectrometry in chronic patients with complex regional pain syndrome. Journal of Proteome Research 2009, 8, 5559-5567 (Chapter 7).

7. Ramautar R, Ratnayake CK, Somsen GW, de Jong GJ, Capillary electrophoresis-

mass spectrometry using an in-line sol-gel concentrator for the determination of methionine enkephalin in cerebrospinal fluid. Talanta 2009, 78, 638-642.

8. Ramautar R, Somsen GW, de Jong GJ, CE-MS in metabolomics. Electrophoresis

2009, 30, 276-291 (Chapter 2).

174

List of Publications

175

9. Ramautar R, Mayboroda OA, Deelder AM, Somsen GW, de Jong GJ, Metabolic analysis of body fluids by capillary electrophoresis using noncovalently coated capillaries. Journal of Chromatography B 2008, 871, 370-374 (Chapter 4).

10. Ramautar R, Mayboroda OA, Derks RJ, van Nieuwkoop C, van Dissel JT,

Somsen GW, Deelder AM, de Jong GJ, Capillary electrophoresis-time of flight-mass spectrometry using noncovalently bilayer-coated capillaries for the analysis of amino acids in human urine. Electrophoresis 2008, 29, 2714-2722 (Chapter 6).

11. Ramautar R, Somsen GW, de Jong GJ, Direct sample injection for capillary

electrophoretic determination of organic acids in cerebrospinal fluid. Analytical and Bioanalytical Chemistry 2007, 387, 293-301 (Chapter 3).

12. Ramautar R, Demirci A, De Jong GJ, Capillary electrophoresis in metabolomics,

Trends in Analytical Chemistry 2006, 25, 455-466. 13. Kool J, Ramautar R, van Liempd SM, Beckman J, de Kanter FJ, Meerman JH,

Schenk T, Irth H, Commandeur JN, Vermeulen NP, Rapid on-line profiling of estrogen receptor binding metabolites of tamoxifen. Journal of Medicinal Chemistry 2006, 49, 3287-3292.

14. Kool J, Van Liempd SM, Ramautar R, Schenk T, Meerman JH, Irth H,

Commandeur JN, Vermeulen NP, Development of a novel cytochrome P450 bio-affinity detection system coupled on-line to gradient reversed-phase high-performance liquid chromatography. Journal of Biomolecular Screening 2005, 10, 427-436.

Curriculum Vitae

Curriculum Vitae Rawi Ramautar werd op 22 juli 1979 geboren te Paramaribo in Suriname. Na het behalen van het VWO diploma aan het oecumenische scholengemeenschap “Het Baken” te Almere in 1998, volgde het propedeutisch examen Scheikunde gevolgd door de doctoraalstudies farmacochemie en scheikunde aan de Vrije Universiteit te Amsterdam. Tijdens de farmacochemie studie heeft hij een hoofdvakstage verricht bij het Leiden/Amsterdam Center for Drug Research bij de afdeling moleculaire toxicologie o.l.v. prof. dr. Vermeulen en dr. Kool. Het onderzoek was gericht op de on-line koppeling van vloeistofchromatografie en cytochroom P450 bio-affiniteit detectie systemen. Tijdens de scheikunde studie heeft hij een hoofdvakstage gelopen bij Kiadis B.V. o.l.v. prof. dr. Irth en dr. Schenk. Het onderzoek was gericht op de ontwikkeling van bio-analytische technieken voor de analyse van bioactieve verbindingen in complexe matrices. In 2004 werd het doctoraal farmacochemie (richting moleculaire toxicologie) en het doctoraal scheikunde (richting analytische chemie) behaald. Van september 2004 tot januari 2006 was hij werkzaam als research analyst metabolomics in de vakgroep Biomedische Analyse van prof. dr. G.J. de Jong. Vanaf 2006 tot en met 2009 was hij werkzaam als assistent in opleiding (A.I.O.) bij dezelfde vakgroep. Het promotie-onderzoek had als doel het ontwikkelen van capillaire electroforese-massaspectrometrie technieken voor het globaal profileren van endogene metabolieten in lichaamsvloeistoffen op een reproduceerbare wijze. Dit onderzoek, o.l.v. prof. dr. G.J. de Jong en dr. G.W. Somsen, werd deels uitgevoerd op de biomoleculaire massaspectrometrie afdeling van prof. dr. Deelder op het Leids Universitair Medisch Centrum o.l.v. dr. Mayboroda. Gedurende dit promotie-onderzoek heeft Rawi Ramautar verscheidene nationale en internationale lezingen gehouden, waaronder een lezing op Microscale Bioseparations 2008 in Berlijn en een lezing op het Metabolomics congres 2009 in Edmonton (Canada) die door de Metabolomics Society gehonoreerd werd met een oorkonde en bijbehorende prijs van $500. Sinds januari 2010 werkt hij als post-doc op de biomoleculaire massaspectrometrie afdeling van het Leids Universitair Medisch Centrum. Het onderzoek richt zich op de ontwikkeling en toepassing van een analytisch platform voor metabolomics-studies aan urine en bloed van patiënten met nierafwijkingen, om op deze wijze meer inzicht te krijgen in de biochemische processen die ten grondslag liggen aan nierziekten.

176

Dankwoord Na vijf uitdagende jaren is mijn proefschrift afgerond. Een proefschrift dat tot stand is gekomen met hulp van veel mensen die ik bij deze hartelijk bedank voor hun moeite, maar in het bijzonder wil ik de volgende mensen bedanken die direct dan wel indirect hebben bijgedragen aan dit resultaat. Mijn lieve ouders, zusje Artie en zwager Saties, wil ik graag bedanken voor hun onvoorwaardelijke steun. Pa en ma, jullie hebben mij gestimuleerd om te gaan studeren en om zover mogelijk te komen qua opleiding. Bedankt hiervoor! Mijn promotor, Prof. Dr. Gerhardus J. de Jong. Beste Ad, hartelijk bedankt dat ik onder jouw begeleiding mocht werken als promovendus in je onderzoeksgroep. Ik heb ontzettend veel van je geleerd en met veel plezier met je samengewerkt. Samen zijn we begonnen met het ontwikkelen en evalueren van CE-MS technieken voor het vakgebied metabolomics en ik denk dat wij mooie resultaten hebben behaald. Je had altijd tijd voor me om het onderzoek en de vele manuscripten te bespreken. Vooral het schrijven van review artikelen is mij ontzettend goed bevallen en ik hoop dat ik in de toekomst nog vele review artikelen met je mag schrijven. Ook hoop ik dat we op een aantal onderzoeksterreinen nog verder zullen samenwerken. Hierbij wil ik ook je vrouw bedanken voor de jaarlijkse barbecue-bijeenkomsten bij jou thuis in Mijdrecht. Mijn co-promotoren, Dr. Govert W. Somsen en Dr. Oleg A. Mayboroda. Beste Govert, ik heb met veel plezier met je samengewerkt. Je was altijd heel erg snel met je kritische commentaren op manuscripten, abstracts en presentaties en dat heb ik zeer gewaardeerd. Ik bewonder je schrijftalent en ik hoop dat ik een deel van dat talent heb meegekregen. Je was er van ’s ochtends vroeg tot ’s avonds laat voor me. Biertjes drinken met jou was ontzettend gezellig en ik hoop dat we in de toekomst nog een aantal samenwerkingsprojecten zullen hebben. Dear Oleg, thanks a lot for your guidance, scientific discussions, patience and friendship. I am really looking forward to work with you as a postdoctoral researcher at the department of Biomolecular Mass Spectrometry Unit at the LUMC. Your expertise in many different scientific fields is amazing and I hope to learn much more from you the upcoming years. Prof. Dr. André M. Deelder. Beste André, bedankt voor de mogelijkheid dat ik op jouw afdeling een deel van mijn promotie-onderzoek heb mogen uitvoeren. Ik heb dat met veel plezier gedaan en het is voor mij toch een voorrecht geweest om rond te wandelen in een omgeving met state-of-the-art analytische meetinstrumenten. Bedankt voor je vertrouwen in mijn werk. De samenwerking tussen de groepen Farmaceutische Analyse (Prof. De Jong) en Parasitologie-Biomoleculaire Massaspectrometrie (prof. Deelder) heeft mooie wetenschappelijke resultaten en publicaties opgeleverd.

177

178

Prof. Dr. Bob van Hilten en Drs. Anton van der Plas. Beste Bob en Anne, bedankt voor jullie samenwerking op het gebied van CRPS. Het houden van een presentatie op TREND 2009 en de discussies met de vele artsen daar heb ik als zeer leerzaam ervaren. De samenwerking heeft geresulteerd in een mooie publicatie. Hierbij wil ik ook Prof. Dr. Hubertus Irth, Prof. Dr. Coral Barbas, Prof. Dr. Van Solinge en Prof. Dr. Jos Beijnen bedanken voor het lezen en beoordelen van het proefschrift. Al mijn collega’s van Biomedische Analyse: Johan, Javier, Linda, Wout, Gerard, Edwin, Rob, Paul, Roelof, Xander, Joop, Willy Underberg, Lidija, Yvonne, Bregje, Miranda, Michiel, Frits, Thierry, Irma, Rolf, Irene, Andrew, Roel, Dragana, Andrea en Remco…Allen zeer bedankt voor een ontzettend gezellige tijd! All my colleagues at the Parasitology dept.: Katja N., Alegria, Bart, Maurice, Rico, Tiziana, Paul, Ollie, Crina, Liam, Irina, Sibel,, Carolin, Carolien, Rob, Simone, Magnus, Yuri, Hans, Janine, Alexandra, Katja M., Renee, Ralf, Aswin, Manfred, Gerhild, Rene, Dick-Paul, Jean-Marc, Axel, Alex, Cees , Marco en Aswin… Thanks! I would also like to thank Dr. Chitra Ratnayake, Dr. Jerry Feitelson, Dr. Mark Lies, Dr. Jeff Chapman and Dr. Jean-Marc Busnel from Beckman Coulter (USA). Mijn vrienden Stefan, Huub, Antoni, Sander, Jelle en Ferry wil ik graag bedanken voor de vele discussies, miljoenen borrels, feestjes, vakanties, etc. Ik hoop dat we dit nog vele jaren zullen doen. Graag wil ik de mensen/vrienden bedanken die mij enthousiast hebben gekregen voor het wetenschappelijk onderzoek: Jeroen Kool en Tim Schenk. In hetzelfde rijtje wil ik ook Sebastiaan van Liempd, Peter Keizers, Chris de Graaf, Jan Commandeur en Nico Vermeulen noemen. Ook wil ik een aantal mensen van de ACAS-groep bedanken: Mark E., Jon, Niels M., Henk, Freek en Cees. Mijn paranimfen: Varsha en Stefan, mijn vrouw en mijn beste vriend, jullie stonden heel dichtbij mijn onderzoek de afgelopen jaren en het is voor mij een eer om jullie als paranimf te mogen hebben. De laatste woorden zijn voor mijn vrouw, Dr. Varsha Ramautar: nu we allebei gepromoveerd zijn, kunnen we nog meer van het leven gaan genieten en zo snel mogelijk de dingen gaan doen die op ons verlanglijstje staan: het afstrepen kan beginnen. Je hebt me ontzettend gesteund en geholpen op allerlei fronten en jij geeft me de inspiratie plus motivatie om nog verder te bloeien in het wetenschappelijk onderzoek. Ik wil niets liever dan naast jou wakker worden en samen met jou in slaap vallen…….

Explorative Analysis of Urine by Capillary Electrophoresis-Mass Spectrometry in Chronic Patients with Complex Regional Pain Syndrome

Documents