Advertisement

Chromatographia

, Volume 82, Issue 1, pp 325–345 | Cite as

“Pseudostationary Ion-Exchanger” Sweeping as an Online Enrichment Technique in the Determination of Nucleosides in Urine via Micellar Electrokinetic Chromatography

  • Azza H. Rageh
  • Ute PyellEmail author
Original
  • 84 Downloads
Part of the following topical collections:
  1. 50th Anniversary Commemorative Issue

Abstract

The presented study aims to develop a new online enrichment strategy [“pseudostationary ion-exchanger” (PSIE) sweeping] for the analysis of highly hydrophilic nucleosides in urine samples with a special focus on the fundamental aspects regarding the enrichment process itself. In the first method, we employ the ionic liquid (IL)-type surfactant 1-tetradecyl-3-methylimidazolium bromide (C14MImBr) as micelle forming agent under alkaline pH conditions. It is shown that maximum enrichment efficiency can be obtained by keeping the retention factors very high within the sample zone and very low within the background electrolyte (BGE) while maintaining a sufficient resolution for the studied analytes. With this method, detection limits as low as 0.1 µg mL−1 are obtained for all analytes studied. For the nucleosides, adenosine and cytidine, a second method is developed using sodium dodecyl sulfate (SDS) as micelle forming agent under acidic pH conditions. In addition, we investigate the effect of replacing ionic buffering constituents with a zwitterionic/isoelectric buffering compound (aspartic acid) with regard to separation and enrichment efficiency. With the second method, the achieved limits of detection are as low as 0.1 µg mL−1 for Ado and 0.2 µg mL–1 for Cyd. The applicability of the two complementary methods to the analysis of the nucleosides under investigation is shown for blank and spiked human urine samples after their extraction using the commercially available phenylboronate affinity gel.

Graphical Abstract

Keywords

Micellar electrokinetic chromatography Ionic liquid-type surfactant Urinary nucleosides Borate complexation “Pseudostationary ion-exchanger” sweeping Phenylboronate affinity gel 

Introduction

The increasing interest in metabolomics and the diagnostic value of urinary nucleosides as cancer biomarkers have drawn the attention of many working groups to develop analytical methods [1, 2, 3, 4, 5, 6, 7] for their analysis in the urine of healthy and cancer patients. Micellar electrokinetic chromatography (MEKC) is one of the capillary electromigration separation techniques that is widely involved in the analysis of these highly polar metabolites (as neutral compounds) using mostly sodium dodecyl sulfate (SDS) as surfactant [8, 9, 10, 11, 12, 13, 14, 15]. Due to the low interaction between SDS and these analytes under neutral pH conditions, a high concentration of the pseudostationary phase is required for their separation, a problem that is typically associated with high electric current, Joule heating, noisy base line, irreproducible migration times and long analysis time [16]. With the need for new surfactants that are capable of interacting more strongly with these highly hydrophilic analytes, providing better resolution while maintaining reproducible migration times and short run time, we have reported an MEKC method for the analysis of nucleosides using the ionic liquid (IL)-type surfactant 1-tetradecyl-3-methylimidazolium bromide (C14MImBr). It was shown that the separation of these polar analytes can be achieved in less than 6 min at low surfactant concentration using C14MImBr as cationic surfactant under alkaline pH conditions, where the nucleosides are negatively charged [16].

C14MImBr as PSP in MEKC provides a higher repeatability in migration times than the conventional cationic surfactants (e.g., n-alkyltrimethylammonium salts) and a modified selectivity, which are attributed by us to the efficiency by which C14MImBr dynamically coats the inner capillary wall and to the versatility of the interaction sites provided by the imidazolium head group [16]. Using only 20 mmol L−1 C14MImBr in the BGE (5 mmol L−1 tetraborate, pH 9.38), a complete separation of all the studied analytes (existing as negatively charged complexes) is achieved. In our previous work, we have demonstrated the dependency of kBGE (retention factor in the BGE) on the borate concentration of the BGE [16]. We have found that the kBGE values are the highest at the lowest borate concentration. Therefore and based on the classical theory of ion-exchange chromatography, the micelles formed by C14MImBr can be regarded as a pseudostationary ion-exchanger that provides a fixed concentration of ion-exchange sites and enables the regulation of the retention factor by variation of the competing ion concentration. Besides, also the pH of the BGE strongly influences the kBGE values of the nucleosides. The highest values of kBGE are obtained at the pH that enables a full deprotonation of these analytes and hence permits strongest interaction with the oppositely charged PSP.

Extending the study of the fundamental aspects underlying the separation of the nucleosides using C14MImBr [16], we want to integrate the outcome of our previous results, with emphasizing on the applicability of C14MImBr as a novel PSP in the analysis of the hydrophilic nucleosides in real urine samples. We especially consider the involvement of online enrichment techniques for a further improvement of the detection limits. “Pseudostationary ion-exchanger” sweeping (using C14MImBr) will be investigated as an online enrichment technique for the focusing of the investigated nucleosides when employing a sample matrix void of PSP. The possibility of the involvement of other enrichment mechanisms will be also taken into consideration.

It is noteworthy to mention that in [16] we have increased kBGE of the studied nucleosides by increasing the pH and reducing the competing ion (co-ion) concentration in the BGE. However, for the nucleosides Ado and Cyd (the first-migrating nucleosides), kBGE (and consequently also the retention factor in the sample kS) is extremely low even under optimized conditions (low tetraborate concentration and high pH). This will negatively affect their sweeping efficiency with C14MImBr. As an alternative to this approach, we will show that the analysis of these two compounds under acidic conditions permits their efficient “pseudostationary ion-exchanger” sweeping when using the oppositely charged surfactant SDS and a BGE containing a zwitterionic/isoelectric buffering component (as defined by [17]). Under these conditions, it is possible to minimize the dramatic effect of the competing ion concentration on kS and hence on the sweeping efficiency.

Optimization of the ionic strength (buffer concentration) and the pH of the sample matrix with regard to its significant effect on the sweeping efficiency and the sensitivity of the developed method has not been considerably discussed in the literature [18] and will be the main focus of the presented study. According to the best of our knowledge, RFGE-“pseudostationary ion-exchanger” sweeping using: (1) C14MImBr under alkaline pH conditions [16] (Method 1) or (2) SDS under acidic pH conditions (Method 2) for the analysis of polar nucleosides as charged compounds in real urine samples has not been reported so far.

In this work, we investigate the optimization of the sample matrix and the BGE with respect to maximum focusing efficiency while maintaining adequate resolution. Adjustment of kBGE is studied by variation of the pH, the buffer concentration, or via the use of a zwitterionic/isoelectric buffering compound. In addition, we optimize the sample injection volume by variation and selection of those parameters (composition of sample matrix and injection conditions) that give rise to the highest peak height and highest peak area, while maintaining at the same time, acceptable peak shapes and resolution. The underlying focusing mechanisms are discussed. The developed methods are validated according to the ICH guidelines [19]. With several blank and spiked urine samples, the applicability of the developed and validated methods is demonstrated after the selective extraction of the target analytes with a phenylboronate affinity gel (PBA).

Theoretical Considerations

Sweeping of Neutral Analytes

Sweeping is one of the online enrichment techniques that is used to overcome the poor concentration sensitivity in CE. It is defined as the accumulation of analyte by the pseudostationary phase (PSP) that penetrates the sample zone (being void of PSP) under an applied voltage. Based on the concept presented by Terabe and co-workers [20, 21], the length of the sample zone after sweeping Lsweep depends on the initial sample-plug length Linj and on the retention factor in the sample zone kS during sweeping. The enrichment factor (= Linj/Lsweep) is then given by 1 + kS:
$${L_{sweep}}=\frac{1}{{1+{k_S}}}{L_{inj}}.$$
(1)

According to Eq. (1), the adjustment of the retention factors within the sample zone is the key factor to maximize the sweeping efficiency and hence the sensitivity of the developed method.

In the following, we will distinguish between Lsweep (the zone length after sweeping) and Lfocus (the zone length after focusing that might involve sweeping and other focusing principles). Counterintuitively, it is not possible to combine sweeping and field-amplified sample stacking (FASS) [22]. El-Awady et al. [23] verified experimentally for neutral parabens with samples of varied electric conductivity that under the conditions of their measurements the enrichment factors are in first approximation independent of the electric conductivity of the sample matrix.

However, because of the retention factor gradient effect (RFGE), Lfocus is not independent of the ratio KBGE/KS (ratio partitioning coefficient of the solute in the BGE to partitioning coefficient of the solute in the sample). This ratio might be different from unity, if the sample contains a solvent that is not contained in the BGE or if the sample has another pH than the BGE. Ideally, kS should be maximized (by maximizing KS), while kBGE will be adjusted to the optimum value for the separation (via reducing KBGE) [24]:
$${L_{focus}}=\frac{{{K_{BGE}}}}{{{K_S}}} \cdot \frac{1}{{1\;+\;{k_{BGE}}}}{L_{inj}}.$$
(2)
In another form, we can also write [24, 25]:
$${L_{focus}}=\frac{{{k_S}{k_{BGE}}+\;{k_{BGE}}\;}}{{\;{k_S}\,{k_{BGE}}\;+\;{k_S}\;}} \cdot \frac{1}{{1\;+\;{k_S}}}{L_{inj}},$$
(3a)
$${L_{focus}}\;=\;\;\frac{1}{f} \cdot \frac{1}{{\;(1\;+\;{k_S})}}\;{L_{inj}}\;=\;\frac{{\;{L_{sweep}}\;}}{f}.$$
(3b)
Here, f is the additional focusing/defocusing factor due to the RFGE. El-Awady and Pyell [25] reported that the factor f could reach values up to three for neutral analytes under conditions of an additional focusing due to RFGE. The validity of Eqs. (3a, 3b) was confirmed experimentally [25]. Pyell et al. [24] also studied the case that the injection zone of a neutral analyte is focused by sweeping under inhomogeneous electric field conditions in the presence of RFGE. The derived equation is identical to Eq. (2). This result confirms that for neutral solutes under typical sweeping conditions and under non-homogeneous electric field conditions, the enrichment factor is independent of the electric conductivity of the sample solution and influenced by RFGE as in sweeping under homogeneous electric field conditions.
RFGE is principally independent of sweeping. There are conditions, in which the sample zone and the BGE have identical concentration of the PSP, while KS and KBGE can differ, e.g., because of the addition of an organic solvent to the BGE that is not contained in the sample solution. In that case, we obtain
$${L_{focus}}\;=\;\frac{{\;{k_S}\,{k_{BGE}}\;+\;{k_{BGE}}\;}}{{\;{k_S}\,{k_{BGE}}\;+\;{k_S}\;}}\;{L_{inj}}\;.$$
(4)

Sweeping of Charged Analytes

In [24] we have shown for the general case that the enrichment factor for a charged analyte due to sweeping equals that of a neutral analyte (see Eq. 1). However, this statement is only valid under homogeneous electric field conditions in the absence of RFGE. Under homogeneous electric field conditions in the presence of RFGE, we obtain [24]
$${L_{focus}}\;=\;\frac{{\;\left( {{k_{BGE}}\;+\;\xi } \right)\;}}{{\;\left( {{k_S}\;+\;\xi } \right)\;}}\; \cdot \frac{1}{{1\;+\;{k_{BGE}}}}\;{L_{inj}}\;,$$
(5)
where \({\xi}={{{\mu _{epA}}} \mathord{\left/ {\vphantom {{{\mu _{epA}}} {{\mu _{PSP}}}}} \right. \kern-0pt} {{\mu _{PSP}}}}\)(effective electrophoretic mobility of the analyte divided by the electrophoretic mobility of the pseudostationary phase).
We have also shown for inhomogeneous electric field conditions in the absence of RFGE and for \({\mu _{epA}}\) = const. and \({\mu _{PSP}}\) = const.’ (considering the case that dynamic pH junction is absent) that the following equation is valid [24]:
$${L_{focus}}\;=\;\frac{{\;{k_{BGE}}\;+\;\xi \;}}{{\;\theta \; \cdot \;{k_{BGE}}\;+\;\xi \;}}\; \cdot \;\frac{1}{{\;\gamma \;}}\; \cdot \;\frac{{\;1\;}}{{\;1\;+\;{k_{BGE}}\;}}\; \cdot \;{L_{inj}}\;,$$
(6)
where γ is the ratio of electric conductivity of the BGE to electric conductivity of the sample solution, θ ≈ 1/γ. For γ > 1 and positive ξ, the first factor will be smaller than 1/θ, for γ > 1 and negative ξ, the first factor will be larger than 1/θ, for γ < 1 and positive ξ, the first factor will be larger than 1/θ, and for γ < 1 and negative ξ, the first factor will be smaller than 1/θ [24]. Additional focusing is reached, when the first factor is smaller than 1/θ, while additional defocusing is reached, when the first factor is larger than 1/θ.
The situation is becoming more complicated for the more general case of a charged analyte focused by sweeping under inhomogeneous electric field conditions in the presence of RFGE considering the case that dynamic pH junction is absent [24]:
$${L_{focus}}\;=\;\frac{{\;{k_{BGE}}\;+\;\xi \;}}{{\;\alpha \; \cdot \;\theta \; \cdot \;{k_{BGE}}\;+\;\xi \;}}\; \cdot \;\frac{1}{{\;\gamma \;}}\; \cdot \;\frac{{\;1\;}}{{\;1\;+\;{k_{BGE}}\;}}\; \cdot \;{L_{inj}}\;,$$
(7)
where α = KBGE/KS (ratio partitioning coefficient of the analyte in the BGE to partitioning coefficient of the analyte in the sample). If α is very large, ξ can be neglected and we obtain Eq. (2), while assuming that θ ≈ 1/γ. Hence, we can conclude that for a charged analyte, the enrichment factor due to sweeping under inhomogeneous electric field conditions in the presence of RFGE is mostly influenced by the retention factor of the analyte in the sample zone as with a neutral analyte. For high α, the impact of the coupling with intrinsic stacking and/or destacking can be neglected.

It should be also noted that in the case of a sample with an electric conductivity adapted to that of the BGE (sweeping under homogenous electric field conditions), the retention factor in the sample zone kS is the retention factor that is obtained in a buffer, which contains the PSP in a concentration identical to that of the BGE in a matrix which is identical to that of the injected sample solution [23]. However, in the case of a sample with an electric conductivity not adapted to that of the BGE (sweeping under inhomogeneous electric field conditions), kS cannot be obtained with the same method. The reason behind this is the fact that sweeping under inhomogeneous electric field conditions is a multistep enrichment process including (1) stacking or destacking of the PSP when entering the sample zone, (2) sweeping of the neutral analytes by the stacked or destacked PSP, and (3) destacking or stacking of the swept analyte zone [23].

“Pseudostationary Ion-Exchanger” Sweeping

One approach to reach a high α for acidic, amphoteric or basic analytes is to employ a sample solution that has a pH different from that of the BGE. In the sample zone, KS will be maximized by selection of a pH, for which the analytes have a high degree of dissociation or a high degree of protonation. Simultaneously, in the BGE compartment, KBGE will be reduced with selection of a pH, for which the analytes have a low degree of dissociation or a low degree of protonation. This pH difference allows maximizing kS and brings kBGE in the optimum range for separation. At the same time, the analytes will be present in the sample zone and the BGE compartment in two different species that do not only differ in their partition coefficient with the PSP but also in their effective charge number and consequently in their effective electrophoretic mobility and in their pseudoeffective electrophoretic mobility (that is the observed effective electrophoretic mobility due to interaction with the PSP).

Under these conditions, the interaction of the analyte with the PSP in the sample zone will be via Coulombic interaction or via hydrophobically assisted Coulombic interaction [26]. Hydrophobically assisted Coulombic interaction allows to reach a very high kS, when the analyte has a high effective charge number and a hydrophobic structure unit. The PSP in the sample zone must have a charge of opposite sign than that of the analyte. In addition, the concentration of the co-ion (having a charge of same sign as the analyte, also denoted the competing ion) in the sample solution must be minimized, as it strongly influences the obtainable enrichment efficiency (a mechanism known from ion-exchange chromatography) [27]. We therefore suggest to call this type of focusing “pseudostationary ion-exchanger” sweeping (PSIE-sweeping), as it can best be described as sweeping with a pseudostationary ion-exchanger.

However, with this strategy, it will be unavoidable to introduce a pH difference between the sample zone and the BGE compartment. The two boundaries at the beginning and the end of the sample zone are then not only “starting lines” for moving concentration boundaries (transient moving step-wise change of the concentration of the PSP) but also “starting lines” for moving pH boundaries (transient moving step-wise change of the pH). In PSIE-sweeping, sweeping is unavoidably combined with dynamic pH junction [28].

In principle, sweeping and dynamic pH junction are very similar online focusing methods as they both are based on the modification of the (pseudo-)effective electrophoretic mobility of the analyte that passes through a moving reaction boundary (regarding associate formation and/or protonation/deprotonation as reactions). While sweeping is induced by a difference in the concentration of the PSP between the sample zone and the BGE compartment, focusing by dynamic pH junction [29] is induced by a difference in the pH between the sample zone and the BGE compartment. Principally, dynamic pH junction can be performed either with low-pH sample/high-pH BGE or with high-pH sample/low-pH BGE systems [30]. In all cases, a moving reaction boundary (at which protonation/deprotonation takes place) is developed.

Several authors report that the coupling of sweeping and dynamic pH junction permits to realize higher enrichment factors than with exclusively sweeping or exclusively dynamic pH junction [31, 32, 33, 34, 35, 36, 37]. In contrast to other inhomogeneities introduced by sample injection (e.g., inhomogeneity of the electric conductivity, inhomogeneity of the organic solvent content), the produced pH boundaries cannot be stationary boundaries. In all instances, dynamic pH junction-sweeping includes two moving boundaries which are modelled with two moving acceleration or deceleration planes that traverse the sample zone with different velocities (in co- or in counter-direction) [24].

The modelling of the processes involved in dynamic pH junction is more sophisticated than the modelling of the processes involved in sweeping. The exact mathematical description of moving pH disturbances (e.g., H+ eigenpeaks or OH eigenpeaks [38]) requires the numerical solution of inherently non-linear equations [39]. It is apparent that the optimization of online focusing schemes based on a combination of dynamic pH junction with sweeping is mostly done on an empirical basis.

If borate is a component of the BGE, while the sample is void of borate, PSIE-sweeping of nucleosides combines dynamic pH junction with two sweeping steps. The first sweeping step is due to the formation of a charged tetrahydroxyborate/cis-diol-complex. The second sweeping step is due to the formation of an associate of the charged tetrahydroxyborate/cis-diol-complex with the oppositely charged micellar PSP. Modelling of this “dynamic pH junction-dual-step-sweeping” comprises then the assumption of three different moving acceleration or deceleration planes [24] that traverse the sample zone with different velocities.

One limitation of PSIE-sweeping should not be overlooked: (in most cases) the involvement of sections of the capillary filled with solutions differing in their pH unavoidably will induce large differences in the local electroosmotic mobility. Simultaneously, differences in the local electroosmotic mobility will evoke local pressure differences. These local pressure differences generate a partial parabolic flow profile that causes band broadening and strongly counteracts focusing via PSIE-sweeping [40]. Hence, the enrichment efficiency obtainable with PSIE-sweeping will be dependent on a very large set of experimental parameters and might be best optimized by a combination of theoretical considerations and (empirical) experimental design.

Experimental

Chemicals and Background Electrolytes

Cytidine (Cyd), adenosine (Ado), 5-methyluridine (5MeUrd), uridine (Urd), guanosine (Guo), inosine (Ino), and xanthosine (Xao) were purchased from Sigma–Aldrich, Steinheim, Germany; chemical structures, pKa [41] and lg Pow [42] see Fig. S1, supplementary data. Abbreviations are given according to the IUPAC-IUB commission on biochemical nomenclature [43]. Sodium dodecyl sulfate (SDS), hydrochloric acid, orthophosphoric acid (85%), l-aspartic acid, and sodium hydroxide were from Fluka, Buchs, Switzerland. Disodium tetraborate decahydrate (borax) and sodium dihydrogen phosphate monohydrate were from Merck, Darmstadt, Germany. Methanol, HPLC grade was from VWR-BDH-Prolabo, Leuven, Belgium. Formic acid (98–100%), sodium chloride and ammonium hydroxide (25%) were from Sigma–Aldrich. The Affi-gel 601, used as solid phase for the extraction of nucleosides from urine, was purchased from Bio-Rad (Hercules, CA, USA). Ammonium acetate buffer (0.25 mol L−1, pH 8.80) was prepared by dissolving 9.6350 g of the salt in 400 mL water, adjusting with concentrated ammonium hydroxide (25%) to pH 8.80 and then diluting to 500 mL with water. The synthesis and characterization of C14MImBr are described in detail in [16].

Stock solutions of tetraborate buffer, phosphate buffer and l-aspartic acid were prepared in water and further diluted for the preparation of the background electrolytes. Stock disodium tetraborate buffer (100 mmol L−1, pH 9.48) was prepared by dissolving 9.5342 g disodium tetraborate decahydrate in 250 mL of water. Stock phosphate buffer (50 mmol L−1, pH 2.66) was prepared by dissolving 1.3799 g (40 mmol L−1) of sodium dihydrogen phosphate monohydrate in 200 mL of water, adding of 0.17 mL of concentrated orthophosphoric acid (10 mmol L−1), and diluting to 250 mL with water. Stock l-aspartic acid (25 mmol L−1, pH 2.86) was prepared by dissolving 0.8319 g l-aspartic acid in 250 mL of water.

BGEs were: (1) 20 mmol L−1 C14MImBr in 5 mmol L−1 sodium tetraborate, pH 9.38 (without any pH adjustment), (2) 20 mmol L−1 C14MImBr in 10 mmol L−1 disodium tetraborate adjusted to pH 9.02 using 1 mol L−1 HCl, (3) 100 mmol L−1 SDS in 50 mmol L−1 phosphate buffer, pH 2.82, and (4) 100 mmol L−1 SDS in 25 mmol L−1l-aspartic acid, pH 3.22 (without any pH adjustment). All buffer solutions were filtered prior to use through a 0.45-µm nylon membrane filter (WICOM, Heppenheim, Germany). BGEs were replaced after every four runs.

Instrumentation

All measurements were done using the ATI Unicam CE System, Crystal 300 Series, Model 310 equipped with UV/Vis detector Spectra 100 (with deuterium lamp) from Thermo Separation Products, San Jose, USA, set to a wavelength of 257 nm (optimized wavelength). Data acquisition was done using an A/D-converter (USB-1280FS, Measurement Computing, Middleborough, USA). Data were recorded using CE-Kapillarelektrophorese software (development of the electronic workshop of the Department of Chemistry, University of Marburg based on Delphi). Data analysis was performed with Origin 8.5 software (OriginLab Corporation, Northampton, USA). Fused silica-capillaries (50 µm I.D., 360 µm O.D.) were obtained from Polymicro Technologies (Phoenix, AZ, USA), with a total length of 649 mm and a length to the detector of 501 mm (if not stated otherwise). InoLab pH 720 (WTW, Weilheim, Germany) was used for pH measurements. Solid phase extractions were performed on a vacuum manifold column processor (J.T. Baker, Griesheim, Germany). The flow rate during sample loading and elution is 0.5 mL/min. The eluate obtained after the extraction procedure was lyophilized in a Christ Alpha 2–4 LSC Freeze Dryer (Martin Christ, Osterode am Harz, Germany).

New capillaries were conditioned by flushing them first with NaOH solution (1 mol L−1) 60 min, water 60 min, and BGE 15 min using an applied pressure of 800 mbar. For Method 1, the capillaries were rinsed between runs with methanol 2 min, HCl (1 mol L−1) 2 min, water 2 min, NaOH solution (1 mol L−1) 2 min, water 2 min and finally with BGE for 2 min using an applied pressure of 800 mbar. For Method 2, the capillaries were rinsed between runs with BGE for 5 min. Peak identities were confirmed by spiking.

Urine Samples and Extraction Conditions

Samples of human urine were obtained from a female 27-year-old healthy volunteer and were collected in 100-mL plastic bottles and frozen immediately until analysis. Before use, the samples were thawed at room temperature. For the study of spiked urine samples, samples were pretreated with a phenylboronate affinity gel (PBA) column. The extraction conditions are based on what was previously reported [11, 15, 44, 45]. The exact extraction procedure that was performed in the current work is described in detail in [1]. In case of Method 1, the eluate after drying was redissolved in 2 mL 2.5 mmol L−1 disodium tetraborate, pH 10.45, whereas it was reconstituted in 2 mL of water in case of Method 2.

Preparation of Standard Solutions and Calibration Curves

All single analyte stock solutions (800.0 mg L−1 of Ado, Cyd, Urd, 5MeUrd, Ino, 400.0 mg L−1 of Xao, Guo) were prepared in water and stored in the refrigerator (stock solutions of nucleoside standards are used within 1 month). The working standard solutions were prepared daily (concentration of each of the studied nucleosides in the sample solution mixture 20 mg L−1, unless otherwise specified).

It is unsuitable to prepare the reference samples in urine as the tested analytes are endogenously present in urine [46]. Distilled water was used as a surrogate or artificial matrix for the preparation of the standard solutions, as the studied analytes are highly polar [15]. The concentration ranges for the investigated analytes are listed in Table 1.

Table 1

Linear regression parameters of the developed methods for calibration standards after SPE

Studied analytesa

No. of calib. standards

Linearity range (µg mL−1)

Intercept (a) ± SDb

Slope (b) ± SDc

SSEd

S yx e

S x0 f

Sr%g

Confidence interval of (a)h

Confidence interval of (b)h

r i

Mandel’s test valuej

LODk (µg mL−1) S/N = 3

LOQl (µg mL−1) S/N = 10

5MeUrd

 PHm

8

0.50–10.0

0.05 ± 0.03

0.30 ± 0.01

94.01

0.05

0.17

4.31

± 0.10

± 0.02

0.9991

6.31

0.1

0.2

 PAn

9

0.50–12.0

0.03 ± 0.04

0.49 ± 0.01

151.07

0.08

0.17

3.46

± 0.15

± 0.02

0.9993

0.36

  

 Corr. PAo

9

0.50–12.0

0.01 ± 0.01

0.13 ± 0.00

151.07

0.02

0.18

3.71

± 0.04

± 0.01

0.9992

0.76

  

Urd

 PH

7

0.50–8.00

0.08 ± 0.02

0.26 ± 0.01

52.68

0.04

0.16

4.98

± 0.09

± 0.02

0.9988

16.66

0.1

0.2

 PA

9

0.50–12.0

0.00 ± 0.05

0.64 ± 0.01

151.02

0.09

0.14

2.87

± 0.16

± 0.03

0.9995

2.20

  

 Corr. PA

9

0.50–12.0

0.00 ± 0.01

0.16 ± 0.00

151.02

0.02

0.12

2.54

± 0.04

± 0.01

0.9996

1.49

  

Guo

 PH

8

0.50–10.0

0.07 ± 0.04

0.30 ± 0.01

93.74

0.07

0.24

5.99

± 0.15

± 0.03

0.9982

11.25

0.1

0.2

 PA

9

0.50–12.0

− 0.07 ± 0.07

0.85 ± 0.01

150.73

0.13

0.15

3.07

± 0.23

± 0.04

0.9995

0.00

  

 Corr. PA

9

0.50–12.0

− 0.01 ± 0.01

0.19 ± 0.00

150.73

0.03

0.14

2.91

± 0.05

± 0.01

0.9995

0.14

  

Ino

 PH

7

0.50-8.00

0.09 ± 0.05

0.35 ± 0.01

52.59

0.08

0.22

7.09

± 0.18

± 0.04

0.9977

3.64

0.1

0.2

 PA

9

0.50–12.0

0.03 ± 0.10

1.14 ± 0.02

150.88

0.19

0.17

3.48

± 0.35

± 0.06

0.9993

0.18

  

 Corr. PA

9

0.50–12.0

0.01 ± 0.02

0.23 ± 0.00

150.88

0.04

0.17

3.56

± 0.07

± 0.01

0.9993

0.48

  

Xao

 PH

7

0.50–8.48

0.09 ± 0.02

0.24 ± 0.01

59.61

0.04

0.15

4.60

± 0.08

± 0.02

0.9990

9.09

0.1

0.2

 PA

8

0.50–10.6

0.03 ± 0.08

1.00 ± 0.01

106.16

0.15

0.15

3.53

± 0.30

± 0.05

0.9994

7.78

  

 Corr. PA

8

0.50–10.6

0.01 ± 0.02

0.19 ± 0.00

106.16

0.03

0.14

3.34

± 0.05

± 0.01

0.9994

15.34

  

Ado

 PH

9

0.20–16.0

0.05 ± 0.03

0.27 ± 0.00

263.82

0.06

0.23

3.74

± 0.11

± 0.01

0.9993

0.84

0.1

0.2

 PA

9

0.20–16.0

− 0.05 ± 0.09

0.88 ± 0.01

263.82

0.18

0.21

3.36

± 0.32

± 0.04

0.9994

0.65

  

 Corr. PA

9

0.20–16.0

0.00 ± 0.01

0.11 ± 0.00

263.82

0.02

0.20

3.25

± 0.04

± 0.01

0.9995

0.39

  

Cyd

 PH

7

0.40–16.0

0.03 ± 0.01

0.04 ± 0.00

181.39

0.02

0.49

6.29

± 0.05

± 0.01

0.9967

0.14

0.2

0.4

 PA

7

0.40–16.0

0.07 ± 0.04

0.13 ± 0.00

181.39

0.05

0.40

5.14

± 0.14

± 0.02

0.9978

0.22

  

 Corr. PA

7

0.40–16.0

0.01 ± 0.00

0.02 ± 0.00

181.39

0.01

0.37

4.76

± 0.02

± 0.00

0.9981

0.14

  

aThe linear regression parameters for 5MeUrd, Urd, Guo, Ino, and Xao are obtained using Method 1, whereas the linear regression parameters for Ado and Cyd are obtained using Method 2

bStandard deviation of the intercept

cStandard deviation of slope

dSum of square errors

eResidual standard deviation [standard error of estimate (SEE)]

fMethod standard deviation (= Syx/b)

gRelative standard deviation of the method (= Sx0/\(\overline {x}\))

hConfidence interval calculated at P = 0.99

ir is the correlation coefficient

jTabulated f values at P = 0.99 are 13.74. 16.25, 21.19 at [df1 (1), df2 (n − 3) = (1,6), (1,5), (1,4)], respectively

kLimit of detection

lLimit of quantitation

mPeak height

nPeak area

oCorrected peak area

Following optimized method parameters were used: capillary 649(501) mm × 50.2 µm I.D., separation voltage − 20 kV, injection pressure 50 mbar for 1 min, detection 257 nm (Method 1 and Method 2). BGEs are 20.0 mmol L−1 C14MImBr in 5 mmol L−1 disodium tetraborate (Method 1, pH 9.38) and 100 mmol L−1 SDS in 25 mmol L−1l-aspartic acid (Method 2, pH 3.22). Oven temperature is 35 °C (Method 1) or 25 °C (Method 2). Calibration curves were constructed by plotting peak height, peak area, or corrected peak area against the corresponding analyte concentration in the sample solution (in µg mL−1).

Results and Discussion

PSIE-Sweeping Using C14MImBr Micelles (Method 1)

MEKC separation of negatively charged nucleosides with oppositely charged IL-based surfactant C14MImBr is based on both electrophoretic and chromatographic phenomena [16]. A detailed description of the optimization of separation conditions is given in [16]. In the present study, we will focus on the optimization of the conditions for the online enrichment of the studied nucleosides via PSIE-sweeping. Here three important points must be highlighted and taken into consideration. (1) The degree of complex formation between tetrahydroxyborate and the cis-diol moieties of the nucleosides is close to one even at very low borate concentration (2.5 mmol L−1 tetraborate) [16], which gives an indication about the very high complex formation constants of the formed complexes. (2) Tetrahydroxyborate in the micellar BGE acts as the buffering, complexing and competing ion [16]. As the concentration of tetrahydroxyborate in the BGE increases, it reduces the degree of association of the nucleosides (associated to the PSP by electrostatic forces), which results in a lowering of their kBGE and a reduction of their migration times [16]. Lowering of the kBGE and reduction of migration times will both have a negative impact on the quantitative analysis in urine samples, as under these conditions there is a high possibility of coelution of the studied analytes with matrix components. Based on our previous discussion, we will therefore restrict our investigations to BGEs, which contain a maximum of 10 mmol L−1 disodium tetraborate. (3) The pH of the sample solution and the BGE affect both the degree of analyte protonation/deprotonation (in the respective compartment) and the degree of borate complexation (in the respective compartment), which consequently affects both the kS and kBGE values of the nucleosides and their sweeping by C14MImBr.

The following equations are describing the equilibria that are involved in the separation of the nucleosides investigated (protonation/deprotonation, borate complexation, and association with the charged PSP):
$$N \rightleftharpoons N - {O^ - }+{H^+},$$
(8a)
$$N+{B^ - } \rightleftharpoons {\left[ {BN} \right]^ - },$$
(8b)
$$N - {O^ - }+{B^ - } \rightleftharpoons {\left[ {BN - {O^ - }} \right]^{2 - }},$$
(8c)
$$N+{M^{x+}} \rightleftharpoons \left[ {N{M^{x+}}} \right],$$
(8d)
$$N - {O^ - }+{M^{x+}} \rightleftharpoons \left[ {(N - {O^ - }){M^{x+}}} \right],$$
(8e)
$${\left[ {BN} \right]^ - }+{M^{x+}} \rightleftharpoons \left[ {{{(BN)}^ - }{M^{x+}}} \right],$$
(8f)
$${\left[ {BN - {O^ - }} \right]^{2 - }}+{M^{x+}} \rightleftharpoons \left[ {{{(BN - {O^ - })}^{2 - }}{M^{x+}}} \right],$$
(8g)
where N is the nucleoside, Bˉ is the tetrahydroxyborate anion, [BN]ˉ is the complexed neutral form of the nucleoside, N–O is the deprotonated form of the nucleoside, [BN–O]2− is the complexed deprotonated form of the nucleoside, Mx+ is the micelle.

In the following discussion, the given kS values are taken from [16] (refer to Tables S1 and S2, supplementary data). The parameter kS of the analytes in the sample solution was determined via two different experimental series. The first experimental series was performed in the CZE mode (using 2.5 mmol L−1 disodium tetraborate pH 9.03 and 10.43 as BGEs) to calculate µeff of each of the studied nucleosides at the given pH value and the second series was conducted in the MEKC mode under exactly the same conditions as in the CZE mode using 20 mmol L−1 C14MImBr in 2.5 mmol L−1 disodium tetraborate as BGE to calculate \(\mu _{{{\text{eff}}}}^{P}\) (the pseudoeffective electrophoretic mobility of the analyte in micellar BGE) and µMC (the electrophoretic mobility of the micelles) (for more details see [16]). Then the parameter kS can be determined from the three mobilities (\(\mu _{{{\text{eff}}}}^{P}\), µeff, and µMC) obtained in the two experimental series as described in [16].

The enrichment efficiency of the nucleosides is mainly dependent on kS [Eqs. (17)]. Very high kS values can be obtained by a high pH and a low competing ion concentration in the sample zone [16]. The enrichment efficiency is also dependent on kBGE via RFGE and on the pH of the sample zone and the BGE compartment via coupling with dynamic pH junction. The pH of the BGE was preselected to be 9.02 [16] as this value permits a good resolution of the analyte zones.

Our first trials were carried out by varying the pH and the composition of the sample matrix. In Fig. 1, four cases were studied (each at three different sample injection volumes, see figure caption) using Guo and Xao as representative analytes. In all cases, the composition of the BGE is 20 mmol L−1 C14MImBr in 10 mmol L−1 disodium tetraborate, pH 9.02. We are varying the pH of the sample matrix (10.45 or 9.02) either in the presence (non-sweeping conditions) or in the absence of C14MImBr micelles (sweeping conditions), while under sweeping conditions the electric conductivity of the sample was considerably lower than that of the BGE. The parameter enhancement factor (EF) will be used to compare between the four studied cases. Calculation of this factor was carried out using the approach that was described in detail in our previously published work [1]. According to this approach:

Fig. 1

Electropherograms obtained with samples containing Guo and Xao (each 20.0 mg L−1) dissolved in four different sample matrices including a 2.5 mmol L−1 sodium tetraborate pH 10.45, b 2.5 mmol L−1 sodium tetraborate pH 9.02, c 20 mmol L−1 C14MImBr in 2.5 mmol L−1 sodium tetraborate pH 10.45 and d BGE (no FASS + no dynamic pH junction + non-sweeping condition) employing three different injection parameters. CE conditions: 20 mmol L−1 C14MImBr in 10 mmol L−1 disodium tetraborate, pH 9.02 as BGE, hydrodynamic injection using pressure (A1, B1, C1, D1) 40 mbar for 0.4 min, (A2, B2, C2, D2) 50 mbar for 0.8 min, (A3, B3, C3, D3) 70 mbar for 2 min, capillary 649(501) mm × 50 µm I.D, applied voltage − 20 kV, oven temperature 35 °C. Peak designation: 4 = Guo, 5 = Xao

$${\text{EF}}={h_2}/{h_1},$$
(9)
where h1 is the limiting peak height in the plateau region under non-focusing conditions (analytes are dissolved in BGE), and h2 is the limiting peak height in the plateau region under focusing conditions (analytes are dissolved in sample matrix void of C14MImBr micelles). However, with the developed enrichment method, it is impossible to reach the volume overload region. Therefore, h2 was defined to be the peak height obtained for a given concentration of the nucleoside using the maximum sample injection volume that provides an acceptable peak shape. On the other hand, h1 was calculated from the saturation curve (using identical nucleoside concentration) by non-linear regression (for further details, see [1]). This definition of h1 and h2 assures that the resulting EF constitutes a minimum value, i.e., the correct EF value will be slightly higher [24].

In Fig. 1a, 2.5 mmol L−1 disodium tetraborate (adjusted to pH 10.45 using 1 mol L−1 NaOH) is used as sample matrix (online focusing via “pseudostationary ion-exchanger” sweeping). From previous studies [16], we know that kS [with c(C14MImBr) = 20 mmol L−1 and c(borax) = 2.5 mmol L−1] is higher with pH 10.45 than with pH 9.02: kS at pH 10.45 is 5.1 for Guo and 6.2 for Xao and kS at pH 9.02 is 1.4 for Guo and 5.2 for Xao.

As seen in Fig. 1a, the peak heights for Guo and Xao increase by increasing the sample injection volume (Fig. 1A1, A2 and A3). Sample injection volumes higher than that used in Fig. 1A3 can be applied without affecting peak shape and resolution between the two studied analytes. Hydrodynamic injection using a pressure of 80 mbar for 3.0 min is the maximum sample injection volume that can be applied without loss of resolution (Fig. S2, supplementary data). The parameter h2 was determined with the maximum sample injection volume resulting from hydrodynamic injection at 80 mbar for 3.2 min with Guo and 80 mbar for 4 min with Xao. This parameter was then divided by h1 (Fig. S3, supplementary data). The resulting EF is 32.7 for Guo and 22.8 for Xao.

In Fig. 1b, the sample matrix is 2.5 mmol L−1 sodium tetraborate adjusted to pH 9.02 using 1 mol L−1 HCl. Under these conditions, focusing due to RFGE still exists due to different concentrations of tetrahydroxy borate in the sample matrix and in the BGE. This difference causes KS > KBGE. Deterioration of the peak shape of Guo was obtained at larger sample injection volume (70 mbar for 2 min, Fig. 1B3). The reason behind this very characteristic peak distortion might be boundary stacking of the PSP [24, 25]. On the other side, the peak height of Xao in Fig. 1B3 is only slightly lower than that obtained with a BGE of higher pH, in full agreement with the similar kS of Xao at the two pH studied (see above). We ascribe this observation to the fact that the amidic group of Xao is fully dissociated within the pH range investigated. However, we also know that Xao-tetrahydroxyborate does not have the full twice negative charge at pH 9.02 [16]. From electrophoretic measurements we know that the degree of complex formation reaches one at a pH that is higher than the pKa value of boric acid [refer to Eq. (8c)]. This conclusion follows from a comparison of effective electrophoretic mobilities for Xao determined in 2.5 mmol L−1 borate buffer at varied pH (in the absence of PSP) [16]. Via CE, we obtained the following values: µeff(Xao at pH 9.02) = − 3.06 × 10−4 cm2 V−1 s−1 and µeff(Xao at pH 10.45) = − 3.39 × 10−4 cm2 V−1 s−1.

As already described above, also for a sample matrix with pH 9.02, the limiting height h2 was calculated for both Guo and Xao from the peak height at maximum sample injection volume (70 mbar for 1.5 min in case of Guo and 80 mbar for 3 min in case of Xao), while h1 was obtained from the saturation limit (Fig. S3, supplementary data). EF is 13.4 for Guo and 10.8 for Xao. The comparison of EF values shows that a sample matrix containing 2.5 mmol L−1 sodium tetraborate, pH 10.45 has advantages over the conditions employed in Fig. 1b. This sample matrix composition will therefore be used in further investigations.

The sample matrices employed in Fig. 1c (dynamic pH junction + RFGE + non-sweeping conditions) and 1D (no dynamic pH junction + no RFGE + non-sweeping conditions) are 20 mmol L−1 C14MImBr in 2.5 mmol L−1 sodium tetraborate, adjusted to pH 10.45 (using 1 mol L−1 NaOH solution) and 20 mmol L−1 C14MImBr in 10 mmol L−1 sodium tetraborate, adjusted to pH 9.02 (using 1 mol L−1 HCl), respectively. These matrices were studied to elucidate whether dynamic pH junction in combination with RFGE plays a role in the combined focusing process and how much it can contribute to the overall enrichment mechanism. It is important to see that focusing due to RFGE exists also in the absence of sweeping [refer to Eq. (4)]. The peak heights for Xao and Guo (in the volume overload region) are higher in Fig. 1c than those obtained in Fig. 1d, whereas Guo is much more enriched than Xao. This is because the shift in the pseudoeffective electrophoretic mobility (\(\mu _{{{\text{eff}}}}^{P}\)) due to an increase in the pH from 9.02 to 10.45 (accompanied by an increase in c(borax) from 2.5 to 10 mmol L−1) is much higher for Guo than that for Xao (refer to Table S3, supplementary data). Under these conditions, it is possible to determine the saturation limits of the peak heights under focusing and under non-focusing conditions. Consequently, the parameter h2 was determined for both Guo and Xao at the maximum sample injection volume of 70 mbar for 2 min (being taken as an estimate of the saturation limit). This estimate of h2 was divided by h1 obtained via regression analysis (for details refer to Fig. S3, supplementary data). Following this strategy, EF is 2.4 for Guo and 1.4 for Xao. This clear difference in EF found for the two nucleosides follows directly from the different shifts in the degree of dissociation when increasing the pH from 9.02 to 10.45 (see Tables S1 and S2, supplementary data). From these considerations, we conclude that for the two studied analytes, the contribution of dynamic pH junction (produced by a variation in the pH between sample zone and BGE compartment) in combination with RFGE (produced by both a variation in the pH and a variation in the borate concentration between sample zone and BGE compartment) to the overall enrichment process via PSIE-sweeping is relatively small. The EF of 1.4 determined for Xao (via both dynamic pH junction and RFGE) can be regarded as an estimate of the maximum EF that can be reached via RFGE due to a variation in the borate concentration.

Applying the experimental conditions optimized for both Guo and Xao to all investigated nucleosides (using 50 mbar for 1 min as sample injection parameters and 2.5 mmol L−1 sodium tetraborate pH 10.45 as sample matrix), results in a deterioration of the peak efficiency for Ado and Cyd (Fig. 2a). The kS values of these two compounds are very low (kS values at pH 10.45 are 0.33 and 0.17 for Ado and Cyd, respectively [16], the two nucleosides are eluted before the peak of the EOF marker), as these compounds gain their negative charge only via borate complexation.

Fig. 2

Electropherograms obtained with a standard solution containing 20.0 mg L−1 of each of the studied analytes dissolved in different sample matrices including a and c 2.5 mmol L−1 sodium tetraborate pH 10.45 and b water. CE conditions: BGEs are a and b 20 mmol L−1 C14MImBr in 10 mmol L−1 disodium tetraborate, pH 9.02 and c 20 mmol L−1 C14MImBr in 5 mmol L−1 disodium tetraborate, pH 9.38, capillary 649(501) mm × 50 µm I.D, hydrodynamic injection using pressure 50 mbar for 1 min, applied voltage − 20 kV, oven temperature 35 °C. Peak designation: 1 = Cyd, 2 = Urd, 3 = 5MeUrd, 4 = Guo, 5 = Xao, 6 = Ado, 7 = Ino

Besides, for these two analytes, the difference in µeff between the sample zone and the BGE is very low. (− 1.46 × 10−4 cm2 V−1 s−1 at pH 9.02 and − 1.76 × 10−4 cm2 V−1 s−1 at pH 10.45 [16], both Ado and Cyd in 10 mmol L−1 sodium tetraborate), which will be associated with an insignificant focusing by dynamic pH junction. We therefore conclude that FASS (due to a difference in the electric conductivity between that of the sample solution and that of the BGE) must be the main principle of (the small) focusing of the Ado and Cyd zones. Lower sample injection volumes or using water as sample matrix (Fig. 1b, further discussion is given in the supplementary data, Fig. S4) can be applied to improve their peak shapes (via borate sweeping). However, the determination of Ado and Cyd via Method 1 will not be considered further, as with real samples there is a coelution of urine matrix constituents with these two compounds (as will be shown later).

As shown in Fig. 2a, the peaks of Urd and 5MeUrd coelute. Therefore, the next trial was to improve the resolution by increasing the pH of the BGE to a value of 9.38 and simultaneously by reducing the concentration of tetraborate in the BGE (via increase in the retention factors). Using 20 mmol L−1 C14MImBr in 5 mmol L−1 tetraborate at a pH of 9.38, improves the resolution between the two compounds when applying a pressure injection of 50 mbar for 1 min (Fig. 2c). Moreover, the peak heights of all the studied analytes are higher in Fig. 2c than in Fig. 2a, which we ascribe to the negative influence of chloride ions (from HCl used for the pH adjustment of the BGE) entering the sample zone from the injection end during the sweeping step and decreasing kS of the studied analytes. Based on these investigations, 20 mmol L−1 C14MImBr in 5 mmol L−1 tetraborate, pH 9.38 was identified as optimum BGE composition for the determination of Urd, 5MeUrd, Guo, Xao, and Ino in urine. In addition, hydrodynamic injection using a pressure of 50 mbar for 1 min was employed in further studies to avoid the coelution of Urd and 5MeUrd.

Employing the final optimized enrichment conditions, the mechanism of focusing (using a nucleoside with a deprotonation site (e.g., Urd) as a representative example) can be summarized as follows: (1) after filling the capillary with the BGE (20 mmol L−1 C14MImBr in 5 mmol L−1 sodium tetraborate, pH 9.38), the sample (nucleosides are dissolved in 2.5 mmol L−1 sodium tetraborate, pH 10.45) is injected hydrodynamically by applying a pressure of 50 mbar for 1 min (Fig. 3a). A negative voltage is applied, whereas the complexed deprotonated nucleosides (BN–O)2− migrate with a high velocity towards the detection end (Fig. 3b1). Simultaneously, C14MImBr micelles sweep the analytes (from the detection end) within the sample zone causing their migration towards the injection end [(BN-O)2−Mx+] (Fig. 3b2). The completely swept analytes are enriched at the rear of sample zone/BGE compartment boundary (Fig. 3c). Once the swept analytes pass the BGE/sample zone boundary, their velocity is reduced (kBGE < kS) resulting in their additional focusing by RFGE (Fig. 3d). Finally, the swept analyte zones are separated via MEKC (Fig. 3e).

Fig. 3

Schematic representation of the suggested enrichment mechanism using Method 1. a Capillary filled with BGE followed by hydrodynamic injection of the sample plug. b1 Application of a negative voltage, where EOF is towards the anode. b2 Simultaneously, C14MImBr commences to sweep the complexed deprotonated nucleosides within the sample zone. c Sweeping is completed and analytes are focused at the rear boundary of the sample zone/BGE compartment. d Additional focusing of the swept analyte zone by RFGE. e Separation of the focused analyte zones. The length of the arrow represents the magnitude of the velocity. It should be taken into consideration that step b1 and b2 are taking place concomitantly, however, they are separated here for illustrative purpose

PSIE-Sweeping Using SDS Micelles (Method 2)

The low retention factor for Ado and Cyd with respect to C14MImBr and the low sweeping efficiency obtained for these two compounds when using C14MImBr in addition to the unsuitability of Method 1 for the determination of Ado and Cyd in urine samples (due to interfering matrix constituents) necessitate the development of an alternative strategy to enable their sensitive and selective analysis in urine samples. Recently, we have reported for selected nucleosides that the addition of a boronic acid (replacing borate as complexing agent) to the BGE can increase their partitioning coefficients regarding the distribution between the micelles formed by C14MImBr and the surrounding aqueous phase via hydrophobic interaction between the alkyl/aryl group of the boronic acid added and the hydrophobic core of the cationic micelles [47]. Consequently, the retention factors of these highly hydrophilic metabolites are increased with respect to those found with C14MImBr in borate buffer.

Another approach is based on the fact that Ado and Cyd are weak bases. Under acidic pH conditions they are existing as positively charged species, which constitutes an ideal basis for their sweeping by the anionic surfactant SDS using electrostatic (Coulombic) forces between the positively charged species and the negatively charged PSP.

Following this strategy, we started our investigations by employing an acidic BGE (with negligible electroosmotic flow (EOF)) for the separation of Ado and Cyd with a BGE consisting of 50 mmol L−1 SDS in 50 mmol L−1 phosphate buffer, pH 2.82. By application of a negative voltage and in the presence of SDS micelles (reversed elution mode), Ado migrates faster than Cyd, although the effective charge number of Ado is significantly lower than that of Cyd (see pKa values in the supplementary data). Additional hydrophobic interaction of Ado (Ado is more hydrophobic than Cyd, see log Pow, supplementary data) with SDS explains the faster migration of Ado than Cyd due to the stronger interaction of the former with SDS [i.e., k(Ado) > k(Cyd)]. Optimum concentration of SDS in the BGE with regard to both peak shape and analyte resolution was found to be 100 mmol L−1 (see Fig. 4A1 and A2 using two different sample injection volumes). Concentration of SDS > 100 mmol L−1 generates an excessive electric current as well as unfavorably increased analyte migration times.

Fig. 4

Electropherogram obtained with a sample containing Ado and Cyd (20.0 mg L−1 each) dissolved in water employing two different sample injection volumes using a 100 mmol L−1 SDS in 25 mmol L−1 phosphate buffer, pH 2.82 and b 100 mmol L−1 SDS in 25 mmol L−1 aspartic acid buffer, pH 3.22 as BGEs. Other CE conditions: hydrodynamic injection using pressure A1, B1 50 mbar for 0.5 min, A2, B2 50 mbar for 1 min, capillary 649(501) mm × 50 µm I.D, applied voltage − 20 kV, oven temperature 25 °C. The figure inset in B1 is the electropherogram which was obtained with the same conditions as that in B1 but with BGE as sample matrix. The figure inset in B2 is the electropherogram which was obtained with the same conditions as in B2 but after adjusting the pH of the BGE with HCl to pH 2.18. Peak designation: see Fig. 2

The concentration of the co-ion (competing ion) influences the retention factors of the charged species with respect to the oppositely charged PSP [16, 25, 27], and consequently, the achievable sweeping efficiency [25] to a large extent. However, also the type of the co-ion plays a role. El-Awady and Pyell [25] reported that the sweeping efficiency for some aromatic amines is higher when they used glutamic acid, pH 3.35 as buffering sample constituent than when they used phosphoric acid as sample matrix, pH 3.50. The authors have ascribed this result to the small difference in the degree of protonation of the analytes between the sample matrix and the BGE and accordingly slightly different ks and slightly different sweeping efficiency. Based on this report [25], we studied the impact of l-aspartic acid as buffering sample constituent on the obtained sweeping efficiency of Ado and Cyd.

Buffering a solution with a zwitterionic/isoelectric buffering compound produces a buffered electrolyte (1) with low electrical conductivity, (2) with satisfactory buffering capacity, and (3) without the concomitant introduction of co-ions [17] that would be detrimental to the sweeping efficiency. Amino acids with two acidic groups and one basic group (or one acidic group and two basic groups) and with an isoelectric point (pI) close to two of its pKa values (within ~ ± 1.5 pH units) [17, 48] can provide these advantages (e.g., aspartic acid, glutamic acid or lysine). The buffer is prepared simply by dissolving the amino acid in water and the pH is close to the pl of the amino acid [48].

l-aspartic acid is one of the non-essential amino acids that is normally synthesized in the body. It consists of two carboxylate groups with pKa1 and pKa2 of 1.95 and 3.71, and one amino group (pKa3 of 9.66). Its isoelectric point (pI) equals 2.77 [41]. Based on these properties, l-aspartic acid can be used as a low conductivity buffer constituent providing a good buffering capacity at acidic pH values. In subsequent studies 100 mmol L−1 SDS in 25 mmol L−1 aspartic acid buffer, pH 3.22 (without any pH adjustment) was employed as BGE for the separation of Ado and Cyd using water or the BGE as sample matrix (see electropherograms in Fig. 4B1 and B2). The figure inset in Fig. 2B1 demonstrates the separation of the two compounds under non-sweeping conditions. It is interesting to see that the peak heights, at two different sample injection volumes (Fig. 4B1 and B2), are significantly higher for both analytes in aspartic acid buffer than those obtained using phosphate buffer (Fig. 4A1 and A2). The peak height in aspartic acid buffer is higher than that in phosphate buffer, although the degree of protonation α of Ado and Cyd in aspartic acid buffer (pH 3.22) is lower [α(Ado) = 0.656 and α(Cyd) = 0.895] than that in phosphate buffer [here pH = 2.82, α(Ado) = 0.827, and α(Cyd) = 0.955]. The lower sweeping efficiency of Ado and Cyd in phosphate buffer results from the co-ions which migrate within the sample zone (Na+ and H+ ions) reducing ks and lowering the sweeping efficiency. Trials to increase the degree of protonation of Ado and Cyd in aspartic acid buffer via addition of HCl resulted in a dramatic decrease in the peak height due to the negative influence of the chloride ions on the retention factors and hence the sweeping efficiency (see figure inset in Fig. 4B2). Therefore, 100 mmol L−1 SDS in 25 mmol L−1 aspartic acid buffer, pH 3.22 was used as optimum BGE.

We have tested both water and 2.5 mmol L−1 aspartic acid as sample matrix. The peak heights of Ado and Cyd are slightly higher when using water as sample matrix than those obtained when using 2.5 mmol L−1 aspartic acid. This difference can be attributed to the initial zone focusing by dynamic pH junction (very fast migrating H+ boundary) when using water as sample matrix that is not existing when using 2.5 mmol L−1 aspartic acid, therefore the sample matrix water will be used for subsequent investigations. When using 2.5 mmol L−1 aspartic acid as sample matrix, Ado and Cyd are both positively charged within the sample zone, therefore, their focusing by dynamic pH junction is not possible and their sweeping by SDS micelles entering from the injection end (under negligible EOF conditions) commences immediately once a negative voltage is applied.

The theoretical description of the focusing process under the conditions of pure water as sample matrix is more complicated, because two moving reaction boundaries have to be taken into account. The first moving reaction boundary is the boundary of hydrogen ions that traverses the sample zone very fast because of the absence of buffering matrix components and the low electric conductivity in this segment. This boundary starts at the stationary boundary sample zone/BGE compartment and migrates versus the injection end. The second moving reaction boundary is the boundary of micelles that causes sweeping under reversal of the direction of velocity as soon as the two boundaries cross each other. This moving boundary starts at the injection end of the capillary and migrates versus the detector end. The second moving reaction boundary will have a velocity that is much slower than that of the hydrogen ions because of the different electrophoretic mobilities of the involved species. This difference in velocities (due to a difference in electrophoretic mobilities) is a key element in the understanding of the complete focusing process, as sweeping of the neutral analyte species is of very low efficiency, so that there is inevitably a loss of (neutral) analyte due to the EOF before the two moving reaction boundaries cross each other. This difference in velocities explains why there is only a very small loss of analyte within this focusing scheme.

We also have to consider that the bulk EOF velocity is the weighted average of the local EOF velocities. The bulk EOF velocity will be continuously decreasing until the moving boundary of hydrogen ions has reached the injection end of the sample zone and a large part of the initial sample zone has left the capillary at the injection end. In this moment, the pH in all segments of the capillary is so low that the resulting EOF velocity can be regarded to be negligible and the bulk EOF is quasi-eliminated. Part of the sweeping process takes place under nearly EOF-free conditions. Because of the influence of all segments of the capillary on the bulk EOF, the absolute velocity of the two moving boundaries will be permanently higher than that of the EOF during the complete process (whilst the length of the injected sample zone is lower than a critical length). After having defined the sample matrix composition, we varied the injected sample volume by increasing both the pressure and the time of sample injection. Sample injection volumes higher than those produced by pressure injection of 50 mbar for 1 min did not result in a further increase of the peak height. (Fig. S5, supplementary data).

Hydrodynamic injection using a pressure of 50 mbar for 1 min will be considered for further studies. In addition, a voltage of −20 kV was applied as optimum providing an acceptable running time, resolution and electric current. The effect of the oven temperature on the separation was investigated in the range of 15–35 °C. A capillary temperature of 25 °C was selected as optimum. In an attempt to study the transient change of the EOF during the focussing process, we recorded the electric current under the optimized injection conditions. Under these conditions, the electric current was about − 4.8 µA at the start, reaches − 30 µA after 24 s and remains at − 32.4 µA from 80 s to the end of the run. We conclude that after 24 s the front of hydrogen ions reaches the end of the capillary, while the initial sample zone has completely left the capillary after 80 s (very low EOF velocity).

The suggested focusing mechanism of Ado and Cyd using SDS can be described as follows: (1) At the beginning, the capillary is filled with the BGE (100 mmol L−1 SDS in 25 mmol L−1l-aspartic acid, pH 3.22), followed by hydrodynamic injection of the sample containing the nucleosides dissolved in water (Ado and Cyd are neutral in water) (Fig. 5a). (2) A negative voltage is applied, whereas the presence of a low conductivity sample matrix (of higher pH) induces a bulk EOF, which results in an initial loss of the neutral analytes from the injection end (Fig. 5b). (3) Due to the migration of hydrogen ions within the sample zone, Ado and Cyd are protonated and migrate with a high velocity towards the injection end that results in their focusing by dynamic pH junction (Fig. 5c). (4) When the front of the protonated nucleosides reaches the front of the migrating micelles, SDS starts to sweep the protonated nucleosides leading to the reversal of their migration direction [N+Mx−] (Fig. 5d). After completion of the focusing process (and removal of a large part of the initial sample zone as a result of the bulk EOF), the swept analyte zones are separated via MEKC (Fig. 5e).

Fig. 5

Schematic representation of the suggested focusing mechanism using Method 2. a Capillary filled with BGE followed by hydrodynamic injection of the sample plug. b Application of a negative voltage, whereas hydrogen ions migrate towards the injection end and SDS migrates towards the detection end. c Hydrogen ions protonate the nucleosides within the sample zone starting their focusing at the migrating reaction boundary, while the front of micelles migrates from the injection end of the sample zone versus the injection end at much lower velocity than the front of protons. d SDS commences sweeping of the positively charged nucleosides as soon as the two in opposite direction migrating concentration boundaries meet within the sample zone, while simultaneously the matrix is partially pumped out from the injection end until the front of protons reaches the injection side end of the capillary. e Separation of the swept analyte zones starts. The length of the arrow represents the magnitude of the velocity

Method Performance and Validation Study

After applying the calibration standards to the phenylboronate affinity gel (PBA), drying of the eluate and reconstitution in the appropriate solvent (as described under Section “Urine samples and extraction conditions”), the calibration plots were constructed using the full optimized parameters (see Section “Preparation of standard solutions and calibration curves”) for Method 1 or Method 2. The linear regression parameters for 5MeUrd, Urd, Guo, Ino, and Xao are obtained using Method 1, whereas the linear regression parameters for Ado and Cyd are obtained using Method 2. To determine the linearity of the calibration functions, seven to nine calibration standards of the studied nucleosides were taken to construct the calibration curves (i.e., seven to nine different concentrations of the studied compounds with six replicates for each concentration). Peak height, peak area and corrected peak area (peak area/migration time) were used as the response factors. The results of the statistical analysis of the data are summarized in Table 1, showing the linearity range of the developed methods for each analyte.

As stated by the ICH guidelines [19], the correlation coefficient r, the y-intercept a, the slope of the regression line b, and the sum of squared errors SSE are required for the evaluation of linearity. The previously mentioned parameters in addition to the linearity range, the standard deviation of the intercept and the slope, the standard deviation of the residuals Syx, the confidence interval of the slope and the intercept, the method standard deviation Sx0 and the relative standard deviation of the method Sr are listed in Table 1. The correlation coefficient r of the calibration curve is higher in all cases than 0.9967, which indicates excellent linearity of the developed methods. In addition to the determination of correlation coefficients, the linearity was assessed by performing Mandel’s fitting test [49]. At a significance level of P = 0.99, all Mandel’s test values are lower than the critical F values (Table 1), which indicates that the chosen linear regression model is adequate.

As given in Table 1, the use of the corrected peak area provides no significant improvement when compared to the uncorrected peak area (see r values in Table 1), which implies a high reproducibility of the migration times and a very good precision of the injection system. In addition, the linearity range in most cases is wider when using the peak area or the corrected peak area than when using the peak height for the construction of the calibration function. This is because the peak height is more affected by peak broadening due to electromigration dispersion (concentration overload) at higher sample concentration (leading to nonlinear calibration functions) than the peak area or the corrected peak area [1, 50, 51].

Limits of detection (LOD) and limits of quantitation (LOQ) were determined based on a signal to noise ratio (S/N) of 3 and 10, respectively. The S/N is calculated (according to the European Pharmacopeia [52]) as given in our previously published work [1]. As shown in Table 1, the LOD and LOQ for the five studied nucleosides using Method 1 were found to be 0.1–0.2 µg mL−1. The LOD for Ado and Cyd using Method 2 was found to be 0.1 and 0.2 µg mL−1, and the LOQ to be 0.2 and 0.4 µg mL−1, respectively. The LOD of Cyd can be further reduced if the selected wavelength is changed to 272 nm, which is for Cyd the wavelength of maximum absorbance coefficient.

The repeatability of migration times, peak heights and peak areas (intra-day variation) was determined for the developed optimized methods with 12 replicate injections of a sample containing (1) 6.0 mg L−1 of 5MeUrd, Guo, Urd, Ino, and Xao (Method 1) or (2) 12 mg L−1 Ado and Cyd, (Method 2). The inter-day variation was evaluated by analyzing the same sample over a period of 3 days. Precision was expressed as RSD (%).

As given in Table 2, the repeatability of the migration times of Method 1 is ≤ 0.49 and 1.91% for intra- and inter-day precision, respectively, whereas the repeatability of the migration times of Method 2 is ≤ 0.48 and 1.33% for intra- and inter-day precision, respectively. The highest values for RSD (%) of the peak height and peak area using Method 1 (intra-day precision) are 1.85 and 2.78%, respectively. With Method 2, the highest values for RSD (%) of the peak height and peak area (intra-day precision) are 3.78 and 4.62%, respectively. For inter-day precision, all RSD (%) values using Method 1 are lower than 3.54 and 3.24% for the peak height and the peak area, respectively, which indicates in total a good repeatability of Method 1. In contrast to this result, the RSD (%) values for the peak height and the peak area with Method 2 (inter-day precision) are ≤ 10.42 and 9.42%, respectively. These relatively high RSD (%) values for the peak height and the peak area can be attributed to the fluctuation of the EOF velocity under acidic pH conditions [53].

Table 2

Intra- and inter-day precision of the developed methods

 

Method

Intra-day precision (n = 12), RSDa (%)

Inter-day precision, 3 days (n = 37), RSD (%)

MTb

PHc

PAd

MT

PH

PA

5MeUrd

1

0.45

1.84

1.83

1.42

3.14

2.75

Urd

1

0.46

1.27

2.78

1.48

2.76

2.67

Guo

1

0.48

0.52

1.78

1.62

3.44

2.94

Ino

1

0.45

0.83

1.48

1.81

3.54

3.10

Xao

1

0.49

1.85

2.39

1.91

2.78

3.24

Ado

2

0.48

1.91

2.35

1.21

5.29

9.10

Cyd

2

0.46

3.78

4.62

1.33

10.42

5.81

aRelative standard deviation

bMigration time

cPeak height

dPeak area

Although the obtained LODs are higher than those reported in [1], the developed methods provide a similar reproducibility and they are more simple and faster than the method presented in [1] as no polarity switching is needed.

Extraction Using a PBA Column and Application to Human Urine Samples

The extraction procedure using a PBA column was used only for the clean-up of the urine sample and not for off-line sample concentration as the freeze-dried eluate was reconstituted in a volume which corresponds to the starting urine sample volume (see Section “Urine samples and extraction conditions”). However, the extraction process has a positive influence on the focusing efficiency obtainable with Method 2. To clarify the last point, we have compared the electropherograms obtained with a standard solution containing Ado and Cyd dissolved in water without carrying out SPE with PBA (injected in CE directly, Fig. 6a) and the same standard solution but with carrying out the SPE step (Fig. 6b). This comparison was made for two different analyte concentrations. The peak heights and areas for Ado and Cyd are higher for those runs in which the nucleosides were first extracted on the PBA column although there is no off-line focusing. The ionic constituents which remain in the eluate after extraction (either the ions present in urine or brought in by the washing solutions) reduce the initial analyte loss at the beginning of the negative voltage application. This is due to a reduction of the averaged bulk EOF velocity caused by the increase of the ionic strength within the sample compartment. This result is in good agreement with what we have recently reported [1]. In agreement with our previous results, we expect that the observed effect does not impede the quantitative analysis. We have shown that the developed focusing technique tolerates a variation in the electric current at the start of the focusing process in the range of − 1.4 to − 5.1 µA (related to a varied electric conductivity of the injected pre-treated sample) without having an impact on the quantitative result [1]. Hence, correct quantitative analysis is possible with suitable matrix-adapted calibration standards.

Fig. 6

Electropherograms obtained using Method 2 for a nucleosides dissolved in water and injected in CE directly or b nucleosides extracted with a PBA column and reconstituted in the same volume of water. Sample solution is 4.0 mg L−1 of Ado and 8.0 mg L−1 Cyd in A1, B1 or 8.0 mg L−1 of Ado and 16.1 mg L−1 Cyd in A2, B2. CE conditions: 100 mmol L−1 SDS in 25 mmol L−1 aspartic acid buffer, pH 3.22 as BGE, hydrodynamic injection using pressure 50 mbar for 1 min, applied voltage − 20 kV, current − 30 µA, oven temperature 25 °C. Peak designation: see Fig. 2

Figures 7 and 8 show the application of the developed methods to the analysis of urinary nucleosides in blank or spiked urine samples. In each case, a standard solution containing the nucleosides dissolved in water was also analyzed for comparison. The injected sample is the sample obtained after pre-treatment of the standard solution or the blank/spiked urine sample with a PBA column, freeze drying of the eluate and reconstitution in 2 mL in an aqueous solution of optimized matrix composition (See Section “Urine samples and extraction conditions”). Figures 7a and 8a show the application of Methods 1 and 2, respectively, to blank urine samples. The presence of nucleosides as endogenous metabolites in the urine samples can be confirmed by (1) comparing the electropherograms in Figs. 7a and 8a by those obtained using standard solutions as sample (Figs. 7b, 8b) or (2) by observing an appropriate increase in peak height or in peak area (e.g., for Ino and Xao in Fig. 7c and Ado in Fig. 8c) when comparing the result obtained for the spiked sample with the result obtained for the unspiked sample (cf., corresponding peaks in Figs. 7b, 8b). It is clear from Figs. 7 and 8 that there is an excellent extraction selectivity obtained with PBA as there are only few peaks that have to be assigned to unknown compounds present in the urine matrix. In addition, these unknown constituents are in general well separated from the studied nucleosides. There is one important exception: As demonstrated in Fig. 7a, c, with Method 1 eluted matrix constituents coelute with Ado and Cyd. This method is therefore unsuitable for the determination of Ado and Cyd in human urine. Therefore, Method 1 has to be complemented with Method 2. For clinical studies, the determined nucleoside concentrations have to be normalized on the creatinine content. Based on a previous report [15], we assume that the peak marked with (?) can be ascribed to pseudouridine, which is the most predominant modified nucleoside in tRNA.

Fig. 7

Electropherograms obtained using Method 1 for a a blank urine sample, b standard solution of nucleosides in water, and c a spiked urine sample after extraction on PBA column and reconstitution in 2 mL 2.5 mmol L−1 sodium tetraborate pH 10.45. Nucleoside concentration in b and c is 4 mg L−1 each. CE conditions: 20 mmol L−1 C14MImBr in 5 mmol L−1 disodium tetraborate, pH 9.38 as BGE, capillary 649(501) mm × 50 µm I.D, hydrodynamic injection using pressure 50 mbar for 1 min, applied voltage − 20 kV, current − 9.9 µA, oven temperature 35 °C. Peak designation: see Fig. 2. *Unidentified peak, ? = pseudouridine

Fig. 8

Electropherograms obtained using Method 2 for a a blank urine sample, b standard solution of nucleosides in water, and c a spiked urine sample after extraction on PBA column and reconstitution in 2 mL water. Nucleoside concentration in b and c is 8.0 mg L−1 Ado and 16.1 mg L−1 Cyd. CE conditions and peak designation: see Fig. 6

Conclusions

Two simple, fast, reproducible and sensitive MEKC methods have been developed and validated for the analysis of urinary nucleosides. Combined with RFGE-sweeping or dynamic pH junction-sweeping as online enrichment techniques, the developed methods can be successfully applied to the analysis of nucleosides in real urine samples using either C14MImBr or SDS as PSP. For charged analytes separated with a charged PSP (acting as a pseudostationary ion-exchanger), adjustment of the retention factors both within the sample zone and within the BGE compartment, is the prerequisite for obtaining an adequate focusing efficiency, especially if low limits of detections are needed. This adjustment of the retention factors can be realized by adjustment of the pH and the ionic strength within both the sample zone and the BGE compartment (while simultaneously providing an adequate resolution during the separation step). In this regard, isoelectric buffers due to zwitterionic buffer constituents (providing a very low co-ion concentration) are particularly favorable either in the optimization of the composition of the sample solution or in the optimization of the BGE composition.

Notes

Acknowledgements

We thank the workshops of the Department of Chemistry, University of Marburg for the development of the data recording unit.

Funding

A. H. Rageh thanks the Egyptian Ministry of Higher Education and the Ministry of State for Scientific Research and the Deutscher Akademischer Austauschdienst (DAAD) for funding her PhD scholarship through German Egyptian Research Long-Term Scholarship program (GERLS).

Compliance with ethical standards

Conflict of interest

A. H. Rageh declares that she has no conflict of interest. U. Pyell declares that she has no conflict of interest.

Ethical approval

All procedures performed in  this study involving a human participant were in accordance with the 1964 Helsinki Declaration and its later amendments. For experiments with human urine, informed consent was obtained from the volunteer. This article does not contain any studies with animals performed by any of the authors.

Supplementary material

10337_2018_3570_MOESM1_ESM.pdf (140 kb)
Supplementary material 1 (PDF 139 KB)

References

  1. 1.
    Rageh AH, Kaltz A, Pyell U (2014) Determination of urinary nucleosides via borate complexation capillary electrophoresis combined with dynamic pH junction-sweeping-large volume sample stacking as three sequential steps for their online enrichment. Anal Bioanal Chem 406:5877–5895CrossRefGoogle Scholar
  2. 2.
    Rodriguez-Gonzalo E, Hernandez-Prieto R, Garcia-Gomez D, Carabias-Martinez R (2014) Development of a procedure for the isolation and enrichment of modified nucleosides and nucleobases from urine prior to their determination by capillary electrophoresis-mass spectrometry. J Pharm Biomed Anal 88:489–496CrossRefGoogle Scholar
  3. 3.
    Rodriguez-Gonzalo E, Garcia-Gomez D, Carabias-Martinez R (2011) Development and validation of a hydrophilic interaction chromatography-tandem mass spectrometry method with online polar extraction for the analysis of urinary nucleosides. Potential application in clinical diagnosis. J Chromatogr A 1218:9055–9063CrossRefGoogle Scholar
  4. 4.
    Cho SH, Choi MH, Lee WY, Chung BC (2009) Evaluation of urinary nucleosides in breast cancer patients before and after tumor removal. Clin Biochem 42:540–543CrossRefGoogle Scholar
  5. 5.
    Wang S, Zhao X, Mao Y, Cheng Y (2007) Novel approach for developing urinary nucleosides profile by capillary electrophoresis-mass spectrometry. J Chromatogr A 1147:254–260CrossRefGoogle Scholar
  6. 6.
    Helboe T, Hansen SH (1999) Separation of nucleosides using capillary electrochromatography. J Chromatogr A 836:315–324CrossRefGoogle Scholar
  7. 7.
    Sasco AJ, Rey F, Reynaud C, Bobin JY, Clavel M, Niveleau A (1996) Breast cancer prognostic significance of some modified urinary nucleosides. Cancer Lett 108:157–162CrossRefGoogle Scholar
  8. 8.
    Cohen AS, Terabe S, Smith JA, Karger BL (1987) High-performance capillary electrophoretic separation of bases, nucleosides, and oligonucleotides: retention manipulation via micellar solutions and metal additives. Anal Chem 59:1021–1027CrossRefGoogle Scholar
  9. 9.
    Liebich HM, Xu G, Di S, Lehmann R, Häring HU, Lu P, Zhang Y (1997) Analysis of normal and modified nucleosides in urine by capillary electrophoresis. Chromatographia 45:396–401CrossRefGoogle Scholar
  10. 10.
    Liebich HM, Lehmann R, Xu G, Wahl HG, Häring HU (2000) Application of capillary electrophoresis in clinical chemistry: the clinical value of urinary modified nucleosides. J Chromatogr B Biomed Sci Appl 745:189–196CrossRefGoogle Scholar
  11. 11.
    Xu G, Liebich HM, Lehmann R, Müller-Hagedorn S (2001) Capillary electrophoresis of urinary normal and modified nucleosides of cancer patients. Methods Mol Biol 162:459–474Google Scholar
  12. 12.
    Zheng YF, Xu GW, Liu DY, Xiong JH, Zhang PD, Zhang C, Yang Q, Shen L (2002) Study of urinary nucleosides as biological marker in cancer patients analyzed by micellar electrokinetic capillary chromatography. Electrophoresis 23:4104–4109CrossRefGoogle Scholar
  13. 13.
    Liebich HM, Müller-Hagedorn S, Klaus F, Meziane K, Kim KR, Frickenschmidt A, Kammerer B (2005) Chromatographic, capillary electrophoretic and matrix-assisted laser desorption ionization time-of-flight mass spectrometry analysis of urinary modified nucleosides as tumor markers. J Chromatogr A 1071:271–275CrossRefGoogle Scholar
  14. 14.
    Zheng YF, Kong HW, Xiong JH, Shen L, Xu GW (2005) Clinical significance and prognostic value of urinary nucleosides in breast cancer patients. Clin Biochem 38:24–30CrossRefGoogle Scholar
  15. 15.
    Szymanska E, Markuszewski MJ, Bodzioch K, Kaliszan R (2007) Development and validation of urinary nucleosides and creatinine assay by capillary electrophoresis with solid phase extraction. J Pharm Biomed Anal 44:1118–1126CrossRefGoogle Scholar
  16. 16.
    Rageh AH, Pyell U (2013) Imidazolium-based ionic liquid-type surfactant as pseudostationary phase in micellar electrokinetic chromatography of highly hydrophilic urinary nucleosides. J Chromatogr A 1316:135–146CrossRefGoogle Scholar
  17. 17.
    Rodemann T, Johns C, Yang WS, Haddad PR, Macka M (2005) Isoelectric buffers for capillary electrophoresis. 2. Bismorpholine derivative of a carboxylic acid as a low molecular weight isoelectric buffer. Anal Chem 77:120–125CrossRefGoogle Scholar
  18. 18.
    El-Awady M, Belal F, Pyell U (2013) Robust analysis of the hydrophobic basic analytes loratadine and desloratadine in pharmaceutical preparations and biological fluids by sweeping—cyclodextrin-modified micellar electrokinetic chromatography. J Chromatogr A 1309:64–75CrossRefGoogle Scholar
  19. 19.
    ICH Harmonised Tripartite Guidelines, Validation of analytical procedures: text and methodology Q2(R1) (1996) http://www.ich.org/products/guidelines/quality/article/quality-guidelines.html. Accessed 24 Apr 2018
  20. 20.
    Monton MR, Quirino JP, Otsuka K, Terabe S (2001) Separation and online preconcentration by sweeping of charged analytes in electrokinetic chromatography with nonionic micelles. J Chromatogr A 939:99–108CrossRefGoogle Scholar
  21. 21.
    Quirino JP, Terabe S (1998) Exceeding 5000-fold concentration of dilute analytes in micellar electrokinetic chromatography. Science 282:465–468CrossRefGoogle Scholar
  22. 22.
    Quirino JP, Terabe S, Bocek P (2000) Sweeping of neutral analytes in electrokinetic chromatography with high-salt-containing matrixes. Anal Chem 72:1934–1940CrossRefGoogle Scholar
  23. 23.
    El-Awady M, Huhn C, Pyell U (2012) Processes involved in sweeping under inhomogeneous electric field conditions as sample enrichment procedure in micellar electrokinetic chromatography. J Chromatogr A 1264:124–136CrossRefGoogle Scholar
  24. 24.
    Pyell U, Rageh AH, El-Awady M (2017) The concept of stationary and moving boundaries modelled as accelerating or decelerating planes in the understanding of sweeping processes employed for online focusing in capillary zone electrophoresis and electrokinetic chromatography. Chromatographia 80:359–382CrossRefGoogle Scholar
  25. 25.
    El-Awady M, Pyell U (2013) Sweeping as a multistep enrichment process in micellar electrokinetic chromatography: the retention factor gradient effect. J Chromatogr A 1297:213–225CrossRefGoogle Scholar
  26. 26.
    Yang X, Dai J, Carr PW (2003) Analysis and critical comparison of the reversed-phase and ion-exchange contributions to retention on polybutadiene coated zirconia and octadecyl silane bonded silica phases. J Chromatogr A 996:13–31CrossRefGoogle Scholar
  27. 27.
    Orentaitė I, Maruška A, Pyell U (2011) Regulation of the retention factor for weak acids in micellar electrokinetic chromatography with cationic surfactant via variation of the chloride concentration. Electrophoresis 32:604–613CrossRefGoogle Scholar
  28. 28.
    Kazarian AA, Hilder EF, Breadmore MC (2011) Online sample pre-concentration via dynamic pH junction in capillary and microchip electrophoresis. J Sep Sci 34:2800–2821CrossRefGoogle Scholar
  29. 29.
    Britz-McKibbin P, Chen DDY (2000) Selective focusing of catecholamines and weakly acidic compounds by capillary electrophoresis using a dynamic pH junction. Anal Chem 72:1242–1252CrossRefGoogle Scholar
  30. 30.
    El-Awady M, Pyell U (2014) Processes involved in sweeping as sample enrichment method in cyclodextrin-modified micellar electrokinetic chromatography of hydrophobic basic analytes. Electrophoresis 35:605–616CrossRefGoogle Scholar
  31. 31.
    Britz-McKibbin P, Otsuka K, Terabe S (2002) Online focusing of flavin derivatives using dynamic pH junction-sweeping capillary electrophoresis with laser-induced fluorescence detection. Anal Chem 74:3736–3743CrossRefGoogle Scholar
  32. 32.
    Britz-McKibbin P, Terabe S (2002) High sensitivity analyses of metabolites in biological samples by capillary electrophoresis using dynamic pH junction-sweeping. Chem Rec 2:397–404CrossRefGoogle Scholar
  33. 33.
    Britz-McKibbin P, Markuszewski MJ, Iyanagi T, Matsuda K, Nishioka T, Terabe S (2003) Picomolar analysis of flavins in biological samples by dynamic pH junction-sweeping capillary electrophoresis with laser-induced fluorescence detection. Anal Biochem 313:89–96CrossRefGoogle Scholar
  34. 34.
    Britz-McKibbin P, Ichihashi T, Tsubota K, Chen DDY, Terabe S (2003) Complementary online preconcentration strategies for steroids by capillary electrophoresis. J Chromatogr A 1013:65–76CrossRefGoogle Scholar
  35. 35.
    Su AK, Chang YS, Lin CH (2004) Analysis of riboflavin in beer by capillary electrophoresis/blue light emitting diode (LED)-induced fluorescence detection combined with a dynamic pH junction technique. Talanta 64:970–974CrossRefGoogle Scholar
  36. 36.
    Yu L, Li SFY (2005) Dynamic pH junction-sweeping capillary electrophoresis for online preconcentration of toxic pyrrolizidine alkaloids in Chinese herbal medicine. Electrophoresis 26:4360–4367CrossRefGoogle Scholar
  37. 37.
    Chen Y, Zhang L, Cai Z, Chen G (2011) Dynamic pH junction-sweeping for online focusing of dipeptides in capillary electrophoresis with laser-induced fluorescence detection. Analyst (Cambridge, UK) 136:1852–1858CrossRefGoogle Scholar
  38. 38.
    Štědrý M, Jaroš M, Včeláková K, Gaš B (2003) Eigenmobilities in background electrolytes for capillary zone electrophoresis: II. Eigenpeaks in univalent weak electrolytes. Electrophoresis 24:536–547CrossRefGoogle Scholar
  39. 39.
    Gaš B, Kenndler E (2004) System zones in capillary zone electrophoresis. Electrophoresis 25:3901–3912CrossRefGoogle Scholar
  40. 40.
    Huhn C, Pyell U (2010) Diffusion as major source of band broadening in field-amplified sample stacking under negligible electroosmotic flow velocity conditions. J Chromatogr A 1217:4476–4486CrossRefGoogle Scholar
  41. 41.
    Lundblad RL, MacDonald F (2010) Handbook of biochemistry and molecular biology, 4th edn. CRC Press, Taylor and Francis Group, LLC, Boca RatonGoogle Scholar
  42. 42.
    Marrubini G, Mendoza BEC, Massolini G (2010) Separation of purine and pyrimidine bases and nucleosides by hydrophilic interaction chromatography. J Sep Sci 33:803–816CrossRefGoogle Scholar
  43. 43.
    IUPAC-IUB Commission on Biochemical Nomenclature (CBN) (1974) Abbreviations and symbols of nucleic acids, polynucleotides, and their constituents. Pure Appl Chem 40:277–290CrossRefGoogle Scholar
  44. 44.
    Kuo KC, Phan DT, Williams N, Gehrke CW (1990) In: Gehrke CW, Kuo KC (eds) Chromatography and modification of nucleosides. Part C modified nucleosides in cancer and normal metabolism methods and applications. Elsevier, OxfordGoogle Scholar
  45. 45.
    Gehrke CW, Kuo KC, Davis GE, Suits RD, Waalkes TP, Borek E (1978) Quantitative high-performance liquid chromatography of nucleosides in biological materials. J Chromatogr 150:455–476CrossRefGoogle Scholar
  46. 46.
    van de Merbel NC (2008) Quantitative determination of endogenous compounds in biological samples using chromatographic techniques. Trends Anal Chem 27:924–933CrossRefGoogle Scholar
  47. 47.
    Rageh AH, Pyell U (2015) Boronate affinity-assisted MEKC separation of highly hydrophilic urinary nucleosides using imidazolium-based ionic liquid type surfactant as pseudostationary phase. Electrophoresis 36:784–795CrossRefGoogle Scholar
  48. 48.
    Hjertén S, Valtcheva L, Elenbring K, Liao JL (1995) Fast, high-resolution (capillary) electrophoresis in buffers designed for high field strengths. Electrophoresis 16:584–594CrossRefGoogle Scholar
  49. 49.
    Reichenbächer M, Einax JW (2011) Challenges in analytical quality assurance. Springer, Berlin HeidelbergCrossRefGoogle Scholar
  50. 50.
    Wätzig H (1995) Appropriate calibration functions for capillary electrophoresis I. Precision and sensitivity using peak areas and heights. J Chromatogr A 700:1–7CrossRefGoogle Scholar
  51. 51.
    Wätzig H, Degenhardt M, Kunkel A (1998) Strategies for capillary electrophoresis. Method development and validation for pharmaceutical and biological applications. Electrophoresis 19:2695–2752CrossRefGoogle Scholar
  52. 52.
    European Pharmacopoeia (7.8) (2013) European directorate for the quality of medicines & healthcare (EDQM), 7th edn, online version. StrasbourgGoogle Scholar
  53. 53.
    Quirino JP, Terabe S (1998) Online concentration of neutral analytes for micellar electrokinetic chromatography. 3. Stacking with reverse migrating micelles. Anal Chem 70:149–157CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of ChemistryUniversity of MarburgMarburgGermany
  2. 2.Department of Pharmaceutical Analytical Chemistry, Faculty of PharmacyUniversity of AssiutAssiutEgypt

Personalised recommendations