Matrix influence on derivatization and ionization processes during selenoamino acid liquid chromatography electrospray ionization mass spectrometric analysis

Page 1

Msm

RI

a

ARRAA

KADML

1

btila(tic

uitao

h1

Journal of Chromatography B, 955–956 (2014) 34–41

Contents lists available at ScienceDirect

Journal of Chromatography B

jou rn al hom epage: www.elsev ier .com/ locate /chromb

atrix influence on derivatization and ionization processes duringelenoamino acid liquid chromatography electrospray ionizationass spectrometric analysis

iin Rebane ∗, Maarja-Liisa Oldekop, Koit Herodesnstitute of Chemistry, University of Tartu, Ravila 14a, 50411 Tartu, Estonia

r t i c l e i n f o

rticle history:eceived 5 October 2013eceived in revised form 11 January 2014ccepted 8 February 2014vailable online 18 February 2014

eywords:mino acidserivatizationatrix effects

a b s t r a c t

Considering the importance of derivatization in LC/ESI/MS analysis, the objective of this work was todevelop a method for evaluation of matrix effect that would discriminate between matrix effect due tothe derivatization reaction yield and from the ESI. Four derivatization reagents (TAHS, DEEMM, DNS,FMOC-Cl) were studied with respect to matrix effects using two selenoamino acids and onion matrix asmodel system. A novel method for assessing matrix effects of LC/ESI/MS analyses involving derivatizationis proposed, named herein post-derivatization spiking, that allows evaluating effect of matrix on ESIionization without derivatization reaction yield contribution. The proposed post-derivatization spikingmethod allowed to demonstrate that the reason of reduced analytical signal can be signal suppression inESI (as in case of DNS derivatives with matrix effects 38–99%), alteration of derivatization reaction yield

C/ESI/MS (TAHS, matrix effects 92–113%, but reaction yields 20–50%) or both (FMOC-Cl, matrix effects 28–88%and reaction yields 50–70%). In case of DEEMM derivatives, matrix reduces reaction yield but enhancesESI/MS signal.

A method for matrix effect evaluation was developed. It was also confirmed that matrix effects can bereduced by dilution.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

Amino acids are analyzed for various reasons in all types ofiological samples ranging from human bodily fluids and tissueso various foods. For example, in clinical biochemistry, changesn amino acid concentrations in human serum can be corre-ated to certain diseases [1]. Similarly, selenoamino acids suchs Se-methylselenocysteine (Se-MeSeCys) and selenomethionineSeMet) are under interest in clinical studies [2] (Fig. 1) and sincehey have anticarcinogenic properties, these are determined in var-ous foods, such as garlic [3,4]. In all mentioned cases, very lowoncentrations of amino acids are targeted in complex matrices.

Due to the constant pursuit for more sensitive analysis, liq-id chromatography mass spectrometry (LC/MS) with electrospray

onization (ESI) has become one of the most popular analysis

echniques for many analytes, including selenoamino acids [5]. Inddition to low limits of detection, it enables to confirm the identityf analytes and identify unknown compounds [6].

∗ Corresponding author.E-mail address: [email protected] (R. Rebane).

ttp://dx.doi.org/10.1016/j.jchromb.2014.02.016570-0232/© 2014 Elsevier B.V. All rights reserved.

When analyzing amino acids, traditionally, derivatizationhas been applied in order to enhance the sensitivity ofultraviolet–visible detection (UV–Vis) and also to improve thechromatographic separation [7]. Employing derivatization forLC/ESI/MS is a newer approach, but the aims are similar: to increasedetection sensitivity and selectivity by means of MS/MS technique,improve chromatographic retention or peak shape, eliminate car-ryover, facilitate sample cleanup, and to form a stable derivativefor unstable analytes [8].

For the LC/ESI/MS analysis of amino acids, “classical” deriva-tization reagents are often used, for example dansyl chloride(DNS) [9], 9-fluorenylmethyl chloroformate (FMOC-Cl), and diethylethoxymethylenemalonate (DEEMM) [10], which have beendesigned for UV absorbance or/and fluorescence detectors. Inrecent years, there has been a rapid growth in designingand developing amino acid derivatization reagents that arespecially meant for LC/ESI/MS applications for lowering detec-tion limits, e.g. 3-aminopyridyl-N-hydroxysuccinimidyl carbamate

(APDS) [11], N-hydroxysuccinimide ester of N-alkylnicotinicacid (Cn-NA-NHS) [12] and p-N,N,N-trimethylammonioanilylN′-hydroxysuccinimidyl carbamate iodide (TAHS) [13]. How-ever, these reagents are not commercially available and in

Page 2

R. Rebane et al. / J. Chromatogr.

tt

aFieaiommpretna

Eapccar

M

scs

ctmae

ipdsaat

Fig. 1. Structures of Se-methylselenocysteine and selenomethionine.

horoughly optimized conditions, commercially available deriva-ization reagents can provide similar sensitivity [10].

When combination of derivatization and LC/ESI/MS analysis ispplied on real samples, multiple aspects are to be considered.irstly, in the ESI source, the efficiency of generating gas phaseons from a compound in solution depends mainly on the prop-rties of the compound [14]. Importantly, compounds other thannalyte in the ESI source can have a considerable effect on theonization of the analyte. If the compounds causing suppressionr enhancement of an analyte signal originate from the sampleatrix, the effect is called the matrix effect [15,16]. In general,atrix effects may be caused by the compounds of current or

revious injections, either as late-eluting peaks (bands) or as impu-ities depositing on the internal surfaces of ESI source [17]. Kingt al. analyzed biological samples and concluded that the ioniza-ion suppression is most likely resulted by the high concentration ofonvolatile compounds present in the spray concurrently with thenalyte [18].

The matrix effect (ME%) can be quantitatively expressed byq. (1), where Amatrix and Astandard are peak areas of the equalmount of analyte, respectively, in presence and in absence ofossibly interfering compounds. The ME% value of 100% indi-ates that the ionization of an analyte is not affected by otherompounds present in the ESI source. ME% values below andbove 100% indicate ionization suppression and enhancement,espectively [19].

E% = Amatrix

Astandard× 100% (1)

This technique is also called post-extraction spiking: analyticalignal of blank sample extract spiked with the analyte (Amatrix) isompared with the signal of the equal amount of analyte in pureolvent (Astandard) [17].

Another technique for evaluation of matrix effects is post-olumn infusion, a method where using a syringe pump and aee-piece, a continuous flow of an analyte is mixed with the chro-

atographic effluent of sample blank analysis [17]. However, thispproach does not give quantitative information about the matrixffect [20].

For certain types of analytes, such as amino acids, a blank samples sometimes not available. For example, in case of honey or bloodlasma analysis, amino acid free sample does not exist and it is quiteifficult to get an estimation of matrix effect. Spiking technique can

ometimes help, but there are problems related to this approachlso [19]. On the other hand, for selenoamino acids, blank matricesre available, since selenium concentration in plants depends onhe soil it is grown in. Therefore, onions grown in regions such as

B 955–956 (2014) 34–41 35

Middle and North Europe do not contain selenoamino acids [21].These samples can be used for matrix effect estimation of methodsanalyzing selenoamino acids in onion samples.

With methods that contain derivatization step, the matrix effectis often studied as in case of a regular analytical method. For exam-ple, for �-alanine propyl chloroformate derivative, matrix effectshave been evaluated with continuously infusing purified deriva-tive with LC-effluent through tee-piece after analytical column. Formatrix effect assessment, signal of the �-alanine propyl chlorofor-mate derivative was monitored [22]. This approach is somewhatlimited since in order to get information for all amino acids, deriva-tives purified from derivatization reaction byproducts and buffercomponents must be obtained, and even then quantitative infor-mation about the matrix effects is not obtained.

Derivatization step adds complexity to the quantitative assess-ment of matrix effects since derivatization mixture containsadditional components such as nonvolatile borate buffer [1], whichhas been shown to cause signal suppression in the ESI source [23].Moreover, previously described approaches of the matrix effectassessments do not take into account the reaction yield of thederivatization (it is assumed to be the same for the standard solu-tions and the matrix). With derivatization reactions, there is alwaysa possibility that part of the analytes remain unreacted. This is wellillustrated by the fact that there are many publications dedicatedto the optimization of derivatization reactions [24–26]. Therefore,while investigating matrix effects in the ESI source, there is a riskthat poor derivatization reaction yield may be regarded as matrixeffect.

Considering the importance of derivatization in LC/ESI/MS anal-ysis, evaluation of its matrix effects deserves in-depth analysis.The objective of this work is to develop a method for the evalua-tion of matrix effect in complex matrices. The method assesses thetwo matrix effect components that are 1) due to the derivatizationreaction yield and 2) from the ESI-ionization.

2. Materials and methods

2.1. Chemicals

HPLC-grade methanol and acetonitrile were obtained fromSigma-Aldrich. Derivatization reagents diethyl ethoxymethylen-emalonate (DEEMM) and dansyl chloride (DNS) were purchasedfrom Fluka and 9-fluorenylmethyl chloroformate (FMOC-Cl) waspurchased from Aldrich. Se-MeSeCys was kindly donated by LGCLimited (United Kingdom) and SeMet was purchased from Sigma.Other chemicals: sodium hydroxide (Chemapol); acetic acid (Lach-Ner); sodium dihydrogensulfate (Merck); hydrochloric acid, boricacid, and ammonium hydroxide were from Reakhim; formic acidand ammonium acetate from Fluka. All reagents were of analyticalgrade unless otherwise stated.

All aqueous solutions were prepared with ultrapure water puri-fied by Millipore Milli-Q Advantage A10 (Millipore).

Onion samples were obtained from the local market and thestated country of origin was the Netherlands.

2.2. Preparation of standard solutions

Se-MeSeCys stock solution (0.55 mg g−1) was prepared by dis-solving the amino acid in 0.1 M hydrochloric acid with 30% MeOH.Stock solution of SeMet (0.4 mg g−1) was prepared in 0.5% 2-mercaptoethanol aqueous solution. Stock solutions were prepared

once and stored at −20 ◦C. All dilutions (0.5–12 �g g−1) were madewith 0.004% aqueous solution of 2-mercaptoethanol to avoid oxi-dation of SeMet [27]. Working standard solutions were prepareddaily.

Page 3

36 R. Rebane et al. / J. Chromatogr. B 955–956 (2014) 34–41

DEEM

2

hew2

cet4t1tflwu

2

r

padd

Fig. 2. Derivatization reactions for a)

.3. Sample preparation

The method previously applied for determining amino acids inoney [28] was adapted for onion. Approximately 2 g of homog-nized (1 min, 5000 rpm, Retch Grindomix GM200) onion waseighed and suspended in 25 mL of phosphate buffer (0.03 M, pH

.12). The solution was filtered through a wide-pore paper filter.For the extraction of amino acids, solid phase extraction (SPE)

artridges with styrene–divinylbenzene polymeric strong cationxchange sorbent, 500 mg (Alltech, USA) were used. SPE car-ridge was conditioned with 10 mL of HCl (0.1 M) at flow rate

mL min−1. The buffered onion sample was applied to the car-ridge at ∼1.5 mL min−1 flow rate. The analytes were eluted with5 mL of 2.5 M ammonium hydroxide containing 10% of ace-onitrile. The eluate was evaporated to dryness using nitrogenow and redissolved in 1 mL of 0.004% 2-mercaptoethanol inater. For experiments discussed in this article, this extract issed.

.4. Derivatization procedure

Derivatization reactions with amino acids for all derivatizationeagents are presented in Fig. 2.

DEEMM derivatization: to 1 mL of sample solution (either sam-

le extract or standard solution) 30 �L of DEEMM, 1.5 mL methanol,nd 3.5 mL of 0.75 M sodium borate buffer (pH 9.0) was added. Theerivatized mixture was kept at room temperature protected fromirect light for 24 h. LC/MS analysis has to be carried out at least

M; b) FMOC-Cl; c) DNS and d) TAHS.

24 h but not more than 48 h after the derivatization [28]. Samplesolutions were filtered through 0.45 �m cellulose acetate syringefilter (Whatman). The molar ratio of DEEMM to selenoamino acidsin most concentrated solution was 3770.

FMOC-Cl derivatization: to 300 �L of sample solution 300 �Lof 0.75 M sodium borate buffer (pH 9.0) and 300 �L of FMOC-Cl(10 mg mL−1 in acetonitrile) was added and vigorously mixed. Themixture was kept at room temperature for 2 h and then 300 �L ofHis solution (8 mg g−1 in water) was added to bind excess FMOC-Cl and vigorously mixed again [29]. Sample solutions were filteredthrough 0.45 �m regenerated cellulose syringe filter (Agilent). Themolar ratio of FMOC-Cl to selenoamino acids in most concentratedsolution was 1040.

DNS derivatization: to 100 �L of sample solution, 20 �L of 2 MNaOH and 500 �L of DNS solution (10 mg mL−1 in acetone) wereadded. Reaction mixture was kept at 4–6 ◦C in the dark for 45 min.Reaction was stopped with 10 �L of 25% NH4OH [30]. Sample solu-tions were filtered through 0.45 �m regenerated cellulose syringefilter (Agilent). The molar ratio of DNS to selenoamino acids in mostconcentrated solution was 4970.

TAHS derivatization: with minor modifications from Ref. [13]. To10 �L of sample solution 30 �L of 0.2 M sodium borate buffer (pH9.0) and 20 �L of TAHS solution (approximately 20 mg mL−1 in ace-tonitrile) was added. Reaction was carried out at room temperature

and stopped after 30 min with 200 �L of 0.2% acetic acid in water.Heating was not necessary since Tyrosine is not analyzed [13]. Themolar ratio of TAHS to selenoamino acids in most concentratedsolution was 3670.

Page 4

R. Rebane et al. / J. Chromatogr. B 955–956 (2014) 34–41 37

ke me

2

Tdsp(tCu

w(tF5A

2

sanDuFw

135

(pgrtd

2

o

Fig. 3. Schemes for matrix effect evaluations using 1) post-extraction spi

.5. Instrumentation

HPLC system Agilent Series 1100 LC/MSD Trap XCT (Agilentechnologies, Santa-Clara, USA) was equipped with an in-lineegasser, a binary pump, an autosampler and a column thermo-tat. For detection photodiode array detector (PDA) with 6 mmath length flow cell and electrospray interface mass spectrometerESI/MS) were used. The system was controlled with Chemsta-ion (Rev.A.10.02) and LCMSD Trap Control (Version 5.2) software.hemstation (Rev.A.10.02) and DataAnalysis (Version 3.2) weresed for UV and MS chromatograms analysis and peak integration.

Chromatographic analysis of DEEMM and TAHS derivativesas performed using an analytical column Synergi Hydro-RP 80A

4.60 mm × 250 mm, 4 �m) (Phenomenex, USA) with guard car-ridge 4.0 mm × 2.0 mm, polar endcapped C18 (Phenomenex). ForMOC-Cl and DNS derivatives, Eclipse XDB-C18 4.6 × 250 mm,

�m analytical column with guard column (4.6 × 12.5 mm, 5 �m;gilent) was used.

.6. LC/UV/MS analysis

For DEEMM and TAHS, mobile phase component A was bufferolution (pH = 3.2; 1 mM ammonium acetate in 0.1% formic acid)nd for FMOC-Cl and DNS 0.1% formic acid. Mobile phase compo-ent B was acetonitrile for all derivatization reagents. For FMOC-Cl,NS and TAHS, the eluent flow rate was 0.8 mL min−1 and the col-mn was maintained at 30 ◦C and 10 �L of the sample was injected.or DEEMM, the eluent flow rate was 0.9 mL min−1 and the columnas maintained at 40 ◦C and 20 �L of the sample was injected [28].

HPLC conditions for DEEMM were 0–12 min, 20–25% B;2–20 min, 25% B; 20–50 min, 25–60% B; for FMOC-Cl 0–45 min,0–100% B; for DNS 0–45 min, 10–100% B; and for TAHS 0–30 min,–35% B.

For all measurements, UV absorbance was recorded at 280 nmfull spectra were acquired for additional confirmation). ESI sourcearameters were same for all derivatization reagents: nebulizeras (nitrogen) pressure 50 psi (345 kPa), drying gas (nitrogen) flowate 12 L min−1 and drying gas temperature 350 ◦C. MS parame-ers were optimized for each derivatization reagent and amino aciderivative separately.

.7. Experimental plan

Important aspect of the investigation of the matrix effects fornion samples was that SeMet and Se-MeSeCys were added after

thod (ME%*) and 2) the new post-derivatization spike approach (dME%).

the sample preparation in order to be sure that there is no lossof selenoamino acids due to the SPE recovery—the approach com-monly known as post-extraction spiking. Using post-extractionspiking, the concentration of underivatized analyte in sample extractand standard solution will be equal. However, in order to evalu-ate ESI matrix effect, concentrations of derivatized analytes must beequal. Post-extraction spike technique cannot guarantee this as thesample matrix can affect the yield of the derivatization reaction.

For better understanding of matrix effects for derivatization,series of experiments were carried out. First, a re-evaluation of theanalysis methods was carried out, including LC/ESI/MS and deriva-tization procedures.

For the matrix effect investigation, following experiments werecarried out:

a) evaluation of matrix effects using conventional post-extractionspike method (Fig. 3-1). In case of derivatization LC/ESI/MS anal-ysis, the peak area of derivatized analyte from sample extractdepends on the derivatization reaction yield and ESI matrixeffect. We use asterisk, ME%*, to indicate this difference fromconventional post-extraction spike method;

b) a novel approach was developed for matrix effect investigationfor derivatization procedures: analyte is derivatized separatelyand then combined with the derivatized sample extract (Fig. 3-2). This method guarantees equal concentrations of derivatizedanalyte in standard solution and in sample extract, which allowsdirect assessment of ESI matrix effect. We call the approachpost-derivatization spike method and denote the respectivederivatization matrix effect as dME%;

c) investigation of how onion extract amount influences thematrix effect (concentration levels 4 �g g−1 for Se-MeSeCys and3 �g g−1 for SeMet);

d) evaluation of matrix effects at various selenoamino acids con-centrations when the amount of onion extract remains the same(matrix-matched calibration with concentrations 0.8–12 �g g−1

for Se-MeSeCys and 0.6–9 �g g−1 for SeMet).

From ME%* and dME%, relative yield of derivatization reactionin presence of sample matrix can be calculated as follows:

ME%∗

Yrel =

dME%× 100% (2)

where Yrel is the yield of derivatization reaction in presence of sam-ple matrix relative to the yield in standard solution.

Page 5

38 R. Rebane et al. / J. Chromatogr. B 955–956 (2014) 34–41

Table 1Retention times, parent and product ions used for the MS2 analysis.

DEEMM FMOC-Cl DNS TAHS

Aminoacid

Ret. time(min)

Parention (m/z)

Production (m/z)

Ret. time(min)

Parention (m/z)

Production (m/z)

Ret. time(min)

Parention (m/z)

Production (m/z)

Ret. time(min)

Parention (m/z)

Production (m/z)

63 25.0 416 252 15.2 359 17763 26.7 430 252 18.5 373 177

3

3

gtftMucsfl

3

tudwp

dpdpprmstwiTw

3

woss

sC(wtepsd

Table 2ME%* for Se-MeSeCys and SeMet in onion samples with relative standard deviationsin brackets (n = 3).

TAHS (%) DNS (%) FMOC-Cl (%) DEEMM (%)

100% onionSe-MeSeCys 22 (19) 64 (24) 14 (6) 65 (17)SeMet 17 (15) 58 (23) 25 (9) 58 (15)50% onionSe-MeSeCys 27 (17) 79 (14) 34 (12) 92 (8)SeMet 28 (14) 36 (33) 50 (17) 74 (16)

some of the selenoamino acids remain unreacted. This, however,is not the case, since the amount of reagent was optimized andhigher concentrations were also tested. For further confirmation,matrix-matched calibration curves were constructed for the

Se-MeSeCys 32.4 354 306; 308 23.2 428 2SeMet 37.2 368 320; 322 24.5 442 2

. Results and discussion

.1. Method adjustment for selenoamino acids

Since previously published LC/ESI/MS methods [27] (by ourroup) were developed for selected amino acids, some parame-ers were re-evaluated. Separation of Se-MeSeCys and SeMet peaksrom all other peaks was achieved without any major modifica-ions to the methods [27]. Retention times are presented in Table 1.

S optimization was also carried out and parent and product ionssed for MS2 analysis are presented in Table 1. In addition, it wasonfirmed that onion sample used as a blank did not contain theelenoamino acids or interfering components. Concentration levelsor selenoamino acids were chosen such that signals would be ininear range for all derivatization reagents.

.2. Optimization of derivatization procedures

The derivatization procedures were optimized with respect tohe concentration of the derivatization reagent and reaction timesing spiked onion extracts. For that, the maximum amount (asescribed in Section 2.4) of onion extract was used. Concentrationsere 4 �g g−1 for Se-MeSeCys and 3 �g g−1 for SeMet (spiking leveler gram of raw sample).

Two aspects were considered for changing: concentration of theerivatization reagent and the reaction time. For DNS and DEEMM,rocedures developed earlier for standard solutions (conditionsescribed in Section 2.4) [10] also provided the highest signal inresence of onion matrix. It could be concluded that the matrixresent in the derivatization mixture less affects the derivatizationeactions of DNS and DEEMM. However, for FMOC-Cl and TAHS, fewodifications were made compared to the methods developed for

tandard solutions [10]. Due to the complex matrix, higher concen-ration of FMOC-Cl was needed (original 1 mg mL−1 was replacedith 10 mg mL−1). Moreover, derivatization was carried out for 2 h

n order to complete the derivatization of selenoamino acids. ForAHS, modification was only made in derivatization reaction time,here now 45 min was used instead of 10 min.

.3. Matrix effects from classical post-extraction spike experiment

First, the matrix effects (ME%*) for all derivatization reagentsere calculated using the equation in Figure 3-1. Same amount

f underivatized selenoamino acids was present in the standardolution and post-extraction spiked onion samples. Results are pre-ented in Table 2.

Results show a signal loss when derivatized post-extractionpiked onion samples are analyzed. In the case of TAHS and FMOC-l (Fig. 4a), peak areas decrease by 75% in presence of onion matrixME%* < 25%). For DNS and DEEMM, ME%* values are a bit higher,ith signals up to 65% of the original, but still significant amount of

he signal is lost. To gain further information, the amount of onion

xtract was reduced and it was assumed that if the signal sup-ression is due to the compounds in the onion matrix, increasedignal should be observed. This approach models the conventionalilution method used to reduce matrix effects [31].

20% onionSe-MeSeCys 43 (46) 72 (16) 37 (15) 98 (6)SeMet 57 (13) 38 (10) 57 (10) 83 (9)

Two lower amounts of onion extract were tested; first, only halfof the original onion extract was used and then 20% of the origi-nal amount (rest of the sample was replaced with water). Resultsshowed (Table 2) that as the concentration of the onion decreased,the signal of selenoamino acids increased. For DEEMM, with smalleronion concentrations, almost no signal loss was observed (ME%*approaching 100%). As for other three reagents, no clear trend wasobserved. Moreover, interestingly, for 50% and 20% onion amount,results were very similar except for TAHS, where matrix effects aresignificantly reduced with each onion extract amount reduction.

One of the reasons for the loss of signal could be the insuffi-ciency of the reagent for the derivatization of all the amino groupcontaining compounds present in the onion samples and therefore

Fig. 4. Extracted ion chromatograms of standard solutions (black) and correspond-ing spiked onion extract (gray) for FMOC-Cl: a) post-extraction spike method (SeMet8.2 �g g−1 and Se-MeSeCys 11.4 �g g−1) and b) post-derivatization spike approach(SeMet 4.1 �g g−1 and Se-MeSeCys 5.7 �g g−1). (Note: * is for a peak from the onionmatrix with the same precursor ion as Se-MeSeCys).

Page 6

R. Rebane et al. / J. Chromatogr.

Table 3dME% for Se-MeSeCys and SeMet with a novel approach for onion samples withrelative standard deviations in brackets (n = 3).

TAHS (%) DNS (%) FMOC-Cl (%) DEEMM (%)

100% onionSe-MeSeCys 105 (5) 60 (28) 28 (53) 128 (6)SeMet 103 (12) 38 (23) 46 (44) 144 (41)50% onionSe-MeSeCys 92 (12) 77 (25) 46 (9) 112 (12)SeMet 96 (4) 66 (23) 70 (5) 127 (23)20% onion

f0sdtDefvrdt

3o

std(ttcay

rMada

vedp1fi

TE

Se-MeSeCys 110 (9) 99 (4) 68 (6) 109 (3)SeMet 113 (13) 92 (10) 88 (16) 104 (16)

ollowing concentration ranges: 0.8–12 �g g−1 for Se-MeSeCys and.6–9 �g g−1 for SeMet. Results showed that the ME%* remained theame for all concentration levels. Consequently, availability of theerivatization reagent does not limit the process. It is noteworthyhat dilution is efficient for reducing matrix effects only in case ofEEMM. It is very likely that reducing the amount of onion extractven further, at one point, ME%* would approach 100%. However,or samples where the concentration of the analytes is alreadyery low, this kind of dilution is often not possible. Based on theseesults it is not possible to decide whether the signal reduction isue to the interference during the derivatization procedure or ishere a signal suppression in the ESI-source.

.4. Novel post-derivatization spike approach for determinationf derivatization matrix effects

In order to gain more information about the reasons for theignal loss, a novel approach was developed. Two separate deriva-ization reactions were carried out for the elimination of the effectsue to the derivatization procedure itself. First, standard solution0% onion sample) of selenoamino acids was derivatized and thenhe blank onion extract (without selenoamino acids) was deriva-ized independently. These two derivatization mixtures were thenombined before the LC/ESI/MS analysis (Fig. 3-2). This approachllows eliminating any effects of onion matrix on the derivatizationield. As a result “pure” ESI matrix effect can be evaluated.

For comparability, this approach was used for all derivatizationeagents and also the reduced onion extract amounts were tested.atrix effects obtained with this novel approach are referred to

s dME% (Table 3). Moreover, the effect of onion sample on theerivatization reaction yield is also calculated (Eq. (2)) and resultsre presented in Table 4.

Results of the novel post-derivatization spike approach showery different dME% values than post-extraction spike matrixffects (ME%*). For TAHS, in ESI there are no signal suppressionsue to the onion matrix since signals with and without onion sam-

le extracts are almost the same (indicated by dME% values close to00% in Table 3). This allows a conclusion that low signals for therst approach are due to the effects of onion matrix components

able 4ffect of onion sample on the relative yield of derivatization reaction (Yrel).

TAHS (%) DNS (%) FMOC-Cl (%) DEEMM (%)

100% onionSe-MeSeCys 20 110 50 50SeMet 20 150 50 4050% onionSe-MeSeCys 30 100 70 80SeMet 30 50 70 6020% onionSe-MeSeCys 40 70 50 90SeMet 50 40 50 80

B 955–956 (2014) 34–41 39

on the derivatization reaction yield. Indeed, relative yields of TAHSderivatives in presence of onion matrix are poor (Table 4).

For DEEMM, results show that onion matrix actually has sig-nal enhancing effect (dME% > 100%, Table 3). At the same time, thepresence of matrix components lowers the reaction yield (Table 4).These effects – ESI signal enhancement and lowered yield – com-pensate each other and the ME%* obtained from post-extractionspiking is not too far from 100% (Table 2). At lower onion concen-trations both matrix effects – ME%* and dME% – are reduced, i.e.ME%* increases towards 100% and dME% decreases towards 100%.

In the case of DNS, ME%* and dME% values are below 100%, butsimilar for higher onion extract concentrations meaning that sig-nal suppression is present, but no effect from the sample matrixto the derivatization reaction yield is observed. With lower onionextract concentration, there is minimal suppression in the ESI-source (Table 3). Interestingly, reaction yields appear higher insolutions richer in onion matrix (Table 4). One hypothesis for thisis that components in the onion matrix are contributing to thereaction yield.

As for FMOC-Cl, post-derivatization method reveals ratherstrong ionization suppression (Fig. 4b) even when only 20% of theonion extract is present (Table 3). Also, the relative reaction yieldsare moderate (Table 4). Therefore, rather poor overall matrix effect,ME%*, is observed.

In conclusion, post-derivatization approach gives a clearer viewwhy derivatization of selenoamino acids in presence of onionextract causes the loss of signal. It also shows that different deriva-tization reagents are differently influenced by these components.

3.5. Calibration curves for two approaches

For more complete overview and also in order to obtain moreconfidence about the previously described results, for the 20%onion extract amount, matrix matched calibration curves were con-structed for all cases. Results supported all the previous results andwere reproducible for a chosen concentration range and, more-over, calibration curves were linear with R2 between 1.00 and 0.93for standard solutions and for matrix matched calibrations, R2 was1.00–0.96 (except DNS for Se-MeSeCys with 0.80). R2 values wereslightly poorer for the post-extraction spiked calibration curvesthan for the novel approach. Comparison of the slopes also showedthat in the case of TAHS and DEEMM, novel approach gave sim-ilar slopes for standards and matrix-matched calibrations (this issupported by the results obtained from dME%). For FMOC-Cl andDNS, slopes for the standard solution calibration curves and matrix-matched calibration curves were different, showing the presenceof matrix effects.

Representative calibration curves based on SeMet are presentedin Fig. 5, where two separate cases are observed for the matrix-matched calibrations: 1) for post-extraction spiked method wheresignal obtained with the onion extract is lower than the signal with-out the onion extract (FMOC-Cl) (a) or signal remains the same(DEEMM) (b) and 2) for the novel approach where signal obtainedwith the onion extract is lower than the signal without onionextract (FMOC-Cl) (c) or signal remains the same (TAHS) (d).

3.6. Results in the scope of designing new derivatization reagents

Results obtained in this work are also valuable when designingnew derivatization reagents for LC/ESI/MS. Usually, the new deriva-tization reagents are targeted towards more sensitive analysis [13]and the possibility for matrix effects is not directly considered.

Though more sensitive derivatization reagents enable to reducematrix effects indirectly, samples can be more diluted prior to anal-ysis, which reduces matrix effects [13]. It would be reasonable todesign the structures of the new derivatization reagents such as

Page 7

40 R. Rebane et al. / J. Chromatogr. B 955–956 (2014) 34–41

F ed meF

ta

Ftwwatcdc

4

aons

dtd

sTcTcS

i

[[

[[

[[[

ig. 5. Calibration curves for matrix-matched calibrations. For post-extraction spikMOC-Cl and d) TAHS.

o lower matrix effects. From the applicability point of view, thisspect is important when analysis in complex matrices is targeted.

In the present work more problems were encountered withMOC-Cl and DNS derivatization as the onion matrix affectedhe derivatization yield and ESI signal intensity. For TAHS, issuesith derivatization yield were observed but signal suppressionas not a problem. Reaction yield and signal appeared to be less

ffected by the matrix components in case of DEEMM deriva-ives. This suggests that for simpler matrices, TAHS is a goodandidate for derivatization. However, for more complex matrices,erivatization reagents similar to the DEEMM structure should beonsidered.

. Conclusions

Comparison of matrix effects for selenoamino acid LC/ESI/MSnalysis was carried out using four derivatization reagents andnion extract. Results showed that in some cases analytical sig-al is decreased due to the lower reaction yield and not due to theignal suppression in the ESI-source.

As a result, a new approach – post-derivatization spiking – waseveloped that enables determination of the signal loss due tohe matrix effects in ESI-source leaving aside the loss due to poorerivatization yield.

Thus, two aspects were found with the derivatization ofelenoamino acids in onion samples. Firstly, FMOC-Cl, DNS andAHS have problems with derivatization yields in complex matri-es. The second aspect is the ESI ionization point of view, whereAHS and DEEMM are very good derivatization reagents since the

omplex onion matrix does not suppress signals of derivates ofeMet and Se-MeSeCys.

Altogether, as the yield of derivatization reaction of TAHSs strongly affected by the onion matrix, out of the four

[[

[

thod a) FMOC-Cl and b) DEEMM and for the post-derivatization spike approach c)

derivatization reagents, DEEMM is most suitable for analysis forcomplex matrices.

Acknowledgment

This work was supported by Grant no. 8572 from the EstonianScience Foundation.

References

[1] Y.V. Tcherkas, L.A. Kartsova, I.N. Krasnova, J. Chromatogr. A 913 (2001) 303–308.[2] C. Gabel-Jensen, J. Odgaard, C. Skonberg, L. Badolo, B. Gammelgaard, J. Anal. At.

Spectrom. 24 (2009) 69–75.[3] E. Dumont, Y. Ogra, F. Vanhaecke, K.T. Suzuki, R. Cornelis, Anal. Bioanal. Chem.

384 (2006) 1196–1206.[4] S. McSheehy, W. Yang, F. Pannier, J. Szpunar, R. Lobínski, J. Auger, M. Potin-

Gautier, Anal. Chim. Acta 421 (2000) 147–153.[5] F. Gosetti, P. Frascarolo, S. Polati, C. Medana, V. Gianotti, P. Palma, R. Aigotti, C.

Baiocchi, M.C. Gennaro, Food Chem. 105 (2007) 1738–1747.[6] W.M.A. Niessen, Liquid Chromatography–Mass Spectrometry, 2nd ed., Marcel

Dekker, New York, 1999.[7] T. Toyo’oka (Ed.), Modern Derivatization Methods for Separation Sciences, John

Wiley & Sons, Chichester, UK, 1999.[8] M. Jemal, Y.-Q. Xia, Curr. Drug Metab. 7 (2006) 491–502.[9] A.M. Timperio, M. Fagioni, F. Grandinetti, L. Zolla, Biomed. Chromatogr. 21

(2007) 1069–1076.10] R. Rebane, M.-L. Oldekop, K. Herodes, J. Chromatogr. B 904 (2012) 99–106.11] K. Shimbo, T. Oonuki, A. Yahashi, K. Hirayama, H. Miyano, Rapid Commun. Mass

Spectrom. 23 (2010) 1483–1492.12] W.-C. Yang, H. Mirzaei, X. Liu, F.E. Regnier, Anal. Chem. 78 (2006) 4702.13] K. Shimbo, A. Yahashi, K. Kirayama, M. Nakazawa, H. Miyano, Anal. Chem. 81

(2009) 5172–5179.14] M. Oss, A. Kruve, K. Herodes, I. Leito, Anal. Chem. 82 (2010) 2865–2872.15] P.J. Taylor, Clin. Biochem. 38 (2005) 328–334.16] W.M.A. Niessen, P. Manini, R. Andreoli, Mass. Spectrom. Rev. 25 (2006)

881–899.17] T.M. Annesley, Clin. Chem. 49 (2003) 1041–1044.18] R. King, R. Bonfiglio, C. Fernandez-Metzler, C. Miller-Stein, T. Olah, J. Am. Soc.

Mass Spectrom. 11 (2000) 942–950.19] A. Kruve, A. Künnapas, K. Herodes, I. Leito, J. Chromatogr. A 1187 (2008) 58–66.

Page 8

atogr.

[

[[

[[

[[[

R. Rebane et al. / J. Chrom

20] M. Yadav, V. Trivedi, V. Upadhyay, G. Shah, G.A. Baxi, S. Goswami, P.S. Shrivastav,J. Chromatogr. B 885–886 (2012) 138–149.

21] M. Stadlober, M. Sager, K.J. Irgolic, Food Chem. 73 (2001) 357–366.

22] P. Uutela, R.A. Ketola, P. Piepponen, R. Kostiainen, Anal. Chim. Acta 633 (2009)

223–231.23] R. Rebane, K. Herodes, Eur. J. Mass Spectrom. 18 (1) (2012) 71–75.24] A.A. Descombes, J.-L. Veuthey, W. Haerdi, Fresenius J. Anal. Chem. 339 (1991)

480–484.

[[

[[

B 955–956 (2014) 34–41 41

25] X. Kang, J. Xiao, X. Huang, Z. Gu, Clin. Chim. Acta 366 (2006) 352–356.26] A. Jámbor, J. Molnár-Perl, J. Chromatogr. A 1216 (2009) 3064–3077.27] R. Rebane, K. Herodes, I. Leito, Analyst 136 (2011) 5241–5246.

28] R. Rebane, K. Herodes, Anal. Chim. Acta 672 (2010) 79–84.29] J.W. Kim, S.U. Kim, H.S. Lee, I. Kim, M.Y. Ahn, K.S. Ryu, J. Chromatogr. A 1002

(2003) 93–110.30] K. Kivirand, R. Rebane, T. Rinken, Sens. Lett. 9 (2011) 1794–1800.31] A. Kruve, I. Leito, K. Herodes, Anal. Chim. Acta 651 (2009) 75–80.