Journal of Chromatography A, 1104 (2006) 209–221

Stainless steel electrospray probe: A dead endfor phosphorylated organic compounds?

R. Tuyttena, F. Lemierea,b,∗, E. Wittersb,c, W. Van Dongena,b, H. Slegersd,R.P. Newtone, H. Van Onckelenb,c, E.L. Esmansa,b

a Department of Chemistry, Nucleoside Research and Mass Spectrometry Unit, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgiumb Centre for Proteome Analysis and Mass Spectrometry, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgium

c Department of Biology, Laboratory of Plant Biochemistry and Plant Physiology, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, Belgiumd Department of Biochemistry, Cellular Biochemistry, University of Antwerp, Universiteitsplein 1, B-2610 Wilrijk, Belgium

e Biochemistry Group, School of Biological Sciences and Biomolecular Analysis Mass Spectrometry Facility, University of Wales, Swansea, UK

Received 9 September 2005; received in revised form 30 November 2005; accepted 1 December 2005Available online 27 December 2005

Abstract

ectrosprayi s of phos-p iomoleculess sage throught tup were ablt heir releasef H 12). Fromt nt to realisea the moleculesa pray processw©

K spho;C

1

mmvtfl

v

owa-com-ndsntsgh-ight

t be

ofsomeron-lves.luck-nce

0d

A study of the interaction of phosphorylated organic compounds with the stainless components of a liquid chromatography–elonisation–mass spectrometry system (LC–ESI–MS) was carried out to disclose a (forgotten?) likely pitfall in the LC–ESI–MS analysihorylated compounds. The retention behaviour of some representative compounds of different important classes of phosphorylated buch as nucleotides, oligonucleotides, phosphopeptides, phospholipids and phosphorylated sugars was investigated during their pashe injector and the stainless steel electrospray capillary. It became clear that the stainless steel components within the LC–ESI–MS seeo retain and trap phosphorylated compounds when these compounds were introduced under acidic conditions (0.1% acetic acid). Trom these stainless steel parts was accomplished by applying an extreme basic mobile phase (25–50% ammonium hydroxide, ca. phe data collected one could conclude that the availability of a primary phosphate group appeared imperative but was not always sufficiedsorption on a stainless surface. Furthermore, the number of phosphate moieties seemed to enhance the adsorption properties ofnd hence roughly correlated with the analyte fraction lost. Corrosion of the inner surface caused by the mobile phase and the electrosas found to be an important factor in the course of these adsorption phenomena.2005 Elsevier B.V. All rights reserved.

eywords: Electrospray; Phosphorylated compounds; Metal affinity; Adsorption; Phosphopeptides; Nucleotides; Oligonucleotides; Phospholipids; Phosugarsorrosion

. Introduction

It was known that stainless steel parts mounted in liquid chro-atography (LC) equipment can corrode under the influence ofany of the conventionally used solvents[1–3]. Even passi-

ated surfaces can be activated by reversed-phase solvents andhe resulting corrosion can be extensive[1] leading to the mal-unction of some parts of the instrument which may result in e.g.ess accurate flow rates.

Yet more important was the observation of the irre-ersible adsorption of compounds to (corroded) stainless steel

∗ Corresponding author. Tel.: +32 3 2653406; fax: +32 3 2653233.E-mail address: filip.lemiere@ua.ac.be (F. Lemiere).

surfaces[4]. This phenomenon should be remembered as ndays analysts are confronted with samples of increasingplexity. Moreover, the (partial) loss of specific compouprior to detection is particularly important if limited amouof sample are available. Within the context of high throuput methods producing massive amounts of data one mforget this problem and incorrect/incomplete data mighgenerated.

The culprit in the case of LC equipment is the availabilityiron in the system. This issue was already addressed bymanufacturers of (nano-)LC instrumentation by designing ifree chromatographic systems and injection/switching vaUnfortunately, these designs are much more expensive andily the majority of analytes are indifferent towards the preseof iron.

021-9673/$ – see front matter © 2005 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2005.12.004

210 R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221

Although the phenomena of corrosion and adsorption arewell documented their impact upon the acquisition of analyticaldata and more specific electrospray ionisation–mass spectrom-etry (ESI–MS) data are frequently overlooked. In general thephysical situation in ESI capillaries has been described as acontrolled current electrolytic (CCE) flow cell[5–8]. The occur-rence of emitter corrosion was nicely exemplified by Van Berkel:the Fe(II) and Cu(II) ions formed by anodic corrosion of a nanostainless steel emitter produced a complex with phenanthrolineand were detected as such by both MS and photo diode array[9,10]. Oxidation/reduction processes in the ESI interfaces wereexploited and even optimised for analytical benefits as recentlyoutlined by Van Berkel et al.[11] (and many references citedtherein).

Yet the impact of the corrosion of the internal surface ofthe ESI capillary during ESI–MS analyses is – to the best ofour knowledge – not yet explicitly accounted for[12]. One candeduce that in the course of time the surface of the emitter isaltered by both solvents and redox processes resulting in anexpanded contact surface. This would increase the prospect ofcompound losses by adsorption or even chemical binding.

In a study using Fe(III)-immobilised metal affinity chro-matography (IMAC) as a tool for the on-line sample clean up andpreconcentration of nucleoside mono-, di- and triphosphates, theloading and elution of the samples was monitored by ESI(−)-MS in order to probe the effectiveness of the trapping process[ inalp tureac ith-o ingo aseo ), tht K))a capab rapp

conc omp mass

2

2

fromS bleg ericd osys( itivec dsa mS d oa edp id

was a mixture containing combinations of different fatty acidresidues (palmitic, stearic, oleic, linoleic and traces of otherfatty acids). Raffinose was also obtained from Sigma–Aldrich.Trypsin was purchased from Promega (Madison, WI, USA).

Acetic acid (HOAc; analytical-reagent grade) and ammo-nium hydroxide 28–30 wt.% (NH4OH; analytical-reagentgrade) were from Acros (Geel, Belgium). Ammonium acetate(NH4OAc; analytical-reagent grade) was purchased from Merck(Overijse, Belgium). Ammonium bicarbonate (NH4HCO3;analytical-reagent grade) was from Fluka (Bornem, Belgium).Acetonitrile (ACN; HPLC supra gradient), methanol (MeOH;HPLC Absolute) and water (H2O; HPLC) were from Bio-solve (Valkenswaard, The Netherlands). Twenty-five millime-tres HPLC Syringe Filters were obtained from Alltech (Lokeren,Belgium).

2.2. Sample preparation

Prior to a series of measurements, stock solutions of thedifferent compounds were prepared in HPLC grade water atconcentrations of 10−3 and 10−4 M and stored at−21◦C untiluse. Just before analysis they were diluted further according tothe experimental needs.

For the experiments described under Section3.1 an indi-vidual compound concentration of 2.5× 10−6 M was main-tained throughout all measurements. An exception was madef rcialp( ed int (v/v)Hss Fort 0.1%( -t ono-,d em (all2 thed ll2 f thet ITR( QRK( )( stsa ofg

w f1

2( 00t lew )M

13]. The latter is the result of the selective chelation of termhosphate groups by Fe(III) under acidic conditions, a fealready put to use to enrich phosphopeptides[14–20]. In theourse of these experiments analogous “IMAC effects” wut an IMAC precolumn present were noticed. This intrigubservation prompted us to start additional experiments bn the assumption that the injector (needle, loop and valve

ubing (either fused silica or poly(ether ether ketone) (PEEnd/or the stainless steel capillary of the ESI source werele of mimicking the Fe(III)-IMAC behaviour and hence thosphorylated organic compounds.

In this paper we wish to demonstrate that this is a tangibleern for the analysis of a variety of phosphorylated organic counds by liquid chromatography–electrospray ionisation–pectrometry system (LC–ESI–MS).

. Experimental

.1. Chemicals

All nucleosides and nucleotides were purchasedigma–Aldrich (Bornem, Belgium) in the highest availarade. The nucleotides were in the sodium salt form. Dimeoxyoligonucleotides were obtained from Sigma-GenPampisford Cambridge, UK). The phosphopeptide posontrol set, the natural phospholipid 3-sn-phosphatidic aciodium salt egg yolk lecithin, the glycerophosphate, the�-caseins well as the [Glu1]-fibrinopeptide B (human) were also froigma–Aldrich. The phosphopeptide control set consistetetraphosphopeptide and a monophosphopeptide obtain

urification of a tryptic digest of�-casein. The phospholip

de

-

--s

fby

or the sn-phosphatidic acid sodium salt: as the commeroduct is a mixture itself a concentration of ca. 5× 10−5 Mbased on the average mass of the mixture) was ushe samples. The samples were dissolved in 0.1%OAc (50:50 (v/v) MeOH/H2O) for the “injector + stainless

teel capillary” (INJ + SSC), “injector” (INJ) and “stainlessteel capillary” (SSC) effect experiments (see below).he reference samples the compounds were made up inw/v) NH4OAc (25:25:50 NH4OH/H2O/MeOH). The nucleoide test set contained adenosine (Ado), adenosine mi-, triphoshate (AMP/ADP/ATP), uridine (Urd), uridinono-, di-, triphoshate (UMP/UDP/UTP) and raffinose.5× 10−6 M). The oligonucleotide test set consisted ofimeric deoxyoligonucleotides GpC, 5′-pGpC and raffinose (a.5× 10−6 M). The phosphopeptide test set was made o

etraphosphopeptide RELEELNVPGEIVEpSLpSpSpSEESTppept), the monophosphopeptide FQpSEEQQQTERELMppept), [Glu1]-fibrinopeptide B (EGVNDNEEGFFSARglu-fib) and raffinose (all 2.5× 10−6 M). The phospholipids teet contained thesn-phosphatidic acid sodium salt (5× 10−5 M)nd raffinose (2.5× 10−6 M). The last test set was made uplycerophosphate and raffinose (all 2.5× 10−6 M).

The samples measured under Section3.2 (Ado and ATP)ere dissolved in 0.1% (v/v) HOAc (H2O) in a concentration o0−5 M.

For the experiment described under Section3.3 ca.mg �-casein was dissolved in 1 mL 100 mM NH4HCO3

90:10 H2O/ACN) pH 8.5 and digested for 3 h with 1:1rypsin (1�g/1�L) at 37◦C. Following digestion the sampas diluted to ca. 2.5× 10−6 M in 0.1% (v/v) HOAc (50:50 (v/veOH/H2O) and used as such.

R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221 211

Table 1Abbreviations, compositions and purpose of mobile phases used

Mobile phase Composition Purpose

A 0.1% (v/v) HOAc (50:50MeOH/H2O)

Sample loading, rinsing,acidic

B 100% ACN RinsingC 0.1% (w/v) NH4OAc

(25:25:50NH4OH/H2O/MeOH)

Elution of adsorbedcompounds, basic

D 0.1% (v/v) HOAc (H2O) Sample loading, rinsing,acidic

E 50% (v/v) NH4OH (H2O) Elution of adsorbedcompounds, basic

2.3. LC conditions (Table 1)

For the experiments described in Section3.1a ternary gradi-ent pump (CapLC, Waters, Manchester, UK) with integratedinjector was used. The mobile phases were all delivered at10�L/min and were 0.1% (v/v) HOAc (50:50 MeOH/H2O)(A), 100% ACN (B) and 0.1% (w/v) NH4OAc (25:25:50NH4OH/H2O/MeOH) (C). They were, respectively used as theloading, rinsing and elution solvent. Prior to use they were fil-tered on a 0.2�m nylon 25 mm syringe filter (Alltech). Sampleswere injected via a 10�L PEEK loop mounted on a 6 port valvetype stainless steel (SS) Nitronic 60 injector (Valco InstrumentsHouston, TX, USA) The injector was connected to the electro-spray probe via PEEK tubing (Upchurch Scientific, Oak Harbor,WA, USA). The total of all the components associated with theinjection of the sample and its deliverance to the mass spectrometer (actual injector and tubings) was called injector (INJ)throughout the text. An amount of 10�L of the analyte wasinjected (2.5× 10−6 M (25 pmol/injection)). Samples were elec-trosprayed by means of the standard stainless steel ESI capilla(SSC) (184 mm× 100�m I.D.).

In all experiments described under Section3.1sample injec-tions and LC effluents were monitored by ESI–MS in the nega-tive ion mode (ESI(−)) unless specified.

Se

syst ingw The8 a. 5t

Sc

theE withb Jusb ed tt wasa alvem

Study of the SSC effect (Section 3.1): experimentalconditions

During injection and subsequent flushing with mobile phaseA the INJ was attached to the MS (8 min). Subsequent the INJwas manually disconnected and rinsed with mobile phases C(8 min) and B (10 min). After that the MS was again hooked upwith the INJ. An additional wash of the INJ + SSC with mobilephase B preceded the definitive application of mobile phase C.

Reference (Section 3.1): experimental conditionsSamples were dissolved in 0.1% (w/v) NH4OAc (25:25:50

NH4OH/H2O/MeOH) (C) and flow injected in mobile phase C.The applied solvent sequences for the different experiments

described under Section3.1were summarized inFig. 1.

A Waters CapLC (Waters) system equipped with a 996 diodearray detection (DAD) system (Waters) with capillary flow cellwas used for the experiments described under Section3.2. Thetotal flow path consisted of a SS injector, a 20�L PEEK loop,75�m I.D. silica tubing and the SSC of the ESI interface. Sam-ples were dissolved in 0.1% HOAc (H2O) (10−5 M) and anamount of 20�L was injected (ca. 200 pmol). The carrier solventwas 0.1% (v/v) HOAc (H2O) (D) and was delivered at a flowrate of 10�L/min. After injection of the sample, the completesetup was flushed for at least 20 min with mobile phase D at thesame flow rate. Elution of adsorbed compounds was achievedb( byE attroI

2

a ripleq ith ap ndardS sprayp

Q d sep-a rs tS turew nda ectraw st setE 0 Daa rdedb phos-p vereda lipidm were,r wasm 30 Vw

ss ed

tudies of the INJ + SSC effects (Sections 3.1 and 3.3):xperimental conditions

Samples were injected in mobile phase A and the wholeem was flushed for 8 min followed by 8 min of additional rinsith mobile phase B. Finally mobile phase C was applied.min intervals were arbitrarily chosen and represented c

imes the internal volume of the SSC.

tudy of the INJ effect (Section 3.1): experimentalonditions

Prior to sample injection the INJ was disconnected fromSI–MS setup. After sample injection the INJ was rinsedoth mobile phases A and B, respectively for 8 and 10 min.efore switching to mobile phase C the INJ was reconnect

he MS. Coupling and decoupling of the injector to the MSlways done manually as the use of an extra switching vight involve extra adsorption sites.

,

-

ry

-

5

to

y multiple injections of 20�L of 50% (v/v) NH4OH (H2O)E). Eluting compounds were monitored by UV (DAD) orSI(−)-MS on a triple quadrupole mass spectrometer (Qu

I, Waters) or both.

.4. MS conditions

Electrospray mass spectra were recorded in ESI(−) either onQ-TOF II mass spectrometer (Waters) or a Quattro II t

uadrupole mass spectrometer (Waters) both equipped wneumatically assisted Z-spray source (Waters) and staS electrospray probes. The type of SS used for the electrorobes was not released by the manufacturer.

In the experiments described under Sections3.1 and 3.3the-TOF was used. For each test set the instrument was tunerately to maximise response for the particularm/z range undetudy. The ionisation voltage applied was−3.0 kV throughouections3.1 and 3.3. The source and desolvation temperaere both set at 100◦C. A cone gas flow rate of ca. 60 L/h adesolvation gas flow rate of ca. 150 L/h were applied. Spere recorded in full scan mode. In case of the nucleotide teSI(−) mass spectra were recorded in the range of 200–75t a cone voltage of 30 V. The oligonucleotide set was recoetween 200 and 800 Da at a cone voltage of 30 V. For thehopeptide test set a wide range of 300–2000 Da was cond a cone voltage of 40 V was applied. For the phosphoixture the mass range covered and cone voltage applied

espectively, 200–1000 Da and 30 V. The glycerophospateonitored between 100 and 600 Da and a cone voltage ofas used.For the experiments under Section3.2 the Quattro II mas

pectrometer was used for ESI(−) spectra recording; the cover

212 R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221

Fig. 1. Summary of the different solvent, detection and manipulation sequences following sample injections implemented in the experiments described in Section3.1.

mass range was 350–650 Da at a cone voltage of 40 V. Flow ratesfor the cone and desolvation gases were similar to those of theQ-TOF experiments.

3. Results

3.1. Hardware contributions to retention/losses ofphosphorylated analytes

In a first series of experiments we studied the behaviour ofa limited set of terminally phosphorylated and non-terminallyphosphorylated representatives of important biochemical com-pound classes, i.e. nucleotides, oligonucleotides, phospholipids,phosphosugars and phosphopeptides. An equimolar amount ofa non-phosphorylated compound, i.e. raffinose was added to allsamples as reference. InTable 2the molecular mass and relevantm/z values of all compounds used were summarised, togetherwith some structures.

We subdivided the sample/solvent path of the analytical sys-tem in two parts: the SSC and the INJ, the latter covering all thehardware components associated with the injection of the sam-ple and its delivery to the mass spectrometer. The idea was toprobe the retention behaviour of the phosphorylated compoundsin both SSC and INJ. Any retention was investigated by plottingthe extracted ion chromatograms for [M-H]− of the analytes([M-2H]2− or [M-3H]3− in the case of the (phospho)peptides),b s anb

ndsu samp cycl

involving: (1) mobile phase A; (2) mobile phase B and (3) mobilephase C, as outlined inFig. 1. Preliminary experiments (data notshown) learned that only mobile phase C (high NH4OH concen-tration) was able to release compounds adsorbed.

Next the contribution of each individual component (INJ andSSC effect) to the retention of phosphorylated compounds wasexamined.

To probe only the INJ effect the LC and MS were decou-pled during sample injection and rinsing of the PEEK tube. Justbefore applying mobile phase C the LC system was coupled tothe MS. Any signal then observed by ESI(−)-MS must origi-nate from compounds previously adsorbed in the INJ under theacidic loading conditions.

The contribution of the SSC (SSC effect) to the adsorption ofphosphorylated compounds was investigated in a similar way.During injection and subsequent washing with the mobile phaseA the INJ was attached to the MS. In this way eventual elution ofcompounds could be detected. Subsequently the INJ was man-ually disconnected from the MS system and rinsed with mobilephase C followed by mobile phase B. By doing so all com-pounds adsorbed in the INJ were removed. After that the INJwas reinstalled and coupled to the electrospray SSC. An addi-tional rinsing step of the INJ and SSC with 100% ACN (mobilephase B) preceded the definitive delivery of mobile phase C. Ifany compounds were adsorbed on the SSC ions should now bedetected.

entsd ible ar reten-t ndsw elu-

y comparing the areas under the chromatographic peaky interpretation of the according mass spectra.

Initially the cumulative INJ + SSC effects on the compounder test were investigated: after injection of a particularle the complete instrumental setup was subjected to a

d

-e

In order to validate all the data gathered in the experimescribed above and make a relative quantification posseference experiment was needed. It was assumed that noion on the LC–ESI–MS hardware would occur if the compouere dissolved and in-flow injected in mobile phase C. The

R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221 213

Table 2Molecular masses (monoisotopic) and relevantm/z values of the deprotonated compounds used, together with some structures

tion patterns and mass spectra obtained under these conditionswere used as reference data to obtain a rough idea of the amountof compound adsorbed under acidic conditions.

3.1.1. NucleotidesThe nucleotide test mixture containing uridine, adenosine,

uridine 5′-mono-, di-and triphosphate, adenosine 5′-mono-, di-and triphosphate together with raffinose was subjected to theexperiments described above.

The extracted ion chromatograms of the cumulativeINJ + SSC effect experiment showed a clear difference betweenthe behaviour of the non-phosphorylated and phosphorylatedcompounds. This is illustrated inFig. 2 showing the recon-structed ion chromatograms for the [M-H]− of raffinose (m/z503), Urd (m/z 243),UMP (m/z 323), UDP (m/z 403) and UTP

(m/z 483). In general the non-phosphorylated compounds (raffi-nose and Urd) emerged immediately after injection. During thefollowing steps, i.e. rinsing with mobile phases A and B and sub-sequent change to mobile phase C no raffinose or Urd eluted. It isclear that within 5 min after injection of the sample plug also sig-nals of the phosphorylated compounds were observed. However,the question remained whether a fraction of these compoundswas retained by the INJ and/or the SSC, since this could have atremendous effect on the LODs of phosphorylated compounds.

When the mobile phase was changed to C no increment ofthe signals of the raffinose and the nucleosides were observed.However, [M-H]− signals were recorded pointing to the elutionof the nucleoside 5′-mono-, di- and triphosphates.

As can be seen fromFig. 2the UMP signal at ca. 18.5 min isquite low in intensity. Therefore, we injected a solution of solely

214 R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221

Fig. 2. The extracted ion chromatograms of uridine, UMP, UDP and UTP andraffinose when monitoring the INJ + SSC effect. The increments in the traces ofUDP and UTP at ca. 18.5 min matches the mobile phase C elution front.

UMP. This experiment showed that UMP was not adsorbed bythe stainless steel surfaces. The signal observed around 18.5 mincould therefore be explained by in source CAD of UDP orUTP. Analogous results were obtained for AMP. The SSC andINJ effect experiments yielded similar elution profiles. In eithercase the extracted ion chromatograms and the according spectraunambiguously indicated considerable amounts of the nucleo-side di- and triphosphates were recovered when applying thebasic C mobile phase. In the case of the SSC effect experi-ment only the electrospray SSC itself could account for theselosses. For the INJ effect two potential contributors could beresponsible: the stainless steel injection valve and the PEEKtubing linking the valve with the ESI probe. Since there is noreason/literature suggesting interactions of nucleoside di- andtriphosphates with PEEK tubing the injection valve remainedthe only suspect.

In order to get an insight in the relative amount of compoundadsorbed on the stainless steel surface the intensity of the diferent [M-H]−-signals observed in the experiments describedabove was compared to the intensity of the corresponding[M-H] −-signal obtained in the reference experiment. Theseareas were plotted in a normalised bar graph (Fig. 3). Fromthese data it can be seen that the non-phosphorylated compoun(nucleosides + raffinose) experienced no adsorption whatsoeveYet the results for the di- and triphosphates were striking: recoveries from the different hardware components –compared to thr n thc up to6 nt io

Fig. 3. Normalised areas under ion chromatograms of the nucleotide testset compounds associated with their recoveries from hardware components(INJ + SSC, INJ and SSC) and their injection in mobile phase C eluent con-ditions (reference), the latter was equated to 100%.

suppression phenomena and should as such be interpreted withcaution; nevertheless the adsorption effect cannot be neglected.Furthermore, there seemed to be a certain correlation betweenthe aptitude to adsorption and the number of phosphate moietiespresent in the molecule.

To exclude any possible electrical effects in the SSC similarexperiments were conducted but under ESI(+) (data not shown).The same effects and tendencies were observed.

3.1.2. OligonucleotidesTwo dimeric 2′-deoxyoligonucleotides of guanosine and cyti-

dine together with raffinose made up the oligonucleotide test set.The selected dimers were GpC ([M-H]− = 555) and 5′-pGpC([M-H] − = 635), the former only having a 3′-5′-phosphodiesterbond, the latter having an terminal 5′-phosphate group.

In search for the INJ + SSC effect it was found that, by exam-ining the extracted ion chromatograms for [M-H]−, 5′-pGpCexperienced already some retention in the liquid pathway andexhibited a delayed return to the background values whilst flush-ing the solvent path with the acidic mobile phase A. During theACN wash (B) none of the traces showed a signal increment. Theapplication of the mobile phase C led again to important incre-ment atm/z 635 for 5′-pGpC, yet also a signal atm/z 555 wasseen as is illustrated inFig. 4. The observed pattern for the tracesof raffinose and GpC showed a square wave form as expected forfl ioni by ins wasr rovedt

SCw iment( ncedc h ther andI thati hatew le inL d ben t.

eference experiment-varied between 12 and 31%. Wheumulative effect of INJ + SSC was considered recoveries4% were recorded. These values did not take into accou

f-

dsr.

-ee

n

ow injection. As GpC showed no retention following injectts presence in the trace at 18.35 min was explained againource CAD of 5′-pGpC. In order to be sure the experimentepeated with GpC as the sole compound injected, which phe previous assumption.

Again the contributions of both the INJ and the Sere investigated and compared with the reference exper

Fig. 5). The observed adsorption effects were less pronouompared to the nucleoside-di-and triphosphates, thougecoveries of 5′-pGpC were about 9 and 26% for the SSCNJ, respectively. From these results one could concluden the case of dimeric oligonucleotides, a primary phospas imperative for adsorption to the SS surfaces availabC–ESI–MS hardware. However, more experiments woulecessary to determine whether this is a general rule or no

R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221 215

Fig. 4. The extracted ion chromatograms of GpC, 5′-pGpC and raffinose whenmonitoring the INJ + SSC effect. The increments in the traces of GpC, 5′-pGpCat ca. 18.5 min matches the mobile phase C elution front.

Fig. 5. Normalised areas under ion chromatograms of the oligonucleotide tesset compounds associated with their recoveries from hardware componen(INJ + SSC, INJ and SSC) and their injection in mobile phase C eluent con-ditions (reference), the latter was equated to 100%.

3.1.3. PhosphopeptidesWithin the context of the current worldwide focus on pro-

teomics it was important to investigate whether phospho-peptides were also prone to (ir)reversible interactions withLC–ESI–MS hardware. For that reason a monophosphory-lated peptide (Mppept), a tetraphosphorylated peptide (Tppept),[Glu1]-fibrinopeptide B (Glu-fib) and raffinose were mixed tomake up another test set. The commercially available phospho-peptides originated from a tryptic digest of�-casein (cf. Section2). The sample was injected and subjected to the INJ + SSC sol-vent sequence. The different reconstructed ion chromatogramsof the negatively charged deprotonated molecules were gen-erated and plotted (Fig. 6). For the raffinose the [M-H]− atm/z 503 was used and for the peptides the summed isotopiccluster was used (Mppept:� m/z 1029.5–1031; Tppept:� m/z1039.5–1040.5; Glu-fib:� 783.8–785.3). For evaluation of thephosphopeptides, respectively [M-2H]2− and [M-3H]3− wereselected and for the Glu-fib: [M-2H]2−.

The effect seen for the Tppept was impressive: the signaltrace exhibited a huge increment when mobile phase C reachedthe ESI interface. Furthermore, the Tppept exhibited retentioncompared to raffinose once injected. The effects for Mppept wereless pronounced: a slight retention was observed together witha limited recovery by mobile phase C. The Glu-fib trace agreedwell with the behaviour expected for a non-retained compound,yet some Glu-fib was also recovered when the mobile phase wass nalet tiple)p

F itoringowaw

t

witched to mobile phase C. This was contradicting our ratiohat correlated the SS interaction with the presence of (mulhosphorylations.

ig. 6. The extracted ion chromatograms of the peptide test set when monf the INJ + SSC effect. For the peptides isotopic envelopes spanning 4m/z units

tsere summed. The charge states were as indicated. The increment in the tracest ca. 12 min agrees with the ACN front (mobile phase B), the one at ca. 18.5 minith the mobile phase C elution front.

216 R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221

Fig. 7. Part of the spectra (m/z 800–1600) obtained from recovery of peptidesfrom the different hardware components after injection of the peptide test set andinjection of the test set made up in solvent C. The spectra are dominated by thedouble and triple charged isotopic envelopes of the peptides, sodium adductionis also observed. Intensities of the different spectra are mutually compared andnormalised to 100%. The complete spectrum of the reference is also given, whichmakes the dominance of Glu-fib in the spectrum clear. Ion suppression of thephosphopeptide signals by Glu-fib in the reference results in the overestimationof the recoveries in the other experiments.

On the other hand it was observed that Glu-fib is far moreprominent in the spectra compared to the other compounds eventhough it was present in an equimolar quantity. This preferentialdetection might overestimate the actual (absolute) presence ofGlu-fib. Nevertheless, it did not alter the fact that Glu-fib wasrecovered to some extent and thus previously adsorbed.

Additionally the individual INJ and SSC effects were theninvestigated for the peptide test set, followed by the referenceexperiment (Fig. 7). The latter confirmed that Glu-fib was themost abundant ion when recorded under the eluent conditions(mobile phase C) and that at the same time the signals of theother compounds were suppressed. More interesting was theobservation that again from both the SSC and INJ a significantamount of the phosphorylated compounds, especially Tppept,was recovered. At the same time an impurity in the commer-cially phosphopeptides sample showed: the signal atm/z 987.7agrees with another triply charged tetraphosphorylated peptideoriginating from�-casein, with mass 2965.15[21]. The Glu-fibcontribution from the INJ if compared to the SSC contribution

Fig. 8. Normalised areas under ion chromatograms of the peptide test set com-pounds associated with their recoveries from hardware components (INJ + SSC,INJ and SSC) and their injection in the mobile phase C eluent conditions (refer-ence), the latter was equated to 100%. Most probable the Glu-fib is quenchingthe signals of the phosphopeptides in the reference resulting in their under-representation.

was more pronounced. This might ensue from a memory on thePEEK tubing rather than from interaction with the stainless steelsurfaces as Glu-fib is rather lypophilic. The bar graphs derivedfrom the areas under the elution peaks of the different extractedion chromatograms were not as informative as before because ofthe Glu-fib ion suppression effect. This gave rise to an apparentrecovery for Tppept in the INJ + SSC effect of >100% com-pared to the reference data (Fig. 8). The quenching was evenmore pronounced when a similar experiment was conducted inthe positive ion mode, though the same tendencies were there:the loss of phosphorylated compounds in acidic conditions toSS hardware components was significant and correlated withthe number of phosphate groups present.

3.1.4. Phospholipids and phosphosugarsSome other classes of monophosphorylated biomolecules

were briefly investigated, namely terminally phosphorylatedphospholipids and a phosphorylated sugar. The phospholipidsample, 3-sn-phosphatidylcholine contained a typical fatty acidester content of mainly palmitates and oleates and considerableamounts of stearates, linoleates, etc. When subjected to the ini-tial INJ + SSC experiment the bulk only eluted if mobile phaseB was applied. This agreed well with their hydrophobic prop-erties and as such their anticipated interaction with the PEEKtubing. Subsequent application of the mobile phase C releaseds ouro artial

ome more of the phospholipids, which would agree withbservations for the other studied compounds. However, p

R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221 217

Fig. 9. Left traces: the UV traces at 258 nm of two 20�L injections: a blank (0.1% HOAc) and ATP (10−5 M in 0.1% HOAc). In the upper trace the analytical systemconsisted of an SS injector, 50�m I.D. silica tubing and the DAD system. In the lower trace the SSC is mounted as path of the solvent path before the DAD system.Right traces: The UV profiles between 230 and 300 nm of the according ATP injections. In the upper trace (no SSC in the flow path) the typical ATP UV profile isapparent. The lower UV-profile (SSC in the flow path) the typical ATP UV profile is not seen.

hydrolysis of the available fatty acid esters gave rise to phos-pholipids in the sample with only one instead of two fatty acidresidues. The shape of the extracted ion chromatograms of thesecompounds resembled a square wave form as anticipated fora non-retained compound and minimal recoveries were foundafter elution with mobile phase C. Further testing (cf. INJ andSSC effects, reference) made clear that the hydrophobic interac-tions were predominant and as such these concise experimentsmade deductions whether adsorption to SS surfaces based onthe presence of a phosphate group occurred or not, impossible.It is not inconceivable that the different effects (hydrophobicityand interactions with SS) reinforced each other. In view of thisall no further experiments were performed.

As target compound for a phosphorylated sugar glyc-erophospate was chosen. The associated [M-H]− trace atm/z171 mimicked perfectly the unretained compound trace (cf. raf-finose). This agreed well with the observations made for thenucleoside monophosphates, which also had no retention.

3.2. Trapping capacity of the SSC

The above experiments showed interactions with SS surfacesof the LC–ESI–MS hardware to be responsible for the (partial)loss of (multi) phosphorylated compounds. In this process theelectrospray SS capillary was an important contributor. It wasassumed that the physical state of the inner wall of the capillaryw orei ble)a mass ct ana SSCo sed

in different LC–ESI–MS studies in our laboratory was evalu-ated. The analytical system used for this purpose consisted of aninjector, silica tubing and the aforementioned SSC. MS and/orDAD were used to monitor the compounds.

To study the trapping capacity solely associated with the SSCof the ESI probe ATP, adenosine and blank solutions were usedas test samples. In all experiments injections of ATP or adenosinewere always preceded by a number of 20�L injections of 50%NH4OH (H2O) (E) and a blank injection, the latter being 0.1%HOAc (H2O), mobile phase D. If necessary additional injectionsof mobile phase E were made to reduce the signal of the adsorbedcompounds to back ground level.

In a first experimental setup ca. 200 pmol ATP (in 0.1%HOAc) was injected twice via a SS valve directly connectedto the DAD system via a fused silica line. The elution of ATPwas monitored at 258 nm and identified by its UV-absorptionprofile (Fig. 9, upper half). A different result was obtained whenthe same ATP sample was injected on the system in which theSSC was mounted between injector and DAD system: none ofthe ATP reached the detector (Fig. 9, lower half). As both sys-tems only differed in the presence/absence of the SSC, it wasconcluded that the SSC had a considerable trapping capacity forATP.

In a third experiment the solvent path consisted of the SSinjection valve, silica tubing, the DAD system and the SSC. Thelatter was now mounted in the ESI probe. Identical injectionsw MSs asd[ gedt gousc

ould be an important factor in this phenomenon. Furthermt felt expedient to establish additional proof for the (possictual adsorption properties of the SSC without using apectrometer to bypass any unforeseen instrumental effet the same time excluding all electrical effects. To do so anf an ESI probe which was, intensively (and trouble free) u

,

sd

ere made and monitored via both the DAD and the ESI–ystems. Following its injection in mobile phase D ATP wetected by DAD, yet no signal was observed by ESI(−)-MS at

M-H] − (m/z 506) unless the solvent composition was chano mobile phase E. If adenosine was injected under analoonditions no adsorption was observed.

218 R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221

Fig. 10. The monitoring of INJ + SSC after injection of a crude tryptic digest of�-casein. In the upper trace the extracted ion chromatogram corresponding to theisotopical envelope of the [M-3H]3− signal of Tppept is plotted. In the lower tracethe according TIC is given. The large increment at ca. 18.5 min coincides thearrival of the mobile phase C eluent front. The accumulated spectrum associatedwith this front is also given. The spectrum is dominated by the triply chargedTppept atm/z 1039, whereas the other signals were attributable to other multi-phosphorylated peptides originating from�S1-and�S2-caseins.

3.3. The adsorption process and the analysis of realsamples

To gain a deeper appreciation of the possible implicationsof these findings for “real” samples we injected a crude tryp-tic digest of the phosphorylated protein�-casein. This trypticdigest would contain Mppept and Tppept. Prior to ESI–MS anal-ysis the sample was diluted to ca. 2.5 pmol/�L in 0.1% (v/v)HOAc (50:50 MeOH/H2O). The above described INJ + SSC sol-vent sequence (cf. Section3.1) was used as an evaluation tool.

Soon after injection a huge increment in the TIC trace wasseen at the moment the injection plug reached the detecto(Fig. 10). When the mobile phase was changed to mobile phasB the total ion current (TIC) showed some more incrementsdue to both change in solvent composition and the release osome hydrophobic species. After an additional 4 min no morecompounds eluted until mobile phase C was applied. Taken thprevious results into consideration the isotopic envelopes of thdouble and triple charged states of the respective Mppept anTppept were extracted and plotted. For the Mppept (m/z 1029.4)the trace corresponded well with that of a non retained compound (not shown). However, the trace of Tppept (m/z 1039.4)was more interesting: following injection a considerable reten-tion was observed with a slight increment when changing to100% ACN (B). A major release of Tppept was seen whenmobile phase C was applied. The accumulated spectrum wad

1039.4) ions of Tppept confirming the selectivity of the availableSS hardware towards this tetraphosphorylated species. How-ever, from the spectrum it was evident that besides Tppept someother compounds were trapped on the LC–ESI–MS hardware.Inspection of them/z values and comparison with literature[22] clarified that all other signals corresponded with multi-phosphorylated peptides following tryptic digestion of�S1-and�S2-caseins. For example the doubly charged peptide signalat m/z 962.3 corresponds with DIGpSEpSTEDQAMEDIK, adiphosphorylated tryptic peptide of�S1-casein. This implies theused�-casein was cross-contaminated with�-caseins, thoughmore important it demonstrates once more the selectivity of the(corroded) stainless steel surfaces towards multiphosphorylatedcompounds.

4. Discussion

The observation of irreversible compound losses due to SShardware is not new: already in the 1980s Sadek et al. demon-strated irreversible protein losses due to the presence of stainlesssteel surfaces and especially to stainless steel column frits[4].Just recently, Cosman et al. also demonstrated adsorption ofcomplete�-casein proteins to SS surfaces[23].

From our semi-quantitative data it was obvious that the result-ing losses were significant if compared to values generallyaccepted for memory effects in instruments (typically carry overv

ntialp at thee pho-r o theb bar-r preadt

rfacet eadyi sur-f e ofm flowr tp ad-d tableq ofsi e ESIp obilep thee r-e lreadyo

ners ed byeu illaryw e inners faceb howed

ominated by the doubly (m/z 1559.6) and triply charged (m/z

re

f

eed

-

s

alues for injectors range between 0.02 and 0.2%).Furthermore, our preliminary data indicated that an esse

art of the detector can bias the results: The observation thlectrospray capillary contributes to the removal of phosylated compounds from the solvent stream has not yet, test of our knowledge, been reported. This is especially emassing since electrospray mass spectrometry is a widesechniques in (phospho)peptide studies.

In this process the degree of corrosion of the inner suhe SSC seemed to be an important contributor. Mowery alrllustrated that even passivated “inert” stainless steel 316aces could become “active” and corroded under influenc

any reverse phase blends especially in the high linearegions (valves, connectors, capillary tubing)[1]. As a resulitting and the formation of metal oxide layers occurred. Had and Foley reported that under acidic conditions detecuantities of Fe(III) were observed following the corrosiontainless steel parts in the solvent path (mainly frits)[24]. Theres no reason to believe that the inner surface of the SSC of throbe would not be susceptible to this phenomenon of mhase corrosion, which will add to the corrosion inherent tolectrospray process itself[10]. It has to be noted that “diffences” in appearance for SS nanospray emitters were abserved after 48 h of use[25].

To illustrate the detrimental impact of corrosion the inurfaces of a used and a brand new SSC were examinlectron microscopy. As can be seen inFig. 11 the tip of thesed capillary was heavily corroded whereas the new capas undamaged. The same observations were made for thurface close to the tip. Away from the spraying tip the surecame more alike to the surface seen in a new SSC. This s

R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221 219

Fig. 11. Electron microscopy images of a used SSC (left) and a new SSC (right). Images (a) and (b) show the outlook of the tips of the respective SSCs. It isclearthe used one (a) is heavily corroded even at the outside. Images (c) and (d) show lengthwise cross-sections (of parts) of both capillaries near the tip.The new one isobviously not entirely smooth but already displays small lobes (d). The inner surface near the tip showed an heavily damaged inner layer (c). When a cross-sectionof the used SSC further away from the spraying tip was evaluated its appearance was more alike the new SSC inner surface (not shown). Image (e) displays the innersurface at the very end of the used SSC: it is obviously heavily corroded and an increase in inner surface is apparent.

220 R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221

that corrosion and/or metal consumption in the ESI probe wasmost pronounced in the “electrospray area”.

The corrosion products likely to occur are metal(hydr)oxides[1,3]. In soil science it was well established that inorganic phos-phates adsorb well to iron (hydr)oxides[26–29]. A maximumsorption of phosphates was observed at low pH and the sorp-tion decreased at higher pH[29]. In the same paper it wasstated that sorption was rapid at pH < 8.5; above this value thesorption became kinetically slow. Furthermore, ligand exchangebetween the (inorganic) phosphate and the surface hydroxylgroups of the oxide surfaces will – according these studies –lead to the (pH dependent) bidentate binding of the phosphateby the metal ion (chemisorption). Not by coincidence this biden-tate formation is believed to be the underlying mechanism inthe purification of phosphorylated analytes by Fe(III)-IMAC[30–32]. This information side by side with the here presenteddata and the electron microscopy images suggested that SSsurfaces and especially activated(/corroded) SS surfaces canchemisorb phosphorylated analytes, most probably via mecha-nisms also governing Fe(III)-IMAC. However, our observationsconcerning the monophosphorylated analytes do not fit thismodel completely. So additional effects as the possibility of mul-tiple interactions, pH conditions, interaction times, etc. might beinvolved. It is worthwhile mentioning that Fe(III)-complexationwith phospholipids in HPLC analyses was also reported[33].

n op esto icati eref ongla erabls illaryp

nalys eptw inR rgeq I(M hi f theM

acide ac itivei ievep idet iviraa RPE tesW si-t

. Yet itiono patioo ). In

turn this process prevents the adsorption of the phosphorylatedanalytes.

Kim et al. came recently to a similar conclusion for phospho-peptides, only they suggested silanophilic interactions betweenthe free silanol groups of both capillary tubing and C18 station-ary phase and the phosphopeptides. The addition of phosphoricacid was believed to suppress these interactions[39]. Our currentstudy cannot contradict this conclusion, however in one of our“silica free” experiments (only PEEK tubing) retention couldonly be explained by interaction with the SS surface.

These few literature examples indicate that interactions oreven reactions (cf. bidentate formation) of multiply phospho-rylated analytes with LC and MS hardware might occur morefrequently and should be taken into account when analysing suchcompounds. In addition to this it might be necessary to reassesscritically earlier (quantitative) data concerning phosphorylatedcompounds, especially when low concentrated and/or minutesample amounts were involved. On the other hand ample adoptedhardware devices are already available which would preventthe described phenomena: e.g. the usage of silica ESI emittersand the aforementioned biocompatible LC hardware. The elec-tropolishing of stainless steel electrospray needles is reported toadd to its corrosion resistance resulting in more robust electro-spray emitters[42,43]. Alternatively derivatisation of the ana-lytes prior to the actual analysis might make them less prone toadsorption phenomena[44]. When considering the LC–ESI–MSc res-s erestb nd/orh ta.

5

avail-a t ina ana-l boths steelc pro-c facei com-p ssibleb reatb des,o hos-p lablep pped.C weret n onS

phe-n s thatL ory-l hesee com-p ts ofs

Despite the fact this phenomena of retention/adsorptiohosphorylated compounds on SS surfaces were – to the bur knowledge – not yet clearly described, some data ind

ng that other scientists were coping with similar effects wound. Shi et al. reported that nucleoside triphosphates strdsorbed on SS surfaces and that peak tailing was consid

ess when using a 3.5 cm instead of a 20 cm SSC[34]. They alsotated that after replacing the SSC with a fused silica capeak tailing was suppressed.

Several authors reported that for the RPLC–ESI–MS ais of a tryptic digest of�-casein only the above used Mppas observed[19,35] and that detection of the TppeptPLC–ESI–MS was only possible when significantly lauantities were injected or when alkaline conditions and ES−)-S detection were applied[36–38]. Acidic conditions led bot

n positive as in negative ion mode only to the detection oppept[38].Recently it was also shown the addition of phosphoric

nhanced the detection of both Tppept and Mppept from�-asein tryptic digest when using RPLC–ESI–MS in the poson mode[39]. St. Claire added ammonium phosphate to acherformant ion-pair RPLC (IPRP)-ESI–MS for a nucleos

riphosphate and the (tri)phosphorylated metabolite of antgent[40]. We experienced the same for the simultaneous IPSI(−)-MS analysis of nucleoside mono-, di- and triphosphahen adding NH4H2PO4 to the mobile phase both MS sen

ivity as chromatographic performance were enhanced[41].At the time we had no explanation for these observations

he current study supports the following hypothesis: the addf phosphates to the mobile phase leads to a complete occuf the active sites on the SS surfaces (valve/tubing/SSC

fof-

yly

-

l-:

t

n

ompatibility of mobile phases not only minimising ion suppion and background increments should be a point of intut also suppression of (phospho)compound adsorption aardware corrosion. This would enhance the validity of da

. Conclusion

This study demonstrates that the stainless steel surfacesble in LC–ESI–MS hardware can play a principal pardsorption and even trapping of (multi) phosphorylated

ytes under acidic conditions. We have demonstrated thattainless steel valves as well as the commonly stainlessapillary used in ESI sources can contribute to this “trappingess”. The latter implies that the electrospray LC–MS intertself can severely bias the analysis of phosphorylatedounds. The release of the trapped compounds was only poy implying harsh basic conditions. Some compounds of giochemical and biological significance including nucleotiligonucleotides (with a terminal phosphate group) and phopeptides are certainly affected. The number of avaihosphate moieties correlates with the aptitude to be traorrosion by LC mobile phases and the ESI process itself

he main contributors for a process of possible chemisorptioS hardware.It was not intended to produce an in depth study of this

omenon. Our objective was to add to the (re)awarenesC–ESI–MS data generated from the analysis of phosph

ated compounds should be interpreted in the light of txperiments; it is essential to realize that some importantounds may be “lost” during the analysis of minute amounample.

R. Tuytten et al. / J. Chromatogr. A 1104 (2006) 209–221 221

Acknowledgements

We would like to thank Jan Van Daele, Ludo Rossou and J.Eysermans for acquiring the microscopic images of the electro-spray capillaries and resolving the technical problems involved.We thank Professor Timmermans and Professor G. Van Tendeloofor the opportunity to use their electron microscopy equipment.

This research was funded by a Ph.D. grant of the Institutefor the Promotion of Innovation through Science and Technol-ogy in Flanders (IWT Vlaanderen). The MS equipment wasfunded by the Flemish Foundation for Scientific Research (FWOVlaanderen), the Flemish Government (GOA action) and theUniversity of Antwerp (RAFO).

References

[1] R.A. Mowery, J. Chromatogr. Sci 23 (1985) 22.[2] K.E. Collins, C.H. Collins, C.A. Bertran, LC-GC North Am. 18 (2000)

688.[3] K.E. Collins, C.H. Collins, C.A. Bertran, LC-GC North Am. 18 (2000)

600.[4] P.C. Sadek, P.W. Carr, L.D. Bowers, L.C. Haddad, Anal. Biochem. 144

(1985) 128.[5] G.J. Van Berkel, F.M. Zhou, Anal. Chem. 67 (1995) 2916.[6] N.B. Cech, C.G. Enke, Mass Spectrom. Rev. 20 (2001) 362.[7] M.G. Ikonomou, A.T. Blades, P. Kebarle, Anal. Chem. 62 (1990) 957.[8] J.F. de la Mora, G.J. Van Berkel, C.G. Enke, R.B. Cole, M. Martinez-

Sanchez, J.B. Fenn, J. Mass Spectrom. 35 (2000) 939.

[ trom

[ ass

[ .P.Masrom-

[ .L.

[[ 600.[[

[18] S. Li, C. Dass, Eur. Mass. Spectrom. 5 (1999) 279.[19] L.M. Nuwaysir, J.T. Stults, J. Am. Soc. Mass Spectrom. 4 (1993)

662.[20] J.D. Watts, M. Affolter, D.L. Krebs, R.L. Wange, L.E. Samelson, R.

Aebersold, J. Biol. Chem. 269 (1994) 29520.[21] A. Stensballe, O.N. Jensen, Rapid Commun. Mass Spectrom. 18 (2004)

1721.[22] W. Zhou, B.A. Merrick, M.G. Khaledi, K.B. Tomer, J. Am. Soc. Mass

Spectrom. 11 (2000) 273.[23] N.P. Cosman, K. Fatih, S.G. Roscoe, J. Electroanal. Chem. 574 (2005)

261.[24] P.R. Haddad, R.C.L. Foley, J. Chromatogr. 407 (1987) 133.[25] B.A. Sinnaeve, J.F. Van Bocxlaer, J. Chromatogr. A 1058 (2004) 113.[26] T. Hiemstra, W.H. Van Riemsdijk, J. Colloid Interface Sci. 210 (1999)

182.[27] M.A. Butkus, D. Grasso, Environ. Geol. 40 (2001) 446.[28] G. Lefevre, Adv. Colloid Interface Sci. 107 (2004) 109.[29] Y.M. Tzou, M.K. Wang, R.H. Loeppert, Arch. Environ. Contam. Toxicol.

44 (2003) 445.[30] S.A. Lopatin, V.P. Varlamov, Appl. Biochem. Microbiol. 31 (1995)

221.[31] E. Sulkowski, Bioessays 10 (1989) 170.[32] V. Gaberc-Porekar, V. Menart, J. Biochem. Biophys. Methods 49 (2001)

335.[33] Y.H. Wang, I.S. Krull, C. Liu, J.D. Orr, Biomed. Chromatogr. 17 (2003)

149.[34] G. Shi, J.T. Wu, Y. Li, R. Geleziunas, K. Gallagher, T. Emm, T. Olah,

S. Unger, Rapid Commun. Mass Spectrom. 16 (2002) 1092.[35] M. Allen, J. Anacleto, R. Bonner, P. Bonnici, B. Shushan, L. Nuwaysir,

Rapid Commun. Mass Spectrom. 11 (1997) 325.[36] S. Visser, C.J. Slangen, F.M. Lagerwerf, W.D. Vandongen, J. Haverkamp,

[[ .G.

. 15

[ 8.[[ apid

[ 002)

[ Mass

[ G.

[9] G.J. Van Berkel, J. Anal. At. Spectrom. 13 (1998) 603.10] G.J. Van Berkel, K.G. Asano, P.D. Schnier, J. Am. Soc. Mass Spec

12 (2001) 853.11] G.J. Van Berkel, V. Kertesz, M.J. Ford, M.C. Granger, J. Am. Soc. M

Spectrom. 15 (2004) 1755.12] R. Tuytten, F. Lemiere, E. Witters, W. Van Dongen, H. Slegers, R

Newton, E. Esmans, in: Proceedings of the 53rd Conference onSpectrometry and Allied Topics. American Society for Mass Spectetry, San Antonio, TX, 2005. paper ThP 250.

13] R. Tuytten, F. Lemiere, W. Van Dongen, H. Slegers, R.P. Newton, EEsmans, J. Chromatogr. B 809 (2004) 189.

14] P. Cao, J.T. Stults, J. Chromatogr. A 853 (1999) 225.15] P. Cao, J.T. Stults, Rapid Commun. Mass Spectrom. 14 (2000) 116] F.Y. Lin, W.Y. Chen, S.Y. Huang, Bioprocess Eng. 23 (2000) 467.17] S. Li, C. Dass, Anal. Biochem. 270 (1999) 9.

.

s

J. Chromatogr. A 711 (1995) 141.37] L. Leadbeater, F.B. Ward, J. Chromatogr. 397 (1987) 435.38] A. Beck, M. Deeg, K. Moeschel, E.K. Schmidt, E.D. Schleicher, W

Voelter, H.U. Haring, R. Lehmann, Rapid Commun. Mass Spectrom(2001) 2324.

39] J. Kim, D.G. Camp, R.D. Smith, J. Mass Spectrom. 39 (2004) 2040] R.L. St Claire, Rapid Commun. Mass Spectrom. 14 (2000) 1625.41] R. Tuytten, F. Lemiere, W. Van Dongen, E.L. Esmans, H. Slegers, R

Commun. Mass Spectrom. 16 (2002) 1205.42] A. Subramanian, S. Mohan, T. Vasudevan, Bull. Electrochem. 18 (2

97.43] Y. Ishihama, H. Katayama, N. Asakawa, Y. Oda, Rapid Commun.

Spectrom. 16 (2002) 913.44] A. Nordstrom, P. Tarkowski, D. Tarkowska, K. Dolezal, C. Astot,

Sandberg, T. Moritz, Anal. Chem. 76 (2004) 2869.