RESEARCH PAPER

Development and characterization of a small electromembraneextraction probe coupled with mass spectrometry for real-timeand online monitoring of in vitro drug metabolism

Helene Bonkerud Dugstad & Nickolaj Jacob Petersen &

Henrik Jensen & Charlotte Gabel-Jensen &

Steen Honoré Hansen & Stig Pedersen-Bjergaard

Received: 3 June 2013 /Revised: 11 September 2013 /Accepted: 16 September 2013# Springer-Verlag Berlin Heidelberg 2013

Abstract A small and very simple electromembrane extrac-tion probe (EME-probe) was developed and coupled directlyto electrospray ionization mass spectrometry (ESI-MS), andthis system was used to monitor in real time in vitro metabo-lism by rat liver microsomes of drug substances from a smallreaction (incubation) chamber (37 °C). The drug-related sub-stances were continuously extracted from the 1.0 mL meta-bolic reactionmixture and into the EME-probe by an electricalpotential of 2.5 V. The extraction probe consisted of a 1-mmlong and 350-μm ID thin supported liquid membrane(SLM) of 2-nitrophenyl octyl ether. The drugs and formedmetabolites where extracted through the SLM and directlyinto a 3 μL min−1 flow of 60 mM HCOOH inside the probeserving as the acceptor solution. The acceptor solution wasdirected into the ESI-MS-system, and the MS continuouslymonitored the drug-related substances extracted by the EME-probe. The extraction efficiency of the EME-probe was de-pendant on the applied electrical potential and the length of theSLM, and these parameters as well as the volume of the

reaction chamber were set to the values mentioned above toavoid serious depletion from the reaction chamber (soft ex-traction). Soft extraction was mandatory in order not to affectthe reaction kinetics by sample composition changes inducedby the EME-probe. The EME-probe/MS-system was used toestablish kinetic profiles for the in vitro metabolism ofpromethazine, amitriptyline and imipramine as modelsubstances.

Keywords Electromembrane extraction probe .Massspectrometry . Drugmetabolism . Real-timemeasurement

Introduction

Recently, electromembrane extraction (EME) was proposedas a new principle for sample preparation in chromatography,electrophoresis, and mass spectrometry [1]. In EME, chargedanalytes are extracted from the sample, through a thin sup-ported liquid membrane (SLM) immobilized in the pores inthe wall of a porous hollow fiber and into an acceptor solutionpresent inside the lumen of the hollow fiber. The driving forcefor the mass transfer is an electrical potential sustained overthe SLM by an external power supply. For cationic analytes,the negative electrode is located in the acceptor solution, andthe positive electrode is located in the sample. For EME ofanionic analytes, the electrical potential is reversed, and thepositive electrode is placed in the acceptor solution.

In EME, the selectivity of the extraction can easily becontrolled by the external power supply. At low voltages,many substances are discriminated by the organic SLM, andmainly non-polar and singly charged compounds are extracted.

Published in the topical collection Challenges and New Directions inAnalytical Sample Preparation with guest editors Astrid Gjelstad andStig Pedersen-Bjergaard.

H. B. Dugstad :N. J. Petersen (*) :H. Jensen : C. Gabel-Jensen :S. H. Hansen : S. Pedersen-BjergaardSchool of Pharmaceutical Sciences, Faculty of Health and MedicalSciences, University of Copenhagen, Universitetsparken 2,2100 Copenhagen, Denmarke-mail: nickolaj.petersen@sund.ku.dk

H. B. Dugstad : S. Pedersen-BjergaardSchool of Pharmacy, University of Oslo, P.O. Box 1068 Blindern,0316 Oslo, Norway

Anal Bioanal ChemDOI 10.1007/s00216-013-7378-z

However, as the voltage is increased, more polar and multiplecharged species can penetrate the SLM, and the voltage can beused very effectively to control the selectivity of the system[2]. In addition, the selectivity can be controlled by the chem-ical properties of the organic solvent used as the SLM. Non-polar analytes can be extracted by relatively non-polar SLMs,like 2-nitrophenyl octyl ether [3] or 1-octanol [4], whereasaddition of an ion-pair reagent is required for EME of morepolar substances [5]. Up to date, the selection of the SLM hasmainly been performed based on trial and error, but a morescientific approach to this is expected in the near future.

In EME, target analytes are typically extracted from samplevolumes of 100–2.000 and into 10–25 μL acceptor solution.Because of this volume difference and because of the highefficiency of EME, the technique can be used for pre-concentration of target analytes. Thus, pre-concentration fac-tors up to 190 have been reported in the literature [6]. Inaddition, EME has been demonstrated to give efficient sampleclean-up from biological fluids [7–10] and environmentalsamples [3, 4, 11]. The efficient sample clean-up is based onthe fact that most matrix substances cannot penetrate theorganic SLM and are prevented from entering the acceptorsolution. For each sample, a small piece of porous hollowfiber is used, and the consumption of organic solvent for theSLM is at the low microliter level. Thus, EME is a low-costapproach to sample preparation. Up to date, EME has mostlybeen used for extraction of pharmaceuticals from biologicalfluids and of pollutants from environmental samples, andseveral reviews have been presented [12, 13].

Most EME up to date has been accomplished as an off-lineprocess with the acceptor solution located inside a poroushollow fiber. Thus, the SLM has been immobilized in thepores in the wall of a porous hollow fiber. After EME, theacceptor solution has been collected by a syringe or a pipetteand transferred to an auto-sampler for chromatographic orelectrophoretic analysis. Very recently, the EME principlewas transferred to a micro-chip format, where the SLM wasimmobilized in the pores of a flat membrane [14–16]. Withthis on-chip EME system, the sample solution was pumpedinto a small channel in the micro-chip. From this channel,analytes were extracted across the flat SLM and into anotherchannel with the acceptor solution. This acceptor solution waspumped through the channel and directed to a mass spectrom-eter (MS). Thus, an online system was developed where themass spectrometer continuously and in real-time measuredtarget compounds extracted over the SLM. With this onlinesystem, in vitro metabolic reactions were monitored to inves-tigate the metabolism of drug substances by rat liver micro-somes [15, 16].

On-chip EME was the first approach to continuouselectromembrane extraction. The system was highly efficientbecause the SLM was very thin (25 μm) and because theanalyte diffusion distances within the chip were short,

typically less than 50 μm. Thus, recoveries were high, andexcellent sample clean-up was achieved due to the presence ofthe SLM. However, the SLMwas located inside a micro-chip,and regeneration of the SLM or replacement of the polymericmembrane support was difficult. Also, construction of thehome-built micro-chip was relatively complicated, whichmay limit the possibility for other researchers to performrelated experiments. To circumvent these disadvantages, avery simple EME-probe was developed for the first time inthe present work. The EME-probe represented a new technicalapproach to continuous electromembrane extraction and de-viated from the on-chip approach in several ways. Due tosimplicity, the new EME-probe can easily be constructedand reproduced by other researchers. Also, with very easyaccess to the SLM and to the polymeric membrane support,maintenance and replacement of the membrane is straightfor-ward. The EME-probe was coupled to electrospray ionizationmass spectrometry (ESI-MS), the system was optimized andcharacterized, and the applicability was demonstrated for thestudy of in vitro metabolism of several drug substances.

Experimental

Chemicals and solutions

Amitriptyline hydrochloride, promethazine hydrochloride,and imipramine hydrochloride were all obtained fromSigma-Aldrich (St. Louis, MO, USA). 2-Nitrophenyl octylether (NPOE) was obtained from Fluka (Buchs, Switzerland).Rat liver microsomes (male Sprague–Dawley, pooled) 20 mg/mL was obtained from BD Biosciences (San Jose, CA, USA),and β-nicotinamide adenine dinucleotide 2′-phosphate re-duced tetrasodium salt hydrate (NADPH) was obtained fromSigma-Aldrich.

Stock solutions containing 1 mg/mL of each of amitripty-line, promethazine, and imipramine was prepared in 10% (v/v)ethanol and stored protected from light at 4 °C. Pseudo-reaction mixtures containing the different compounds wereprepared by dilution of the stock solutions with 50 mM phos-phate buffer pH 7.4. Formic acid of 60 mM was used as theacceptor solution. For the metabolism experiment, 35 μMstock solutions of promethazine, amitriptyline, and imipra-mine, 10 mMNADPH, 100 mMMgCl2, and 1.0 M potassiumphosphate buffer (pH 7.4) were prepared individually in de-ionized water.

Principle of the electromembrane extraction probe(EME-probe)

The basic principle of the entire system is illustrated in Fig. 1.A 3-μL min−1 flow of acceptor solution was pumped by amicro-syringe pump through a short length of steel capillary,

H.B. Dugstad et al.

which comprised the first part of the EME-probe. From the endof this steel capillary, the acceptor solution entered a smallpiece of porous hollow fiber made of polypropylene. Prior to ametabolic experiment, a small volume of NPOE wasimmobilized by capillary forces in the pores in the wall of thishollow fiber piece. The immobilization was accomplished byapplying a small droplet of NPOE onto the porous fiber,thereby filling the pores of the membrane by capillary forces,and excess NPOE was removed with a medical wipe. Theimmobilized NPOE served as a SLM in the wall of the hollowfiber. This SLM was placed into a reaction chamber for themetabolic reaction. The volume of the reaction chamber was 1,000 μL, the reaction chamber was temperature controlled at37 °C, and in the reaction chamber, different drug substanceswere metabolized in vitro in phosphate buffer pH 7.4 by ratliver microsomes (RLM). During the metabolic experiment,the drug substance and some of the metabolites were continu-ously extracted by the EME-probe, into the SLM and furtherinto the acceptor solution inside the piece of hollow fiber. Thedriving force for the extraction was an electrical potentialsustained over the SLM. Thus, in this study, where focus wasdirected towards basic drug substances, an anode was locatedin the reaction mixture, and a cathode was coupled to the steelcapillary which was in electrically contact with the acceptorsolution. The anode and the cathode were coupled to a powersupply which enabled careful tuning and control of the elec-trical potential (extraction voltage). The outlet of the hollowfiber piece was connected to a fused-silica capillary, and thiscapillary was directed towards the mass spectrometer coupledwith electrospray ionization. Thus, parent drugs and drugmetabolites extracted into the acceptor phase were continuous-ly pumped into the mass spectrometer. In this way, with theEME-probe coupled directly to ESI-MS, the system enabledreal-time and online measurement of the metabolic reaction.

Construction of the EME-probe

The acceptor solution was delivered by a micro-syringe pump(KDS-100-CE, kdScientific, Holliston, MA, USA) operatedwith a 1,000 μL gastight #1001 syringe (Hamilton, Bonaduz,Switzerland). The probe inlet was a steel capillary of 120 μminternal diameter and 260 μm outer diameter (nebulizer nee-dle, part # G1946-20177, Agilent technologies). The end ofthe steel capillary was coupled to a 2-cm piece of poroushollow fiber (Plasmaphan P1LX polypropylene hollow fibre,330 μm internal diameter, 150 μmwall thickness, and 0.4 μmpore size purchased from Membrana, Wuppertal, Germany).The steel tubing was inserted 1 cm into the hollow fiber, andthis intersection was heated with a lighter to seal the connec-tion. The heat made a seal tight connection to the steel capil-lary and as well closed the pores in the part of the fiberexposed to the heat. Since the pores were closed, this part ofthe steel capillary was electrically insulated when inserted intothe reaction mixture. The probe outlet was a 7-cm piece offused-silica capillary (internal diameter of 100 μm and outerdiameter of 375 μm, Polymicro Technologies, Phoenix, AZ,USA). The fused-silica capillary was inserted into the hollowfiber, and the intersection was heated to seal the connection. A1–5-mm space was made between the steel capillary and thefused-silica capillary, and this space with the surroundinghollow fiber served as the support for the SLM.

Since the SLM was highly flexible, the steel capillary andthe fused silica capillary could be kept in parallel, therebyforming a probe with the SLM at the bottom (Fig. 1). TheEME-probewas dipped into a 1.5-mLEppendorf vial, and thisserved as the reaction chamber. The reaction chamber wastemperature controlled to 37 °C by circulating water aroundthe vial with an in-house build device. The reaction mixtureinside the reaction chamber was stirred using a magnetic

Fig. 1 Basic principle of theEME-probe/MS-system

Electromembrane extraction probe coupled directly with mass spectrometry

stirring bar (Spinvane®, Sigma-Aldrich). The extraction volt-age was delivered by a variable DC power supply (ISO-TechIPS-603 power supply, RS components Ltd, Northlands, UK).The cathode was coupled directly to the steel capillary inlet ofthe EME-probe, while the anode (platinum wire) was locatedin the reaction chamber (Fig. 1).

Mass spectrometry (MS)

To allow stable electrospray ionization the 100 μm ID and375 μm OD fused silica capillary was drawn in a flame,thereby forming a sharp pointed spray tip required for stableelectrospray ionization. After pulling the spray tip in theflame, the tip was slightly etched (5 min) in 48% hydrofluoricacid (HF). Extreme caution should be taken using HF as anetchant since it can cause severe chemical burns and the fumesare toxic as well. The etch rate of the fused silica was approx-imately 2 μmmin−1, and the etching slightly expanded the IDat the spray tip, thereby minimizing the counter pressure in thefinal setup. A 90° curvature was afterwards made in thecapillary by bending the capillary in a flame. The bendingallowed the probe to be dipped into the reactionmixture, whilekeeping the ESI tip horizontal towards the MS inlet.

The probe consisting of the EME probe with attached ESIspray tip was mounted on an X, Y, Z stage in front of the MSinlet. Stable spray was obtained with flow rates down to 1 μL/min and with an ESI voltage of −1.5 kV applied at the MSinlet. The MS used was a Bruker-Esquire LC-MS ion trap(model G1979A, Bruker Daltronics Inc., Billerica, MA).

Operation of EME-probe/MS-system

Prior to each metabolic experiment, 50 μL of 100 mMMgCl2,100 μL 1.0 M potassium phosphate buffer (pH 7.4), and600 μL deionized water was delivered to the reaction chamber.Stirring at 800 rpm and the circulating water (37 °C) was turnedon, and the system was allowed to temperature equilibrate. TheEME-probe (1–5-mm piece of hollow fiber) was wetted withNPOE for a few seconds, and excess of NPOE was removedwith a paper tissue. The EME-probe was then dipped into thereaction chamber and coupled to the power supply. Thepumping of acceptor solution was turned on, and the extractionvoltage was turned on. The MS background signal for the drugsubstance was allowed to stabilize for a few minutes, before100 μL of 35 μM drug substance was added to the reactionchamber. The MS signal for the drug substance was allowed tostabilize, before 50μL of 20mgmL−1 rat liver microsomes wasadded to the reaction chamber. After another fewminutes, whenthe MS signal for the drug substance had stabilized, 100 μL of10 mM NADPH was added as co-factor, and the metabolicreaction was initiated. During the next 30–60 min, the MS wasoperated in the full scan mode to monitor both the parent drugand potential metabolites extracted by the EME-probe.

Capillary electrophoresis

Capillary electrophoresis (CE) was performed during the ini-tial characterization of the EME-probe and during the optimi-zation of operational parameters. For these experiments, theacceptor solution was collected after certain time intervals atthe outlet of the EME-probe and subsequently analyzed byCE. The equipment used was an Agilent Technologies HP3D

CE instrument (Agilent Technologies, Waldbronn, Germany)equipped with a UV-detector operated at 200 nm. The runningbuffer was 25 mM sodium dihydrogen phosphate adjusted topH 2.7 with ortho-phosphoric acid. Separations wereperformed at 20 kV in a 75 μm-I.D. fused-silica capillary(TSP075375, Polymicro Technologies, Phoenix, AZ, USA)with an effective length of 24.5 cm.

Results and discussion

Figure 2 illustrates the type of information acquired with theEME-probe/MS-system, with the anti-depressant drug ami-triptyline as model substance. In the first 3 min of the exper-iment, only sodium phosphate buffer and MgCl2 were presentin the reaction chamber, and no signal was measured foramitriptyline at m/z 278. Subsequently, amitriptyline wasadded to the reaction chamber, and the mass spectrometricsignal at m/z 278 increased significantly. After another 2 min,when the MS-signal was stabilized, rat-liver microsomes(enzymes) were added to the reaction chamber, and the signalfor amitriptyline decreased to some extent. Since 50 μL ofRLMwas added to the 850 μL reaction mixture, a decrease of6 % was expected due to the dilution factor of the drug. Theobserved decrease was much higher than expected from thedilution, and the decrease was attributed to partial binding ofamitriptyline to the proteins of the rat-liver microsomes.

After another 5 min, NADPHwas added as co-factor to thereaction chamber, and this initiated the metabolic reaction.The metabolic reaction was followed by the MS in full-scanmode (m/z 100–300) with an approximately delay of 30 s,equivalent to sum of the transfer time across the SLM and thetransfer time in the capillary to the MS. As seen from Fig. 2,the concentration of amitriptyline (m/z 278) in the reactionchamber decreased as function of time, and a correspondingincrease in the concentration was observed for the metaboliteshydroxy-amitriptyline (m/z 294) and nortriptyline (m/z 264).The principal function of the EME-probe was to avoid buffersalts and proteins to enter the mass spectrometer, which maycause serious ion-suppression and contamination [16]. TheSLM effectively discriminated the later type of components,and only relatively non-polar and positively charged sub-stances passed across the SLM and entered the mass spec-trometer. Since the SLM prevented sodium and potassiumfrom the incubation mixture to pass the membrane, no metal

H.B. Dugstad et al.

adduct ions of the drugs and metabolites was observed in theMS spectra.

Characterization of the EME-probe and optimizationof operational parameters

In different series of experiments, principal operational param-eters were investigated to obtain a fundamental understandingon how to optimize the EME-probe for studying metabolism.The idea was to use the EME-probe for studying in vitro drugmetabolism over time (30 min) from small reaction volumes. Inorder not to disturb or affect the metabolic reactions by theextraction, the investigation was focused on soft extraction,where only a small amount of drug substances was removedby extraction as compared with the rate of metabolism. Theexperiments were accomplished with the reaction chamberfilled with a solution of the basic drug substance amitriptylinedissolved in phosphate buffer pH 7.4 (pseudo-reactionmixture).No rat-liver microsomes were present, and therefore no metab-olism occurred during the investigation of the principal opera-tional parameters. The solution present in the reaction chamberwas continuously stirred during the experiments. Based onearlier experiences [14–16], NPOE was used for the SLM,and 60 mM HCOOH was selected as the acceptor solution.

In a first set of experiments, the effect of extraction voltagewas investigated. The voltage was varied between 0 and 60 V.The results are illustrated in Fig. 3a, where the amount of drugleft in the reaction chamber was plotted as function of extractiontime with different voltages (0, 2.5, 10, and 15 V). Someextraction was observed even at 0 V. This was due to the pH-gradient across the SLM since the reaction chamber wasmaintained at physiological pH (7.4) and 60 mM formic acid

was used as the acceptor solution on the backside of the SLM tobe compatible with the ESI. At pH 7.4, some of the drugmolecules were unionized, and they were extracted due topassive diffusion across the SLM. With acidic conditions inthe acceptor solution, the drugmolecules were ionized andwereprevented from re-entering the SLM. As the voltage across theSLM was applied, and gradually increased above 2.5 V, theextraction efficiency of the EME-probe increased, until itreached a maximum value at voltages in the range 15–25 V.Above this, there was no further gain in extraction efficiency.This voltage dependency was in accordance with recent andrelated EME work (on-chip EME) [15]. These experimentsdemonstrated a very important feature of the EME-probe,namely that the extraction performance to a large extent wascontrolled by the applied voltage, and this was easily adjustedon the power supply. With a voltage of 2.5 V, the EME-probeprovided soft extraction, and about 11 % of the investigateddrugwas removed from the 1mL reaction chamber after 30minof continuous extraction. Soft extraction at this level was con-sidered as acceptable for the metabolic reactions studied below.As seen from Fig. 3a, the extraction performance of the EME-probe was similar at open circuit and 2.5 V. However, thetransfer time for the drug substance across the SLMwas shorterusing 2.5V.With open circuit, there was no electro-kinetic masstransfer contribution across the SLM, and slow transfer only bypassive diffusion increased the time to pass the membranesubstantially. Slow transfer across the SLM was undesirable ifthe system should be able to monitor fast concentration changesinduced by the metabolism, and all remaining experiments inthis work were performedwith 2.5 Vapplied to the EME-probe.With this voltage, the EME-probe/MS-system responded quick-ly to compositional changes in the reaction chamber.

Fig. 2 In vitro metabolic profiles for amitriptyline using the EME-probe/MS-system. Extraction voltage, 2.5 V; length of SLM, 1 mm; volume ofthe reaction mixture, 1,000 μL; initial concentration of amitriptyline,3.5 μM. Amitriptyline (AT) was measured at m/z 278, the metabolitesnortriptyline (NT) at m/z 264, and hydroxy-amitriptyline (AT-OH) at m/z294. At a , amitriptyline was added, at b , rat liver microsomes were

added, and at c NADPH was added. The inserted MS specter shows thesignal for amitriptyline and the formed metabolites at t =14 min. Due tothe low S/N caused by the soft extraction, the MS specter has beenbackground-subtracted with the MS signal only having buffer in theincubation chamber (t=1.7 min)

Electromembrane extraction probe coupled directly with mass spectrometry

In a second set of experiments, the length of the SLM wasinvestigated. The length was varied between 1 and 5 mm asillustrated in Fig. 3b. As seen from the figure, the extractionefficiency increased proportionally to the length of the SLM.This behavior was in accordance with recent theoreticalmodels developed in general for EME [17], where the extrac-tion efficiency was found to be dependant of the surface areaof the SLM. These experiments clearly demonstrated that thelength of the SLM is an additional and principal parameter forcontrolling the efficiency of the EME-probe. In order tooperate the EME-probe as soft as possible, to remove onlyminimal amount of the drug substance from the reactionchamber, a 1 mm SLM was used during the rest of this study.

In a third set of experiments, the effect of the volume of thereaction mixture was investigated. The volume was variedbetween 500 and 1,000 μL. As illustrated in Fig. 3c, the

depletion in concentration from the reservoir decreased asthe volume of the reaction mixture increased from 500 to 1,000 μL. This was expected, and the observation was inaccordance with recent theoretical models developed in gen-eral for EME [17]. These experiments demonstrated that theperformance of the EME-probe was also controlled by thevolume of the reaction mixture. In order to operate the EME-probe as soft as possible, a reaction mixture volume of1,000 μL was used during the rest of this study.

In a final set of optimization experiments, the effect of theacceptor solution flow rate was investigated. The flow ratewas varied between 1 and 15 μL min−1, and the extract wasanalyzed both by CE and by direct coupling to MS. Theextraction efficiency (amount of compound extracted per unittime) was found to be independent of the acceptor flow rate.Higher enrichment factors were achieved the lower the

Fig. 3 Studies of operationalparameters for the EME-probe/MS-system. a Extraction voltagewas varied between 0 and 15 V;length of SLM, 1 mm; volumeof the reaction mixture, 1,000 μL.b Length of SLM was variedbetween 1 and 5 mm; extractionvoltage, 2.5 V; volume of reactionmixture, 1,000 μL. c Volume ofthe reaction mixture was variedbetween 500 and 1,000 μL;extraction voltage, 2.5 V; lengthof SLM, 1 mm

H.B. Dugstad et al.

accepter flow rate. With a high flow rate, the transport timefrom the EME-probe and to the mass spectrometer was short,giving fast response times. On the other hand, with a high flowrate, the back-pressure on the SLM increased. At too-highflow rates, the SLMwas pressed out of the pores of the hollowfiber, and the MS signal became suppressed by the samplematrix. Therefore, as a compromise 3 μL min−1 was selectedfor the remaining experiments.

Metabolic profiles for amitriptyline

A metabolic profile obtained for amitriptyline was presentedin Fig. 2. In this case, the EME-probe was operated with a1 mm SLM and at 2.5 V to avoid depletion from the reactionmixture. Because amitriptyline metabolized rapidly, the ex-periment was completed in 22 min. From Fig. 2, the half timefor amitriptyline was estimated from the slope of the concen-tration profile for amitriptyline at m/z 278. Based on threereplicate experiments, the half time for amitriptyline wasestimated to 1.6 min with a RSD of 32 %. This value was in

agreement with previous reported data on an EME-chip sys-tem [16]. The profiles shown in Fig. 2 also detected themetabolites hydroxy-amitriptyline (m/z 294) and nortriptyline(m/z 264). Nortriptyline exhibited maximum concentration inthe reaction chamber 2.3 min after initialization of the reac-tion, whereas the maximum for hydroxyl-amitriptyline wasobserved after 3.8 min. The subsequent decrease of the signalsfor the metabolites was most probably due to further enzy-matic degradation. More polar metabolites were not observedin the present system due to the hydrophobic nature of theNPOE SLM.

In a separate experiment, the linearity of the EME-probe/MS-system was checked for amitriptyline. The purpose of thisexperimentwas to verify the reliability of themeasured signals.The concentration in the reaction mixture was varied between0.1 and 2.5 μM, and the MS-signal of the extract increasedlinearly with increasing concentration (r2=0.9947). The limitof quantitation for amitriptyline in the extraction system wasestimated to be 96 nM and the limit of detection (LOD) 29 nM.The LOD is a compromise with the requirements for the soft

Fig. 4 In vitro metabolic profilesfor promethazine and imipramineusing the EME-probe/MS-system. Extraction voltage, 2.5 V;length of SLM, 3 mm; volume ofthe reaction mixture, 1,000 μL;initial concentration of drugsubstance, 3.5 μM. At a , drugsubstance was added; at b rat livermicrosomes were added, and at cNADPH was added. Inserted MSspectra have been background-subtracted for the MS signal onlyhaving buffer in the incubationchamber (t =0.7 min)

Electromembrane extraction probe coupled directly with mass spectrometry

extraction. With this given sensitivity and starting with a drugconcentration of 3.5 μM, it should be possible to monitor themetabolism for approximately seven halftimes.

Metabolic profiles for other drug substances

The EME-probe/MS-system was also tested for promethazineand imipramine. All experiments were conducted with a startconcentration of 3.5 μM for the drug substance. The resultsare summarized in Fig. 4. As in the amitriptyline experiment,the half-life of promethazine was estimated from the slope ofthe semi-log profile to 4.4 min (Fig. 4). Promethazine isknown to be extensively metabolized by cytochrome P450sin human liver microsomes. The reactions observed are S-oxidation, aromatic oxidation, and N -demethylation resultingin metabolites that might be detected by MS at m/z 301, 301,and 271, respectively [18] In the EME-probe/MS-system, thedisappearance of promethazine was followed by appearanceof a metabolite with m/z 301 corresponding to either S-oxidation or aromatic hydroxylation of promethazine. Theoxidized metabolite increased in concentration during the first10 min and reached maximum at 30 min. This concentrationwas observed throughout the remaining incubation time indi-cating that the EME-probe/MS-system was stable for longerexperiments with respect to unwanted depletion of analytesdue to extensive extraction from the metabolic reactionmixture.

As for amitriptyline and promethazine, imipramine clear-ance followed first-order kinetics with a half-life of 4.8 min(Fig. 4). The known major imipramine metabolites formedby cytochrome P450 N -demethylation or hydroxylation werenot detected. These preliminary experiments indicate that thesystem is suited for metabolism studies of basic and fairlyapolar drug compounds. Further validation on the quantita-tive performance of the system will reveal the prospects forrapid online determination of in vitro clearance and metabo-lism in particular with respect to extraction efficiency ofmore hydrophilic metabolites through the hydrophobicNPOE SLM.

When microsomes were added to the incubation chamber,the online monitoring revealed an instant decrease in bothimipramine and promethazine intensity of 33 % and 54 %,respectively. This decrease in intensity was larger than theanticipated decrease due to the inherent dilution that occurredwhen the microsomes were added. This was tentativelyinterpreted to be due to microsomal protein binding. Otherstudies on imipramine microsomal protein binding, measuredby different conditions, reported a higher degree of proteinbinding [19]. The potential use of the EME-probe/MS-systemfor protein binding studies will be the subject of further studiesas it may exclude some of the challenges in common usedmethods for protein binding, for example, volume shiftscaused by osmotic flow during dialysis experiments.

Conclusions

The present work has developed a very simplemicro-scale probefor continuous electromembrane extraction (EME-probe) andcoupled this EME-probe directly to electrospray ionizationMS. Also, the present work demonstrated the application of thisEME-probe/MS-system for real-time in vitro metabolic profilingof drug substances. Thus, the system enabled direct and onlinemass spectrometric measurement of compositional changes in areaction chamber, where drug substances were gradually metab-olized by enzymes from rat liver microsomes. The EME-probeeffectively served to control the transfer of drug-related sub-stances from the reaction chamber and into the mass spectrom-eter. Also, the EME-probe isolated the drug-related substancesfrom buffer salts and proteins to avoid ion-suppression andserious contamination of the mass spectrometer. The extractionefficiency of the EME-probe was easily controlled by the mag-nitude of the extraction voltage and by the length of the support-ed liquid membrane. “Soft” extractions were performed in thepresent work in order not to induce analyte depletion in theincubation chamber that would otherwise influence the observedreaction kinetics. The EME-probe/MS-system showed a greatpotential for in vitro drugmetabolism studies in the present workbut has to be further validated and compared against acceptedstandard methods before implemented for real applications. Thisis in progress, along with experiments to identify other applica-tion areas of the technology as well.

Acknowledgments Bin Li andDr.María D. Ramos Payán are acknowl-edged for their technical help during this project.

References

1. Pedersen-Bjergaard S, Rasmussen KE (2006) Electrokinetic migra-tion across artificial liquid membranes—new concept for rapid sam-ple preparation of biological fluids. J Chromatogr 1109:183–190

2. Domínguez NC, Gjelstad A, Nadal AM, Jensen H, Petersen NJ,Honoré Hansen S, Rasmussen KE, Pedersen-Bjergaard S (2012)Selective electromembrane extraction at low voltages based on ana-lyte polarity and charge. J Chromatogr A 1248:48–54

3. Seidi S, Yamini Y, Heydari A, Moradi M, Esrafili A, Rezazadeh M(2011) Determination of thebaine in water samples, biological fluids,poppy capsule, and narcotic drugs, using electromembrane extractionfollowed by high-performance liquid chromatography analysis. AnalChim Acta 701:181–188

4. Payán MR, Bello López MÁ, Torres RF, Navarro MV, Mochón MC(2011) Electromembrane extraction (EME) and HPLC determinationof non-steroidal anti-inflammatory drugs (NSAIDs) in wastewatersamples. Talanta 85:394–399

5. Gjelstad A, Rasmussen KE, Pedersen-Bjergaard S (2006) Electroki-netic migration across artificial liquid membranes—tuning the mem-brane chemistry to different types of drug substances. J ChromatogrA 1124:29–34

6. EskandariM, Yamini Y, Fotouhi L, Seidi S (2011)Microextraction ofmebendazole across supported liquid membranes forced by pH gra-dient and electrical field. J Pharm Biomed Anal 54:1173–1179

H.B. Dugstad et al.

7. Strieglerová L, Kubáň P, Boček P (2011) Electromembrane extrac-tion of aminoacids from body fluids followed by capillary electro-phoresis with capacitively coupled contactless conductivity detec-tion. J Chromatogr A 1218:6248–6255

8. Rezazadeh M, Yamini Y, Seidi S (2011) Electromembrane extractionof trace amounts of naltrexone and nalmefene from untreated biolog-ical fluids. J Chromatogr B 879:1143–1148

9. Hosseiny Davarani SS, Najarian AM, Nojavan S, Tabatabei M-A(2012) Electromembrane extraction combined with gas chromatog-raphy for quantification of tricyclic antidepressants in human bodyfluid. Anal Chim Acta 725:51–56

10. Basheer C, Tan SH, Lee HK (2008) Extraction of lead ions byelectromembrane isolation. J Chromatogr A 1213:14–18

11. Kiplagat IK, Oanh Doan TK, Kubáň P, Boček P (2011) Trace deter-mination of perchlorate using electromembrane extraction and capil-lary electrophoresis with capacitively coupled contactless conductiv-ity detection. Electrophoresis 32:3008–3015

12. Bello-LópezMÁ, Ramos-PayánM, Ocana-González JA, Fernández-Torres R, Callejón-Mochón M (2012) Analytical applications ofhollow fiber liquid phase microextraction (HF-LPME): a review.Anal Lett 45:804–830

13. Gjelstad A, Pedersen-Bjergaard S (2011) Electromembrane extrac-tion: a new technique for accelerating bioanalytical sample prepara-tion. Bioanalysis 3:787–797

14. Petersen NJ, Jensen H, Honoré Hansen S, Taule Foss S, SnakenborgD, Pedersen-Bjergaard S (2010) On-chip electro membrane extrac-tion. Microfluid Nanofluid 9:881–888

15. Petersen NJ, Taule Foss S, Jensen H, Honoré Hansen S, Skonberg C,Snakenborg D, Kutter JP, Pedersen-Bjergaard S (2011) On-chipelectro membrane extraction with online ultraviolet and mass spec-trometric detection. Anal Chem 83:44–51

16. Petersen NJ, Sønderby Pedersen J, Nørgård Poulsen N, Jensen H,Skonberg C, Honoré Hansen S, Pedersen-Bjergaard S (2012) On-chip electromembrane extraction for monitoring drug metabolism inreal time by electrospray ionization mass spectrometry. Analyst 137:3321–3327

17. Seip KF, Jensen H, Sønsteby MH, Gjelstad A, Pedersen-Bjergaard S(2013) Electromembrane extraction: distribution or electrophoresis?Electrophoresis 34:792–799

18. Nakamura K, Yokoi T, Inoue K, Shimada N, Ohashi N, Kume T,Kamataki T (1996) CYP2D6 is the principal cytochrome P450responsible for the metabolism of the histamine H-1 antagonistpromethazine in human liver microsomes. Pharmacogenetics 6:449–457

19. Obach RS (1997) Nonspecific binding to microsomes: Impact onscale-up of in vitro intrinsic clearance to hepatic clearance as assessedthrough examination of warfarin, imipramine, and propranolol. DrugMetab Dispos 25:1359–1369

Electromembrane extraction probe coupled directly with mass spectrometry