Article pubs.acs.org/ac
Real Time Extraction Kinetics of Electro Membrane Extraction Verified by Comparing Drug Metabolism Profiles Obtained from a Flow−Flow Electro Membrane Extraction-Mass Spectrometry System with LC−MS David Fuchs, Henrik Jensen, Stig Pedersen-Bjergaard, Charlotte Gabel-Jensen, Steen Honoré Hansen, and Nickolaj Jacob Petersen* Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark ABSTRACT: A simple to construct and operate, “dip-in” electromembrane extraction (EME) probe directly coupled to electrospray ionization-mass spectrometry (ESI-MS) for rapid extraction and real time analysis of various analytes was developed. The setup demonstrated that EME-MS can be used as a viable alternative to conventional protein precipitation followed by liquid chromatography−mass spectrometry (LC−MS) for studying drug metabolism. Comparison of EME-MS with LC−MS for drug metabolism analysis demonstrated for the first time that real time extraction of analytes by EME is possible. Metabolism kinetics were investigated for three different drugs: amitriptyline, promethazine, and methadone. By comparing the EME-MS extraction profiles of the drug substances and formed drug metabolites with the metabolism profiles obtained by conventional protein precipitation followed by LC−MS good correlation was obtained with only very limited time delay in the extraction. The results indicate that, by tuning the electromembrane properties, for example, by optimizing the extraction voltage, extremely fast extraction kinetics can be obtained. A metabolic profile could be generated while the drug was metabolized offering a significant time saving as compared to conventional LC−MS where laborious protein precipitation or other sample pretreatments are required before analysis. This makes the developed EME-MS setup a highly promising sample preparation method for various kinds of applications where fast and real-time analysis of analytes is of interest.
S
Not at least due to its potentially fast extraction kinetics, EME has recently been proposed as a valuable method for realtime analysis of formation of drug metabolites by coupling the EME probe to ESI-MS. In its first demonstration, ESI-MS was directly coupled to an EME microchip extracting drugs and their formed metabolites from a metabolic reaction mixture.11 In another approach, a simpler probe was prepared using a porous hollow fiber instead of a chip for immobilization of the membrane. In this setup, the hollow fiber was coupled to a capillary connecting the EME-probe to the MS.12 These setups were both able to monitor in-line the formation of drug metabolites but suffered from some drawbacks. In the chip setup, the SLM was located inside the microchip making regeneration or replacement of the SLM after usage difficult. Further, the fabrication process of the chip was rather timeconsuming. In the more simple hollow fiber setup, the extraction probe was placed directly into the metabolic reaction mixture which led to rapid sample depletion when higher voltages were applied. A sufficiently high electric potential,
upported liquid membranes (SLMs) used as a highly selective sample preparation method have been explored for several years as a promising alternative to conventional sample pretreatment methods such as solid phase extraction (SPE), liquid−liquid extraction (LLE), or protein precipitation. In SLM extraction samples are cleaned up by extraction through an organic liquid membrane, immobilized in the pores of a polymeric support, and into an acceptor phase on the other side of the membrane. Extraction using SLMs is primarily based on passive diffusion across the membrane making it a rather slow extraction process not suitable for applications where fast analysis of compounds is of interest.1−4 To accelerate the extraction process, electromembrane extraction (EME) can be utilized. In EME an electric potential is applied across the SLM making charged analytes migrate toward an oppositely charged electrode in the acceptor solution. The major driving force is thereby no longer passive diffusion but an electric potential difference across the membrane making the extraction process potentially faster and more efficient.5 Since its first introduction, EME has been successfully applied for efficient cleanup of a wide range of analytes from complex matrixes including waste- and drinking water, blood plasma, and urine.6−10 © XXXX American Chemical Society
Received: March 13, 2015 Accepted: April 28, 2015
A
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry however, was predicted to be crucial for fast extraction kinetics and is therefore necessary if fast analysis of compounds is of interest.13 Although the setups were able to extract and analyze drug metabolites, both setups could not prove that the drug metabolites were extracted in real time. In the current work a new EME probe for in-line extraction and analysis of drug compounds was developed. It utilizes a flow−flow system by having the extraction probe placed inside a tube slowly draining the sample solution and at the same time having the acceptor solution being pumped through the inside of the probe and delivered toward the ESI-MS. The current probe configuration is, even at high electric potentials and long analysis times, not prone to sample depletion as it extracts only from the drawn sample. With the developed probe setup, EME kinetics of various analytes were for the first time directly investigated. This was achieved by direct analysis of extracted drugs and its metabolites formed in a metabolic reaction by MS and comparing the obtained extraction profiles with conventional protein precipitation followed by LC−MS.
Figure 1. Schematic drawing of the EME probe setup.
■
MATERIALS AND METHODS Chemicals and Sample Solutions. Amitriptyline hydrochloride and promethazine hydrochloride, carbamazepine, nitrophenyl octyl ether (NPOE), and reduced β-nicotinamide adenine dinucleotide 2′-phosphate tetrasodium salt hydrate (NADPH) were obtained from Sigma-Aldrich (St. Louis, MO). Methadone hydrochloride was obtained from Nordisk Droge og Kemikalie A/S (Copenhagen, Denmark). Rat liver microsomes (RLM, male Sprague−Dawley, pooled; 20 mg/mL) were obtained from BD Biosciences (San Jose, CA). Stock solutions containing 1 mg/mL each of amitriptyline hydrochloride, promethazine hydrochloride, and methadone hydrochloride, respectively, were prepared in 10% (v/v) ethanol and stored at 4 °C. The 100 mM of potassium phosphate buffer, 10 mM of NADPH, and 10 mM of formic acid were prepared in deionized water. Working solutions of 10 μM or 100 μM of each drug compound were prepared by adequate dilutions from stock solutions in 100 mM potassium phosphate/5 mM MgCl2 (pH 7.4). Working solutions of 5 μM carbamazepine used as internal standard for LC−MS were prepared in LC−MS grade acetonitrile (ACN) and stored at 4 °C. EME Probe Preparation and Operation. The principal setup of the EME probe is illustrated in Figure 1. For suction of sample solution, a silicone tubing having an 760 μm i.d. and an 2.5 mm o.d. (SCP Science, Québec, Canada) was coupled to a microsyringe pump (KDS-100-CE, kdScientific, Holliston, MA) operated at a suction of 10 μL/min with a 5 mL Gastight no. 1005 syringe (Hamilton, Bonaduz, Switzerland). The silicone tube was interrupted by a 1 cm long piece of stainless steel tubing (1.2 mm o.d.) and connected to a platinum electrode placed inside the sample flow tube (76 μm diameter, SigmaAldrich) by twisting the wire around the stainless steel tubing. The platinum electrode provided uniform electric contact close to the membrane independent of the conductivity of the sample solution or on potential bubbles located in the drain tube. Electric potential was applied using a HVS448, High Voltage Sequencer (Labsmith, Livermore, CA). Usage of a silicone tube for sample flow had the advantage that the fused silica capillary used for the acceptor phase flow could be leak tight when punctured through the tube. To avoid possible sample absorption by the silicone tube the part of the tube in
which the SLM was located was replaced by a polytetrafluoroethylene (PTFE) i.d. 750 μm tubing (Mikrolab Aarhus, Denmark). The EME probe was located inside the PTFE tubing 2−4 mm from the opening. A hollow fiber made of porous polypropylene (Plasmaphan P1LX, purchased from Membrana, Wuppertal, Germany) was used as support for the liquid membrane. The fiber had a 330 μm i.d., 150 μm wall thickness, 0.4 μm pore size. It was on both ends coupled to fused silica capillaries (Polymicro Technologies, 100 μm i.d., 245 μm o.d.) by heat shrinking (at approximately 200 °C) of the intersection of fiber and capillary leaving approximately 3 mm of intact porous hollow fiber in between the two capillaries. The heat closes the pores and shrinks the diameter of the fiber and thereby provides seal tight coupling of the fiber to the silica capillaries. To allow a compact probe design that could be dipped into narrow sample vials, the fused silica capillary connecting the syringe pump and the SLM was sharply bent in a flame. As the flame removed the external polyimide coating of the capillary (making the fused silica more prone to break), a small piece of heat shrinkable PTFA tubing (Zeus, Orangeburg, SC) was applied to substitute the removed polyimide coating. This was done by slipping the PTFA tubing over the capillary and shrinking it by heating it up to approximately 340 °C. The applied heat shrinkable tubing gave the capillary sufficient robustness and flexibility to avoid breaking the capillary. The two fused silica capillaries interrupted by the porous hollow fiber were on one end coupled to a syringe pump (operated with a 1 mL Hamilton gastight no. 1001 syringe at a flow rate of 10 μL/min) providing continuous flow of acceptor solution and on the other end to the interface of an ESI-MS instrument (Agilent 1100 series LC/MSD ion trap, G2445, with a standard HPLC−ESI interface, G1945A, Agilent Technologies, Santa Clara, CA). Prior to an experiment, a small volume of NPOE was immobilized by capillary forces in the pores of the hollow fiber. This was done by applying a small droplet of NPOE onto the porous hollow fiber. Successful immobilization of the membrane was visually inspected as the appearance of the hollow fiber changed from white to transparent. Excess of NPOE was removed from the fiber by a medical wipe, and the membrane was slid into the PTFE part of the sample flow tube, approximately 2−4 mm away from the sample inlet. The end of the capillary that coupled the probe to B
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
was initiated by addition of 100 μL of 10 mM of NADPH. Formation and extraction of drug metabolites was in-line monitored by ESI-MS operated in full scan mode. Protein Precipitation Procedure. A volume of 20 μL of sample taken at various time points of each metabolic reaction were immediately resuspended in ice cold ACN (spiked with 5 μM of carbamazepine as internal standard for the HPLC−MS runs) in a 1:1 ratio (equal amounts of sample and ACN). Solutions were precipitated at 4 °C for 15 min and afterward centrifuged for 10 min at 18 000g. Supernatants were carefully transferred into high recovery polypropylene vials (250 μL sample vials, Agilent Technologies) and analyzed by LC−MS. LC−MS Method. The chromatographic separation was performed on a reversed phase column (Phenomenex, Torrance, CA, Kinetex XB C18 100 mm × 2.1 mm, 2.6 μm particle size) using a Thermo Scientific (Waltham, MA) Accela HPLC system coupled to a Finnigan (Waltham, MA) TSQ Quantum Triple Quadrupole LC−MS. Separation was performed with gradient elution using 5% ACN/0.1% HCOOH as mobile phase A and 95% ACN/0.1% HCOOH as mobile phase B at a flow rate of 500 μL/min. A linear gradient from 0% to 70% of mobile phase B was applied for 3.5 min followed by re-equilibration with 100% mobile phase A for 3 min. The column was operated at 40 °C and the samples were kept at 4 °C in the autosampler. Detection of analytes was performed in positive ionization mode and with single ion monitoring (SIM) for the parent drug, the known metabolites of the drug as well as of carbamazepine (used as an internal standard). The relative signal intensities were calculated by dividing the peak areas of the analytes by the obtained peak areas of carbamazepine. Sample Depletion Testing. To confirm that the developed EME probe setup is not prone to sample depletion when extracting for longer time periods at a high voltage (+200 V), the probe was dipped in the reaction chamber filled with 10 μM of amitriptyline in 100 mM potassium phosphate/5 mM MgCl2 (pH 7.4). The probe was coupled to the acceptor phase pump and the sample suction pump both set to a flow rate of 10 μL/min. An electric potential of +200 V was applied and signal intensity was monitored by ESI-MS for 35 min. Production of Metabolites for Evaluating Extraction Speed. A volume of 200 μL of 100 μM of drug substance and 1400 μL of 100 mM phosphate buffer/5 mM MgCl2 (pH 7.4) was added to the reaction chamber and allowed to reach a temperature of 37 °C. After approximately 5 min, 200 μL of 20 mg/mL of RLM and 200 μL of 10 mM of NADPH were added. The metabolism reaction was stopped after 5 min (for amitriptyline) by resuspending the solution in equal volumes of ice cold ACN. The solution was precipitated at 4 °C for 15 min and afterward centrifuged for 10 min at 18 000g. The supernatant was transferred to a new test vial and ACN was evaporated in a nitrogen evaporator set to 50 °C and with constant blow down of nitrogen onto the sample. After approximately 15 min, the sample was removed from the evaporator, filled up to a final volume of 2000 μL with deionized water, and stored at 4 °C until use. Investigation of Voltage Dependency on Extraction Kinetics. The EME probe was dipped into the reaction chamber and coupled to the acceptor phase pump and the sample suction pump both operated at 10 μL/min. Prior to use, the amitriptyline metabolite solution was spiked with 2.5 μM (final concentration) of amitriptyline. A volume of 450 μL of metabolite solution spiked with amitriptyline was added to the
the MS could easily be pierced leak tight through the silicone tube and was directly coupled to the ESI-MS interface. At the end of each working day, the membrane was cleaned by removing the NPOE from the hollow fiber. This was done by coupling a syringe filled with ethanol to the fused silica capillary and flushing the hollow fiber by applying sufficient pressure to wash out the NPOE. After removing of NPOE the probe was stored dry until use for another day of experiments. Applying the cleaning procedure, the same probe could easily be reused for several days of experiments. Reaction chamber for drug metabolism. As metabolism reaction chamber, the cell body of a Franz Cell Chamber with a water jacket for temperature control was used. The sampling port of the chamber was used as an EME probe inlet, and the water jacket was coupled to a water bath providing continuous water circulation at 37 °C. A photograph of the Franz Cell Chamber with inserted EME probe is shown in Figure 2. The
Figure 2. Photograph of a Franz Cell Chamber cell body with inserted EME-probe.
reaction chamber was kept open during the experiments to ensure free access to oxygen during the metabolic reactions and to allow easy addition of reagents. For continuous mixing of the solution, an approximately 8 mm long stir bar and a magnetic stirrer placed below the reaction chamber were used. Probe Setup. For EME-MS experiments, the EME probe was dipped into the metabolic reaction chamber as illustrated in Figure 2 and coupled to two syringe pumps. One for pumping the acceptor solution (10 mM of formic acid) through the inner part of the probe and one for continuous delivery of sample along the outside of the SLM probe by drawing of sample solution through the PTFE tube where the probe was located. Further the probe was coupled directly to ESI-MS for in-line analysis of extracted analytes. Syringe pumps for sample solution drawing and acceptor phase pumping were both set to 10 μL/min. An electric potential of +200 V was applied at the anode in the sample flow tube relative to the acceptor solution that was grounded. The applied voltage served as driving force for extraction of positively charged analytes from the sample into the acceptor solution. Drug Metabolism Experiments. A volume of 700 μL of 100 mM phosphate buffer/5 mM MgCl2 (pH 7.4) was first added to the reaction chamber and allowed to reach a temperature of 37 °C. After 5 min of operation, 100 μL of 100 μM of drug sample was added and the MS signal was allowed to stabilize. Next, 100 μL of 20 mg/mL of RLM was added. Signal was again allowed to stabilize and the metabolic reaction C
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
potential of +200 V a stable extraction signal could be obtained and as expected no sample depletion was observed. Theoretical Considerations for an Optimal Extraction Voltage. The extraction kinetics in EME has previously been modeled theoretically.14 An important parameter describing the extraction kinetics is the distribution coefficient of the extracted (charged) drug compound, Kd*:14
reaction chamber, and extraction kinetics of amitriptyline and dihydroxylated amitriptyline (Di-OH-amitriptyline) were monitored. After 5 min the sample solution was replaced by 100 mM phosphate buffer/5 mM MgCl2 (pH 7.4). The experiment was repeated applying extraction voltages of 0 V, +60 V, +200 V, and +400 V.
■
RESULTS AND DISCUSSION Quality Control of Probe Preparation Process and Interday Probe Stability Testing. As differences in the extraction kinetics across the membrane of different probes were observed, extraction performance testing of every newly prepared EME probe was done. This was performed by monitoring the extraction kinetics at +200 V when adding 500 μL of a 10 μM of amitriptyline solution to the reaction chamber. The time points after sample addition when a first increase in signal intensity and when 80% of the stable signal intensity was reached were monitored (ΔY0−80%). Only probes showing ΔY0−80% < 30 s were used for further metabolism experiments. Variations in extraction performance of different probes were most probably caused by slight variations in the production process where heat was used to attach the porous hollow fibers to the capillaries. Interday stability of the developed probe was tested by performing the same experiment on three consecutive days using the same probe applying the removal and regeneration procedure of the liquid membrane as described in the Materials and Methods section. The probe showed highly reproducible extraction profiles with the same response times and signal intensities and could therefore be easily reused for several days. Sample Depletion Testing. To demonstrate that the developed probe setup was, different to earlier dip probe EMEMS setups,12 not prone to any kind of sample depletion at higher extraction voltages, a sample solution of 10 μM of amitriptyline in 100 mM potassium phosphate/5 mM MgCl2 (pH 7.4) was filled into the reaction chamber and extraction was done as described in the Materials and Methods part. As shown in Figure 3, the signal intensity for amitriptyline immediately increased after addition of sample and remained at a constant level for the entire extraction time of 30 min. The experiment therefore confirmed that even at a high electric
⎛zF ⎞ Kd* = exp⎜ i (Δow ϕ − Δow ϕi0)⎟ ⎝ RT ⎠
(1)
Δwo ϕ is the Galvani potential difference across the sample solution−membrane interface, which is proportional to the total applied extraction potential difference. Δwo ϕ0i is a standard extraction potential related to how hydrophobic the extracted compound i is. zi is the charge on the compound i (+1 for the compounds in the present study), F is Faradays constant, R is the gas constant, and T is the absolute temperature. A high Kd* is required for effective extraction kinetics.14 According to eq 1, the optimal extraction voltage is thus dependent on the drug compound via Δwo ϕ0i ; hydrophobic compounds will require a relatively low extraction voltage for optimal extraction, whereas more hydrophilic compounds will require a comparatively higher extraction voltage. In the simple model discussed here, ion pairing effects in the membrane are not taken into consideration. Recent studies have indicated that such effects may be in play for some compounds.15,16 Ion pairing in the SLM will affect the extraction process as it will lead to compound accumulation in the membrane and a slower extraction kinetics since it will tend to reduce the apparent charge on the compound being extracted. In the worst case (100% ion paring in the membrane), the extraction will thus proceed by passive diffusion through the membrane which is much slower than normal EME. Investigation of Voltage Dependency on Extraction Kinetics. To evaluate the response time for the different generated metabolites, it was necessary to have stock solutions of the formed metabolites. Since the formed metabolites were either unknown or hard to purchase, the metabolites were generated in house according to the procedure described in the Materials and Methods section. The produced metabolites were subsequently used for investigating the voltage dependency on extraction kinetics. The final optimized extraction experiments were all performed using an electric potential of +200 V. This relatively high extraction voltage, compared to previous applications of EME,6,17,18 was necessary as extraction kinetics of some drug metabolites of amitriptyline were shown to be highly dependent on the applied electric potential. While the more hydrophobic parent drug even at low voltages was rapidly extracted, it was shown that for some more hydrophilic drug metabolites a higher electric potential was necessary in order to obtain fast extraction kinetics. This observation (illustrated in Figure 4) is in accordance with eq 1 as the more hydrophilic metabolites have a higher Δwo ϕ0i . In the experiment amitriptyline and one of its metabolites, Di-OH-amitriptyline were simultaneously extracted from 100 mM phosphate buffer/5 mM MgCl2 (pH 7.4) at different extraction voltages. The compounds showed as expected no (for Di-OH-amitriptyline) or very slow (for amitriptyline) extraction at 0 V. At an electric potential of +60 V, amitriptyline was already rapidly extracted from the solution while its more polar metabolite Di-OH-amitriptyline still
Figure 3. Monitoring of signal stability and possible depletion from the sample reservoir while extracting with a voltage of +200 V; sample phase pumping was initiated after 5 min. Sample: 1 mL of 10 μM of amitriptyline prepared in 100 mM potassium phosphate/5 mM MgCl2 (pH 7.4). Acceptor solution: 10 mM formic acid; acceptor and sample phase flow both operated at 10 μL/min. D
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 4. Influence of extraction voltage on the extraction kinetics. Parallel extraction of amitriptyline (blue) and Di-OH-amitriptyline (red) from 100 mM phosphate buffer/5 mM MgCl2 (pH 7.4) at varying extraction voltages; sample solution was exchanged with 100 mM phosphate buffer/5 mM MgCl2 (pH 7.4) after 5 min.
factor that was observed. This drop was most probably caused by partial protein binding of the drug and therefore less availability of the analyte for extraction. Signal was again allowed to stabilize for 5 min, and 100 μL of 10 mM of NADPH was added to initiate the metabolic reaction. Formation of the drug metabolites was in-line monitored as they were extracted and analyzed by ESI-MS. As signals for drug metabolites started to increase, signal intensities for parent drugs decreased indicating that they were converted into their metabolites. To compare the obtained metabolic profiles with conventional protein precipitation followed by LC−MS, samples of 20 μL were taken during the metabolic reaction at various time points. The metabolic reactions in the samples were stopped immediately by protein precipitation in 20 μL of ice cold ACN. After protein precipitation samples were analyzed by LC−MS and metabolic profiles of parent drugs and its metabolites were compared to metabolic profiles obtained in-line by EME-MS. As shown in Figures 5−7, EME-MS and LC−MS signals show very good overlays for all tested model drugs and their metabolites. The signal decrease after addition of RLM observed by EME-MS probably caused by protein binding of the drug was not observed by LC−MS. Protein denaturation most likely released the bound drugs leading to similar signal intensities before and after RLM addition. This finding confirms that the primary reason for the signal drop after addition of RLM was most likely protein binding of the drug. The time delay at an acceptor phase flow rate of 10 μL/min caused by the dead volume of the system (volume from the membrane to the ESI spray tip) was calculated to be 20 s. To account for the dead volume in the overlays, the signal obtained from EME-MS was offset with the time delay of 20 s. For promethazine, formation of its two main metabolites S-oxide promethazine and S-oxide OH-promethazine was monitored. Ion intensity plots obtained from EME-MS overlapped for both metabolites as well as for the parent drug very well with the ion intensity plots measured by LC−MS (Figure 5). Formation of
exhibited slow extraction kinetics. An electric potential of +200 V was necessary to reach comparable extraction kinetics for DiOH-amitriptyline and amitriptyline. Notably, the extraction rate of Di-OH-amitriptyline seemed to decrease again when an even higher electric potential of +400 V was applied. A possible reason could be ion pairing in the membrane which would lead to slower extraction kinetics. Although a higher extraction potential will lead to a better partitioning into the membrane, it may also potentially lead to counterions being extracted from the acceptor solution into the membrane. In the present example, the counterion is the formate ion (HCOO−) which is relatively hydrophilic and therefore only likely to partition into the membrane at relatively high extraction potentials. A similar tendency of acceptor solution counterions to lower extraction efficiency was recently observed in a related system.15 A similar set of experiments comparing the extraction kinetics at different voltages was also done with methadone and its main metabolite ethyl-dimethyl-diphenyl-pyrrolidine (EDDP). Both compounds however showed almost immediate extractions already at +60 V. Further increasing the electric potential had no effect on the extraction (data not shown). Comparison of EME-MS with LC−MS on Drug Metabolism Profiles. Metabolism experiments were performed according to the same protocol for all model drugs. The EME probe was dipped into the reaction chamber containing 700 μL of 100 mM potassium phosphate/5 mM MgCl2 (pH 7.4) and coupled to the acceptor phase pump, sample phase pump (both set to a flow rate of 10 μL/min), and ESI-MS. Extraction of analytes was initiated by applying an electric potential of +200 V. A volume of 100 μL of parent drug was added to the reaction chamber, and the signal of the parent drug started to increase as it was continuously extracted from the sample. Signal was allowed to stabilize for 5 min, and 100 μL of 20 mg/mL of RLM was added. A signal decrease of around 11% was expected due to dilution of the drug solution with 100 μL of RLM. For all drugs, however, a signal drop of the parent drug much higher than expected from the dilution E
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
obtained from LC−MS for both methadone and EDDP (Figure 6). Formation of ethyl-methyl-diphenyl-pyrroline (EMDP),
Figure 6. Comparison of metabolic profiles for methadone obtained using EME-MS and LC−MS: (A) metabolic profile of methadone using EME-MS; (B, C) overlay of EME-MS ion intensity plots of methadone (B) and its main metabolite EDDP (C) with LC−MS data. Relative ion intensities from LC−MS were obtained by dividing the ion intensities of analytes by ion intensities of carbamazepine (used as internal standard). The Y-axes of EME-MS and LC−MS were scaled to have peak maximums overlapped. To account for the dead volume of the EME-MS system (volume from membrane to ESI spray tip), the signal obtained from EME-MS was offset with the expected time delay of 20 s.
Figure 5. Comparison of metabolic profiles for promethazine obtained using EME-MS and LC−MS. (A) metabolic profile obtained using EME-MS ; (B−D) overlay of EME-MS ion intensity plots of promethazine (B) and its two main metabolites S-oxide promethazine (C) and S-oxide OH-promethazine (D) with LC−MS data. Relative ion intensities from LC−MS were obtained by dividing the ion intensities of analytes by ion intensities of carbamazepine (used as internal standard). The Y-axes of EME-MS and LC−MS were scaled to have peak maximums overlapped; to account for the dead volume of the EME-MS system (volume from membrane to ESI spray tip), the signal obtained from EME-MS was offset with the expected time delay of 20 s.
another metabolite of methadone20 was detected in small amounts by LC−MS but could not be detected using EME-MS. At the used drug concentration, EMDP was most probably formed in amounts below the LOD of the used MS. For amitriptyline, all four known main metabolites,21 OHamitriptyline, Di-OH-amitriptyline, nortriptyline, and OHnortriptyline were extracted (Figure 7). A short delay of around 30 s beyond the delay expected from the dead volume of the system was observed for all metabolites of amitriptyline by EME-MS compared to the data obtained by the LC−MS. Experiments on extraction kinetics using lower extraction voltages showed that this delay was even higher when applying lower extraction voltages (Figure 3). The results therefore indicate that especially for the more hydrophilic metabolites a relatively high extraction voltage is necessary to obtain fastest possible extraction kinetics. Further optimization of the EME
demethylated promethazine, an additional metabolite of promethazine,19 was detected neither by LC−MS nor by EME-MS. At the used concentration of the parent drug, demethylated promethazine was most probably formed at a too low concentration to be detected by the present analytical methods. In a next experiment metabolism of methadone into its main metabolite, EDDP was monitored. Similar to promethazine intensity plots from EME-MS correlated very well with data F
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 7. Comparison of metabolic profiles of amitriptyline obtained with EME-MS and LC−MS: (A) metabolic profile of amitriptyline using EMEMS, (B−F) overlay of ion intensity plots of amitriptyline and its main metabolites obtained from EME-MS vs LC−MS. Relative ion intensities from LC−MS were obtained by dividing the ion intensities of analytes by ion intensities of carbamazepine (used as internal standard): (B) amitriptyline, (C) OH-amitriptyline; (D) di-OH-amitriptyline, (E) nortriptyline, (F) OH-nortriptlyine. To account for the dead volume of the EME-MS system (volume from membrane to ESI spray tip), the signal obtained from EME-MS was plotted with the expected time delay of 20 s. The Y-axes of EMEMS of panels B−E were scaled to have base lines as well peak maximums overlapped with the data obtained by LC−MS. Ion suppression at m/z 280 (m/z ratio for OH-nortriptyline, panel F) was observed in the EME-MS trace after addition of the parent drug. The Y-axis for LC−MS was therefore not adjusted to the baseline signal prior to addition of RLM but to the background signal obtained before parent drug was added.
probe, for example, by using hollow fibers with thinner walls (and therefore thinner liquid membranes) might lead to even faster extraction kinetics.
It was found that some of the more polar metabolites require much higher extraction voltages than the parent drug compound in order to show real-time concentration changes. Previous studies with stationary EME probes where only end point measurements of extracted analytes were performed suggested that relatively low electrical potentials (+5 V to +50 V) are sufficient as the determined extraction recoveries reached a maximum at relatively low voltages.6,17,18 In the current study however it could be shown that for some compounds relatively high electric potentials (+200 V) are necessary in order to obtain fastest possible extraction kinetics. This is especially important for applications where real-time analysis of compounds is of interest. The current work demonstrated that after optimization of the applied electric potential the investigated compounds can be extracted from the solutions with no or only very little time delay. The experiments therefore revealed that extremely fast extractions can be obtained by EME and that the applied electric potential has a great impact on the extraction kinetics of some compounds.
■
CONCLUSION The developed EME probe is simple and straightforward to fabricate and can be assembled for an experiment within only a few minutes. As the hollow fiber is easily accessible by sliding it in and out of the suction tube, simple removal and regeneration of the liquid membrane is possible. The same probe can thereby be reused for several experiments. The design with the SLM located in a suction tube allows a compact probe design that can be dipped in sample vials even with a small opening of approximately 3 mm in diameter. Since only the sample drawn inside the tube is extracted, the composition (analyte concentrations) in the sample reservoir or metabolic reaction chamber is not altered by the extraction even at high extraction voltages. In the current work the probe was used for monitoring dynamic composition changes in a metabolic reaction chamber. G
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry With the newly developed probe it was for the first time demonstrated that real time extraction of analytes by EME is possible. By comparing the EME-MS extraction profiles of the drug metabolites with the metabolism profiles obtained from the same samples by conventional LC−MS after protein precipitation, it was demonstrated that EME-MS shows almost no time delay between formation and extraction of analytes. Compared to conventional protein precipitation followed by LC−MS, the EME setup allows extremely rapid analysis of metabolic profiles of drug substances. It is however not capable of analyzing different isomers of formed metabolites as only analytes with different m/z ratios are distinguishable by ESIMS. Furthermore, only semiquantitative quantification of extracted analytes is possible by comparing relative ion intensities. As all extracted compounds are analyzed in parallel by ESI-MS, the system is potentially prone to ion suppression making an absolute quantification very difficult. In addition ion suppression leads to a decrease in sensitivity limiting the detectability of low-abundance drug metabolites. To overcome these limitations, work is in progress to implement micro-LC or CE for additional separation preceding ESI-MS analysis of extracted compounds. Further, work with drug substances with varying physicochemical properties is in progress to further broaden the analytical potential of the developed system.
■
(16) Seip, K. F.; Jensen, H.; Kieu, T. E.; Gjelstad, A.; PedersenBjergaard, S. J. Chromatogr. A 2014, 1347, 1−7. (17) Balchen, M.; Gjelstad, A.; Rasmussen, K. E.; Pedersen-Bjergaard, S. J. Chromatogr. A 2007, 1152, 220−225. (18) Petersen, N. J.; Foss, S. T.; Jensen, H.; Hansen, S. H.; Skonberg, C.; Snakenborg, D.; Kutter, J. P.; Pedersen-Bjergaard, S. Anal. Chem. 2011, 83, 44−51. (19) Katsunori Nakamura, T. Y.; Kazuaki, I.; Noriaki, S.; Noriko, O.; Toshiyuki, K.; Tetsuya, K. Pharmacogenetics 1996, 6, 449−457. (20) Kelly, T.; Doble, P.; Dawson, M. J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 2005, 814, 315−323. (21) Rousu, T.; Herttuainen, J.; Tolonen, A. Rapid Commun. Mass Spectrom. 2010, 24, 939−957.
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge ARIADME, a European FP7 ITN Community’s Seventh Framework Program under Grant Agreement No. 607517.
■
REFERENCES
(1) Audunsson, G. Anal. Chem. 1986, 58, 2714−2723. (2) Jönsson, J. Å.; Mathiasson, L. Trends Anal. Chem. 1992, 11. (3) Jönsson, J. Å.; Mathiasson, L. Trends Anal. Chem. 1999, 18. (4) Jönsson, J. Å.; Mathiasson, L. Trends Anal. Chem. 1999, 18. (5) Pedersen-Bjergaard, S.; Rasmussen, K. E. J. Chromatogr. A 2006, 1109, 183−190. (6) Dominguez, N. C.; Gjelstad, A.; Nadal, A. M.; Jensen, H.; Petersen, N. J.; Hansen, S. H.; Rasmussen, K. E.; Pedersen-Bjergaard, S. J. Chromatogr. A 2012, 1248, 48−54. (7) Payan, M. R.; Lopez, M. A.; Torres, R. F.; Navarro, M. V.; Mochon, M. C. Talanta 2011, 85, 394−399. (8) Basheer, C.; Tan, S. H.; Lee, H. K. J. Chromatogr. A 2008, 1213, 14−18. (9) Strieglerova, L.; Kuban, P.; Bocek, P. J. Chromatogr. A 2011, 1218, 6248−6255. (10) Kuban, P.; Bocek, P. Anal. Chim. Acta 2014, 848, 43−50. (11) Petersen, N. J.; Pedersen, J. S.; Poulsen, N. N.; Jensen, H.; Skonberg, C.; Hansen, S. H.; Pedersen-Bjergaard, S. Analyst 2012, 137, 3321−3327. (12) Dugstad, H. B.; Petersen, N. J.; Jensen, H.; Gabel-Jensen, C.; Hansen, S. H.; Pedersen-Bjergaard, S. Anal. Bioanal. Chem. 2014, 406, 421−429. (13) Gjelstad, A.; Rasmussen, K. E.; Pedersen-Bjergaard, S. J. Chromatogr. A 2007, 1174, 104−111. (14) Seip, K. F.; Jensen, H.; Sonsteby, M. H.; Gjelstad, A.; PedersenBjergaard, S. Electrophoresis 2013, 34, 792−799. (15) Huang, C.; Eibak, L. E.; Gjelstad, A.; Shen, X.; Trones, R.; Jensen, H.; Pedersen-Bjergaard, S. J. Chromatogr. A 2014, 1326, 7−12. H
DOI: 10.1021/acs.analchem.5b00981 Anal. Chem. XXXX, XXX, XXX−XXX
