Journal of Chromatography A, 1341 (2014) 1–7
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Optimization of solid phase microextraction coatings for liquid chromatography mass spectrometry determination of neurotransmitters夽 Erasmus Cudjoe, Janusz Pawliszyn ∗ Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
a r t i c l e
i n f o
Article history: Received 29 December 2013 Received in revised form 10 March 2014 Accepted 11 March 2014 Available online 18 March 2014 Keywords: Solid phase microextraction Liquid chromatography Mass spectrometry Bioanalytical applications
a b s t r a c t A simple solid phase microextraction method coupled to liquid chromatography mass spectrometry is introduced for the analysis of neurotransmitter compounds with a wide range of polarities in biological matrices. A novel “reversed” reverse-phase chromatographic method was developed without pre-column derivatization for the analysis of dopamine, serotonin, gamma aminobutyric acid and glutamate. New solid phase microextraction “in house” coatings using mixed-mode solid phase extraction particles were prepared, and used for the extraction of polar neurotransmitters. The polymer-support base reverse phase mixed-mode sorbents with strong ion exchange properties generally had higher extraction efficiencies compared to similar sorbents with weak ion exchange properties. The linear range was determined to be between 0.01 and 150 ng/mL for all the analytes, except for GABA, which was from 0.1 to 100 ng/mL. The limit of detection range was from 6 to 10 pg/mL for all the neurotransmitters, and the limits of quantitation were in the range of 20–35 pg/mL. The results demonstrate the potential of the SPMELC–MS/MS technique for bioanalysis of small polar endogenous compounds, such as neurotransmitters, from various biological matrices using the mixed-mode sorbents as the extraction phase. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The development of bioanalytical methods for quantitation of polar compounds is critical for analysis of pharmaceuticals, as most drugs have polar characteristics. Polar endogenous compounds are also often intermediates of various biological processes, and as such, can provide insight into some of the mechanisms of these processes. In clinical applications, more than one polar compound may be used to assist in disease diagnosis or as a biomarker. Even in food analysis, polar compounds such as melamine [1,2] and folic acid [3,4] can also form indicators in food safety and nutrition. Neurotransmitters are low molecular weight endogenous polar chemical substances found in the brain, and used to communicate information throughout the brain and body. They are known for their significant role in brain function, and are linked to our behavior, cognition, mood, and health, to mention a few. For
夽 Presented at the 40th International Symposium on High Performance Liquid Phase Separations and Related Techniques (HPLC 2013 Hobart), Hobart, Tasmania, Australia, 18–21 November 2013. ∗ Corresponding author. Tel.: +1 519 888 4641. E-mail address: [email protected] (J. Pawliszyn). http://dx.doi.org/10.1016/j.chroma.2014.03.035 0021-9673/© 2014 Elsevier B.V. All rights reserved.
example, changes in the function of dopamine (DA), a monoamine neurotransmitter, have been linked to Parkinson’s disease and schizophrenia [5]. Amino acid neurotransmitters, such as gamma aminobutyric acid (GABA) and glutamate, are most abundant in the brain and also constitute important building blocks for proteins [6]. Glutamate and GABA are neurotransmitters for fast excitatory and inhibitory synaptic transmissions respectively, and are involved in various functions of the central nervous system (CNS), as well as associated with several neurological diseases. Numerous clinical conditions, including psychiatric disorders, appear to involve an imbalance in excitation and inhibition [7]. Basically, abnormal neurotransmission has been linked to a wide range of conditions, including depression, drug dependence, and degenerative diseases, among many others. Measurements and analyses of neurotransmitters on the liquid chromatographic mass spectrometry (LC–MS) platform have undeniably improved our understanding of the relationship between the chemistry in the CNS and the behavioral, cognitive, and emotional state of an organism [8]. However, as endogenous polar compounds coupled with the dynamics of neurotransmission, development of quantitative sample preparation and sampling methods using LC–MS still remain a challenge. Separation and analysis of polar neurotransmitters by LC–MS can be quite a daunting analytical task. This is because these
2
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
compounds are usually found in lower concentration in biological matrices and very hydrophilic in nature. Sampling of highly hydrophilic molecules in biological matrices are characterized by low recoveries and relatively poorer chromatographic separation of the analytes when analyzed in their underivatized form. Microdialysis, a known analytical sampling tool with good selectivity for polar hydrophilic compounds, is commonly used for in vivo applications. However, the dialysate can also have a significant matrix impact when coupled to the electrospray ionization technique in LC–MS. For most in vitro applications, protein precipitation, liquid-liquid extraction and solid phase extraction (SPE) are typically used as sampling tools. Whereas huge matrix effects, due to the large presence of extracted phospholipids, characterize PPT and LLE, these methods suffer from a lack of appropriate solvent systems for the separation of polar hydrophilic compounds. Alternatively, sorptive methods such as SPE have been applied to the extraction of polar chemical substances in various matrices, especially for in vitro bio-applications. The principal interest in sorptive methods derives from the ability to modify the properties of the sorbent material for enhanced selectivity toward the target compounds [9]. Several hydrophilic sorbents with affinity for polar analytes have been prepared either by copolymerization of appropriate functional monomers, or by chemically modifying the hydrophobic polymer with a polar moiety [9]. For example, the Oasis HLBTM , a divinylbenzene-based copolymer used in the SPE method, has been reported for its potential to extract polar compounds; in this study, an online SPE method coupled to LC–MS/MS was reported for separating antibiotics in an aqueous matrix [10]. Other SPE-LC–MS applications have also been reported in literature [11–13]. Another sorptive method that has achieved considerable success for the analysis of compounds in various matrices is solid phase microextraction (SPME). Unlike the column-like packing for SPE cartridges, where a wider range of sorbents can be easily applied, SPME coatings are typically immobilized on a rigid support. This limits SPME applicability to sorbents having appropriate morphology, which allows direct deposition unto a rigid support. Despite this challenge, various types of commercially available SPME coatings have been applied to the extraction of analytes from complex matrices. For example, polypyrrole (PPY) coating has been used for the determination of -blockers [14,15], phenols [16], and aliphatic alcohols [17]. Other SPME approaches, such as online intube SPME [18–20], have been used for the analysis of polar analytes in various matrices, using LC for separation. Subsequently, recently developed “in-house” coating methods pioneered the fabrication of new SPME extraction phases for LC–MS applications for both in vitro and in vivo complex systems [21,22]. Effective chromatographic retention of small polar neurotransmitters cannot be attained without analyte derivatization. Typical reverse phase C8 and C18 stationary phases have been used for separation of many analytes using the appropriate solvent composition, temperature, pH, etc., [23]. However, chromatographers sometimes encounter difficult separations for which selectivity, ruggedness, or reproducibility are not easily obtained with conventional C8 and C18 phases. Thus, for improved retention, reverse phase separations often require the use of ion-pairing agents as one of the approaches to improve the retention factor for polar compounds in separation [23]. Derivatization agents such as o-phthalaldehyde (OPA) [24], naphthalene-2,3-dicarboxaldehyde (NDA) [25], and 1dimethylaminonaphthalenesulfonyl (DANSYL) [26], to mention a few, have been used for the analysis of neurotransmitters. The hydrophilic interaction chromatography (HILIC) column offered an alternative for the separation of various polar compounds, which were difficult to retain with conventional reverse phase columns. To date, HILIC separation continues to attract a lot of interest because it solves various hitherto difficult separation problems, such as the separation of small organic acids, basic drugs, and
many other neutral and charged substances [27]. However, the characteristics of the stationary phase may, in some cases, limit the choice of mobile phase composition, ion strength, or buffer pH value [27]. The technique often requires careful manipulation of mobile phase pH and buffer salt concentration. The resultant effect is often signal or ion suppression, depending on the analyte type when coupled to ESI-MS. An alternative to the HILIC separation is pentafluorophenyl (PFP) bonded to a silica stationary phase. PFP stationary phases have demonstrated unique retention for small polar analytes. PFP stationary phase separates compounds based upon selective interactions such as steric recognition, charge by transfer, or by – interactions. By using a PFP, it is often possible to improve separation and elution of difficult polar compounds in LC for easier quantitation. The PFP stationary phases also offer the flexibility of using common mobile phases, thus avoiding the use of ion-pairing reagents, concentrated buffer systems, strong pH conditions and complex mobile phase preparations. Since most of the mobile phases used with the PFP stationary phase do not require strong buffer conditions, enhanced MS signals with improved sensitivities can be observed. In this study, the use of silica- or polymer-support base mixedmode as new SPME coatings for quantitative LC–MS/MS analysis of selected polar neurotransmitters has been demonstrated. The selected neurotransmitters encompass polar organic compounds with a wide range of pKa values, and their LC separation was attained without the need for derivatization. To improve extraction efficiency, coatings with a higher surface to volume ratio were prepared using the flat blade/thin film configuration. The study also compares the extraction efficiencies of various “in-house” mixedmode SPME coatings for the analysis of neurotransmitters. Mixed mode sorbents where chosen due to their ability to offer multiple modes of interaction. This study offers a robust LC–MS/MS separation technique for quantification of polar neurotransmitters using mixed-mode coatings as SPME extraction phases. Chromatographic optimization and retention of the polar neurotransmitters were performed using the HILIC and PFP stationary phases were compared. The PFP stationary phase was finally chosen for retention and separation of both amino acid (glutamate and ␥-aminobutyric acid) and monoamine (dopamine and serotonin) neurotransmitters due to higher signal-to-noise ratio.
2. Experimental 2.1. Materials and methods HPLC grade acetonitrile (ACN) was purchased from EMD Chemicals Inc., Ontario. All mixed-mode SPE sorbent particles were obtained through the assistance of Chromatographic Specialties® , Ontario as research samples but are also available commercially. MCX, MAX, WCX and WAX particles were available from Waters® , (Milford, MA, U.S.A.), Discovery DPA-6S, C18 + B, C8 + B and C18 particles were available from Supelco® (Oakville, Ontario), Clean Screen DAU and GHB respectively were available from United Chemical Technologies (Bristol, PA, U.S.A.) and Chromabond® solid phase extraction particles were available from Machenery-Nagel (Bethlehem, PA, U.S.A.). The Loctite 349 impruvTM (R. S. Hughes Company, Plymouth, MI) and Kasil 1® (PQ Corporation, Valley Forge, PA) were used as adhesives. Medical grade stainless steel tubes were used for making flat surface blades through the assistance of University of Waterloo science machine shop, and were purchased from Small Parts® Inc., Miami, FL. The SK-300 mechanical shaker was obtained from JEIO Tech, Korea. Glutamate, GABA, DA and serotonin (5-HT) were purchased from Supelco® , Oakville, Ontario. Diazepam, used as an internal standard was purchased from Cerilliant Corporation,
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
Round Rock, TX, U.S.A. Artificial cerebrospinal fluid (aCSF) used for preparing samples and method development was obtained from Harvard Apparatus (Holliston, MA). HPLC grade formic acid was also purchased from Supelco® , Oakville, Ontario. Rat brain samples were obtained through the assistance of NoAb BioDiscoveries, Mississauga, ON. Deionized water for preparation of standards and LC mobile phases were from Barnstead/Thermodyne NANO-pure ultra-water system (Dubuque, IA, U.S.A.) and the Thermodyne® Maxi mix plus vortexer was also from Barnstead/Thermodyne (Dubuque, IA, U.S.A.). The HPLC columns for chromatographic separation were obtained from Supelco Discovery® (HS F5; 100 mm × 2.1 mm × 3 m) and Phenomenex (Torrance, CA), Kinetex® core shell (100 mm × 2.1 mm × 2.6 m). The C18 (150 mm × 2.1 mm × 3 m) and HILIC (50 mm × 2.1 mm × 5 m) columns were also obtained from Phenomenex. Cerebrospinal fluid (CSF) obtained from Sprague Dawley rat was purchased from Bioreclamation, Hicksville, NY. Individual stock standard (1 mg/mL) solutions were all prepared in a final solution of acetonitrile/water/formic acid in amber vials, and kept refrigerated for a maximum of four weeks until discarded. With the exception of glutamate, which was initially dissolved in acidified water (0.1% formic acid), all other standards were directly prepared in acetonitrile/water 2:3 (v/v) mixture with 0.1% formic acid. Instrument calibration standard solutions were freshly prepared by serial dilution of 1 g/mL solution prepared from the stock to cover a concentration range of 0.006–200 ng/mL. All samples and working calibration standards were prepared in physiological fluid (aCSF) while maintaining an organic content of less than 1% and subsequently extracted using the SPME coatings prepared. 2.1.1. Chromatographic procedure and mass spectrometry conditions Three main types of LC columns (reverse phase C18 , HILIC and PFP) were examined for their effectiveness in separating GABA, glutamate, DA and 5-HT. Various mobile phase compositions appropriate for a specific column type/stationary phase were used to determine the efficiency in retaining and separating the analytes. The C18 column, PFP and HILIC columns were assessed using acetonitrile/water mobile phase systems spiked with 0.1% formic acid and acetonitrile/ammonium formate buffer respectively. The AccelaTM autosampler (Thermo Scientific, USA), equipped with a temperature-controlled tray chamber, was used to introduce a 10 L sample for chromatographic separation using the AccelaTM HPLC Pump with a dual piston pump. The autosampler tray was maintained at 5 ◦ C throughout the entire analysis. The TSQ VantageTM triple quadrupole mass spectrometer (Thermo Scientific, USA) in positive mode was used to identify and quantify all analytes. The source voltage was optimized and set at 1000 V, while the capillary and vapourizer temperatures were both set at 250 ◦ C. Optimized sheath and auxiliary gases were 55 and 15 arbitrary units respectively. The parent and daughter ions monitored for each compound were as follows: GABA (104.1; 69.1), glutamate (148.1; 84.1), DA (154.1; 91.2) and 5-HT (177.1; 115.1). All data analyses were performed with the Xcalibur® software version 2.0.7. 2.1.2. Preparation of SPME coatings Two different coating preparation methods were explored to establish the effective coating procedure to be adopted for the study. For the first coating method, Loctite 349 impruvTM was used as glue, while Kasil 1® was used as adhesive to immobilize a thin film of extraction phase on the flat stainless steel blade. Prior to the coating process, the surfaces of the metal blades were etched in concentrated nitric acid for approximately an hour in order to improve adhesion. Subsequently, the blades were washed thoroughly with tap water followed by deionized water. After drying the surface of the metal for a few minutes, the metal blades were then placed into
3
Table 1 Types of SPE sorbents used for SPME coatings. Sorbent
Support
Type of interaction
Chromabond SA Discovery DPA-6S C18 particles Clean screen DAU Clean screen GHB SSBCX C18 + B
Silica n/a Silica Silica Silica Silica Silica
C8 + B
Silica
MCX
Polymer
MAX
Polymer
WCX
Polymer
WAX
Polymer
Strong ion exchange Polyamide resin Reverse phase Reverse phase and strong ion exchange n/a Strong ion exchange Reverse phase with mixed-mode strong ion exchange Reverse phase with mixed-mode strong ion exchange Reverse phase with strong mixed-mode ion exchange Reverse phase with strong mixed-mode ion exchange Reverse phase with weak mixed-mode ion exchange Reverse phase with weak mixed-mode ion exchange
* All particles are commercially available as solid phase extraction phases. MCX, MAX, WCX and WAX particles were copolymers of divinylbenzene).
acetone in a beaker and agitated for 30 min to remove any possible organic contaminants introduced during the washing process. Later, the blades were thoroughly dried under a gentle stream of nitrogen gas. With the Loctite 349 impruvTM , the first set of metal blades were dipped into the adhesive inside a 2 mL vial and subsequently rotated several times in a pile of particles on a clean paper. The prepared coatings were cured under an UV lamp for an hour while rotating the blade every 10 min to ensure all sides of the coatings were exposed to the UV lamp. In the case of the Kasil 1TM , the treated metal blade was dipped into the adhesive for about 15 s, followed by rotating the adhesive coated blade in the pile of particles. The blade coated with particles was subsequently passed over fumes of concentrated nitric acid for a few seconds. Later, the blades were kept inside a desiccator overnight. The immobilized coatings on the blade were 1.5 cm long and 2.0 mm wide. Sorbents used for this study were categorized into two main groups: silica- and polymer-support base (Table 1). For the purpose of evaluation, 3 replicates of each coating were selected, initially evaluated and later assessed for their extraction efficiency prior to use for further extractions. This pre-screening was carried to minimize variations between coatings. Prior to SPME extractions, all coatings were pre-conditioned overnight in 1:1 (v/v) methanol/water, with agitation on at 150 rpm on a mechanical shaker. Subsequently, to reduce the organic, the coatings were placed for a couple of minutes in an aCSF solution diluted 10× with deionized water content prior to extraction. As part of the objective, the reusability of each coating type was also monitored to ascertain its robustness. 2.1.3. SPME Extraction Procedure All extractions and desorptions were carried out in aCSF and water/acetonitrile 3:2 (v/v) with 0.1% formic acid, respectively. An hour extraction of 50 ng/mL solution of the analytes was carried out under static conditions with each sorbent type; subsequently, the neurotransmitters were desorbed in a 180 L desorption solution in a 300 L amber vial. During the one-hour desorption process, the blades were agitated at 800 rpm on the SK-300 mechanical shaker. After the desorption process, the samples were further subjected to liquid chromatographic separation and tandem MS analysis using the conditions stipulated in the previous section. Details on the chromatographic method are presented in subsequent section.
4
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
Fig. 1. Chromatographic separation of underivatized polar neurotransmitters in a regular reverse phase chromatography mode on a PFP column. GABA and glutamate eluted within the void volume.
2.1.3.1. Extraction of neurotransmitters from CSF and rat brain samples. Approximately 2.0 g of the rat brain samples were weighed into previously cleaned 20 mL vials and subjected to a high speed vortex for about 3–5 min at approximately 30 s intervals to avoid heating of the samples beyond laboratory room temperature. All the brain samples were later pooled into a laboratory petri-dish, covered and vortexed again. After this process, approximately 1.0 g of the macerated brain tissue sample was transferred into 3 separate vials and distinctly spiked at different concentration levels (50 ng/mL and 500 ng/mL). Next, 6 vials were spiked at 5 ng/mL and vortexed for about 2 min. Each of the spiked samples were prepared in replicates of 3 and extracted with the C18 + B SPME blade coating for 30 min, with agitation at 750 rpm. In addition, unspiked macerated brain tissue samples were also extracted and treated as blank correction. After the extraction process, the SPME blades were wiped with KimwipesTM to physically remove the deposits of brain tissue from the surface of the coatings, dipped into deionized water for about 2 s, and then desorbed in 300 L of desorption solution containing 5 ng/mL diazepam as an internal standard. The same set of SPME blades were used for all extractions at a particular concentration level. Prior to the extraction process, preliminary extractions from aCSF in steady state conditions revealed that all the analytes reached equilibrium within 20 min. A relatively longer extraction time was chosen, despite the agitation, to ensure that all analytes reached equilibrium due to the tortuosity of the brain tissue samples. The average area response of each analyte for the blank samples (unspiked samples) was subtracted from their respective responses at each concentration level. A 6-point calibration curve was prepared by extracting spiked samples of aCSF with neurotransmitters at concentrations ranging from 0.1 to 200 ng/mL. A plot of the area ratio of the each analyte to the internal standard (diazepam) versus the nominal concentration of the standard was used for quantification. Extractions were also carried out using the rat CSF obtained from Bioreclamation® . The sample and desorption solution volumes used in this experiment were 750 and 150 L respectively. However, all CSF experiments were carried out in a steady
state and without spiking, and experiments were carried out in triplicate.
3. Results and discussions 3.1. Chromatographic method optimization As a proof of concept, the C18 column used for retaining and separation of the neurotransmitters did not provide any retention of the analytes for isocratic reverse phase separation method with a 90% acetonitrile and 10% water mixture spiked with 0.1% formic acid using a 300 L/min flow rate (data not shown). This experiment was carried out to confirm that retention of the selected neurotransmitters without derivatization in a typical reverse phase C18 column was not feasible as the compounds are likely to elute in the column void volume. Subsequently, two types of PFP stationary phase columns were also evaluated for their effectiveness in separating and retaining the polar molecules, using the same mobile phase system as previously mentioned. However, the Discovery HS F5 gave a huge backpressure at relatively higher flow rates, and thus a flow rate of 300 L/min in a reverse phase chromatographic separation was used. Although retention of separation was attained with the column, the chromatographic runtime was longer (15 min), including column pre-conditioning. In addition, DA, which eluted at about 11.5 min, and had a characteristic broader peak (data not shown). However, the Kinetex® core shell PFP column was able to resolve the peak broadening observed for DA. This observation could be attributed to the relatively smaller particle sizes of the column, and with the core shell technology, was able to withstand higher flow rates with reasonable backpressures. The overall effect was narrower peak shapes, with an improvement in the signal-to-noise ratio for each analyte. Mobiles phases used for the Kinetex® column consisted of A (90% water, 5% methanol, 5% acetonitrile and 0.1% formic acid) and B (10% water, 90% acetonitrile and 0.1% formic acid) at 450 L/min
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
5
Fig. 2. Chromatographic separation of underivatized polar neurotransmitters in “reversed” reverse phase chromatography mode without the need for buffers on a PFP column.
flow rate for both normal and reverse phase modes. The typical reverse phase gradient separation mode, started with 90% A, which was held for 1 min and mobile phase B was increased gradually to 90% by 3 min. The high organic content was maintained for 0.5 min and the column equilibrated for the next 1.5 min. With this gradient approach, DA and glutamate compounds eluted with the volume and most of the analytes were tailing as shown in Fig. 1. From the chromatogram, most of the compounds eluted with the stronger aqueous solvent front due to the high hydrophilicity. Diazepam eluted much later because it is relatively more hydrophobic than the neurotransmitters. To increase the retention capacity of the PFP column, chromatographic separation was started with a mobile phase B (high organic content), followed by a gradual increase in the mobile phase A. Retention and separation was achieved within a 5-min run-time (Fig. 2), starting with 0% mobile phase A, held for 1 min, and then gradually increased to 100% A by 3.5 min. Mobile phase A (100%) was maintained for half a minute before re-conditioning of the column for another minute. Diazepam, used as an internal standard for correction of any potential injection errors, eluted earlier due to its characteristic higher hydrophobicity compared to Fig. 1 where it eluted much later in a typical reverse phase separation method. However, with this particular separation, in a normal phase separation method, the aqueous mobile phase, which acted as the stronger eluting solvent during the chromatographic run-time, effectively separated the highly polar neurotransmitters. In the case of the HILIC column, mobile phase A consisted of 1:1 (v/v) water and ammonium formate buffer with pH adjusted to 3.5, and B contained 95% acetonitrile and 5% ammonium formate. The chromatographic method started with 0% mobile phase A, held for 1 min, gradually increased to 95% A by 4 min, and then the column was re-conditioned for the next minute. Although retention and separation of all the neurotransmitters was attained within the 5 min run-time on the HILIC column (data not shown), an approximately 50-fold decrease in sensitivity was observed compared to the response with the PFP column. The reduction in sensitivity most likely resulted from signal suppression at the MS ion source due to the presence of a high concentration of buffer ions in the mobile
phase, which might have affected the effective ionization of the neurotransmitters in their underivatized forms. 3.2. SPME coatings evaluation The evaluation of SPME coatings focused on the overall extraction efficiency of analytes from the aCSF solution by each coating type. The pH of the aCSF solution was maintained at physiological conditions to mimic a typical biological system. This was done so that the optimized method will be amenable to in vivo applications or biological matrices where the physiological pH cannot be adjusted. With the exception of the ChromabondTM , Clean ScreenTM sorbents, DPA-6STM and C18 particles, the rest were all mixed-mode sorbents on either a silica- or polymer-support base as shown in Table 1. The mixed-mode sorbents were included because of their characteristic multi-interactions of hydrophobic and hydrophilic mechanisms. All the analytes were typically polar, with pKa values ranging from 2.13 for glutamate to 9.8 for 5-HT in aqueous media. 3.2.1. Evaluation of coating procedures To compare the two coating procedures, the extraction efficiencies, robustness, and the inter- and intra-coating reproducibility of the selected fibers were compared. Each coating was prepared in triplicate and used for 5 extractions of glutamate, DA and GABA from aCSF. Fig. 3 shows the percentages of glutamate that were extracted by each sorbent using the two different coating approaches. Of the two methods, coatings made using Kasil 1TM as an adhesive extracted relatively higher amounts of glutamate, DA and GABA. The difference in the amount extracted could be attributed to the difference in particle sizes. Smaller particle sizes (≤10 m) were used with the Kasil 1TM adhesive method, while larger particle sizes (30 m ≤ particle size ≤ 60 m) were used with the Loctite adhesive method. It is important to note that with smaller particle sizes, a larger surface area of the coating would be obtained, and thus improve the overall extraction efficiency. The smaller error bars may also be due to the uniformity of the particle sizes used with the Kasil 1® adhesive compared to the greater variation in the particle sizes used with the Loctite 349 impruvTM
6
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
Fig. 3. Comparison of the efficiencies of two SPME coating methods for the extraction of glutamate from artificial cerebrospinal fluid for selected sorbents (n = 5).
adhesive. Thus, variations between the blades were minimal. The %RSD for the 3 replicates of coatings were respectively 15% and 9% for the Loctite 349 impruvTM and the Kasil 1TM adhesive. Subsequently, all analyses were performed using the Kasil 1TM adhesive coating procedure (Fig. 3). 3.2.2. Evaluation of sorbent extraction efficiency The initial criterion used in the evaluation process involved the pre-selection of sorbent(s) able to extract quantitative amounts of all four neurotransmitters, i.e., the ability of the developed SPME coating to extract analytes with a wide range of varying pKa values. Therefore, the sorbents were initially screened for their ability to extract all 4 neurotransmitters using spiked samples of aCSF containing 50 ng/mL of each analyte. Table 2 shows that most of the mixed-mode sorbents were able to extract all 4 neurotransmitters. Chromabond SA, Clean screen GHB, SSBCX, DPA-6S and C18 sorbents extracted ≤3 analytes with DPA-6S and C18 sorbent extracting only glutamate and DA respectively. This observation may be due to the fact that the DPA-6S and the C18 do not exhibit multiple interaction modes (hydrophobic and hydrophilic interactions) with the analytes. However, in the case of the mixed-mode SPE particles, due to their multiple interactions with compounds, the coatings were able to extract all 4 neurotransmitters in quantitative amounts. Subsequently, the efficiencies of the selected mixed-mode SPME coatings (C18 + B; C8 + B; MCX; MAX; WCX; WAX) were determined and compared by determining the amounts extracted in a separate set of extractions. The comparison process entailed triplicate one hour SPME extractions of 50 ng/mL neurotransmitters in aCSF samples with subsequent desorption for another hour. A second desorption of the same SPME coating was carried out to ascertain any carryover amounts after the initial desorption. The percent amount of each extracted analyte was determined for the selected Table 2 Screening of sorbents used for SPME coatings for their ability to extract neurotransmitters. Sorbent type
Chromabond SA DPA-6S Clean screen DAU Clean screen GHB SSBCX C18 + B C8 + B MCX MAX WCX WAX C18 particles
Neurotransmitter GLU
GABA
DA
5-HT
+ + + + + + + + + + +
+ + + + + + + + + +
+ + + + + + + + + +
n/a n/a n/a + + + + + + n/a
n/a, not available at the time of experiment. (+), extracted quantitative amount; (), detected but not quantitative.
Fig. 4. Comparison of the extraction efficiency of selected mixed-mode SPE sorbents used as SPME coatings for the extraction of neurotransmitters spiked in artificial cerebrospinal fluid.
coatings. Results (Fig. 4) showed that C18 with benzenesulphonic acid group (C18 + B) extracted relatively equal amounts of GABA, DA and 5-HT, with glutamate being the analyte with the highest extracted amount. Glutamate was most extracted the by all the sorbents, while DA relatively least extracted. With the exception of C18 + B, there were no significant differences in the amounts of GABA and glutamate extracted by the MAX and C8 + B sorbents. MCX sorbent showed relatively higher extraction efficiency for 5HT; however, there was no significant difference in the amount extracted when compared to that of the C18 + B sorbent. Whereas WCX sorbent did not show any significant difference in its extraction efficiency for all the analytes, WAX showed higher extraction efficiency for GABA and glutamate only, despite the fact that both sorbents had similar weak mixed-mode ion exchange properties. Relatively, mixed-mode coatings with relatively stronger ion exchange properties performed better in the extraction of the polar neurotransmitters compared to their corresponding sorbents with weak ion exchange properties. This may be due to the fact that hydrophilic interactions between the coatings and the analytes were stronger. However, the changes in the amount extracted by the sorbents were not significantly different. Finally, in terms of base support (silica- or polymer-based sorbents), there were no observable patterns in the extraction efficiencies of the sorbents for these analytes. This implies that the extraction efficiency of the sorbent was not necessarily dependent on the type of base support. 3.3. Application of mixed-mode coatings for extraction of selected neurotransmitters from CSF and rat brain tissue samples The C18 + B SPE coating was selected among the mixed-mode particles and applied for the extraction of all 4 neurotransmitters in two biological matrices (CSF and rat brain tissue). To ensure method validity, the extractions were carried out with the brain samples were at three different concentration levels. Results obtained are shown in Table 3. The precision (%RSD) for each analyte at each concentration level determined was less than 12% (n = 3) for the rat brain tissue, which is acceptable for sample preparation processes for complex Table 3 Results obtained for the extraction of spiked rat brain tissue sample and CSF. Neurotransmitter (ng/mL) GA Spiked brain samples
50 ng/mL (n = 3) 48.0 (4) 500 ng/mL (n = 3) 510 (11) 5 ng/mL (n = 6)
CSF sample nd: not detected.
GABA
DA
5-HT
45.0 (6) 493 (9)
52.0 (5) 505 (8)
43.0 (4) 496 (10)
5.50 (1.2) 4.80 (1.0) GABA GA
5.10 (0.7) 4.50 (1.0) DA 5-HT
0.92 (0.1) 0.53 (0.06) nd
nd
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
biological matrices. The analytical figures of merit determined were the limit of detection and quantitation, the linear range using neat solutions of standards prepared in desorption solution, and method robustness. With the exception 5-HT, which gave 86% accuracy for extractions using 50 ng/mL and 500 ng/mL rat brain samples, accuracy all the analytes were ≥90% for 3 replicate extractions. The RDS% obtained for 6 replicates of the lowest concentration (5 ng/mL) spiked brain samples were used to measure method robustness. The linear range was determined to be between 0.01 and 150 ng/mL for all the analytes, except for GABA, which was from 0.1 to 100 ng/mL. The limit of detection range was from 6 to 10 pg/mL for all the neurotransmitters. Glutamate and GABA recorded 6 and 10 pg/mL LODs whereas DA and 5-HT were the same 8 ng/mL. The limits of quantitation (LOQ) were in the range of 20–35 pg/mL. The LODs and LOQs were determined based on 3× the S/N of a blank extraction of the aCSF and the LOQ was 10× the S/N of the blank extract. The method sensitivity, reliability and robustness have also been demonstrated elsewhere in an in vivo extractions of neurotransmitters in the brain extracellular fluid of freely moving rats [28]. 3.4. Conclusion The mixed-mode sorbents showed superior extraction efficiencies over other coatings in their ability to quantitatively extract polar neurotransmitters of varying polarities from biological matrices. By combining a “reversed” reverse-phase chromatographic separation on a PFP stationary phase column, a robust LC–MS/MS quantitative method was developed for the retention, separation and analysis of neurotransmitters without analyte derivatization. The project has demonstrated that by choosing the appropriate sorbent, the SPME-LC–MS/MS technique is a viable alternative bioanalytical method for the quantitative determination of both endogenous and exogenous analytes in various biological matrices.
7
References [1] G. Venkatasami, J.R. Sowa, Anal. Chim. Acta 665 (2010) 227. [2] H. Sun, L. Wang, L. Ai, S. Liang, H. Wu, Food Control 21 (2010) 686. ´ [3] A. Lebiedzinska, M. Da¸browska, P. Szefer, M. Marszałł, Toxicol. Mech. Methods 18 (2008) 463. [4] European Food Safety Authority, Scientific opinion on the safety and efficacy of folic acid as a feed additive for all animal species, EFSA J. 10 (5) (2012) 2674. [5] P. Uutela, L. Karhu, P. Piepponen, M. Käenmäki, R.A. Ketola, R. Kostiainen, Anal. Chem. 81 (2009) 427. [6] E. Herlenius, H. Lagercrantz, Early Hum. Dev. 65 (2001) 21. [7] R.W. Olsen, GABA, in: Neuropsychopharmacol. Fifth Gener. Prog., Lippincott Williams & Wilkins, 2002. [8] M. Perry, Q. Li, R.T. Kennedy, Anal. Chim. Acta 653 (2009) 1. [9] N. Fontanals, R.M. Marcé, F. Borrull, J. Chromatogr. A 1152 (2007) 14. [10] V.R. Panditi, S.R. Batchu, P.R. Gardinali, Anal. Bioanal. Chem. 405 (2013) 5953. [11] L.G. Freitas, C.W. Götz, M. Ruff, H.P. Singer, S.R. Müller, J. Chromatogr. A 1028 (2004) 277. [12] R. Carabias-Martínez, E. Rodríguez-Gonzalo, P. Revilla-Ruiz, J. Chromatogr. A 1056 (2004) 131. [13] M. Smith, G.E. Collins, J. Wang, J. Chromatogr. A 991 (2003) 159. [14] J. Wu, J. Pawliszyn, J. Chromatogr. A 909 (2001) 37. [15] J. Wu, H.L. Lord, J. Pawliszyn, H. Kataoka, J. Microcolumn Sep. 12 (2000) 255. [16] R. Wissiack, E. Rosenberg, M. Grasserbauer, J. Chromatogr. A 896 (2000) 159. [17] D. Djozan, S. Bahar, Chromatographia 59 (2004) 95. [18] H. Kataoka, J. Pawliszyn, Chromatographia 50 (1999) 532. [19] H. Bagheri, A. Salemi, Chromatographia 59 (2004) 501. [20] M.E.C. Queiroz, E.B. Oliveira, F. Breton, J. Pawliszyn, J. Chromatogr. A 1174 (2007) 72. [21] D. Vuckovic, I. de Lannoy, B. Gien, Y. Yang, F.M. Musteata, R. Shirey, L. Sisisky, J. Pawliszyn, J. Chromatogr. A 1218 (2011) 3367. [22] N. Raikos, G. Theodoridis, E. Alexiadou, H. Gika, H. Argiriadou, H. Parlapani, H. Tsoukali, J. Sep. Sci. 32 (2009) 1018. [23] J. Henriksen, P. Roepstorff, L.H. Ringborg, Carbohydr. Res. 341 (2006) 382. [24] J.F. Stobaugh, A.J. Repta, L.A. Sternson, J. Pharm. Biomed. Anal. 4 (1986) 341. [25] F.D. Swanek, B.B. Anderson, A.G. Ewing, J. Microcolumn Sep. 10 (1997) 185. [26] F. Beaudry, S.A. Guénette, A. Winterborn, J.-F. Marier, P. Vachon, J. Pharm. Biomed. Anal. 39 (2005) 411. [27] B. Buszewski, S. Noga, Anal. Bioanal. Chem. 402 (2012) 231. [28] E. Cudjoe, B. Bojko, I. de Lannoy, V. Saldivia, J. Pawliszyn, Angew. Chem. Int. Ed. 52 (2013) 12124.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Optimization of solid phase microextraction coatings for liquid chromatography mass spectrometry determination of neurotransmitters夽 Erasmus Cudjoe, Janusz Pawliszyn ∗ Department of Chemistry, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
a r t i c l e
i n f o
Article history: Received 29 December 2013 Received in revised form 10 March 2014 Accepted 11 March 2014 Available online 18 March 2014 Keywords: Solid phase microextraction Liquid chromatography Mass spectrometry Bioanalytical applications
a b s t r a c t A simple solid phase microextraction method coupled to liquid chromatography mass spectrometry is introduced for the analysis of neurotransmitter compounds with a wide range of polarities in biological matrices. A novel “reversed” reverse-phase chromatographic method was developed without pre-column derivatization for the analysis of dopamine, serotonin, gamma aminobutyric acid and glutamate. New solid phase microextraction “in house” coatings using mixed-mode solid phase extraction particles were prepared, and used for the extraction of polar neurotransmitters. The polymer-support base reverse phase mixed-mode sorbents with strong ion exchange properties generally had higher extraction efficiencies compared to similar sorbents with weak ion exchange properties. The linear range was determined to be between 0.01 and 150 ng/mL for all the analytes, except for GABA, which was from 0.1 to 100 ng/mL. The limit of detection range was from 6 to 10 pg/mL for all the neurotransmitters, and the limits of quantitation were in the range of 20–35 pg/mL. The results demonstrate the potential of the SPMELC–MS/MS technique for bioanalysis of small polar endogenous compounds, such as neurotransmitters, from various biological matrices using the mixed-mode sorbents as the extraction phase. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The development of bioanalytical methods for quantitation of polar compounds is critical for analysis of pharmaceuticals, as most drugs have polar characteristics. Polar endogenous compounds are also often intermediates of various biological processes, and as such, can provide insight into some of the mechanisms of these processes. In clinical applications, more than one polar compound may be used to assist in disease diagnosis or as a biomarker. Even in food analysis, polar compounds such as melamine [1,2] and folic acid [3,4] can also form indicators in food safety and nutrition. Neurotransmitters are low molecular weight endogenous polar chemical substances found in the brain, and used to communicate information throughout the brain and body. They are known for their significant role in brain function, and are linked to our behavior, cognition, mood, and health, to mention a few. For
夽 Presented at the 40th International Symposium on High Performance Liquid Phase Separations and Related Techniques (HPLC 2013 Hobart), Hobart, Tasmania, Australia, 18–21 November 2013. ∗ Corresponding author. Tel.: +1 519 888 4641. E-mail address: [email protected] (J. Pawliszyn). http://dx.doi.org/10.1016/j.chroma.2014.03.035 0021-9673/© 2014 Elsevier B.V. All rights reserved.
example, changes in the function of dopamine (DA), a monoamine neurotransmitter, have been linked to Parkinson’s disease and schizophrenia [5]. Amino acid neurotransmitters, such as gamma aminobutyric acid (GABA) and glutamate, are most abundant in the brain and also constitute important building blocks for proteins [6]. Glutamate and GABA are neurotransmitters for fast excitatory and inhibitory synaptic transmissions respectively, and are involved in various functions of the central nervous system (CNS), as well as associated with several neurological diseases. Numerous clinical conditions, including psychiatric disorders, appear to involve an imbalance in excitation and inhibition [7]. Basically, abnormal neurotransmission has been linked to a wide range of conditions, including depression, drug dependence, and degenerative diseases, among many others. Measurements and analyses of neurotransmitters on the liquid chromatographic mass spectrometry (LC–MS) platform have undeniably improved our understanding of the relationship between the chemistry in the CNS and the behavioral, cognitive, and emotional state of an organism [8]. However, as endogenous polar compounds coupled with the dynamics of neurotransmission, development of quantitative sample preparation and sampling methods using LC–MS still remain a challenge. Separation and analysis of polar neurotransmitters by LC–MS can be quite a daunting analytical task. This is because these
2
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
compounds are usually found in lower concentration in biological matrices and very hydrophilic in nature. Sampling of highly hydrophilic molecules in biological matrices are characterized by low recoveries and relatively poorer chromatographic separation of the analytes when analyzed in their underivatized form. Microdialysis, a known analytical sampling tool with good selectivity for polar hydrophilic compounds, is commonly used for in vivo applications. However, the dialysate can also have a significant matrix impact when coupled to the electrospray ionization technique in LC–MS. For most in vitro applications, protein precipitation, liquid-liquid extraction and solid phase extraction (SPE) are typically used as sampling tools. Whereas huge matrix effects, due to the large presence of extracted phospholipids, characterize PPT and LLE, these methods suffer from a lack of appropriate solvent systems for the separation of polar hydrophilic compounds. Alternatively, sorptive methods such as SPE have been applied to the extraction of polar chemical substances in various matrices, especially for in vitro bio-applications. The principal interest in sorptive methods derives from the ability to modify the properties of the sorbent material for enhanced selectivity toward the target compounds [9]. Several hydrophilic sorbents with affinity for polar analytes have been prepared either by copolymerization of appropriate functional monomers, or by chemically modifying the hydrophobic polymer with a polar moiety [9]. For example, the Oasis HLBTM , a divinylbenzene-based copolymer used in the SPE method, has been reported for its potential to extract polar compounds; in this study, an online SPE method coupled to LC–MS/MS was reported for separating antibiotics in an aqueous matrix [10]. Other SPE-LC–MS applications have also been reported in literature [11–13]. Another sorptive method that has achieved considerable success for the analysis of compounds in various matrices is solid phase microextraction (SPME). Unlike the column-like packing for SPE cartridges, where a wider range of sorbents can be easily applied, SPME coatings are typically immobilized on a rigid support. This limits SPME applicability to sorbents having appropriate morphology, which allows direct deposition unto a rigid support. Despite this challenge, various types of commercially available SPME coatings have been applied to the extraction of analytes from complex matrices. For example, polypyrrole (PPY) coating has been used for the determination of -blockers [14,15], phenols [16], and aliphatic alcohols [17]. Other SPME approaches, such as online intube SPME [18–20], have been used for the analysis of polar analytes in various matrices, using LC for separation. Subsequently, recently developed “in-house” coating methods pioneered the fabrication of new SPME extraction phases for LC–MS applications for both in vitro and in vivo complex systems [21,22]. Effective chromatographic retention of small polar neurotransmitters cannot be attained without analyte derivatization. Typical reverse phase C8 and C18 stationary phases have been used for separation of many analytes using the appropriate solvent composition, temperature, pH, etc., [23]. However, chromatographers sometimes encounter difficult separations for which selectivity, ruggedness, or reproducibility are not easily obtained with conventional C8 and C18 phases. Thus, for improved retention, reverse phase separations often require the use of ion-pairing agents as one of the approaches to improve the retention factor for polar compounds in separation [23]. Derivatization agents such as o-phthalaldehyde (OPA) [24], naphthalene-2,3-dicarboxaldehyde (NDA) [25], and 1dimethylaminonaphthalenesulfonyl (DANSYL) [26], to mention a few, have been used for the analysis of neurotransmitters. The hydrophilic interaction chromatography (HILIC) column offered an alternative for the separation of various polar compounds, which were difficult to retain with conventional reverse phase columns. To date, HILIC separation continues to attract a lot of interest because it solves various hitherto difficult separation problems, such as the separation of small organic acids, basic drugs, and
many other neutral and charged substances [27]. However, the characteristics of the stationary phase may, in some cases, limit the choice of mobile phase composition, ion strength, or buffer pH value [27]. The technique often requires careful manipulation of mobile phase pH and buffer salt concentration. The resultant effect is often signal or ion suppression, depending on the analyte type when coupled to ESI-MS. An alternative to the HILIC separation is pentafluorophenyl (PFP) bonded to a silica stationary phase. PFP stationary phases have demonstrated unique retention for small polar analytes. PFP stationary phase separates compounds based upon selective interactions such as steric recognition, charge by transfer, or by – interactions. By using a PFP, it is often possible to improve separation and elution of difficult polar compounds in LC for easier quantitation. The PFP stationary phases also offer the flexibility of using common mobile phases, thus avoiding the use of ion-pairing reagents, concentrated buffer systems, strong pH conditions and complex mobile phase preparations. Since most of the mobile phases used with the PFP stationary phase do not require strong buffer conditions, enhanced MS signals with improved sensitivities can be observed. In this study, the use of silica- or polymer-support base mixedmode as new SPME coatings for quantitative LC–MS/MS analysis of selected polar neurotransmitters has been demonstrated. The selected neurotransmitters encompass polar organic compounds with a wide range of pKa values, and their LC separation was attained without the need for derivatization. To improve extraction efficiency, coatings with a higher surface to volume ratio were prepared using the flat blade/thin film configuration. The study also compares the extraction efficiencies of various “in-house” mixedmode SPME coatings for the analysis of neurotransmitters. Mixed mode sorbents where chosen due to their ability to offer multiple modes of interaction. This study offers a robust LC–MS/MS separation technique for quantification of polar neurotransmitters using mixed-mode coatings as SPME extraction phases. Chromatographic optimization and retention of the polar neurotransmitters were performed using the HILIC and PFP stationary phases were compared. The PFP stationary phase was finally chosen for retention and separation of both amino acid (glutamate and ␥-aminobutyric acid) and monoamine (dopamine and serotonin) neurotransmitters due to higher signal-to-noise ratio.
2. Experimental 2.1. Materials and methods HPLC grade acetonitrile (ACN) was purchased from EMD Chemicals Inc., Ontario. All mixed-mode SPE sorbent particles were obtained through the assistance of Chromatographic Specialties® , Ontario as research samples but are also available commercially. MCX, MAX, WCX and WAX particles were available from Waters® , (Milford, MA, U.S.A.), Discovery DPA-6S, C18 + B, C8 + B and C18 particles were available from Supelco® (Oakville, Ontario), Clean Screen DAU and GHB respectively were available from United Chemical Technologies (Bristol, PA, U.S.A.) and Chromabond® solid phase extraction particles were available from Machenery-Nagel (Bethlehem, PA, U.S.A.). The Loctite 349 impruvTM (R. S. Hughes Company, Plymouth, MI) and Kasil 1® (PQ Corporation, Valley Forge, PA) were used as adhesives. Medical grade stainless steel tubes were used for making flat surface blades through the assistance of University of Waterloo science machine shop, and were purchased from Small Parts® Inc., Miami, FL. The SK-300 mechanical shaker was obtained from JEIO Tech, Korea. Glutamate, GABA, DA and serotonin (5-HT) were purchased from Supelco® , Oakville, Ontario. Diazepam, used as an internal standard was purchased from Cerilliant Corporation,
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
Round Rock, TX, U.S.A. Artificial cerebrospinal fluid (aCSF) used for preparing samples and method development was obtained from Harvard Apparatus (Holliston, MA). HPLC grade formic acid was also purchased from Supelco® , Oakville, Ontario. Rat brain samples were obtained through the assistance of NoAb BioDiscoveries, Mississauga, ON. Deionized water for preparation of standards and LC mobile phases were from Barnstead/Thermodyne NANO-pure ultra-water system (Dubuque, IA, U.S.A.) and the Thermodyne® Maxi mix plus vortexer was also from Barnstead/Thermodyne (Dubuque, IA, U.S.A.). The HPLC columns for chromatographic separation were obtained from Supelco Discovery® (HS F5; 100 mm × 2.1 mm × 3 m) and Phenomenex (Torrance, CA), Kinetex® core shell (100 mm × 2.1 mm × 2.6 m). The C18 (150 mm × 2.1 mm × 3 m) and HILIC (50 mm × 2.1 mm × 5 m) columns were also obtained from Phenomenex. Cerebrospinal fluid (CSF) obtained from Sprague Dawley rat was purchased from Bioreclamation, Hicksville, NY. Individual stock standard (1 mg/mL) solutions were all prepared in a final solution of acetonitrile/water/formic acid in amber vials, and kept refrigerated for a maximum of four weeks until discarded. With the exception of glutamate, which was initially dissolved in acidified water (0.1% formic acid), all other standards were directly prepared in acetonitrile/water 2:3 (v/v) mixture with 0.1% formic acid. Instrument calibration standard solutions were freshly prepared by serial dilution of 1 g/mL solution prepared from the stock to cover a concentration range of 0.006–200 ng/mL. All samples and working calibration standards were prepared in physiological fluid (aCSF) while maintaining an organic content of less than 1% and subsequently extracted using the SPME coatings prepared. 2.1.1. Chromatographic procedure and mass spectrometry conditions Three main types of LC columns (reverse phase C18 , HILIC and PFP) were examined for their effectiveness in separating GABA, glutamate, DA and 5-HT. Various mobile phase compositions appropriate for a specific column type/stationary phase were used to determine the efficiency in retaining and separating the analytes. The C18 column, PFP and HILIC columns were assessed using acetonitrile/water mobile phase systems spiked with 0.1% formic acid and acetonitrile/ammonium formate buffer respectively. The AccelaTM autosampler (Thermo Scientific, USA), equipped with a temperature-controlled tray chamber, was used to introduce a 10 L sample for chromatographic separation using the AccelaTM HPLC Pump with a dual piston pump. The autosampler tray was maintained at 5 ◦ C throughout the entire analysis. The TSQ VantageTM triple quadrupole mass spectrometer (Thermo Scientific, USA) in positive mode was used to identify and quantify all analytes. The source voltage was optimized and set at 1000 V, while the capillary and vapourizer temperatures were both set at 250 ◦ C. Optimized sheath and auxiliary gases were 55 and 15 arbitrary units respectively. The parent and daughter ions monitored for each compound were as follows: GABA (104.1; 69.1), glutamate (148.1; 84.1), DA (154.1; 91.2) and 5-HT (177.1; 115.1). All data analyses were performed with the Xcalibur® software version 2.0.7. 2.1.2. Preparation of SPME coatings Two different coating preparation methods were explored to establish the effective coating procedure to be adopted for the study. For the first coating method, Loctite 349 impruvTM was used as glue, while Kasil 1® was used as adhesive to immobilize a thin film of extraction phase on the flat stainless steel blade. Prior to the coating process, the surfaces of the metal blades were etched in concentrated nitric acid for approximately an hour in order to improve adhesion. Subsequently, the blades were washed thoroughly with tap water followed by deionized water. After drying the surface of the metal for a few minutes, the metal blades were then placed into
3
Table 1 Types of SPE sorbents used for SPME coatings. Sorbent
Support
Type of interaction
Chromabond SA Discovery DPA-6S C18 particles Clean screen DAU Clean screen GHB SSBCX C18 + B
Silica n/a Silica Silica Silica Silica Silica
C8 + B
Silica
MCX
Polymer
MAX
Polymer
WCX
Polymer
WAX
Polymer
Strong ion exchange Polyamide resin Reverse phase Reverse phase and strong ion exchange n/a Strong ion exchange Reverse phase with mixed-mode strong ion exchange Reverse phase with mixed-mode strong ion exchange Reverse phase with strong mixed-mode ion exchange Reverse phase with strong mixed-mode ion exchange Reverse phase with weak mixed-mode ion exchange Reverse phase with weak mixed-mode ion exchange
* All particles are commercially available as solid phase extraction phases. MCX, MAX, WCX and WAX particles were copolymers of divinylbenzene).
acetone in a beaker and agitated for 30 min to remove any possible organic contaminants introduced during the washing process. Later, the blades were thoroughly dried under a gentle stream of nitrogen gas. With the Loctite 349 impruvTM , the first set of metal blades were dipped into the adhesive inside a 2 mL vial and subsequently rotated several times in a pile of particles on a clean paper. The prepared coatings were cured under an UV lamp for an hour while rotating the blade every 10 min to ensure all sides of the coatings were exposed to the UV lamp. In the case of the Kasil 1TM , the treated metal blade was dipped into the adhesive for about 15 s, followed by rotating the adhesive coated blade in the pile of particles. The blade coated with particles was subsequently passed over fumes of concentrated nitric acid for a few seconds. Later, the blades were kept inside a desiccator overnight. The immobilized coatings on the blade were 1.5 cm long and 2.0 mm wide. Sorbents used for this study were categorized into two main groups: silica- and polymer-support base (Table 1). For the purpose of evaluation, 3 replicates of each coating were selected, initially evaluated and later assessed for their extraction efficiency prior to use for further extractions. This pre-screening was carried to minimize variations between coatings. Prior to SPME extractions, all coatings were pre-conditioned overnight in 1:1 (v/v) methanol/water, with agitation on at 150 rpm on a mechanical shaker. Subsequently, to reduce the organic, the coatings were placed for a couple of minutes in an aCSF solution diluted 10× with deionized water content prior to extraction. As part of the objective, the reusability of each coating type was also monitored to ascertain its robustness. 2.1.3. SPME Extraction Procedure All extractions and desorptions were carried out in aCSF and water/acetonitrile 3:2 (v/v) with 0.1% formic acid, respectively. An hour extraction of 50 ng/mL solution of the analytes was carried out under static conditions with each sorbent type; subsequently, the neurotransmitters were desorbed in a 180 L desorption solution in a 300 L amber vial. During the one-hour desorption process, the blades were agitated at 800 rpm on the SK-300 mechanical shaker. After the desorption process, the samples were further subjected to liquid chromatographic separation and tandem MS analysis using the conditions stipulated in the previous section. Details on the chromatographic method are presented in subsequent section.
4
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
Fig. 1. Chromatographic separation of underivatized polar neurotransmitters in a regular reverse phase chromatography mode on a PFP column. GABA and glutamate eluted within the void volume.
2.1.3.1. Extraction of neurotransmitters from CSF and rat brain samples. Approximately 2.0 g of the rat brain samples were weighed into previously cleaned 20 mL vials and subjected to a high speed vortex for about 3–5 min at approximately 30 s intervals to avoid heating of the samples beyond laboratory room temperature. All the brain samples were later pooled into a laboratory petri-dish, covered and vortexed again. After this process, approximately 1.0 g of the macerated brain tissue sample was transferred into 3 separate vials and distinctly spiked at different concentration levels (50 ng/mL and 500 ng/mL). Next, 6 vials were spiked at 5 ng/mL and vortexed for about 2 min. Each of the spiked samples were prepared in replicates of 3 and extracted with the C18 + B SPME blade coating for 30 min, with agitation at 750 rpm. In addition, unspiked macerated brain tissue samples were also extracted and treated as blank correction. After the extraction process, the SPME blades were wiped with KimwipesTM to physically remove the deposits of brain tissue from the surface of the coatings, dipped into deionized water for about 2 s, and then desorbed in 300 L of desorption solution containing 5 ng/mL diazepam as an internal standard. The same set of SPME blades were used for all extractions at a particular concentration level. Prior to the extraction process, preliminary extractions from aCSF in steady state conditions revealed that all the analytes reached equilibrium within 20 min. A relatively longer extraction time was chosen, despite the agitation, to ensure that all analytes reached equilibrium due to the tortuosity of the brain tissue samples. The average area response of each analyte for the blank samples (unspiked samples) was subtracted from their respective responses at each concentration level. A 6-point calibration curve was prepared by extracting spiked samples of aCSF with neurotransmitters at concentrations ranging from 0.1 to 200 ng/mL. A plot of the area ratio of the each analyte to the internal standard (diazepam) versus the nominal concentration of the standard was used for quantification. Extractions were also carried out using the rat CSF obtained from Bioreclamation® . The sample and desorption solution volumes used in this experiment were 750 and 150 L respectively. However, all CSF experiments were carried out in a steady
state and without spiking, and experiments were carried out in triplicate.
3. Results and discussions 3.1. Chromatographic method optimization As a proof of concept, the C18 column used for retaining and separation of the neurotransmitters did not provide any retention of the analytes for isocratic reverse phase separation method with a 90% acetonitrile and 10% water mixture spiked with 0.1% formic acid using a 300 L/min flow rate (data not shown). This experiment was carried out to confirm that retention of the selected neurotransmitters without derivatization in a typical reverse phase C18 column was not feasible as the compounds are likely to elute in the column void volume. Subsequently, two types of PFP stationary phase columns were also evaluated for their effectiveness in separating and retaining the polar molecules, using the same mobile phase system as previously mentioned. However, the Discovery HS F5 gave a huge backpressure at relatively higher flow rates, and thus a flow rate of 300 L/min in a reverse phase chromatographic separation was used. Although retention of separation was attained with the column, the chromatographic runtime was longer (15 min), including column pre-conditioning. In addition, DA, which eluted at about 11.5 min, and had a characteristic broader peak (data not shown). However, the Kinetex® core shell PFP column was able to resolve the peak broadening observed for DA. This observation could be attributed to the relatively smaller particle sizes of the column, and with the core shell technology, was able to withstand higher flow rates with reasonable backpressures. The overall effect was narrower peak shapes, with an improvement in the signal-to-noise ratio for each analyte. Mobiles phases used for the Kinetex® column consisted of A (90% water, 5% methanol, 5% acetonitrile and 0.1% formic acid) and B (10% water, 90% acetonitrile and 0.1% formic acid) at 450 L/min
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
5
Fig. 2. Chromatographic separation of underivatized polar neurotransmitters in “reversed” reverse phase chromatography mode without the need for buffers on a PFP column.
flow rate for both normal and reverse phase modes. The typical reverse phase gradient separation mode, started with 90% A, which was held for 1 min and mobile phase B was increased gradually to 90% by 3 min. The high organic content was maintained for 0.5 min and the column equilibrated for the next 1.5 min. With this gradient approach, DA and glutamate compounds eluted with the volume and most of the analytes were tailing as shown in Fig. 1. From the chromatogram, most of the compounds eluted with the stronger aqueous solvent front due to the high hydrophilicity. Diazepam eluted much later because it is relatively more hydrophobic than the neurotransmitters. To increase the retention capacity of the PFP column, chromatographic separation was started with a mobile phase B (high organic content), followed by a gradual increase in the mobile phase A. Retention and separation was achieved within a 5-min run-time (Fig. 2), starting with 0% mobile phase A, held for 1 min, and then gradually increased to 100% A by 3.5 min. Mobile phase A (100%) was maintained for half a minute before re-conditioning of the column for another minute. Diazepam, used as an internal standard for correction of any potential injection errors, eluted earlier due to its characteristic higher hydrophobicity compared to Fig. 1 where it eluted much later in a typical reverse phase separation method. However, with this particular separation, in a normal phase separation method, the aqueous mobile phase, which acted as the stronger eluting solvent during the chromatographic run-time, effectively separated the highly polar neurotransmitters. In the case of the HILIC column, mobile phase A consisted of 1:1 (v/v) water and ammonium formate buffer with pH adjusted to 3.5, and B contained 95% acetonitrile and 5% ammonium formate. The chromatographic method started with 0% mobile phase A, held for 1 min, gradually increased to 95% A by 4 min, and then the column was re-conditioned for the next minute. Although retention and separation of all the neurotransmitters was attained within the 5 min run-time on the HILIC column (data not shown), an approximately 50-fold decrease in sensitivity was observed compared to the response with the PFP column. The reduction in sensitivity most likely resulted from signal suppression at the MS ion source due to the presence of a high concentration of buffer ions in the mobile
phase, which might have affected the effective ionization of the neurotransmitters in their underivatized forms. 3.2. SPME coatings evaluation The evaluation of SPME coatings focused on the overall extraction efficiency of analytes from the aCSF solution by each coating type. The pH of the aCSF solution was maintained at physiological conditions to mimic a typical biological system. This was done so that the optimized method will be amenable to in vivo applications or biological matrices where the physiological pH cannot be adjusted. With the exception of the ChromabondTM , Clean ScreenTM sorbents, DPA-6STM and C18 particles, the rest were all mixed-mode sorbents on either a silica- or polymer-support base as shown in Table 1. The mixed-mode sorbents were included because of their characteristic multi-interactions of hydrophobic and hydrophilic mechanisms. All the analytes were typically polar, with pKa values ranging from 2.13 for glutamate to 9.8 for 5-HT in aqueous media. 3.2.1. Evaluation of coating procedures To compare the two coating procedures, the extraction efficiencies, robustness, and the inter- and intra-coating reproducibility of the selected fibers were compared. Each coating was prepared in triplicate and used for 5 extractions of glutamate, DA and GABA from aCSF. Fig. 3 shows the percentages of glutamate that were extracted by each sorbent using the two different coating approaches. Of the two methods, coatings made using Kasil 1TM as an adhesive extracted relatively higher amounts of glutamate, DA and GABA. The difference in the amount extracted could be attributed to the difference in particle sizes. Smaller particle sizes (≤10 m) were used with the Kasil 1TM adhesive method, while larger particle sizes (30 m ≤ particle size ≤ 60 m) were used with the Loctite adhesive method. It is important to note that with smaller particle sizes, a larger surface area of the coating would be obtained, and thus improve the overall extraction efficiency. The smaller error bars may also be due to the uniformity of the particle sizes used with the Kasil 1® adhesive compared to the greater variation in the particle sizes used with the Loctite 349 impruvTM
6
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
Fig. 3. Comparison of the efficiencies of two SPME coating methods for the extraction of glutamate from artificial cerebrospinal fluid for selected sorbents (n = 5).
adhesive. Thus, variations between the blades were minimal. The %RSD for the 3 replicates of coatings were respectively 15% and 9% for the Loctite 349 impruvTM and the Kasil 1TM adhesive. Subsequently, all analyses were performed using the Kasil 1TM adhesive coating procedure (Fig. 3). 3.2.2. Evaluation of sorbent extraction efficiency The initial criterion used in the evaluation process involved the pre-selection of sorbent(s) able to extract quantitative amounts of all four neurotransmitters, i.e., the ability of the developed SPME coating to extract analytes with a wide range of varying pKa values. Therefore, the sorbents were initially screened for their ability to extract all 4 neurotransmitters using spiked samples of aCSF containing 50 ng/mL of each analyte. Table 2 shows that most of the mixed-mode sorbents were able to extract all 4 neurotransmitters. Chromabond SA, Clean screen GHB, SSBCX, DPA-6S and C18 sorbents extracted ≤3 analytes with DPA-6S and C18 sorbent extracting only glutamate and DA respectively. This observation may be due to the fact that the DPA-6S and the C18 do not exhibit multiple interaction modes (hydrophobic and hydrophilic interactions) with the analytes. However, in the case of the mixed-mode SPE particles, due to their multiple interactions with compounds, the coatings were able to extract all 4 neurotransmitters in quantitative amounts. Subsequently, the efficiencies of the selected mixed-mode SPME coatings (C18 + B; C8 + B; MCX; MAX; WCX; WAX) were determined and compared by determining the amounts extracted in a separate set of extractions. The comparison process entailed triplicate one hour SPME extractions of 50 ng/mL neurotransmitters in aCSF samples with subsequent desorption for another hour. A second desorption of the same SPME coating was carried out to ascertain any carryover amounts after the initial desorption. The percent amount of each extracted analyte was determined for the selected Table 2 Screening of sorbents used for SPME coatings for their ability to extract neurotransmitters. Sorbent type
Chromabond SA DPA-6S Clean screen DAU Clean screen GHB SSBCX C18 + B C8 + B MCX MAX WCX WAX C18 particles
Neurotransmitter GLU
GABA
DA
5-HT
+ + + + + + + + + + +
+ + + + + + + + + +
+ + + + + + + + + +
n/a n/a n/a + + + + + + n/a
n/a, not available at the time of experiment. (+), extracted quantitative amount; (), detected but not quantitative.
Fig. 4. Comparison of the extraction efficiency of selected mixed-mode SPE sorbents used as SPME coatings for the extraction of neurotransmitters spiked in artificial cerebrospinal fluid.
coatings. Results (Fig. 4) showed that C18 with benzenesulphonic acid group (C18 + B) extracted relatively equal amounts of GABA, DA and 5-HT, with glutamate being the analyte with the highest extracted amount. Glutamate was most extracted the by all the sorbents, while DA relatively least extracted. With the exception of C18 + B, there were no significant differences in the amounts of GABA and glutamate extracted by the MAX and C8 + B sorbents. MCX sorbent showed relatively higher extraction efficiency for 5HT; however, there was no significant difference in the amount extracted when compared to that of the C18 + B sorbent. Whereas WCX sorbent did not show any significant difference in its extraction efficiency for all the analytes, WAX showed higher extraction efficiency for GABA and glutamate only, despite the fact that both sorbents had similar weak mixed-mode ion exchange properties. Relatively, mixed-mode coatings with relatively stronger ion exchange properties performed better in the extraction of the polar neurotransmitters compared to their corresponding sorbents with weak ion exchange properties. This may be due to the fact that hydrophilic interactions between the coatings and the analytes were stronger. However, the changes in the amount extracted by the sorbents were not significantly different. Finally, in terms of base support (silica- or polymer-based sorbents), there were no observable patterns in the extraction efficiencies of the sorbents for these analytes. This implies that the extraction efficiency of the sorbent was not necessarily dependent on the type of base support. 3.3. Application of mixed-mode coatings for extraction of selected neurotransmitters from CSF and rat brain tissue samples The C18 + B SPE coating was selected among the mixed-mode particles and applied for the extraction of all 4 neurotransmitters in two biological matrices (CSF and rat brain tissue). To ensure method validity, the extractions were carried out with the brain samples were at three different concentration levels. Results obtained are shown in Table 3. The precision (%RSD) for each analyte at each concentration level determined was less than 12% (n = 3) for the rat brain tissue, which is acceptable for sample preparation processes for complex Table 3 Results obtained for the extraction of spiked rat brain tissue sample and CSF. Neurotransmitter (ng/mL) GA Spiked brain samples
50 ng/mL (n = 3) 48.0 (4) 500 ng/mL (n = 3) 510 (11) 5 ng/mL (n = 6)
CSF sample nd: not detected.
GABA
DA
5-HT
45.0 (6) 493 (9)
52.0 (5) 505 (8)
43.0 (4) 496 (10)
5.50 (1.2) 4.80 (1.0) GABA GA
5.10 (0.7) 4.50 (1.0) DA 5-HT
0.92 (0.1) 0.53 (0.06) nd
nd
E. Cudjoe, J. Pawliszyn / J. Chromatogr. A 1341 (2014) 1–7
biological matrices. The analytical figures of merit determined were the limit of detection and quantitation, the linear range using neat solutions of standards prepared in desorption solution, and method robustness. With the exception 5-HT, which gave 86% accuracy for extractions using 50 ng/mL and 500 ng/mL rat brain samples, accuracy all the analytes were ≥90% for 3 replicate extractions. The RDS% obtained for 6 replicates of the lowest concentration (5 ng/mL) spiked brain samples were used to measure method robustness. The linear range was determined to be between 0.01 and 150 ng/mL for all the analytes, except for GABA, which was from 0.1 to 100 ng/mL. The limit of detection range was from 6 to 10 pg/mL for all the neurotransmitters. Glutamate and GABA recorded 6 and 10 pg/mL LODs whereas DA and 5-HT were the same 8 ng/mL. The limits of quantitation (LOQ) were in the range of 20–35 pg/mL. The LODs and LOQs were determined based on 3× the S/N of a blank extraction of the aCSF and the LOQ was 10× the S/N of the blank extract. The method sensitivity, reliability and robustness have also been demonstrated elsewhere in an in vivo extractions of neurotransmitters in the brain extracellular fluid of freely moving rats [28]. 3.4. Conclusion The mixed-mode sorbents showed superior extraction efficiencies over other coatings in their ability to quantitatively extract polar neurotransmitters of varying polarities from biological matrices. By combining a “reversed” reverse-phase chromatographic separation on a PFP stationary phase column, a robust LC–MS/MS quantitative method was developed for the retention, separation and analysis of neurotransmitters without analyte derivatization. The project has demonstrated that by choosing the appropriate sorbent, the SPME-LC–MS/MS technique is a viable alternative bioanalytical method for the quantitative determination of both endogenous and exogenous analytes in various biological matrices.
7
References [1] G. Venkatasami, J.R. Sowa, Anal. Chim. Acta 665 (2010) 227. [2] H. Sun, L. Wang, L. Ai, S. Liang, H. Wu, Food Control 21 (2010) 686. ´ [3] A. Lebiedzinska, M. Da¸browska, P. Szefer, M. Marszałł, Toxicol. Mech. Methods 18 (2008) 463. [4] European Food Safety Authority, Scientific opinion on the safety and efficacy of folic acid as a feed additive for all animal species, EFSA J. 10 (5) (2012) 2674. [5] P. Uutela, L. Karhu, P. Piepponen, M. Käenmäki, R.A. Ketola, R. Kostiainen, Anal. Chem. 81 (2009) 427. [6] E. Herlenius, H. Lagercrantz, Early Hum. Dev. 65 (2001) 21. [7] R.W. Olsen, GABA, in: Neuropsychopharmacol. Fifth Gener. Prog., Lippincott Williams & Wilkins, 2002. [8] M. Perry, Q. Li, R.T. Kennedy, Anal. Chim. Acta 653 (2009) 1. [9] N. Fontanals, R.M. Marcé, F. Borrull, J. Chromatogr. A 1152 (2007) 14. [10] V.R. Panditi, S.R. Batchu, P.R. Gardinali, Anal. Bioanal. Chem. 405 (2013) 5953. [11] L.G. Freitas, C.W. Götz, M. Ruff, H.P. Singer, S.R. Müller, J. Chromatogr. A 1028 (2004) 277. [12] R. Carabias-Martínez, E. Rodríguez-Gonzalo, P. Revilla-Ruiz, J. Chromatogr. A 1056 (2004) 131. [13] M. Smith, G.E. Collins, J. Wang, J. Chromatogr. A 991 (2003) 159. [14] J. Wu, J. Pawliszyn, J. Chromatogr. A 909 (2001) 37. [15] J. Wu, H.L. Lord, J. Pawliszyn, H. Kataoka, J. Microcolumn Sep. 12 (2000) 255. [16] R. Wissiack, E. Rosenberg, M. Grasserbauer, J. Chromatogr. A 896 (2000) 159. [17] D. Djozan, S. Bahar, Chromatographia 59 (2004) 95. [18] H. Kataoka, J. Pawliszyn, Chromatographia 50 (1999) 532. [19] H. Bagheri, A. Salemi, Chromatographia 59 (2004) 501. [20] M.E.C. Queiroz, E.B. Oliveira, F. Breton, J. Pawliszyn, J. Chromatogr. A 1174 (2007) 72. [21] D. Vuckovic, I. de Lannoy, B. Gien, Y. Yang, F.M. Musteata, R. Shirey, L. Sisisky, J. Pawliszyn, J. Chromatogr. A 1218 (2011) 3367. [22] N. Raikos, G. Theodoridis, E. Alexiadou, H. Gika, H. Argiriadou, H. Parlapani, H. Tsoukali, J. Sep. Sci. 32 (2009) 1018. [23] J. Henriksen, P. Roepstorff, L.H. Ringborg, Carbohydr. Res. 341 (2006) 382. [24] J.F. Stobaugh, A.J. Repta, L.A. Sternson, J. Pharm. Biomed. Anal. 4 (1986) 341. [25] F.D. Swanek, B.B. Anderson, A.G. Ewing, J. Microcolumn Sep. 10 (1997) 185. [26] F. Beaudry, S.A. Guénette, A. Winterborn, J.-F. Marier, P. Vachon, J. Pharm. Biomed. Anal. 39 (2005) 411. [27] B. Buszewski, S. Noga, Anal. Bioanal. Chem. 402 (2012) 231. [28] E. Cudjoe, B. Bojko, I. de Lannoy, V. Saldivia, J. Pawliszyn, Angew. Chem. Int. Ed. 52 (2013) 12124.
