Design and implementation of an automated liquid-phase microextraction-chip system coupled on-line with high performance liquid chromatography

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ “This is an Accepted Manuscript of an article published by Elsevier in: TALANTA on 2014, available at: https://doi.org/10.1016/j.talanta.2013.12.016” 1 Design and implementation of an automated liquid-phase 1 microextraction-chip system coupled on-line with high 2 performance liquid chromatography 3 4 Bin Lia, Nickolaj Jacob Petersena,*, María D. Ramos Payánb, Steen Honoré Hansena, and Stig 5 Pedersen-Bjergaarda,c 6 a Department of Pharmacy, Faculty of Health and Medical Sciences, University of Copenhagen, 7 2100 Copenhagen, Denmark 8 b Department of Analytical Chemistry, Faculty of Chemistry, University of Seville, P.O. Box 9 41012, Seville, Spain 10 c School of Pharmacy, University of Oslo, P.O Box 1068 Blindern, 0316 Oslo, Norway 11 12 13 14 Keywords: Automated liquid-phase microextraction 15 On-line sample pretreatment 16 Microfluidic chip 17 High performance liquid chromatography 18 Opium alkaloids 19 20 *Corresponding author at: Department of Pharmacy, Faculty of Health and Medical Sciences, 21 University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark, Tel.: +45 22 35336184; E-mail address: nickolaj.peter[email protected] 23 24 2 Abstract 25 An automated liquid-phase microextraction (LPME) device in a chip format has been developed 26 and coupled directly to high performance liquid chromatography (HPLC). A 10-port 2-position 27 switching valve was used to hyphenate the LPME-chip with the HPLC autosampler, and to 28 collect the extracted analytes, which then were delivered to the HPLC column. The LPME-chip-29 HPLC system was completely automated and controlled by the software of the HPLC instrument. 30 The performance of this system was demonstrated with five alkaloids i.e. morphine, codeine, 31 thebaine, papaverine, and noscapine as model analytes. The composition of the supported liquid 32 membrane (SLM) and carrier was optimized in order to achieve reasonable extraction 33 performance of all the five alkaloids. With 1-octanol as SLM solvent and with 25 mM sodium 34 octanoate as anionic carrier, extraction recoveries for the different opium alkaloids ranged 35 between 17 and 45 %. The extraction provided high selectivity, and no interfering peaks in the 36 chromatograms were observed when applied to human urine samples spiked with alkaloids. The 37 detection limits using UV-detection were in the range of 1-21 ng/mL for the five opium alkaloids 38 presented in water samples. The repeatability was within 5.0-10.8 % (RSD). The membrane 39 liquid in the LPME-chip was regenerated automatically between every third injection. With this 40 procedure the liquid membrane in the LPME-chip was stable in 3-7 days depending on the 41 complexity of sample solutions with continuous operation. With this LPME-chip-HPLC system, 42 series of samples were automatically injected, extracted, separated, and detected without any 43 operator interaction. 44 45 46 3 1. Introduction 47 The high complexity of biological samples and low concentrations of target analytes are two 48 of the main challenges for analytical detection and quantitation. Therefore, clean-up and 49 enrichment procedures in order to resolve those analytical limitations are important, preferably in 50 an automated way that is able to handle low sample volumes. For many years, liquid-liquid 51 extraction (LLE), solid phase extraction (SPE), and solid-phase microextraction (SPME) have 52 been the standard methods for sample preparation [1, 2]. In recent years, substantial interest has 53 also been devoted to extractions across supported liquid membranes (SLM) where an organic 54 liquid is immobilized in the pores of a porous hydrophobic membrane. Analytes of interest can be 55 selectively extracted across the SLM driven by either a pH gradient as used in the format of 56 liquid-phase microextraction (LPME) [3-6] or a voltage gradient termed electromembrane 57 extraction (EME) [7]. With LPME or EME, membrane microextraction has demonstrated a 58 significant potential in pharmaceutical analysis [8], environmental [9-11] and food analysis [12]. 59 Due to the high versatility of SLM based extraction techniques, they are readily incorporated 60 into different platforms and coupled directly with high performance analytical instruments such 61 as liquid chromatography (LC) [13], gas chromatography (GC) [14], capillary electrophoresis 62 (CE) [15], or flame atomic absorption spectrometry (AAS) [16]. Chip-based SLM- systems have 63 been explored and coupled on-line with LC since the 1980s [17] due to their significant 64 advantages in terms of miniaturization and automation [18]. Previous SLM-chip modules were 65 made by packing a flat sheet membrane in between two grooved polymer holders, which were 66 then clamped with bolts. The volume of the channels was generally in the range of 10-20 μL [19, 67 20]. The automated SLM-chip systems have been explored and applied for a wide range of 68 biosamples, such as anaesthetics (SLM-GC) [21], bambuterol in human plasma (SLM-CE) [22], 69 and peptides in spiked plasma (SLM-HPLC) [19]. 70 Recently, SLM extraction has been successfully downscaled to a microfluidic chip for sample 71 enrichment and clean-up [23-26]. The advantages of such microchip membrane extraction 72 include minimal organic solvent consumption, the ability to handle a wide range of sample 73 volumes, ease of use, potentially high enrichment factors from small sample volumes, and the 74 ability to provide selective extraction of analytes depending on their polarity and charge. The 75 chemical binding of flat sheet membranes into polymethyl methacrylate (PMMA) blocks was 76 4 developed in our group and high performance of this SLM-chip unit has been demonstrated by 77 both EME [24, 25, 27] and LPME [23] work reported previously. In this microchip membrane 78 extraction module, the sample solution was pumped into a 50 µm deep micro channel where the 79 analytes were extracted through the SLM and into an acceptor channel located on the other side 80 of the SLM. The driving force for the microchip membrane extraction was either a DC electrical 81 potential [25], or a pH gradient [23]. With microchip membrane extraction, dynamic extraction 82 was performed in which the samples were delivered continuously to the chip by a microsyringe 83 pump. The enrichment factor (EF) was controlled by the ratio of the sample volume delivered to 84 the device and the volume of the acceptor solution that could either be stationary (stopped flow) 85 or delivered continuously [23, 24]. In addition, in the microchip EME system, the EF was also 86 controlled by the applied extraction voltage [24]. Both the microchip EME and LPME systems 87 have been used for online and real-time measurement of in vitro metabolism of drug substances 88 by rat liver microsomes [23, 27]. 89 The objective of this study was to integrate a microchip LPME system directly to a 90 commercial high performance liquid chromatography (HPLC) system, and to fully automate the 91 system. This report describes the design, construction, operation, and optimization of such a 92 LPME-chip-HPLC system. The system was developed to automatically perform sample injection, 93 LPME, SLM liquid regeneration, and fast HPLC separation. Different alkaloids were used as 94 model analytes. The intention was not to develop an analytical method for the alkaloids, but 95 rather to investigate fundamental aspects of the LPME-chip-HPLC system. 96 97 2. Experimental 98 2.1 Chemicals and sample solutions 99 Morphine (pKa (base) = 8.2, pKa (acid) = 9.7; log P = 0.89) was obtained from Nycomed 100 DAK (Copenhagen, Denmark), codeine (pKa = 8.2; log P = 1.19) and noscapine (pKa = 6.3; log P 101 = 1.5) were obtained from Nordisk Droge and Kemikalie (Copenhagen, Denmark), thebaine (pKa 102 = 8.4; log P = 2.0) was obtained from Nomeco (Copenhagen, Denmark), and papaverine (pKa = 103 6.3; log P = 3.0) was obtained from Mecobenzon (Copenhagen, Denmark). All these substances 104 were hydrochlorides and with purities >99 %. LC-MS grade formic acid, acetonitrile, 105 tricaprylmethyl-ammonium chloride (Aliquat 336), and sodium octanoate were purchased from 106 Sigma–Aldrich (St. Louis, MO, USA). 1-Octanol and 2-nitrophenyl octyl ether (NPOE) were 107 5 obtained from Fluka (Buchs, Switzerland). All water used was prepared with a Millipore Direct-108 Q3 UV system (Billerica, MA, USA). 109 Stock solutions containing 1 mg/mL of each model analyte were prepared in 10 % (v/v) 110 acetonitrile in 100 mM HCOOH and stored protected from light at 277 K (4°C). Sample solutions 111 of the compounds were prepared daily by adequate dilutions from the 1 mg/mL stock solutions 112 by pure water or urine. 113 114 2.2 Instrumentation of the automated LPME-chip-HPLC 115 As show in Figure 1A, the integrated LPME-chip-HPLC consisted of three main parts: (1) an 116 Agilent 1100 HPLC system (Agilent, Palo Alto, CA, USA) including an autosampler (model 117 G1329A), a binary pump system (G1312A), and a UV detector (G1314A); (2) the home-built 118 LPME-chip attached to a Valco Instrument (EHAM model, Houston, TX, USA) two position 10-119 port valve actuator control module; and (3) two microsyringe pumps (Kd Scientific, Holliston, 120 MA). The HPLC software (Chemstation B.04.02) was applied for programming the sample 121 injection, separation, and UV detection. The 10-port valve was used to synchronize the sample 122 pretreatment, and to separate the low pressure of LPME-chip module from the high pressure of 123 HPLC system by switching the positions between sample loading and injection. The automated 124 operation of this 10-port valve was controlled by the remote control output (RS232 plug) on the 125 HPLC system. The valve switching flow diagram for the LPME-chip-HPLC system is 126 schematically illustrated in Figure 1B. In this setup, the two switch positions A and B in the 10-127 port valve were alternatively changed for microchip LPME sample pretreatment and on-line 128 HPLC analysis, respectively. The two microsyringe pumps were used to deliver the sample 129 carrier buffer solution and the acceptor phase through the LPME-chip for the dynamic extraction, 130 respectively. 131 The construction of the LPME-chip was published recently and only a short description is 132 given here [23, 27]. The porous polypropylene membrane (Celgard 2500 micro porous membrane; 133 Celgard, Charlotte, NC, USA) with a 25 μm thickness (55% porosity, and 0.21 μm × 0.05 μm 134 pores) used for the SLM was placed between two polymethyl methacrylate (PMMA) (53 mm × 135 53 mm × 2.1 mm) plates having 6 mm long channels with a depth of 50 μm and a width of 2.00 136 mm. The whole assembly was fixed by solvent-assisted bonding with ethanol and cured in a 137 6 343K (70°C) oven. At both ends of the channels, 1.6 mm I.D. holes were drilled through the plate 138 to serve as inlet and outlet for the sample carrier liquid and the acceptor phase. 139 Prior to connection of the tubing to the chip, the supported liquid membrane was immobilized 140 in the polypropylene membrane by filling approximately 0.2 μL of organic solvent (1-octanol or 141 NPOE) into one end of the extraction channel using a micropipette. The solvent immediately 142 immobilized into the polypropylene membrane by capillary forces, and this process was visually 143 inspected as the appearance of the membrane changed from white to transparent during 144 immobilization of membrane liquid. Subsequently, the tubings for the donor and acceptor flow 145 were connected to the LPME-chip. 146 147 2.3 Procedure of carrier mediated LPME-chip-HPLC 148 Sample was loaded in the autosampler tray of the HPLC instrument in 2 mL LC vials 149 (Microlab, Aarhus, Denmark), and extractions were normally carried out according to the 150 following procedure; sample solution was prepared by the mixture of 500 μL analyte solution 151 with 500 μL 50 mM sodium octanoate (ion-pair reagent) prepared in 25 mM pH 7.0 phosphate 152 buffer. By means of the autosampler, 50 μL sample solution was draw into the injection needle 153 and then directed back to the HPLC needle seat connected with the HPLC 6-port valve (all part of 154 the Agilent® autosampler) (Figure 1A). The microsyringe on the donor side was connected with 155 the 6-port valve of the autosampler, and was filled with 25 mM phosphate buffer (pH 7.0) used as 156 the sample carrier liquid. With a flow rate of 5 μL/min, the sample plug was transferred to the 157 LPME-chip and the analytes were extracted through the SLM. The tubings used for connecting 158 the LPME-chip to the autosampler had a small dead volume in the order of 1 µL, also on the 159 accepter side there was a small dead volume of approximately 0.5 µL for transferring the extracts 160 to the 5 µL HPLC loop. The total extraction time was set for the requirement of total injected 161 sample reached the chip, and also that the extract was transferred to the 5 µL sample loop. The 162 time delay due to the dead volume was taken into account. Therefore, the on-chip membrane 163 microextraction of 50 µL sample solution will take 15 minutes with a flow rate of 5 μL/min on 164 the donor side. During the 15 minutes extraction, the analytes extracted into the accepter solution 165 were continuously delivered to the 5 μL sample loop by the continuous flow of acceptor phase 166 (50 mM HCl), which was pumped with a second microsyringe pump on the acceptor side at a 167 flow rate of 0.5 μL/min. 168 7 Analysis of the extracts, collected by the loop, was performed by switching the 10-port valve 169 to position B after 15 min extraction (Figure 1B). The software triggered the valve switching. In 170 position B the mobile phase from the HPLC pump was directed to the 5 μL sample loop inserted 171 on 10-port valve and thereby the enriched analytes was transferred into the HPLC column. 172 173 2.4 HPLC equipment 174 An Agilent 1100 series HPLC system (Agilent Technologies) was applied for the on-line 175 LPME analysis as described in section 2.2. The LC separation was performed on a Zorbax 176 Eclipse XDB-C18 column (Agilent Technologies) (4.6 × 50 mm, 1.8 μm particle size). The flow 177 rate was 0.8 mL/min. Using a short column packed with 1.8 μm porous particles coupled with 178 HPLC will shorten the analysis time without loss of the separation resolution compared to 179 traditional 3-5 μm based columns [28]. Acidified water (100 mM HCOOH) and ACN were used 180 as the mobile phases A and B, respectively. The solvent gradient adopted was as follows: 5% B at 181 0-2 min, 5-20% B at 2–4 min, 20% B at 4-10 min, 20–100% B at 10-12 min, followed by wash 182 and equilibration. The analytes were detected using a UV detector at 282 nm. Baseline separation 183 of the opium alkaloids was obtained in 10 min. The 10-port valve completely separated the low 184 pressure of LPME-chip device from the HPLC system and the organic mobile phase never came 185 in contact with the polymer chip. Acetonitrile in the mobile phase would otherwise have 186 dissolved the PMMA and also modified the SLM used for the extraction. 187 188 2.5 Extraction efficiency 189 To determine the extraction efficiency of the model analytes, 50 μL of diluted standard 190 solution with 25 mM sodium octanoate was extracted as described above. The same standard 191 solution (5 μL unextracted) was also injected directly into the HPLC. Percentage recovery (R %) 192 was calculated as follows: 193 𝑅𝑅=𝑉𝑉 𝑎𝑎 ∙ 𝐶𝐶𝑎𝑎 𝑓𝑓𝑓𝑓𝑓𝑓𝑎𝑎𝑓𝑓 𝑉𝑉 𝑠𝑠 ∙ 𝐶𝐶𝑠𝑠 𝑓𝑓𝑓𝑓𝑓𝑓𝑖𝑖𝑓𝑓𝑎𝑎𝑓𝑓 194 Where Va is the volume of accepter solution (5 µL) injected into the HPLC having the 195 concentration Ca final. Vs is the sample volume (50 µL) injected from the autosampler into the 196 LPME device having the concentration Cs initial. Compared to the normal way of calculating the 197 recovery for LPME, this equation also takes into account that not all the extracted compounds are 198 8 collected by the HPLC injection loop and the calculated recoveries reflect the amount of analytes 199 collected and analysed by the HPLC. 200 The enrichment factor (EF) for the analyte was calculated according to the following equation: 201 𝐸𝐸𝐸𝐸 =𝐶𝐶𝑎𝑎 𝑓𝑓𝑓𝑓𝑓𝑓𝑎𝑎𝑓𝑓 𝐶𝐶𝑠𝑠 𝑓𝑓𝑓𝑓𝑓𝑓𝑖𝑖𝑓𝑓𝑎𝑎𝑓𝑓 202 Where Ca final is the concentration of the enriched analyte solution injected into the HPLC and 203 Cs initial is the concentration of analytes in the untreated sample. Since only 50 µL of sample was 204 injected into the LPME device and the accepter volume injected into the HPLC was 5 µL, the 205 theoretical maximum enrichment factor was 10. 206 A standard curve using the LPME-chip–HPLC system was constructed for all analytes. For 207 calculating the recovery, the standards prepared in 100 mM formic acid were injected directly 208 into the 5 µL loop on the 10-port valve by filling the standard solution directly into the loop with 209 a microsyringe. 210 211 3. Results and discussion 212 3.1 Principle of operation 213 The primary purpose of coupling the LPME-chip directly to the HPLC was to provide on-line 214 clean-up, enrichment, and analysis in micro-scale without time-consuming off-line sample 215 preparation. All autosamplers that allow the control of an external valve can be used in 216 combination with the LPME-chip. In this work the entire system was controlled from the Agilent 217 Chemstation software. The basic setup and a photo of the automated LPME-chip-HPLC system 218 are illustrated in Figure 1. Initially the HPLC 6-port valve was in the bypass position and the 219 donor phase pumped directly through the chip. Meanwhile, the 10-port valve was in position A, 220 where HPLC mobile phase was passed directly through the HPLC column. When the LPME 221 process was initiated, the HPLC 6-port valve was triggered by the injection program and 222 switched to the mainpass position (Table 1). In this mainpass position, the donor phase was 223 directed through the injection needle and delivered 50 μL sample solution directly towards the 224 chip. The analytes were extracted across the SLM and into the acceptor phase. The acceptor 225 phase was continuously pumped into a 5 μL sample loop with a flow rate of 0.5 μL/min. After 15 226 minutes of LPME, the 10-port valve was programmed to switch to the position B, and the mobile 227 phase was switched to pass by the sample loop. Thus, sample injection, transportation, membrane 228 15 Table 3 371 Analytical performance of LPME-chip-HPLC system 372 Analyte Calibration curve a Linear range (μg/mL) R2 LOD (μg/mL) morphine y = 0.5005x + 0.174 0.1-10 0.9998 0.021 codeine y = 0.8842x + 0.1847 0.1-10 0.9999 0.021 thebaine y = 10.932x - 0.7379 0.05-5 0.9969 0.001 papaverine y = 8.4735x + 0.785 0.01-5 0.9991 0.001 noscapine y = 2.1787x - 0.2943 0.1-10 0.9959 0.021 373 a y: peak area (mAU), x: sample concentration (μg/mL) 374 SLM: 1-octanol, Injection volume: 50 μL. 375 Sample: 0.1-10 μg/mL of the five opiates prepared in 25 mM sodium octanoate, pH 7.0. 376 Donor phase: 5 μL/min, 25 mM phosphate buffer pH 7.0, 377 Acceptor phase: 0.5 μL/min, 50 mM HCl; extraction time: 15 min. 378 379 380 16 Figure caption 381 382 Figure 1 Photo (A) and schematic illustration (B) of automated LPME-chip-HPLC system. In 383 Fig 1A, left panel (1) presented the overview of complete LPME-chip-HPLC device, and the 384 right two panels showed the close-up view of the autosampler 6-port valve (2) as well as the chip 385 system directly coupled to the external automated 10-port valve (3). 386 387 Figure 2 On LPME-chip-HPLC for spiked human urine. a) Direct HPLC analysis of spiked urine 388 sample. b) Spiked urine sample after 15 min extraction on the LPME-chip-HPLC. Urine sample: 389 spiked with 5 opiates at 5 μg/mL; SLM liquid: 1-octanol; injection volume 50 μL; donor phase:, 390 5 μL/min, 25 mM phosphate buffer (pH 7.0); acceptor phase: 0.5 μL/min, 50 mM HCl. 391 392 Figure 3 Repeatability tests of system with 18 runs were conducted in 3 days (6 runs per day) 393 with the same channel cleaned with ethanol and dried after extraction every day. Urine sample: 394 spiked with 5 opiates at 2.5 μg/mL; SLM liquid: 1-octanol; injection volume 50 μL; donor phase: 395 5 μL/min, 25 mM phosphate buffer (pH 7.0); acceptor phase: 0.5μL/min, 50 mM HCl; extraction 396 time: 15 min. 397 398 399 17 400 401 Figure 1 402 403 18 404 Figure 2 405 406 19 407 Figure 3 408