Anal. Chem. 1998, 70, 3304-3314
Analysis of Nicotine and Its Oxidation Products in Nicotine Chewing Gum by a Molecularly Imprinted Solid-Phase Extraction A ° sa Zander,† Paul Findlay,‡ Thomas Renner,§ and Bo 1 rje Sellergren*
Department of Inorganic Chemistry and Analytical Chemistry, Johannes Gutenberg University Mainz, J.J.-Becherweg 24, 55099 Mainz, Germany Aleksander Swietlow⊥
Pharmacia & Upjohn, Consumer Healthcare, Box 941, S25109 Helsingborg, Sweden
Chromatographic stationary phases showing exceptional selectivity for nicotine can be prepared by the technique of molecular imprinting. Such phases were used in the search for a rapid cleanup step for nicotine and some of its oxidation products in chewing gum formulations. Thus, using an organic mobile phase, the nicotine analytes from chewing gums dissolved in nonpolar solvent were retained, whereas the nonpolar matrix eluted close to the void peak. A subsequent switch to an acidic mobile phase resulted in elution of the analytes as one sharp peak. Due to weak binding of the less basic oxidation products, other imprinted polymers were tested, and the solid-phase extraction procedure was optimized. Polymers were prepared using various functional and cross-linking monomers, templates, porogens and thermal treatments. This resulted in phases that, when compared with a nonimprinted or a C18 reversed-phase column, showed significantly higher recoveries of the analytes. Furthermore, no bleeding of template from the phases could be detected. The cleanup step was coupled off-line to reversed-phase HPLC, and the efficiency of the analysis was compared with and without the cleanup step. Three out of four analytes were quantitatively recovered using the imprinted phase, whereas, using the nonimprinted phase, only nicotine was recovered. Without the cleanup step, none of the analytes could be determined using the reversedphase HPLC method.
printed materials3-5 have been used for chiral recognition of a variety of small molecules, including therapeutic drugs, recognition of sugars, nucleotide bases, and pesticides as well as steroid and peptide hormones. The high affinity and selectivity for the target analyte exhibited by some of the imprinted materials have justified a comparison with the corresponding immunoaffinity phases.6-8 In contrast to the latter phases, however, the materials are straightforward to prepare, stable in most media, and reusable over long periods of time. Applications of the materials in chromatography, chemical sensing, or specific assays are, therefore, being investigated. A particularly promising application is solid-phase extraction (SPE)9 of analytes present in low concentrations or in complex matrixes. This may lead to selective enrichments and cleanup of the analytes to levels not achievable with existing methods. During recent years, molecularly imprinted solid-phase extractions (MISPE) have been used in bioanalysis,10-13 food analysis,14 and environmental analysis.15 In these cases, selective enrichment and cleanup of the analyte is obtained, resulting in a higher accuracy and a lowering of the detection limit in the subsequent chromatographic quantification. In most examples of recognition with imprinted phases, the highest selectivity is seen in solvents of low to medium polarity, where electrostatic forces dominate the binding.3-5 Changing to an aqueous medium, the binding and selectivity decreases for hydrophilic templates, whereas, for hydrophobic templates, a specific or nonspecific hydrophobic
Present address: Ferring AB, P.O. Box 30047, 20061 Malmo¨, Sweden. Present address: Department of Pure and Applied Chemistry, University of Strathclyde, Thomas Graham Building, 295 Cathedral St., Glasgow G11XL, UK. § Present address: ABF, Goethestrasse 20, 80336 Mu ¨ nchen, Germany. ⊥ Present address: Analysis Department, Polypeptide Laboratories (Sweden) AB, P.O. Box 30047, 20061 Malmo¨, Sweden. (1) Wulff, G. Angew. Chem., Int. Ed. Engl. 1995, 34, 1812. (2) Shea, K. J. Trends Polym. Sci. 1994, 2, 16. (3) Mosbach, K. Trends Biochem. Sci. 1994, 19, 9.
(4) Sellergren, B. Trends Anal. Chem. 1997, 16, 310. (5) Mayes, A. G.; Mosbach, K. Trends Anal. Chem. 1997, 16, 321. (6) Vlatakis, G.; Andersson, L. I.; Mu ¨ ller, R.; Mosbach, K. Nature 1993, 361, 645. (7) Andersson, L. I.; Mu ¨ ller, R.; Vlatakis, G.; Mosbach, K. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 4788. (8) Andersson, L. I. Anal. Chem. 1996, 68, 111. (9) Berrueta, L. A.; Gallo, B.; Vicente, F. Chromatographia 1995, 40, 474. (10) Sellergren, B. Anal. Chem. 1994, 66, 1578. (11) Andersson, L. I.; Paprica, A.; Arvidsson, T. Chromatographia 1997, 46, 57. (12) Martin, P.; Wilson, I. D.; Morgan, D. E.; Jones, G. R.; Jones, K. Anal. Commun. 1997, 34, 45. (13) Rashid, B. A.; Briggs, R. J.; Hay, J. N.; Stevenson, D. Anal. Commun. 1997, 34, 303-305. (14) Muldoon, M. T.; Stanker, L. H. Anal. Chem. 1997, 69, 803. (15) Matsui, J.; Okada, M.; Tsuruoka, M.; Takeuchi, T. Anal. Commun. 1997, 34, 85.
3304 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
S0003-2700(97)01272-9 CCC: $15.00
During recent years, numerous reports of selective recognition of small molecules with materials prepared by molecular imprinting (MIPs) have appeared.1-5 Noncovalently molecularly im† ‡
© 1998 American Chemical Society Published on Web 06/23/1998
Scheme 1
Table 1. Polymers Prepared for Use in Nicotine Analysis and Composition of the Monomer Mixtures Prior To Polymerizationa polymer PMAA(1,2) PMeCN PMAAhb PMAABL PTFM PTFMBL PTRIM PTRIMBL PCOT PMYO PβNIC
template NIC (1 mmol) NIC (1 mmol) NIC (1 mmol) NIC (1 mmol) NIC (4.3 mmol) COT (0.4 mmol) MYO (0.4 mmol) βNIC (0.4 mmol)
monomer
cross-linker
porogen
MAA (4 mmol) MAA (4 mmol) MAA (4 mmol) MAA (4 mmol) TFM (4 mmol) TFM (4 mmol) MAA (27.8 mmol) MAA (27.8 mmol) MAA (4 mmol) MAA (4 mmol) MAA (4 mmol)
EDMA (20 mmol) EDMA (20 mmol) EDMA (20 mmol) EDMA (20 mmol) EDMA (20 mmol) EDMA (20 mmol) TRIM (27.8 mmol) TRIM (27.8 mmol) EDMA (20 mmol) EDMA (20 mmol) EDMA (20 mmol)
CH2Cl2 (5.6 mL) MeCN (5.6 mL) CH2Cl2 (5.6 mL) CH2Cl2 (5.6 mL) CH2Cl2 (5.6 mL) CH2Cl2 (5.6 mL) CH2Cl2 (20 mL) CH2Cl2 (20 mL) CH2Cl2 (5.6 mL) CH2Cl2 (5.6 mL) CH2Cl2 (5.6 mL)
a The polymers were prepared as described in the Experimental Section using the above monomer compositions. MAA, methacrylic acid; EDMA, ethyleneglycol dimethacrylate; TFM, trifluoromethylacrylic acid; TRIM, trimethylolpropanetrimethacrylate (see Scheme 1 for structures); COT, cotinine; MYO, myosmine; β-NIC, β-nicotyrine (see Chart 1 for structures). b Polymer PMAA2 heat-treated at 120 °C under vacuum for 24 h followed by column repacking.
contribution to the binding can be seen.16 The latter depends on the structure and hydrophobicity of the template. The MISPE step can, therefore, rely on a selective adsorption step, as can be expected in the extraction of analytes in media of low polarity,11,14 or a selective desorption step,12,13,15 as in the extraction of hydrophobic analytes from aqueous media. We report here an example where MISPE could be used for the analysis of pharmaceutical formulations. Some of the latter are soluble only in nonpolar media and, thus, are suited for a direct MISPE step coupled with an analytical HPLC quantification. We have evaluated this possibility as a cleanup step in the analysis of nicotine and some of its oxidation products in nicotine chewing gum. This is the first example of the use of MISPE for multianalyte analysis. MATERIALS AND METHODS Chemicals. The monomers methacrylic acid (MAA) (Aldrich), trimethylolpropane trimethacrylate (TRIM) (Aldrich), and ethyleneglycol dimethacrylate (EDMA) (Fluka) and the solvents were purified as previously described.24 Trifluoromethylacrylic (16) Dauwe, C.; Sellergren, B. J. Chromatogr. 1996, 753, 191. (17) Sudan, B. J. L.; Brouillard, C.; Strehler, C.; Strub, H.; Sterboul, J.; SainteLaudy, J. J. Chromatogr. 1984, 288, 415. (18) Dash, A. K.; Wong, S.-T. J. Chromatogr. 1996, 749, 81-85. (19) Gru ¨ n, M.; Kurganov, A. A.; Schacht, S.; Schu ¨ th, F.; Unger, K. K. J. Chromatogr. 1996, 740, 1-9. (20) Watson, I. D. J. Chromatogr. 1977, 143, 203-206. (21) Matsui, J.; Miyoshi, Y.; Takeuchi, T. Chem. Lett. 1995, 1007.
acid (TFM) was purchased from Aldrich and used as received. The initiator azobis(isobutyronitrile) (AIBN) was obtained from Janssen and purified by recrystallization from methylene chloride. (S)-(-)-Nicotine in the free base form (Aldrich) was purified by distillation and used as template. The nicotine standards were prepared using nicotine hydrogen (+)-tartrate dihydrate (BDH Chemicals Ltd.). The nicotine oxidation products and the chewing gum samples were provided by Pharmacia & Upjohn and used as received. The HPLC-grade solvents were purchased from either Aldrich or Merck, and the water was collected from a Millipore water purification system. Equipment. The UV lamp used in the photopolymerization was a high-pressure mercury vapor lamp (Original Hanau 800). The initial chromatographic evaluations of the imprinted polymers were done using a Bischoff HPLC pump, a Rheodyne injector, an LKB 2151 variable-wavelength monitor as UV detector, and a Kipp and Zonen DB41 plotter. The subsequent optimization, the normal-phase HPLC, and the off-line SPE were carried out using a Hewlett-Packard instrument (HP1100 or HP1050) equipped with a binary pump (HP1100) or a quaternary pump (HP1050), an autosampler, a diode array detector (HP1100), and an HP workstation. The chewing gum workup was carried out according (22) Matsui, J.; Doblhoff-Dier, O.; Kugimiya, A.; Takeuchi, T. Anal. Chim. Acta 1997, 343, 1-4. (23) Kempe, M. Anal. Chem. 1996, 68, 1948. (24) Sellergren, B.; Shea, K. J. J. Chromatogr. 1993, 635, 31.
Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3305
Chart 1
to a method developed by Pharmacia & Upjohn. The MISPE was carried out using repacked Bondelut (Waters) polypropylene cartridges and a Baker SPE 12G vacuum unit. Polymer Synthesis. The polymers were synthesized according to the general imprinting protocol shown in Scheme 1 and the monomer compositions in Table 1. A typical procedure, such as the preparation of polymer PMAA, would be as follows. Nicotine (162 mg, 1 mmol), MAA (344 mg, 4 mmol), EDMA (3.96 g, 20 mmol), and AIBN (40 mg, 0.24 mmol) were dissolved in methylene chloride (5.6 mL). The solution was transferred to a thick-walled glass polymerization tube, degassed by sparging with nitrogen gas for 20 min, and then sealed with Parafilm. The tube was then immersed into a water bath maintained at 10 °C and allowed to equilibrate for 10 min. The tube was irradiated using the UV lamp for 24 h, rotating the tube periodically to ensure even polymerization. Following polymerization, the polymer monolith was ground in wetted state by means of a mechanical ball mill. The procedure was optimized so as to obtain the maximum yield of the required size fraction, 25-35 µm. This fraction was sieved and washed thoroughly with water prior to slurry packing in 80% aqueous methanol at a maximum pressure of 500 bar into an HPLC column (125 × 4 mm, Merck). Chromatographic Evaluation of the Imprinted Polymers. A primary evaluation of the polymer selectivity was done using an organic mobile phase: MeCN/H2O/HOAc 92.5/2.5/5 (v/v). Thereafter, the polymers were evaluated using an aqueous mobile phase: MeCN/potassium phosphate (KP, 0.05 M) 70/30 (v/v) at various pH values. The flow rate was 1 mL/min, the injection volume 10 µL, the UV detector wavelength 254 nm, and the chromatography run at room temperature with duplicate injections unless otherwise stated. The retention, k′, was calculated as k′ ) (t - to)/to, where to is the elution time of the void marker, acetone, which normally eluted as a sharp peak with a maximum plate number, N, of approximately 10 000/m. HPLC Analysis Using MISPE Off-Line. SPE CARTRIDGES. After on-line chromatographic optimization of the SPE conditions, the polymers (0.27 g of 25-35-µm-diameter particles) were removed from the columns and slurry packed in acetonitrile by vacuum suction into emptied standard SPE polypropylene car3306 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
tridges using an SPE vacuum unit. After packing, the inlet frit was replaced and the column conditioned in acetonitrile. MISPE. A citrus placebo chewing gum was dissolved in ethyl acetate (80 mL) + 25% ammonia (900 µL). After sedimentation, 40 mL of the supernatant was removed and 225 µL of acetic acid added. After filtration, the gum solution was added to stock solutions of nicotine, cotinine, β-nicotyrine, and myosmine in 10mL measuring cylinders to make-up the concentration to 0.04 mg/ mL for nicotine and 0.004 mg/mL for the oxidation products. The SPE column was conditioned prior to each run by washing with 3 mL of acetonitrile, 1 mL of a solution of acetonitrile + 2.5% water + 0.2% trifluoracetic acid (elution solvent), and 6 mL of acetonitrile. The spiked gum solutions were filtered, 500 µL was applied to the conditioned SPE column, and the column was allowed to run dry. The column was washed with 3 × 3 mL acetonitrile and elution done with 3 × 1 mL of elution solvent, making sure that the column did not run dry. All fractions were collected in graduated vials and the volumes adjusted with acetonitrile. The fractions were then diluted with one part mobile phase used in the HPLC analysis and directly analyzed by reversed-phase HPLC. This procedure was repeated using a blank. The recovery in each fraction was calculated using the external calibration curve and the average recovery calculated based on three identical runs using the same cartridge. RESULTS AND DISCUSSION Method Requirements. Strict health and environmental regulations concerning the development and manufacturing of new pharmaceuticals have increased the demand for quality control. Among the nicotine formulations for smoking cessation therapy, nicotine chewing gum (Nicorette) is the most widely used product. At present, the quality control includes monitoring of the level of nicotine (2 or 4 mg per gum) and the oxidation products: cotinine, nicotine cis-N-oxide, and nicotine trans-N-oxide. However, some other oxidation products are also of interest, and the method development has targeted myosmine and β-nicotyrine as well (Chart 1). The presently used HPLC methods for this analysis lack the speed required in order to handle an increasing number of samples and often require the use of environmentally harmful solvents. Our goal was to develop a fast, robust, accurate, and environmentally friendly HPLC method in accordance with the producers’ requirements. To achieve this, we have considered a number of possible alternatives. The study has been limited to HPLC-based methods due to considerations concerning analyte stability and method robustness. Analytical Strategies. Knowledge of the chemical and physical properties of the analytes is necessary for a successful method development. Nicotine (NIC) is a moderately strong base with pKa1 ) 8 and pKa2 ) 3, whereas the other analytes are weak bases with the following pKa’s: myosmine (MYO), 5.5; nicotine N-oxide (NOX), 5.0; β-nicotyrine (βNIC), 4.7; cotinine (COT), 4.5. The polarity of the nonprotonated forms, estimated from reversedphase HPLC, increases in the order β-nicotyrine , nicotine < myosmine < cotinine < nicotine N-oxide. The oxidation of nicotine is catalyzed by light, and nicotine is more stable in its protonated form than in its unprotonated, more volatile, form.17 It is chiral and occurs naturally as the S isomer. In the gum, nicotine (2 or 4 mg) is bound to polyacrilex, a weak cationexchanger. The gum further contains a nonpolar polymeric
Scheme 2
Scheme 3
matrix, flavors, and sodium bicarbonate. For a fully automated procedure, the HPLC method should be directly compatible with the sample preparation i.e., similar solvents, pH, and temperatures (Scheme 2). Thus, samples in organic solvent are suited for normal-phase or imprinted-phase chromatography, whereas samples in water are suited for reversed-phase chromatography. The sample preparation can be divided into several steps. First, the gum base has to be dissolved in order to allow release of the analytes into solution. To accomplish this, a solvent of low to medium polarity is required (e.g., hexane, heptane, THF, ethyl acetate, chloroform). Second, to release the analytes from the ion-exchanger, either acid or base has to be added (Scheme 3). Two-Phase Strategy. Addition of acid will lead to release of the analytes from the ion-exchanger in the more stable protonated form. The analytes can then be directly extracted into an aqueous phase, followed by analysis using ion pair or reversed phase HPLC.18 However, this procedure requires a quantitative extraction yield, which can be difficult to achieve for the less polar analytes. One-Phase Strategy. Due to this problem, a one-phase strategy was investigated. Addition of base will lead to deprotonation of the analytes into a form that is soluble in the nonpolar medium used for dissolving the gum matrix. Provided that the matrix components can be separated from the analytes, this onephase strategy should allow quantification with a higher recovery of the less polar analytes. We evaluated the one-phase strategy by considering the following alternative methods: (1) direct injection of the dissolved and filtered gum sample using a normalphase column in order to elute the matrix components with the void followed by direct quantification of all analytes and (2) SPE with the goal of removing the matrix components by a selective
Figure 1. Normal-phase chromatography using a SiO2 column (Merck Purosphere Si80 (125 × 4), 5 µm) with 92.3% ethyl acetate, 7% 2-propanol, and 0.7% ammonia (25% aqueous) as mobile phase. The flow rate was 1 mL/min, the injection volume 10 µL, and the UV detection wavelength 254 nm. Sample: (a) nicotine, 0.2 mg/mL, + oxidation products, 0.005 mg/mL, in heptane/acetone 1/1 (v/v) and (b) an extract of a Nicorette Classic gum dissolved in 40 mL of heptane/acetone 1/1 (v/v).
extraction step prior to reversed phase HPLC. This includes comparison of different SPE columns, therein, a MISPE column. Normal-Phase HPLC Method. In normal-phase chromatography, nitrogen bases often elute with pronounced peak asymmetry due to interactions between the lone pair electrons of the nitrogen and electrophilic sites on the surface of the stationary phase. On metal oxide stationary phases, these are divided into Bro¨nsted acid sites (i.e., silanol groups) and Lewis acid sites (Al, Ti). The influence of these properties on the HPLC of nitrogen bases was recently investigated by comparison of three metal oxide normal-phase columns (TiO2, Al2O3, and SiO2).19 We compared such phases as well as a cyano-bonded-phase column for the separation of the nicotine-related analytes. Using ethyl Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3307
Figure 2. Results from imprinted-phase chromatography using columns packed with various imprinted or blank polymers as described in the Experimental Section. The mobile phase was MeCN/H2O/HOAc 92.5/2.5/5 (v/v/v) in (a), (b) and (c) and MeCN in (d). The compounds were injected separately: nicotine, 100 µg, and the other analytes, 10 µg. The separation factor, R, was calculated as the ratio of the capacity factor (k′) on the imprinted column to k′ on the nonimprinted blank column: R ) k′imp/k′bl.
acetate/2-propanol/ammonia as mobile phase20 and a silica column (Merck Purosphere), we obtained elution profiles with acceptable retention times, plate numbers, and peak symmetries for myosmine, nicotine, and cotinine (Figure 1a). The other normal-phase columns gave either unacceptable peak shapes or insufficient retention of the analytes. As can be seen in Figure 1b, injection of a spiked chewing gum solution on the silica column resulted in elution of the nonpolar matrix close to the void peak and nicotine and cotinine as resolved peaks. This method should, therefore, be suitable for rapid quantification of these analytes alone. However, quantification of β-nicotyrine and myosmine would require a prior cleanup step, whereas the nicotine N-oxides could not be quantified since they were totally retained on this column. Synthesis of Nicotine-Imprinted Polymers. Following an imprinting procedure (Scheme 1) known to give rise to polymers with high affinity and selectivity for nitrogen bases,3-5 we prepared a series of polymers imprinted with nicotine and some of the other analytes (Table 1). The polymers were ground in a ball mill, where the grinding was optimized to achieve a maximum yield 3308 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
of polymer particles in the desired size range. The resulting particles were then washed and packed into chromatographic columns. The goal was to find a polymer that, as stationary phase, would retain and resolve the analytes from the matrix components and quantitatively elute them in a small volume. This would result in an efficient cleanup and enrichment of the analytes. Ideally, the analytes should elute as baseline-resolved Gaussian peaks, allowing direct quantification using the MISPE procedure.10 Due to the many variables affecting the performance of the materials, i.e., polymerization temperature, functional monomer, concentration of template and monomers, and amount and type of porogen (polymerization solvent), a careful optimization of the synthesis of molecularly imprinted polymers is necessary in order to achieve the desired chromatographic efficiency and affinity for the target compound or compounds. Guided by recent advances in the field, we have compared the functional monomers methacrylic acid (MAA) and trifluoromethylacrylic acid (TFM),21,22 cross-linking monomers ethyleneglycol dimethacrylate (EDMA) and trimethylolpropanetrimethacrylate (TRIM),23 various porogens, and differ-
Figure 3. Elution profile of a mixture of nicotine (50 µg), nicotine N-oxides, β-nicotyrine, myosmine, and cotinine (each 5 µg) separated on a column packed with a heat-treated nicotine-imprinted polymer (PMAAh) using an aqueous mobile phase: MeCN/potassium phosphate buffer 0.05 M, pH 4, 70/30 (v/v).
ent thermal treatments.24 Moreover, different batches of the same polymer were compared. Chromatographic Evaluation in Organic Mobile Phases. In Figure 2a, the capacity factors of the analytes injected on columns packed with imprinted phases (Table 1) are shown. The retention of nicotine increased in the order PMAA, PTRIM, PTFM. These results are reasonable in view of the higher concentration of functional monomer used in the preparation of PTRIM and the stronger acidic character of PTFM in comparison with PMAA. The retention of the analytes decreased in the order nicotine > nicotine N-oxides (except on PTFM) > myosmine > cotinine + β-nicotyrine. All analytes were more retained on the nicotineimprinted columns than on the nonimprinted blank columns (Figure 2b). This is reflected in the separation factor (R), which is calculated as the ratio of the capacity factor on the imprinted phase to the capacity factor on the blank, nonimprinted phase. The PMAA column exhibited the highest selectivity, followed by the PTRIM column and then the PTFM column. The low R-values seen on the PTFM column indicate pronounced nonspecific binding. This contrasts with the recent report by Matsui et al.,22 where the selectivity was higher when using TFM than when using MAA as monomer. Since more polar hydrogen-bonding porogens were used in their work, the difference may be related to the solvent used as porogen. An estimate of the polymer batchto-batch reproducibility was made by comparing two polymers (PMAA1 and PMAA2) prepared under identical conditions. At a high sample load (nicotine, 100 µg; other analytes, 10 µg), the difference between the capacity factors obtained on the two columns was less than 10% of the mean, except in the case of cotinine and β-nicotyrine (Figure 2c). The weak retention of the latter two analytes causes a large uncertainty in the calculation of the capacity factors. At lower sample loads, the difference between the nicotine capacity factors was larger. For instance,
at 10 µg sample load, k′ on PMAA1 was 5.4, whereas on PMAA2 it was 4.3, giving a spread of around 20%. Clearly, the reproducibility needs to be improved. This may be done by more carefully controlling the conditions during synthesis and evaluation. Furthermore, these results stress the importance of studying the retention at several sample loads when characterizing imprinted materials, or, preferably, measuring the complete adsorption isotherms.25 As another attempt to enhance the binding of the less retained analytes, polymers imprinted with cotinine, β-nicotyrine, and myosmine were prepared (Figure 2d). All analytes were less retained on these phases than on the nicotineimprinted phases. However, PCOT and PMYO clearly recognized their templates in a weaker mobile phase (MeCN), whereas PβNIC showed no recognition. These results make sense, considering the Bro¨nsted basicity and hydrogen-bonding properties of the templates. Nicotine is the most basic template leading to strong interactions with MAA due to proton transfer and hydrogen bonding. For the less basic templates, hydrogen bonding is likely to be the predominant interaction with the functional monomer. Here, both the lactam group in cotinine and the imine group in myosmine are stronger hydrogen bond acceptors than the pyrrole ring in β-nicotyrine. The extent of ring opening of myosmine under these conditions (to the more basic primary amine) is not known. Chromatographic Evaluation in Aqueous Mobile Phase. In the case of Bro¨nsted basic templates, a higher column efficiency is usually observed using aqueous buffered mobile phases.26 The elution profile of a mixture of the five analytes (Figure 3) shows that all analytes except the cis- and trans-nicotine N-oxides were partly resolved, with the latter and nicotine appearing as broad and strongly tailing peaks. This broadening and asymmetry are related to slow kinetics and overloading of high-energy binding sites.25 The ability of these sites to recognize nicotine, however, appears from Figure 4a. As is obvious from Figure 3, the efficiency is insufficient for the columns to be useful for direct quantifications. In Figure 4, the retention at different mobile-phase pH values can be seen for the PMAA-, PTRIM-, and PTFM- columns. Most strikingly was that the PTFM column now exhibited the weakest retention of all analytes. Due to the higher acidity of the TFM carboxylic acid, the degree of ionization will be high at all of the pH values studied. Although this will allow the positively charged analytes to interact electrostatically, the hydrogen-bonding capacity is weak, which is likely to cause the weak retention. When comparing the PTRIM and PMAA columns, a reversal in the relative retention at the different pH values is observed. On PMAA, nicotine is more strongly retained at pH 7 than at pH 4, whereas the opposite was observed on PTRIM. At pH 7, the retentions on both polymers were almost the same. Higher sample load capacities have been observed using TRIM-imprinted polymers in comparison with EDMA-imprinted polymers.23 A higher abundance of imprinted sites containing carboxylic acid groups, which are presumed to be more acidic,27 may explain the stronger retention at low pH values. In agreement with previous results,24 the retentions on nicotine-imprinted polymers prepared (25) Sajonz, P.; Kele, M.; Zhong, G.; Sellergren, B.; Guichon, G. J. Chromatogr., in press. (26) Sellergren, B.; Shea, K. J. J. Chromatogr. 1995, 690, 29. (27) Sellergren, B.; Shea, K. J. J. Chromatogr. 1993, 654, 17.
Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3309
Figure 4. Retention times of the nicotine analytes (10 µg) injected separately on the columns used in Figure 2 using an aqueous mobile phase: MeCN/potassium phosphate buffer 0.05 M, 70/30 (v/v), at different pH values of the buffer.
using either acetonitrile or methylene chloride as porogens (Figure 4e) were similar. This is important since it may be desirable to prepare materials with different swelling properties and morphologies. Thermal treatments of imprinted polymers were previously shown to result in a higher column efficiency.26 3310 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
PMAA was thus heated at 120 °C under vacuum for 24 h. This procedure only resulted in small changes in the chromatographic retention (Figure 4f) and a small increase in the chromatographic efficiency. However, this phase showed no detectable bleeding of template in the subsequent solid-phase extraction (vide infra).
Figure 5. Optimization of elution conditions for nicotine (4 µg) adsorbed in MeCN on a nicotine-imprinted phase (PMAAh). At 4 min, the mobile phase was switched to MeCN containing various amounts of TFA and water as indicated in the figure.
Figure 6. Peak area in the elution step versus amount of nicotine applied on a column packed with a nicotine-imprinted polymer (PMAA) or a nonimprinted blank polymer (PMAABL) using the elution conditions optimized for the MISPE (MeCN, 0-8 min; MeCN + 0.2% TFA + 2.5% H2O, 8-12 min; MeCN, 12-16 min) or elution using an aqueous elution solvent (MeCN, 0-8 min; MeCN/H2O 70/30 (v/v) + 0.1% TFA, 8-12 min; MeCN, 12-16 min).
Due to this fact and since PMAA exhibited the highest selectivity for nicotine, PMAAh was used in the following method development. SPE Optimization. To find conditions allowing high recovery of the analytes and minimum elution volumes, the elution conditions were optimized on-line. The mobile phase in this step must be chosen considering the subsequent analytical method as well as the solvent used in the chewing gum preparation. Reversed-phase HPLC methods are usually more robust than normal-phase methods; therefore, a reversed-phase method was
Figure 7. Elution profiles of the nicotine analytes on a C18 reversedphase column, Prodigy 5-µm ODS3 (125 × 4.6 mm), using the optimized elution conditions described in Figure 6.
developed, allowing determination of four of the analytes: cotinine, myosmine, nicotine, and β-nicotyrine. In this method, a mobile phase consisting of 50% methanol was used. Two additional considerations were made. The chewing gum is dissolved using a water-immiscible solvent requiring either evaporation and redissolution of the sample or solvent change during the SPE step. The use of acetonitrile as wash solvent was considered a suitable choice due to the miscibility with the sample preparation solvent Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3311
Scheme 4
Figure 8. Adsorption and elution of 100 µL of a placebo chewing gum extract spiked with nicotine (0.5 mg/mL) and β-nicotyrine, cotinine, and myosmine (0.05 mg/mL). The extract was prepared and spiked as described in the Experimental Section. The step elution was as follows: MeCN, 0-8 min; MeCN + 0.2% TFA + 2.5% H2O, 8-12 min; MeCN, 12-15 min. The peak identity was established by separate injections and spectral analysis using a diode array detector. The rest refers to still-not-eluted β-nicotyrine, cotinine, and myosmine. In MISPE, only β-nicotyrine could be detected in the adsorption and wash steps. 3312 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
as well as the HPLC mobile phase. Due to the strong affinity for nicotine exhibited by nicotine-imprinted polymers (a binding constant of 2.7 × 105 M-1 was previously determined for a similar nicotine-imprinted polymer),28 a strong acid (TFA) had to be added to the elution solvent (Figure 5) in order to desorb bound nicotine in a small volume. A slight sharpening of the elution peak was obtained by adding small amounts of water. The final optimum elution solvent was found to be 0.2% TFA and 2.5% water in acetonitrile. Comparing the peak areas in the elution step using the nicotine-selective column with those using the blank column (Figure 6), the former responded linearly to the amount of injected nicotine, whereas a considerable breakthrough was observed on the blank column, leading to a nonlinear response. Larger amounts of water in the elution solvent gave a lower recovery and a pronounced nonlinearity. The solvent switch may, in this case, shrink the polymer, leading to entrapment of the analyte in the smaller pores of the matrix.24 We then investigated whether the cleanup could be performed using a standard C18 reversedphase column (Figure 7). Most of the nicotine broke through with the void, and the major part of the other analytes eluted before the elution step. We therefore concluded that C18 SPE would not be suited for cleanup in this method. Based on the results of the normal-phase evaluation, none of these phases as well is suited for the cleanup step. Injections of spiked chewing gum extracts on the PMAAh and PMAABL phases clearly show the effect of imprinting (Figure 8). Whereas, on the blank column, most of β-nicotyrine, cotinine, and myosmine break through before the elution step, only β-nicotyrine breaks through on the nicotine-imprinted column. This allows quantitative elution of three of the four analytes as one sharp peak. In view of the linear response observed in Figure 6 and the previously determined saturation capacity of a nicotineimprinted column,28 the imprinted binding sites are not likely to be saturated in these experiments. Off-Line MISPE Coupled with Reversed-Phase HPLC. Based on the results from the on-line SPE-optimization, an offline procedure was developed using standard SPE cartridges slurry packed with the polymers (Scheme 4). Figure 9 shows clearly the difference between the elution profiles obtained before and after SPE on a blank polymer or a nicotine-imprinted polymer. (28) Matsui, J.; Kaneko, A.; Miyoshi, Y.; Yokoyama, K.; Tamiya, E.; Takeuchi, T. Anal. Lett. 1996, 29, 2071.
Figure 9. Reversed-phase HPLC analysis of (a) a nicotine standard mixture (nicotine, 0.02 mg/mL; cotinine, β-nicotyrine, and myosmine, 0.002 mg/mL) and of a placebo chewing gum extract spiked with nicotine (0.04 mg/mL) and β-nicotyrine, cotinine, and myosmine (0.004 mg/ mL) before SPE (b), after SPE on PMAABL (c), and after SPE on PMAAh (d). The MISPE procedure was repeated using a nonspiked blank sample (e). The extract was prepared as described in the Experimental Section. The mobile phase was MeOH/0.04 M potassium phosphate buffer, pH 8.5, 1/1 (v/v), the flow rate 0.8 mL/min, the detection wavelength 254 nm, the injection volume 10 µL, and the column a Prodigy 5-µm ODS3 (125 × 4.6 mm).
Direct injection of dried and redissolved extract gave matrix peaks coeluting with myosmine and nicotine, preventing accurate quantification. Furthermore, the column lifetime will be short due to uneluted matrix components. However, a prior MISPE step resulted in a cleaner elution profile with an almost quantitative recovery of nicotine, cotinine, and myosmine (Figure 10). The quantification of cotinine was less reliable due to the poor resolution of this peak from a matrix impurity eluting close to the void peak. On the blank polymer, the acetonitrile wash
fractions showed a considerable breakthrough of most analytes, and only nicotine was recovered in the elution step. Although β-nicotyrine was retained more strongly on PMAAh than on PMAABL, other wash steps will have to be introduced in order to quantitatively recover this analyte. According to several independent reports, a quantitative recovery of the template is important for the use of MISPE for analysis of low concentrations of the template.11-14 In cases where the accessible binding sites are in large excess compared to the Analytical Chemistry, Vol. 70, No. 15, August 1, 1998
3313
Figure 10. Recovery of the analytes in the various fractions collected as shown in Scheme 4 after SPE on PMAAh and PMAABL. Conditions are as described for Figure 9. Results of three runs on the same cartridge gave the following average recoveries: nicotine, 101 ( 4%; myosmine, 111 ( 6%; cotinine, 106 ( 11%.
amount of adsorbed analyte, the problem is due to bleeding of residual “nonextractable” template. Even when the latter constitutes
