Terbutaline Enantiomer Separation and Quantification by

Apr 30, 2008 - The limit of detection and analysis time per sample compared favorably to literature values for chiral terbutaline separation by HPLC a...
1 downloads 9 Views 302KB Size
Download PDF
Anal. Chem. 2008, 80, 4133–4140

Terbutaline Enantiomer Separation and Quantification by Complexation and Field Asymmetric Ion Mobility Spectrometry-Tandem Mass Spectrometry Axel Mie,*,† Andrew Ray,‡ Bengt-Olof Axelsson,§ Magnus Jo¨rnte´n-Karlsson,§ and Curt T. Reimann† Department of Analytical Chemistry, Chemical Centre, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden, Analytical Development, AstraZeneca R&D, Loughborough, United Kingdom, and Analytical Development, AstraZeneca R&D, SE-221 87 Lund, Sweden Recently, we introduced a new approach to chiral separation and analysis of amino acids by chiral complexation and electrospray high-field asymmetric waveform ion mobility spectrometry coupled to mass spectrometry (ESIFAIMS-MS). In the present work, we extended this approach to the separation of the drug compound terbutaline. Terbutaline enantiomers were complexed with metal ions and an amino acid to form diastereomeric complexes of the type [MII(L-Ref)2((+)/(-)-A)-H]+, where MII is a divalent metal ion, L-Ref is an amino acid in its L-form, and A is the terbutaline analyte. When metal and reference compound were suitably chosen, these complexes were separable by FAIMS. We also detected and characterized larger clusters that were transmitted at distinct FAIMS compensation voltages (CV), disturbing data analysis by disintegrating after the FAIMS separation and forming complexes of the same composition [MII(L-Ref)2((+)/(-)A)-H]+, thus giving rise to additional peaks in the FAIMS CV spectra. This undesired phenomenon could be largely avoided by adjusting the mass spectrometer skimmer voltages in such a way that said larger clusters remained intact. In the quantitative part of the present work, we achieved a limit of detection of 0.10% (-)-terbutaline in a sample of (+)-terbutaline. The limit of detection and analysis time per sample compared favorably to literature values for chiral terbutaline separation by HPLC and CE. Enantiomers of drug substances are considered as distinct compounds, because they may have differing biological interactions, effects, and consequences.1 As the majority of the newly approved chiral drugs are single enantiomers,2 rapid methods for the analysis of the “wrong” enantiomer are required. From ICH guidelines, any impurities above 0.05% (or 0.1% dependent on * Corresponding author. E-mail: [email protected]. † Lund University. ‡ AstraZeneca R&D, Loughborough, United Kingdom. § AstraZeneca R&D, SE-221 87 Lund, Sweden. (1) De Camp, W. H. Chirality 1989, 1, 2–6. (2) Caner, H.; Groner, E.; Levy, L.; Agranat, I. Drug Discov. Today 2004, 9, 105–110. 10.1021/ac702262k CCC: $40.75  2008 American Chemical Society Published on Web 04/30/2008

dose) are required to be identified and reported.3 Enantiomeric impurities are excluded from these thresholds, reflecting the fact that due to the similarity in physical properties of enantiomers, separating them is a challenge.4 Typically, enantiomeric impurities would be measured to 0.5% or 1.0%. Many methods for chiral analysis have been developed:5 nuclear magnetic resonance (NMR) spectroscopy with chiral shift reagents,6 supercritical fluid chromatography (SFC),7 capillary electrochromatography (CEC),8–11 mass spectrometry (MS),12,13 and recently ion mobility spectrometry (IMS) with chiral drift gas dopants14 have all been used. However, most chiral separations are carried out by high-performance liquid chromatography (HPLC),7,15 capillary electrophoresis (CE),7,16–18 or gas chromatography (GC).19 Recently, we demonstrated that amino acid enantiomers can be separated by complexation and electrospray ionization (ESI) followed by high-field asymmetric waveform ion mobility spectrometry (FAIMS, also referred to as differential mobility spec(3) Anonymous Guidance for Industry: Q3A(R2) Impurities in New Drug Substances. The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH), 2006. (4) Anonymous Guidance for Industry: Q6A Specifications: test procedures and acceptance criteria for new drug substances and new drug products. The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH), 1999. (5) Ward, T. J.; Hamburg, D.-M. Anal. Chem. 2004, 76, 4635–4644. (6) Wenzel, T. J.; Wilcox, J. D. Chirality 2003, 15, 256–270. (7) Gubitz, G.; Schmid, M. G. Biopharm. Drug Dispos. 2001, 22, 291–336. (8) Zarbl, E.; Lammerhofer, M.; Woschek, A.; Hammerschmidt, F.; Parenti, C.; Cannazza, G.; Lindner, W. J. Sep. Sci. 2002, 25, 1269–1283. (9) Constantin, S.; Bicker, W.; Zarbl, E.; Lammerhofer, M.; Lindner, W. Electrophoresis 2003, 24, 1668–1679. (10) Hebenstreit, D.; Bicker, W.; Laemmerhofer, M.; Lindner, W. Electrophoresis 2004, 25, 277–289. (11) Schweitz, L.; Andersson, L. I.; Nilsson, S. Anal. Chim. Acta 2001, 435, 43–47. (12) Tao, W. A.; Zhang, D.; Wang, F.; Thomas, P. D.; Cooks, R. G. Anal. Chem. 1999, 71, 4427–4429. (13) Schug, K. A.; Lindner, W. J. Sep. Sci. 2005, 28, 1932–1955. (14) Dwivedi, P.; Wu, C.; Matz, L. M.; Clowers, B. H.; Seims, W. F.; Hill, H. H., Jr. Anal. Chem. 2006, 78, 8200–8206. (15) Thompson, R. J. Liq. Chromatogr. Relat. Technol. 2005, 28, 1215–1231. (16) Blomberg, L. G.; Wan, H. Electrophoresis 2000, 21, 1940–1952. (17) Stalcup, A. M.; Gahm, K. H. J. Microcolumn Sep. 1996, 8, 145–150. (18) Zhu, W.; Vigh, G. J. Chromatogr., A 2003, 987, 459–466. (19) Schurig, V. J. Chromatogr., A 2001, 906, 275–299.

Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

4133

trometry, DMS) with mass spectrometric (MS) detection.20 With the aid of this emerging gas-phase ion separation technique, enantiomer impurities of some amino acids could be detected at a better than 0.1% level. In the first place, this article is aimed at extending the method to an analyte, which is not an amino acid, and furthermore which is a known drug substance. Moreover, this article is aimed at further investigating the reported phenomenon, in which a putative single type of ion gives rise to two or three peaks in compensation voltage (CV) spectra acquired in FAIMS separations.20 This is referred to as the “multiple peak” phenomenon throughout this work. A final goal is studying the quantitative aspects of chiral analysis with the reported method. The target compound of this study was terbutaline, a β2-adrenergic receptor agonist used as a fast-acting bronchodilator (for shortterm asthma treatment) and to delay premature labor.21 Terbutaline contains one chiral center and has been used as a test compound for investigation of many chiral separation techniques, particularly those using HPLC22–24 and CE.25–30 Comparative HPLC and CE methods for the quantitation of (+)-terbutaline in (-)-terbutaline have also been published.31 This allowed for the comparison of the new FAIMS-based separation method with a more conventional liquid-phase separation for enantiomers. EXPERIMENTAL SECTION All experiments were performed using a beta unit of a “Selectra” FAIMS system (Ionalytics Corp., Ottawa, Canada; Thermo Finnigan, San Jose, CA) coupled with a custom-built PEEK support to a Bruker Esquire LC ion trap mass spectrometer (Bruker, Bremen, Germany), as shown in Figure 1. The actual FAIMS interface was identical to the interface described previously.20 Samples were supplied using either an Agilent 1100 series autosampler, a syringe pump, or the autosampler together with a syringe pump, coupled together via a T-junction. All samples were supplied to the ion source by direct infusion, without prior liquid chromatographic separation. The original Bruker electrospray ion source is rather large, and due to spatial restrictions it was not mechanically compatible with the FAIMS interface. Therefore, a smaller custom-built ESI source was used incorporating the original Bruker electrospray needle and nebulizer. The electrospray needle was kept at +5000 V, the ESI front plate was kept at +1800 V, and the spraying was roughly (20) Mie, A.; Jo ¨rnte´n-Karlsson, M.; Axelsson, B.-O.; Ray, A.; Reimann, C. T. Anal. Chem. 2007, 79, 2850–2858. (21) Lam, F.; Elliott, J.; Jones, J. S.; Katz, M.; Knuppel, R. A.; Morrison, J.; Newman, R.; Phelan, J.; Willcourt, R. Obstet. Gynecol. Surv. 1998, 53, 85– 95. (22) Andersson, M. E.; Aslan, D.; Clarke, A.; Roeraade, J.; Hagman, G. J. Chromatogr., A 2003, 1005, 83–101. (23) Ye, Y. K.; Stringham, R. W. Chirality 2006, 18, 519–530. (24) Stringham, R. W.; Ye, Y. K. J. Chromatogr., A 2006, 1101, 86–93. (25) Aboul-Enein, H. Y.; Efstatiade, M. D.; Baiulescu, G. E. Electrophoresis 1999, 20, 2686–2690. (26) Westall, A.; Malmstroem, T.; Petersson, P. Electrophoresis 2006, 27, 859– 864. (27) Hedeland, Y.; Lehtinen, J.; Pettersson, C. J. Chromatogr., A 2007, 1141, 287–294. (28) Yang, G. S.; Chen, D. M.; Yang, Y.; Tang, B.; Gao, J. J.; Aboul-Enein, H. Y.; Koppenhoefer, B. Chromatographia 2005, 62, 441–445. (29) Servais, A.-C.; Fillet, M.; Chiap, P.; Dewe, W.; Hubert, P.; Crommen, J. Electrophoresis 2004, 25, 2701–2710. (30) Ivanyi, R.; Jicsinszky, L.; Juvancz, Z.; Roos, N.; Otta, K.; Szejtli, J. Electrophoresis 2004, 25, 2675–2686. (31) Kim, K. H.; Kim, H. J.; Kim, J. H.; Lee, J. H.; Lee, S. C. Chromatographia 2001, 53, 334–337.

4134

Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

Figure 1. Schematic of the ion source, FAIMS electrodes, and MS intermediate pressure stage used in this work.

orthogonal to the ESI front plate ion inlet. The nebulizer gas pressure was typically 8 psi. The FAIMS front plate was kept at +1200 V. N2 at 200 °C and 3L/min was used as the countercurrent drying gas. The FAIMS carrier gas flow rate was 4 L/min. The FAIMS carrier gas composition was 60% N2 and 40% He unless otherwise specified. The FAIMS outer electrode was set to ground potential. The radial distance between the inner and outer electrodes was 2.0 mm, and the axial (tip) distance was kept at 2.5 mm throughout this work, except during the investigation of the origin of “multiple peaks”, when the tip distance was set to 2.3 mm in order to increase ion transmission. The FAIMS inner electrode dispersion voltage (DV) was -4350 V. One focus of this work was studying the effect of the declustering/desolvation potentials in the intermediate pressure stage of the MS on the CV spectra of target complexes. The capillary exit voltage and the skimmer 1 voltage were thus varied with respect to the skimmer 2 voltage, which was held constant at 6 V. Software-suggested “smart settings” of the capillary exit and skimmer 1 voltages, depending on the variation of a “compound stability” software parameter by the user, were utilized as a convenient means of varying the declustering conditions in the “multiple peak” investigation and in the refined reference compound screening below. For the initial reference compound screening and for the optimization of successful separations, these potentials were manually adjusted. The capillary exit voltage was varied within an overall range of 20.0-250.1V, and the skimmer 1 voltage was varied within an overall range of 5.0 to 100.0 V. The capillary exit and skimmer 1 voltages actually used, together referred to as the “declustering potentials”, are specified in the figures or figure texts associated with the respective experiments. All other MS parameters were set to generic tune values: the MS inlet capillary tip was set to ground potential, lens 1 was set to -5 V; the octopole dc offset was set to 2.76 V; the octopole ∆ was 2.4 V; and the octopole rf amplitude was set to 100 Vpp. Lens 2 was kept at -60 V; and the trap drive was 51.3 V. Terbutaline HBr, in (+), (-), and racemic forms, was supplied by AstraZeneca (Lund, Sweden). Bromide ions were removed from the terbutaline samples and exchanged for acetate ions on Isolute PE-AX anion exchange columns (Biotage, Sweden), according to the instructions supplied with the columns. These stock solutions of terbutaline and the amino acids (10 mM) were prepared in 50:50 methanol/water. Stock solutions of the metal

Figure 2. Separation of (+)/(-)-terbutaline as trimeric cluster ions [Cu(L-Trp)2(terb)-H]+. (a) Mixture of (+) and (-)-terbutaline; (b) (+)terbutaline individually; and (c) (-)-terbutaline individually. The curves show extracted ion chromatograms of m/z 226 from MS/MS fragmentation of parent ion m/z 695. Fragment mass corresponds to [terb + H]+. Capillary exit, 75.0 V; skimmer 1, 35.0 V. Carrier gas N2/He 56%:44%. CV scan rate 0.33 V/min.

salts (25 mM) were prepared in water. Other chemicals were obtained and used as specified elsewhere.20 In all experiments, solutions containing 0.133 mM total terbutaline, 0.133 mM reference amino acid, and 0.033 mM metal ions were directly infused to the electrospray ion source. When electrosprayed, these solutions formed, among other ions and clusters, ions of the type [MII(L-Ref)2(A)-H]+ (where MII is a divalent metal ion, L-Ref is a chiral reference amino acid of the L-form, and A is a chiral analyte, here (+) or (-)-terbutaline). It is this complex that may be employed for FAIMS separations of terbutaline enantiomers, as demonstrated below. RESULTS AND DISCUSSION Initial Screening. In order to investigate whether our reported method for chiral separations of amino acids20 is viable for the enantiomers of the drug terbutaline, combinations of metal ions and chiral amino acids were screened for their suitability as chiral reference compounds in conjunction with this analyte. These initial experiments showed that terbutaline enantiomers can be separated in FAIMS as [Cu(L-Trp)2(terb)-H]+ ions (Figure 2). Explanation of Multiple Peaks in CV Spectra. A confusing observation is that each terbutaline enantiomer gives rise to (or is somehow associated with) more than one peak in Figure 2; e.g., the ion [CuII(L-Trp)2((+)terb)-H]+ in Figure 2b seems to have transmission maxima at three distinct CV values, -9.2, -10.9, and -18.5 V. We have observed the same phenomenon with amino acid analytes.20 This could be explained by either (a) the presence of different conformations of ions of the same formula (atomic composition), each conformation being transmitted through FAIMS at a different CV; or (b) by the transmission of larger clusters through FAIMS at different CVs, with these different larger clusters disintegrating in the declustering region of the

mass spectrometer and releasing ions of the type [CuII(LTrp)2((+)terb)-H]+. Indeed, evidence ranging from speculative to fairly conclusive has been presented for both ideas. For example, FAIMS separation of different conformers of certain proteins32–34 and a peptide35 has been demonstrated. On a basic level, the observation of a unique “bump” in a CV spectrum is attributed to a “conformer” or family of “closely related” conformers. These ideas are borne out by combining FAIMS with complementary techniques such as low-field drift-tube mobility where the expected width of a “conformer” is rigorously known,36 or electron-capture dissociation, a more complex technique which can probe conformational differences on different levels,37 or other methods.32,33 It is also possible to study FAIMS performance on polymer ions (like PEG) known to occupy dominantly one conformation, and by a calibration procedure, to then use FAIMS to ferret out multiple conformations of another kind of ion (like the protein lysozyme) in a given charge state.38 Though it appears well established that conformations of (bio)polymers can be separated by FAIMS, such molecules are much larger than the trimeric metal-bound complexes studied in the present work. Moreover, (bio)polymers have a vast conformational space with relatively small energy barriers for conformational transitions. So these ideas might not be directly applicable in the present work. Obviously, in FAIMS separations, every effort is made to ensure that a well-defined species is being analyzed in the FAIMS analytical gap. However, especially in high-flow-rate applications like LC-MS, analyte ions might be incompletely desolvated and could pass through the FAIMS at different CVs before final desolvation in the MS, leading to apparent CV shifting.39 This has been observed as a function of mobile phase composition for chemical warfare agents and rationalized in terms of clustering/ aggregation and incomplete desolvation, though putative clusters/ aggregates were not actually observed.39 In a study of a family of disaccharides, it was recommended that the best FAIMS separation can be obtained under gentle interface conditions, essentially leaving higher-order clusters/aggregates intact so that they do not interfere with MS measurements acquired at specific m/z values of desolvated analyte.40 Finally, use of a methanol driftgas modifier was explored for a similar analyte system to actually get rid of aggregate ions in the FAIMS, leading to more clear MS data about singly protonated analytes.41 (32) Purves, R. W.; Barnett, D. A.; Ells, B.; Guevremont, R. J. Am. Soc. Mass Spectrom. 2000, 11, 738–745. (33) Purves, R. W.; Barnett, D. A.; Ells, B.; Guevremont, R. J. Am. Soc. Mass Spectrom. 2001, 12, 894–901. (34) Shvartsburg, A. A.; Li, F.; Tang, K.; Smith, R. D. Anal. Chem. 2007, 79, 1523–1528. (35) Purves, R. W.; Barnett, D. A.; Ells, B.; Guevremont, R. Rapid Commun. Mass Spectrom. 2001, 15, 1453–1456. (36) Shvartsburg, A. A.; Li, F.; Tang, K.; Smith, R. D. Anal. Chem. 2006, 78, 3304–3315. (37) Robinson, E. W.; Leib, R. D.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2006, 17, 1469–1479. (38) Robinson, E. W.; Sellon, R. E.; Williams, E. R. Int. J. Mass Spectrom. 2007, 259, 87–95. (39) Kolakowski, B. M.; McCooeye, M. A.; Mester, Z. Rapid Commun. Mass Spectrom. 2006, 20, 3319–3329. (40) Gabryelski, W.; Froese, K. L. J. Am. Soc. Mass Spectrom. 2003, 14, 265– 277. (41) Levin, D. S.; Vouros, P.; Miller, R. A.; Nazarov, E. G. J. Am. Soc. Mass Spectrom. 2007, 18, 502–511.

Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

4135

Figure 3. Effect of the variation of declustering potentials on the observed intensity of [Cu(L-Trp)2((+)terb)-H]+. All curves have the same intensity scale. All curves are extracted ion chromatograms of m/z 695 + 697. CV scan rate 5 V/min. Axial electrode distance 2.3 mm.

For the understanding and acceptance of our chiral separation method, it is imperative to reveal in some detail the origin of the multiple peaks seen in Figure 2. We chose the case of [Cu(LTrp)2((+)terb)-H]+ (m/z 695 and 697) for detailed analysis. To probe the role of clusters/aggregates in the FAIMS separation, we varied the MS declustering potential during consecutive CV scans. If some of the multiple peaks originated from larger clusters, one could expect that these larger clusters would remain intact when a low MS declustering potential was chosen, possibly yielding a dramatic effect on the CV spectra observed. It should be noted that the variation of the declustering potential has a dual effect: apart from influencing the declustering efficiency by varying the average energy of ion/neutral collisions in the intermediate stage of the mass spectrometer, a lensing effect can also impact the ion transmission efficiency and intensities. For example, a very low declustering potential will leave most clusters entering the intermediate step intact but will also lead to relatively inefficient ion transmission due to a weak lensing effect. It can be seen from Figure 3 that at very low declustering potentials ( no. 1 and no. 2), only one peak appeared in the CV spectrum, at CV2 ) -11.7 V. As the declustering potential was increased (nos. 3-6), two more peaks appeared, at CV1 ) -9.6 V and CV3 ) -18.0 V. (The slight variations in CV, compared to 4136

Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

the spectra shown in Figure 2, are due to different FAIMS tip distances). At even higher declustering potentials (nos. 7-10), the peaks at CV1 and CV2 disappeared and the peak at CV3 gradually decreased in intensity. We interpret the peaks at CV1 and CV3 as originating from larger clusters that release ions of the type [Cu(L-Trp)2((+)terb)-H]+ upon application of higher declustering potentials, while the peak at CV2 is interpreted as representing the “naked” complex ion [Cu(L-Trp)2((+)terb)-H]+ transmitted through FAIMS. The fact that the intensity of the peak at CV2 is much smaller at declustering potential no. 1 than at no. 2 is attributed to overall poorer electrostatic focusing at declustering potential no. 1, resulting in inefficient ion transmission. (The asymmetry of the peak at CV2 is discussed in the Supporting Information.) In order to verify this interpretation, we tried to find and characterize the proposed larger clusters. This involved setting the FAIMS compensation voltage to the value CV3 and acquiring mass spectra for different declustering voltage settings. Results are presented in Figure 4. Comparing the mass spectra of Figure 4a,b, it is evident that in the low declustering potential case (a) a number of ions in the m/z range 930-1250 are present that do not appear at the higher declustering potential as shown in part b. Instead, the ion with m/z 695, representing [Cu(L-Trp)2((+)terb)H]+, is much more prominent in part b. Consequently, the ions in the m/z range 930-1250 are suspected to be clusters composed of at least Cu2+, (+)-terbutaline, and L-Trp, which decompose upon application of higher declustering potentials to form [Cu(LTrp)2((+)terb)-H]+ and other fragments. We verified this by subjecting the putative clusters to analysis by MS/MS, with results shown in Figure 5. It is apparent from Figure 5 that some of the clusters in the m/z range 930-1250 release [Cu(L-Trp)2((+)terb)-H]+ with m/z 695 upon subjecting the ions to MS/MS fragmentation conditions. The same has been shown for ions at MS/MS isolation m/z 930.0, 981.0, 1012.0, 1125.0, 1032.0, 1114.0, 1165.0, 1227.0 (arrows in Figure 4a, data not shown). Thus, many of the ions in Figure 4a are clusters which contain Cu2+, (+)-terbutaline, and L-Trp. The spacing of clusters in the mass spectrum of Figure 4a and their isotopic patterns give additional information. A portion of the spectrum of Figure 4a has been studied at high resolution (see right inset in Figure 4a; for the same mass spectrum covering a wider m/z range, see Supporting Information). There, the doubly charged clusters are seen to be spaced at an interval of 10.5 m/z, corresponding to a mass difference of 21 units between terbutaline (225) and tryptophan (204). The peaks can be tentatively assigned to a series of clusters with the general formula [Cu3(L-Trp)10-n ((+)terb)n-4H]2+, with experimental m/z values in good agreement with theoretical ones. The left inset in Figure 4a illustrates a good agreement in the isotopic pattern and the m/z value between the experimental high-resolution MS data and the theoretical values for the cluster [Cu3(L-Trp)8((+)terb)2-4H]2+. Particularly consistent with the presence of three copper atoms in the cluster are the relative intensities of the first three isotopic peaks. Specifically, the second isotopic peak (at m/z 1134.4) is much smaller than one would expect for ions consisting of only amino acids, terbutaline, and solvents, due to the distinct isotopic pattern of copper, its main isotope being 63Cu (69.1% isotopic

Figure 5. MS/MS on some clusters appearing in Figure 4a but not part b. (a) MS/MS of m/z 1134.0, (b) MS/MS of m/z 1145.0. CV ) -17.5 V.

Figure 4. MS of a mixture of Cu2+, (+)-terbutaline, and L-Trp around CV3 (average of the range -16 to -19 V). (a) Capillary exit, 83.3 V; skimmer 1, 16.1 V (as in curve no. 2, Figure 3). (b) Capillary exit, 144.8 V; skimmer 1, 58.5 V (as in curve no. 6, Figure 3). Left inset in part a: experimental high-resolution mass spectrum of the peak appearing at m/z 1134.4 (above) and simulated (theoretical) mass spectrum of the cluster [Cu3(L-Trp)8((+)terb)2-4H]2+ (below). Right inset in part a: experimental high resolution-mass spectrum of some peaks apperaring in part a, with a spacing of 10.5 m/z, with assigned cluster formulas. The desired complex is located at m/z 695.1 and 697.1. Arrows or assigned m/z values in part a indicate that these peaks were further investigated by MS/MS (see text).

abundance) but having a fairly abundant isotope 65Cu (30.9% isotopic abundance). Together with the tentative assignment of formulas to the fragments in the MS/MS spectrum of Figure 5a, the collected information appears to be strong enough to unambiguously conclude that the formula of the cluster ion of m/z 1134 is [Cu3(L-Trp)8((+)terb)2-4H]2+. The same procedure as above, identifying cluster ions by MS/ MS and simulation of isotopic patterns, has also been performed for the clusters with m/z 1124 and 1145 (results not shown), confirming the formulas [Cu3(L-Trp)9((+)terb)-4H]2+ and [Cu3(LTrp)7((+)terb)3-4H]2+, respectively. Another cluster, a singly charged ion at m/z 1164.4, could be identified in the same manner

as [Cu2(L-Trp)4((+)terb)-3H]+; and yet another prominent cluster at m/z 1247 could be identified as [Cu3(L-Trp)8((+)terb)3-4H]2+. We were able to conclude that in the CV range around CV3, namely, between -16 and -19 V, a number of larger clusters containing Cu2+, L-Trp and (+)-terbutaline were transmitted through FAIMS and then either passed the intermediate step of the MS or disintegrated therein, depending on the declustering potential. When these cluster ions disintegrated, they formed a variety of product ions, among these the complex ion of the formula [Cu(L-Trp)2((+)terb)-H]+, which also is the target complex for the separation of the terbutaline enantiomers. We also examined the nature of ion species contributing to signal detected at a FAIMS compensation voltage setting of CV1 (Figure S2, see Supporting Information). The number of detected clusters was smaller at CV1 than at CV3, but similar clusters were identified. We also performed similar experiments as described above using (-)-terbutaline instead of (+)-terbutaline, with very similar results. A total of three peaks appeared in the CV spectra in analogy with Figure 3, at CV4 ) -10.2 V, CV5 ) -14.4 V, and CV6 ) -17.5 V. The peak at CV5 was present even at very low declustering potential, in analogy to the peak at CV2 in Figure 3, while the peaks at CV4 and CV6 only appeared at higher declustering potentials (results not shown). Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

4137

Table 1. Metal and Reference Compound Combinations Leading to Successful Terbutaline Enantiomer Separationsa reference compound

Rmof (%)

Rchiral

Cu2+, L-Trp Cu2+, L-Asn Cu2+, L-Gln Mg2+, L-Met

1.1 f 0.11b 1.1 4.2 0.12

1.17 1.60 1.06

c

a Peak resolution in FAIMS CV spectra according to Figure 3 and Figures S3-S5 (see Supporting Information) and chiral selectivity according to Cooks. Minimum overlap factors (Rmof) of the successful terbutaline separations in FAIMS,20 with Rmof ) 0% indicating perfect separation (see also Suppporting Information). Chiral selectivity of MS/ MS kinetic method Rchiral.42 Rchiral very different from 1 indicates very good chiral selectivity. b The 1.1% are calculated from the data shown in Figure 2, while the improved value of 0.11% was calculated from data optimised for suppressed multiple peaks, Figure 3, second curve. c MS/MS fragments used for calculating the chiral selectivity Rchiral according to the kinetic method were only observed very weakly, so that Rchiral could not be determined accurately.

For the purpose of sensitive analysis of a minority enantiomeric species in the presence of a large excess of the opposite enantiomer, the ideal situation to strive for is one in which each of the two enantiomers gives rise to only one peak in the CV spectrum for the target complex ion and that these two peaks be suitably shifted depending on enantiomericity of the analyte. Unfortunately such a “clean” situation might not be achievable in general. The next best strategy, originally proposed in ref 40, could be to choose a declustering potential low enough to leave most of any occurring larger clusters intact, in the hopes that these appear at quite different CVs than the naked complex ions of interest, especially the one originating from the minority enantiomeric species. However, the declustering potential should also be chosen to be high enough to promote an efficient ion transmission through the intermediate pressure stage of the MS. In the present study, for the analyte terbutaline and using L-Trp and Cu for chiral reference compounds, the settings capillary exit (83.3 V) and skimmer 1 (16.1 V) (as in curve no. 2, Figure 3) appeared to be a good compromise between those two requirements and were thus used in quantification experiments described in the Quantification section below. Screening for Suitable Reference Compounds for Terbutaline. Taking our improved understanding of the origin and behavior of larger clusters into account, we modified the screening procedure for suitable metal ion and chiral reference compounds, described earlier,20 including the MS declustering potentials in the list of variables to be optimized. Specifically, each combination of metal ion and chiral reference compound was tested with four different declustering potentials (capillary exit/skimmer 1 settings, 75.8/9.7; 87.4/19.4; 99.8/28.9; 113.0/38.3 V) spanning all changes of interest in the CV spectra of Figure 3. The CV was scanned at 15 V/min from 0 to -30 V. Confirmed separations were then further refined with respect to carrier gas composition and declustering potential, in order to optimize peak separation in the CV spectra and suppress “multiple peaks”. This screening procedure led to the discovery of three additional successful combinations of metal ion and reference 4138

Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

Figure 6. Calibration curve for 0-50% (-)-terbutaline of total terbutaline, measured as fragment m/z 226 of the parent ion [Cu(LTrp)2 (+)/(-)terb)-H]+ (m/z 695 + 697). Error bars denote (standard deviation, based on five injections per concentration. rfrel ≈ 1.05.

compound, summarized in Table 1. (For details on the separation quality of these methods, see Supporting Information.) Quantification. We have previously shown some indication on the performance of quantitative methods based on our chiral complexation/FAIMS separation method.20 In this work, we aim at characterizing some quantitative aspects in further detail. The parameters precision and limit of detection were of chief interest. The measurements for the calibration curves have been performed by alternately monitoring the CV values CV(+) ) -11.7 V and CV(-) ) -14.4 V, corresponding to peak maxima of [Cu(LTrp)2((+)terb)-H]+ and [Cu(L-Trp)2((-)terb)-H+, respectively. The response on the y-axis for each calibration sample is the ratio of the signal intensities for CV(-) and CV(+). The calibration curve follows the equation

( )( ) (

I(-) x(-) ) I(+) x(+)

)

rf(-) x(-) +b) rf + b rf(+) 100 - x(-) rel

(1)

where I denotes signal intensity, x denotes the fraction (in %) of total terbutaline that is present in the respective enantiomeric form, rf the MS/MS response factors of the measured complexes, (-) and (+) the enantiomeric form, rfrel the relative response factor of the (-) and (+) enantiomers, and b the y-axis intercept. For small fractions of (-)-terbutaline, a linear calibration curve is a good approximation. Equation 1 is derived in the Supporting Information. Calibration curves have been established for the (-)-terbutaline of total terbutaline fraction range 0-50% (see Figure 6) with calibration samples at 5% intervals and for the 0-5% range with calibration samples at 0.5% intervals (see Supporting Information), with five replicates for each calibration sample. From the latter curve, the limit of detection (lod) for (-)-terbutaline of total terbutaline was determined to be 0.10% (calculated as the concentration resulting in a signal 3 times that of the background (blank) signal). Because of the lack of reference information on the purity of the (+)-terbutaline at this high chiral purity, it is in this case unclear whether the observed blank signal is due to a contamina(42) Tao, W. A.; Gozzo, F. C.; Cooks, R. G. Anal. Chem. 2001, 73, 1692–1698.

Table 2. Comparison of Some Figures of Merit of Different Techniques for Chiral Separation of Terbutaline Enantiomers technique

limit of detection (fraction of minority enantiomer of total terbutaline)a (%)

analysis time per sample (not including sample preparation)

precision (relative standard deviation, RSD) (%)

CE HPLC FAIMS

0.03b 0.05b 0.10

38 minc 13 min 6 min/90 sd

3.05e 3.08e 7.7f

a For CE and HPLC, the minority enantiomer was (+)-terbutaline, for FAIMS (-)-terbutaline. b CE and HPLC values are taken from work31 published earlier, while FAIMS refers to results from the present work. c Including 13 min column conditioning per sample. d Including/excluding sample delivery time. e Calculated from peak area, six replicates at 10% minority enantiomer of total terbutaline. f Calculated from MS/MS signal intensity, average of RSD of 11 different concentrations of minority enantiomer, between 0% and 5% of total terbutaline, five replicates at each concentration.

tion of (+)-terbutaline with (-)-terbutaline or due to peak overlap in the CV spectrum. If this blank signal is due to contamination, the content of (-)-terbutaline in the used (+)-terbutaline is less than the calculated lod (i.e.,