Unexpected Behavior of Diastereomeric Ions in the GasPhase: A ...

B American Society for Mass Spectrometry, 2013

J. Am. Soc. Mass Spectrom. (2013) 24:573Y578 DOI: 10.1007/s13361-012-0575-8

RESEARCH ARTICLE

Unexpected Behavior of Diastereomeric Ions in the GasPhase: A Stimulus for Pondering on ee Measurements by ESI-MS Caterina Fraschetti,1 Antonello Filippi,1 Maria Elisa Crestoni,1 Tadashi Ema,2 Maurizio Speranza1 1 2

Dipartimento di Chimica e Tecnologie del Farmaco, Università La Sapienza, Rome, Italy Graduate School of Natural Sciences and Technology, Okayama University, Okayama, Japan Abstract. The most common protocols for the quantitative determination of the enantiomeric excess (ee) of raw mixtures by ESI-MS reveal inadequate in cases where the distribution of diastereomeric derivatives diverges from the ee of original solutions. This phenomenon is attributable to a matrix effect, i.e., to the stereospecific formation of high order noncovalent adducts in the ESI droplets, which alters the actual availability of the diastereomeric species under MS analysis. In this frame, the assumption of classic protocols that the ionization correction factor q is independent on the composition of the mixture submitted to analysis is questionable. An alternative methodology is presented in this paper, which is aimed at circumventing the problem by excluding any chemical derivatization of the original raw mixture. It is based on the measurement of the actual distribution of ESIformed proton-bound diastereomeric complexes from the enantiomeric mixture through a careful analysis of their reaction kinetics with a suitable reactant. Key words: Enantiomeric excess, Noncovalent complexes, ESI-FT-ICR mass spectrometry, Gas-phase kinetics, High-throughput parallel screening Received: 6 November 2012/Revised: 21 December 2012/Accepted: 21 December 2012/Published online: 14 March 2013

Introduction

T

he crucial importance of asymmetric synthesis is clearly witnessed by the ever-growing interest in high-throughput parallel screening of chiral catalysts for industrial applications, especially in the pharmaceutical field. Normally, combinatorial methods are widely used for this purpose, including the techniques combining mass spectrometry (MS) and labeling/coding strategy [1, 2]. Most classical MS ionization sources have been employed for this purpose [3, 4]. Among these, electrospray ionization (ESI) proved to be a fast, selective technique for screening a large number of chiral products if discernible by molecular weight or by reactivity [5–7]. Dedicated to Detlef Schröder who passed away too soon Electronic supplementary material The online version of this article (doi:10.1007/s13361-012-0575-8) contains supplementary material, which is available to authorized users. Correspondence to: Maurizio Speranza; e-mail: [email protected]

Several ESI-MS protocols have been devised in recent years for the quantitative determination of the enantiomeric excess (ee) of raw AR/AS mixtures from enantioselective catalytic reactions through molecular weight discrimination [8–12]. The procedure, illustrated in Scheme 1, is essentially based on two concepts, both regarding step (a) of the sequence, i.e., the mass tagging of the raw AR/AS mixture with suitable BR and BS* auxiliaries and the kinetic resolution of the corresponding quasi-diastereomeric derivatives. The BS* reactant differs from its BR quasi-enantiomer for the presence of a substituent group or isotope label (the “mass-tag”) located away from the chiral center. When their equimolar mixture reacts in solution with a raw AR/AS mixture, two pairs of differently stable “mass-tagged” derivatives (i.e. ARBR/ASBR, and ASBS*/ARBS*) are formed at rates that may be different (kinetic resolution). If their formation can be considered irreversible and insensitive to “mass-tag” effects, then kRR 0 kSS* 0 khomo and kSR 0 kRS* 0 khetero. These assumptions are taken for granted if a large excess of the BR/BS* racemate is used in step (a). It should be noted, however, that even in the presence of a large excess of the BR and BS* reactants, a reversible step (a) would lead to a product distribution that does not necessarily

574

C. Fraschetti et al. : Enantiomeric Excess by ESI-MS

Scheme 1. Kinetic pattern and relative equations of the protocol for the quantitative determination of the composition of raw AR/AS mixtures through molecular weight discrimination by ESI-MS [refs. 8–12]

reflect the composition of the original AR/AS mixture unless the corresponding equilibrium constants are very large or very similar. Under such conditions, a calibration curve is drawn using AR/ S A samples of known enantiomeric excess. The measured ESIMS intensity ratio of the ESI-formed [ARBR•H]+/[ASBR•H]+ (Imass1) and [ASBS*•H]+/[ARBS*•H]+ (Imass2) products, obtained from the AR/AS racemate, is taken equal to a fixed correction factor q, which accounts for the differences in their ESI efficiencies. The actual mass ratio y of the [ARBR•H]+/ [ASBR•H]+ and [ASBS*•H]+/[ARBS*•H]+ products from nonracemic AR/AS mixtures is derived from the corresponding Imass1/Imass2 values divided by the q factor taken as independent of the composition of the mixture. The same assumption is used in the expression of the s and ee terms of Scheme 1. In view of frequently observed matrix-effects on MS response [13, 14], several considerations arise from a careful examination of the protocol of Scheme 1. First, the assumption that q factor is independent of the composition of the mixture implicitly excludes any diastereomeric effect on the ESI protonation of the ARBR/ASBR and ASBS*/ ARBS* pairs [8–12]. Furthermore, the protocol requires the use of a large excess of the BR/BS* auxiliaries, which can be considered as abundant chiral “impurities” in the ARBR/ ASB R and ASB S*/AR BS* mixture as well as in the corresponding ESI nanodrops (the “grey zone” of Scheme 1). Here, the possible formation of transient supramolecular aggregates (e.g., [ARBR•H•BS*]+ versus [ASBR•H•BS*]+) may appreciably alter the availability of diastereomeric structures (e.g., [A R B R •H] + versus [ASBR•H]+) and, therefore, introduce a further element against the q invariance [15–17]. Therefore, if the q factor is affected by the composition of the mixture, the value of the actual mass ratio y and, therefore, of the estimated s ratio, both present in the ee calibration equation, would be affected as well. A symptom that this is the case may be offered by the significant deviations (up to ca. 20%) observed between the measured and the actual ee values [8–12].

To eliminate conceivable sources of errors of the sequence of Scheme 1, an alternative procedure has been devised, which avoids any wet chemical transformation of the raw AR/AS mixture prior to ESI-MS analysis [18–21]. The general strategy, illustrated in Scheme 2, is based on: i- the ESI-formation of proton-bound complexes between the AR/AS enantiomers and a suitable receptor M [step (b) of Scheme 2]; ii- the isolation in the gas phase of their isobaric mixture; and iii- the measure, after a fixed delay time t, of their reaction extent towards a suitable reactant Y, present at defined concentrations in the cell of a FTICR mass spectrometer. Several variants can be used that employ (1) a chiral M and an achiral Y; (2) an achiral M and a chiral Y; and (3) chiral M and Y. The method requires the construction of a calibration curve obtained by preparing standard AR/AS/M mixtures of defined composition and by comparing the peak height of the [AR•H•M]+ and [AS•H•M]+ isobars (Ireactants) with that of their [Y•H•M]+ product (Iproduct) at a reaction time that provides the optimum between the shortest analysis time and the largest difference in reaction extent. Figure S1 in the Electronic Supplementary Material (ESM) reports a simulation of the first-order Y-to-AR (or AS) exchange reaction involving several mixtures of the [AR•H•M]+ (k′01.0 s–1) and [AS•H•M]+ (k′′00.5 s–1) isobars. Figure 1 illustrates the linear dependence of the corresponding reaction extent (100×Ireactants/(Ireactants + Iproduct)) as a function of the [AR•H•M]+ molar fraction, calculated at different reaction times t. As expected, the slope of the curves depends on the reaction time and at tmax 01.4 s it reaches the maximum value, i.e., the largest difference in reaction progress between [AR•H•M]+ and [AS•H•M]+ (see the inset of Figure 1). Once defined at tmax, the calibration curve allows of the estimate of the composition of an unknown AR/AS mixture through the analysis of the 100×Ireactants/(Ireactants + Iproduct) ratio from a single MS spectrum, provided that the[AR•H•M]+/[AS•H•M]+ ratio from ESI coincides with the [AR]/[AS] one in the original solution. In general, this condition applies to mixtures of simple mono- or bifunc-


575

Scheme 2. Schematic protocol for the quantitative determination by ESI-MS of the composition of raw AR/AS mixtures through the reactivity of their noncovalent complexes [refs. 8–12]

tional enantiomers, whereas it may fail for more complex multifunctional species, which can give rise to a variety of isomeric complexes. The present paper will show what happens when the [AR•H•M]+/[AS•H•M]+ ratio arising from ESI appreciably deviates from the [AR]/[AS] one in the original solution and how to manage the problem.

Experimental Materials The pure enantiomers of phenylalanine (pheD and pheL) and of 2-aminobutane (YR and YS) were obtained from Sigma-Aldrich and used without further purification. The macrocycles Chirabite-AR (1R in Figure 2) was synthesized and purified as described elsewhere [22, 23].

Mass Spectrometry The experiments were performed at room temperature in an APEX 47e FT-ICR mass spectrometer equipped with a nano-ESI source (Bruker Spectrospin, Bremen, Germany)

Figure 1. Typical linear calibration plots, corresponding to the decay curves in Figure S1 of ESM, obtained at different reaction times, from t01 s to t05 s. The reaction timedependence of the slopes of the linear curves is reported in the inset (see Equations S1–S4 in ESM)

and a resonance cell (“infinity cell”) situated between the poles of a superconducting magnet (4.7 T). Stock methanolic solutions of 1R (1×10–5 M), containing a 3-fold excess of phenylalanine, either the pure enantiomers or their calibrated mixtures, were nano-electrosprayed through a heated capillary into the external source of the FT-ICR mass spectrometer. The operating conditions of the nano-ESI source have been the following: cap exit 100 V; dry gas: 50 °C. The resulting ions were transferred into the resonance cell by use of a system of potentials and lenses and were quenched by collisions with Argon, pulsed into the cell through a magnetic valve. Abundant signals, corresponding to the natural isotopologues of the proton-bound complex between the macrocycle and the phenylalanine, were monitored and isolated by broadband ejection of the accompanying ions. The proton-bound complexes were then allowed to react with a 2-aminobutane enantiomer (either YR or YS), present in the cell at the corrected pressure of 3.8×10–7 mbar.

Results and Discussion When the isobaric proton-bound complexes [pheD•H•1R]+ or [pheL•H•1R]+ are allowed to react with the amine, either YR or YS, the dependence of the 100×Ireactants/(Ireactants + Iproduct) ratio as a function of the pheD molar fraction χ appeared nonlinear (see the points in Figure 3). It should be noted, in addition, that for χG0.6 values, the 100×Ireactants/ (Ireactants + Iproduct) values tend to merge towards 100, i.e.,

Figure 2. Structure of Chirabite-AR (1R)

576

Figure 3. Reaction extent of the nano-ESI-formed [pheD•H•1R]+ and [pheL•H•1R]+ isobars with R-(−)-2-aminobutane (YR) as a function of Χ, measured at different reaction times t [2.5 s (open circles); 5.0 s (filled circles); 7.5 s (diamonds)]. For the meaning of the nonlinear curves, see text

that the reaction efficiency tends to decrease progressively. This implies that the calibration curves of Figure 3 can be used only in the 0.6≤χG1.0 range, whereas they lose their discriminating power at χG0.6 since becoming essentially flat [calibration curves in the 0.0≤ χ G0.4 range, similar to those of Figure 3, can be obtained by using as host the enantiomer of Chirabite AR].


The nonlinear trends of the points of Figure 3 may find an explanation in a marked difference between the [pheD]/ [pheL] ratio in the starting methanolic solutions and the [pheD•H•1R]+/[pheL•H•1R]+ ratio formed in the nano-ESI droplets (henceforth denoted as the [phe D •H•1 R ] 0 + / [pheL•H•1R]0+ ratio). To substantiate this hypothesis, it is necessary to estimate the [pheD•H•1R]0+/[pheL•H•1R]0+ ratio through a detailed study of their reaction kinetics towards YR. The time-dependence of the relative abundance of the [pheD•H•1R]+ and [pheL•H•1R]+ reactants, individually generated from the corresponding pheD/1R and pheL/1R solutions, are shown in Figure 4a and b, respectively. The kinetic curves of Figure 4a are consistent with a very slow Y R -induced substitution of the phenylalanine in the [pheL•H•1R]+ complex. In contrast, those of Figure 4b suggest the occurrence of several [pheD•H•1R]+ isomeric structures. The most abundant one (ca. 80%) reacts with YR through a complicated pattern yielding both the substitution product [Y R •H•1 R ] + and the addition derivative [pheD•H•1R•YR]+. Minority isomers (ca. 20%) are essentially inert towards YR. Similar patterns were observed by using YS, instead of YR, as the amine reactant (Figure S2 of ESM). The continuous lines in Figure 4a and b represent the best fitting between the experimental points and relative abundance of reactants and products calculated from the exact rate expressions of the corresponding reaction patterns (see Equations S5–S8 of ESM). The best fit procedure led to the rate constant values reported in Table 1.

Figure 4. (a) Time-dependent relative abundances of [pheL•H•1R]+ (circles) and [YR•H•1R]+ (diamonds) from the reactions between YR and [pheL•H•1R]+; (b) time-dependent relative abundances of [pheD•H•1R]+ (open circles), [pheD•H•1R•YR]+ (filled circles), and [YR•H•1R]+ (diamonds) from the reactions between YR and [pheD•H•1R]+. The continuous lines represent the relative abundance of reactants and products calculated from the rate constants of Table 1


577

Table 1. Rate constants for the reaction of 2-aminobutane with the proton-bound diastereomeric complexes between phenylalanine and the chiral macrocycle 1R Complex

Amine

%a

Rate constants 10 add

1010ksubst

103kdiss

cm3 mol–1 s–1

cm3 mol–1 s–1

s–1

0.0868±0.0015

0.197±0.053 0.0017±0.0005 0.199±0.046 0.0012±0.0003

10 k

[pheD•H•1R]+ [pheL•H•1R]+ [pheD•H•1R]+ [pheL•H•1R]+

YR YR YS YS

80 100 80 100

0.0810±0.0021

95±13 25±7

a

Percent of reactive complex

Inspection of Table 1 indicates that the configuration of phenylalanine not only determines the reaction pattern but also the magnitude of the corresponding rate constants. The rate constants of ligand displacement on [pheD•H•1R]+ are over 100 times larger than those on [pheL•H•1R]+. Kinetic curves qualitatively similar to that of Figure 4b were obtained for the isobaric [phe D •H•1 R ] 0 + and [pheL•H•1R]0+ complexes from pheD/pheL mixtures of defined composition (see Figures S3–S6 of ESM). Best fit of the experimental data by using kinetic parameters of Table 1 leads to the relative abundance of the [pheD•H•1R]0+ and [pheL•H•1R]0+ diastereomers, which can be compared to the composition of the starting pheD/pheL methanolic solutions (Figure 5 and Table S1 of ESM). Because of the difference between the [pheD]/[pheL] ratio in the starting solutions and the [pheD•H•1R]0+/ [pheL•H•1R]0+ one in the nano-ESI droplets, the correlation between the molar fraction of [pheD•H•1R]0+ and χ is nonlinear, as expected (Figure 5a). However, a satisfactory linear correlation can be established between the experimental [phe D •H•1 R ] 0 + /[phe L •H•1 R ] 0 + ratios and the corresponding [pheD]/([pheL] ones (Figure 5b). The slope of the curve of Figure 5b gives the measure of the discrepancy between the two ratios by effect of ESI.

Accordingly, the abundance of the [pheD•H•1R]0+ adduct is found to be 5–6 times lower than expected. Going back to Figure 3, the curves represents the χdependence of the 100×Ireactants/(Ireactants +Iproduct) ratios calculated from the kinetic equations S9 and S10 of ESM by introducing the [pheD•H•1R]0+ and [pheL•H•1R]0+ molar fraction derived from Table S1. Although the agreement between the experimental points and the calculated correlation curves is not perfect (r2 00.962), nevertheless both follows the same trend with χ. This further confirms the soundness the best fitting procedure used to determine the reaction patterns of Figure 4a, b and the relative abundance of the [pheD•H•1R]0+ and [pheL•H•1R]0+ compexes from ESI. The marked effect of ESI on the relative abundance of diastereomeric species, shown ìn Table S1, may operate in the ESI of raw mixtures of the quasi-enantiomers and diastereomers of Scheme 1 as well. In the corresponding calibration runs, the q parameter and, therefore, the y term may be influenced by the large excess of the BR and BS* reactants in the starting AR/AS mixtures [8–12]. This effect is particularly evident in the calibration curves of Figures S7 and S8 of ESM. If the points corresponding to the methodological constrains eecalc 0eeactual 00% (racemate) and 100% (pure enantiomer) are removed from the correla-

Figure 5. (a) Nonlinear correlation between the molar fraction of [pheD•H•1R]+ and χ; (b) linear correlation between the [pheD•H•1R]0+/[pheL•H•1R]0+ ratio and the [pheD]/[pheL] one in the starting methanolic solutions (see also Table S1 of ESM)

578

tion, a significant deviation of the slope is observed, which reflects the presence of a systematic error. In conclusion, the present study indicates that the highthroughput screening of the enantiomeric composition of raw AR/AS mixtures based on MS molecular weight discrimination of ESI-formed ions may be liable to systematic errors because of the presence of chiral impurities (mostly BR and BS*) in the analyzed samples [8–12]. A more reliable procedure is based on: i- ESI of raw AR/AS mixtures containing a suitable M receptor; ii- the isolation of corresponding [AR•H•M]0+ and [AS•H•M]0+ isobars; iiitheir gas-phase reaction with a suitable reactant Y; and, ivthe determination of their reaction yield after a given delay time t (Figure 1). If the yield is plotted against the molar fraction χ of one between AR and AS in their mixture, a linear calibration curve is generally obtained, provided that ESI does not alter appreciably the [AR•H•M]0+/[AS•H•M]0+ ratio relative to the [AR]/[AS] one in the liquid mixture. If this requirement is not fulfilled, a nonlinear calibration curve is obtained, which may make the determination of the composition of raw AR/AS mixtures much less straightforward. A possible solution could be to estimate the [AR•H•M]0+/[AS•H•M]0+ ratio through a careful study of the kinetics and mechanism of their reaction with the reactant Y. By this approach, a linear calibration curve is obtained whose slope corresponds to the departure of the [AR•H•M]0+/[AS•H•M]0+ ratio relative to the [AR]/[AS] one in the liquid mixture (Figure 5b). An alternative route could be to employ an achiral receptor M and chiral reactants Y (variant 2) of the protocol of Figure 1, provided that corresponding [AR•H•M]+ and [AS•H•M]+ enantiomers exhibit an appreciably different reactivity towards Y. Work is in progress along this line. A final consideration concerns the good versatility of variant1 (1) and (3) of the procedure of Scheme 2, which can be employed to determine the composition of mixtures of structurally complex, multifunctional enantiomers, like those of Figure 2, by using simple molecules, such as phenylalanine, acting as ”receptors.”

Acknowledgments The authors acknowledge support for this work by the Ministero dell’Istruzione dell’Università e della Ricerca. Annito Di Marzio is gratefully acknowledged for his technical support.

References 1. Vedejs, E., Chen, X.: Parallel kinetic resolution. J. Am. Chem. Soc. 119, 2584–2585 (1997) 2. Schug, K.A., Lindner, W.: Chiral molecular recognition for the detection and analysis of enantiomers by mass spectrometric methods. J. Sep. Sci. 28, 1932–1955 (2005)


3. Shizuma, M., Imamura, H., Takai, Y., Yamada, H., Takeda, T., Takahashi, S., Sawada, M.: Facile ee-determination from a single measurement by fast atom bombardment mass spectrometry: a double labeling method. Int. J. Mass Spectrom. 210, 585–590 (2001) 4. So, M.P., Wan, T.S.M., Chan, T.W.D.: Differentiation of enantiomers using matrix-assisted laser desorption/ionization mass spectrometry. Rapid Commun. Mass Spectrom. 14, 692–695 (2004) 5. For comprehensive surveys on chiral recognition by MS methods, see: Chiral Recognition in the Gas Phase, Anne Zehnacker, Ed.; CRC Press, Taylor & Francis Group, Boca Raton, FL 33487–2742 (2010) 6. Tao, W.A., Cooks, R.G.: Chiral analysis by MS. Anal. Chem. 75, 25A– 31A (2003) 7. Filippi, A., Giardini, A., Piccirillo, S., Speranza, M.: Gas-phase enantioselectivity. Int. J. Mass Spectrom. 198, 137–163 (2000) 8. Guo, J., Wu, J., Siuzdak, G., Finn, M.G.: Measurement of enantiomeric excess by kinetic resolution and mass spectrometry. Angew. Chem. Int. Ed. Engl. 38, 1755–1758 (1999) 9. Reetz, M., Becker, M.H., Klein, H.W., Stöckigt, D.: A method for highthroughput screening of enantioselective catalysts. Angew. Chem. Int. Ed. Engl. 38, 1758–1761 (1999) 10. Thurow, K., Gördes, D.: High-throughput screening application for the determination of enantiomeric excess using ESI-MS. JALA 2006(11), 128–133 (2006) 11. Fleisher, H., Gördes, D., Thurow, K.: High-throughput screening. Applications for enantiomeric excess determination using ESI-MS. Am. Lab. 41, 21–24 (2009) 12. Fleisher, H., Thurow, K.: Rapid enantiomeric excess determination of D- and L-proline using electrospray ionization-mass spectrometry. Am. Lab. 43, 34–36 (2011) 13. Souverain, S., Rudaz, S., Veuthey, J.L.: Matrix effect in LC-ESI-MS and LC-APCI-MS with off-line and on-line extraction procedures. J. Chromatogr. A 1058, 61–66 (2004) 14. Dams, R., Huestis, M.A., Lambert, W.E., Murphy, C.M.: Matrix effect in bio-analysis of illicit drugs with LC-MS/MS: influence of ionization type, sample preparation, and biofluid. J. Am. Soc. Mass Spectrom 14, 1290–1294 (2003) 15. Aschi, M., Fraschetti, C.: Filippi. A., Speranza, M.: Reaction diastereoselectivity of chiral aminoalcohols/[Co(II)NO3]+ complexes in evaporating ESI nanodroplets: new insights from a joint experimental and computational investigation. J. Mass Spectrom. 44, 1038–1046 (2009) 16. Flick, T.G., Leib, R.D., Williams, E.R.T.: Standard-free quantitation of mixtures using clusters formed by electrospray mass spectrometry. Anal. Chem. 81, 8434–8440 (2009) 17. Schug, K.A., Joshi, M.D., Fryčák, P., Maier, N.M., Lindner, W.: Investigation of monovalent and bivalent enantioselective molecular recognition by electrospray ionization-mass spectrometry and tandem mass spectrometry. J. Soc. Mass Spectrom. 129, 1629– 1642 (2008) 18. Liang, Y., Bradshow, J.S., Izatt, R.M., Pope, M., Dearden, D.V.: Analysis of enantiomeric excess using mass spectrometry: fast atom bombardment/sector and electrospray ionization/Fourier transform mass spectrometric approaches. Int. J. Mass Spectrom. 185/186/187, 977–988 (1999) 19. Grigorean, G., Ramirez, J., Ahn, S.H., Lebrilla, C.B.: A mass spectrometry method for the determination of enantiomeric excess in mixtures of D,L-amino acids. Anal. Chem 72, 4275–4281 (2000) 20. Grigorean, G., Lebrilla, C.B.: Enantiomeric analysis of pharmaceutical compounds by ion/molecule reactions. Anal. Chem. 73, 1684–1691 (2001) 21. Gerbaux, P., De Winter, J., Cornit, D., Ravicini, K., Pesesse, G., Cornil, J., Flammang, R.: Noncovalent interactions between ([18]crown-6)tetracarboxylic acid and amino acids: electrospray-ionization mass spectrometry investigation of the chiral-recognition processes. Chem. Eur. J. 14, 11039–11049 (2008) 22. Ema, T., Tanida, D., Sakai, T.: Synthesis and characterization of a persistent paramagnetic rotaxane based on α-cyclodextrin. Org. Lett. 8, 3773–3775 (2006) 23. Ema, T., Tanida, D., Sakai, T.: Versatile and practical macrocyclic reagent with multiple hydrogen-bonding sites for chiral discrimination in NMR. J. Am. Chem. Soc. 129, 10591–10596 (2007)