Anal. Chem. 2005, 77, 5311-5316
Compound Screening for the Presence of the Primary N-Oxide Functionality via Ion-Molecule Reactions in a Mass Spectrometer Michael A. Watkins,† Danielle V. WeWora,† Sen Li,† Brian E. Winger,‡ and Hilkka I. Kentta 1 maa*,†
Department of Chemistry, Purdue University, West Lafayette, Indiana 47907-2038, and Eli Lilly and Company, Indianapolis, Indiana 46285
A mass spectrometry method is presented for the identification of compounds that contain the primary N-oxide functional group. This method utilizes a gas-phase ionmolecule reaction with dimethyl disulfide that rapidly and selectively derivatizes the protonated primary N-oxide functional group in a mass spectrometer to yield an ionic reaction product (with 31 Da higher mass than that of the protonated molecule) that is diagnostic for the presence of a primary N-oxide functionality. A variety of protonated analytes containing different functional groups were tested in Fourier transform ion-cyclotron resonance and triple quadrupole mass spectrometers to probe the selectivity of the reaction. Only molecules containing the protonated primary N-oxide functional group yielded the diagnostic reaction product; all other protonated molecules gave protonated dimethyl disulfide or no reaction products. The feasibility of this method for compound screening was tested by examining six analytes with the same molecular formula but different atom connectivity. The one analyte that contained the primary N-oxide functional group was readily differentiated from the other analytes. The oxidation of nitrogen-containing species to nitrogen oxides (N-oxides) is a commonly observed biotransformation and stressinduced oxidative degradation reaction in pharmaceuticals.1 The ability to screen for and identify metabolites and degradation products with an N-oxide functionality is of interest since they are typically considered to be genotoxic.2,3 Traditional structure elucidation methods, such as NMR and mass spectrometry (MS), have difficulties in identifying the N-oxide functionality. NMR typically provides detailed structural information for degradation products and metabolites but has difficulties detecting nitrogencontaining species (the natural abundance of 15N is 0.37% relative to 14N). Furthermore, differences in chemical shifts of many nitrogen compounds are very small, which complicates NMR spectral interpretation.4 Tandem mass spectrometric methods * To whom correspondence should be addressed. E-mail:
[email protected]. † Purdue University. ‡ Eli Lilly and Co. (1) Clement, B. Biomed. Health Res. 1998, 25, 59-71. (2) Ashby, J.; Tennant, R. W. Mutat. Res. 1988, 204, 17-115. (3) Ashby, J.; Tennant, R. W.; Zeiger, E.; Stasiewicz, S. Mutat. Res. 1989, 223, 73-103. 10.1021/ac050324b CCC: $30.25 Published on Web 07/07/2005
© 2005 American Chemical Society
involving collision-activated dissociation (CAD) have been widely used for structure elucidation of unknowns, but are generally ineffective in the identification of N-oxides due to the fact that similar CAD spectra are also generated for many other nitrogencontaining species.5 Recently, a mass spectrometry method was introduced for the differentiation of two types of isomeric species that commonly arise during drug metabolism: N-oxides (formed from nitrogen oxidation) and nitrogen-containing alcohols (formed from aliphatic or aromatic carbon hydroxylation). This method utilizes the characteristic oxygen atom loss ([M + H - O]+) from protonated N-oxide-containing species during thermal degradation in a heated atmospheric pressure ionization (API) source to distinguish them from the protonated hydroxyl-containing isomers.5-7 Both the atmospheric pressure chemical ionization (APCI) source and the electrospray ionization (ESI) source with a heated transfer tube can be used to obtain the product diagnostic for protonated N-oxides. However, as noted by the authors, thermal degradation methods must be approached cautiously as they may hinder quantitative analysis.5 Furthermore, it was recently shown that some MS analyses that require heating can cause such severe degradation of N-oxide metabolites that the parent ion is not observed, thus masking the presence of the compound. For example, until recently, cocaine N-oxide had not been directly observed or quantified in biological fluids because the traditional technique used for cocaine analyses, GC/MS, results in 100% thermal conversion of the cocaine N-oxide into cocaine.8 Even more recently, it has been shown that thermal processes in LC/APCI-MS can cause the conversion of up to 74% of cocaine N-oxide into cocaine and norcocaine (most of the degradation was attributed to the heated APCI vaporizer).9 Therefore, mass spectrometric methods that do not require the use of thermal degradation, and that provide direct evidence for the (4) Nelson, J. H. Nuclear Magnetic Resonance Spectroscopy; Pearson Education: Upper Saddle River, NJ, 2003. (5) Peiris, D. M.; Lam, W.; Michael, S.; Ramanathan, R. J. Mass Spectrom. 2004, 39, 600-606. (6) Ramanathan, R.; Su, A. D.; Alvarez, N.; Blumenkrantz, N.; Chowdhury, S. K.; Alton, K.; Patrick, J. Anal. Chem. 2000, 72, 1352-1359. (7) Tong, W.; Chowdhury, S. K.; Chen, J.-C.; Zhong, R.; Alton, K. B.; Patrick, J. E. Rapid Commun. Mass Spectrom. 2001, 15, 2085-2090. (8) Wang, P. P.; Bartlett, M. G. J. Anal. Toxicol. 1999, 23, 62-66. (9) Lin, S.-N.; Walsh, S. L.; Moody, D. E.; Foltz, R. L. Anal. Chem. 2003, 75, 4335-4340.
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005 5311
presence of compounds with an N-oxide functionality, are of interest. We report here a mass spectrometric method for the screening of molecules for the presence of primary N-oxide functionalities (nitroso compounds) via a functional group selective ion-molecule reaction. Any method can be used to evaporate and protonate the analyte, including the most popular API technique, ESI. Protonated analytes are allowed to undergo ion-molecule reactions with dimethyl disulfide that generates a diagnostic product only for primary N-oxides. The method is rapid, requiring less than 3-s analysis time in a FT-ICR mass spectrometer and even less in a triple quadrupole mass spectrometer (once the analyte has been introduced). EXPERIMENTAL SECTION Two mass spectrometers were used for these experiments, a Finnigan model FTMS 2001 FT-ICR mass spectrometer and a Finnigan TSQ 700 triple quadrupole mass spectrometer. The FTICR mass spectrometer is equipped with a 3-T solenoidal superconducting magnet. It was modified to accommodate various inlets for introduction of solid, liquid, and gaseous reagents.10 The two cells in this differentially pumped dual cell11 FT-ICR share a common trap plate that can be temporarily held at 0 V, allowing ions to be transferred from one cell into the other through a 2-mm hole in the center of this plate. Quadrupolar axialization (QA) was used for radial ion cloud compression prior to transfer in order to increase its efficiency.12 The inherently mass-selective nature of QA was used to isolate the ion population of interest prior to transfer. In these experiments, one cell region of the dual cell was used for ion generation while the other was used for ion-molecule reactions and detection. Vapors from the liquid analytes were introduced into a cell of the FT-ICR via a Varian variable leak valve (∼3 × 10-8 Torr nominal pressure in the ICR cell; all pressures were monitored by Bayard-Alpert ionization gauges). Methanol vapor, used as a chemical ionization (CI) reagent, was introduced into the same cell via an Andonian Cryogenics adjustable leak valve (∼3 × 10-8 Torr nominal pressure in the ICR cell). Methanol was ionized by electron ionization (EI; 25-eV electron energy for 100 ms at 8-µA emission current) to form the molecular ion and ionic fragments that reacted (∼3 s) with the analyte to form the protonated analyte. The protonated analyte was subsequently transferred into the other cell where dimethyl disulfide was present at a constant pressure of (4-6) × 10-8 Torr (nominal). Time was provided for the analyte ions to react with the dimethyl disulfide reagent, followed by detection of the remaining analyte ions and ion-molecule reaction products (reaction times were typically 0.5-10 s; however, up to 300 s was used to verify that no reaction took place for non-N-oxide analytes). All ion-molecule reaction mass spectra were subjected to background subtraction. The background spectra were generated by stored waveform inverse Fourier transform13 ejection of the ion of interest prior to reaction time. Collision-activated dissociation studies on a deriva(10) Thoen, K. K.; Smith, R. L.; Nousiainen, J. J.; Nelson, E. D.; Kentta¨maa, H. I. J. Am. Chem. Soc. 1996, 118, 8669-8676. (11) Littlejohn, D. P.; Ghaderi, S. U.S. Patent, 4,581,533, 1986. (12) Schweikhard, L.; Guan, S.; Marshall, A. G. Int. J. Mass Spectrom. Ion Processes 1992, 120, 71-83. (13) Marshall, A. G.; Wang, T. C. L.; Chen, L.; Ricca, T. L. ACS Symp. Ser. 1987, 359, 21-33.
5312
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
tized analyte were carried out by pulsing argon into the analyzer cell (nominal peak pressure 1 × 10-6 Torr) in the end of the experiment described above. The analyte ions were translationally excited by an off-resonance (at (1000 Hz from the ion’s cyclotron frequency) rf field, allowing for energetic collisions to occur with argon for 0.3-0.6 s. The TSQ 700 triple quadrupole instrument used in this work was modified to allow for the introduction of volatile liquid reagents into the rf-only quadrupole (Q2) collision chamber. These modifications include the addition of a valve manifold to the collision gas inlet, and a rotary vane vacuum pump to allow freeze, pump, and thaw cycles needed to purge all dissolved gases from the liquid reagents. This triple quadrupole instrument is equipped with a CI source that was used for ion generation with methanol vapor (50 eV electron energy, 400 µA emission current). Methanol was introduced into the source via the CI reagent gas inlet (1.5 Torr nominal pressure of methanol in the ion source; measured by a Granville-Phillips Convectron gauge). The analyte was introduced into the ion source via a home-built probe that was inserted into the solids probe inlet and used a Varian variable leak valve to control the analyte pressure in the source (10 mTorr nominal pressure). After the protonated analyte was mass-selected by the first quadrupole (Q1), it was allowed to undergo reactive collisions with the dimethyl disulfide reagent present in the second quadrupole (Q2; 3 mTorr nominal pressure; measured by a Granville-Phillips Convectron gauge) at laboratory collision energies less than 0.5 eV (set by Q2 offset voltage relative to the source). Reaction products were monitored by scanning the third quadrupole (Q3). All reagents used in these experiments were purchased from Sigma-Aldrich and used as received. RESULTS AND DISCUSSION Previously we have shown that methods based on ionmolecule reactions in a FT-ICR mass spectrometer can be useful for the identification of the functional group in protonated monofunctional oxygen-containing compounds.14 Furthermore, we have demonstrated that these methods can be used to identify and count the number of functional groups in polyols.15 The first reaction step in these methods involves the transfer of a proton from the protonated analyte to a methoxy functionality in a neutral boron-containing reagent. Unfortunately, these reagents are not basic enough to deprotonate protonated N-oxides. Therefore, a different neutral reagent is needed for the identification of analytes with the N-oxide functional group. A desired property of a functional group selective ion-molecule reaction is that the selected functional group be modified while all other functional groups remain unchanged. For the work presented here, a ligand exchange reaction between a neutral reagent and the protonated N-oxide functional group was the targeted derivatization reaction. A Wittig-type reaction was sought (for an example of a Wittig reaction, see Scheme 1). Peroxides were examined as possible reagent candidates because they could potentially react via ligand exchange with protonated primary N-oxides. Di-tert-butyl peroxide was found to (14) Watkins, M. A.; Price, J. M.; Winger, B. E.; Kentta¨maa, H. I. Anal. Chem. 2004, 76, 964-976. (15) Watkins, M. A.; Winger, B. E.; Shea, R. C.; Kentta¨maa, H. I. Anal. Chem. 2005, 77, 1385-1392.
Figure 1. Product ion mass spectrum measured after 5-s reaction of protonated nitrosobenzene with di-tert-butyl peroxide present at a nominal pressure of 6.0 × 10-8 Torr. The most abundant product ion corresponds to the derivatized analyte. A small abundance of an ion of m/z 73 is generated as a side product. The ion of m/z 146 is the result of electron transfer from the peroxide to the protonated analyte.
Scheme 1
react with protonated nitrosobenzene (PA ) 204 kcal/mol)16 as anticipated (Figure 1). Furthermore, the ligand exchange product was not observed for the reaction of di-tert-butyl peroxide with protonated pyridine (PA ) 222 kcal/mol)16 or protonated N,Ndiethylacetamide (PA ) 221 kcal/mol).16 In fact, no reaction was observed between di-tert-butyl peroxide and these protonated analytes. These findings indicate that di-tert-butyl peroxide has limited selectivity for protonated primary N-oxides. However, a ligand exchange product is formed for several protonated compounds other than N-oxides, such as nitrobenzene (PA ) 191 kcal/ mol),16 ethanol (PA ) 186 kcal/mol),16 methanol (PA ) 180 kcal/ mol),16 and acetone (PA ) 194 kcal/mol).16 Therefore, a reagent more selective than di-tert-butyl peroxide is required. The di-tert-butyl peroxide reagent was found to undergo ligand exchange with many protonated oxygen-containing analytes with low PAs. Therefore, the selectivity of the reaction may be improved by using a reagent with a greater basicity. This may (16) Hunter, E. P.; Lias, S. G. Proton Affinity Evaluation. In NIST Chemistry WebBook, NIST Standard Reference Database; Linstrom, P. J., Mallard, W. G., Eds.; National Institute of Standards and Technology: Gaithersburg MD, March 2003; Number 69 (http://webbook.nist.gov).
allow the reagent to preferabily deprotonate analytes with low PAs, be unreactive toward analytes with high PAs, and react with N-oxides by ligand exchange. A disulfide reagent can be expected to undergo ligand exchange via the same mechanism as peroxides, but the sulfur atoms are more basic than peroxide oxygens (e.g., PA of hydrogen peroxide is 161 kcal/mol and PA of dimethyl disulfide is 195 kcal/mol).16 Indeed, dimethyl disulfide was found to undergo a ligand exchange reaction similar to that of di-tertbutyl peroxide with protonated nitrosobenzene (Figure 2), except that the ligand exchange is accompanied by a proton-transfer reaction (Scheme 2). The ligand exchange product was verified to have the correct elemental composition by exact mass measurements. Further support for the proposed structure comes from collision-activated dissociation experiments. The product ion (m/z 139) fragments by loss of a methyl radical to form an ion of m/z 124 with an elemental composition corresponding to the expected product ion C6H6sNH+dS, and by loss of CH2dS to yield an ion of m/z 93 with the elemental composition of the aniline radical cation. The selectivity of the dimethyl disulfide reagent for protonated primary N-oxide functionalities was probed by examining the reactivity of a protonated alcohol, ketone, ether, ester, amide, amine, carboxylic acid, pyridine, nitrile, N-nitrosoamine, tertiary N-oxide, secondary N-oxides, primary N-oxides, and a nitro compound (Table 1). None of the protonated analytes listed above, other than the primary N-oxides (nitrosobenzene and 2-nitrosotoluene), were found to react by ligand exchange accompanied by proton transfer. The only reaction observed was proton transfer to dimethyl disulfide (m/z 95; and a secondary reaction, a nucleophilic addition/elimination reaction of dimethyl disulfide with protonated dimethyl disulfide to yield an ion of m/z 141) for analytes that have PAs at or below the PA of dimethyl disulfide (195 kcal/mol)16 (Table 1). No reaction products were observed Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
5313
Figure 2. Product ion mass spectrum measured after 2-s reaction of protonated nitrosobenzene with dimethyl disulfide present at a nominal pressure of 3.9 × 10-8 Torr. The most abundant product ion corresponds to the derivatized analyte.
Scheme 2
for protonated analytes that have PAs greater than dimethyl disulfide, except for the two primary N-oxides, nitrosobenzene (PA ) 204 kcal/mol)16 and 2-nitrosotoluene (PA unknown) (Table 1). Therefore, the ligand exchange accompanied by proton transfer appears to be unique for primary N-oxides and thus can be used as a diagnostic test for the presence of the primary N-oxide functionality in unknown analytes. Since very few products (predominantly ligand exchange and proton-transfer products) are observed for the reaction of dimethyl disulfide with protonated analytes, the ion-molecule reaction mass spectra are easy to interpret. The feasibility of using the above ion-molecule reaction to screen for molecules with the N-oxide functionality was probed by studying the reactivity of dimethyl disulfide toward several protonated analytes that have the same molecular formula, but different connectivity. Six commercially available analytes with the molecular formula C7H7NO, including one primary N-oxide (Figure 3), were studied. Each analyte was protonated (vide supra) and subsequently allowed to react with dimethyl disulfide. All protonated analytes, except for the primary N-oxide, were found to be unreactive toward dimethyl disulfide. The selective derivatization of the protonated N-oxide, 2-nitrosotoluene, provides 5314
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
Table 1. Reaction of Various Protonated Analytes with Dimethyl Disulfide in a FT-ICR
analyte ethanol acetonitrile nitrobenzene acetone butanol benzoic acid diethylether ethyl acetate nitrosobenzene 2-nitrosotoluene aniline tetramethylpiperidine nitroxide N,N-diethylacetamide pyridine isoquinoline N-oxide di-tert-butylnitroxide nitrosopiperidine
proton affinity (PA) (kcal/mol)a
observed reaction
186 186 191 194 195 196 198 200 204 unknown 211 211
proton transfer to dimethyl disulfideb proton transfer to dimethyl disulfideb proton transfer to dimethyl disulfideb proton transfer to dimethyl disulfideb proton transfer to dimethyl disulfideb no reaction no reaction no reaction ligand exchange and proton transfer ligand exchange and proton transfer no reaction no reaction
221 222 unknown unknown unknown
no reaction no reaction no reaction no reaction no reaction
a PA values are from ref 16. b Protonated dimethyl disulfide reacts with neutral dimethyl disulfide present in the cell to form a secondary product (m/z 141).
Figure 3. Analytes with molecular formula C7H7NO that were protonated and allowed to react with dimethyl disulfide.
Figure 4. Product ion mass spectrum measured for the reaction of protonated nitrosobenzene with dimethyl disulfide in the second quadrupole (Q2) of a triple quadrupole mass spectrometer. Abundant ligand exchange reaction product is evident at m/z 139. Electron transfer from dimethyl disulfide and analyte ion dissociation was observed to form products of m/z 94 and m/z 78, respectively.
support for the usefulness of this ion-molecule reaction for screening for molecules with a primary N-oxide functional group. In an effort to demonstrate the applicability of the above method to mass spectrometers other than FT-ICR, it was adapted to a Finnigan TSQ 700 triple quadrupole mass spectrometer. The effectiveness of the triple quadrupole mass spectrometer as a tool for ion-molecule reaction based compound screening was established by analyzing the six commercially available analytes shown in Figure 3. These analytes were protonated in a CI source (vide supra), mass selected with the first quadrupole (Q1), and screened via reactive collisions with dimethyl disulfide in the second quadrupole (Q2; 3 mTorr, nominal reagent pressure) while the third quadrupole (Q3) was scanned. Of the six protonated analytes, only the primary N-oxide, 2-nitrosotoluene, produced the ligand exchange/proton-transfer reaction product (Figure 4). The only reactions observed for the other five protonated analytes was electron transfer from dimethyl disulfide and minor fragmentation of the analyte ion (note: these reactions were not observed in the FT-ICR since the reactions occur at near-thermal energies in
this instrument whereas the collision energies in the triple quadrupole are near 0.5 eV (11.5 kcal/mol)). CONCLUSIONS The ability to use functional group selective ion-molecule reactions in a mass spectrometer to screen for compounds with a primary N-oxide functionality has been demonstrated. Since this method involves the examination of protonated analytes, it allows for analytes to be introduced into the mass spectrometer by any common API (e.g., ESI or APCI) or CI source. Furthermore, this method has been shown to be robust in that it can distinguish primary N-oxides from secondary and tertiary N-oxides as well as from isomers that have similar connectivities as primary N-oxides. All protonated analytes examined in this work, except for the primary N-oxides, were found to be unreactive toward or react by proton or electron transfer with dimethyl disulfide. Only protonated primary N-oxides were found to undergo a ligand exchange accompanied by proton transfer with dimethyl disulfide. The excellent agreement between the results obtained by using Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
5315
FT-ICR and a triple quadrupole mass spectrometer suggests that less expensive and more readily available instruments can be adapted to perform this ion-molecule reaction based screening method.
donation of the TSQ 700 mass spectrometer. The authors also thank Dr. Steven W. Baertschi of Eli Lilly for useful discussions and guidance during this project.
ACKNOWLEDGMENT
Received for review February 22, 2005. Accepted May 9, 2005.
The authors gratefully acknowledge Eli Lilly and Company for financial support of this work and Lubrizol Corporation for
5316
Analytical Chemistry, Vol. 77, No. 16, August 15, 2005
AC050324B
