Introduction

Cinchona alkaloids are natural products found in the bark of trees of the genus Cinchona. Besides their well-known medicinal use, various cinchona alkaloids and their derivatives have been employed as key structural subunits in the synthesis of other complex natural products [1, 2]. Recently, a new class of cinchona alkaloid-based phase-transfer catalysts (PTC), which are quaternized at both the quinuclidine and quinoline nitrogens, were reported [3]. These catalysts demonstrated excellent reactivity and enantioselectivity in spirocyclization reactions [3].

Mass spectrometry (MS) has become a preferred analytical method for quaternary ammonium salts owing to its high specificity and sensitivity [4, 5]. Tandem mass spectrometry (MS/MS) is able to provide further molecular level structural information for individual ammonium ions [6, 7]. In early studies, the MS and MS/MS analysis of various mono-quaternary ammonium salts were done using different ionization techniques such as thermospray [8], fast atom bombardment (FAB) [9], and atmospheric pressure ionization (API) [10–15]. Generally, these reports have focused either on detection of the quaternary cation, or on selected reaction monitoring (SRM) of a characteristic transition through MS/MS. Only recently have a few mass spectrometric studies of surfactants, pesticides, and pharmaceuticals that have symmetric bisquaternary ammonium centers been reported [16–23]. Among these studies, most of the bisquaternary compounds display differing extents of in-source fragmentation in MS spectra, primarily driven by strong Columbic repulsion between the two charged sites [16, 17]. The distance between two preformed charged sites is essential to the stability of the doubly charged ion in the gas phase [16, 17]. In addition to observing the doubly charged ion M2+, counter anion (A–), and ion pair [M2+ + A-]+, the formation of an [M – H]+ ion was observed as a minor product ion in the MS spectrum of a few symmetrical bisquaternary compounds. This was attributed to a gas-phase Hofmann elimination [16–20]. However, to the best of our knowledge, no previous study had focused on the mechanistic understanding of this reaction in the gas phase.

Mass spectrometry was employed in the present study for a series of novel PTC compounds. The two quaternary ammonium centers in each PTC compound make them an ideal model for the study of gas-phase Hofmann eliminations. As shown in Figure 1, all compounds contain three substituents: R1 to R3. Besides the expected doubly charged M2+ ions and common fragment ions [M2+ – R1 + or R2 +]+, all doubly charged PTCs exhibited a unique [M – H]+ ion as the major product ion in the MS spectra. To better understand these observations, the in-source and MS/MS fragmentation patterns of the novel doubly quaternized cinchona alkaloid-based phase-transfer catalysts (with the general formula depicted in Figure 1) were further studied using electrospray ionization high resolution tandem mass spectrometry (ESI-HRMS/MS).

Figure 1
figure 1

Cinchona-alkaloid based phase-transfer catalysts used in the present study

Full size image

Experimental

Chemicals

All PTC related compounds were synthesized and fully characterized by NMR in a previous report [3]. All PTC was prepared as dibromo-salt. Deuterium oxide (D2O) and deuterated trifluoroacetic acid-d (TFA-d) were purchased from Sigma Aldrich (St. Louis, MO, USA). All solvents used for LC/UV/MS analysis were of OPTIMA LC/MS grade purchased from Fisher Scientific (Pittsburgh, PA, USA).

Mass Spectrometric Analysis

Mass spectrometric analysis was performed on a Waters Synapt G1 Q-ToF operating in positive or negative ion mode. Samples were dissolved in 100% ACN at the concentration of 0.1 mg/mL to avoid any possible anion exchange during ionization. To study potential solvent effect, samples were also dissolved in 50% ACN/water at the same concentration. Samples were introduced into the instrument using direct infusion at the flow rate of 20 μL/min. For the H/D exchange experiments, samples were dissolved in 50% ACN in D2O. Source temperature and desolvation temperature were set at 130 °C and 350 °C, respectively. Nitrogen was used as both cone gas (40 L/h) and desolvation gas (600 L/h), and argon was used as the collision gas. The capillary voltage was set to 3 KV. A cone voltage of 5 V was applied. Leucine enkephalin was used as the lock mass (m/z of 556.2771) for accurate mass calibration and was introduced using the lock spray interface at 20 μL/min at a concentration of 0.5 μg/mL in 50% aqueous acetonitrile containing 0.1% formic acid. During mass spectrometric scanning, data were acquired in centroid mode from m/z 50 to 1000. For MS/MS fragmentation of target ions, collision energies ranging from 10 to 20 V were applied.

Computational Protocol

All calculations were performed with the Gaussian 09 package [24]. Geometry optimization, energy calculations, and transition state identification were carried out with density functional theory, using the M062X functional [25]. The LANL2DZ basis set [26–28] with an effective core potential was used for bromine, and the 6-31+G** basis set [29] was used for other atoms. Frequency analysis was conducted at the same level of theory in order to verify saddle points and obtain thermodynamic energy corrections without scaling at 298.15 K. Computed structures are illustrated using Maestro and Agui [30, 31].

Results and Discussion

MS Spectra of PTCs

Fourteen PTCs (1–14) were studied and their MS in-source fragment ions and relative abundances are summarized in Table 1. Compounds 1–12 were bisquaternized at both the quinuclidine and the quinoline nitrogens. Compounds 13–14 had only one quaternized center at the quinuclidine nitrogen. For bisquaternized compounds 1–12, loss of the substituted benzyl cation R1 + or R2 + led to formation of [M2+ – R1 + or -R2 +]+ as the major product ions. The benzyl cations R1 + or R2 + were also observed as fragment ions. The relative abundances of the doubly charged ion M2+ was less than 30% for most doubly charged compounds. Lowering the cone voltage to 5 V did not increase the amount of doubly charged ion M2+. All PTC compounds with bisquaternized ammonium centers produced the Hofmann elimination product ions [M – H]+. The relative abundance of the [M – H]+ ion was about 30%. Interestingly, no Hofmann elimination product was found for compounds 13–14, which only have one quinuclidine ammonium center. These observations suggested that the major driving force for the extensive in-source fragmentation of the PTCs was strong Columbic repulsion between the two charged sites. Compound 1 was selected as an example to illustrate the MS and MS/MS fragmentation pattern since all other PTC derivatives with different R1 to R3 groups displayed similar MS patterns. As shown in Figure 2, [M2+ – R1 + or R2 +]+ with m/z 385 was the base peak in the MS spectrum. The relative abundance of the doubly charged ion M2+ with m/z 238 was only about 2%. The relative abundance of [M – H]+ product ion was about 30%.Similarly, lowering of both ESI and cone voltage did not increase the relative abundance of doubly charged ions in MS spectrum of 1. Next, efforts were made to locate where the in-source fragmentation was occurring within the API interface.

Table 1 All Ions of 14 PTC Compounds with Different R1, R2, and R3 Groups are Listed with Their m/z Values and Relative Abundances in the MS Spectra. The In-Source Hofmann Elimination Product Ions are Marked in Red
Full size table
Figure 2
figure 2

(a) MS and (b) H/D exchange spectra of compound 1

Full size image

Is the ESI Probe or MS Inlet Responsible for Inducing In-Source Fragmentation of PTCs?

In order to understand the possible location(s) for in-source fragmentation, three experiments were performed. The first experiment was the temperature variation study of the ESI probe and ion inlet. The ESI desolvation gas temperature was lowered from 350 °C to 40 °C and the source temperature was also lowered to 40 °C. However, low desolvation gas and source temperature did not change the MS spectrum of 1 (Supplementary Figure S1 in Supporting Information). The second experiment was NMR confirmation of the ESI spray analyte. A sample of 1 was dissolved in ACN and was directly infused through the (+)-ESI probe. All the analyte was collected by a glass vial before it entered the MS inlet. The collected sample was analyzed by NMR, which showed the collected fraction to be identical to compound 1. These two experiments indicated that the ESI spray alone did not induce fragmentation. The second experiment employed direct infusion of 1 under negative ESI conditions. The bromide anion adduct of the intact PTC dibromo salt was observed under negative ESI conditions when the sample was dissolved in ACN (Supplementary Figure S2), indicating the molecule had not fragmented before it entered the ESI. Hence, it is likely that the electrostatic interaction between the negatively charged bromide anion and the positively charged ion path of the MS inlet induced dissociation of the bromide/ammonium bond. However, due to the potential charge–charge repulsion between the two ammonium cations, the second bromide anion was likely to leave as a neutral molecule such as HBr or benzyl bromide (BnBr) rather than in the anionic form.

MS/MS and H/D Exchange Experiments Reveal the Position of the Double Bond in the Hofmann Elimination Product Ion

For the H/D exchange experiment, the sample was dissolved in 50% D2O/ACN before direct infusion MS analysis. Hence, all the exchangeable protons such as hydroxyl group in the molecule of interest were expected to be replaced by deuterium. Based on the H/D exchange data (Figure 2), deuterium was still present in the product ions [M2+ – R1 + or R2 +]+ and [M – H]+, indicating that the hydroxyl group was intact in the [M – H]+ ion. MS/MS was performed on the [M – H]+ ion with m/z 475 and M2+ ion with m/z 238. As shown in Figure 3, the benzyl quinolinium fragment ion, m/z 220, was found in the MS/MS spectra of both [M – H]+ and M2+, indicating that the additional proton loss was not occurring on the quinolone part of the molecule. The fragment ion with m/z 385 of M2+ and m/z 383 of [M – H]+ established that the proton loss was coming from the quinuclidine part of the molecule. Further H/D exchange followed by MS/MS (Supplementary Figure S3) also indicated that the hydroxyl group was present in the fragment ions observed at m/z 383 and m/z 385 and was not involved in the process of forming the [M – H]+ ion. The MS/MS of the doubly charged ion M2+ did not yield the Hofmann elimination product ions [M – H]+. Taken together, these results support Hofmann elimination occurring on the quinuclidine.

Figure 3
figure 3

MS/MS on M2+ with m/z 238 of compound 1 (top) and [M – H]+ with m/z 475 (bottom)

Full size image

Proposed MS In-Source Fragmentation Mechanisms of PTCs

The proposed gas-phase fragmentation mechanism was supported by DFT calculations. We first compared several DFT functionals (B3LYP, B3LYP-D2, B3LYP-D3, M06-2X) and basis sets (6-31G**, 6-31+G**, LANL2DZ, SDD) to assess their ability to replicate the crystal structure of an analogue of 1. We identified the M06-2X functional and 6-31+G** basis set with the LANL2DZ basis and effective core potential on Br as the optimal combination, given its good agreement with the crystal data. As shown in Figure 4, the calculated low energy conformation of 1 overlaid well with the crystal structure of an analogue. The crystallographic Br anion and the DFT Br anion are separated by a distance of 2.77 Å. This difference is likely due to the presence of a water molecule in the crystal, which is oriented (O-Br distance 3.59 Å, HO-Br angle 154°) to interact with the anion.

Figure 4
figure 4

Calculated low energy conformation of 1 (orange) overlaid onto the small molecule crystal structure (CCDC 1479610) of a close analogue (gray) (heavy atom RMSD~0.3 Å, only the quinuclidine bromide is shown). The hydrogen bonding interaction between the OH group and quinuclidine bromide anion is evident for 1, while the crystal interactions induce a shift of the anion (orange to brown)

Full size image

The proposed in-source fragmentation pathways for compound 1 are summarized in Figure 5. Under positive ESI, the most likely pathway to form the base peak m/z 385 is through bromide anion loss followed by neutral loss of BnBr or HBr in the gas phase. The BnBr and HBr eliminations may be driven by two factors: (1) the resulting singly charged product ion m/z 385 is lower in energy compared with the doubly charged ion m/z 238 (Figure 6); (2) since ionic bond cleavage is calculated to be endothermic in the gas phase, concerted HBr or BnBr elimination could substantially lower the transition state energy compared with Br anion loss alone [32, 33]. The other pathway to form m/z 385 was through benzyl cation loss from the doubly charged m/z 238. The calculation results indicated that losing a benzyl cation, R1, was more favorable than the loss of R2 in the case of compound 1 (Figure 5). Fragmentation leads to both product ions being identified, as shown in Table 1.

Figure 5
figure 5

Proposed in-source fragmentation pathway of 1

Full size image
Figure 6
figure 6

Reaction profile for loss of benzyl cations from the doubly charged cation M2+ (products do not include benzyl cation for clarity)

Full size image

Gas Phase Hofmann Elimination of 1

Solution Hofmann elimination usually involves the concerted bimolecular E2 elimination of the least substituted β proton [34]. In the transition state for Hofmann elimination in solution, the β proton and the quaternary ammonium cation (leaving group) are anti-periplanar [34]. Bulkiness of the quaternary ammonium cation is expected to cause energetically unfavorable gauche interactions with substituents on the adjacent carbon. In order to avoid gauche interactions, elimination proceeds via the least substituted β proton.[35] In the case of gas phase Hofmann eliminations in the studied PTCs, cations 1a and 1b resulting from loss of one bromine anion represent the two possible intermediates that can further undergo elimination (Figure 5). Calculation of relative energies indicates 1b (0.0 kcal/mol) is more stable than 1a (9.73 kcal/mol) due to hydrogen bonding between the OH group and the Br anion (Figure 4). During the ionization process, the quinolone bromide anion is more likely to be lost first to form 1b followed by loss of the quinuclidine bromide anion as a neutral molecule. For the gas-phase Hofmann elimination of 1b, the most likely basic anion available to abstract a β proton was the bromide anion. Moreover, there were four different β hydrogens available for elimination leading to four possible [M – H]+ product ions (2a–2d) (Figure 5).

Reaction profiles for Hofmann elimination reactions (2a–2d) were calculated at M06-2X/6-31+G** level of theory (Figure 7). The results indicated product ion 2b had the lowest activation barrier and lowest relative product energy followed by 2a, 2d, and 2c. Calculations are in good agreement with the MS/MS data in Figure 3. A key fragment ion with m/z 200 could only be formed from Hofmann elimination product ions 2a and 2b but not from 2c and 2d (Figure 8). Taken together, experimental fragmentation patterns and computed energetics indicate product ion 2b is the most likely gas-phase Hofmann elimination product. Interestingly, according to standard interpretation of solution Hofmann eliminations, product 2a should be the major elimination product since 2a is formed through elimination of a β proton on the less-hindered secondary carbon, whereas product 2b is formed through elimination of a β proton on the more hindered tertiary carbon. There are two possible explanations for this: (1) product ion 2b is more stable than 2a due to conjugation between the olefin and aromatic ring of the quinolinium (Figure 7); (2) the transition state leading to product 2b was lower than 2a since the bromide anion was closer to the tertiary β hydrogen due to a hydrogen bonding interaction with the hydroxyl group. The transition state calculations also indicate that the elimination proceeds through a unimolecular Ei elimination pathway [36], in which the bromide anion and the ammonium cation leaving group are syn-periplanar. The unimolecular nature of the gas-phase elimination reaction [37–40] and the tight coordination between the bromide anion and the ammonium cation in the gas phase leads to the formation of 1b as the major gas phase Hofmann elimination product of 1.

Figure 7
figure 7

Reaction pathway for the four possible Hofmann elimination product ions a to d starting from 1b of the transition state for formation of b is shown, in which the Br anion and the ammonium cation are syn-periplanar

Full size image
Figure 8
figure 8

Only Hofmann elimination product 2a and 2b are able to form product fragment ion m/z 200. The conjugation between the olefin and quinolinium is highlighted in bold

Full size image

Conclusions

In this present study, the in-source fragmentation pathways of 14 doubly quaternized cinchona-alkaloid-based phase-transfer catalysts (1–14) were studied under (+)-ESI. The results showed all compounds give [M2+ – R1 + or R2 +]+, [M – H]+, and M2+ ions in the MS spectra. The likely driving force for the formation of these fragment ions was to mitigate the strong Columbic repulsion in the gas phase between the two quaternized ammonium cations. The mechanism for formation of the gas-phase Hofmann elimination [M – H]+ products was elucidated by application of HRMS/MS, H/D exchange, and DFT calculation using compound 1 as an example. The proposed formation of 2b as the major Hofmann elimination product was supported by both HRMS/MS and DFT calculations. Product ion 2b is proposed to form through a concerted unimolecular Ei elimination pathway rather than the bimolecular E2 elimination pathway for common solution Hofmann eliminations. The mechanistic understanding of this unusual in-source fragmentation may aid in the interpretation of structural elucidation efforts for compounds with similar structures.