Discriminating quinine stereoisomers through collision cross-section differences enhanced by cyclodextrin complexation

2.1. Reagents and chemicals

QN and QD (97.00%, MW: 324.42), HQN and HQD (97.00%, MW: 326.42), and cinchonine (CCN, 99%, MW: 294.39) were purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). α-CD (98.00%, MW: 972.84), β-CD (98.00%, MW: 1134.98), and γ-CD (98%, MW: 1297.12) were bought by Aladdin Co., Ltd. (Shanghai, China). All inorganic metal salts (analytical grade) were obtained from Shanghai Macklin Biochemical Co., Ltd (China), while LC-MS grade methanol was sourced from Fisher Scientific (USA). The ESI tuning standard was procured from Agilent Technologies (USA), and ultrapure water was generated using a Millipore Milli-Q system (USA).

2.2. Samples preparation

Considering solubility requirements, 1.0 mmol L^-1 stock solutions of QN, QD, HQN, HQD, and α/β/γ-CD were prepared in methanol-water (1:1, v/v). For analysis, working solutions were obtained by combining these stock solutions with metal ions in the same solvent system.

2.3. TIMS-MS and data analysis

The analyses were performed on a Bruker timsTOF system (Bremen, Germany) featuring an ESI interface for sample introduction [25]. Complete experimental parameters and corresponding analytical results are provided in the Supplementary Materials (Table S1).

2.4. Standard curve

Calibration curves were constructed using both relative and absolute quantification approaches. For relative quantification, sample series covering molar ratios from 10:1 to 1:10 were prepared; for absolute quantification, standard solutions were prepared in a concentration range of 1-50 μg L⁻¹, with CCN as the internal standard at a fixed concentration of 50 μg L⁻¹. The relative amounts of the enantiomers were measured based on their ion mobility peak areas, whereas absolute quantification was performed by calculating the ratio of the MS peak intensities of the analyte to that of the internal standard. The Limit of Detection (LOD) was determined based on a signal-to-noise ratio (S/N) of 3:1. The noise (N) was calculated as the standard deviation of the baseline response from at least 10 blank injections in the vicinity of the analyte’s retention time, while the signal (S) was measured as the peak height of the analyte.

In the recovery analyses, three concentrations, low, medium, and high, corresponding to 5 μg L⁻¹, 25 μg L⁻¹, and 45 μg L⁻¹, respectively, with a QN/QD and HQN/HQD molar proportion of 1:1 were added to the blank practical samples. The concentrations were determined by applying the measured MS intensities to pre-established relative standard curves through back-calculation. Additionally, blank serum and urine samples were spiked with three different molar ratios (1:3, 2:1, and 9:1) at concentrations within the range of 10⁻⁵ mol L⁻¹, followed by relative quantification analysis. The recovery percentage was calculated using the formula: Recovery (%) = (Detected concentration of QN/QD and HQN/HQD in spiked sample / Added concentration) × 100%. Each analysis was conducted in duplicate to obtain the average recovery rate and relative standard deviation (RSD).

2.5. Sampe preparation of practical samples

The sample preparation for the practical samples (serum and urine) was carried out according to a previously reported method [26], with modifications as detailed below. The internal standards were added to the samples prior to the preparation procedure. Briefly, 1 mL of each collected blood and urine sample was centrifuged at 2500 g for 15 min at 4°C to obtain the supernatant. Protein precipitation was then performed using 2 mL of methanol as the organic solvent, followed by centrifugation at 2500 g for 10 min to remove the precipitate. The resulting solution was filtered through a 0.22 μm organic membrane, dried under a stream of N₂, and stored at -80°C for future use.

2.6. Theoretical calculations

Theoretical calculations were employed to predict the conformations of the isomer complexes [QN/QD+γ-CD+H]⁺ and [HQN/HQD+γ-CD+H]⁺. The 3D conformations of QN, QD, HQN, HQD, and γ-CD were retrieved from the PubChem database, with compound identifiers (CIDs) as follows: QN (3034034), QD (441074), HQN (121515), HQD (91503), and γ-CD (5287407), respectively (references are provided in the Supporting Information). Due to the existence of numerous stable, low-energy structures for the complexes, their initial conformations were generated using a molecular docking procedure. For each complex, the ten structures with the lowest energy were selected for subsequent analysis [27]. These conformations were then optimized at the density functional theory (DFT)/UB3LYP/6-31+G(d) level of theory using the Gaussian 09 program [28]. Frequency analyses were performed to confirm that the optimized conformations represented true energy minima. Computational collision cross section (CCS) predictions for the energy-minimized conformers were obtained through trajectory analysis performed with IMoS (version 1.10) [29]; detailed parameters are provided in the Supplementary Information. Experimental CCS data were matched with theoretical predictions to establish the predominant conformations of the complexes, providing a rationale for the experimental observations.

Furthermore, to analyze weak intermolecular interactions (e.g., van der Waals forces, hydrogen bonds), we performed independent gradient model based on Hirshfeld partitioning (IGMH) calculations in Multiwfn, complemented by 3D isosurface visualizations and 2D scatter plots for quantitative assessment [30,31].

3.1. Direct analysis of stereoisomers by TIMS-MS

TIMS-MS analysis was performed on the target enantiomers, yielding MS and extracted ion mobilograms (EIMs) for the two pairs of isomers. As shown in Figures 1(a, b), distinct mass peaks at m/z 325.19 and 327.21 were observed, corresponding to the ions [QN/QD+H]⁺ and [HQN/HQD+H]⁺, respectively. Subsequently, mobility-based separation of the isomeric ions was attempted; however, the extracted mobility peaks were nearly overlapping Figures 1(c, d). Specifically, the mobility values of [QD+H]⁺ and [QN+H]⁺ were 0.803 V·s·cm⁻² and 0.811 V·s·cm⁻², resulting in an Rₚ₋ₚ value of 0.313, indicating that the QD and QN isomers could not be separated. Similarly, the mobility values of [HQD+H]⁺ and [HQN+H]⁺ were 0.811 V·s·cm⁻² and 0.820 V·s·cm⁻², yielding an Rₚ₋ₚ value of 0.311 (specific parameters are provided in Table S2). In summary, the QN/QD and HQD/HQN isomers could not be separated by TIMS-MS under these conditions, indicating that alternative strategies, such as increasing the structural differences between the isomers are required.

Close

Figure 1.

Mass spectra and ion mobility spectra of (a, c) [QN/QD+H]⁺; (b, d) [HQN/HQD+H]⁺.

Export to PPT

3.2 Separation of the isomers by complexing with CD

Since the isomers QD/QN and HQD/HQN could not be separated directly by TIMS-MS, CDs were added as separation reagents to form diastereomeric complexes. As shown in Figure 2(a), three different CDs (α-, β-, and γ-CD) were added as separation agents to the QN/QD mixture, and the corresponding binary complexes were observed. Specifically, the peaks at m/z 1297.38, 1459.36, and 1621.39 correspond to the binary complexes [QN/QD+α-CD+H]⁺, [QN/QD+β-CD+H]⁺, and [QN/QD+γ-CD+H]⁺, respectively. Similarly, the mass spectrum for the mixture of HQD/HQN and CDs has been shown in Figure 2(b), where the peaks at m/z 1299.36, 1461.41, and 1623.43 correspond to the complexes [HQN/HQD+α-CD+H]⁺, [HQN/HQD+β-CD+H]⁺, and [HQN/HQD+γ-CD+H]⁺, respectively.

Close

Figure 2.

Mixed mass spectra of (a) HQN/HQD+α-CD, β-CD, γ-CD; (b) QN/QD+α-CD, β-CD, γ-CD; and Mixture mobility diagram of (a-1) QN/QD+α-CD; (a-2) QN/QD+β-CD; (a-3) QN/QD+γ-CD; (b-1) HQN/HQD+α-CD; (b-2) HQN/HQD+β-CD; (b-3) HQN/HQD+γ-CD.

Export to PPT

Mobility separation of the QD/QN and HQD/HQN isomers was systematically investigated through their CD complexes. As shown in Figures 2(a1-a3), the EIMs of [QN+α-CD+H]⁺ and [QD+α-CD+H]⁺ are broad and nearly overlapping, both concentrated in the mobility range of 1.62 to 1.65 V·s·cm⁻². Moreover, the EIMs for [QN+β-CD+H]⁺ and [QD+β-CD+H]⁺ are largely overlapping with only slight differences, exhibiting a Δmobility value of merely 0.03 V·s·cm⁻². However, when β-CD was replaced by γ-CD, which has a larger cavity, QN and QD showed clear mobility separation, with mobility values of [QN+γ-CD+H]⁺ and [QD+γ-CD+H]⁺ being 1.770 and 1.751 V·s·cm⁻², respectively, and an Rₚ₋ₚ value of 1.12. Regarding the complexes [HQN/HQD+CD+H]⁺ shown in Figures 2(b1-b3), the separation between HQD and HQN improved as the cavity size of the CD increased. For example, the mobility values of [HQN+γ-CD+H]⁺ and [HQD+γ-CD+H]⁺ were 1.772 and 1.751 V·s·cm⁻², respectively, yielding an Rₚ₋ₚ of 1.23 (specific parameters have been provided in Table S2). Overall, the results show that the QD/QN and HQD/HQN isomers can be mobility separated by non-covalent interaction with γ-CD. That is, the interaction between QD/QN or HQD/HQN and γ-CD forms complexes with distinct structural characteristics, thereby enabling their mobility-based separation.

3.3. Theoretical calculation by DFT

Considering that the enantiomers QN/QD and HQN/HQD can be separated by mobility through the formation of binary complexes with γ-CD, the conformations of these complexes warrant further investigation. In this study, theoretical modeling approaches were applied to decipher the mechanism underlying the differential ion mobility of the [QN/QD+γ-CD+H]⁺ and [HQN/HQD+γ-CD+H]⁺ complexes at the molecular level, using the DFT/UB3LYP/6-31+G(d) method. Figure 3 shows the front view (left) and top view (right) of the complexes. The enantiomeric molecules QN/QD and HQN/HQD are embedded within γ-CD, but their interaction sites differ.

Close

Figure 3.

The favored conformation for the complexes of (a) [QN+γ-CD+H]⁺; (b) [QD+γ-CD+H]⁺; (c) [HQN+γ-CD+H]⁺; (d) [HQD+γ-CD+H]⁺.

Export to PPT

Specifically, QN and QD penetrate directly into the cavity of γ-CD, with the quinoline ring interacting inside the hydrophobic cavity of the γ-CD, while the terminal alkene double bond and N-heterocycle remain outside the cavity. This spatial arrangement suggests that the quinoline ring is the primary site of interaction with the γ-CD, likely through non-covalent interactions such as hydrophobic effects and π-π stacking. In contrast, the terminal alkene and N-heterocycle, being positioned outside the cavity, may interact with the hydrophilic exterior of the CD or remain solvent-exposed.

Compared to [γ-CD+QD+H]⁺, the complex [γ-CD+QN+H]⁺ exhibits a more compact structure, which can be attributed to the specific orientation and binding affinity of QN within the CD cavity, enabling tighter packing and more efficient interactions. Similarly, comparing the complexes [γ-CD+HQN+H]⁺ and [γ-CD+HQD+H]⁺, it is evident that the interaction sites between HQN and γ-CD are more concentrated near the upper rim of the CD cavity. The top view of the complexes shows that HQD has more interaction sites with γ-CD than HQN. Thus, HQD forms a more compact structure than HQN, possibly because the interactions of HQD with γ-CD are more spatially concentrated, leading to a more favorable arrangement within the CD cavity.

The predicted CCS values for these molecular systems show strong consistency with empirically determined data obtained from TIMS-MS (Marked in Figure 3 and concluded in Table S3). The error deviations between the theoretical and experimental CCS values range from 4.12% to 8.84%, indicating a high level of consistency. This agreement not only validates the theoretical models used to predict the structures of the complexes but also supports the proposed binding mechanisms and spatial arrangements of QN, QD, HQN, and HQD within the γ-CD cavity.

3.4. Visualization analysis of interaction sites

Moreover, the weak interactions between γ-CD and the isomers were further investigated through visualization analysis in Multiwfn to clarify the interaction differences (Figure S2). To more clearly visualize their interaction sites, we extracted the isomers and their respective binding sites separately. In the [QD+γ-CD+H]⁺ complex, QD exhibits more interaction sites with γ-CD compared to the [QN+γ-CD+H]⁺ complex. Specifically, QD forms weak interactions with γ-CD at four distinct sites: Site 1: an N···O(CD) interaction, and Sites 2–4: H···O(CD) interactions. In contrast, QN does not exhibit these interactions with γ-CD. The additional binding sites in [QD+γ-CD+H]⁺ likely contribute to its more compact structure, resulting in a smaller CCS value (see the inset Figure in Figure 4a, b).

Close

Figure 4.

Visual study of weak interaction between γ-CD and the isomers and their scatter plots (a) [QN+γ-CD+H]⁺; (b) [QD+γ-CD+H]⁺; (c) [HQN+γ-CD+H]⁺; (d) [HQD+γ-CD+H]⁺.

Export to PPT

Similarly, in the [HQN+γ-CD+H]⁺ and [HQD+γ-CD+H]⁺ complexes, HQN and HQD share similar interaction sites with γ-CD (see the inset Figure in Figure 4c, d). However, HQD forms an additional weak H···O(CD) interaction at Site 1, which is absent in HQN. Furthermore, the scatter plots reveal that [HQD+γ-CD+H]⁺ exhibits stronger interactions compared to [HQN+γ-CD+H]⁺. Specifically, in the range of −0.06 to −0.02 a.u., [HQD+γ-CD+H]⁺ shows additional blue points, possibly indicating stronger hydrogen bonds (Figure S3); in the 0.00 to 0.02 a.u. range, [HQD+γ-CD+H]⁺ also displays more pronounced weak hydrogen bonds or van der Waals interactions (Figures 4c,d). These findings provide valuable insights into the molecular interactions between CDs and QN stereoisomers, which could facilitate the development of more effective separation and analytical techniques in pharmaceutical research.

3.5. Diastereomeric structural stability comparison

MS/MS, as an analytical technique based on mass spectrometry, provides more detailed structural and chemical information about complexes. It is hypothesized that structural differences in the complexes may lead to variations in their stability, which could be reflected in their MS/MS behavior. Therefore, we performed a comparative analysis of the fragment ion peaks of the complexes and investigated the stability of intermolecular non-covalent interactions.

As shown in Figures 5(a, b), the diastereomeric complexes exhibit identical fragment ion peaks, both displaying peaks corresponding to [γ-CD+H]⁺ and the free analyte, but no peaks corresponding to γ-CD monosaccharide fragments bound to the analyte. Additionally, the conformational stability of the enantiomeric complexes was evaluated using survival yield (SY) curves, which revealed differences in dissociation collision energies among the complexes. Specifically, the 50% dissociation energy for [QN+γ-CD+H]⁺ was 41.94 eV, while that for [QD+γ-CD+H]⁺ was 43.91 eV (Figure 5c). Similarly, the 50% dissociation energy for [HQN+γ-CD+H]⁺ was 40.73 eV, compared to 42.85 eV for [HQD+γ-CD+H]⁺ (Figure 5d). These results indicate that the denser structure of [QD+γ-CD+H]⁺ requires higher energy for collisional dissociation compared to [QN+γ-CD+H]⁺. Likewise, [HQD+γ-CD+H]⁺ exhibits higher dissociation energy than [HQN+γ-CD+H]⁺, suggesting that HQD forms a more stable complex with γ-CD.

Close

Figure 5.

MS/MS spectra and survival yield plots of ternary complex ions of (a,c) [QN/QD+γ-CD+H]⁺; (b,d) [HQN/HQD+γ-CD+H]⁺.

Export to PPT

By integrating theoretical calculations, we observed a strong correlation between the collision energy of the diastereomeric complexes and their experimental CCS values. Specifically, complexes with higher dissociation energies corresponded to smaller CCS values and more extensive weak interaction networks with γ-CD ([QD+γ-CD+H]⁺ and [HQD+γ-CD+H]⁺). This suggests that these complexes possess more compact structures, highlighting the potential of this approach as an enantiomer recognition method.

3.6. Quantitative analysis of enantiomers

The isomers were quantitatively analyzed by the TIMS-MS method, including absolute quantification by MS with internal standards and relative quantification by the TIMS method. Relative quantitative analysis for the diastereomeric complexes was performed with a solution concentration of 10⁻⁵ mol L⁻¹, with the molar ratios of QN/QD and HQN/HQD ranging from 1:10 to 10:1. As shown in Figure S4, as the molar ratio of the isomers increases, the intensity of the corresponding EIM peak also increases. The relative intensity of the enantiomers was determined by comparing the corresponding mobility peak area to their molar concentration. Linear regression equations have been summarized in Table 1, yielding good linear coefficients (R² > 0.99) with RSD ≤ 2.16%. Moreover, LODs for the complexes were calculated based on S/N = 3, yielding relative quantification ranges of 1:22.3–23.3:1 for QN/QD and 1:24.4–25.2:1 for HQN/HQD.

Meanwhile, absolute quantification was measured to determinate the total composition of the isomers. To enhance quantification precision, CCN was employed as an internal standard at a fixed concentration of 50 μg L^-1. A 1:1 mixture of the isomers was prepared and quantified over a concentration range of 1 to 50 μg L^-1 for the two isomer groups. The analytes were quantified based on the MS peak intensity ratio of [QN/QD+H]⁺ (m/z 325.14) or [HQN/HQD+H]⁺ (m/z 327.16) to [CCN+H]⁺ (m/z 295.26), which was then correlated with their respective concentration ratios. Table 1 presents the calibration curve equations, obtaining acceptable R² of 0.9976 for QD/QN and 0.9993 for HQD/HQN, with RSD ≤2.32%. Moreover, the LODs (S/N=3) for the isomers were calculated 0.183 and 0.256 μg L^-1 for QN/QD and HQN/HQD, respectively. These results demonstrate that the proposed method provides accurate quantification for the QN/QD and HQN/HQD isomers.

Additionally, the accuracy of the quantitative analysis was further evaluated through quality control (QC) at three different concentrations, with back-calculation using the calibration curves. For absolute quantification, QC was performed at low (5 μg L⁻¹), medium (25 μg L⁻¹), and high (45 μg L⁻¹) concentrations. For relative quantification, QC was conducted at low (1:3), medium (2:1), and high (9:1) molar ratios. The detection results are presented in Table 2, demonstrating a recovery rate of ≥86.0%, with an acceptable accuracy indicated by an RSD of ≤3.78%. The standard deviation of recoveries (SR) for the isomers ranged from 1.08 to 1.22 [32] (the calculation formula is provided in the Supporting Information). Overall, the method exhibited robust accuracy and precision across both absolute and relative quantification, making it suitable for reliable quantitative analysis in complex matrices.

3.7. Recovery analysis in serum and urine sample

Considering that QN and HQN are important drugs present in patient serum and urine after administration, it is crucial to detect their levels in these biological samples. Therefore, evaluating the feasibility of the proposed method in artificial serum and urine is essential. As shown in Figure 6, the target isomers were undetectable in blank matrices (artificial serum/urine). After adding QN/QD (molar ratio 1:1) and HQN/HQD (molar ratio 1:1) at 50 μg L⁻¹, along with γ-CD, prior to sample pretreatment, ions corresponding to [QN/QD+H]⁺ and [HQN/HQD+H]⁺ were clearly detected, and the binary complexes [QN/QD+γ-CD+H]⁺ and [HQN/HQD+γ-CD+H]⁺ were also observed in both artificial serum and urine. Meanwhile, mobility separation of the complexes was successfully achieved, as demonstrated in the inset figure, showing well-resolved peaks consistent with those in Figure 2.

Close

Figure 6.

Mass spectra for (a) blank urine sample; (b) blank serum sample; (c) the added QN/QD and γ-CD in urine sample; (d) the added QN/QD and γ-CD in serum sample; (e) the added HQN/HQD and γ-CD in urine sample; and (f) the added HQN/HQD and γ-CD in serum sample.

Export to PPT

Accordingly, recovery analyses for both relative and absolute quantification using this method in the matrices were performed. Specifically, equimolar standard solutions at three different concentrations, low (5 μg L⁻¹), medium (25 μg L⁻¹), and high (45 μg L⁻¹), were added to blank serum and urine samples to determine the absolute quantification recovery rate. CD-mediated chiral discrimination further facilitated identification through characteristic host-guest complex formation. Similarly, QN/QD and HQN/HQD mixtures at three different molar ratios (1:3, 2:1, and 9:1) and a concentration of 10⁻⁵ mol·L⁻¹ were added to blank serum and urine samples to determine the relative quantification recoveries. The detected results, calculated using the corresponding linear regression equations for the two practical samples, have been shown in Table S4. The absolute quantification recoveries for the two isomer groups were ≥85.6% with RSD ≤ 4.89% (n = 5), while the relative quantification recoveries were ≥84.0% with RSD ≤ 4.64% (n = 5). These recovery rates are comparable to those obtained for standard samples. In summary, the recovery results indicate that the proposed method exhibits acceptable accuracy, making it a feasible approach for detecting QD/QN and HQD/HQN in serum and urine samples.