Precise analysis of thyroxine enantiomers in pharmaceutical formulation by mobility difference based on cyclodextrin

2.1 Chemicals and samples

D-T4 (MW: 776.87, 98%), L-T4 (MW: 776.87, 98%), L-Triiodothyronine (L-T3, MW: 650.97, 98%), α-CD (MW: 972.84, 98.00%), β-CD (MW: 1134.98, 98.00%), and γ-CD (MW: 1297.12, 98.00%) were obtained from Aladdin Co., Ltd. (Shanghai, China). All inorganic metal salts reported in this work are analytical grade and bought by Shanghai Macklin Biochemical (Shanghai, P. R. China). Methanol (LC-MS grade) and acetonitrile (ACN, LC-MS grade) were acquired from Fisher Scientific Inc. (Pittsburgh, PA). The ESI-Low Concentration Tuning Mix was purchased from Agilent Technologies (Santa Clara, CA). Deionized water used in this work was freshly prepared using a Milli-Q water purification system (Bedford, MA).

2.2 Instrumentation and software

All ion mobility analysis for the complexes was performed on a TIMS-TOF instrument from Bruker Daltonik (Bremen, Germany) equipped with electrospray ionization (ESI) source. And the extensive experimental data analysis and processing is performed using the DataAnalysis 5.0 software. And the details are described in the Supporting Information section 2.

2.3 Peak-to-peak resolution

The peak-to-peak resolution (R_p-p) is used to quantify the degree of separation of the two neighboring peaks, which can be calculated by the following formula: (Wu et al., 2022) R_p-p = 2.35 × (A₂ − A₁) / (2 × W_FWHM1 + 2 × W_FWHM2) (3)

Where A₁ and A₂ represent the different 1/K₀ of two adjacent peaks, and W_FWHM1 and W_FWHM2 is their corresponding full peak width at half-maximum.

2.4 Chemical calculation

Theoretical structures of the diastereomer complexes of [β-CD + D-/L-T4 + H]⁺ and [β-CD + D-/L-T4 + Ca]²⁺ were optimized using the Gaussian program (Weedbrook et al., 2012; Wu et al., 2022). The candidate structures for [β-CD + D-/L-T4 + H]⁺ were first proposed by AutoDock Vina program to get all molecules of interest (Eberhardt et al., 2021), and the preferred structures were further optimized using Gaussian 09 with M062X/LANL2DZ. Besides, the interest structures of [β-CD + D-/L-T4 + Ca]²⁺ were measured by AutoDock Vina program via mixing Ca²⁺ to the favored structure of [β-CD + D-/L-T4 + H]⁺. And the favored structures of [β-CD + D-/L-T4 + Ca]²⁺ were optimized using Gaussian 09 with M062X/LANL2DZ as well. Vibrational frequencies were calculated to guarantee that the optimized structures corresponded to a stable energy space, and zero-point energy corrections were applied to calculate the relative stability between the structures. Finally, theoretical CCSs for the favored diastereomer complexes were calculated using IMoS software v1.10 (Shrivastav et al., 2017).

2.5 Sample pretreatment for the tablet drug

The levothyroxine tablets were bought from the local drugstore in Ningbo (Zhejiang, China), and stored at 4 °C for detection. To extract the analytes, drugs are first removed from the thin film coating and ground to a uniform powder using an agate mortar and pestle. 1.0 g power was dissolved in deionized water (10.0 ml) and sonicated for 2 min. The extract was then filtered by a 0.22 μm drainage filter membrane as the stock solution and stored at 4 °C. The detection solution was obtained by taking 1.0 ml of stock solution and diluted 10-folds with ACN/deionized water (50%:50%, v/v).

2.6 Method validation

The proposed method, based on TIMS-MS for D-T4 and L-T4 chiral enantiomer quantitative analysis was validated, which measured by identifying the different molarities for the isomers (each having ten data points). Relative quantitative were constructed by measuring different isomer mixture ratios, which comprise the molality ratios measured for D-T4: L-T4 ratios range from 10:1 to 1:10 at the concentration level of 10^-6 mol L^-1. The mixture was then incubated for 15 min at room temperature to equilibrate before executing the TIMS-MS detection process. Here, the calibration curves of relative quantification for the isomer complexes were constructed by comparing their peak area intensity ratio (y) against their corresponding relative concentration (x). Moreover, absolute quantification for T4 were also measured by a series of concentrations ranged from 1.00 to 10⁴ ng mL⁻¹ containing 10.0 ng mL⁻¹ of T3 as internal standard, and the calibration curves of absolute quantification was measured on the ratio of the intensities of the mass spectrum peaks (T4/T3). Furthermore, the accuracy of the method was tested by spiking with a known amount of D-T4 and L-T4, which encompassed three different concentrations of a low, medium, and high sensitivity, and was performed in duplicate to determine the average recovery.

3.1 Mobility separation of T4 enantiomer in complexing with CDs

The MS and mobility analyses were first performed for the T4 enantiomers by complexing with CDs. As shown in Fig. 2, the m/z 777.00, 1750.02, 1912.08, and 2074.15 are contributed to the ions of [T4 + H]⁺, [α-CD + T4 + H]⁺, [β-CD + T4-H]⁺, and [γ-CD + T4-H]⁺, respectively. The results revealed that the T4 enantiomers can non-covalent interaction with CDs and form the corresponding binary complex of [TH + CD-H]⁺. And then their mobility analysis was studied. As shown in Fig. 1, the identical extraction ion mobilities (EIMs) were found for the complexes of [D-T4 + H]⁺ and [L-T4 + H]⁺, revealing that the D and L-T4 enantiomers cannot be mobility separated by their own protonated ions.

Close

Fig. 1

The extracted ion mobility spectra and structure for the D/L-T4 enantiomers.

Export to PPT

Besides, TIMS analysis for the binary complexes of [CD + T4-H]⁺ were studied and shown in the right of Fig. 2. The EIMs between the [CD + D-T4-H]⁺ and [CD + L-T4 + H]⁺ have certain difference. Specifically, the mobility value for [α-CD + D-T4 + H]⁺ is 1.767 V·s·cm⁻² and for [α-CD + D-T4 + H]⁺ is 1.774 V·s·cm⁻², which the calculated R_P-P is 0.411. Identical, the mobility value for [β-CD + D/L-T4 + H]⁺ and [γ-CD + D/L-T4 + H]⁺ are 1.754/1.758 and 1.847/1.849 V·s·cm⁻², with the R_p-p for D/L-T4 separation are 0.235 and 0.118, respectively. The results reveal that the separation effect R_p-p for D/L-T4 enantiomer was decreased as the size of the inner cavity of the CD increasing, which probably because the large CD inner cavity allows more entry of the target analyte that resulting in less structural difference for the diastereomer complex.

Close

Fig. 2

Mass spectra for TH in complex with CDs and their corresponding EIM (a) [α-CD + T4 + H]⁺; (b) [β-CD + T4 + H]⁺; (c) [γ-CD + T4 + H]^+.

Export to PPT

3.2 Mobility separation of T4 in complexing with CDs and metal ions

Through the D/L-T4 enantiomer can be mobility recognized by their formed binary complexes, but the resolution is not satisfactory when their enantiomeric complexes coexist. In this case, the center ligand of metal ions was added to study their separation effect. As shown in the left of Fig. 3, Ca²⁺ was served as an example to complex with the CDs and the D/L-T4, which m/z 894.83, 975.52, 1056.55 are attributed to the complex ions of [α-CD + T4 + Ca]²⁺, [β-CD + T4 + Ca]²⁺, and [γ-CD + T4 + Ca]²⁺, respectively. The MS results reveal that Ca²⁺ can as a ligand to coordinate with CD and T4, and the corresponding ternary complexes can be formed. For other MS peaks in the mass spectra, which are the complexes formed by the interaction of CD with other substances, because metal salts are only in analytical grade and the contaminants may be present in the mass spectrometer. And then the mobility study was measured for the ternary complexes. As shown in the right of Fig. 3, significantly different EIMs were observed for the D- and L-ternary complexes. For example, mobility values for [α-CD + D-T4 + Ca]²⁺ and [α-CD + L-T4 + Ca]²⁺ are 0.941 and 0.951 V·s·cm⁻², with the calculated R_P-P of 0.529. Similarly, mobility values for [β-CD + D/L-T4 + Ca]²⁺ and [γ-CD + D/L-T4 + Ca]²⁺ are 0.967/0.981 and 1.015/1.023 V·s·com⁻², resulting in R_P-P of 0.764 and 0.470, respectively.

Close

Fig. 3

Mass spectra for TH in complex with CDs and Ca²⁺ and their corresponding extracted ion mobility (a) [α-CD + T4 + Ca]²⁺; (b) [β-CD + T4 + Ca]²⁺; (c) [γ-CD + T4 + Ca]^2+.

Export to PPT

In addition, other divalent metal ions of Mg²⁺, Mn²⁺, Fe²⁺, Co²⁺, and Ni²⁺ were also studied as ligand for D/L-T4 separation. As shown in Fig. 4, the studied metal ions all have different effect for D/L-T4 separation, which the R_P-P can reach to 1.59 by the formed complex of [α-CD + D/L-T4 + Ni]²⁺. In comparison, α-CD related complex seems has the best separation effect for D/L-T4 enantiomers in most case. However, the separation resolution of the D/L-T4 may be reduced with the coordinated of the metal ions (Mn²⁺), which probably because Mn²⁺ was interact with the narrow rim of the CD, and majority of the D/L-T4 is in the CD chamber, that reduced their separation effect and have essentially the same separation resolution. But this is not the focus of this work just an empirical guess, and the specific results require further theoretical calculations. And the accuracy for the method was measured in quintuplicate in consecutive days, the calculated RSD% was in a range of 0.22 to 0.50 %, reveling the method has good repeatability and stability. Moreover, considering D/L-T4 can be baseline separation by complexing with β-CD and Ca²⁺ (R_p-p = 0.764) (Fig. 3b-2), and the advantages of high selectivity, non-toxic, chemically stable of β-CD, that [β-CD + D/L-T4 + Ca]²⁺ was selected for further studying.

Close

Fig. 4

Separation effect R_p-p for the D/L-T4 by complex with different metal ions and CDs.

Export to PPT

3.3 Parameter optimization for detection

Detection parameters such as solvent, molar ratio among CD, metal salts and T4, and ionization voltage all can affect the formation of the complexes. In this case, the formation of [β-CD + D/L-T4 + Ca]²⁺ was studied. As shown in Fig. 5a and Fig. 5b, methanol and ACN as solvent were investigated, revealing ACN was more benefited for the formation of [β-CD + D/L-T4 + Ca]²⁺ than methanol. Besides, the relative intensity of the complex was increased when ACN content was increased from 30% to 50%, and when ACN content continued to increase, its intensity decreased. Moreover, the molar ratio among the β-CD, T4 and Ca²⁺ was studied, with the molar ratio of Ca²⁺ to T4 increased from 0.4 to 0.8, the relative intensity of [β-CD + D/L-T4 + Ca]²⁺ was increasing, but its intensity of the complexes was decreased as the molar concentration of the salt was continued to increase (Fig. 5c). Similarly, the molar ratio of β-CD to T4 was also studied, a molar ratio of 1.0:0.8:0.8 for T4: β-CD: Ca²⁺ was best (Fig. 5d). The flow rate was further optimized, it was shown that the strength of the complex decreases with the increase of the flow rate (Fig. 5e). Finally, the ionization voltage was optimized in Fig. 5f, as the ionization voltage increase, the detection efficiency of the target decreases instead, revealing that 3000 V is the favored ionization efficiency in this case. Thus, the favored detection parameters were as follows, the molar ratio of T4: β-CD: Ca²⁺ was 1.0:1.2:0.8, solvent was 50% ACN /water, detection flow rate was 2 ul min⁻¹, and the ionization voltage was 3000 V.

Close

Fig. 5

Parameter optimization for the detection of complex [β-CD + T4 + Ca]²⁺ (a) Methanol content in the solvent; (b) ACN content in the solvent; (c) Molar ratio of Ca²⁺ to T4; (d) Molar ratio of β-CD to T4; (e) flow rate of detection; (f) ionization voltage;

Export to PPT

3.4 Chemical theoretical calculation

The mobility differences of the diastereomeric complexes revealed D/L-T4 have different interaction effects with the CDs. The favored conformation for the complexes were optimized to match and explain their micro-interactions (Campuzano, et al., 2012; Shrivastav, et al., 2017). The possible structures for each complex were first optimized by AutoDock software, and the structures with low energy were selected and further optimized to obtain the final structure by Gaussian. Herein, the favored conformations of D/L-T4 in complexing with β-CD and Ca²⁺ were optimized as examples for interpretation by chemical theoretical calculations using M062X/LANL2DZ (Detailed in the Supporting Information), and the energy for the complexes are shown in Table S1 and Table S2. As showed in Fig. 6a and Fig. 6b, similar modes of interaction were observed for [β-CD + D-T4 + H]⁺ and [β-CD + L-T4 + H]⁺. The CCS values between experimental and chemical theoretical for the complexes were displayed in Table 1, theoretical CCSs for [β-CD + D-T4 + H]⁺ and [β-CD + L-T4 + H]⁺ are 387.63 and 383.56 Å², which have good consistent trend with the experimental values. Meanwhile, the structural difference for the complexes was expanded with the addition of Ca²⁺. As shown in Fig. 6c and Fig. 6d, Ca²⁺ interacts with β-CD and D/L-T4 at different locations, where Ca²⁺ wrapped inside D-T4 and β-CD, while Ca²⁺ was interact with T4 on the small orifice surface of β-CD. Accordingly, the theoretical CCSs for [β-CD + D-T4 + Ca]²⁺ and [β-CD + L-T4 + Ca]²⁺ are 440.03 and 431.65 Å², compared with their experimental CCSs values of 399.0 and 392.0 Å² (Table 1), similar trend of the CCS value changes (RSD ≤ 13.87%) were obtained.

Close

Fig. 6

Structure for the complexes of (a) [β-CD + D-T4 + H]⁺; (b) [β-CD + L-T4 + H]⁺; (c) [β-CD + D-T4 + Ca]²⁺; (d) [β-CD + L-T4 + Ca]^2+.

Export to PPT

Indeed, CCS value errors between the experimental and theoretical were between 10.11 and 13.87%, which were caused by the structural form difference between the calibrated tuning mixture and the diastereomer complexes, the trajectory method for CCS calculation, and the parameter setting of the IMoS program. However, the CCS trends are consistent between experiment and theory (Olajide et al., 2021), the conformation difference for D-/L-T4 was increased by interaction with β-CD and Ca²⁺ and resulted in different effects on the mobility separation.

3.5 Quantitative analysis for the D/L-T4 enantiomers

As the D/L-T4 can be separated by the proposed method, that their quantitative analysis of absolute and relative quantification was evaluated. Herein, the absolute quantitative was measured with L-T3 as the internal standard, keeping L-T3 in a fixed concentration (10.00 ng mL⁻¹), and changing the concentration of D/L-T4 from 1.00 to 10⁴ ng mL⁻¹ in seven points. As shown in Fig. 7a, MS spectra of mixtures of L-T3 (10.00 ng mL⁻¹), D/L-T4 (1:1, 50.0 ng mL⁻¹), and Na⁺ (100.0 ng mL⁻¹) were studied, the ions were present as the form of Na⁺ coordination, such as the ions of [T4/T3 + Na]⁺, [T4/T3 + 2Na-H]⁺, [T4/T3 + 3Na-2H]⁺. In this case, the absolute quantitative was measured by the intention of [T4 + H]⁺, [T4 + Na]⁺, [T4 + 2Na-H]⁺, [T4 + 3Na-2H]⁺ to the intensity of [L-T3 + H]⁺, [L-T3 + Na]⁺, [L-T3 + 2Na-H]⁺, [L-T3 + 3Na-2H]⁺. The results are shown in Fig. 7b, yielding good linear (R² = 0.9904) with an RSD slop of 1.26% (n = 5); the limit of detection (LODs) and limit of quantitation (LOQ) calculation for the T4 by the signal-to-noise ratio (S/N = 3) and (S/N = 10) were 0.11 ng mL⁻¹ and 0.49 ng mL⁻¹, respectively.

Close

Fig. 7

Absolute quantification using the T3 as the interior label (a) Mass spectra for the mixture TT (100.0 ng/mL) T4 (5000 ng/mL); (b) standard curve; Relative quantification for the D/L-T4 enantiomers (c) extracted ion mobility D: L = 10:1–1:10; (d) standard curve.

Export to PPT

Besides, a series of different molar ratios of D-T4 and L-T4 changed from 10:1 to 1:10 in complexing with β-CD and Ca²⁺ were measured. EIMs for the diastereomer complexes [β-CD + T4 + Ca]²⁺ obtained by the mixtures are shown in Fig. 7c, the mobility intensity for the ions was increased with its ratios increasing. And the calibration curve for D/L-T4 relative quantification was assessed by plotting the mobility peak area intensity of D-T4/L-T4 versus the relative concentration of D-T4/L-T4. As displayed in Fig. 7d, good linear relationship between the relative intensity and the relative concentration with a R² value of 0.9901, and RSD (n = 5) of 2.78% were obtained. Moreover, the detection scope of the method was also measured, which in the range of 45.6:1 to 1:59.8 by the proposed method.

Moreover, accuracy of quantitative analysis by the method was further measured by quality control (QC) in three different concentrations and back-calculation using the calibration curves. For the accuracy of absolute quantification, QC in three concentrations of low (3.0 ng mL⁻¹), medium (500.0 ng mL⁻¹) and high (9000.0 ng mL⁻¹), and three molar ratios for relative quantification were low (10:3), medium (10:9), and high (1:9) were measured. The detection results are shown in Table 2, accept accuracy with RSD ≦5.67%. And a simple comparison of the proposed method with the prior methods was measured in Table S3, this method has the advantages of high sensitivity, short time and simple operation. Thus, the relative quantification results revealed that the proposed method based on complexing with β-CD and Ca²⁺ and detected on TIMS-MS, can be a promising approach for D/L-T4 enantiomer detection.

3.6 Detection of D/L-T4 enantiomers in practical samples

The proposed method for D/L-T4 detection in practical sample of drug tablets as example for method validation. Firstly, the total T4 content (including D-T4 and L-T4) was detected by using L-T3 as the internal standard. And the content of the T4 was back calculated by the absolute quantitative standard curve in Fig. 7b via their mass peak intensity. Accordingly, T4 detected in the tablets was 45.3–52.6 ug with the RSD ≤ 4.65%, which was basically the same as the amount descripted in the manual. Moreover, the sample was further mobility separating to determine whether L-T4 contains D-T4 in the tablets. The TIMS-MS results was shown that the D-T4 was basically undetectable in the drug tablets, revealed this drug tablets with better purity.

Additional, 5.0 ug D-T4 was added in the tablet, and TIMS-MS detected by complexing with β-CD and Ca²⁺ in a diluted concentration of 10^-6 mol L ^-1. The detected molar ratios of D-T4 and L-T4 were back calculated by the measured relative quantitative standard curves, the detected molar concentration ratio was 1.0 ± 0.32:10.0 ± 0.54 for the D-T4: L-T4. These results reveal that the proposed method can be applied to D/L-T4 enantiomer analysis and take the fake test in practical samples.