2.1 Materials

(±)-ETO (>98.0%) was obtained from TCI (Tokyo, Japan). R-MCF (>99.0%), pyridine, acetic acid, formic acid (FA), L-proline, and S-naproxen sodium (S-NAP; >98%) were purchased from Sigma–Aldrich (St. Louis, MO, USA). Ethanol (EtOH) and sodium chloride (NaCl) were purchased from Merck (Darmstadt, Germany). HPLC-grade water, methanol (MeOH), and acetonitrile (ACN) were obtained from J.T. Baker (Center Valley, PA, USA). A few lots of mouse serum, each of which was a pooled preparation from a normal mouse population and pooled for the method development and validation processes, were purchased from Sigma–Aldrich. Other chemicals were of analytical grade and purchased from Sigma–Aldrich unless otherwise indicated.

2.2 Instruments and operation conditions

2.3 Preparation of standard and reagent solutions

A standard stock solution of racemic ETO was prepared at a concentration of 1000.0 μg mL⁻¹ in ACN. Standard working solutions were prepared in the range of 1.0–100.0 μg mL⁻¹ by subsequently diluting the stock solution in ACN. Stock solution and working solution of S-NAP, the internal standard (IS), were prepared in EtOH. All standard working and stock solution were stored at −20 °C until use. The derivatization reagent, 200 mM R-MCF, was prepared in ACN and stored at 4 °C.

2.4 Chiral derivatization reaction procedure

Equal volumes of ETO working solution and mouse serum were combined to yield 100 μL of a mixture which was then spiked with 20 μL of a 1.0 μg mL⁻¹ solution of S-NAP. After 880 μL of ACN was added to precipitate protein, the mixture was vortexed for 3 min and centrifuged at 12,300 g for 5 min. A total of 950 μL of the supernatant was removed and dried under nitrogen gas purge. The residues were derivatized as follows: 125 μL of 200 mM R-MCF, 58 μL of ACN, and 17 μL of pyridine were added, and the reaction mixture was capped, followed by ultrasonic radiation for 13 min. The reaction was quenched by the addition of 170 μL of 200 mM L-proline solution and subsequently vortexed for 3 min followed by incubation at room temperature for 10 min. Phase separation was readily achieved by the addition of 20 mg of NaCl to the reaction mixture and subsequent brief vortexing. The upper organic phase of 140 μL was removed and filtered through 0.45 μm pore membrane filters (Millipore, Tullagreen, Ireland) before injection into the HPLC system.

2.5 RSM for the optimization of derivatization conditions

RSM based on a central composite design was performed using Design-Expert 8 (Stat-Ease Inc., Minneapolis, MN, USA). The optimized variables were as follows: (1) A, volume of 200 mM R-MCF, (2) B, ultrasonic radiation time, and (3) C, pyridine volume. The total sum of two peak areas of derivatized ETO diastereomers was used as the response.

2.6 Assessment of the derivatization reaction rate

The derivatization reaction products of R- and S-ETO were examined for 0–20 min of ultrasonic radiation. The reaction conditions were based on the optimized conditions except for the reaction time. The total peak areas of the R- and S-ETO derivatives were used for the total reaction yields. Enantiomeric fractions (EFs) were measured for the R-ETO derivatives using the following equation (Polo-Diez et al., 2015; Ribeiro et al., 2014): $$\mathit{Enantiomaric}\mathit{fraction}(\mathit{EF})=\frac{\text{peak area of}R\text{enantiomer}}{\text{peak area}R\text{enantiomer}+\text{peak area of}S\text{enantiomer}}\text{.}$$

2.7 Density functional theory method for computing dipole moment of derivatives

A computational study using density functional theory (DFT) was performed to compute dipole moment of the ETO derivatives. The old, but popular B3LYP (Becke, 1993; Lee et al., 1988) functional, and the more recent M06-2X (Zhao and Truhlar, 2008) functional were considered. The basis set used for all atoms was 6-31G(d) (Hehre et al., 1986). All computations were performed using Gaussian09 (Frisch et al., 2009). The diastereomers, R-ETO:R-MCF and S-ETO:R-MCF were fully optimized in MeOH and ACN as the solvents. Solvent effects were taken into account in the framework of self-consistent reaction field (SCRF) (Cances et al., 1997; Cossi et al., 1996; Barone et al., 1998) based on Polarizable Continuum Model (PCM) as developed by Tomasi’s group (Tomasi and Persico, 1994). Frequency computations were performed on the optimized diastereomers using the same methods as for the optimization to check the nature of the stationary points.

2.8 Method validation

The established method was validated with respect to its constituent parameters according to the United States Food and Drug Administration bioanalytical method validation guideline (Food and Drug Administration, 2001).

2.9 Animal test for real sample analysis

Eight-week-old male ICR mice were purchased from Koatech Co., Ltd (Seoul, Korea). All animal care procedures were conducted in accordance with the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. The procedures were reviewed and approved by the Institutional Animal Care and Use Committee of the School of Pharmacy, Sungkyunkwan University (Approval No. SKKUIACUC-20150034). A single dose of ETO racemates at 20 mg kg⁻¹ was administered via intraperitoneal injection, and serum was collected 30 and 60 min after administration.

3.1 Derivatization of ETO using R-MCF, and characterization of its reaction chemistry

Chloroformates can react with aliphatic alcohols, amines, and carboxylic acids to form carbonates, carbamates, and esters, respectively (Matzner et al., 1964; Lee et al., 2014). These facts suggest that both the secondary amine and the carboxylic acid group of ETO are the functional groups capable of reaction with R-MCF (Matzner et al., 1964). Pyridine has been used as a catalyst and scavenger of the hydrogen chloride formed during the derivatization reaction of chloroformates with amines and carboxylic acids (Lee et al., 2014; Farsam et al., 1999).

Based on the above facts, we examined the derivatization reactions between ETO racemates and R-MCF with and without the addition of pyridine. The multiple conjugated structure of ETO enables its detection by fluorescence emission (El Kousy, 1999) and ultraviolet absorption (Helmy and El-Bedaiwy, 2014). In this study, we used an FL detection system to attempt to selectively and sensitively detect ETO and its derivatives. For quantification, enantiomerically pure S-NAP, which is an NSAID with a carboxylic acid moiety and that possesses a native fluorescent property, was used as the IS.

First, derivatization was performed using 200 μL of 20 mM R-MCF in the absence of pyridine. As a result, the intact ETO peak was decreased, while no new peak was detected in the chromatogram of HPLC–FL analysis (Supplementary Fig. S1). On the other hand, UHPLC–Q-TOF–MS analysis of the derivatization products displayed one new peak, which had not been visible by the FL detection, in addition to the native ETO peak, which decreased. The accurate masses of the molecular ion and its fragment ions revealed that the new peak corresponded to a carbamate derivative of ETO formed between the indole amine of ETO and R-MCF (Table 1 and Supplementary Fig. S2). The ETO diastereomeric derivatives could not be resolved using a typical reversed phase C₁₈ column in the UHPLC system. It is likely that the menthyl group attached directly on the indole amine induced fluorescence quenching in the ETO moiety. Thus, we concluded that the R-MCF derivatization without pyridine was inapplicable for producing ETO diastereomers suitable for enantioseparation using HPLC–FL.

In contrast, pyridine added to the derivatization reaction at a volume of 30 μL yielded very different results. The native ETO peak was decreased, and two incompletely resolved peaks were found to elute later in the HPLC–FL chromatogram under a simple gradient elution with an aqueous MeOH mixture (data not shown). Based on the UHPLC–Q-TOF–MS analysis of the reaction products, it was confirmed that the two peaks were ETO ester diastereomers that had been produced by the reaction of R-MCF with the carboxylic acid group of ETO (Table 1 and Supplementary Fig. S2). Fluorescence of the derivatives produced in the presence of pyridine yielded levels similar to that of the native ETO under the same detection conditions. Accordingly, we used the same excitation and emission wavelengths for the derivatives as for native ETO, with which the peak of S-NAP diastereomer also yielded a signal similar to that of the native S-NAP. The two derivatization reactions, differing according to the presence of pyridine, are shown in Fig. 1.

Close

Figure 1

Schemes for the derivatization reactions with R-MCF. (a) ETO derivatization in the absence of pyridine, (b) ETO derivatization in the presence of pyridine, and (c) IS (S-NAP) derivatization in the presence of pyridine. The asterisk indicates a chiral center.

Export to PPT

3.2 Chromatographic resolution of the ETO diastereomers

Since diastereomers have very similar but not identical chemical and physical properties, they can be resolved on conventional achiral columns. In our initial study, the two diastereomeric derivative peaks were not completely resolved and an interfering endogenous peak slightly overlapped with the peak of the second diastereomer that eluted under the aqueous MeOH solution gradient. Accordingly, we attempted to optimize the chromatographic conditions to achieve improved resolution of these overlapping peaks.

To this end, we investigated the effects of the properties of aqueous and organic solvents in the mobile phase, and of column temperature, while modifying gradient elution conditions as appropriate. First, we found that changing the mobile phase component from water to 10 mM acetic acid as eluent A yielded a baseline separation of the two ETO diastereomers. However, the problem of the second diastereomer peak overlapping with the following impurity peak still remained. Thus, we changed the organic solvent of the mobile phase from MeOH to ACN, which resulted in a distinct separation between the second diastereomer and the interfering peaks, although the two ETO diastereomers could not be separated at baseline. We then combined MeOH and ACN at various ratios, and found that a mixture of MeOH and ACN at 10:1 (v/v) yielded baseline separation of the two diastereomers and impurity peaks (data not shown). All of these modifications were performed at a column temperature 25 °C. Finally, the effect of column temperature was examined, varying from 25 °C to 45 °C. It was found that the shortest run time without compromising resolution was achieved at 45 °C, which was accordingly selected as the column temperature for the final chromatographic conditions as described in Section 2.2.1. The total chromatographic run time was 23 min.

3.3 Differentiation between R- and S-ETO diastereomers by in vivo assay and computational study

Direct identification of the R- and S-ETO diastereomers was not achievable because enantiomerically pure ETO standards could not be obtained. Nonetheless, resolution of the R- and S-ETO is still meaningful as previously reported (Higashi et al., 2012; Singh and Bhushan, 2015). Differentiation between the two diastereomers was inferred from in vivo assay results. Serum samples taken from ICR mice 30 min and 60 min after administration of (±)-ETO were prepared using the initial derivatization conditions involving pyridine, and were promptly analyzed by HPLC–FL. Based on previous stereoselective pharmacokinetic studies of ETO that showed a predominant chiral disposition of the R-form in vivo due to the rapid metabolism of the S-form (Boni et al., 2000; Brocks and Jamali, 1990), we concluded that the first eluting diastereomer peak (peak 2), which was always much larger than the second peak (peak 3), was the R-ETO diastereomer, followed by the second peak, corresponding to the S-ETO diastereomer (Supplementary Fig. S3).

To complement our experimental investigations, we also performed a computational study using DFT to compute dipole moment, which is a useful parameter to understand the order of elution in LC. The coordinates of the optimized structures of the two diastereomers, R-ETO:R-MCF and S-ETO:R-MCF, are given in Supplementary Table S1. The electric dipole moments of the two fully optimized diastereomers are summarized in Table 2. As per the experimental conditions, the eluting solvents are relatively polar and the stationary phase is non-polar in the reversed phase HPLC. Therefore, as a general rule, a solute that is more polar is expected to have a shorter elution time. The computed dipole moments indicate that the R-ETO derivative is more polar than the S-ETO derivative, and that the R-form is likely to elute ahead of the S-form. These results are consistent with those by in vivo assay. Therefore, it was concluded that the peaks 2 and 3 corresponded to the R- and S-ETO diastereomer, respectively.

3.4 Optimization of R-MCF derivatization conditions and the assessment of the derivatization reaction rates

Biological fluids usually require sample cleanup steps due to matrix complexity. In this study, protein precipitation method, which is one of the most popularly used methods, was used because it is simple and rapid. The two commonly used deproteination solvents, MeOH and ACN (Kole et al., 2011), both of which are easy to remove and have good dissolution capability for ETO, were tested, with the vortexing time fixed at three min to ensure complete mixing of serum and solvent. It was found that ACN yielded significantly higher total peak areas of the ETO derivatives than MeOH (p < 0.001) (Supplementary Fig. S4). As a result, ACN was used as the protein precipitation solvent in the subsequent experiments.

It was necessary to optimize the reaction conditions for R-MCF and ETO because previously published conditions employing R-MCF had only been established for reactions with amines, not carboxylic acids. A literature search and our own experiences with derivatization reactions involving chloroformate (Matzner et al., 1964; Lee et al., 2014) suggested that R-MCF derivatization could be affected by a number of variables, including the ratio of R-MCF to ETO and pyridine volume. Recently, alternative energy inputs such as microwave and ultrasound-assisted mixing have been proven to improve conventional organic synthesis (Baig and Varma, 2012). With regard to this, our preliminary experiments indicated that derivatization efficiency was significantly influenced by the external factor of whether and how contact between R-MCF and ETO was facilitated. Specifically, ultrasonic radiation of the reaction mixture was more efficient than incubation without agitation or vortexing for the same given period (data not shown). Thus, ultrasound-assisted derivatization was selected and the ultrasonic radiation time was included as a variable to be optimized while the ultrasonic power was set to maximum. In order to efficiently determine the optimal reaction conditions and explore the interactions between these variables, we applied RSM using a central composite design.

The following three variables were varied from the level of −1 to +1, and a sufficient volume of ACN was added to the reaction mixture to bring the final volume to 200 μL: (1) variable A, volume of 200 mM R-MCF (50–140 μL), (2) variable B, ultrasonic radiation time (5–18 min), and (3) variable C, pyridine volume (7–24 μL). The sum of the peak areas of R- and S- ETO derivatives was used as the response. The experimental design included 20 independent experiments including 6 center points, 8 factorial points, and 6 axial points, and the experiments were performed in random order. As a result, the response was fitted to a polynomial quadratic equation using the coded factors as follows: $$\text{Peak area}=+1080.31+258.07\times A+188.75\times B+113.94\times C-116.37\times \mathit{AB}+72.98\times \mathit{AC}-42.3\times \mathit{BC}-106.79\times A^2-155.93\times B^2-159.01\times C^2$$

Based on the results of analysis of variance (ANOVA) (Supplementary Table S2), the constructed model was highly significant (p = 0.0003) with a satisfactory adjusted coefficient ( $R_{\text{adj}}^2=0.8639$ ). The regression model showed that all first-order and second-order terms for the three variables had significant coefficients. A negative interactive effect was observed between the R-MCF volume and ultrasonic radiation time. The results indicated that larger volumes of R-MCF and pyridine, and longer ultrasonic radiation time, generally contributed to higher peak areas of the diastereomers (Supplementary Fig. S5).

Based on the constructed model, the optimized conditions were determined to be 125 μL, 13 min, and 17 μL, for the R-MCF volume, ultrasonic radiation time, and pyridine volume, respectively. The derivatization reaction time of 13 min is much shorter than those by Jin et al. (2008) and Singh and et al. (1986) and it is comparable to or slightly longer than the reactions using (S)-anabasine (Higashi et al., 2012), (R)-(+)-α-methylbenzylamine (Singh and Bhushan, 2015). The derivatization was reproduced under these optimized conditions, and the resulting peak areas, 1180.6 (±90.7, n = 3), were found to be reasonably close to the predicted values (1244.8; 95% PI, 899.9–1589.6).

Under the optimized conditions, the reaction rates for R- and S-ETO were examined in the time course of derivatization reaction facilitated by ultrasonic radiation. As displayed in Fig. 3a, the derivatization reaction yields increased sharply for the beginning five min and continued to increase until 10 min, after which no significant changes were observed. EFs for R-ETO also reached the maximum after 10 min of reaction and remained unchanged thereafter (Fig. 3b). These results indicate that the reaction rate of S-ETO is faster than R-ETO in the beginning and that after 10 min of ultrasonic radiation, the reaction rates of the two enantiomers reached the steady state. At the optimal derivatization conditions of which the derivatization reaction time was 13 min, it can be inferred that the derivatization reaction rates were consistent. Consequently, the subsequent method development was conducted under these optimized conditions.

3.5 Salting-out assisted liquid–liquid extraction of the derivatives

After derivatization, 170 μL of aqueous L-proline solution was added to 200 μL of the reaction mixture to quench the reaction. This step introduced an unintended dilution of the sample and therefore decreased sensitivity. We adopted a salting-out assisted liquid–liquid extraction (SA-LLE) strategy to improve sensitivity (Farajzadeh et al., 2013; Wang et al., 2013). SA-LLE is performed by adding an electrolyte to an aqueous mixture to result in the phase separation between water-miscible organic solvent and water. SA-LLE has been widely applied to achieve increased analyte concentrations as well as sample cleanup (Valente et al., 2013; Wen et al., 2013). Specifically, we added NaCl to the quenched reaction mixture as a salting-out agent. Of various amounts tested from 0 to 20 mg, 20 mg of NaCl was found to be sufficient to readily induce phase separation of consistent volumes of the organic phase (data not shown). The derivatized products were found to be enriched in the resulting upper organic phase of more than 140 μL. While the entire SA-LLE procedure could be performed within less than 2 min, it helped achieve a relatively high sensitivity for ETO analysis in biological fluids.

3.6 Method validation

3.6.1

3.6.1 Specificity, sensitivity, and linearity

Specificity of the developed method was confirmed in the chromatograms of samples prepared from blank and spiked sera. No endogenous interfering peaks overlapped with the peaks of the ETO and IS derivatives (Fig. 2). The LOQ values measured based on the signal-to-noise ratio were 0.3 μg mL⁻¹ for both R- and S-ETO. The linearity of the calibration curves that were established using spiked mouse sera was found to be in the range of 0.5–50.0 μg mL⁻¹ for both enantiomers with correlation coefficients (r²) above 0.998. The parameters of the calibration curves are summarized in Table 3.

Close

Figure 2

HPLC–FL chromatograms of samples prepared from blank serum (a), serum spiked with IS (b), and serum spiked with (±)-ETO and IS (c). The spiked concentrations for (±)-ETO and IS were 50.0 and 1.0 μg mL⁻¹, respectively. Peak identification: peaks 2, 3, and 4 correspond to compound 2 (ester derivative of R-ETO), 3 (ester derivative of S-ETO), and 4 (ester derivative of IS) in Fig. 1.

Export to PPT

Close

Figure 3

Total reaction yields of R- and S-ETO derivatives (a) and the enantiomeric fractions of the R-ETO derivative (b) in the time course of derivatization facilitated by ultrasound radiation. Error bars indicate the SEM (n = 3).

Export to PPT

Table 3 Linearity and LOQ of the developed method.

Analyte	Calibration curve	r2	Linear range (μg mL−1)	LOQ (μg mL−1)
R-ETO	y = 0.1205x − 0.02000	0.9987	0.5–50.0	0.3
S-ETO	y = 0.1275x − 0.02207	0.9991	0.5–50.0	0.3

3.7 Comparison of the developed method to other methods and application of the developed method

ETO has been analyzed by HPLC–UV in chiral quantification methods (Boni et al., 2000; Jin et al., 2008; Singh and Bhushan, 2015). In this study, we compared the two detection methods, UV and FL with respect to sensitivity and selectivity. In the chiral derivatization reaction, pyridine that was added as a catalyst was unavoidably retained in the sample after SA-LLE. Because pyridine has a strong UV absorbing property, its peak overwhelmed the HPLC–UV chromatograms at 226 nm that was the maximal absorption wavelength for the ETO diastereomers upon wavelength scanning. The LOQ values of the HPLC–UV method were 5.0 μg mL⁻¹ for both R- and S-ETO, which is 17-fold higher than those of our developed method based on the FL detection. These results support that HPLC–FL analysis provides selective and sensitive determination of ETO enantiomers.

As summarized in Supplementary Table S4, the LOQ value of the current method (0.3 μg mL⁻¹) is comparable to or lower than those of other quantification methods for biological fluids, and this sensitivity was achieved using only 50 μL of sample volume. Specifically, the method reported by Hewala and et al. (2014) consumed 500 μL of plasma to obtain the LOQ values of 0.19–0.20 μg mL⁻¹, while slightly higher LOQ values than that of our method were achieved using a plasma volume of 150–200 μL (Jin et al., 2008). Although our method involves the derivatization procedure that may elongate the analysis time compared to direct chiral separation methods, its derivatization time is reasonably acceptable, and it is much shorter than that of the bioanalytical study reporting 2 h of chiral derivatization (Jin et al., 2008). In addition, our study applied the protein precipitation in combination with SA-LLE for sample cleanup and enrichment, which is simpler and more rapid than solid phase extraction (Hewala et al., 2014) or LLE (Becker-Scharfenkamp and Blaschke, 1993; Jin et al., 2008).

The developed method was applied to real serum samples taken from ICR mice administered (±)-ETO at a dose of 20 mg kg⁻¹ by intraperitoneal injection (n = 3). The serum concentrations 30 and 60 min after administration were determined to be 26.7 and 13.9, 27.8 and 19.4, and 30.1 and 15.7 μg mL⁻¹, respectively, for R- ETO; and 2.4 and 1.2, 2.2 and 3.2, and 3.3 and 1.6 μg mL⁻¹, respectively, for S-ETO.