Enantioseparation of DL-tryptophan by spiral tube assembly counter-current chromatography and evaluation of mass transfer rate for enantiomers

Abstract

Spiral tube assembly counter-current chromatography was successfully applied in enantioseparation of DL-tryptophan using bovine serum albumin as chiral selector. An improved biphasic aqueous-aqueous solvent system 12.0% (w/w) polyethyleneglycol 8000–9.0% (w/w) dibasic potassium phosphate-0.1% ammonia-78.9% water was used as the solvent system for counter-current chromatography, in which bovine serum albumin was predominantly distributed in the lower phase of the two-phase aqueous system. The aqueous-aqueous solvent system gave a very high enantioselectivity for D- and L-tryptophan at α=2.605 along with distribution ratio D_D=1.200 and D_L=0.461. High peak resolution was obtained for enantioseparation of 2.0 mg of DL-tryptophan by spiral tube assembly counter-current chromatography under room temperature. It was found that 0.1% ammonia added in the aqueous-aqueous solvent system greatly improved the enantioseparations. An unusual extremely broad peak for L-tryptophan was observed during enantioseparations. In order to give an explanation, mass transfer rates of D- and L-enantiomers through the interface between the two phases were measured. It was found that L-tryptophan showed lower mass transfer rate than D-tryptophan. Further discussions were proposed for possible reasons for mass transfer rate difference between the enantiomers.

1. Introduction

Early in 1973 it was demonstrated for the first time that DL-tryptophan could be enantioseparated by chromatography on a bovine serum albumin-agarose column following the findings that bovine serum albumin shows higher affinity for L-tryptophan of the DL-pair [1, 2]. S. Allenmark, et al. improved the column performance by immobilizing bovine serum albumin onto the silica support [3]. Since the 1980s a number of papers have been published on direct liquid chromatographic separation of various kinds of enantiomers using immobilized bovine serum albumin stationary phases [4–6]. Conventional liquid chromatography for enantioseparation was generally used for analytical purpose in which high cost is generally required because of the complex process for manufacture of the chiral column.

Partition in an aqueous biphasic solvent system for chiral separation using bovine serum albumin as a chiral selector in one phase could be accomplished by counter-current extraction and counter-current chromatography. The aqueous biphasic solvent systems can be prepared by either mixing two polymers, or one polymer and an inorganic salt in water [7] in which albumin could be distributed almost completely in one of the two phases, while the racemic small molecule is freely partitioned in both phases. In the past, bovine serum albumin has been used as a chiral selector for enantioseparation of DL-tryptophan [8], DL-kynurenine [9] and ofloxacin [10] by counter-current extractions (distributions), and enantioseparation of DL-kynurenine by counter-current chromatography [11]. However, peak resolution was very poor in each case because of low separation efficiency of counter-current distribution technique and low enantioseparation factor for DL-tryptophan with the reported solvent systems. In order to achieve better peak resolution, higher enantioseparation factors and improved counter-current chromatography techniques are necessary.

Counter-current chromatography (CCC) belongs to liquid-liquid partition chromatography, which eliminates the use of a solid support which may cause various complications in conventional liquid chromatography. When it comes to chiral separations, a chiral selector could be directly added in either of the two phases without an expensive immobilizing process. The cost for modern CCC apparatus is usually lower than that for conventional liquid chromatography. And because of its easy scale-up properties, there are an increasing number of papers about chiral separations by counter-current chromatography [12–14].

For conventional high-speed CCC, the retention of the stationary phase is solely provided by the Archimedean screw effect by rotating the multilayer coiled column in the centrifugal force field (planetary motion). The system using this conventional multilayer coil separation column, however, fails to retain enough of the stationary phase for polar solvent systems such as the aqueous-aqueous polymer phase systems. To address this problem, the geometry of the coiled channel was modified to a spiral configuration in our lab so that the system could also utilize the radially acting centrifugal force to improve the retention of the stationary phase. Two different types of spiral columns were fabricated: the spiral disk assembly and the spiral tube assembly [15, 16]. Here, we report a complete enantioseparation of DL-tryptophan using the spiral tube assembly. And an improved aqueous biphasic solvent system was found which provided a higher enantioselectivity for DL-tryptophan (Fig. 1). Furthermore, evaluations of mass transfer rate for each tryptophan enantiomer were conducted by a rotary device to provide an explanation for an unusual extremely broad peak of L-enantiomer observed during the enantioseparation.

Fig. 1.

Chemical structure of DL-tryptophan

2. Experimental Section

2.1. Apparatus

The instrument for counter-current chromatography used was a type-J coil planet centrifuge (PC Inc., Potomac, MD, USA) with a 10 cm revolution radius. It holds a spiral tube assembly (consisting of multiple spiral layers of coil embedded in a spiral tube support) and a counterweight. The spiral tube support was custom-made by CC Biotech, Maryland, USA. It has 4 spiral grooves, each 2.8 mm wide and ca 5 cm deep with 12 transfer radial grooves. The corner of each spiral was rounded to prevent kinking of the tubing. PTFE (polytetrafluoroethylene) tubing of 1.6 mm ID (SW 14) (Zeus Industrial Products, Orangeburg, SC, USA) was flat-twisted and accommodated tightly into the spiral grooves by squashing with a tool which fitted to the radial grooves. Total length of the tubing is estimated as 46 m, the number of layer of the spiral is 12, and the total capacity is 85 ml. The column temperature was maintained at room temperature (25°C) and the revolution rate of the separation column was set at 800 rpm. The solvents were pumped into the column with a model LC-10ADvp constant-flow pump (Shimadzu Corporation, Kyoto, Japan). Continuous monitoring of the effluent was achieved with a model LKB Uvicord SII detector (LKB Instruments, Bromma, Sweden) at 280 nm, and a strip-chart recorder (Pharmacia, Sweden) was used to record the chromatogram (Attenuation: 0.5, chart speed: 1cm/20 min). A model 7000 Ultropac fraction collector (LKB Instruments,) was used to collect the CCC fractions. The Shimadzu HPLC system used in the research was comprised of a Shimadzu SPD-6AV UV detector, a Shimadzu LC-6A constant-flow pump, and a CR501 chromatopac recorder (Shimadzu Corporation, Kyoto, Japan). This analytical HPLC system was used to analyze the fractions from counter-current chromatography. A model Genesys 10 UV ultraviolet spectrophotometer was used to determine partition coefficient of the chiral selector and distribution ratio of tryptophan enantiomer (Thermo Spectronic, Rochester, NY, USA). The pH value was determined with an Accumet AP110 portable pH meter (Fisher Scientific Co., NJ, USA).

2.2. Reagents

DL-tryptophan (≥99%), D-tryptophan (≥99%), L-tryptophan (≥99%), L-isoleucine, bovine serum albumin (A-7906, ≥98%), dibasic potassium phosphate, dibasic sodium phosphate, polyethyleneglycol (PEG) 8000 and PEG 1000 were purchased from Sigma-Aldrich, USA. Dextran 70 (DX, MW 70,000 DA) was purchased from Tokyo Chemical Industry CO., LTD, Japan. Ammonium hydroxide (assay 30%) was purchased from Fisher Scientific, New Jersey, USA. Copper (II) sulfate, anhydrous, was a product of Janssen Chimica (Janssen Pharmaceutica Nv, GEEL, B 2440, Belgium).

Methanol used for HPLC analysis was of chromatographic grade (Fisher Scientific). Water used for CCC and HPLC study was obtained from a model of Picosystem equipment, Hydro Service and Supplies, Inc., Baltimore, MD, USA.

2.3. Enantioselective Liquid-Liquid Extraction Experiments

Selection of a suitable two-phase solvent system is the most important step for successful enantioseparation by CCC. Generally, for the conventional commercial high-speed CCC apparatus, the settling time of a selected two-phase solvent system needs to be less than 30s in order to obtain a satisfactory level of stationary phase retention; and the distribution ratio of enantiomers should fall within the range of 0.2–5.0. Furthermore, the chiral selector should be distributed mostly in one phase of the two-phase solvent system and at the same time the chiral selector should retain the capability of its chiral recognition for an enantiomer in the applied solvent system. Combination of the above requirements contributes to the difficulty of successful chiral separation using CCC. DL-tryptophan is a highly polar small organic molecule and it is difficult to be dissolved in organic solvent, but it shows high solubility in the aqueous phase. Therefore, all the two-phase solvent systems investigated were mainly aqueous-aqueous solvent systems but with one exception, a solvent system consisting of ethanol-saturated ammonium sulfate-water. However, long settling time, i.e., ≥5min, is always required for aqueous-aqueous solvent system because of its low interfacial tension of the two phases and high viscosity caused by the high concentration of polymers such as PEG and dextran. Hence, it is difficult to conduct the enantioseparation experiment by the conventional multilayer coil separation column in the type-J high-speed CCC apparatus. Fortunately, spiral tube assembly CCC could retain a satisfactory level of the stationary phase for the two-phase solvent systems even with a very long settling time, i.e., even longer than 30 min. Thus, settling time could be overlooked during the selection of biphasic solvent systems. The values of distribution ratio and enantioselectivity for enantiomers were the only factors that need to be investigated.

Series of enantioselective liquid-liquid extractions were conducted to select the two-phase solvent systems, in which the partition coefficient of bovine serum albumin, distribution ratios and enantioseparation factor of D- and L-tryptophan were determined. The two-phase solvent system containing bovine serum albumin was prepared in advance and allowed to equilibrate over two hours, in which 6 g of albumin was used per 100 g of solvent system. The K value of bovine serum albumin was measured by pipetting 100 μL of each phase into a separate 10 mL test tube and diluting each phase with 2 mL of water. UV absorbance was determined by the ultraviolet spectrophotometer for each phase and partition coefficient of bovine serum albumin was calculated with the following conventional formula:

$$K_{\mathit{BSA}}=\frac{A_{\mathit{UP}}}{A_{\mathit{LP}}}$$

where A_UP and A_LP were the absorbance values for upper phase and lower phase respectively, and subscript BSA indicates bovine serum albumin. The chiral selector bovine serum albumin was expected to be mainly partitioned into one of the two phases depending on the different solvent systems used, but it could be slightly dissolved in the other phase. Since bovine serum albumin has high UV absorbance under 280 nm, the UV absorbance in either phase could be easily measured by spectrophotometer. On the other hand, determination of distribution ratio and enantioselectivity of D- and L-tryptophan could be disturbed by the high UV absorbance of bovine serum albumin. Therefore, a modified subtraction method was necessary to determine the distribution ratio of D- and L-tryptophan. Take the solvent system 12% (w/w) PEG 8000–9% (w/w) dibasic potassium phosphate-79% water in which 6 g of albumin was added per 100 g of system as an example, the overwhelming majority of bovine serum albumin would distribute in the lower phase. In this case the determination of the distribution ratio for D- and L-tryptophan was conducted as follows: A_UP was determined first as follows: 2 mLof upper phase of the solvent system was placed in a 10 mL stoppered test tube into which 1–2 mg of tryptophan was dissolved. 100 μL of this solution was then pipetted into 2 mL of water and its UV absorbance A₁ was determined. Add two milliliters of lower phase into the solution followed by vigorous shaking for several minutes. After centrifugation 100 μL of upper phase was pipetted into a 2 mL of water and determine its UV absorbance A₂. The distribution ratio D and enantioselectivity α for D- and L-tryptophan was calculated with the following expressions:

$$D_{\mathit{\text{Tryptophan}}}=\frac{A_2-A_{\mathit{UP}}}{(A_1-A_{\mathit{UP}})-(A_2-A_{\mathit{UP}})}=\frac{A_2-A_{\mathit{UP}}}{A_1-A_2}$$

$${\alpha }_{\mathit{\text{Tyrptophan}}}=\frac{D_{D-\mathit{\text{Tryptophan}}}}{D_{L-\mathit{\text{Tryptophan}}}}$$

2.4. Spiral CCC Procedure

The stationary phase retention in the conventional multilayer coil separation column for a typical type-J CCC conducted with aqueous-aqueous solvent system is usually very low due to the high viscosity caused by the polymers and it is difficult to achieve a high peak resolution. In order to overcome this problem, cross-axis coil planet centrifuge (type-X CCC system) could be employed to provide higher retention of stationary phase. But the design of a cross-axis coil planet centrifuge apparatus is very complex and its fabrication required much higher cost than the type-J CCC apparatus. Spiral tube assembly CCC belongs to an improved type-J CCC system with the modified separation columns. This separation column provides a satisfactory level of stationary phase retention for all the existing two-phase solvent systems, even with very long settling time (i.e., ≥30 min). Retention of stationary phase is one of the important parameters because the amount of chiral selector retained in the column was directly related with retention percentage of the stationary phase. This system also allows a flexible choice of the mobile phase so that either phase could be efficiently used as a stationary phase. Usually the phase containing the chiral selector is preferred to be used as the stationary phase.

The solvent system consisting of 12.0% (w/w) PEG 8000–9.0% (w/w) dibasic potassium phosphate-0.1% ammonia-78.9% water was used. It was prepared with the following procedures: 72 g of PEG 8000, 54 g of dibasic potassium phosphate, and 2 mL of 30% ammonium hydroxide were dissolved in 474 g of water. The solvent mixture was thoroughly equilibrated in a separatory funnel, and the two phases were separated shortly before use. The enantioseparation was initiated by filling the column with the lower phase free of chiral selector. Then, 20 mL of the lower phase containing 3.0 g of bovine serum albumin was injected into the column, discharging about 20 mL of excess stationary phase without chiral selector from the column outlet. The column therefore contained 65 mL of lower phase with no chiral selector at its outlet. During the separation, a portion of this blank stationary phase was retained in the column to absorb any chiral selector that might be carried over by the mobile phase, thus preventing UV absorbance caused by bovine serum albumin and its contamination into the CCC fractions. The upper mobile phase was pumped into the column at 0.5 mL min⁻¹ while the column was rotated at 800 rpm in the tail-to-head elution mode.

The sample solutions were prepared as follows: 2.0 mg of DL-tryptophan was dissolved in 1.0 mL of mixture of upper phase and lower phase containing albumin (1:1, v/v). Sample solution was injected after the hydrodynamic equilibrium was reached, as indicated by a clear mobile phase eluting at the outlet. The effluent from the separation column needs to be diluted with 0.1 mL min⁻¹ of water before entering the detector cell in order to obtain a smooth recording of the chromatogram. The on-line dilution was realized by a Microkros™ cross flow syringe filters with hollow fibers membrane (Spectrum laboratories, Inc., CA, USA). The absorbance of the effluent was continuously monitored at 280 nm and 2 mL fractions were collected.

2.5. Chiral Analytical Method by HPLC

The chiral analysis of DL-tryptophan and their CCC fractions was performed by HPLC with a chiral ligand exchange mobile phase additive using a Phenomenex Luna 5μ C18 (250×4.60 mm i.d., 5 μm) reverse phase column. The mobile phase was water-methanol (90:10, v/v) solution containing 10 mmol L⁻¹ L-isoleucine and 5 mmol L⁻¹ cupric sulfate eluted at a flow rate of 1.0 mL min⁻¹. The mobile phase was filtered through a 0.45 μm filter and sonicated for 20 min prior to use. The wave length of the UV detector was set at 280 nm. The column temperature was set at 25°C. Before HPLC injection, the CCC fractions needed to be treated with Ultrace YM-10 Mirocon centrifugal filter devices to remove polymers (Regenerated cellulose 10,000 MWCO, Millipore). The fractions were centrifuged at ca. 13000 g for 15 min and 20 μL of each was injected into the HPLC column.

2.6. Determination of Mass Transfer Rate for Enantiomers

An unusual broad peak for L-tryptophan was observed during enantioseparation of racemic tryptophan by spiral tube assembly. Great difference in peak height was found between D- and L-enantiomer. The calculated number of theoretical plates for D-enantiomer peak was more than 12 times higher than that of L-enantiomer peak. Usually, peak resolution is mainly determined by separation factor, retention time is mainly determined by the distribution ratio, whereas peak width is usually dependent on the mass transfer rate of the analyte through the interface between the two phases. In our previous studies [17], the cause of excessive band broadening of protein samples separated in polymer phase system by CCC was investigated using a simple rotary device, in which the mass transfer rates of several samples were determined using an aqueous-aqueous polymer phase system. The results indicated that the mass transfer rates of the analyte are closely correlated with their molecular masses: the higher the molecular mass, the lower the mass transfer rate. For a small molecule like DL-tryptophan, equilibration between two phases could quickly be reached within 5 min when the rotation speed of the rotary device was set at 30 rpm, while equilibration time was highly dependent on the rotation speed of the rotary device for a given analyte.

In the present study we used the same rotary device to determine the mass transfer rate for D- and L-tryptophan molecules partitioned in the aqueous biphasic solvent system. However, a lower rotation speed of the rotary device (10 rpm) was applied to prolong the equilibration time to facilitate the accurate measurement of the concentration of D- and L-tryptophan in both phases. This slow rotation results in mixing within each phase while preserving the shape of the interface between the two phases. Consequently, the enantiomer present in the lower phase is gradually transferred into the upper phase through the interface until the distribution equilibrium is reached between the two phases. For both of D- and L-enantiomers, the following mathematical treatment of mass transfer rates could be applied:

$$-\ln [(C_t-C_{\infty })/(C_o-C_{\infty })]=R(A/V)t$$ (1)

where C_t is the concentration of enantiomer in the lower phase at time t, C_o is initial concentration of the enantiomer in the lower phase, C_∞ is equilibrium concentration of the enantiomer in the lower phase, R is the mass transfer rate, A is the interfacial area and V, the volume of the lower phase. If the left term, -ln[(C_t − C_∞)/( C_o − C_∞)], is plotted against time t in an x-y coordinate system, a straight line should be obtained and mass transfer rate R could be calculated from the slope of the straight line.

The apparatus consists of a motor which drives a rotary shaft symmetrically holding a set of five glass test tubes (100×15 mm O.D.) around its axis. The rotary shaft is inclined at 18° against the horizontal line and the rotation is optimized at 10 rpm with a speed regulator. The aqueous-aqueous solvent system used for determination was 12.0% (w/w) PEG8000–9.0% (w/w) potassium phosphate dibasic-0.1% ammonia-78.9% water (6 g of bovine serum albumin to 100 g of this system). In each measurement, the sample was dissolved in 2 mL of the lower phase of the above solvent system which was then delivered into one of the glass test tubes. The concentration of racemic tryptophan was limited so that the molar ratio of bovine serum albumin to racemic tryptophan was about 1:1. Then 2 mL of the upper phase was gently layered over the lower phase followed by rotation of the tube. The other four test tubes were filled only with solvent system (2 mL of upper phase and 2 mL of lower phase) to keep the rotation balanced. At intervals of 1, 2, 5, 10, 15, 20, 30 and 45 min, the rotation was interrupted and a small aliquot (80 μL) of the lower phase was pipetted into a tube in preparation for HPLC analysis, in which the same amount of upper phases was removed to maintain the 1:1volume ratio of two phases in the test tube.

3. Results and Discussion

3.1. Selection of Two-Phase Solvent Systems

The following two-phase solvent systems were investigated: 10% (w/w) PEG 8000–5% (w/w) dibasic sodium phosphate; 8% (w/w) dextran70–7% (w/w) PEG 8000–0.1 mol L⁻¹ sodium chloride and 50 mmol L⁻¹ sodium bicarbonate buffer at pH 9.5; 12%(w/w) polyethyleneglycol 8000–9% dibasic potassium phosphate; 16% (w/w) PEG 1000–12.5% (w/w) dibasic potassium phosphate and ethanol-saturated ammonium sulfate-water (1:1:1, v/v). Bovine serum albumin (6.0g) was added per 100 g of the solvent system. The extremely hydrophilic organic-aqueous two-phase solvent system composed of ethanol-saturated ammonium sulfate-water (1:1:1, v/v) [18,19], provided high solubility for DL-tryptophan in both phases as well as high stationary phase retention, but it was found that an enough amount of the chiral selector (6%, w/w) bovine serum albumin could not dissolve in this system due to its incompatibility with the organic solvent, besides the system showed an extremely long settling time after the addition of bovine serum albumin. So this solvent system was not further studied.

In the literature [8], direct chiral resolution of DL-tryptophan with an aqueous polymer two-phase solvent system by an automatic thin layer counter-current distribution apparatus was reported, in which the phase system was composed of 10% dextran 40, 7% PEG 6000, 0.1 mol L⁻¹ sodium chloride, 50 mmol L⁻¹ sodium bicarbonate buffer at pH 9.2, and bovine serum albumin (6.5 g) was added per 100 g of the phase system. But no complete chiral resolution was obtained in the literature [8]. Experiments showed that more than 94% of bovine serum albumin was distributed in the lower phase of this system. Hence, this solvent system was firstly tested by spiral tube assembly CCC in our research. Results showed that a very broad peak (elution time 90–330 min) was observed for racemic DL-tryptophan during separations with upper phase as the mobile phase (tail-to-head elution) without peak resolution. And no separation was achieved with lower phase as the mobile phase (head-to-tail elution). Also, it was very difficult to conduct the experiments due to its extremely high viscosity due to the high concentration of dextran, although over 45% stationary phase volume retention was achieved using the spiral tube assembly CCC system.

The solvent system composed of 10% (w/w) PEG8000–5% (w/w) dibasic sodium phosphate-6%bovine serum albumin-79% water has been used for enantioseparation of DL-kynurenine by a cross-axis coil planet centrifuge [11]. More than 98% of bovine serum albumin distributed in the lower phase of this solvent system. However, dibasic sodium phosphate is easily precipitated from the solvent system due to its low solubility in the system. Furthermore, the composition of this solvent system of 10% (w/w) PEG8000–5% (w/w) dibasic sodium phosphate was close to the critical point and it easily forms a single phase with a slight change of temperature [7]. Then, dibasic sodium phosphate was replaced by dibasic potassium phosphate for the aqueous-aqueous solvent system but with a slightly different weight percentage. The system with 12% (w/w) PEG8000–9% (w/w) dibasic potassium phosphate-79% water was obtained by a trial and error method. Here it is necessary to point out that in the selection of the two-phase solvent system for CCC, it is preferred that the system produces almost equal volumes of the upper and lower phases so that either phase could be used as the mobile phase. The system with the weight percentage 12% (w/w) PEG8000–9% (w/w) dibasic potassium phosphate-79% water leads to almost equal volume ratio between the two phases. Also when 6.0 g of bovine serum albumin was added to 100 g of the system, it was found that over 97% of bovine serum albumin was distributed in the lower phase. However, only very limited enantioseparation was obtained for DL-tryptophan with this solvent system by spiral tube assembly CCC, in which the upper phase was used as the mobile phase. But higher theoretical plate (sharp peak) was observed compared with the previously used solvent system containing 10% dextran 70.

Another solvent system consisting of 16% (w/w) PEG1000–12.5% (w/w) dibasic potassium phosphate-71.5% water was also tested. In this case, over 91% of bovine serum albumin partitioned into the upper phase. But the too large distribution ratio for D- and L-enantiomers (7.655 and 10.300, respectively) made this solvent system useless.

Finally, much attention was paid to the system 12% (w/w) PEG8000–9% (w/w) dibasic potassium phosphate-79% water since it produced high enantioselectivity as well as suitable distribution ratio. Generally, lower separation temperature that suppresses the molecular motion would improve the chiral selectivity while the partition efficiency is decreased [20]. No obvious improved peak resolution was found yet even when the column temperature was decreased to 6–8°C. At last, it was found that much higher enantioselectivity, α=2.605, was observed when to the solvent system was added a small amount of ammonia (2 mL of 30% ammonium hydroxide added to 600 g of the solvent system), in which the pH value of the system increased from 9.20 to 10.16. With this improved solvent system, over 98% of bovine serum albumin was dissolved in the lower phase and high peak resolution was obtained for enantioseparation of DL-tryptophan by spiral tube assembly. The partition coefficient values (K) of bovine serum albumin, distribution ratio D and enantioseparation factor α of D- and L-tryptophan in the studied aqueous-aqueous solvent systems are listed in Table 1.

Table 1.

The K (partition coefficient) values of bovine serum albumin (BSA), D (distribution ratio) and α (enantioseparation factor) of DL-tryptophan in different aqueous-aqueous solvent systems

Solvent system	KBSA	DD	DL	αDL-tryptophan
1. 8% (w/w) Dextran70–7% (w/w) PEG8000–85% 0.1 mol L−1 sodium chloride and 50 mmol L−1 sodium bicarbonate buffer	0.053	1.186	0.838	1.416
2. 10%(w/w) PEG8000–5%(w/w) dibasic sodium phosphate-6% (w/w) bovine serum albumin-79% water	0.013	2.328	1.430	1.628
3. 12% (w/w) PEG8000–9% (w/w) dibasic potassium phosphate-79% water	0.029	1.210	0.714	1.694
4. 16% (w/w) PEG1000–12.5% (w/w) dibasic potassium phosphate-71.5% water	9.605	7.655	10.300	1.346
5. 12.0% (w/w) PEG8000–9.0% (w/w) potassium phosphate dibasic-0.1% ammonia-78.9% water	0.021	1.200	0.461	2.605

Note: all the values were determined after bovine serum albumin (6.0g) was added per 100 g of each solvent system except system 2. Temperature: 25°C.

3.2. Enantioseparation of DL-tryptophan by Spiral Tube Assembly Counter-current Chromatography

The aqueous-aqueous solvent system was optimized at 12.0% (w/w) PEG8000–9.0% (w/w) potassium phosphate dibasic-0.1% ammonia-78.9% water at pH=10.16, where to 600 g of the system was added 2 mL of 30% ammonium hydroxide. It was found that after separation the retention of stationary phase was 61% when the upper phase was used as the stationary phase (head-to-tail elution), while it decreased to 40% when lower phase as the stationary phase (tail-to-head elution). Fig. 2 shows a typical chromatogram for enantioseparation of DL-tryptophan by spiral CCC with the above solvent system. Tail-to-head elution of the upper phase was applied since the chiral selector was mostly dissolved in the lower aqueous phase. The sample was injected after hydrodynamic equilibrium was reached inside the column, t=0 min, as shown in Fig. 2. The signal of UV absorbance might be disturbed if a small amount of lower stationary phase was carried over in the upper mobile phase. Therefore, the sample should be injected after hydrodynamic equilibration. DL-tryptophan was almost enantioseparated with baseline resolution. No obvious peak resolution for DL-tryptophan was obtained if the solvent system 12.0% (w/w) PEG8000–9.0% (w/w) potassium phosphate dibasic-79% water without addition of ammonia. Therefore, it was speculated that pH of the solvent system might play a critical role in this chiral recognition. The pH in the solvent system might alter the conformation of bovine serum albumin and electric charge of DL-tryptophan. Consequently, the optimum pH value would lead to a great improvement for its enantiorecognition. Generally, complete separation of two components could be accomplished on the modern commercial CCC system if the separation factor was over 1.4. As shown in Table 1, high enantioselectivity, greater than 1.4, could be obtained with several solvent systems. But no successful enantioseparations could be achieved by spiral tube assembly CCC with the other solvent systems. The resolution was poor in each case. Since the polymer phase systems are rather viscous, the mass transfer process during the separation might be worse than in the common organic/aqueous solvent systems, hence it requires higher enantioselectivity for successful enantioseparations.

Fig. 2.

Chromatogram of enantioseparation of DL-tryptophan by spiral tube assembly CCC. Solvent system: 12.0% (w/w) PEG 8000–9.0% (w/w) dibasic potassium phosphate −0.1% ammonia-78.9% water, in which 20 mL of the lower phase contained 3.0 g of bovine serum albumin (see text); elution: tail-to-head; stationary phase: lower phase; mobile phase: upper phase; sample solution: 2 mg of DL-tryptophan dissolved in 1 mL of mixture of equal volumes of upper and lower phases containing albumin; flow rate: 0.5 mL min⁻¹; revolution: 800 rpm; column temperature: 25°C; retention of stationary phase after separation: 40%.

The purity of collected peak fractions from CCC was analyzed by analytical HPLC using a chiral ligand exchanger as mobile phase additive. Each peak fraction was also compared with standard reference of D- and L-tryptophan. HPLC results showed that the first peak was D-tryptophan with the over 98% purity and the second peak was L-tryptophan with over 96% purity. The elution sequence obtained in the CCC chromatogram was in agreement with the reported affinity results [1, 2] where L-tryptophan was proved to have much higher affinity to bovine serum albumin than D-tryptophan.

3.3. Evaluation of Mass Transfer Rate for Enantiomers

Fig. 3(a) shows concentration of D- and L-tryptophan in the lower phase plotted against time during this mass transfer process. Equilibration concentration of D- and L-tryptophan was reached in 45 min with the rotation speed at 10 rpm. It was found that concentration decreasing rate of D-enantiomer was much higher than that of L-tryptophan in the first 10 min. Fig. 3(b) was obtained by plotting the left term of equation (1), −ln[(C_t − C_∞)/( C_o − C_∞)], against time t. The results showed that, for D-tryptophan, the variation of -ln[(C_t − C_∞)/( C_o − C_∞)] against time t fitted very well with our suggested mathematical equations, producing a perfect straight line. But as for L-tryptophan, it showed two divided straight lines, a short lower-inclined line in the first 10 min followed by a long straight line parallel to that of D-tryptophan, indicating that the mass transfer process was affected in first 10 min by some additional factor(s) beside the two-phase solvent system itself, probably by the low transfer rate of albumin-L-tryptophan through the interface. This hypothesis is supported by an additional experiment to determine the mass transfer rate of bovine serum albumin through the interface using the rotary device, which showed that the equilibrium was reached just in 10 min. Fig. 3(b) clearly indicated that mass transfer rate of D-enantiomer was much greater than that of L-enantiomer in the first 10 min which might explain the formation of an extremely broad peak of L-tryptophan.

Fig. 3.

Mass transfer rates of D- and L-enantiomer of tryptophan determined by a rotary device. (a) concentration of D-enantiomer and L-enantiomer in the lower phase during evaluation of mass transfer rate; (b) −ln[(C_t − C_∞)/(C_o − C_∞)] vs. time plot.

According to the well established mathematical model, the mass transfer rate of an analyte in a liquid-liquid partition system can be controlled by either diffusion processes or the kinetics of the chemical reactions taking place in the system [21]. The diffusion processes are enhanced by increasing the rotation speed, but rotation speed would have no effect on mass transfer rate if it was controlled by chemical reaction. In other words, mass transfer rate was determined by either rotation speed (diffusion process) or a combination reaction in the system. However, it was found that there is no difference of mass transfer rate of both D- and L-tryptophan after vortex shaking of the aqueous-aqueous system, in which both enantiomers could reach equilibration concentration within few minutes. These results demonstrated that mass transfer rate of D- and L-tryptophan in our specific case was solely controlled by the diffusion process with no influence of the combination reaction between bovine serum albumin and L-tryptophan.

Fig. 4 schematically shows a cross-sectional view through a portion of the separation column where two aqueous phases are arbitrarily separated, where bovine serum albumin (chiral selector) is distributed between the two phases. As shown in Table 1, the chiral selector is mainly in the lower phase, and only about 2% is present in the upper phase. According to our previous work [17], the mass transfer rate of large molecules like albumin was much lower than that of small molecules. Since D-tryptophan has no affinity with the chiral selector, it could be quickly partitioned between both phases with a high mass transfer rate. However, for L-tryptophan, mass transfer process should be more complicated since it has high affinity for bovine serum albumin. When the upper mobile phase was eluted through the column, 2% of bovine serum albumin continuously needs to be re-equilibrated in the column with much lower mass transfer rate than free enantiomers. Thus, a long equilibration time was necessary for mass transfer rate of 2% of bovine serum albumin, which decreased the mass transfer rate of L-tryptophan during the separation to produce its extremely broad peak. In order to verify this speculation, enantioseparation of DL-tryptophan by spiral tube assembly CCC was conducted at a much lower flow rate of the upper mobile phase (0.2 mL min⁻¹), which allowed much longer equilibration time for both 2% of bovine serum albumin and enantiomers. Experimental results showed that compared with D-tryptophan, the peak width of L-tryptophan became much less broader and the ratio of theoretical plates between D-/L-tryptophan peaks decreased to 4 : 1 from 12 : 1 in the original separation at a flow rate of 0.5 ml/min (the chromatogram is not shown here). Therefore, the major difference of mass transfer rate between D- and L-tryptophan for the first 10 min, as showed in Fig. 3(b), indicated that about 10 min was necessary for equilibration time of 2% of bovine serum albumin in the separation column during enantioseparation by spiral tube assemble CCC with the present biphasic aqueous solvent system. The number of theoretical plates or peak width of the two enantiomers tends to become much closer at a lower flow rate of mobile phase.

Fig. 4.

Schematic diagram of chemodynamic equilibrium between DL-tryptophan and the chiral selector bovine serum albumin (BSA) in the spiral CCC separation column. D and L: D- and L-enantiomers of tryptophan; K_D, K_L, K_BSA-L and K_BSA: partition coefficient, k_f: formation constant of combination reaction.

Another speculation for the lower mass transfer rate of L-tryptophan is that there might be several chiral recognition sites each with different activities in each molecule of bovine serum albumin since it has a large molecular weight, which would lead to a different elution time for the L-tryptophan molecule. It is reported that the exact number of distinct binding locations in bovine serum albumin is not clear but there are two principal sites for small heterocyclic and aromatic carboxylic acids [22]. If this is true, similar broad peak for L-tryptophan should be observed when bovine serum albumin was bonded to the immobilized stationary phase. However, no noticeable broad peaks for L-tryptophan were reported in the literature on enantioseparation by liquid chromatography with bovine serum albumin used as a chiral stationary phase [23–25]. Compared with liquid chromatography, the major difference presented in enantioseparation by spiral tube assembly CCC was 98% of bovine serum albumin partitioned in stationary phase while 2% of bovine serum albumin would distributed in the mobile phase. Therefore, 2% of bovine serum albumin combined with L-tryptophan could move with a very slow transfer rate due to its large molecular characteristics, which decreased the mass transfer rate of L-enantiomer. We have tried to look for an aqueous-aqueous solvent system, in which bovine serum albumin is entirely distributed in one phase, to see if this would sharpen the peak of L-tryptophan. Unfortunately, no such biphasic aqueous-aqueous solvent system could be obtained.

4. Conclusion

Using an improved biphasic aqueous solvent system composed of 12.0% (w/w) PEG8000–9.0% (w/w) dibasic potassium phosphate −0.1% ammonia-78.9% water using bovine serum albumin as a chiral selector, the enantioseparation factor for enantiomers of DL-tryptophan was increased from 1.694 to 2.605 by adding ammonium hydroxide to the solvent system. Spiral tube assembly CCC was successfully used for complete resolution of 2.0 mg of DL-tryptophan with the above system. Optical purities for both enantiomers were over 96% as determined by chiral HPLC. However, recovery of the enantiomers from fractions was not investigated in our present research. Using a simple rotary device, mass transfer rates of L-tryptophan was found to be much lower than that of D-tryptophan which is most likely the cause of the extremely broad peak formation of L-enantiomer.