Anulyst, October 1996, Vol. 121 (1397-1400) 1397 Turbidimetric Flow Method for the Enantiomeric Discrimination of L- and D-Aspartic Acid Monika Hosse, Evaristo Ballesteros, Mercedes Gallego and Miguel Valcarcel* Department qf Anulytical Chemistry, Faculty oj’ Sciences, University of Cdrdoha, E-14004 Cbrdoba, Spain The proposed enantiomeric resolution of aspartic acid is based-on the inhibition of the crystallization of L- and D-histidine. A flow-through system permits the turbidimetric multi-detection of the signal produced in the crystallization of histidine from a supersaturated solution. The presence of L- or maspartic acid delays the growth of L- or D-histidine crystals, respectively, the delay being proportional to the concentration of aspartic acid. Calibration graphs are linear down to 40 mg 1 - 1 of L- and D-aspartic acid, with a precision (repeatability, as RSD, n = 11) of 2.5%.The method was applied to the determination of I,-aspartic acid in pharmaceutical preparations (spiked with D-aspartic acid) and the resolution of a racemic sample of D,L-aspartic acid. The results obtained were consistent with the nominal contents. Keywords: Tui-hidimetry; jlow-tlil-ougli system; L- and D-aspartic acid; c h i d resolution ; pharmaceuticul prepaimkm; racemic. sunzple Introduction From Pasteur’s very first optical resolution of a racemate’ to today’s fast chromatographic techniques, there has been a formidable accumulation of stereochemical knowledge. The importance of optically active compounds for the elucidation of reaction mechanisms and the dynamic behaviour of chiral molecules in organic chemistry has grown dramatically.The pharmaceutical industry is becoming increasingly interested in methods for resolving racemates into optical antipodes in order to be able to subject them individually to pharmacological testing2 In recent years, HPLC has bccome the most popular choice for the resolution of enantiomeric compounds.3--” However, this technique has some disadvantages, such as the high price of chiral stationary phases. Alternative techniques for the resolu- tion of racemic mixtures are based on fractional crystallization of conglomerates. Growing crystal surfaces can be thought of as being composed of ‘active sites’ that interact stereospecifically with molecules in solution, in a manner similar to the interactions of enzymes and substrates or antibodies and antigens.6 The molecules that interact with the active sites are known as ‘tailor-made’ inhibitors. Grases and March7 reviewed the potential applications of these crystallization inhibitory processes in analytical chemistry.They have developed various methods for the determination of L-amino acids in which a supersaturated solution of a substrate is obtained by changing the solvent composition; the analyte (L-amino acid) inhibits crystallization of the substrate.8.9 The applicability of enantiose- * To whom correspondence should be addressed. lective biosensor systems depends on the availability of suitable group-specific/enantiospecific and D- and I -enantiospecific enzyme pairs. These enzymes can be combined with suitable potentiometric, amperometric, calorimetric and optical trans- ducers.Schugerl et al. l o recently discussed the potential of biosensors for enantio-sensitive analysis with non-enantio- meric, enantiomeric and D- and L-enantiomeric enzyme pairs. They reported that biosensors for enantiomeric analysis can be used for process monitoring and control of the enantiomeric purity of organic chemicals, whereas HPLC is suitable for quality control of the product but not for process monitoring. Recently, our group developed several turbidimetric methods for the continuous determination of s,-lysine. 1,-arginine and I.- ornithine in pharmaceutical preparations.] 1 . i 2 in this work, two enantioiners (L- and D-) of aspartic acid were discriminated by their inhibitory effect on the crystal growth of L- and I)- histidine, respectively.For this purpose, a continuous method was developed that permits the sequential determination of L- and maspartic acid in the presence of other amino acids with no need for a prior separation. The method was applied to the analysis of pharmaceutical samples, where the determination of L-aspartic acid is of special interest. Since the method is integrated in a flow-through system, it can be implemented on- line for process control purposes. Experimental Chemicals and Apparatus Propan-2-01, ethanol, methanol and acetonitrile, all of HPLC grade, were purchased from Romil Chemicals (Loughborough, UK). Amino acids (1,- and maspartic acid, racemate of I),[,- aspartic acid, L- and D-histidine and the other amino acids used for the study of interferences) were supplied by Sigma (St.Louis, MO, USA). Sodium hydroxide and hydrochloric acid were obtained from Merck (Darmstadt, Germany). Stock standard solutions containing 1 g 1-1 of L- or 11- aspartic acid were prepared in Milli-Q-purified water and stored -stoppered bottles. Solutions containing 3.0 or 2.8 g I-’ of L,- or D-histidine were used as substrates. These solutions remained stable for at least 1 week. Turbidimetric measurements were made on a Unicam 8625 UV/vIS spectrophotometer (Unicam, Cambridge, UK) equipped with a Hellma (Jamaica, NY, USA) flow cell (path length 10 mm, inner volume 18 ~ 1 ) . The ab5orbance was continuously recorded at 550 nm by a Radiometer (Copen- hagen, Denmark) REC-80 Servograph recorder. The flow manifold consisted of two peristaltic pumps [ Gilson (Wor- thington, OH, USA) Minipuls-21 fitted with poly(viny1 chlo- ride) and Solvaflex pumping tubes for aqueous and organic solutions, respectively.A Rheodyne (Cotati, CA, USA) Model 504 1 injection valve, two Rheodyne Model 530 1 switching13% Analyst, October 1996, Vol. 121 30 .r 25- E \ -0 a, n .g 2 0 - 1 5 - .- + 0 $ 10- - 5 - valves and PTFE tubing of 0.5 mm id for coils were also used. - Sample Preparation Five tablets (BOl-K aspartic acid, Laboratory BOI, Barcelona, Spain) or ten pills (Aspartono, Laboratory MEDIX, Madrid, Spain), chosen at random from several samples, were placed in a mortar and ground to a fine mesh. A portion of about 0.5 g (pills) or 5 g (tablets) of the resulting powder (containing approximately 300 mg of L-aspartic acid) was accurately weighed and dissolved in I00 ml of Milli-Q-purified water with electromagnetic stirring for 60 min.The solution was filtered, the residue washed with water and the filtrate was diluted to volume with water in a 250 ml calibrated flask. Aliquots of 100-200 p1 of these final solutions were placed in 10 ml calibrated flasks and diluted to volume (pH 3-10) for analy- sis. Procedure for the Resolution of Enantiomers The continuous-flow procedure for the resolution of L- and D- aspartic acid involves several steps (Fig. 1). For the determination of L-aspartic acid [Fig. 1(a)], an aqueous sample containing 3 4 0 mg 1-1 of L- and D-aspartic acid at pH 3-10 is continuously aspirated at 0.3 ml min-1 and mixed with a substrate stream containing 3.0 g 1-1 of L-histidine at 0.3 ml min-1.Then, the mixed stream is merged with another of propan-2-01 circulated at 1.3 ml min- I . The mixed solution is homogenized in coil C2 and then passed through the injection valve (IV). Simultaneously, the open system is filled with carrier (propan-2-01) by means of the second pump (P2) at 0.7 ml min- I . Then [Fig. 1 ( h ) ] , 100 pl of the loop contents of IV are injected into the carrier and SV2 is switched to close the circuit. At that moment, the absorbance curve, which reflects the time course of crystal growth, is recorded at 550 nm. Finally, SV2 is switched to flush the open-closed system with Milli-Q-purified water. For the determination of D-aspartic acid [Fig.l(a)], SV, is switched and the same aqueous sample is mixed with D- histidine solution (2.8 g 1-I). The procedure is then executed as described above for r--aspartic acid. The signal profile obtained during the crystallization of L- or D-histidine is shown in Fig. 1. The induction period ( t ) was used as the analytical parameter. I I L-Histidine D-Histidine R 11 Fig. 1 Manifold for the continuous-flow discrimination of 1,- and D- aspartic acid. ( a ) Introduction of sample (L- and D-aspartic acid), substrate and propan-2-01 into the system. ( h ) Closed system and signal multi- detection of crystal growth of L- or D-histidine. P, pump; SV, switching valve; IV, injection valve; C, coils (C,, C2 and C3,40, 100 and 40 cm long, respectively); W, waste; D, spectrophotometer; R, signal recording; t, induction period.Results and Discussion Selection of the Substrate and Organic Solvent The continous-flow system depicted in Fig. 1 was used to select the most suitable reagent and solvent. Initially, a sample containing 10 mg 1-1 of L- and D-aspartic acid was introduced into the flow system and merged with a solution containing 3.0 g 1 - l of L- or D-substrate solution. Two amino acids (histidine and lysine) were examined as substrates. I,- and D-aspartic acid exhibited an inhibitory effect on the crystal growth of L- and D- histidine, respectively; however, using lysine to determine aspartic acid was inadvisable as L- and D-aspartic acid showed no inhibitory effects on the crystal growth of L- and D-lysine.For this reason, L- and D-histidine were selected as the substrates for the discrimination of L- and u-aspartic acid, respectively . A previous study' 1 revealed that some organic solvents (methanol, ethanol, propan-2-01 and acetonitrile) induce crys- tallization of the substrate. Such solvents, all miscible with water, were used to prepare supersaturated solutions of the substrate. Propan-2-01 was finally selected as the organic solvent because the resulting induction period ( 5 min) was shorter than that obtained with other solvents (longer than 20 min) and the analysis time was thus the shortest. Optimization of the Working Conditions In aqueous solution, amino acids are present as cations. zwitter- ions or anions, depending on the pH; their crystallization is especially favourable in the zwitterion form.The effect of pH on the crystallization of L- and D-histidine was studied between pH 2 and 12 (adjusted with 0.01 mol 1-1 HNO3 or NaOH). For this purpose, aqueous solutions of L- and D-aspartic acid ( 10 mg 1- 1) or blanks (Milli-Q-purified water) at different pH values were processed in the system. As can be seen in Fig. 2, the induction period in the crystallization of L-histidine (for sample and blank) remained constant over the pH range 3-10, outside which the period was considerably longer. Similar results were obtained for the D-enantiomer (Fig. 2). The induction periods were similar (approximately 6 min) for both blanks. A water blank and a sample pH of 5-6 were selected, which were directly obtained by preparing the samples in water.In order to select the best L- and D-histidine concentrations, several calibration graphs for I,- and D-aspartic acids ( 3 4 0 mg 1-1) were run at a various concentrations of the substrate between 2.5 and 3.5 g 1-1. Fig. 3(a) shows the influence of the substrate concentration on the sensitivity (slope of the calibra- B 4 01 I I 1 1 1 3 5 7 9 11 PH Fig. 2 Effect of pH on the crystallization of L- or D-histidine (3.0 and 2.8 g 1 - I , respectively) for a sample containing 10 mg 1-1 of L- and D-aspartic acid (1 and 2, respectively) and for the blanks (water) of the L- and D- enantiomers of histidine (3 and 4, respectively).Anulyst, October 1996, V d . 121 I399 tion graph) of the method. These experiments allowed the following conclusions to be drawn: ( I ) the sensitivity increases as the concentration of substrate decreases; (2) the sensitivity is slightly higher for L-aspartic acid than for the D-enantionier at all substrate concentrations (probably because the crystalliza- tion of whistidine in propan-2-01 is more favourable than that of L-histidine); and (3) increased sensitivity can be obtained at the expense of longer induction periods [Fig.3(h)]. Concentrations of 3.0 and 2.8 g I-' for L- and D-histidine, respectively. were selected because the sensitivity for L- and D-aspartic acid was similar and the sample throughput acceptable. The flow variables affecting the perforniance of the con- tinuous-flow system were optimized by using an aqueous sample containing 10 mg 1- 1 01' 1,- and ixispartic acid.water as the blank and a substrate solution containing 3.0 or 2.8 g I-' of L- or 1,-histidine, respectively. The influence of the flow rates used in this system was studied by varying one at a time while keeping all others constant. Increasing the sample flow rate increased the induction period for the substrate crystallization through an increased concentration of L- and n-aspartic acid and increased dilution of the substrate in the final mixture. Conversely, increasing the substrate flow rate decreased the induction period. Increasing the flow-rate of propan-2-01 had a similar effect to diluting the sample and substrate in the final mixture. Flow rates of 0.3, 0.3 and 1.3 ml min-I were therefore selected for the sample, substrate and propan-2-01, respectively, as compromises between acceptable sensitivity and sample throughput.The effect of the flow rate of the carrier (propan- 2-01) in the open-closed system was studied over the range 0.3-1.5 ml min-I; the signal remained constant above 0.5 ml min-1. A flow rate of 0.7 ml min- 1 was chosen to transfer the contents of the sample loop into the closed system. The crystallization of L- and D-histidine was significantly affected by the injected volume of valve IV throughout the range studied (50-250 pl); the induction period decreased with increasing volume injected through decreased dilution of the sample and substrate. An injected volume of 100 pl was selected. The influence of the length of the coils for mixing the sample and substrate [C, in Fig.I(a)], and both with propan-2-01 (C,) was studied between 25 and 150 cm (0.5 nim id). Lengths of 40 and 100 cm were selected for C I and C2, respectively, as they proved sufficient for homogenizing the solutions. Finally, the length of the coil inserted in the closed system [C, in Fig. I (b)] was varied between 25 and 100 cni. Increasing length of C, increased the dilution of the sample and substrate in the closed system and hence increased the induction period for histidine 0.5 0.4 c '5 0.3 .- c .- v) C $ 0.2 0.1 0 --+ , 2.4 2.6 2.8 3.0 3.2 3.4 Concentration of L - or D-histidine/mg I-' Fig. 3 ( u ) Sensitivity (slope of the calibration graph) for L.- and n-aspartic acid (solid and dashed lines, respectively) and ( h ) induction period for a sample containing 10 nig I - ' of L- and o-aspartic acid at various concentrations of L- or n-histidine.respectively. crystal growth. A length of 40 cm was chosen as a compromise between acceptable sensitivity and sample throughput. Analytical Performance The performance and reliability of the continuous-turbidimetric system (Fig. 1) were tested by determining the sensitivity (slope of the calibration graph), analyte detectability, linearity range and RSD for the determination of L- and D-aspartic acid. For this purpose, solutions containing various concentrations of I.- and waspartic acid were introduced into the system and merged with a substrate stream containing 3.0 or 2.8 g 1-I of L- or D- histidine, respectively. The calibration graphs obtained by plotting the induction period [f, difference between the induction period for the sample and blank (about 6 min), in minutes] against the aspartic acid concentration (mg 1- I ) were t = -0.08 + 0.26 [L-aspartic acid] t = 0.03 + 0.19 [n-aspartic acid] I - = 0.997; linear rangc 3 4 0 mg 1 r = 0.998; linear range 4-40 mg 1-1 The detection limits, calculated a\ three times the standard deviation of the induction period for 10 determinations of the same blank (Milli-Q-purified water), were I and I .8 mg 1.- for L- and D-aspartic acid, respectively. The RSD, obtained by measuring 1 1 samples containing 10 mg 1- I of each enantiomer, was 2.1 and 2.5% for the L- and D-enantiomer, respectively. Study of the Interference of Amino Acids The influence of L-amino acids frequently encountered in pharmaceutical products was studied by adding different amounts of each potential interferent to samples containing 1 0 mg 1- I of L-aspartic acid.As can be seen from Table I , most of the species tested were tolerated at high concentrations in the determination of 1,-aspartic acid. The most serious interferences were those of the L-enantiomers of cysteine, glutamic acid and diaminocarboxylic acids (lysine, ornithine and arginine), which interfered at concentrations below that of L-aspartic acid. The maximum tolerated concentrations of D-amino acids in the determination of 10 mg 1-1 of n-aspartic acid are also included in Table 1. All amino acids that interfered increased the induction period for the substrate crystallization. An amino acid was considered to interfere when the induction period for the 10 mg 1- I standard (about 8 min) was increased by 0.4 min.As can be seen, the greatect interferences were those from diamino- carboxylic acids, as in the determinalion of L-aspartic acid. A more detailed study of potential interferences in the determina- tion of L-aspartic acid was carried out than for D-aspartic acid because pharmaceutical preparations contain predominantly the ].-form. The D-enantiomer did not interfere in the determination of L-aspartic acid at concentrations seven times in excess; on the other hand, L-aspartic acid is tolerated at a concentration only 2.5 times higher than that of D-aspartic acid (analyte). However, this tolerated ratio permits the discrimination of the two isomers. Analysis of Pharmaceutical Preparations and u Racemic Sample Amino acids present in pharmaceutical preparations are L- enantiomers, except for phenylalanine, histidine, methionine and tryptophan, which are also active in the D-configuration.3 Therefore, the proposed method can only be applied to the determination of L-aspartic acid in pharmaceutical samples. Only two commercial samples were available for this purpose, which were dissolved as described under Experimental. Table 2 gives the average results for five determinations of r>-aspartic1400 Analyst, October 1996, VoE. 121 Table 1 Effect of various amino acids on the determination of L- and D-aspartic acid (1 0 mg 1 I ) [.-Aspartic acid determination D-Aspartic acid determination Amino acid Maximum tolerated concentrationlmg 1-' Amino acid Maximum tolerated concentrationlmg 1-1 11-Aspartic acid 70 L-Aspartic acid 25 L-methionine, L-valine, D- Alanine 40 L-Isoleucine, L-phenylalanine, D- V a1 i ne > 100 D- Asparagine 35 L-Alanine, i -leucine 7 s D-Glutaniic acid 5 L-Serine, L-asparagine 50 D-Lysine, n-orni thine, D-arginine 1 L-Threonine, L-glutamine 25 L-Cysteine 5 L -Lysine, L-ornithine 1 .5 L- Arginine 1 L-tyrosine > 100 L-Glutamic acid 4 Table 2 Determination of L- and D-aspartic acid spiked in pharmaceutical preparations L-Aspartic acid D-Aspartic acid Nominal content/ Found/ Added/ Found/ Trade name mg per tablet or pill BOI-K aspktico 350 356+ 10 225 225 f 6 (tablet) 300 290 f 6 mg per tablet or pill mg per tablet or pill mg per tablet or pill 600 610f 12 Aspartono 410 (Pill) 395 + 13 270 360 720 274 k 7 353 * 7 716+ 14 ' 1.-Aspartic is present as asparlate in the original sample.acid and their standard deviations. Because no real samples containing D-aspartic acid were available, the above samples were spiked with this enantiomer. Three standard additions were made for each preparation following dissolution and dilution (the final concentrations of D-aspartic acid in the diluted samples, were 7.5, 10 and 20 mg 1-1). The recoveries obtained from five individual additions of D-aspartic acid were close to 100% in all instances (Table 2). The potential of the proposed method for the discrimination of L- and D-aspartic acid was assessed on a real racemic sample (D,L-aspartic acid from Sigma). For this purpose, solutions containing various concentrations of the racemate were ana- lysed.The mean content and standard deviation ( n = 3 ) of each enantiomer at three different racemate concentrations (20, 30 and 40 mg 1-I) were 10.2 & 0.2, 14.9 & 0.3 and 20.1 & 0.4 mg 1-l (L-aspartic acid): and 9.7 k 0.2, 15.3 f 0.4 and 20.0 & 0.5 mg 1-1 (D-aspartic acid), respectively. Conclusions The inhibitory effect of aspartic acid on the crystallization of histidine can be successfully exploited for the indirect determi- nation of L- and D-aspartic acid. The chief advantages of the proposed method are as follows: ( 1 ) D- and L-aspartic acid can be determined in the same sample simply by changing the reagent (L- or D-histidine, respectively); (2) none of the chiral columns or mobile phases required in HPLC are needed, so analysis costs are reduced; and (3) the method can be used for on-line control of the enantiomeric purity of pharmaceuticals.The principal limitation of the method is that it cannot be applied to amino acid mixtures and can only play a role when the composition of the other amino acids in the material is known . The authors are grateful to the Direccion General de Investiga- ci6n Cientifica y Tkcnica (Project No. PB95-0977) for financial support. References 1 2 3 4 5 6 7 8 9 10 I 1 12 13 Pasteur, L., A m . Chim. Phys., 1850, 28. 56. Allenmark, S. G., Chromut~~grz~phic Enantiosepurotion: Methods urzd Applications, Ellis Horwood, Chichester, 1988. Brueckner, H., Haasmann, S., Langer, M., Weesthauser, T., Witter- ner, R., and Godel. H., J . Chromatogr., 1994, 666, 259. Toyooka, T., and Liu, Y. M., .I. Chromatogr., 1995, 689, 23. Sukbuntherng, J., Hutchaleelaha, A., Chow, H. H., and Mayersohn, M., .I. Anal. To.xicnl., 1995, 19, 139. Weissbuch, I.. Addadi, L., Lahav, M., and Leiserowitz, L., Science, 1991,253, 637. Grases, F., and March, .l. G., Trends Anul. Cheni., 1991, 10, 190. Grases. F., and Genestar, C., Taluntcr, 1993, 40, 1589. Grases, F., Costa-Bauzj, A., Forteza. R., and March, J . G., Afztcl. L c f f . , 1994,27, 278 1. Schugerl, K., Ulber, R., and Scheper, T., Tieizd.7 Anul. Chenz., 1996, 15, 56. Ballesteros, E., Gallego, M., Valcarcel, M., and Grases, F., Anal. Cheni., 1995, 67, 3319. Ballesteros, E., Gallego, M., Valciircel, M., and Grases, F., And. Chim. Acta, 1995, 315, 145. Del Pozo, A., Fcirnzucia GalPnica Esprciul, Romargraf, Barcelona, 1978. Paper 6103425.1 Rec*eivod May I6, 1996 Accepted June 17, I996