UN. HEM. 4/2, 245-249 (1994) #{149} Automation and Analytical Techniques Plasma Amino Acids Determined by Liquid hromatography Within 17 Minutes Tom Teerlink,1 3 Paul A. M. van Leeuwen,2 and Alexander Houdi.jk2 We present an HPL method for the determination of amino acids in plasma. The method is based on automated precolumn derivatization of amino acids with o-phthalaldehyde, separation of the derivatives by reversed-phase chromatography, and quantification by fluorescence detection. omplete separation was achieved within 12 mm. Total analysis time, includingderivatization, chromatography, and reequilibration of the column, was 17 mm. The assay was linear from 5 to 8 mol/l for all amino acids. Recovery of amino acids added to plasma samples was 96-16%, except for tryptophan (89%). Within-run precision (V) was 1.8-6.4%, and betweenrun precision was 2.1-7.2%. The method can be used for determining primary amino acids in plasma and cerebrospinal fluid. The simple sample preparation and short analysis time make the method particularly suitable for routine analysis of large series of samples. IndexIng Terms:chromatography, reversed-phase/fluorometr, Determination of plasma amino acids has become increasingly important in evaluating the nutritional requirements of patients. This has resulted in supplementation of nutrition with amino acids such as glutamine, arginune, and branched-chain amino acids (1-3). The beneficial effects of these feedings on the nutritional status of patients has stimulated clinicians and researchers to study amino acid metabolism in different disease states (6). These developments have led to an increase in the number of plasma amino acid determinations and the need for a rapid determination method. Physiological amino acids are traditionally determined by ion-exchange chromatography in combination with postcolumn ninhydrin detection. These analyses, usually performed on dedicated instruments, are somewhat laborious, costly, and time-consuming. Alternatively, amino acids can be analyzed by reversed-phase HPL after precolumn derivatisation. Various reagents for the precolumn derivatization of amino acids can be used, e.g., dabsylchloride (7), phenylisothiocyanate (8-11), naphthylisocyanate (12), 9-fluorenylmethyl chioroformate (13, 14), and o-phthalaldehyde (OPA) (15-23). Of these reagents, OPA is most widely used. In the presence of a thiol compound, OPA reacts Departments of linical hemistry and 2Surgery, Free University Hospital, P.O. Box 757, 17 MB, Amsterdam, The Netherlands. 3Mdress correspondence to this author. Fax +31 25487673. Nonstandard abbreviations: OPA, o-phthaladehyde; MPA, 3-mercaptopropionic acid; and SSA, 5-sulfosalicylic acid. Received August 16, 1993; accepted October 11, 1993. readily with primary amino acids, forming highly fluorescent derivatives. Although the isoindole derivatives formed are not very stable (24), accurate results can be obtained by automation of the derivatization reaction. Stabifity of the derivatives depends on the choice of the sulfhydryl reagent added to the derivatization mixture (24). 3-Mercaptopropionic acid (MPA) reportedly results in more stable products than does 2-mercaptoethanol (18, 23). Many of the published methods that include OPA derivati.zation require analysis times close to 1 h, limiting the number of samples that can be analyzed in a single run. Furthermore, most reports indicate that not all amino acids in plasma are adequately separated. We have developed a method for the analysis of all major primary amino acids in plasma, using automated precolumn derivatization with OPA in combination with MPA. Separation is achieved in <12 miii with a cycle time of 17 miii. MaterIals and Methods hemicals We obtained OPA and MPA from Fluka (Buchs, Switzerland). 5-Sulfosalicylic acid (SSA), boric acid, potassium dihydrogen phosphate, and HPL-grade methanol were supplied by Merck (Amsterdam, The Netherlands). HPL-grade acetonitrile was obtained from Westburg (Leusden, The Netherlands). Tetrahydrofuran was supplied by J. T. Baker (Deventer, The Netherlands). We prepared HPL-grade water from demuneralized water by using a Mffli-Q UF Plus water purification system (Millipore, Milford, MA). Triethylamine was obtained from Aldrich (Brussels, Belgium). Individual amino acids were obtained from Sigma (St. Louis, MO), Merck, and BDH hemicals (Poole, UK). Amino acid standard solutions were prepared in 1 mmolfl H1. A combined standard solution, prepared by mixing individual standard solutions, was stored in small aliquots at -2#{176}. A separate standard solution of glutamine dissolved in water was prepared and stored at -2#{176}. Equipment The HPL system consisted of a Model 48 pump from Gynkotek (Germering, Germany) and a Model 232-41 autosampler from Gilson (Villiers le Bel, France). The injection valve (Rheodyne, otati, A) was equipped with a 2-&L sample loading loop. To minimize pressure pulses during switching of the injection valve, we fitted it with a pressure-bypass circuit, as described by Dolan and Snyder (25). Eluents were degassed on-line with a Gastorr GT-13 degasser from Lab Quatec (Tokyo, Japan). Peak monitoring was performed with a Model 98 fluorescence detector from Applied LINIAL HEMISTRY, Vol. 4, No. 2, 1994 245
Biosystems (Ramsey, NJ). A Baseline 81 chromatography workstation from Mfflipore, running on a M29S personal computer (Olivetti, Rotterdam, The Netherlands) was used for data processing. Analyses were performed on a 3-pm Microsphere 18 column [1 cm x 4.6 mm (i.d.)] from hrompack (Middelburg, The Netherlands), protected by a reversed-phase guard column [1 cm x 2. mm (i.d.)] from the same supplier. Subjects To determine reference values for amino acids, we analyzed plasma samples from 44 healthy volunteers (mean age 38 years, range 19-77 years). Blood samples were obtained after an overnight fast. The protocol was conducted with approval of the Ethics ommittee of the Free University Hospital and with oral informed consent of each subject. ollection and Processing of Samples Tubes containing 2 mg of SSA were prepared as follows: A solution of 2 g/l SSA in absolute ethanol was divided into.1-ml aliquots in 1.-mL tubes. After evaporation of the ethanol in an oven (overnight at 5#{176}), the tubes were capped and stored at room temperature. These SSA tubes contain sufficient SSA to deproteinize.5 ml of plasma. We also prepared tubes containing 8 mg of SSA, sufficient to deproteinize.2 ml of plasma. Venous blood was collected in heparincontaining tubes. After centrifugation (2g for 1 miii at 4#{176}),.2 or.5 ml of plasma was pipett.ed into a SSA tube. The tubes were vigorously mixed on a vortex-type mixer, snap-frozen in liquid nitrogen, and then stored at -7#{176}. On the day of analysis the tubes were thawed at room temperature and centrifuged (15 miii at 3g and 4#{176}) to remove precipitated protein. The clear supennatants were transferred to polypropylene vials and stored in a refrigerated (4#{176}) sample rack of the autosampler. Preparation of Reagents The OPA-MPA reagent stock solution was prepared every 3 days. We dissolved 25 mg of OPA in.5 ml of methanol, after which we added 4.5 ml of borate buffer (1 mmol/l, ph 1.) and 25 L of MPA. OPA-MPA working solution was prepared on the day of analysis by adding one part of stock solution to 2 parts of 1 mmol/l borate, ph 1.. Final concentrations of OPA and MPA were 1.77 and 2.72 mmol/l, respectively. An internal standard stock solution (1 mmoljl norvaline in 5 mmol/l H1) was prepared every 2 months and stored at room temperature. Internal standard working solution was prepared by diluting the stock solution with water to a final concentration of 1 mmol/l norvaline. Neutralizing buffer consisted of 4 mmol/l potassium dihydrogen phosphate and 1 mill triethylamine and was prepared each week. Derivatization and hromatography The buffer used for the preparation of the mobile phases consisted of 9 mmol/l potassium dihydrogen phosphate and.5 mill triethylainine, adjusted to ph 6.9 with KOH. Mobile phase A was prepared by mixing 1 ml of this buffer, 1 ml of water, and 4 ml of tetrahydrofuran. Mobile phase B was a mixture of the buffer, methanol, and acetonitrile (5/35/15 by vol). Both mobile phases were filtered through a.45-pm filter by suction filtration. Fully automated precolumn derivatization was performed by using the Gilson autosampler. The sample tray contained three racks. One rack contained up to 96 samples in capped vials. A second rack contained an equivalent number of empty reaction vials. Internal standard solution, OPA-MPA reagent, and neutralizing buffer were each stored in multiple capped vials in the third rack. Each set of vials was used for the derivatization of six consecutive samples. Derivitization was started by the transfer of 2 L of water to an empty reaction vial, followed by the addition of 5 L of sample, 5 il of internal standard solution, and 9 L of OPA-MPA reagent. Sample and reagent were mixed by five cycles of aspirating the incubation mixture from the bottom of the tube and dispensing it from a height of 5 mm. After a 3-mm incubation at room temperature, 5 pi of neutralizing buffer was added; after mixing as described above, 3 L of the reaction mixture was injected. The needle of the autosampler was washed with water between all fluidtransfer steps. During chromatography of one sample, the next sample was being derivatized. hromatographic separation was performed at ambient temperature. Table 1 shows the composition of the eluent during analysis. Detection was performed fluorometrically by using an excitation wavelength of 23 nm and an emission cutoff filter of 389 nm. We calculated amino acid concentrations by using the peak area relative to the area of the internal standard peak. Assay Performance Linearity of the analysis was determined from 5 to 8 tmol/l by plotting peak response (area of the amino acid peak divided by the area of the internal standard peak) vs concentration. At concentrations 1 mol/l, a mixture of all amino acids evaluated was used. At higher concentrations, mixtures of only five amino acids were analyzed, to ensure that the OPA-MPA reagent was present in sufficient excess for complete denivatization. Within-run reproducibility was evaluated by replicate analysis of a standard solution (n = 33) and a plasma pool (n = 28) in a single run. Between-run reproducibility was Table 1. HPL gradient condltions. Tim., mln % mobll#{149} phas.. 2 3.5 25 5.2 44 6.9 52 1. 1 11. 2 #{149} Flow rat e was 1.5 mllmin. B In eluent 246 LINIAL HEMISTRY, Vol. 4, No. 2, 1994
tested by analyzing a plasma pool stored at -7#{176} on different days over 4 months (n = 24). Recovery was evaluated by mixing equal volumes of plasma and amino acid standard mixture or water. After mixing, the samples were deproteinized with SSA-coated tubes and analyzed immediately. Recovery was calculated as the difference between samples containing amino acid standards and samples without the mixture, expressed as a percentage of the amount of amino acid added. Results Fig. 1 shows the chromatographic separation of an amino acid standard and a plasma sample. Eluent composition and gradient shape were optimized to obtain baseline resolution of all amino acids. Some critical.7 A I-..6.4,,.3 s-i I- b..1 U. 2 4 6 8 1 12 Time, mm.3 B ).1 IL. 2 4 6 8 1 12 Time, mm Fig. 1. Typicalchromatogramsof aminoacid standard(a) and human plasma sample (B). oncentrations of amino acids inthe standard solutionare 4 &niol/l, except His, 1-mh, and 3-mh (1 tmo1/l) and Tau (2 moi/l). 1-mh, 1.meffhlsfjdjne; 3-mh, 3-melhlhlstidine; Aba, a-aminobutyr$c acid; Noival, noivaline. LINIALHEMISTRY, Vol. 4, No. 2, 1994 247
pairs of amino acids are difficult to separate. To obtain baseline separation of glycine and threonine we found it essential to include a small percentage of tetrahy. drofuran in mobile phase A. Separation of tryptophan and phenylalanine was accomplished by optimizing the ratio of methanol and acetonitrile in mobile phase B. Inclusion of triethylanilne in mobile phase A allowed separation of aspartic acid from SSA used to deproteinize the plasma samples. The elution position of histidine and arginine relative to the other amino acids can be shifted by adjusting the ionic strength of the mobile phase buffer. Separation of 1-methyihistidine and 3- methyihistidine was optimized by varying the ph of the mobile phase buffer. After determining the optimal composition of the eluents, fine-tuning of the separation was performed by modifying the gradient shape. By using a segmented gradient (Table 1), resolution was improved with a concurrent reduction of analysis time. There was a linear relation between peak response (ratio of areas of amino acid and internal standard) and concentration from 5 to 8 moi/l for all amino acids (correlation coefficients all >.999). Within-run precision was determined by replicate analysis of an amino acid standard solution (n = 33) and a plasma sample (n = 28). For the standard solution the V was 1.-1.8%. For the plasma sample the V was 1.8-4.9% with the exception of aspartic acid (6.4%). For quality-control purposes a plasma pool was stored at - 7#{176} in small aliquots; this sample was analyzed in each run. Between-run V, assessed from the results from 24 consecutive runs performed over 4 months, was <5% with the exceptions of aspartic acid (7.2%), taurine (5.8%), and tryptophan (6.1%). Accuracy of the method was evaluated by determining analytical recovery. Recovery of amino acids added to plasma (n = 12) was 96-13% for all amino acids, with the exception of tryptophan (89%) and aspartic acid (16%). Amino acid concentrations determined in plasma of 44 healthy volunteers are summarized in Table 2. DiscussIon The objective of this study was to develop a fast and reliable method for the analysis of plasma amino acids. Precolumn derivatization with OPA was used because the derivatization can be easily automated. To minimize problems due to oxidation of the reagent (26), we kept it in multiple capped vials in the autosampler. Each vial was used for the derivatization of six consecutive samples. A potential source of analytical variation may be the instability of the isoindole derivatives formed by reaction with OPA (24). We used MPA as thiol reagent because it has been shown to form more stable products than mercaptoethanol (18, 23). OPA concentration in the reagent solution was kept rather low (1.76 mmoijl), but excess OPA during derivatization was ensured by adding 18 volumes of reagent to one volume of sample. Assuming a total amino acid concentration of -3 mmol/l for plasma, we used OPA at a 1-fold molar excess over total amino acids. In initial experiments Table 2. Plasma amino acid concentrations subjects (n = 44). oncentration, In healthy &moi/l AmIno acid Mean SD Range Aspartic acid 4 3-9 Glutamic acid 43 2 12-98 Asparagine 51 1 36-71 Serlne 112 2 67-161 Glutamine 681 95 482-938 Glycine 224 47 121-49 Threonine 129 24 7175 Histidine 83 12 48-121 itrulline 32 9 157 Alanlne 356 73 23-518 Taunne 83 4 38-197 Arginine 9 15 5-126 Tyrosine 59 11 43-9 a-aminobutyrlcadd 25 7 9-45 Vallne 237 44 168-35 Methionine 24 5 13-33 Tryptophan 45 1 272 Phenylalanine 55 9 81 Isoleucine 68 15 42-98 Omithine 57 18 2112 Leucine 122 23 73-172 Lyslne 163 33 17-244 with higher OPA concentrations, we observed that some amino acid derivatives (i.e., glycine, histidine, and taurine) degraded, and some additional unidentified peaks probably representing degradation products appeared in the chromatograms. This is in accordance with reports that excess OPA accelerates degradation of isoindole derivatives (24). On-column degradation of the derivatives was minimized by using a short column in combination with a high flow rate, resulting in an analysis time of only 12 mm. To obtain sufficient resolution, we used a column containing 3-nm particles. An additional advantage of these small-diameter particles is that resolution is only marginally decreased at high flow rates, due to a more efficient mass transfer. To obtain reliable values for plasma amino acids, we paid special attention to sample pretreatment. After collection of blood, plasma was immediately separated from blood cells by centrifugation, and plasma proteins were then precipitated by addition of SSA. In a comparison of various deproteinizing agents, Qureshi and Qureshi showed SSA to be superior to other agents (16). We used special tubes containing solid SSA to avoid errors due to sample dilution. After mixing within the tubes, the samples were quickly frozen in liquid nitrogen and then stored at -7#{176}. In our experience, samples prepared in this way can be stored for at least 1 year without change in amino acid concentrations. This method is characterized by high precision and reproducibifity for all amino acids evaluated. With one exception, the recoveries of amino acids added to plasma were close to 1%. The lower recovery of tryptophan 248 LINIALHEMISTRY. Vol. 4, No. 2, 1994
(89%) probably reflects the fact that this amino acid is tightly bound to albumin (19). Plasma amino acid concentrations of healthy subjects determined by the present method (Table 2) are in good agreement with values obtained by other investigators (15-18,2,23). Although we are primarily interested in plasma amino acids, the method can also be applied to the determination of amino acids in cerebrospinal fluid. The method is characterized by speed of analysis and minimal manual sample handling. With proper care the column can be used for at least 5 analyses. If the separation efficiency decreases, column performance can be restored by flushing the column with.2 mol/l phosphate buffer (ph 2.5). References 1. Stehie P, Mertes N, Puchstein, Zander J, Albers S, Lawin P, FUrst P. The effect of parenteral glutamine peptide supplements on muscle glutamine loss and nitrogen balance after major surgery. Lancet 1989;i:231-3. 2. Barbul A. Arginine: biochemistry, physiology and therapeutic implications [Review]. J Parenter Enteral Nutr 1986;1:227-38. 3. Freund H, Hoover H, Atamian S, Fischer JE. Infusion of the branched chain amino acids in postoperative patients. Ann Surg 1979;19:18-23. 4. Riedel E, Hampi H, Scigalla P, NOndel M, Kessel M. orrection of amino acid metabolism by recombinant human erythropoietin therapy in hemodialysis patients. Kidney hit 1989;36(Suppl 27): 5216-21. 5. KubotaA, Meguid MM, Hitch D. Amino acid profiles correlate diagnostically with organ site in three kinds of mslignrnt tumors. ancer 1992;69:2343-8. 6. Gardiner K, Barbul A. Intestinal amino acid absorption during sepsis. J Parenter Enteral Nutr 1993;17:277-83. 7. Stocchi V, Piccoli G, Magnani M, Palms F, Biagiarelli B, uochiarini L. Reversed-phase high-performance liquid chromatography separation of dimethylaminoazobenzene sulfonyl- and dimethylaminoazobenzene thiohydantoin amino acid derivatives for amino acid analysis and microsequencing studies at the picomole level Anal Biochem 1989;178:17-17. 8. Heinrikson RL, Meredith S. Amino acid analysis by reversephase high-performance liquid chromatography: precolumn derivatization with phenylisothiocyanate. Anal Biochem 1984;136: 65-74. 9. ohen SA, Strydom DJ. Amino acid analysis utilizing phenylisothiocyanate derivatives. Anal Biochem 1988;174:1-16. 1. Fierabracci V, Masiello P, Novelli M, Bergamini E. Application of amino acid analysis by high-performance liquid chromatography with phenyl isothiocyanate derivatization to the rapid determination of free amino acids in biological samples. J hromatogr 1991;57:285-91. 11. Buzzigoli G, Lanzone L, iociaro D, Frascerra S, ern M, Scandroglio A, et al. haracterization of a reversed-phase highperformance liquid chromitographic system for the determination of blood amino acids. J hromatogr 199;57:85-93. 12. Neidle A, Banay-Schwartz M, Sacks S, Dunlop DS. Amino acid analysis using 1-naphthylisocyanate as a precolumn high performance liquid chromatography derivatization reagent. Anal Biochem 1989;18:291-7. 13. Haynes PA, Sheumack D, Kibby J, Redmond JW. Amino acid analysis using derivatisation with 9-fluorenylmethyl chloroformate and reversed-phase high-performance liquid chromatography. J hromatogr 1991;54:177-85. 14. Einarason S, Josefsaon B, Lagerkvist S. Determination of amino acids with 9-fluorenylmethyl chloroformate and reversedphase high-performance liquid chromatography. J hromatogr 1983;282:69-18. 15. Fermo I, De Vecchi E, Diomede L, Parom R. Serum amino acid analysis with pre-column derivatization: comparison of the o- phthaldialdehyde and N,N-diethyl-2,dinitro-5-fluoroaniline methods. J hromatogr 199;534:23-35. 16. Qureshi GA, Qureshi AR. Determination of free amino acids in biological samples: problems of quantitation. J hromatogr 1989;491:281-9. 17. Blundeli G, Brydon WG. High performance liquid chromatography of plasma amino acids using orthophthalaldehyde denivatisation. lin him Acta 1987;17:79-84. 18. Godel H, Graser T, Foldi P, Pfaender P, FUrst P. Measurement of free amino acids in human biological fluids by high-performance liquid chromatography. J hromatogr 1984;297:49-61. 19. Uhe AM, offier GR, McLennan BA, Tucker DJ, O Dea K Quantitation of tryptophan and other plasma amino acids by automated precolumn o-phthaldialdehyde derivatization high-performance liquid chromatography: improved sample preparation. J hromatogr 1991;564:81-91. 2. Van EiJk JiM!!, Van der Hejden MAN, Van Berlo LH, Soeters PB. Fully automated liquid-chromatographic determination of amino acids. lin hem 1988;34:251-3. 21. anevari L, Vieira R, Aldegunde M, Dagani F. High-performance liquid chromatographic separation with electrochemical detection of amino acids focusing on neurochemical application. Anal Biochem 1992;25:137-42. 22. Schuster R. Determination of amino acids in biological, pharmaceutical, plant and food samples by automated precolumn derivatization and high-performance liquid chromatography. J hromatogr 1988;431:271-84. 23. Ziegler F, La Boucher J, oudray-lucas, ynober L. Plasma amino-acid determinations by reversed-phase HPL: improvement of the orthophthalaldehyde method and comparison with ion exchange chromatography. J Autom hem 1992;14:145-9. 24. Alvarez-oque MG, Hern#{225}ndezMJM, amafias EMV, Fern#{225}ndezM. Formation and instability of o-phthalaldehyde derivatives of amino adds. Anal Biochem 1989;178:1-7. 25. Dolan JW, Snyder LR. Troubleshooting L systems. lifton, NJ: Humana Press, 1989:258-6. 26. May ME, Brown LL. Instability of orthophthalaldehyde reagent for amino acid analysis. Anal Biochem 1989;181:135-9. LINIALHEMISTRY,Vol.4, No.2, 1994 249