Quantification of Amino Acids in Fermentation Media by Isocratic

Dec 1, 2010 - tion of amino acids. First, r-hydroxy acid derivatives of amino acids were formed after reaction with dinitrogen trioxide by the van Sly...
2 downloads 0 Views 264KB Size
Download PDF
Anal. Chem. 2011, 83, 175–181

Quantification of Amino Acids in Fermentation Media by Isocratic HPLC Analysis of Their r-Hydroxy Acid Derivatives Daniel Pleissner,†,‡ Reinhard Wimmer,† and Niels T. Eriksen*,† Department of Biotechnology, Chemistry and Environmental Engineering, Aalborg University, Sohngaardsholmsvej 49, DK-9000 Aalborg, and Institute of Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark In this paper we describe a novel method for quantification of amino acids. First, r-hydroxy acid derivatives of amino acids were formed after reaction with dinitrogen trioxide by the van Slyke reaction. Second, the r-hydroxy acid derivatives were separated on an Aminex HPX-87H column (Bio-Rad) eluted isocratically with 5 mM H2SO4 and quantified by refractive index detection. We were able to measure the reaction products of 13 of the 20 classical amino acids: glycine, L-alanine, L-valine, Lleucine, L-isoleucine, L-methionine, L-serine, L-threonine, L-asparagine, L-glutamine, L-aspartic acid, Lglutamic acid, and L-proline. We obtained linear relationships between the product peak areas and initial amino acid concentration, whereby the concentrations of these amino acids could be quantified on the basis of the quantification of their products. The method can be used to analyze amino acids in parallel with other small molecules, such as sugars or short chain fatty acids, and was used for parallel quantification of glycine, L-alanine, or L-glutamic acid, and glucose uptake in cultures of the heterotrophic dinoflagellate Crypthecodinium cohnii. The method can also be used to quantify other amines, as demonstrated by detection of Tris (2-amino-2-(hydroxymethyl)propane-1,3-diol). Accurate analysis of metabolites and other small molecules is essential for many biotechnological processes. One of the most versatile methods for quantification of metabolites is based on isocratic separation of the molecules by Aminex columns (BioRad), Eurocat columns (Knauer), or similar columns in combination with refractive index (RI) or UV absorption detection. On the same column, disaccharides, hexoses, pentoses, organic acids, ketones, alcohols, polyols, and other molecules can be separated and simultaneously analyzed in the same sample. The method has been widely used in fermentation studies1,2 and in food * To whom correspondence should be addressed. Fax: +4598141808. E-mail: [email protected]. † Aalborg University. ‡ University of Southern Denmark. (1) Hidayat, B. J.; Eriksen, N. T.; Wiebe, M. G. FEMS Microbiol. Lett. 2006, 254, 324–331. (2) Adigu ¨ zel, A. C.; Bitlisli, B. O.; Yas¸a, I.; Eriksen, N. T. J. Appl. Microbiol. 2009, 107, 226–234. 10.1021/ac1021908  2011 American Chemical Society Published on Web 12/01/2010

analysis,3 but separation is restricted to uncharged molecules. Since the separation takes place at low pH, amino acids will be charged and are therefore not available for analysis. In this paper we describe a method for quantitative analysis of amino acid derivatives which are created using the van Slyke reaction.4 The van Slyke reaction is based on the transformation of amino acids into their corresponding R-hydroxy acids (Scheme 1) by a reaction between amino groups and dinitrogen trioxide, formed from nitrite under acidic conditions, and the formation (nitrosation) of a labile diazonium compound.5-8 The diazonium group is released by an intramolecular lactone formation with a carboxyl group9 predominantly as molecular nitrogen, but also the release of nitrogen monoxide has been observed.10 Hydrolysis of the lactone by an increase of the pH results in the formation of different R-hydroxy acids depending on the original amino acid. In the case of alanine, only an R-lactone is formed, which is further hydrolyzed to lactic acid (Scheme 1). When the reaction is performed in the presence of nucleophilic species, additional nitrosating agents may be formed. Chloride, for example, forms nitrosyl chloride in the presence of nitrite. These nitrosating agents act as catalysts and increase the rates of nitrosation of amino acids,11-13 while we do not expect them to affect which products that are finally formed. From the reaction between glutamic acid and dinitrogen trioxide, R-hydroxyglutaric acid as well as a γ-lactone can be formed (Scheme 1). It has been described that R-hydroxyglutaric acid is formed from glutamic acid when the pH is increased to 7 after the reaction, while the γ-lactone remains the end product if (3) Marsili, R. T.; Ostapenko, H.; Simmons, R. E.; Green, D. E. J. Food Sci. 1981, 46, 52–57. (4) van Slyke, D. D. J. Biol. Chem. 1911, 9, 185–204. (5) Sachsse, R.; Kormann, W. Z. Anal. Chem. 1875, 14, 380–383. (6) Cachaza, J. M.; Casado, J.; Castro, A.; Lo´pez Quintela, M. A. Z. Krebsforsch. Klin. Onkol. 1978, 91, 279–290. (7) Casado, J.; Castro, A.; Leis, J. R.; Quintela, M. A. L.; Mosquera, M. Monatsh. Chem. 1983, 114, 639–646. (8) Garcia Santos, M. d. P.; Calle, E.; Casado, J. Polyhedron 2003, 22, 1059– 1066. (9) Garcia Santos, M. d. P.; Calle, E.; Casado, J. J. Am. Chem. Soc. 2001, 123, 7506–7510. (10) Mishmash, H. E.; Meloan, C. E. Anal. Lett. 1971, 4, 295–299. (11) Silva, d. G.; Kennedy, E. M.; Dlugogorski, B. Z. J. Am. Chem. Soc. 2005, 127, 3664–3665. (12) Silva, d. G.; Kennedy, E. M.; Dlugogorski, B. Z. Ind. Eng. Chem. Res. 2004, 43, 2296–2301. (13) Hildrum, K. I.; Williams, J. L.; Scanlan, R. A. J. Agric. Food Chem. 1975, 23, 439–442.

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

175

Scheme 1. Derivatization of Alanine and Glutamic Acid into Their Corresponding r-Hydroxy Acidsa

a Alanine has only one carboxyl group and forms only one R-lactone intermediate, which is hydrolyzed to the R-hydroxy acid lactic acid. Glutamic acid has two carboxyl groups and can form an R-lactone as well as a γ-lactone, both of which can be hydrolyzed to R-hydroxyglutaric acid.

the pH is not subsequently increased.14 More than one end product may also be possible for amino acids containing additional carboxylic, amino, amide, sulfhydryl, or hydroxyl groups. The reaction between amino acids and dinitrogen trioxide under acidic conditions was originally developed for gasometric measurements of released nitrogen to quantify amides in plant extracts.5 Later, the reaction came to be known as the reaction of van Slyke, who used the released nitrogen as a measurement of free and conjugated amino acids in urine.4,15 Determination of released nitrogen gas by gas chromatography has also been used for total amino acid analysis.10 While analysis of the released nitrogen gas can be used as a measure of the total amino acid content in a sample, analysis of the R-hydroxy acids can be used to quantify individual amino acids. We have therefore investigated the reaction between dinitrogen trioxide and the 20 classical amino acids found in proteins and used the reaction to quantify glycine, L-alanine, and L-glutamic acid in the culture supernatant of the heterotrophic dinoflagellate Crypthecodinium cohnii. EXPERIMENTAL SECTION Reaction Conditions. Derivatizations of amino acids were, in most cases, carried out in duplicate in 2.5 mL polypropylene tubes (Eppendorf) at 45 °C for 90 min containing 1 mL of amino acid solution (0-3 g L-1) mixed with 0.2 mL of 1 M potassium nitrite. The reaction was started by decreasing the pH to 1-2 by addition of 0.04 mL of 12 M HCl and stopped by addition of 0.2 mL of 2 M NaOH. Deviations from this protocol are described in the Results and Discussion. HPLC Analysis of r-Hydroxy Acids and Metabolites. All R-hydroxy acids and other products produced by reactions with dinitrogen trioxide, as well as glucose and phosphate in C. cohnii culture supernatants, were quantified in parallel using isocratic HPLC: A 25 µL volume of the reaction mixture or culture supernatant was, after centrifugation for 15 min at 5000g to remove particles, added to an Aminex HPX-87H column (Bio-Rad) and eluted with 0.4 mL min-1 of 5 mM H2SO4 at 27 °C. Detection was performed by a Knauer K-2300 RI detector at room temperature. The retention times and peak areas of derivatives of L-alanine, L-aspartic acid, and glycine were compared to analyses of known concentrations of their corresponding (14) Bal, D.; Gryff-Keller, A. Magn. Reson. Chem. 2002, 40, 533–536. (15) Levene, P. A.; van Slyke, D. D. J. Biol. Chem. 1912, 12, 301–312.

176

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

R-hydroxy acids: lactic acid, malic acid, and glycolic acid, respectively. Enzymatic Analysis of L-Glutamic Acid. L-Glutamic acid was also quantified by an enzymatic assay based on the oxidation of L-glutamic acid by L-glutamic acid oxidase. The product R-ketoglutaric acid was determined spectrophotometrically using 3-methyl-2-benzothiazolone hydrazone hydrochloride.16 NMR Spectroscopy. NMR spectra were recorded on a Bruker DRX600 NMR spectrometer equipped with a TXI (H/C/N) probe and operating at 14.1 T. The pH* (uncorrected meter readings) of the reaction mixtures and reference compounds was adjusted to 3.3 or 13.1, and 1H NMR and [1H-13C] HSQC spectra were recorded at both pH 3.3 and 13.1 at 298.1 K in D2O and referenced to internal 2,2-dimethyl-2-silapentane-5-sulfonic acid. Cultivation of C. cohnii. C. cohnii CCMP 316 was grown in 250 mL conical flasks in 100 mL of medium containing 30 g L-1 sea salt (Red Sea salt), 0.01 g L-1 FeCl3, 0.05 g L-1 (NH4)2SO4, 85 mg L-1 K2HPO4, 20 g L-1 D-glucose, and 5 g L-1 glycine, L-glutamic acid, or L-alanine. The medium was autoclaved at 121 °C for 30 min. Glucose and K2HPO4 were autoclaved separately. Vitamins and trace metals were supplied as reported for the ATCC medium 460 A2E6. The cultures were inoculated by 5 mL of exponentially growing liquid culture grown in the same medium containing L-glutamic acid as the nitrogen source. Cultivations were carried out at 27 °C, at an initial pH of 6.5, and shaken at 190 rpm in an orbital shaker. Aliquots of 2 mL were taken every 24 h. The optical density was measured at 700 nm, after which the sample was centrifuged for 15 min at 5000g and the supernatant used for analysis of amino acids, glucose, and phosphate. RESULTS AND DISCUSSION Reactions between Dinitrogen Trioxide and L-Alanine, L-Glutamic Acid, or L-Aspartic Acid. Lactic acid was the only product detected from the reaction between L-alanine and dinitrogen trioxide. HPLC chromatograms showed only one product peak at 10.5 min (Figure 1A), which is identical to the retention time of lactic acid. The identity of the product was confirmed by NMR spectroscopy (Figure 2B). Since L-alanine has only one hydroxyl group, only one intermediate R-lactone can be formed, which is spontaneously hydrolyzed to form lactic acid as the end (16) Soda, K. Anal. Biochem. 1968, 25, 228–235.

Figure 1. HPLC chromatograms of products at pH 3.3 (---) and pH 13.1 (s) after derivatization of 33 mM L-alanine (A), L-aspartic acid (B), and L-glutamic acid (C) with dinitrogen trioxide.

product (Scheme 1). The peak seen at 0.5 min in all HPLC chromatograms is predominantly caused by chloride ions from the hydrochloric acid added to the reaction mixture, but also other compounds that were not separated in the column eluted at 0.5 min. For some amino acids, the end product profiles of van Slyke reactions depended on the reaction conditions. The reaction

between dinitrogen trioxide and L-aspartic acid may result in malic acid or a β-lactone, and β-lactones have previously been identified as side products for reactions between dinitrogen trioxide and β-amino acids.9 At pH 3.3, HPLC chromatograms showed a dominating peak at 6.5 min, corresponding to the retention time of malic acid, and a smaller peak at 8 min (Figure 1B). After the pH was increased to 13.1, the peak at 8 min disappeared, while the area of the malic acid peak at 6.5 min increased. Our NMR spectra revealed the presence of an ester bond only under acidic conditions (Figure 2E). This ester bond may very well belong to the ester bond in a β-lactone. Under acidic conditions, L-glutamic acid formed three end products with retention times of 8.5, 10.5, and 13.5 min, with the peak at 13.5 min being the dominant one (Figure 1C). When the pH was increased, the peak at 8.5 min became dominant. The dominating end product formed at low pH with a 13.5 min retention time was the γ-lactone, while the product with an 8 min retention time was R-hydroxyglutaric acid (Scheme 1). NMR spectra of the reaction mixture are shown under acidic (Figure 2H) and basic (Figure 2G) conditions. The observed chemical shifts (both 1H and 13C (not shown)) matched those previously reported for the two compounds.14 When the pH was increased to 13.1 after the van Slyke reaction, the γ-lactone disappeared while the concentration of R-hydroxyglutaric acid increased. None of the R-lactones, which presumably are formed as intermediates during reactions between dinitrogen trioxide and L-alanine, L-glutamic acid, or L-aspartic acid, were observed in the NMR spectra (Figure 2). Rates of Reactions between Dinitrogen Trioxide and L-Alanine and L-Aspartic Acid. Figure 3 shows the formation of lactic acid from 100 mM L-alanine and malic acid from 84 mM L-aspartic acid in the presence of 160 mM nitrite. It has previously been reported that L-aspartic acid requires more time for nitrosation compared to L-alanine.9 Under our experimental conditions, both amino acids showed similar initial reaction rates. In addition, after 20 min the concentration of the formed R-hydroxy acid increased to 50 mM in both cases, and similar reaction times were suitable for derivatization of the different amino acids. The reactions in Figure 3 were carried out with butyric acid as the acidifying agent, and the rates of lactic acid and malic acid formation therefore represent the reaction rates without the presence of additional nitrosating agents, such as nitrosyl chloride. There was a linear relationship between the initial L-alanine concentration (0-47 mM) and lactic acid concentration after 30 min of reaction (Figure 4). In this period, 73% of the L-alanine was converted to lactic acid irrespective of the initial concentration of L-alanine. Therefore, the reaction does not need to be completed to use measurements of lactic acid to quantify L-alanine if determinations are based on comparisons to simultaneously analyzed standards of known alanine concentrations. At higher concentrations of L-alanine, the relative yield of lactic acid decreased, because dinitrogen trioxide became limiting. The nitrite concentration in these experiments was kept low compared to that in the original procedures where gas production was quantified4,5,10,15 to avoid bubble formation in the HPLC column. If sufficient unreacted nitrite and maybe also amino acids are still present at the time of analysis, the acidic eluent will re-form dinitrogen trioxide and may even restart the reaction. Bubble Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

177

Figure 2. 1H NMR spectra: (A) L-alanine at pH 13.1, (B) reaction product of L-alanine at pH 13.1, (C) L-aspartic acid at pH 13.1, (D, E) reaction products of L-aspartic acid at pH 13.1 and 3.3, respectively, (F) L-glutamic acid at pH 13.1, (G, H) reaction products of L-glutamic acid at pH 13.1 and 3.3, respectively.

Figure 3. Formation of lactic acid ([) from L-alanine and malic acid (9) from L-aspartic acid in the presence of 160 mM nitrite. The initial concentrations of L-alanine and L-aspartic acid were 100 and 84 mM, respectively. The reactions were started by decreasing the pH to 3-4 by addition of 0.04 mL of 11 M butyric acid, which also functioned as an internal standard during subsequent HPLC analysis.

formation in the column was observed at initial nitrite concentrations above 300 mM. Derivatization of the 20 Classical Amino Acids. The reaction between dinitrogen trioxide and 13 of the 20 classical amino acids resulted in products which could be separated by the Aminex HPX-87H column and detected and quantified by RI detection. Linear relationships were observed between the peak areas of the end products and initial concentrations of all 13 amino 178

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

Figure 4. Concentration of lactic acid produced from initial L-alanine concentrations from 0 to 95 mM during a 30 min reaction time in the presence of 160 mM nitrite. The slope of the regression line at 0-47 mM L-alanine is 0.73.

acids (Figure 5). The initial nitrite concentration was 160 mM, above the initial amino acid concentrations (0-40 mM), and the reaction time was 90 min. Table 1 gives an overview of the retention times of the products, as well as the slopes of the regression lines and coefficients of determination between the peak areas and initial amino acid concentrations. The lower concentration limit for quantification of the amino acids was on the order of 0.1 g L-1 or below depending on the amino acid: the regression lines for glycine and L-alanine derivatives had the lowest slopes, and these two amino acids are therefore the hardest ones to quantify at low concentrations,

Figure 5. Peak areas from HPLC analysis of the products versus initial concentrations of the amino acids (L-aspartic acid, [; L-serine, 0; L-threonine, 2; L-valine, ×; L-alanine, f; L-glutamine, b; glycine, +; L-methionine, -; L-isoleucine, s; L-asparagine, ]; L-glutamic acid, 9; L-leucine, 4; L-proline, O). The slopes of the regression lines are listed in Table 1. Table 1. Retention Times (RTs) of Derivatives of the 13 Amino Acids That Could Be Quantified after Reaction with Dinitrogen Trioxide, Slopes of the Regression Lines of Peak Area ( Standard Error against the Initial Amino Acid Concentrations (Figure 5), and Coefficients of Determination (r2)a slope amino acid

RT, min

glycine L-alanine L-valine L-leucine L-isoleucine L-methionine L-serine L-threonine L-asparagine L-glutamine L-aspartic acid L-glutamic acid L-proline

10.9 10.5 16.1 26.9 24.1 26.9 7.9 8.5 10.7 8.7 6.4 8.7 21.9

-1

Lg

85 ± 8 93 ± 3 108 ± 4 126 ± 3 138 ± 2 111 ± 3 202 ± 3 212 ± 2 203 ± 4 171 ± 7 181 ± 8 224 ± 4 171 ± 6

mM-1

r2

6±1 8±1 13 ± 1 17 ± 1 18 ± 1 17 ± 1 21 ± 1 25 ± 1 27 ± 1 25 ± 2 24 ± 1 33 ± 1 20 ± 1

0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.99 0.99 0.98

a The initial concentrations of amino acids and nitrite were 0-3 g L-1 (0-40 mM) and 160 mM, respectively.

while the L-glutamic acid derivative R-hydroxyglutaric acid gave the highest RI signal. The slopes of the regression lines may depend on the degree of conversion, refractive index of the amino acid derivative, and formation of more than one end product. Glycine was the only amino acid that resulted in two end products having similar peak areas during HPLC analysis. One of the products was, as expected, glycolic acid with a retention time of 10.9 min (Table 1). The second, unidentified product had a retention time of 15.2 min. Both products were formed in the same ratio irrespective of the initial glycine concentration (Figure 6). Therefore, it was possible to use measurements of glycolic acid to quantify glycine. LGlutamic acid and L-glutamine both formed R-hydroxyglutaric acid as the end product with a retention time of 8.7 min when reacted

Figure 6. Relationship between the peak areas of glycolic acid (10.9 min) and unidentified product (15.2 min) formed from 0, 1, 2, and 3 g L-1 glycine after reaction with dinitrogen trioxide. The inset shows the HPLC spectra of the products after reaction of 1 (---) and 3 (s) g L-1 glycine.

with dinitrogen trioxide. This is expected when the amide group in L-glutamine also undergoes derivatization into a hydroxyl group. Analysis of the end products from van Slyke reactions will therefore provide the sum of the initial concentrations of L-glutamic acid and L-glutamine. In contrast, L-aspartic acid and L-asparagine formed different products when reacted with dinitrogen trioxide. While L-aspartic acid was converted to malic acid with a retention time of 6.4 min, L-asparagine was converted to an unidentified product with a retention time of 10.7 min. However, this product was still suitable for quantification of L-asparagine. Analysis of the end products from van Slyke reactions can therefore quantify L-aspartic acid and L-asparagine separately in the same sample. For seven of the amino acids, no products were observed in HPLC chromatograms despite the fact that they did react with dinitrogen trioxide as seen by formation of gas bubbles. Reactions between aromatic amino acids and dinitrogen trioxide lead to nitration of the aromatic ring,17 and the color of the reaction mixture became brown-red. Products from these three amino acids were not retained by the Aminex HPX-87H column under the analytical conditions. Products of L-cysteine, L-arginine, L-lysine, and L-histidine were also not observed in the chromatograms, possibly because structural properties associated with their additional amino or sulfhydryl group or ring systems prevented them from being retained in the Aminex HPX-87H column. L-Proline, however, did form a detectable product with a retention time of 21.9 min after reaction with dinitrogen trioxide. Despite the fact that this amino acid does not contain a primary amino group, L-proline is known to be a substrate for nitrosation by dinitrogen trioxide as well as nitrosyl chloride.13 Quantification of Other Amines. Quantification of compounds after derivatization is not restricted to amino acids, and the method can also be used to quantity other types of primary amines. As an example we have confirmed that Tris (2-amino-2(hydroxymethyl)propane-1,3-diol), which is commonly used as a buffer, and dinitrogen trioxide form a detectable product with a retention time of 8.4 min. (17) Block, R. J.; Bolling, D. J. Biol. Chem. 1939, 129, 1–12.

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

179

Figure 7. Quantification of substrates: phosphate, glucose, and amino acids in supernatants of C. cohnii cultures grown on glycine (A, B), L-alanine (C, D), or L-glutamic acid (E, F) as the nitrogen source. (A, C, E) RI chromatograms of culture supernatants reacted with 160 mM nitrite at day 0 (s), day 3 (---), and day 6 ( · · · ). Phosphoric acid eluted at 2.7 min, glucose at 4.4 min, and glycolic acid from glycine at 10 min (A), lactic acid from L-alanine at 10.5 min (C), and R-hydroxyglutaric acid from L-glutamic acid at 8.7 min (E). The inset in (C) shows phosphoric acid peaks, analyzed prior to reactions with nitrite on an expanded scale. (B, D, F) Cell density measured as the optical density at 700 nm, OD700 ([), concentrations of phosphate (b), glucose (9), and glycine (B), L-alanine (D), or L-glutamic acid (F) (4), measured from chromatograms A, C, and E and L-glutamic acid measured enzymatically (×).

Quantification of Amino Acids in Microbial Cultures. Parallel detection of glucose, phosphate (measured as phosphoric acid), and glycolic acid from glycine, lactic acid from L-alanine, or R-hydroxyglutaric acid from L-glutamic acid was tested in supernatants of batch cultures of C. cohnii, a marine microorganism that needs an organic nitrogen source. Figure 7 shows chromatograms recorded on culture supernatants after reaction with dinitrogen trioxide and the concentrations of glucose, phosphate, and glycine, L-alanine, or L-glutamic acid in the culture. The measured concentrations of glucose and amino acids at time zero corresponded to the concentrations supplied to the media, and the L-glutamic acid concentrations measured after derivatization into R-hydroxyglutaric acid were in agreement with Lglutamic acid concentrations measured enzymatically by the L-glutamic acid oxidase assay. The ionic strength of the reaction medium is not expected to influence the derivatization reaction,9 and we also observed no effects from the high salt concentration in the C. cohnii seawater180

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

based medium on the conversion of L-glutamic acid into R-hydroxyglutaric acid. However, nucleophilic species in the sea salt medium may have formed nitrosyl chloride and possibly other nitrosating agents, which could have increased the rate of reaction11,12 in particular since the pH in the reaction mixtures was below 2.13 Glucose concentrations measured in the culture supernatant before and after derivatization were identical, and oxidation of glucose to gluconic or glucaric acid by nitrite18 therefore is negligible. Phosphoric acid, coming from the phosphate in the C. cohnii medium, was also detectable by HPLC after 2.7 min, although phosphate was a limiting substrate and the initial phosphate concentrations were low. In addition, part of the phosphate was precipitated at the high pH value used to stop the van Slyke reaction. Accurate determination of phosphate was still possible using the same equipment, but depended on the analysis also of (18) Grigor’eva, I. A.; Chernaya, S. S.; Trusov, S. R. Russ. J. Appl. Chem. 2001, 74, 2021–2026.

supernatant samples that had not been reacted with nitrite (Figure 7C, inset). It is therefore clear that amino acids can be quantified not only in pure solutions but also in complex reaction mixtures, and in parallel or sequentially with the analyses of other small molecules. Analysis of supernatants before and after treatment with nitrite may also be used for sequential quantification of amino acids and volatile compounds such as alcohols that may potentially evaporate into the nitrogen bubbles. CONCLUSION Quantitative analysis of amino acid derivatives after reaction with dinitrogen trioxide by isocratic HPLC is a suitable method for quantification of 13 out of the 20 classical amino acids. The reaction can be performed under various conditions. In most of our experiments, we reacted up to 3 g L-1 amino acid with 160 mM potassium nitrite for 90 min at 45 °C by the following protocol: A 1 mL volume of amino acid solution was mixed with 0.2 mL of 1 M potassium nitrite. The reaction was started by decreasing the pH to 1-2 by 0.04 mL of 12 M HCl, and the reaction mixtures were incubated at 45 °C. After 90 min the reaction was stopped by addition of 0.2 mL of 2 M NaOH. For all 13 amino acids which were detectable, linear and reproduc-

ible relationships between the product peak area and initial amino acid concentration were obtained. The method worked not only in pure solutions of amino acids, but also in culture supernatants, where it was possible to carry out parallel or sequential measurements also of other metabolites. The expected R-hydroxy acids were produced by the reactions between dinitrogen trioxide and the amino acids used most commonly in fermentation media, glycine, L-alanine, and L-glutamic acid, and they could all be separated and quantified in media of high ionic strength, while glucose measurements remained unaffected. The method can also be employed for amino acid analyses in, for example, food, feeds, and health products, as well as in diagnostics and for analysis of other amines. ACKNOWLEDGMENT This work forms part of the MarBioShell project supported by the Danish Agency for Science, Research and Innovation.

Received for review August 20, 2010. Accepted November 12, 2010. AC1021908

Analytical Chemistry, Vol. 83, No. 1, January 1, 2011

181