Development of N-Acetyl Methyl Ester Derivatives for the

Nov 1, 2007 - The principal challenge to the GC/C/IRMS carbon isotope analysis of amino acids relates to their derivatization. Amino acids are a nonvo...
1 downloads 0 Views 626KB Size
Anal. Chem. 2007, 79, 9082-9090

Development of N-Acetyl Methyl Ester Derivatives for the Determination of δ13C Values of Amino Acids Using Gas Chromatography-CombustionIsotope Ratio Mass Spectrometry Lorna T. Corr, Robert Berstan, and Richard P. Evershed*

Organic Geochemistry Unit, Bristol Biogeochemistry Research Centre, School of Chemistry, University of Bristol, Cantock’s Close, Bristol, BS8 1TS, UK

A novel derivatization procedure, N-acetyl methyl (NACME) esterification, was developed to improve the accuracy and precision of amino acid δ13C value determination using gas chromatography-combustion-isotope ratio mass spectrometry (GC/C/IRMS). Standard mixtures of 15 protein amino acids were converted to NACME and N-acetyl-isopropyl (NAIP) esters; the latter established derivative was employed for comparison purposes. Both procedures yielded baseline-resolved peaks for all 15 amino acids when GC columns coated with polar stationary phases were employed. For NACME esters, the methylation conditions governed reaction yields, with highest yields observed when a 1 h, 70 °C methylation procedure (anhydrous MeOH/acetyl chloride, 25:4, v/v) was performed. The mean derivatization yields expressed relative to an underivatized coinjected standard (n-nonadecane) for both NACME and NAIP esters were identical. Likewise, the mean kinetic isotope effects (KIEs) were not significantly different (KIENACME ) 1.036; KIENAIP ) 1.038) and were shown in both cases to be reproducible. The mean reproducibility obtained from 15 replicates (3 × batches of 5) of both derivatives was strong (mean STDVNACME ) 0.3‰ and STDVNAIP ) 0.4‰). The isotopic robustness of both derivatization procedures was observed over a concentration range of 52 500 µg of amino acid. NACME esters displayed low errors ((0.6‰ for phenylalanine to (1.1‰ for serine) due to the higher sample-to-derivative carbon ratio of this derivative. Finally, the integrity of the new NACME procedure was confirmed through analysis of diet and bone collagen amino acids of rats reared on C3 or C4 diets, which indicated the high degree of both accuracy and precision of the δ13C values obtained for individual amino acids. In 1961, liquid chromatographic techniques were combined with off-line isotope ratio mass spectrometry (IRMS) to demonstrate that the individual amino acids comprising proteins possess unique δ13C values, reflecting the respective biochemical pathways available for both synthesis and degradation.1 Since this pioneering * To whom correspondence should be addressed. E-mail r.p.evershed@ bristol.ac.uk. Fax: +44 117 9251295.

9082 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

work, the compound-specific isotope technique has evolved considerably, primarily with the introduction of gas chromatography-combustion-isotope ratio mass spectrometry (GC/C/IRMS) analysis in the early 1990s.2-5 GC/C/IRMS analysis is a highly sensitive technique, enabling rapid on-line separation and carbon isotope determinations of individual compounds, negating the requirement for time-consuming off-line preparation procedures. The method has thus far been employed to probe the amino acid carbon isotope compositions of materials as diverse as meteorites,6 eggshells,7 hawkmoth larvae,8 sediments,9 and archaeological bone collagen.10,11 The principal challenge to the GC/C/IRMS carbon isotope analysis of amino acids relates to their derivatization. Amino acids are a nonvolatile, polyfunctional compound class, requiring derivatization to improve their volatility for GC/C/IRMS analysis. Although several derivatization procedures have been used for the GC/C/IRMS analyses of amino acids, the goal of obtaining an optimal method has remained elusive. Silylated derivatives, e.g., trimethylsilyl and tert-butyldimethylsilyl esters, involve addition of excessive numbers of carbon atoms,12 while the presence of silicon raises the question of catalyst poisoning resulting in incomplete analyte combustion.13,14 Alkoxycarbonyl alkylation, involving the formation of ethoxycarbonyl ethyl esters and methoxycarbonyl methyl esters, has been used occasionally for (1) Abelson, P. H.; Hoering, T. C. Proc. Natl. Acad. Sci. U.S.A. 1961, 47, 623632. (2) Matthews, D. E.; Hayes, J. M. Anal. Chem. 1978, 50 (11), 1465-1473. (3) Meier-Augenstein, W. J. Chromatogr., A 1999, 842, 351-371. (4) Demmelmair, H.; Schmidt, H.-L. Isotopenpraxis 1993, 29, 237-250. (5) Silfer, J. A.; Engel, M. H.; Macko, S. A.; Jumeau, E. J. Anal. Chem. 1991, 63, 370-374. (6) Engel, M. H.; Macko, S. A. Nature 1997, 389, 265-268. (7) Johnson, B. J.; Fogel, M. L.; Miller, G. H. Geochim. Cosmochim. Acta 1998, 62, 2451-2461. (8) O’Brien, D. M.; Fogel, M. L.; Boggs, C. L. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 4413-4418. (9) Brisman, K.; Engel, M. H.; Macko, S. A. Precambrian Res. 2001, 106, 5977. (10) Corr, L. T.; Sealy, J. C.; Horton, M. C.; Evershed, R. P. J. Archaeol. Sci. 2005, 32, 321-330. (11) Fogel, M. L.; Tuross, N. J. Archaeol. Sci. 2003, 30, 535-545. (12) Gross, S.; Glaser, B. Rapid Commun. Mass Spectrom. 2004, 18, 2753-2764. (13) Prevost, S.; Nicol, T.; Monteau, F.; Andre, F.; Le Bizec, B. Rapid Commun. Mass Spectrom. 2001, 15, 2509-2514. (14) Shinebarger, S. R.; Haisch, M.; Matthews, D. E. Anal. Chem. 2002, 74, 6244-6251. 10.1021/ac071223b CCC: $37.00

© 2007 American Chemical Society Published on Web 11/01/2007

GC/C/IRMS analyses,15,16 being attractive because of the low number of carbon atoms introduced. However, chromatographic resolution problems,17,18 nonquantitative reactions,19 and poor derivatization yields for several amino acids have been reported for these derivatives.20 The most widely used approaches to amino acid derivatization for GC/C/IRMS analysis involve a stepwise procedure, involving esterification of the carboxyl group with an acidified alcohol and acylation of amine, hydroxyl, and thiol groups with an anhydride, forming N(O,S)-acyl alkyl derivatives.4,8,10,18,21-23 Three derivatives have been used widely for the GC/C/IRMS analysis of amino acids, namely, trifluoroacetyl isopropyl (TFA-IP) esters, N-acetyln-propyl (NANP) esters, and N-pivaloylisopropyl (NPIP) esters. TFA-IP esters are the most commonly employed derivative for GC/C/IRMS analysis because of their superior chromatographic properties; however, there is increasing concern over their potential as an oxidation catalyst poison.3 Similarly, reported results for NANP derivatives have alluded to poor chromatographic resolution,24 while preparation of NPIP derivatives results in addition of excessive exogenous carbon atoms, which is problematic for carbon isotope determinations at natural abundance, hence their almost exclusive application in 13C-enriched/ tracer studies.24 Our objective was thus to identify or develop a new derivatization procedure that would overcome many of the shortcomings highlighted above, associated with currently available procedures. The excessive number of additional carbon atoms introduced during derivatization is the major disadvantage of most methods, since it reduces the precision associated with amino acid carbon isotope determinations.21,25 However, within the esterification/ acetylation procedures, potential still exists to substitute the widely used propylation step for a methylation procedure, producing N-acetyl methyl (NACME) esters, thereby significantly reducing the number of additional derivative carbon atoms from between +5C and +8C (for NANP and TFA/IP esters) to between +3C and +5C (NACME esters). This range in the number of additional carbon atoms for each derivatization procedure results from the extra functional groups that several amino acids contain; e.g., serine has an extra hydroxyl group (requiring acylation) and aspartic acid contains an extra carboxyl group (requiring esterification). Although the use of NACME derivatives of amino acids would seem to be an appropriate approach, it is noteworthy that their preparation, even for GC analysis, appears unexplored. The lack of use of this derivative for the GC or GC/C/IRMS analysis (15) Montigon, F.; Boza, J. J.; Fay, L. B. Rapid Commun. Mass Spectrom. 2001, 15, 116-123. (16) Reijngoud, D. J.; Hellstern, G.; Elzinga, H.; de Sain-van der Velden, M. G.; Okken, A.; Stellaard, F. J. Mass Spectrom. 1998, 33, 621-626. (17) Husek, P. J. Chromatogr. 1991, 552, 289-299. (18) Metges, C. C.; Daenzer, M. Anal. Biochem. 2000, 278, 156-164. (19) Pelaez, M. V.; Bayon, M. M.; Alonso, J.; Sanz-Medel, A. J. Anal. At. Spectrom. 2000, 15, 1217-1222. (20) MacKenzie, S. L.; Tenaschuk, D.; Fortier, G. J. Chromatogr., A 1987, 387, 241-253. (21) Docherty, G.; Jones, V.; Evershed, R. P. Rapid Commun. Mass Spectrom. 2001, 15, 730-738. (22) Jim, S.; Jones, V.; Copley, M. S.; Ambrose, S. H.; Evershed, R. P. Rapid Commun. Mass Spectrom. 2003, 17, 2283-2289. (23) Macko, S. A.; Ryan, M.; Engel, M. H. Chem. Geol. 1998, 152, 205-210. (24) Metges, C. C.; Petzke, K. J.; Hennig, U. J. Mass Spectrom. 1996, 31, 367376. (25) Rieley, G. Analyst 1994, 119, 915-919.

of amino acids may be due to concern over the stability or volatility of these low molecular weight derivatives; however, we did not consider these perceived challenges as insurmountable. Hence, the aims here were to (i) prepare and purify N-acetyl methyl esters of amino acids and (ii) assess their potential application to the GC/C/IRMS analysis of amino acids. In order to assess their relative advantages and disadvantages for compoundspecific stable carbon isotope analysis, we performed a parallel evaluation of NAIP esters as their nearest propylated homologue. Both derivatives were assessed with respect to the following: (i) derivatization yield and purity, (ii) gas chromatographic performance, (iii) stability, (iv) reproducibility of both yield and carbon isotope determinations for identical replicates and at a range of analyte concentrations, and (v) analytical errors associated with their carbon isotope analysis using GC/C/IRMS analysis. EXPERIMENTAL SECTION Reagents and Standards. Two standard mixtures containing amino acids with differing δ13C values were used throughout this investigation. One solution (0.2 mg g-1 amino acid) containing only L-amino acids, and a second solution containing D,L-amino acids, were prepared in 0.1 M HCl by addition of 15 protein amino acid standards, comprising alanine, aspartic acid, glutamic acid, glycine, L-hydroxyproline (D,L-hydroxyproline was unobtainable), isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tyrosine, and valine (Sigma-Aldrich Co. Ltd., Poole, Dorset, UK). Norleucine (Sigma-Aldrich Co. Ltd.), an amino acid that does not occur naturally, was added as an internal standard to each standard mixture to enable quantification. All solvents were of HPLC grade and purchased from either Rathburn Chemicals or Sigma-Aldrich Co. Ltd. Triethylamine was obtained from Sigma-Aldrich Co. Ltd., acetic anhydride from BDH Laboratory Supplies (Merck Ltd., Poole, Dorset, UK), and anhydrous methanol from Acros Organics. Derivatization. N-Acetyl Methyl Esters. Amino acid NACMEs were prepared using a newly developed procedure.26 Aliquots of each amino acid standard mix were methylated in 1 mL of acidified methanol solution (1.85 M, prepared by addition of 800 µL of acetyl chloride to 5 mL of anhydrous methanol in an ice bath) for 1 h at 70 °C. After heating, reagents were evaporated under a gentle stream of N2 in an ice bath. Dichloromethane (DCM; 2 × 0.25 mL) was added and evaporated at room temperature to remove excess methanol and water. Acetylation was performed by addition of 1 mL of acetic anhydride, triethylamine and acetone (1:2:5, v/v/v, 10 min, 60 °C). The amino acid derivatives were dried under a gentle stream of N2 at room temperature and dissolved in 2 mL of ethyl acetate. Saturated NaCl solution (1 mL) was added, and after vortexing and phase separation, the organic phase was removed and dried under a gentle stream of N2 at room temperature. The vials were then placed in an ice bath, DCM was added and any remaining reagent removed under a gentle stream of N2. N-Acetyl Isopropyl Esters. Aliquots of each amino acid standard mixture were isopropylated in a 1-mL solution of acidified 2-propanol (0.5 mL, 2.8 M with acetyl chloride), for 1 h at 100 °C. The reaction was terminated by placing the reaction tubes in a (26) Corr, L. T.; Berstan, R.; Evershed, R. P. Rapid Commun. Mass Spectrom. In press.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

9083

freezer (-20 °C). The residual 2-propanol was removed under a gentle stream of nitrogen at 40 °C. DCM (2 × 0.25 mL) was added and evaporated at room temperature to remove excess 2-propanol and water. Acetylation was performed by addition of 1 mL of acetic anhydride, triethylamine, and acetone (1:2:5, v/v/v, 10 min, 60 °C). The derivatized amino acids were dried under a gentle stream of N2 at room temperature and dissolved in 2 mL of ethyl acetate. A total of 1 mL of saturated NaCl solution was added, and after vortexing and phase separation, the organic phase was removed and dried under a gentle stream of N2 at room temperature. The vials were then placed in an ice bath, DCM was added, and any remaining reagent removed under a gentle stream of N2. Instrumental Analyses. Gas Chromatography. GC analyses were performed using a Hewlett-Packard Series II 5890 gas chromatograph with H2 carrier gas. Flame ionization detection (FID) was used to monitor the column effluent, and data were acquired and analyzed using Thermo Labsystems Atlas software. The chromatographic performance of the two derivatives has previously been assessed on GC columns with six stationary phases ranging from nonpolar to high-polarity stationary phases, including a chiral phase, in order to optimize conditions for GC and GC/C/IRMS analyses.26 Enhanced resolution and peak shapes were achieved using polar phases such as ZB-WAX (poly(ethylene glycol); Zebron) and VF-23ms (50% cyanopropylmethylpolysiloxane; Varian) coated capillary columns, compared to the low-polarity Ultra-2 column (5% phenyl, 95% dimethyl polysiloxane; J & W Scientific) typically utilized for NAIP ester analyses. Hence, the ZB-WAX (poly(ethylene glycol), 60 m × 0.32 mm × 0.15 µm; Zebron) column was employed throughout this investigation. The temperature program utilized was 40 °C (1 min), 120 °C at 15 °C min-1, 190 °C at 3 °C min-1, and 250 °C (20 min) at 5 °C min-1. GC/Mass Spectrometry (GC/MS). GC/MS analyses were performed on a ThermoQuest Trace GC coupled to a ThermoQuest Trace MS operated in electron ionization (EI) mode (70 eV). The configuration included a programmable temperature vaporizing injector, and sample introduction was either manual or automatic via a CTC A200S autosampler. The carrier gas was helium. Identical GC columns and conditions were employed as for GC analysis. Amino acid standard mixtures and single amino acid standards, where necessary, were used to aid mass spectral identification. GC-Combustion-Isotope Ratio Mass Spectrometry. A ThermoElectron DeltaPlus XP system was used to determine δ13C values of derivatized amino acid standard mixtures. The MS (EI 100 eV, three Faraday cup collectors m/z 44, 45, and 46) was interfaced to a ThermoElectron Trace 2000 GC via a ThermoElectron GC combustion III interface (CuO/NiO/Pt oxidation reactor maintained at 940 °C and Cu reduction reactor maintained at 600 °C). Sample injection was on-column, introduced either manually or by automated on-column injection (CTC Analytics PAL autosampler). Helium was used as the carrier gas, and the MS source pressure was maintained at 9 × 10-4 Pa. Data were acquired and analyzed using ISODAT NT 2.0 software. All δ13C values are reported relative to reference CO2 of known carbon isotopic composition, introduced directly into the ion source in three pulses at the beginning and end of each run. Each reported value is a mean of triplicate determinations of δ13C values. To monitor 9084

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

instrument performance an external standard mixture of five synthetic prederivatized amino acids (alanine, leucine, proline, methionine, and phenylalanine NACME esters) of known isotopic composition were analyzed between each analytical run. Quantification of Amino Acids. Since each amino acid possesses a different structure, inequivalent GC-FID responses are obtained when equivalent weights of each amino acid are analyzed. Hence, FID response factors were determined by performing GC analysis on standard mixtures of amino acids and relating amino acid peak areas to that of an equivalent weight of the internal standard norleucine. FID response factors were then calculated using eqs 1 and (2):

(area of AA/area of IS) × concentration of IS ) apparent concentration of AA (1) actual concentration of AA/ apparent concentration of AA ) correction factor (2) where, AA denotes amino acid, IS denotes internal standard, and area denotes GC peak area. kinetic Isotope Effect (KIE) Calculation. To investigate the occurrence and magnitude of the KIE associated with both derivatization procedures, the δ13C value of each amino acid of the standard mixture was determined (derivastized using GC/ C/IRMS and underivatized using elemental analyzer-IRMS (EAIRMS)) and mass balance equations were employed to determine whether the δ13C values of the derivatized standard amino acid could be calculated using the δ13C values of the additional derivatizing carbon (by measuring reagent δ13C values off-line) in accordance with their molar contribution to the derivatized compounds:

ncdδ13Ccd ) ncδ13Cc + ndδ13Cd

(3)

where n is the number of moles of carbon and the subscript c is the compound of interest, d is the derivative group, and cd is the derivatized compound. The KIE for both derivatization procedures was then calculated using the following equation:

KIE ) 1 + ∆ncd/1000x

(4)

where ∆ is the difference between the measured and predicted δ13C values of the derivatized amino acids, ncd is the total number of carbon atoms in the derivatized compound, and x is the number of functional groups available for acylation. Correction Factors. The occurrence of a KIE precludes the calculation of amino acid δ13C values using stoichiometric mass balance equations. Thus, empirical correction factors were employed to obtain the “effective” δ13C value of the derivative carbon. These correction factors were derived by determining the δ13C values of both the underivatized amino acid standards (off-line via EA-IRMS) and derivatized amino acid standards (on-line via GC/C/IRMS) and then rearranging eq 3 to determine δ13Cd (as in refs 5 and 21). The δ13Cd term can then be replaced with δ13Ccorr.fact, representing the correction factor for the derivatized amino acid (ndδ13Ccorr.fact ) ncdδ13Ccd - ncδ13Cc). Subsequently, δ13C values of individual amino acids from natural materials are

calculated by again rearranging eq 3 to determine δ13Cc, adopting the appropriate value of δ13Ccorr.fact (ncδ13Cc ) ndδ13Ccorr.fact ncdδ13Ccd). Analytical Errors. The errors associated with amino acid δ13C values are higher than those associated with other derivatized compounds, e.g., fatty acid methyl esters, because o: (i) their low sample-to-derivative carbon ratio and (ii) the uncertainty associated with the empirical correction factors used to account for the KIE during derivatization.21 These combined errors propagate as follows, where σ is the standard deviation associated with a given δ13C value, s denotes the standard used for determining correction factors, and sd denotes the derivatized standard:

()

σ2c ) σ2s

ns nc

2

(

+ σ2sd

)

ns + n d nc

2

(

+ σ2cd

)

nc + n d nc

2

(5)

RESULTS AND DISCUSSION Mass Spectral Identification. Amino acid NACME and NAIP esters were identified using GC/MS. Mass spectral identifications were fairly straightforward with cleavage at esterified carboxyl sites and acylated amine and hydroxyl sites affording prominent fragment ions. Additional characteristic fragment ions in the EI spectra of NACME derivatives included the loss of the methyl group attached to the carboxyl moiety, [M - 15]+, and loss of the acetyl group on OH and NH groups, to [M - 43]+, and loss of acetyl-derived ketene to [M - 42]+. In addition, the M+· was typically visible for both NACME and NAIP derivatives. Optimization of N-Acetyl Methyl Ester Protocol. The initial acetylation method used for the preparation of NACME esters was modified from a procedure described by Reid et al.27 This involved dissolving the amino acid methyl esters in ammonium bicarbonate (pH 7.8), followed by acetylation with acetic anhydride/ anhydrous methanol (1/3, v/v, 2 h, 25 °C). Since the latter procedure for the preparation of NACME esters was not optimized for subsequent GC analysis, in our modification, we added more thorough purification steps. Although a clean GC trace then resulted, no peaks were observed for threonine, serine, hydroxyproline, tyrosine, and lysine. Each of these amino acids contains an extra functional group (OH or NH) that requires acylation, indicating the conditions described were insufficient for derivatization of all functional groups. Thus, we modified the acetylation procedure by employing a mixture of acetic anhydride, triethylamine, and acetone (1/2/5, v/v/v, 10 min, 60 °C). Under these conditions, GC peaks and improved yields were observed for all 15 amino acids. Methylation of carboxylic acids can be achieved at either room temperature27 or at elevated temperatures, for example, 70 °C.28 The original methylation procedure for NACMES was performed at room temperature; however, it is possible that a roomtemperature derivatization reaction may not be reproducible, which is a prerequisite for obtaining reliable δ13C values. It is possible also that amino acid yields may differ depending on the methylation temperature. In order to probe this, amino acid standard mixtures were converted to NACME esters (by addition (27) Reid, G. E.; Simpson, R. J.; O’Hair, R. A. J. J. Am. Soc. Mass Spectrom. 1998, 9, 945-956. (28) Islam, A.; Darbre, A. J. Chromatogr. 1972, 71, 223-232.

Figure 1. (a) Influence of methylation conditions (room temperature (RT) for 2 h, 70 °C for 2 h and 70 °C for 1 h) on amino acid NACME ester yields, and (b) comparison of derivatization yields for NACME and NAIP derivatives. Derivatization yields are expressed relative to the GC peak area of the external standard n-nonadecane.

of 1 mL of acidified methanol solution (1.85 M; 1 h, 70 °C)), using different methylation conditions: (i) room temperature for 2 h, (ii) 70 °C for 2 h, and (iii) 70 °C for 1 h. Reaction yields were determined using GC and reported as a ratio to an external underivatized standard of n-nonadecane (Figure 1). Reaction yields increased by as much as 7-fold when the temperature was increased from room temperature to 70 °C (Figure 1a). Although increased yields were highest for isoleucine (×7.4), lysine (×7.1), and valine (×5.6) at 70 °C, increases were observed for all amino acids. In addition, average peak areas doubled when the methylation reaction time at 70 °C was reduced from 2 to 1 h (with the exception of tyrosine), particularly for glycine (×6.6), alanine (×2.6), and proline (×3.0). GC/C/IRMS analysis was subsequently performed on these samples, and the methylation temperature was observed to have a significant affect on amino acid δ13C values. When methylation was performed at room temperature, δ13C values for both the L- and D,L-amino acid standard mixtures were more negative compared to at 70 °C (by 2.4 ( 1.5‰) indicating discrimination against the heavier 13C isotope, Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

9085

Figure 2. GC/C/IRMS chromatograms of a D/L-amino acid standard mixture converted to (a) NACME and (b) NAIP derivatives. Analyses were performed on a polar ZB-WAX-coated capillary column (stationery phase, PEG). D,L-Hydroxyproline is not shown as it was unobtainable.

possibly arising from incomplete methylation. Interestingly, the largest offsets were observed for aspartic acid (4.0 ( 0.5‰) and glutamic acid (4.4 ( 0.4‰), the two amino acids containing a second carboxyl group requiring methylation. Number and Yield of Amino Acids. Two standard mixtures (of either D,L- or L-amino acids) comprising 15 protein amino acids were derivatized as either NAIP or NACME esters and analyzed by GC. Chromatographic peaks for the 15 amino acids and internal standard (norleucine) were observed following preparation of both the NACME and NAIP derivatives (Figure 2a and b, respectively). Both procedures produced GC chromatograms lacking extraneous peaks, indicating an absence of significant byproduct formation or incomplete derivatization with either procedure. A rigorous cleanup procedure was performed for both derivatization methods, such that reagents were thoroughly removed using repeated evaporations over ice and via phase separation. Surprisingly, we believe this to be a novel derivatization of amino acids to NACME esters for GC analysis. Derivatization Yield. Following optimization of the NACME derivatization procedure (i.e., 70 °C methylation, 1 h), the reaction yield of each amino acid for both the NACME and NAIP 9086 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

derivatives was quantified (Figure 1b). Maximum derivatization reaction yields are essential for the application of GC/C/IRMS analysis to materials containing low concentrations of amino acids, such as sediments and poorly preserved archaeological skeletal materials. In order to determine whether differences in absolute yields were achieved for the two procedures, standard mixtures containing 50 µg of each of the 15 amino acids and norleucine were derivatized using both methods and quantified using GC with a coinjection of 40 ng of an underivatized “external” standard, n-nonadecane. The resulting “response” ratios are governed by a combination of the derivatization yield and the GC-FID response for each amino acid and each derivatization procedure. Typically, higher response ratios were observed for aliphatic carbon-rich amino acids, e.g., leucine and isoleucine, while relatively poorer responses were associated with the amino acids containing an extra hydroxyl group, such as threonine, serine, and tyrosine. Although differences were observed between amino acids, the mean response for each of the 15 amino acids relative to the n-nonadecane standard was identical for both NAIP and NACME esters (amino acid derivative/n-nonadecane ) 1.2; Figure 1b). For several amino acids, little difference was observed between the two methods; however, a superior response was observed for the NAIP procedure for alanine, glutamate, and methionine, while the yields achieved for proline and phenylalanine were considerably higher for the NACME procedure. The appreciable yields and favorable GC-FID responses observed for the NACME derivative was very encouraging given the problems anticipated in preparing and handling these low molecular weight derivatives. Poor reaction yields have been reported previously for TFA-methyl esters and were explained by extensive evaporative losses during their preparation.29 From the results obtained herein, it was confirmed that such problems can be overcome by undertaking evaporations in an ice bath using extremely gentle streams of N2. Chromatographic Performance. There is a chromatographic isotope effect associated with δ13C value determinations using GC/ C/IRMS, where the 13C-containing molecules of a given analyte precede those of the 12C-containing molecules by ∼150 ms, resulting in an s-shaped 45/44 signal.3 Thus, for the procurement of reliable δ13C values, it is critical that (i) the whole GC peak is integrated and (ii) baseline resolution from adjacent amino acid peaks is achieved. Metges and Daenzer employed NPIP derivatives to replace NANP for determining δ13C values of amino acids using GC/C/IRMS, because of their enhanced GC resolution compared with the latter.18 The pivaloylated derivatives afforded baseline resolution of alanine and glycine, isoleucine and leucine, aspartate and methionine, threonine and serine, and lysine and histidine, which were previously partially or fully coeluting as their NANP esters. An alternative solution to such coelution problems is to employ a GC column coated with a higher-polarity stationary phase. Similar to Metges and Daenzer,18 we observed the combined problems of extreme tailing and coelution on both nonpolar (ZB1; Phenomenex Zebron) and low-polarity GC columns (Ultra-2; J & W Scientific) for both the NACME and NANP derivatives. However, by switching to high-polarity cyanopropylphenylsubstituted (VF-23ms; Varian) or PEG stationary phases (ZB-FFAP and ZB-WAX; Phenomenex Zebron), Gaussian peak shapes were (29) Darbre, A.; Blau, K. J. Chromatogr. 1965, 17, 31-&.

observed and baseline resolution was achieved for the 15 amino acids as both NACME and NAIP derivatives on ZB-WAX- and VF23ms-coated GC columns, respectively. The excellent chromatographic performance obtained for GC/C/IRMS analyses of both NACME and NAIP derivatives on the ZB-WAX column is illustrated in Figure 2a and b, respectively. Stability. In order to assess the stability of NACME and NAIP esters, amino acid standard mixtures were derivatized and GC analysis was performed weekly over a 12-week period during which they were stored below -5 °C. Neither the absolute concentrations (measured relative to coinjected external standard, n-nonadecane) nor the relative abundances of the various amino acids (determined relative to the derivatized internal standard, norleucine) decreased for either derivative. Following this initial storage period, both derivatives were periodically reanalyzed over several months and the derivatives found to be stable. Thus, despite a slight decrease in overall abundance on prolonged storage, the ratio of alanine to phenylalanine (stable and unstable amino acids, respectively) only changed from 0.83 (day 1) to 0.82 (day 287) for NACME esters and 0.56 (day 1) to 0.52 (day 287) for NAIP esters. We did not determine if there was any change in δ13C values during this extended period; thus, we do not recommend storing either derivative for longer than 12 weeks prior to GC/C/IRMS analysis. KIE and Reproducibility. The association of a KIE with the acylation step of NACME and NAIP derivatization procedures was confirmed by the failure to predict underivatized amino acid δ13C values using mass balance equations (mean KIENACME ) 1.036; mean KIENAIP ) 1.038). In order to achieve the most reproducible KIEs and thus reproducible δ13C values, a number of precautions were taken throughout this work. For example, to ensure both derivatization reactions were complete and reproducible, a considerable excess of both acylation and esterification reagents was always employed for both procedures. Since NACME derivatives have not been previously employed for GC/C/IRMS analysis, several steps were taken here to assess their reproducibility. First, three batches of five replicates (15 replicates in total) of a standard mixture of D,L-amino acids were converted to NACME and NAIP derivatives on three consecutive days and δ13C values determined using GC/C/IRMS. The results are displayed in Figure 3, where it can be seen that the three batches produced very similar δ13C values for both NACME (a) and NAIP (b) derivatives (mean STDV ) 0.3 and 0.4‰, respectively), confirming the reproducibility of the technique when identical conditions are employed. Similarly, reproducible results were then observed for both NACME and NAIP derivatives of a mixture of L-amino acids with very different δ13C values. Since the same reagents and conditions were employed for both the Land D,L-mixtures of amino acids, the reproducibility of the two techniques was further confirmed by calculating the KIEs associated with both mixtures. Despite differences in δ13C values of up to 20‰ between the components of the mixtures of D,L- and L-amino acids, both in derivatized and underivatized forms, almost identical KIEs were calculated for each amino acid for both the NACME (Figure 4a) and NAIP derivatives. Robustness of Derivatization Procedures. The KIEs encountered during acylation reactions result from the nonquantitative reaction of the reagents; i.e., the carbon atom involved in

Figure 3. Reproducibility of amino acid standard mixture δ13C values (uncorrected) following conversion to (a) NACME and (b) NAIP esters. Rep 1, 2, and 3 represent the mean δ13C value of 3 batches of 5 discrete amino acid standard mixtures derivatized on consecutive days. Error bars represent the standard deviation of the 5 replicates within each batch.

the rate-determining step of the reaction is in the derivatizing reagent and not in the compound undergoing derivatization. Hence, it was necessary to explore the possibility that the expression or reproducibility of the KIE may not be constant if different amounts of analyte are derivatized. A range of concentrations between 5 and 2500 µg of a standard mixture of 15 protein amino acids were derivatized using the same volume of reagents, followed by GC analysis. A strong relationship was observed between the amount of amino acid derivatized and the resulting GC peak area for all amino acids, e.g., leucine NACME ester (Figure 4b), which confirms the robustness of both the reaction and GC-FID response for these analytes and reagent concentrations. The robustness of the reaction for stable carbon isotope analysis was then examined by analyzing each concentration of the standard mixture using GC/C/IRMS. For both derivatization procedures, reproducible δ13C values were obtained within the 5-2500 µg concentration range. Figure 5 illustrates the consistency of the δ13C values for both the NACME (a) and NAIP (b) derivatives of alanine (a simple carbon-poor amino acid) and methionine (a more carbon-rich amino acid possessing a thiol group also requiring acetylation) over the range of concentrations. The maximum standard deviation observed, i.e., 0.2‰, is lower than the instrument precision (0.3‰). Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

9087

Figure 4. (a) Comparison of δ13CDerivatized amino acid values and calculated KIE of a D,L- and L-amino acid standard mixture (with differing δ13C values) derivatized to NACME esters, and (b) relationship between concentration of leucine derivatized to NACME ester and GC peak area.

Figure 5. Variation in δ13C values (uncorrected) following derivatization to (a) NACME and (b) NAIP esters with varying amounts (52500 µg) of each amino acid; alanine and methionine shown. The dashed lines represent the mean δ13C value calculated for the range of 5-2500-µg derivatization amounts. Error bars represent the standard deviation of triplicate GC/C/IRMS determinations.

Sample-to-Derivative Carbon Ratio and Isotope Precision. For δ13C value determinations using GC/C/IRMS, it is essential that sample-to-derivative carbon ratios be as high as possible because each additional derivative carbon atom increases the imprecision associated with determined δ13C values. This is particularly important for amino acids that are relatively low molecular weight polyfunctional compounds containing between 2 (glycine) and 11 (tryptophan) carbon atoms, in addition to between 2 and 4 polar functional groups requiring derivatization. Previously, TFA-IP, NANP, or NAIP derivatives have been widely used for δ13C value determinations at natural abundance with sample-to-derivative carbon ratios ranging from 0.4 to 2.2, resulting in associated errors of between 0.7 and 1.5‰. Because of the higher sample-to-derivative carbon ratios of the NACME derivatives (0.6-3.0), the errors associated with carbon isotope determinations are reduced substantially to between 0.6 and 1.1‰. This inverse relationship between sample-to-derivative carbon ratio and errors associated with δ13C value determinations is illustrated in

Figure 6. Thus, the NACME derivative, which introduces as few as three additional carbon atoms, provides the highest precision of all available amino acid derivatives for δ13C value determinations using GC/C/IRMS. Correction Factors. Higher correction factors were observed for the NACME (-49.4 ( 4.5‰) than the NAIP (-41.1 ( 3.1‰) derivatives since these are governed by the carbon loads of the derivative group, with the former derivative introducing fewer carbon atoms (NACME: +3C to +5C; NAIP: +5C to +8C), thus displaying the higher correction factors. This is due to the fact that the KIE results in an extremely low “effective” acetylating carbon δ13C value and the NACME derivative has less remaining carbon atoms to offset it due to its constituent methyl (+1C) rather than propyl group (+3C). The next step was to determine the accuracy with which these correction factors could predict the δ13C values of amino acids of unknown isotopic composition, i.e., sample amino acids. Hence, a D,L- and an L-amino acid standard mixture of known isotopic

9088 Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

Figure 6. Comparison of associated errors (calculated according to eq 5) for the NACME (black bars) and NAIP derivatives (gray bars) showing superior performance of the former due to inherently higher analyte-to-derivative carbon ratio (diamonds and dotted lines).

Figure 8. (a) δ13C values of NACME esters of diet and bone collagen amino acids for rats (n ) 2) reared on pure C3 diet or pure C4 diets (* denotes essential amino acid) and (b) relationship between measured rat bone collagen δ13C values and those calculated from constituent amino acid δ13C values via mass balance equations. The dashed line represents the theoretical x ) y line. Figure 7. Comparison of δ13C values of the L-amino acid mixture (measured off-line using EA-IRMS) with those predicted from correction factors calculated from the comparison of a derivatized (measured on-line using GC/C/IRMS) and underivatized D/L-amino acid mixture. Error bars represent combined error of additional carbon atoms and uncertainty associated with the correction factors employed to correct for the KIE associated with acetylation.

composition (measured off-line using EA-IRMS) were derivatized; correction factors were calculated from the D,L-mixture, which were then used to predict the original, underivatized δ13C values of the L-mixture. A high degree of accuracy was obtained from the correction factors, where the L-mixture δ13C values measured using EA-IRMS and those predicted from the correction factors were accurate to within their associated precisions (Figure 7). This result also highlights the reproducibility of the NACME derivatization procedure and lack of unaccounted for systematic errors. NACME Derivatization of Rat Collagen Amino Acids. Since amino acid NACME derivatives have not been previously em-

ployed for carbon isotope analyses using GC/C/IRMS, it was important to demonstrate that δ13C values obtained for natural abundance determinations were meaningful and precise. Thus, we obtained tissue (bone collagen) and diet samples from a carbon isotope controlled-feeding experiment in which rats were reared on either C3 or C4 diets.22 The compound-specific δ13C values of the amino acid components of these samples have been determined previously as their TFA-IP derivatives.30 By focusing on these previously analyzed samples, the integrity of δ13C values of the NACME derivatives could be accurately assessed. Figure 8a confirms the validity of the NACME derivatization procedure since the C3 and C4 diet, and bone rat collagen amino acid δ13C values are in the anticipated range and are consistently offset (∆13CC4-C3 Diet amino acids ) 11.8 ( 1.2‰; ∆13CC4-C3 Bone collagen amino acids ) 11.0 ( 1.2‰). All δ13C values determined here for the constituent amino acids as their NACME derivatives are very (30) Jim, S.; Jones, V.; Ambrose, S. H.; Evershed, R. P. Br. J. Nutr. 2006, 95, 1055-1062.

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

9089

similar to those previously determined as TFA-IP derivatives.30 As a further assessment of the integrity of this derivatization procedure, mass balance equations, utilizing the individual bone collagen amino acid δ13C values and their known carbon contribution to collagen, were employed to estimate bulk bone collagen δ13C values. Figure 8b shows the relationship between directly determined bone collagen δ13C values (via EA-IRMS) and those calculated from the mass balance equations. An extremely high level of precision was achieved (∆13CMeasured-Calculated ) 0.3 ( 0.4‰), which results from the fact that the NACME derivatization procedure allowed the analysis of 15 amino acids compared to the 12 previously analyzed using TFA-IP esters. This is at least partially attributable to the employment of a polar GC stationary phase in this work, which allowed extension of δ13C value determinations to include methionine, tyrosine, and lysine, thereby allowing compound-specific determinations of δ13C values of 89.1% of collagen carbon compared to 83.8%, as determined previously.30 In the previous analyses using TFA-IP esters, the higher and more negative mean ∆13CMeasured-Calculated values (-1.3 ( 0.6‰)30 resulted largely from the relatively low δ13C value of the undetermined tyrosine and lysine, which together contribute 5.5% of the carbon to collagen. These results both confirm the integrity of amino acid NACME derivative δ13C values and highlight the chromatographic benefits of switching to more polar GC stationary phases for amino acid analyses. CONCLUSIONS In this paper, we have described a novel amino acid derivative for carbon isotope determinations employing GC/C/IRMS. The technique involved methylation of carboxyl groups with acidified

9090

Analytical Chemistry, Vol. 79, No. 23, December 1, 2007

methanol and acetylation of amine, hydroxyl, and thiol groups with acetic anhydride, forming NACME esters. Following much development and refinement of both the derivatization and analytical procedures, we have demonstrated that NACME esters achieve a performance similar to the more commonly employed NAIP esters in several respects, but were advantageous in their lower analytical errors. Currently, further development is being performed in our laboratory to confirm their suitability for nitrogen isotope measurements of amino acids using GC/C/IRMS. Although errors associated with δ15N value determinations are not governed by the carbon weight of the derivative group in nitrogen isotope analysis, it is our assertion that minimizing derivative carbon addition will be beneficial to the performance of the GC/ C/IRMS combustion interface. ACKNOWLEDGMENT We thank Professor Stanley Ambrose at the University of Illinois for provision of diet and rat tissue samples. Rat feeding experiments were supported by the National Science Foundation, USA (BNS 9010937 and SBR 9212466) and the University of Illinois Research Fund. We thank the Wellcome Trust for providing the Bioarchaeology Fellowship to L.T.C. for this research. NERC is thanked for mass spectrometry facilities (GR3/ 2951, GR3/3758, and FG6/36101). We thank Dr. Ian Bull (University of Bristol) for assistance with GC/C/IRMS analyses. An anonymous reviewer is thanked for his/her helpful comments. Received for review June 11, 2007. Accepted August 21, 2007. AC071223B