Anal. Chem. 2001, 73, 799-802
Reduction of Nonpolar Amino Acids to Amino Alcohols To Enhance Volatility for High-Precision Isotopic Analysis Bassem I. Zaideh, Nabil M. R. Saad, Betty A. Lewis, and J. Thomas Brenna*
Division of Nutritional Sciences, Cornell University, Savage Hall, Ithaca, New York 14850
Amino acids are routinely derivatized using carboncontaining groups prior to gas chromatography continuous-flow isotope ratio mass spectrometry (GCC-IRMS). Derivative C contaminates analyte C because the entire derivatized compound is combusted to CO2. Correction procedures are required to extract the analyte isotope ratio. We present a method for reduction of six nonpolar amino acids to their corresponding amino alcohols, demonstrate a GC strategy to produce acceptable peak shapes from the resulting strongly H-bonding analytes, and present isotopic analysis for amino acids and their corresponding amino alcohols to evaluate any possible isotopic fractionation. Alanine, valine, leucine, isoleucine, methionine, or phenylalanine was reduced using NaBH4 in THF with I2 as an electrophile. Reactions were performed with 2 g of analyte to permit isotopic analysis by conventional elemental analysis-IRMS. All reactions were quantitative as assessed by IR spectra, melting points, and GC. Recovery from the reaction mixture was 60-84%. GC separation of a mixture of the six amino alcohols was achieved using a thick stationary-phase (5 µm) capillary column to avoid tailing due to hydrogen bonding to the walls of the fused-silica capillary. The reproducibility of GCC-IRMS determinations of amino alcohols averaged SD(δ13C) ) 0.25 ( 0.19‰. The absolute differences between δ13C of amino acids measured by an elemental analyzer coupled to IRMS and amino alcohols measured by GCC-IRMS was ∆δ13C ) 0.14‰ and showed no general trend. Reactions performed with 2 mg of analyte yielded equivalent chromatograms. These data indicate that the reduction method does not induce isotopic fractionation and can be used for continuous-flow isotopic analysis to avoid addition of contaminating carbon. High-precision carbon isotope analysis of organic compounds is an important analytical technique for determination of natural variability and for tracer applications in fields as diverse as geochemistry and biomedicine. Amino acids have long been a common subject of high-precision isotope analysis in biomedicine for determination of protein metabolism. For example, synthesis rates of proteins are determined in human proteins such as * Corresponding author: (fax) (607) 255 1033; (phone) (607) 255 9128; (email)
[email protected]. 10.1021/ac000604l CCC: $20.00 Published on Web 01/19/2001
© 2001 American Chemical Society
fibrinogen1 and LDL-apo B1002 and in the skeletal muscle of growing piglets.3 In geochemistry, natural variability in amino acid isotope ratio is used as a guide to the origin of amino acids in fossils4 and in extraterrestrial objects such as meteorites.5 Over the past decade, the coupling of gas chromatography (GC) to isotope ratio mass spectrometry (IRMS) using various stages of microchemistry has dramatically improved the sensitivity and speed of high-precision compound-specific isotope analysis for C.6 On-line microchemical steps quantitatively convert the purified analyte emerging from the GC column into an IRMS analysis gas, CO2 in the case of C, and prepare it for on-line admission into an IRMS. Though interfaces between liquid chromatography and IRMS have been introduced,7-9 they are not yet available commercially. Thus, the only alternative for on-line isotopic analysis is GC, which requires volatile analytes. Amino acids are a multifunctional, heterogeneous group of low molecular mass chemical compounds that have at least one carboxyl and one amino group, an attribute that makes them nonvolatile and unsuitable for GC. The conventional means to overcome the nonvolatility limitation of amino acids for continuousflow isotope analysis is to replace the H of the carboxyl group and amino group by a moiety that lowers the polarity of the amino acid and minimizes H-bonding, thus improving volatility.10 At least eight different derivatization schemes for amino acids are in use.11-17 Derivatization adds extraneous C to amino acids, from 4 mol of C for the N-perfluoroacyl-1-propyl group12 to 8 mol of C (1) Velden, M. G.; Kaysen, G. A.; de Meer, K.; Stellaard, F.; Voorbij, H. A.; Reijngoud, D. J.; Rabelink, T. J.; Koomans, H. A. Kidney Int. 1998, 53, 1818. (2) Velden, M. G.; Rabelink, T. J.; Gadellaa, M. M.; Elzinga, H.; Reijngoud, D. J.; Kuipers, F.; Stellaard, F. Anal. Biochem. 1998, 265, 308-12. (3) Reijngoud, D. J.; Hellstern, G.; Elzinga, H.; de Sain-van der Velden, M. G.; Okken, A.; Stellaard, F. J. Mass Spectrom. 1998, 33, 621-6. (4) Engel, M. H.; Goodfriend, G. A.; Qian, Y.; Macko, S. A. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 10475-8. (5) Engel, M. H.; Macko, S. A. Nature 1997, 389, 265-8. (6) Brenna, J. T.; Corso, T. N.; Tobias, H. J.; Caimi, R. J. Mass Spectrom. Rev. 1997, 16, 227-58. (7) Caimi, R. J.; Brenna, J. T. Anal. Chem. 1993, 65, 3497-500. (8) Caimi, R. J.; Brenna, J. T. J. Mass Spectrom. 1995, 30, 466-72. (9) A., B. W.; P., D. Isot. Environ. Health Stud. 1996, 32, 275-83. (10) Klee, M. S. Modern Practice of Gas Chromatography, 2nd ed.: New York, 1985. (11) Holme, D. J.; Peck, H. Analytical Biochemistry; Longman: London, New York, 1983. (12) Iwase, H.; Murai, A. Chem. Pharm. Bull. (Tokyo) 1974, 22, 1455-8. (13) Metges, C. C.; Petzke, K. J. Anal. Biochem. 1997, 247, 158-64. (14) Pietzsch, J.; Julius, U.; Hanefeld, M. Rapid Commun. Mass Spectrom. 1997, 11, 1835-8.
Analytical Chemistry, Vol. 73, No. 4, February 15, 2001 799
for the recently described N-pivaloylsopropyl (NPP) group13 per mole of amino acid. Since amino acids range from 2 to 11 mol of C/mol of amino acid for glycine and tryptophan, respectively, extraneous derivative C contributes a substantial amount of all C in the analyzed product. Carbon isotope analysis of derivatized compounds in gas chromatography continuous-flow (GCC)-IRMS presents two problems. First, analytes are combusted immediately after emerging from the GC column to produce CO2 analysis gas. Extraneous C from the derivative group is indistinguishable from analyte C in the IRMS, and the δ13C of the analyte must be calculated using the mass balance equation6 after an independent assessment of the C isotope contribution from the derivative C. In addition to possible biases in determination of the derivative C, imprecision in the final result is unavoidable due to propagation of errors. In general, this imprecision is larger when the contribution of derivative carbon to the derivatized compound is a large fraction of the derivatized analyte, which is usually the case in compounds with less than 10 carbons, such as amino acids. A second issue is the possibility of kinetic isotope fractionation of the analyte if the reaction is not quantitative and if the procedure involves analyte C bonds.18 This leads to bias that may not be detected with standards. An alternative strategy is to modify the chemical properties of the analyte without introducing extraneous C, which was implemented successfully for conversion of fatty acid methyl esters to fatty alcohols.19 An analogous approach is to convert amino acids to amino alcohols by chemical reduction of the carboxyl group. This approach is most likely to succeed for nonpolar amino acids since most homologous amino alcohols are sufficiently volatile for GC analysis. A number of methods have been reported for chemical reduction of amino alcohols;20-23 however, most suffer from one or more disadvantages such as excessive refluxing time and expensive or relatively toxic reagents. A recent approach for direct reduction of amino acids employs NaBH4 as the reducing agent and an electrophilic catalyst such as H2SO421 or I2.22 We chose the NaBH4/I2 procedure because it is a one-step reaction using inexpensive reagents.22 In this paper, we report an evaluation of this strategy for reduction of nonpolar amino acids to their corresponding amino alcohols, demonstrate a suitable GC method to produce sharp peak shapes from amino alcohols, and evaluate any isotopic effects of the procedure against bulk analysis of the amino acids before and after treatment. Data are reported for isotopic analysis of 2 g of analyte, and the chemical and chromatography results were verified for 2 mg of analyte. (15) Yarasheski, K. E.; Smith, K.; Rennie, M. J.; Bier, D. M. Biol. Mass Spectrom. 1992, 21, 486-90. (16) Calder, A. G.; Garden, K. E.; Anderson, S. E.; Lobley, G. E. Rapid Commun. Mass Spectrom. 1999, 13, 2080-3. (17) Chen, Z.; Landman, P.; Colmer, T. D.; Adams, M. A. Anal. Biochem. 1998, 259, 203-11. (18) Rieley, G. Analyst 1994, 119, 915-9. (19) Corso, T. N.; Lewis, B. A.; Brenna, J. T. Anal. Chem. 1998, 70, 3752-6. (20) Kanth, J. V. B.; Periasamy, M. J. Org. Chem. 1991, 56, 5964-5. (21) Abiko, A.; Masamune, S. Tetrahedron Lett. 1992, 33, 5517-8. (22) McKennon, M. J.; Meyers, A. I.; Drauz, K.; Schwarm, M. J. Org. Chem. 1993, 58, 3568-571. (23) Anand, R. C.; Vimal Tetrahedron Lett. 1998, 39, 917-8.
800
Analytical Chemistry, Vol. 73, No. 4, February 15, 2001
EXPERIMENTAL SECTION Amino Acids, Solvents, and Reagents. All amino acids were analytical grade, >98% purity and were obtained from Sigma Chemical Co. (St. Louis, MO). Ethanol, methanol, and methylene chloride of highest available purity and technical grade dichloromethane (DCM) were obtained from Fisher Scientific Co. (Fair Lawn, NJ). Tetrahydrofuran (THF) and NaBH4 were from Aldrich Chemical Co. (Milwaukee, WI); I2 was from Mallinckrodt Chemical Works. Reduction of Amino Acids. Alanine (Ala), valine (Val), leucine (Leu), isoleucine (Ile), methionine (Met), and phenylalanine (Phe) were tested for reduction by the NaBH4/I2 method. Two grams of each was reduced to yield ample product for isotopic analysis by elemental analyzer-IRMS (EA-IRMS), which requires milligram quantities of sample per analysis, and to conveniently assess purity by scanning infrared spectroscopy (IR). Amino acids were reduced as described previously22 with slight modification. Amino acid (2.0 g) was added to a stirred suspension of NaBH4 (2.5-fold molar excess compared to amino acids) in THF (50 mL). The flask was immersed in an ice bath, and a solution of I2 (equimolar with amino acid) in 10 mL of THF was added dropwise over a period of 15 min to maintain the temperature below 20 °C, resulting in vigorous evolution of H2. After addition of the I2 was complete and gas evolution had ceased, the reaction mixture was refluxed for 15 h and then cooled to room temperature. Methanol was added carefully until the mixture became clear to exhaust excess NaBH4. After 45 min, the solvent was evaporated under an N2 stream leaving a white paste, which was subsequently dissolved by addition of 33 mL of 20% (w/w) KOH. The solution was stirred for 4 h. After three extractions with 50 mL of methylene chloride, the organic extract was dried over sodium sulfate and the solvent was removed by rotary evaporation. To verify that the reaction could be scaled down, the reaction was repeated with six amino acids (Ala, Val, Leu, Ile, Met, Phe) at 2 mg each. The purity of the extracted amino alcohols was evaluated by infrared spectroscopy. Yields were determined by weighing the amino acids before reduction and amino alcohols after extraction and correcting for molecular weights. GCC-IRMS. The on-line isotopic analysis system consisted of a Hewlett-Packard 5890 GC equipped with flame ionization detector (FID) and coupled to a Finnigan-MAT 252 (Bremen, Germany) high-sensitivity, high-precision IRMS via an in-housebuilt combustion/water trap/open split interface. The combustion furnace is made of deactivated fused silica filled with oxidized Cu wire and Pt catalyst and held at 850 °C. The water trap was of the Nafion type (DuPont, Wilmington, DE), which eliminates H2O while retaining CO2. Amino alcohols dissolved in water were injected in split mode onto a CP-sil 5 CB fused-silica capillary column (50 m × 0.53 mm; 5-µm film thickness) (Chrompack Inc.) with He carrier gas. The oven temperature program was increased from 100 to 150 °C at 10 °C/min, held at that temperature for 2 min, increased at 10 °C/min to 250 °C, and finally held at that temperature for 10 min. The injector and detector temperatures were 240 and 250 °C, respectively. After separation, the amino alcohols were either sent to a FID for methods development or to the combustion interface via a rotary valve (Valco Instruments, Houston, TX) for isotope analysis. High-precision isotope analyses are expressed in δ13C notation, which is defined as
δ13C )
[
]
RSPL - RPDB × 103 RPDB
where R is the [13C]/[12C] ratio and SPL and PDB refer to sample
Table 1. Reduction Data for Six Nonpolar Amino Acids
amino acid
amino alcohol
valine alanine leucine isoleucine methionine phenylalanine
valinol alaninol leucinol isoleucinol methioninol phenyalaninol
%
yielda 65 60 84 75 60 60
alcohol IR C-O (cm-1) expb lit.c 1055.02 1055.31 1053.63 1053.63 1058.12 1065.38
1054.4 1059.6 1059.1 1073.3 1056.8 1065.0
alcohol physical state at 25 °C liquid liquid liquid liquid liquid solid
a Extraction yield; extent of reduction is expected to be nearly quantitative. b Experimentally determined maximums of CO stretch. c Literature value of CO stretch maximums.25
and international standard Pee Dee Belemnite, respectively; RPDB ) 0.011 237 2. Isotope ratios were determined for amino acids prior to reduction by an elemental analyzer (EA) coupled to a Europa Scientific 20/20 IRMS (Crewe, U.K.). Samples (1-3 mg) of amino acid were placed into tin cups, crimped, and loaded into the autosampler of the EA. Amino alcohols were loaded onto an inert support (Chromosorb W, 30-60 mesh) in tin cups and placed in the autosampler. Amino alcohols are sufficiently volatile for GCCIRMS analysis and were analyzed by that method as well. RESULTS AND DISCUSSION Derivatization and Chromatography. Table 1 summarizes the amino acids, their corresponding amino alcohols, the yields, and the major IR data and the physical state of the products obtained from the extraction solution. The IR stretches of the CO groups are in good agreement with the literature data and all melting points. Yields range from 60 to 84%, and after evaporation of solvent, the amino alcohols were all colorless products. Losses are likely due to incomplete extraction rather than incomplete reaction. Figure 1 shows the analysis of leucinol on two capillary columns differing only in the thickness of the stationary phase. Figure 1A demonstrates that the leucinol peak on a conventional 0.25-µm-thick phase produces a peak that is more than 100 s wide with a tail that does not return to baseline for >3 min. This tailing for strongly H-bonding compounds such as amino alcohols is well known and is due to H-bonding of analyte to the walls of the fusedsilica capillary column. Figure 1B shows the chromatographic behavior of leucinol on an equivalent column; the stationary-phase thickness for this column was 5 µm. In this case, the peak elutes ∼1 min earlier and it is 5 s wide (fwhh) and suitable for isotopic analysis and quantitative analysis from a mixture. The thick stationary phase masks the walls of the fused silica so that the analyte does not reach the walls of the capillary. Two alternative approaches are sometimes available for analysis of strongly H-bonding analytes. One is use of polar stationary phases. We did not use this approach since the stability of columns tends to decrease with increasing polarity of the stationary phase,24 leading to low maximum operating temperature and resulting in long analysis times. An alternative approach is to use a deactivated fused-silica column, where the walls are treated chemically to eliminate H-bonding; in other words, the walls of the column are derivatized rather than the analyte. A known disadvantage of this approach is that these columns tend to be sensitive to contamination.24 In our hands, this approach did not significantly improve
the chromatography, possibly due to some unknown nonvolatile contaminant. Figure 2 shows the GC analysis of the extraction mixtures for each of the six amino alcohols on the thick-phase column. All peak shapes are sharp and the chromatograms indicate that there are few contaminants to interfere with GC analysis. Retention times shown in the figure are sufficiently different under these chromatographic conditions to permit baseline separation of all six amino alcohols from one another. Equivalent chromatograms (not presented) were obtained for the scaled-down reactions that started with 2 mg of amino acid. A chromatogram of a mixture of the six amino alcohols under the same conditions, also not presented, produced baseline resolution suitable for isotopic analysis for all analytes. Isotopic Analysis. The results of isotopic analyses of amino acids and corresponding amino alcohols measured by EA-IRMS and also by GCC-IRMS are presented in Table 2. The absolute
(24) Grob, R. L. Modern practice of gas chromatography, 2nd ed.; Wiley: New York, 1985.
(25) Pouchert, C., J. The Aldrich library of FT- IR Spectra; Aldrich Chemical Co.: Milwaukee, WI, 1985.
Figure 1. Chromatography according to conditions discussed in the text of leucinol using (a) a 0.25- and (b) a 5-µm stationary phase. The thin film leads to dramatic broadening of the peak due to H-bonding to the walls of the fused-silica capillary, rendering isotopic analysis impossible. The thicker film obviates this problem.
Analytical Chemistry, Vol. 73, No. 4, February 15, 2001
801
Table 2. Isotope Ratios for Amino Acids Measured by EA-IRMS and Amino Alcohols Measured by EA-IRMS and GCC-IRMS amino acid
δ13C ( SD (‰) (EA-IRMS)
amino alcohol
δ13C ( SD (‰) (EA-IRMS)
Ala Val Leu Ile Met Phe
-21.06 ( 0.18 -10.29 ( 0.55 -27.32 ( 0.90 -11.11 ( 0.10 -25.72 ( 0.49 -10.06 ( 0.31
alaninol valinol leucinol isoleucinol methioninol phenylalaninol
-21.51 ( 0.01 -12.2 ( 0.4 -27.90 ( 0.20 -11.98 ( 0.32 -26.05 ( 0.06 -10.62 ( 0.44
∆δ13Ca (EA/EA) 0.45 1.91 0.58 0.87 0.33 0.56 +0.67
av deviation
δ13C ( SD (‰) (GCC-IRMS)
∆δ13Cb (EA/GC)
-20.58 ( 0.56 -10.40 ( 0.37 -27.20 ( 0.08 -10.65 ( 0.13 -25.58 ( 0.10 -10.14 ( 0.29
-0.48 0.11 -0.12 -0.46 -0.14 0.08 -0.14
a ∆δ13C determined as the difference between EA analysis of amino acid and amino alcohol. b ∆δ13C determined as the difference between EA analysis of amino acid and GC analysis of amino alcohol.
Figure 3. δ Values of amino acids (O) measured by EA-IRMS and amino alcohols (0) measured by GCC-IRMS. Standard deviations of replicates are within the symbol sizes.
Figure 2. Chromatography of an amino alcohol mixture using a 5-µm-thick stationary-phase film. Peaks are sharp and suitable for isotopic analysis. Retentions times (in min): alaninol (2-amino-1propanol), 7.7; valinol (2-amino-3-methyl-1-butanol), 11.7; leucinol (2amino-4-methyl-1-pentanol), 12.5; isoleucinol (2-amino-3-methyl-1pentanol), 14.5; methioninol (2-amino-4-methylthio-1-butanol), 19.4; phenylalaninol (2-amino-3-phenyl-1-propanol), 22.1.
differences between δ13C values of amino acids and amino alcohols measured by EA average +0.67‰ (range 0.33-1.91‰). All amino alcohol values were greater than the corresponding amino acid, indicating the presence of enriched C in the analysis mixture, indicative of either a minor isotopically enriched contaminant or isotopic fractionation during the reaction. To distinguish the two possibilities and to demonstrate GCCIRMS analysis of amino alcohols, we analyzed them by GCCIRMS. The results, presented in Table 2 and graphically in Figure 3, show that the mean difference between the EA analysis of amino acids and the GCC-IRMS analysis of amino alcohols produces a mean difference of δ13C ) -0.14‰, which is well within experimental error. Significantly, four amino alcohols now have lower isotope ratios than their corresponding amino acids, while two are higher, as shown in Figure 3. We conclude from these data that the bias in EA analysis of amino acids and amino alcohols results from minor C-containing contaminants probably introduced in the reaction mixture that are included in the EA analyses but are separated from the amino alcohols prior to analysis in the GCC-IRMS analysis. Table 2 also shows that the average precision of the amino alcohol GCC-IRMS analyses was SD(δ13C) ) 0.19‰. This precision is well within that reported for most continuous-flow applications and demonstrates the feasibility of the overall approach for these nonpolar amino acids. While these data were generated on purified standards, numerous methods for cleanup of amino acids are available. For 802 Analytical Chemistry, Vol. 73, No. 4, February 15, 2001
instance, Lecavalier et al. presented a protocol for purification of Phe, Leu, and Ile from plasma intended for subsequent HPLC.26 Nissen and co-workers reported a similar method for isolation of the branched-chain amino acids Leu, Ile, Val, and Met.27 The products of either of these protocols are very similar to our starting materials and in principle the present method could be applied to these mixed analytes. In general, scale down of chemical reactions is straightforward; thus, the reactions reported for 2 mg should be readily achieved for smaller quantities, within the usual constraints of high-precision analysis. In addition to the reductions shown here, we have successfully reduced proline, tryptophan, threonine, and tyrosine, though their amino alcohols were not sufficiently volatile for GC analysis under the conditions reported here. Ongoing work is targeted toward extending GC analysis to these analytes. Reduction of highly water soluble amino acids and those with H-bonding moieties on their side chains will require different reaction conditions for effective reduction and will probably require additional chemical treatment strategies to remove these groups to yield amino alcohols sufficiently volatile for analysis without addition of C. ACKNOWLEDGMENT This work was supported by NIH Grant GM49209.
Received for review May 26, 2000. Accepted December 7, 2000. AC000604L (26) Lecavalier, L.; Horber, F. F.; Haymond, M. W. J. Chromatogr. 1989, 491, 410-7. (27) Nissen, S. L.; Van Huysen, C.; Haymond, M. W. J. Chromatogr. 1982, 232, 170-5.
