15N/14N position-specific isotopic analyses of polynitrogenous amino acids

Article abstract of DOI:10.1021/ac048903o

¹⁵N/¹⁴N isotope ratios are widely used to study processes and systems involving amino acids. Nitrogen isotope fractionation in biological processes occurs primarily at sites of bond-breaking and formation; the finest discrimination for "isotopic fingerprinting" and studies of isotopic fluxes is thus obtained at the position-specific level. While there are numerous reports of natural intramolecular carbon isotope variability, there are no literature reports of ¹⁵N/¹⁴N position-specific isotopic analysis (N-PSIA) of biologically relevant molecules. We report a methodology for high-precision N-PSIA of four polynitrogenous α-amino acids (asparagine, glutamine, lysine, histidine) and the first survey of natural intramolecular ¹⁵N/¹⁴N in these biomolecules. Selective liberation of N-atoms from multiple commercial standards of each parent amino acid was achieved by an appropriate enzymatic reaction or by acid hydrolysis. ¹⁵N/¹⁴N measurements were performed on N-ethoxycarbonyl ethyl ester derivatives of the parent amino acids and their analogues by gas chromatography combustion isotope ratio mass spectrometry, and the average precision for replicate injections was found to be SD(δ¹⁵N) = 0.3‰. Position-specific δ¹⁵N values of the parent amino acid were directly observed or indirectly calculated using mass balance. The average precision obtained for directly measured positions was SD(δ¹⁵N) = 0.2-0.4‰. The average precision for indirectly obtained positions was SD(δ¹⁵N) = 0.6-1.3‰ as a result of propagation of errors. Enrichment in the side chain-N with respect to the peptide-N was observed in nearly all of the amino acid sources, most notably in asparagine (average Δδ_side-peptide = +11‰), which may be indicative of its method of production. In some cases, it was possible to distinguish commercial sources by N-PSIA that could not be distinguished at the compound-specific level.

Full text of DOI:10.1021/ac048903o

Anal. Chem. 2005, 77, 1013-1019
15N/14N Position-Specific Isotopic Analyses of
Polynitrogenous Amino Acids
Gavin L. Sacks and J. Thomas Brenna*
Division of Nutritional Sciences, Cornell University, Savage Hall, Ithaca, New York 14853
struction of contemporary and historical dietary behaviors.4,5 These
applications are based heavily on empirical data and, as such, treat
the isotopic information as a “fingerprint”. However, variations in
isotope ratios are clearly a result of rational (bio)chemical
processes. Isotopic fractionation in biological systems is primarily
a result of kinetic isotope effects associated with bond-breaking
and formation; therefore, nonstatistical isotope distributions are
expressed, and most clearly observed, at the position-specific
level.6 Recent reviews7-9 have called for a robust study of
nonstatistical isotopic distributions within an organism, and the
authors of the reviews stress the importance of developing routine
methods for high-precision position-specific isotope analysis
(PSIA) to facilitate this goal.
While examples of carbon PSIA (C-PSIA) using IRMS are
becoming more commonplace,7 intramolecular 15N/14N isotope
measurements are nearly nonexistent. It has been known for some
time that the R-amino fraction of amino acids is depleted in 15N
with respect to the amide fraction,10 implying that the peptide-N
of glutamine should be depleted with respect to its amide-N.
However, this intramolecular variation has not been measured
directly. In fact, aside from a 1982 extended abstract by Medina
et al.,11 there are no literature reports of N-PSIA on biologically
relevant molecules. The lack of data may have been partly due to
the relative scarcity of small molecules that contain more than
one nitrogen atom, but also because of the lack of facile methods
for rapid, on-line N-PSIA measurements. Recently, several groups
have begun to study isotopomers of N2O, a technique that shows
considerable promise in understanding the cycling of this impor-
tant atmospheric pollutant.12,13 However, these measurements take
advantage of the unique attributes of N2O in the IRMS source,
specifically the dissociation of N2O+ to NO+, and do not provide
a framework for N-PSIA studies of other organic molecules. A
15N/14N isotope ratios are widely used to study processes
and systems involving amino acids. Nitrogen isotope
fractionation in biological processes occurs primarily at
sites of bond-breaking and formation; the finest discrimi-
nation for “isotopic fingerprinting” and studies of isotopic
fluxes is thus obtained at the position-specific level. While
there are numerous reports of natural intramolecular
carbon isotope variability, there are no literature reports
of 15N/14N position-specific isotopic analysis (N-PSIA) of
biologically relevant molecules. We report a methodology
for high-precision N-PSIA of four polynitrogenous r-amino
acids (asparagine, glutamine, lysine, histidine) and the
first survey of natural intramolecular 15N/14N in these
biomolecules. Selective liberation of N-atoms from mul-
tiple commercial standards of each parent amino acid was
achieved by an appropriate enzymatic reaction or by acid
hydrolysis. 15N/14N measurements were performed on
N-ethoxycarbonyl ethyl ester derivatives of the parent
amino acids and their analogues by gas chromatography
combustion isotope ratio mass spectrometry, and the
average precision for replicate injections was found to be
SD(δ15N) ) 0.3‰. Position-specific δ15N values of the
parent amino acid were directly observed or indirectly
calculated using mass balance. The average precision
obtained for directly measured positions was SD(δ15N)
) 0.2-0.4‰. The average precision for indirectly ob-
tained positions was SD(δ15N) ) 0.6-1.3‰ as a result
of propagation of errors. Enrichment in the side chain-N
with respect to the peptide-N was observed in nearly all
of the amino acid sources, most notably in asparagine
(average ∆δside-peptide ) +11‰), which may be indicative
of its method of production. In some cases, it was possible
to distinguish commercial sources by N-PSIA that could
not be distinguished at the compound-specific level.
(3) Schmidt, T. C.; Zwank, L.; Elsner, M.; Berg, M.; Meckenstock, R. U.;
Haderlein, S. B. Anal. Bioanal. Chem. 2004, 378, 283-300.
(4) Mottram, H. R.; Dudd, S. N.; Lawrence, G. J.; Stott, A. W.; Evershed, R. P.
J. Chromatogr., A 1999, 833, 209-221.
(5) Meier-Augenstein, W. Anal. Chim. Acta 2002, 465, 63-79.
(6) Hayes, J. M. In Stable isotope geochemistry; Valley, J. W., Cole, D. R., Eds.;
Mineralogical Society of America: Washington, DC, 2001; pp 225-277.
(7) Brenna, J. T. Rapid Commun. Mass Spectrom. 2001, 15, 1252-1262.
(8) Werner, R. A.; Schmidt, H. L. Phytochemistry 2002, 61, 465-484.
(9) Schmidt, H. L. Naturwissenschaften 2003, 90, 537-552.
Stable isotope analysis by gas chromatography combustion-
isotope ratio mass spectrometry (GCC-IRMS) is a powerful
technique for providing clues to the processes and sources that
formed a sample. Typical applications include the authentication
of flavors and fragrances and the assignment of place of origin
for foodstuffs,1,2 sourcing of environmental pollutants,3 and recon-
(10) Gaebler, O. H.; Vitti, T. G.; Choitz, H. C.; Vukmirov. R Can. J. Biochem.
Physiol. 1963, 41, 1089-&.
* To whom correspondence should be addressed. E-mail: jtb4@cornell.edu.
Voice: (607)255-9182. Fax: (607)255-1033.
(1) Lichtfouse, E. Rapid Commun. Mass Spectrom. 2000, 14, 1337-1344.
(2) Asche, S.; Michaud, A. L.; Brenna, J. T. Curr. Org. Chem. 2003, 7, 1527-
1543.
(11) Medina, R.; Olleros, T.; Schmidt, H. L. 4th International Conference on Stable
Isotopes; Elsevier Scientific Publishing Co.: New York, 1982; pp 77-82.
(12) Toyoda, S.; Yoshida, N. Anal. Chem. 1999, 71, 4711-4718.
(13) Rockmann, T.; Brenninkmeijer, C. A. M.; Wollenhaupt, M.; Crowley, J. N.;
Crutzen, P. J. Geophys. Res. Lett. 2000, 27, 1399-1402.
10.1021/ac048903o CCC: $30.25 © 2005 American Chemical Society
Published on Web 01/11/2005
Analytical Chemistry, Vol. 77, No. 4, February 15, 2005 1013

more widely applicable option is to use on-line pyrolysis (Py)
coupled to a GCC-IRMS system.14 This technique has been
successfully used for C-PSIA measurements of alkanes,15 amino
acids,16 and short-chain organic acids,17,18 and a modification of
this system was used to measure site-specific 18O/16O isotope
ratios in vanillin.19 The primary advantage of the Py-GCC-IRMS
system for C-PSIA is that it permits fragmentation and isolation
of sites that are not readily accessible by off-line chemical methods.
However, Py-GCC-IRMS systems are not commercially available
at this time. Furthermore, pyrolytic fragmentation is nonquanti-
tative, which leads to isotopic fractionation and accompanying
difficulties in measuring absolute isotope ratios.16
from MP Biomedicals (Aurora, OH) and prepared as a 5000 unit/
mL suspension in a pH 9.0 Tris buffer, where 1 unit will deaminate
1 µg of His/min at 25 °C.
The amino acid standards were prepared as 10 mg/mL
solutions in water. A 100-µL aliquot of standard solution, 80 µL of
the appropriate buffer, and 20 µL of the enzyme solution were
added to a 2-mL reaction vial. The mixture was capped with a
Teflon seal, vortexed, and incubated at 37 °C for 24 h. Asparagi-
nase, glutaminase, and histidase generate the amino acid ana-
logues shown in Figure 1 directly. Lysine oxidase converts lysine
to 2-oxo-6-aminohexanoic acid, which spontaneously decarbox-
ylates in the absence of catalase to form DAVA.20
In the case of the polynitrogenous amino acids, the nitrogen
sites are often discrete moieties (such as amine or amide groups)
that can be isolated by chemical means with relative ease. Off-
line fragmentation and isolation is therefore well-suited for N-PSIA.
In this paper, we report a rapid method for N-PSIA of four of the
six polynitrogenous proteinogenic amino acids, asparagine (Asn),
glutamine (Gln), lysine (Lys), and histidine (His), and evaluate
the analytical figures of merit for the overall procedure. We then
apply the procedure to a series of commercially obtained amino
acids to demonstrate the existence of natural variability of
intramolecular δ15N in biomolecules.
N-Ethoxycarbonyl Ethyl Ester (ECEE) Derivatization of
Amino Acids. ECEE derivatives were prepared by a procedure
adapted from Husek21 and from Montigon et al..22 A 200-µL aliquot
of the substrate (either the analogue reaction mixture or a 5 mg/
mL solution of the standard in an appropriate buffer), 100 µL of
EtOH, and 25 µL of pyridine were added to a 2-mL reaction vial.
For derivatization of DAVA, the reaction was acidified with 50 µL
of 2 N HCl. Ethyl chloroformate (ECF, 10 µL) was added. The
mixture was capped, vortexed, and allowed to sit for 5 min. A
300-µL aliquot of saturated NaHCO3 and 1 mL of CH2Cl2 were
added, and the aqueous layer was removed. Anhydrous CaCl2 was
added, and the dried organic layer was transferred to another vial.
The CH2Cl2 was removed under a nitrogen stream, and the
product was redissolved in 50 µL of CH2Cl2 before introduction
to the GCC-IRMS. Identification of the derivatives was performed
by a Varian Saturn 2000 ion trap (Walnut Creek, CA) operated in
positive ion electron impact mode. The identities of the R-amino
acid ECEE derivatives were confirmed by comparing characteristic
ion peaks to data collected by Huang and colleagues.23 ECEE
derivatization of DAVA and urocanic acid has not been previously
reported. Under acidic conditions, ECEE derivatization of DAVA
formed a cyclic product (2-piperidone), and identification was aided
by the Wiley mass spectral library. The urocanic acid-ECEE
derivative showed characteristic ion peaks at m/z ) 238, 193, 166,
121, 94, and 66; these peaks correlate with the fragmentation
pattern observed in His-ECEE following loss of the R-amino-N-
ethoxycarbonyl group.23
GCC-IRMS of Amino Acid Derivatives. The system we
used is similar to that described by Merritt and Hayes.24 Briefly,
the sample was injected, splitless, via an autosampler (Varian, Inc.
8200; Walnut Creek, CA) into a GC (Varian 3400CX). Components
were separated on a 15 m × 0.32 mm × 3.0 µm VB-1 capillary
column (cross-linked dimethylpolysiloxane; VICI; Houston, TX).
The initial GC oven temperature was 80 °C, held for 5 min, ramped
to 260 °C at 15 °C/min, and held for 5 min. The head pressure
was set at 10 psi, resulting in a flow rate of 2 mL/min at a GC
temperature of 80 °C. An electronically controlled Valco rotary
valve was used to divert the solvent and to provide an auxiliary
EXPERIMENTAL SECTION
Amino Acid Standards. Four or five standards of each amino
acid (
respective analogues (
L
-Asn,
L
-Gln,
L
-Lys,
L
-His) and one standard each of their
-Glu, δ-aminovaleric acid (DAVA),
L
-Asp,
L
urocanic acid) were purchased from commercial vendors and used
without further preparation. The vendors used were Acros (Geel,
Belgium; “Acr”), Aldrich (St. Louis, MO; “Ald”), Avocado Organics
(Heysham, England; “Avo”), J. T. Baker (Phillipsburg, NJ; “Bak”),
Fluka (Buchs, Switzerland; “Flu”), and Sigma (St. Louis, MO;
“Sig”). The sources for each amino acid were as follows. Asn: Acr,
Ald, Avo, Flu. Asp: Sig. Gln: Acr, Ald, Avo, Bak, Flu. Glu: Sig.
Lys: Ald, Avo, Flu, Sig. DAVA: Ald; His: Acr, Ald, Avo, Sig.
Urocanic acid: Sig.
Acid Hydrolysis: Asn and Gln. The amino acids were
prepared as 10 mg/mL solutions in water. A 100-µL aliquot of
each standard solution was added along with 100 µL of 2 N HCl
to a 2-mL reaction vial. The mixture was capped with a Teflon
seal, vortexed, and incubated at 80 °C for 24 h.
Enzymatic Reactions: Asn, Gln, Lys, and His. Asparagi-
nase (EC 3.5.1.1), glutaminase (EC 3.5.1.2), and lysine oxidase
(EC 1.4.3.14) were purchased as lyophilized powders from Sigma-
Aldrich and prepared as a 5 unit/mL solution in an appropriate
buffer (50 mM asparaginase, pH 8.6 Tris; glutaminase, pH 4.9
sodium acetate; lysine oxidase, pH 8.0 Tris), where 1 unit will
liberate 1 µmol of NH3 from the substrate amino acid in 1 min at
37 °C in the specified buffer. Histidase (EC 4.3.1.3) was purchased
(14) Corso, T. N.; Brenna, J. T. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 1049-
1053.
(15) Corso, T. N.; Brenna, J. T. Anal. Chim. Acta 1999, 397, 217-224.
(16) Sacks, G. L.; Brenna, J. T. Anal. Chem. 2003, 75, 5495-5503.
(17) Dias, R. F.; Freeman, K. H.; Franks, S. G. Org. Geochem. 2002, 33, 161-
168.
(18) Yamada, K.; Tanaka, M.; Nakagawa, F.; Yoshida, N. Rapid Commun. Mass
Spectrom. 2002, 16, 1059-1064.
(20) Kusakabe, H.; Kodama, K.; Kuninaka, A.; Yoshino, H.; Misono, H.; Soda,
K. J. Biol. Chem. 1980, 255, 976-981.
(21) Husek, P. J. Chromatogr. 1991, 552, 289-299.
(22) Montigon, F.; Boza, J. J.; Fay, L. B. Rapid Commun. Mass Spectrom. 2001,
15, 116-123.
(23) Huang, Z. H.; Wang, J.; Gage, D. A.; Watson, J. T.; Sweeley, C. C.; Husek,
P. J. Chromatogr. 1993, 635, 271-281.
(19) Dennis, M. J.; Wilson, P.; Kelly, S.; Parker, I. J. Anal. Appl. Pyrolysis 1998,
47, 95-103.
(24) Merritt, D. A.; Hayes, J. M. J. Am. Soc. Mass Spectrom. 1994, 5, 387-397.
1014 Analytical Chemistry, Vol. 77, No. 4, February 15, 2005

Figure 1. Structures of R-amino acids and their analogues. The conversion of Lys to DAVA involves a subsequent decarboxylation following
enzymatic deamination. Position-specific δ15N values of the original R-amino acids were calculated either directly or indirectly by mass balance,
using the equations in the rightmost column.
flow of O2 for recharging the furnace, in place of a back-flush
system.25
method for background correction described by Ricci et al.28 was
found to give satisfactory results.
Elemental Analyzer (EA)-IRMS of Amino Acid Derivatives.
The system consisted of an NC2500 Carlo Erba EA interfaced to
a Finnigan MAT Delta Plus IRMS (Bremen, Germany) through
a ConFlo II open split. In the EA, the combustion conditions were
1000 °C over Cr2O3 and CuO, and the reduction conditions were
650 °C over copper filings. Samples (∼1 mg) were loaded into tin
cups (Costech; Valencia, CA) before introduction into the EA by
autosampler. High-purity He (120 mL/min) was used as a carrier
gas, and the N2 reference gas was of highest purity.
Reporting Isotope Ratios. The accepted method for reporting
isotope measurements is to compare the isotope ratio of the
analyte to the isotope ratio of a standard. For nitrogen studies,
analytes are indirectly calibrated against N2 in air and expressed
in delta notation:
The carrier stream exited the rotary valve by fused-silica
transfer line and was coupled within the GC oven to the
combustion furnace using Valco vespel ferrules and zero dead
volume connectors. The combustion furnace consisted of a 30 cm
× 0.5 mm ceramic tube packed with oxidized Cu and resistively
heated to 950 °C by a Fibercraft furnace (Thermcraft; Winston-
Salem, NC). The combustion products were directed to a similarly
constructed reduction furnace, packed with Cu metal, and held
at 550 °C, to convert NxOy to N2. CO2 and H2O were removed by
an LN2 cold trap, and a portion of the effluent stream was admitted
via an open split to an APP2003 IRMS (GV Instruments; Manches-
ter, England). High-purity N2 was calibrated against air and used
as a reference.
Consecutive runs were automated by macros written in the
native APP2003 scripting language. Data acquisition was per-
formed by the APP2003, and the raw data were exported to
SAXICAB,26 a home-written LabVIEW27 software package, for
calculation of δ15N values. The background in the area around
the peaks was low and nearly flat, and the individual summation
δ15Nair ) (Rspl - Rair)/Rair × 1000
Rx ) [15Nx/14Nx]
where Rspl is the isotope ratio of the sample and Rair ) 0.003 678 2.
RESULTS AND DISCUSSION
The structures of the investigated amino acids (Asn, Gln, Lys,
His) and their analogues are shown in Figure 1. The conversion
(25) Goodman, K. J. Anal. Chem. 1998, 70, 833-837.
(26) Sacks, G. L.; Brenna, J. T.; Sepp, J. T. 49th ASMS Conference on Mass
Spectrometry, Chicago, IL, 2001.
(28) Ricci, M. P.; Merritt, D. A.; Freeman, K. H.; Hayes, J. M. Org. Geochem.
1994, 21, 561-571.
(27) National Instruments. Labview 6i; Austin, TX, 2000
Analytical Chemistry, Vol. 77, No. 4, February 15, 2005 1015

Figure 2. General reaction scheme for the preparation of ECEE
derivatives of amino acids.
of Gln, Asn, and His to their corresponding analogues is
straightforward. The enzymatic deamination of Lys generates
2-oxo-6-aminohexanoic acid, which spontaneously decarboxylates
to form DAVA.20 The amino acids were made amenable to GC
analysis by reaction with ECF in a H2O/EtOH/pyridine solvent
system to form ECEE derivatives. The reaction proceeds by the
general reaction scheme shown in Figure 2, with simultaneous
derivatization of the N-containing moieties and the carbonyl
group.21,23 ECEE derivatization for GCC-IRMS 13C/12C measure-
ments of Gln has been reported previously22 and provides several
advantages over more common derivatization schemes. Most
sample preparation techniques for 15N/14N measurements by
GCC-IRMS rely on esterification followed by either trifluoro-
acetylation, acetylation, or pivaloylation.29 Unfortunately, these
methods are inappropriate for the analysis of amides (Asn, Gln)
due to the acidic conditions of the esterification step. Tert-
butylmethylsilation (t-BDMS) derivatization is also routinely used
for GCC-IRMS, and the procedure is amenable to amides.
However, t-BDMS derivatives are unstable at room temperature,
add a large carbon load, and can be damaging to the combustion
furnace.22,30 In contrast, ECEE derivatives are stable for at least 1
week at room temperature, introduce no heteroatoms aside from
oxygen, give excellent chromatography, and are compatible with
all of the proteinogenic amino acids except for arginine.21
Figure 3. GCC-IRMS chromatograms of ECEE derivatives of Asn
standard, Asp standard, and deaminated Asn, showing complete
conversion of Asn to Asp by asparaginase. Three N2 standard pulses
were introduced at the beginning of each run. The additional peaks
in the deaminated bottom run are from contaminants in the aspara-
ginase and are unrelated to the analyte.
Reproducibility of Analogue Preparation and ECEE De-
rivatization. Asn and Gln analogues (Asp and Glu, respectively)
were prepared by both enzymatic reaction and under acidic
conditions; Lys and His analogues (DAVA and urocanic acid,
respectively) were prepared only by enzymatic reaction. In these
procedures, no trace of the starting amino acid was detectable
after 24 h. Representative chromatograms of the conversion of
Asn to Asp by asparaginase are shown in Figure 3, and similar
results were achieved with the other procedures. Meier-Augen-
stein has cautioned that the nonquantitative nature of the chlo-
roformate derivatization is a potential hazard for GCC-IRMS31
and could result in severe and irreproducible fractionation. Other
groups have reported recoveries ranging between 75 and 100%
for ECF derivatization of amino acids,32,33 and we achieved similar
efficiencies in our work (data not shown). Incomplete reactions
in derivatization schemes that cause reproducible isotopic frac-
tionation are well known and, if properly taken into account, do
not seriously degrade the quality of isotope ratio measurement.34
To assess the reproducibility of the fractionation, the ECEE
derivatization was run in duplicate, “prep 1” and “prep 2”, for
multiple commercial sources of each amino acid and its analogue.
δ15N values for the two runs were measured by GCC-IRMS with
an average precision of SD(δ15N) ) 0.3‰ for three or four replicate
injections. The difference between prep 1 and prep 2, ∆prep, was
calculated as
∆
prep ) δ15N(prep 1) - δ15N(prep 2)
The average value and standard deviation of ∆prep are reported
in Table 1 for each compound. Two criteria were defined for
determining whether the analogue preparation and derivatization
procedures behaved acceptably. First, if the average value of ∆prep
was smaller than SD[∆prep], then the difference between the runs
was considered to be within experimental error. In all cases, the
average value of ∆prep was small (<0.3‰), and met this criterion.
Second, the run-to-run precision, SD[∆prep], should be similar to
(or better than) the precision typically associated with δ15N
measurements by GCC-IRMS, SD(δ15N) ) 0.5‰.3 Measurements
of the parent amino acids yielded SD[∆prep] of less than 0.3‰.
However, some of the analogue preparation procedures resulted
in unacceptably high values of SD[∆prep]. For example, enzymatic
hydrolysis of Gln to Glu yielded poor results (SD[∆prep] ) 0.98‰)
compared to acid hydrolysis (SD[∆prep] ) 0.35‰). Husek23
reported that ECF derivatization of Glu forms two products: the
expected ECEE-Glu derivative and a dehydrated ECEE-pyro-
(29) Hofmann, D.; Gehre, M.; Jung, K. Isot. Environ. Health Stud. 2003, 39,
233-244.
(30) Shinebarger, S. R.; Haisch, M.; Matthews, D. E. Anal. Chem. 2002, 74,
6244-6251.
(31) Meier-Augenstein, W. Curr. Opin. Clin. Nutr. Metab. Care 1999, 2, 465-
470.
(32) Nozal, M. J.; Bernal, J. L.; Toribio, M. L.; Diego, J. C.; Ruiz, A. J. Chromatogr.,
A 2004, 1047, 137-146.
(33) Silva, B. M.; Casal, S.; Andrade, P. B.; Seabra, R. M.; Oliveira, M. B.; Ferreira,
M. A. Anal. Sci. 2003, 19, 1285-1290.
(34) Docherty, G.; Jones, V.; Evershed, R. P. Rapid Commun. Mass Spectrom.
2001, 15, 730-738.
1016 Analytical Chemistry, Vol. 77, No. 4, February 15, 2005

Table 1. Reproducibility of Isotope Ratio Analysis for ECEE Derivatization of Parent Amino Acids and Their
Corresponding Analoguesa
parent
prep2
analogue (‰)
enzymatic
amino
acid
acid
source
prep1
∆
prepb
hydrolysis
Asn
Acr
Ald
Avo
Flu
0.74
-1.88
0.16
0.43
-1.99
0.26
0.31
0.11
-0.10
-0.11
-3.25
-3.14
c
av ∆prep
0.05 ( 0.20
-0.02 ( 0.26
-0.34 ( 2.11
-0.26 ( 0.35
(mean(SD, ‰)
Gln
Lys
His
av ∆prep (‰)
av ∆prep (‰)
av ∆prep (‰)
-0.13 ( 0.14
-0.01 ( 0.11
-0.05 ( 0.26
0.15 ( 0.98
-0.07 ( 0.30
-0.42 ( 0.65
n/ad
n/a
a Reactions were done in duplicate, prep 1 and prep 2, and δ15N was measured for each by GCC-IRMS (mean precision, SD(δ15N) ) 0.3‰, for
three or four replicate injections). Complete data for Asn (Parent) are reported and summary data for other amino acids presented for brevity.
b
∆
prep
) δ15NPrep 1-δ15NPrep 2 c Av ∆prep, mean ( SD: calculated as the mean of ∆prep for Asn; SD calculated as the pooled error for the four ∆prep
.
values. Other av ∆prep defined equivalently for the other amino acids. d n/a, not attempted.
Table 2. Comparison of δ15N Values of ECEE Derivatives (GCC-IRMS, n ) 7) and Underivatized Standards (EA-IRMS,
n ) 2)a
amino
acid
GCC-
EA-
R
(‰)
amino
acid
GCC-
EA-
R
(‰)
vendor
IRMS
IRMS
vendor
IRMS
IRMS
Asn
Acr
Ald
Avo
Flu
Sig
Acr
Ald
Avo
Bak
Flu
Sig
0.59
-1.93
0.21
4.17
1.99
-4.04 ( 0.44
Lys
Ald
Avo
Flu
Sig
Sig
Acr
Ald
Avo
Sig
-7.10
-5.46
-6.67
-6.15
-6.02
-9.96
-10.24
-5.05
-9.96
0.88
1.54
-7.45 ( 0.40
4.22
0.58
-3.20
-10.09
-9.34
-3.77
-8.97
-9.05
-9.42
-3.17
1.43
1.33
Asp
Gln
-8.56
-8.33
-2.13
-8.38
-8.16
-8.65
-4.59
-1.53
-0.98 ( 0.40
DAVA
His
-5.93
-7.76
-8.84
-2.91
-7.76
-0.09
-2.09 ( 0.19
Glu
+1.43
urocanic
acid
Sig
-7.16
-8.25
+1.09
a Fractionation factors (R) are reported in the rightmost column where R ) δ15N (GC) - δ15N (EA). For compounds where samples from
multiple vendors were run, the mean ( SD R value is calculated.
glutamic acid derivative. Under alkaline conditions, the two
products are formed in roughly equal amounts, but under acidic
conditions (pH ∼1), we saw nearly exclusive formation of the
ECEE-pyroglutamic acid. The opposite effect was seen for the
conversion of Asn to Asp, which showed far better reproducibility
with the enzymatic reaction (SD[∆prep] ) 0.26‰) than with acid
hydrolysis (SD[∆prep] ) 2.11‰). It is not clear why ECEE-Asp
derivatization yields poor reproducibility under acidic conditions.
ECEE derivatives of DAVA and urocanic acid have not been
previously reported. We observed that at pH 8.0, DAVA forms
both ECEE-DAVA predicted from the mechanism in Figure 2
and 2-piperidone via dehydration. In our work, it was necessary
to acidify the enzymatic reaction mixture following deamination
in order to preferentially form the dehydrated product, from which
good run-to-run precision was achieved (SD[∆prep] ) 0.30‰).
ECEE-urocanic acid, formed through enzymatic deamination of
His, had a reproducibility (SD[∆prep] ) 0.65‰) close to the
precision typically associated with 15N/14N GCC-IRMS. In sum-
mary, in our hands, reproducible fractionation during analogue
formation and ECEE derivatization can be achieved by acid
hydrolysis of Gln and by enzymatic treatment of Asn, Lys, and
His.
and bond-forming steps during the ECF derivatization, two steps
that might be subject to isotopic fractionation are the derivatization
of the parent amino acid or its analogue. A third, less probable
source of fractionation is during the analogue preparation because
the nitrogen sites in the analogue are not part of the bond-breaking
step and thus are at low risk for fractionation;35 consequentially,
we disregard this step as a potential source of fractionation. To
assess the extent of fractionation, δ15N of the underivatized amino
acids and their analogues were measured by EA-IRMS and
compared to the values obtained by GCC-IRMS. The fractionation
factors (R) were calculated for each substrate as R ) δ15NGC
-
δ15NEA and are reported in Table 2. For the parent amino acids,
multiple commercial sources were used, and an average fraction-
ation factor is reported. R was in the range of (2‰ for most
substrates, although it was significantly larger for Asn and Lys.
However, the reproducibility of the fractionation was very good
for the parent amino acids [SD(R) e 0.4‰], even in the case of
the sizable fractionation of Lys (RLys ) -7.45 ( 0.40‰). The
magnitude of the R’s cannot be completely explained by the
fractionation during the ECEE derivatization, which was <0.2‰
for the parent amino acids (Table 1). It is possible that N-
containing impurities present in the bulk samples skewed the EA
Fractionation during Analogue Preparation and Deriva-
tization. Because the nitrogen sites are part of the bond-breaking
(35) Rieley, G. Analyst 1994, 119, 915-919.
Analytical Chemistry, Vol. 77, No. 4, February 15, 2005 1017

Figure 4. Position-specific δ15N values of multiple sources of polynitrogenous amino acids, showing wide variation at the intramolecular level.
The average precision for each site, and for the total molecule, is in the bottom table.
analyses but are excluded from the GC analyses because of the
separation.
lated) position (side chain-N of Asn and Gln, peptide-N of Lys
and His) was greater, averaging 0.6-1.3‰, due to propagation of
errors.
The average and fractionation-corrected position-specific and
compound-specific (“total-N”) isotope ratios for each polynitrog-
enous amino acid are presented in Figure 4. The position-specific
δ15N show a range at least double that shown by total δ15N. In 11
of the 13 amino acids, the peptide-N is depleted with respect to
the side chain-N, and in the two exceptions, the differences are
relatively small.
Calculation of Position-Specific δ15N Values and Error
Analysis. Based on the reliable procedures established in the
reproducibility experiments, data from the enzymatic reactions
of Asn, Lys, and His, and from the acid hydrolysis of Gln, were
used for calculation of PSIA δ15N values. Data from the acid
hydrolysis of Asn and enzymatic hydrolysis of Gln were of
unacceptably poor reproducibility for these duplicate reactions and
were not further explored.
The δ15N measurements from each set of duplicate runs were
pooled, and the mean and standard deviation were recalculated
for each amino acid and analogue. The average δ15N values were
corrected using the experimentally measured fractionation factors.
The position-specific isotope ratios were then calculated for each
amino acid, either directly from the δ15N value of its analogue or
indirectly by subtraction of the isotopic contribution of the
analogue from the parent amino acid. The equations used to
calculate these values are listed in the rightmost column of Figure
1. For Asn, Gln, and Lys, the δ15N of the side chain represents a
single N, while for His, it represents a composite of the two
nonequivalent imidazole nitrogen sites. The average precision
achieved for the directly measured positions (peptide-N of Asn
and Gln, side chain-N of Lys and His) was SD(δ15N) ) 0.2-0.4‰.
The precision achieved for the indirectly measured (i.e., calcu-
More specific interpretation of the variation of position-specific
δ15N is hindered because production methods for these samples
are not known. We can, however, speculate based on published
methods of industrial amino acid production. The side chain-N of
Asn is enriched with respect to the peptide-N in all four sources
(average ∆δside-pep ) +11‰), and in Gln in four of the five sources
(average ∆δside-pep ) +3‰). This is consistent with Gaebler’s
observation that the bulk amide-N fraction is enriched with respect
to the bulk peptide-N fraction in plants10 and with more recent
reports that show enrichment in heteroaromatic compounds that
receive their N from Gln36. Glu acts as a general donor of
peptide-N to the other amino acids by transamination, which favors
15N enrichment in the remaining Glu and depletion in Asp; this
may explain Asn’s prominent depletion in the peptide-N compared
1018 Analytical Chemistry, Vol. 77, No. 4, February 15, 2005

to the side chain amide-N.37 However, it should be pointed out
that industrial production of Gln, Lys, and Asp is typically
accomplished by microbial fermentation,38,39 while Asn is often
produced from Asp by a secondary chemical synthesis step.
Furthermore, the peptide-N of the four Asn sources (-5 to +1‰)
is in a range appropriate for microbial production, while the side
chain amide-N shows much more variability and is quite enriched
(+3 to +13‰). Therefore, the relative enrichment of the side
chain-N of Asn may well be a reflection of the isotopic nature of
the feedstock used in its production.
All four sources of His show enrichment at the imidazole side
chain compared to the peptide-N (average ∆δside-amino ) +9‰).
Three of the four sources of Lys (Acr, Flu, Sig) show enrichment
at the side chain-N (average δ15N ) +4‰) with respect to the
peptide-N (average δ15N ) -3.5‰). The other source (Avo) does
not show a significant difference between the peptide- and side
chain-N positions. Assuming that all four lysine samples were
produced through microbial fermentation, this suggests that Avo
was produced under different physiological conditions or by a
different organism than the other three Lys sources.
The 15N/14N position-specific measurements provide a tool for
distinguishing between sources that cannot be separated by
GCC-IRMS at the compound-specific level. For example, by CSIA,
Gln from Acr was indistinguishable from Gln from Flu at the 95%
confidence limit. Using PSIA, the peptide-N of Acr Gln (δ15N )
-8.85 ( 0.41‰) is distinguishable from the same position on Flu
(-7.66 ( 0.58‰). Similarly, Asn from Acr and Avo are distinguish-
able at peptide-N (δ15N ) -4.31 and +1.18, respectively) although
they are nearly identical at the CSIA level.
CONCLUSIONS
High-precision δ15N measurements of low-microgram quanti-
ties of polynitrogenous amino acids were accomplished by
selective off-line removal of an N-containing moiety, followed by
derivatization with ethyl chloroformate. This represents the first
literature report of intramolecular 15N/14N measurements of
biologically relevant molecules and also provides a general
framework for how other N-PSIA studies of polynitrogenous
compounds may proceed. For example, the peptide-N of tryp-
tophan may be liberated by commercially available tryptophanase
and that of ornithine by ornithine oxo acid transaminase. The
ability to observe isotopic variation at the position-specific level
provides detailed information for refining models of biosynthetic
pathways and their resulting isotopic fluxes, as well as improved
specificity for “isotopic fingerprinting” studies. In this work, we
observed a general pattern of enrichment in the side chain-N with
respect to the peptide-N. In some cases, we could rationalize this
pattern in terms of biosynthetic fractionation (Gln) or as an artifact
of the synthetic process (Asn). Exceptions to this trend (Gln from
Flu, Lys from Avo) may indicate different biological sources or
different production techniques. To facilitate our interpretation
of PSIA data, we intend to study polynitrogenous amino acids from
organisms grown under controlled conditions in future work.
ACKNOWLEDGMENT
This work was supported by NIH Grant GM49209. G.L.S. was
supported by NIH Training Grant DK07158. We thank Art Kasson
at the Cornell Isotope Laboratory (COIL) for his generous
assistance in performing EA-IRMS, and Christopher Wolyniak for
helpful suggestions.
(36) Zalkin, H.; Smith, J. L. In Advances in Enzymology; Purich, D. L., Ed.; John
Wiley & Sons Inc.: New York, 1998; Vol. 72, pp 87-144.
(37) Macko, S. A.; Fogel, M. L.; Hare, P. E.; Hoering, T. C. Chem. Geol. 1987,
65, 79-92.
(38) Rowlands, R. T. Enzyme Microb. Technol. 1984, 6, 3-10.
(39) Kusumoto, I. J. Nutr. 2001, 131, 2552S-2555S.
Received for review July 27, 2004. Accepted November
12, 2004.
AC048903O
Analytical Chemistry, Vol. 77, No. 4, February 15, 2005 1019