Quantitation of phenylalanine and its trans-cinnamic, benzoic and hippuric acid metabolites in biological fluids in a single GC-MS analysis

Article abstract of DOI:10.1002/jms.1218

We describe a sensitive, simple and convenient stable isotope dilution assay developed to study endogenous metabolism of administered stable isotope-labeled phenylalanine (Phe) in phenylketonuric (PKU) mice treated experimentally with phenylalanine ammoni

Full text of DOI:10.1002/jms.1218

JOURNAL OF MASS SPECTROMETRY
J. Mass Spectrom. 2007; 42: 811–817
Published online 18 May 2007 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/jms.1218
Quantitation of phenylalanine and its trans-cinnamic,
benzoic and hippuric acid metabolites in biological
fluids in a single GC-MS analysis
Christineh N. Sarkissian,1 Charles R. Scriver1 and Orval A. Mamer2∗
1
Departments of Biology, Human Genetics and Pediatrics, McGill University and Debelle Laboratory, McGill University-Montreal Children’s Hospital
Research Institute, Montreal, QC, H3H 1P3, Canada
2
The Mass Spectrometry Unit, McGill University, Montreal, QC, H3A 1A4, Canada
Received 28 November 2006; Accepted 21 March 2007
We describe a sensitive, simple and convenient stable isotope dilution assay developed to study
endogenous metabolism of administered stable isotope-labeled phenylalanine (Phe) in phenylketonuric
(PKU) mice treated experimentally with phenylalanine ammonia lyase (PAL). Mouse urine and plasma
containing endogenous and administered labeled Phe together with internal standard Phe bearing a
different pattern of labeling are converted by in situ diazotization to 2-chloro-3-phenylpropionic acid
(CPP). A single solvent extraction is then used to isolate the isotopomers of CPP along with the trans-
cinnamic acid (TCA) produced from Phe by PAL, as well as the TCA metabolites benzoic and hippuric
acids. This procedure eliminates the need for a separate ion-exchange isolation step for Phe on a second
sample aliquot and separate GC-MS analysis. Extracted CPP and the Phe metabolites are then measured by
conversion to the pentafluorobenzyl esters and a single analysis by electron capture negative ion GC-MS.
The estimated lower limit of quantitation is 0.1 µM. Copyright  2007 John Wiley & Sons, Ltd.
KEYWORDS: diazotization; phenylalanine ammonia lyase; metabolite quantitation; phenylketonuria; mouse; plasma; urine
INTRODUCTION
An ongoing study of phenylalanine (Phe) metabolism in mice
requires the combined measurements of urine and serum
concentrations of endogenous Phe, of exogenous labeled Phe
tracer and of their unlabeled and labeled metabolites trans-
cinnamic (TCA), benzoic (BA) and hippuric (HA) acids.
TCA is produced from Phe by phenylalanine ammonia
lyase (PAL, EC 4.3.1.5, Scheme 1), a nonmammalian enzyme
currently being investigated as an alternative treatment of
phenylketonuria (PKU).1
Quantitation of plasma Phe as the butyl ester using
electrospray ionization tandem MS has been reported by
Hardy to have a lower limit of detection of 1 µM.2 Another
tandem MS method with LC inlet of underivatized Phe3
Scheme 1. Conversion of phenylalanine to trans-cinnamic acid
by phenylalanine ammonia lyase and subsequent metabolism
in multiple reaction monitoring with a lower limit of
detection of 1 nM was applied to mass neonatal screening
in Korea. Deng has published an electron ionization GC-MS
method4 for Phe and several other amino acids important in
neonatal screening based on solid-phase microextraction of
the isobutyl chloroformate derivatives made in the aqueous
phase. The lower limits of detection were on the order
of 1 µM. A second method published by Deng5 for Phe
and several other amino acids is applied to dried blood
to benzoic and hippuric acids.
spots collected from neonates and employs microwave-
accelerated formation of trimethylsilyl derivatives. These
are also analyzed in electron ionization GC-MS, with a lower
limit of detection of 0.48 µM for Phe.
Isolation of the acidic metabolites of Phe can be
accomplished by extraction with diethyl ether or ethyl acetate
from the acidified biological fluids. The metabolites may
then be converted into a suitable derivative for analysis
by GC-MS. Since Phe itself cannot be isolated by solvent
extraction, as it is amphoteric and poorly soluble in most
water-immiscible organic solvents, a separate isolation based
ŁCorrespondence to: Orval A. Mamer, The Mass Spectrometry
Unit, McGill University, 740 Dr. Penfield, Montreal, QC, H3A
1A4, Canada. E-mail: orval.mamer@mcgill.ca
Copyright  2007 John Wiley & Sons, Ltd.

812
C. N. Sarkissian, C. R. Scriver and O. A. Mamer
Scheme 2. Conversion of plasma and urinary phenylalanine to
2-chloro-3-phenylpropionic acid by in situ diazotization.
Figure 1. ECNI mass spectra of the PFB derivatives of CPP,
CPP5 and CPP7 (panels A, B and C, respectively) synthesized
from authentic Phe, Phe5 and Phe7, respectively, by
diazotization. The intense ion clusters are the base peaks in the
spectra and reveal the presence of chlorine isotopes, and are
fragments produced by the loss of the 181 Da C6F5 –CH2
radical from the molecular radical anions, m/z 364, 369 and
371, respectively (not detected).
is employed, with isotopically labeled Phe, TCA, BA and
HA as internal standards. Analysis is by electron capture
negative ion (ECNI) GC-MS of the pentafluorobenzyl ester
derivatives.
As an example of the utility and convenience of this
method, we describe here its application to two Pahenu2/enu2
(PKU) mice receiving a deuterium-labeled Phe bolus, one
treated with PAL, the other not. The action of PAL on
endogenous and labeled Phe in the treated mouse is clearly
apparent in the elevations in plasma and urine of labeled
TCA, BA and HA compared with those in the untreated
mouse.
EXPERIMENTAL
Scheme 3. Derivatization of the acidic compounds of interest
to their pentafluorobenzyl esters and fragmentation in electron
capture negative ionization.
Materials
The following labeled compounds were obtained from
CDN Isotopes, Pointe-Claire, Que´bec, Canada: [phenyl-2H5]-
3,3-2H2-phenylalanine (Phe7), [phenyl-2H5]-phenylalanine
(Phe5), [2H7]-trans-cinnamic acid (TCA7), [phenyl-2H5]-trans-
cinnamic acid (TCA5), [phenyl-2H5]-benzoic acid (BA5), ˛-
[13C1]-benzoic acid (BA1), [phenyl-2H5]-hippuric acid (HA5)
and 2,2-2H2-hippuric acid (HA2). Stock solutions of each
(40 µg/ml) were made in deionized water. BioMarin Phar-
maceuticals Inc. CA. USA, kindly provided the recombinant
form of Rhodosporidium toruloides PAL.1 Other chemicals and
solvents were obtained from reliable sources and used as
received.
on ion-exchange resins on a second aliquot is frequently used,
followed by derivatization and GC-MS analysis separately
from the metabolites. This can be problematic when the
sample size is severely limited and the sample numbers are
large.
Analysis can be greatly simplified by in situ diazotization
and conversion of Phe to 2-chloro-3-phenylpropionic acid
(CPP, Scheme 2), which may then be isolated along with
acidic Phe metabolites by a single extraction and GC-MS
analysis. In our study, a stable isotope dilution technique
Copyright  2007 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2007; 42: 811–817
DOI: 10.1002/jms

Phenylalanine diazotization and quantitation
813
Mouse model
Two mice (approximately 40 g each) from the same litter
of the homozygous mutant strain Pahenu2/enu2, acting as
orthologues of human PKU,6 were housed in polycarbonate
cages and placed on a Phe-free diet from Harlan Teklad (2826)
with water containing 30 mg/l L-Phe supplied ad libitum for
three consecutive days. On day 4, the animals were placed in
metabolic cages overnight (1 animal/cage). On day 5, both
were injected subcutaneously between the shoulder blades
with Phe5 (0.2 mg/g body weight) at 0 h. One hour later, one
mouse was injected subcutaneously (between the shoulder
blades) with PAL (2 IU) in 25 mM tris, 150 mM NaCl, pH 7.5
buffer, and the other injected with an equivalent volume of
the buffer vehicle. Plasma and urine were collected 1 h and
4 h after the Phe5 injection.
Calibration sample preparation
Calibration samples were prepared in deionized water over
the physiological ranges for administered Phe5 and its
PAL metabolites (TCA5, BA5, HA5) by serial dilutions of
1 mg/ml stock solutions to 0.002, 0.02, 0.1 and 0.2 mg/ml.
Internal standards Phe7, TCA7, BA1 and HA2 (12.5 µl of
stock solutions) were also added to each calibration sample,
and diluted to a final volume of 100 µl for diazotization as
described below.
Biological sample preparation
Internal standards Phe7, TCA7, BA1 and HA2 (6.25 µl of each
40 µg/ml stock solution) were added to 50 µl aliquots of both
plasma and urine. Samples were then diluted with deionized
water to final volumes of 250 µl and diazotized as described
below.
Diazotization
Diluted samples were made basic (pH approximately 12
with 2 N NaOH), and saturated with solid sodium chloride
by addition of small amounts of solid sodium chloride
until no more would dissolve, extracted with ether and
the ether phase discarded. The remaining samples were then
acidified (pH approximately 2 with 5 N HCl). Aqueous
sodium nitrite (50 µl of a 100 mg/ml solution) was added,
the resulting mixtures were vortex-agitated and then held
uncapped in a fume hood for 10 min to allow the gas
evolution to cease. The solutions were adjusted to pH 7.4
with sodium hydroxide. In this step, labeled Phe5 and
Phe7 and unlabeled endogenous Phe in the calibration and
mouse samples are converted by diazotization to 2-chloro-
3-phenylpropionic acid (CPP5, CPP7 and CPP, respectively,
Scheme 2).
Figure 2. ECNI mass spectra of the PFB derivatives of
authentic TCA5 (A), TCA7 (B), BA1 (C), BA5 (D), HA2 (E) and
HA5 (F). The principal ions are formed by the loss of the 181 Da
PFB radical from the molecular radical anions, m/z 333, 335,
303, 307, 361 and 364, respectively (not detected).
above, vortex-mixed for 2 min and then placed in an
ultrasonic bath for 20 min at room temperature. Hexane
(2 ml) was added, and the samples vortex-mixed for 1 min.
The hexane layers were removed and dried by vortex-mixing
with 10 mg of anhydrous sodium sulfate. This drying step
was repeated twice. Aliquots of the final hexane solutions
containing the PFB esters of labeled and unlabeled Phe,
TCA, BA and HA (Scheme 3) were decanted from the
final sodium sulfate drying agent into autoinjector vials
for GC/MS analysis.
Derivatization
Pentafluorobenzyl (PFB) ester derivatives were made in
the manner described by Hachey.7 The following stock
solutions were prepared: (A) methylene chloride (20 ml)
and pentafluorobenzyl bromide (0.4 ml, Aldrich Chemical
Co.); (B) potassium phosphate buffer (pH 7.4, 100 ml)
and tetrabutylammonium hydrogen sulfate (3.4 g, Aldrich
Chemical Co.). Solutions A (250 µl) and B (250 µl) were
combined with 250 µl of the aqueous samples prepared
Copyright  2007 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2007; 42: 811–817
DOI: 10.1002/jms

814
C. N. Sarkissian, C. R. Scriver and O. A. Mamer
Figure 3. An example of typical SIM chromatograms obtained for a calibrating sample containing CPP5 (A), CPP7 (B), TCA5 (C),
TCA7 (D), BA1 (E), BA5 (F), HA2 (G) and HA5 (H).
peak areas measured for Phe7 (m/z 190) were reduced by the
measured intensity of m/z 190 due to the 37Cl isotope in Phe5.
GC-MS analysis in selected ion monitoring
(SIM)mode
Aliquots (1 µl) of the derivatized mixtures were analyzed in
the ECNI mode with a Hewlett-Packard 5988A GC-MS fitted
with a 30 m ð 0.25 mm i.d. capillary column (J & W Scientific)
coated with a 0.25 µM DB-1 film. The injections were in the
splitless mode with an unpacked liner and employed a 1-
min purge interval. The helium flow rate was 2 ml/min;
RESULTS AND DISCUSSION
ECNI mass spectra obtained for the PFB esters of CPP,
CPP5 and CPP7 synthesized by diazotization of authentic
°
the injector and interface temperatures were 250 C. The
°
column was temperature-programmed from 100 C after a
°
°
°
1-min hold to 120 C at 40 C/min and then at 10 C/min to
°
°
280 C. The column was baked out at 280 C for 5 min at the
completion of each sample analysis. Methane was used as
the moderator gas at an indicated source pressure of 0.6 Torr,
°
and the ion source temperature was 120 C. SIM was used
to measure the intensities of negative ion fragments m/z
183, 188, 190, 147, 152, 154, 121, 122, 126, 178, 180 and 183
with dwell times of 50 ms each. These fragments arise by
the loss of the PFB radical from the molecular anions of the
PFB derivatives of CPP, CPP5, CPP7, TCA, TCA5, TCA7, BA,
BA1, BA5, HA, HA2 and HA5, respectively (Scheme 3). Blank
deionized water samples that were similarly prepared and
analyzed both in SIM and full-scan modes demonstrated the
absence of significant interferences at the required masses
and retention times.
Measured peak areas were corrected for isotopic impu-
rities in the administered Phe5 and the internal standards,
which were determined by prior analysis of the derivatized
labeled materials, and for natural abundance heavy iso-
tope inclusion in endogenous and labeled metabolites. The
Figure 4. The calibration plots for ꢀ, Phe, ♦, TCA, ꢁ, BA and
, HA over the expected physiological ranges for urine and
°
plasma (0.02–2.0 mg/ml). A linear response is demonstrated
for each metabolite over this range (R2 > 0.99 for each
metabolite).
Copyright  2007 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2007; 42: 811–817
DOI: 10.1002/jms

Phenylalanine diazotization and quantitation
815
Figure 5. The area-integrated selected ion chromatograms for Phe, TCA, BA and HA measured in plasma of the untreated and
PAL-treated mice (upper and lower panels, respectively). The rows are respectively the endogenous (i.e. unlabeled) compounds, the
penta-deutero-labeled Phe and its metabolites, and the bottom row illustrates the responses for the internal standards.
Phe, Phe5 and Phe7 are shown in Figs 1(A), (B) and (C),
respectively. The most intense ions correspond to the
carboxylate anions (m/z 183, 188 and 190, respectively)
produced by the loss of the PFB radical (181 Da) from the
diazotized molecular radical anions (m/z 364, 369 and 371,
respectively, not detected). The presence of chlorine in the
fragment ions is demonstrated by the isotopic distribution.
Inspection of Fig. 1 shows that no loss of deuterium labeling
occurs during diazotization.
Negative ion mass spectra obtained for the PFB esters
of authentic TCA5, TCA7, BA1, BA5, HA2 and HA5 are
shown in Fig. 2(A–F). The most intense ions correspond to
the carboxylate anions (m/z 152, 154, 122, 126, 180 and 183,
respectively) produced by the loss of the PFB radical from
the molecular radical anions (m/z 333, 335, 303, 307, 361 and
364, respectively, not detected).
SIM chromatograms obtained for one of the calibrating
mixtures containing Phe5 and labeled metabolites of Phe5
Copyright  2007 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2007; 42: 811–817
DOI: 10.1002/jms

816
C. N. Sarkissian, C. R. Scriver and O. A. Mamer
Table 1. Concentrations (µM) of administered 2H5-labeled Phe, its 2H5-labeled metabolites and unlabeled Phe and its metabolites
measured in mouse plasma and urine with and without PAL treatment
Plasma
Urine
Untreated
Treated
Untreated
Treated
Metabolite
1 h
4 h
1 h
4 h
1 h
4 h
1 h
8.5
120
0.33
1.7
4 h
Phe5
Phe
TCA5
TCA
BA5
BA
2300
450
15
4.0
5.9
130
740
820
3.5
4.0
2.8
57
1700
810
19
11
3.8
200
BQL
BQL
690
1200
50
97
3.3
75
19
31
6.1
33
0.36
BQL
4.3
260
69
59
1.3
1.6
130
190
31
93
220
190
8.2
7.0
290
300
2100
1700
9.3
210
0.87
220
HA5
HA
BQL
BQL
BQL
BQL
BQL
190
Notes: Phe5 was administered at time 0. At 1 h, PAL or buffer was administered.
BQL: Below quantifiable limit (estimated lower level of measurement is 0.1 µM).
and internal standards are shown in Fig. 3(A–H). The SIM
chromatograms represent the intensities of fragments m/z
188, 190, 152, 154, 122, 126, 180 and 183, which are carboxylate
anions formed by the loss of the PFB radical from the
molecular radical anions of the PFB derivatives of CPP5,
CPP7, TCA5, TCA7, BA1, BA5, HA2 and HA5, respectively.
The calibration plots for CPP, TCA, BA and HA over the
expected physiological ranges are shown in Fig. 4, and are
linear over 2 orders of magnitude (R2 > 0.99).
Figure 5 is an example of the SIM chromatograms
obtained for the analysis of the mouse plasma samples by
this technique. The upper set is taken from the untreated
PKU mouse receiving a Phe5 bolus and 1 h later a buffer
injection; the lower set is from the PKU mouse receiving
the Phe5 bolus and 1 h later 2.0 IU of PAL in the buffer.
The upper and lower panels are the area-integrated selected
ion chromatograms for Phe, TCA, BA and HA measured in
plasma of the untreated and PAL-treated mice, respectively.
The top row shows the responses for the endogenous (i.e.
unlabeled) compounds; the middle row, the 2H5-labeled Phe
and its metabolites; and bottom row, the responses for the
internal standards. Similar chromatographic profiles were
also obtained for the corresponding urine samples collected
in the 0–1 h and 1–4 h intervals (data not shown).
Table 1 shows the concentrations calculated from the
SIM data for the plasma and urine samples collected
1 h following Phe5 administration but immediately prior
to PAL or buffer injection, and again at 4 h. In plasma,
1 h after the administration of Phe5, and prior to PAL
injection, concentrations of Phe5 and its metabolites are
comparable for the treated and untreated mice. At 4 h,
the Phe5 concentrations are somewhat lower than in the
1-h samples (one does not expect a large decrease in
the phenylalanine pool size under the current protocol:
Sarkissian, unpublished data), and as expected, the labeled
and unlabeled TCA concentrations in the PAL-treated mouse
are greatly increased compared to those in the untreated
mouse. Plasma BA5 and BA concentrations are similar
in both mice, while both HA5 and HA are substantially
increased in the treated mouse over the untreated one at 4 h.
These observations are consistent with PAL deaminating
Phe5 and Phe to form TCA5 and TCA, respectively, in vivo,
and their subsequent and immediate oxidation to BA5 and
BA followed by conversion to the glycine conjugate HA5 and
HA, respectively.
The 1-h urine samples (Table 1) show comparable
concentrations of Phe5 and its metabolites prior to PAL or
buffer injection. At 4 h, the PAL-treated mouse has excreted
more of the Phe5 bolus than the untreated mouse. Urinary
concentrations of TCA and its metabolites differ markedly,
however, between the untreated and the PAL-treated mice.
Urinary TCA5 and BA5 are higher in the treated mouse, with
the treated mouse excreting more than 600 times as much
HA5 as the untreated mouse.
The findings, in both plasma and urine, are consistent
with the understanding that the liver metabolizes TCA to
BA, which is then excreted principally as HA via urine.8
Unlabeled Phe metabolites followed similar patterns in both
plasma and urine.
CONCLUSIONS
In situ diazotization of endogenous, tracer and internal
standard Phe as described here enables measurement of
plasma and urinary Phe and its acidic metabolites in a single
extraction and GC-MS analysis. The method eliminates Phe
isolation by conventional ion-exchange chromatography and
GC-MS analysis separate from its metabolites, and further
takes advantage of the enhanced sensitivity associated with
ECNI applied to PFB esters. Phe is converted by diazotization
in the sample fluid to an ˛-chloro analog of Phe which
is extractable into water-immiscible organic solvents along
with its acidic metabolites, whereas Phe is not. As proof
of principle, it was employed in two mice participating
in a study in which they received a deuterium-labeled
phenylalanine bolus subcutaneously. When treated with a
recombinant form of PAL, a yeast enzyme not occurring in
mammals, phenylalanine is converted to trans-cinnamic acid,
and its effect on Phe metabolism is clearly seen in the plasma
and urine profiles of treated vs control PKU animals. This
assay is also being used in other ongoing studies of amino
acid catabolism in rodents.
Copyright  2007 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2007; 42: 811–817
DOI: 10.1002/jms

Phenylalanine diazotization and quantitation
817
4. Deng C, Li N, Zhang X. Rapid determination of amino acids in
neonatal blood samples based upon derivatization with isobutyl
chloroformate followed by solid-phase microextraction and gas
chromatography/mass spectrometry. Rapid Communications in
Mass Spectrometry 2004; 18: 2558.
Acknowledgements
We thank Alejandra Ga´mez, Marilyse Charbonneau, Melanie Hur-
tubise and Stephanie Aubin for their technical help and the Canadian
Institutes for Health Research and BioMarin Pharmaceutical Inc.,
CA., USA, for financial support.
5. Deng C, Yin X, Zhang L, Zhang X. Development of microwave-
assisted derivatization followed by gas chromatography/mass
spectrometry for fast determination of amino acids in neonatal
blood samples. Rapid Communications in Mass Spectrometry 2005;
19: 2227.
6. Shedlovsky A, McDonald JD, Symula D, Dove WF. Mouse
models of human phenylketonuria. Genetics 1993; 134: 1205.
7. Hachey DL, Patterson BW, Reeds PJ, Elsas LJ. Isotopic
determination of organic keto acid pentafluorobenzyl esters
in biological fluids by negative chemical ionization gas
chromatography/mass spectrometry. Analytical Chemistry 1991;
63: 919.
REFERENCES
1. Sarkissian CN, Shao Z, Blain F, Peevers R, Su H, Heft R,
Chang TMS, Scriver CR. A different approach to treatment of
phenylketonuria: phenylalanine degradation with recombinant
phenylalanine ammonia lyase. Proceedings of the National Academy
of Sciences of the United States of America 1993; 96: 2339.
2. Hardy DT, Hall SK, Preece MA, Green A. Quantitative
determination of plasma phenylalanine and tyrosine by
electrospray ionization tandem mass spectrometry. Annals of
Clinical Biochemistry 2002; 39: 73.
3. Lee H, Park S, Lee G. Determination of phenylalanine in human
serum by isotope dilution liquid chromatography/tandem mass
spectrometry. Rapid Communications in Mass Spectrometry 2006;
20: 1913.
8. Snapper I, Yu TF, Chiang YT. Cinnamic acid metabolism in man.
Proceedings of the Society for Experimental Biology and Medicine 1940;
44: 30.
Copyright  2007 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2007; 42: 811–817
DOI: 10.1002/jms