Clinical Chemistry 45:4
539 –548 (1999)
Endocrinology and
Metabolism
1
H-NMR Spectroscopy of Body Fluids: Inborn
Errors of Purine and Pyrimidine Metabolism
Ron A. Wevers,1* Udo F.H. Engelke,1 Sytske H. Moolenaar,1 Christa Bräutigam,2
Jan G.N. de Jong,1 Ries Duran,3 Ronney A. de Abreu,1 and Albert H. van Gennip4
Background: The diagnosis of inborn errors of purine
and pyrimidine metabolism is often difficult. We examined the potential of 1H-NMR as a tool in evaluation of
patients with these disorders.
Methods: We performed 1H-NMR spectroscopy on 500
and 600 MHz instruments with a standardized sample
volume of 500 mL. We studied body fluids from 25
patients with nine inborn errors of purine and pyrimidine metabolism.
Results: Characteristic abnormalities could be demonstrated in the 1H-NMR spectra of urine samples of all
patients with diseases in the pyrimidine metabolism. In
most urine samples from patients with defects in the
purine metabolism, the 1H-NMR spectrum pointed to
the specific diagnosis in a straightforward manner. The
only exception was a urine from a case of adenine
phosphoribosyl transferase deficiency in which the accumulating metabolite, 2,8-dihydroxyadenine, was not
seen under the operating conditions used. Similarly,
uric acid was not measured. We provide the 1H-NMR
spectral characteristics of many intermediates in purine
and pyrimidine metabolism that may be relevant for
future studies in this field.
Conclusion: The overview of metabolism that is provided by 1H-NMR spectroscopy makes the technique a
valuable screening tool in the detection of inborn errors
of purine and pyrimidine metabolism.
Proton nuclear magnetic resonance (1H-NMR)5 spectroscopy has been used to measure metabolites, drugs, and
toxic agents in body fluids (1, 2 ). This requires minimal
sample pretreatment and no derivatization or extraction
steps. The technique provides an overview of the quantitatively most important proton-containing low-molecular
mass compounds in a sample. Splitting and relaxation
properties are among the factors that may contribute to
the detection limit for a compound. For most compounds
detected, quantitative analysis is possible. The technique
has been used to study various inborn errors of metabolism (3– 8 ). Disorders of amino acid and organic acid
metabolism have been characterized with 1H-NMR spectroscopy (4 ). To our knowledge, however, there has been
no systematic study of the use of 1H-NMR spectroscopy of
body fluids from patients with inborn errors in purine or
pyrimidine metabolism. 1H-NMR spectra in some individual cases have been published (9 –12 ).
The aim of this study was to collect the 1H-NMR
spectra of various intermediates in purine and pyrimidine
metabolism. Furthermore, we wanted to study the diagnostic performance of the technique in the examination of
body fluids from patients suspected to suffer from inborn
errors of metabolism in these pathways. The nine enzyme
defects studied and their places in the metabolic pathways are explained in Fig. 1.
© 1999 American Association for Clinical Chemistry
1
1
Institutes of Neurology and Paediatrics, University Hospital Nijmegen,
6525 GC Nijmegen, The Netherlands.
2
University Marburg, Department of Neuropediatrics and Metabolic Diseases, D-35037 Marburg, Germany.
3
University Paediatric Hospital Utrecht, Laboratory of Metabolic Diseases,
NL-3512 LK Utrecht, The Netherlands.
4
Laboratory for Genetic Metabolic Disease, Academic Medical Centre,
NL-1105 A2 Amsterdam, The Netherlands.
*Address correspondence to this author at: University Hospital Nijmegen,
Institute of Neurology, Reinier Postlaan 4, 6525 GC Nijmegen, The Netherlands. Fax 31-24-3540297; e-mail [email protected].
Received July 31, 1998; accepted January 27, 1999.
5
Nonstandard abbreviations: 1H-NMR, proton nuclear magnetic resonance; CSF, cerebrospinal fluid; TSP, trimethylsilyl-2,2,3,3-tetradeuteropropionic acid, sodium salt; SAICA, succinylaminoimidazole carboxamide; APRT,
adenine phosphoribosyltransferase; XDH, xanthine dehydrogenase; AO, aldehyde oxidase; HGPRT, hypoxanthine-guanine phosphoribosyltransferase;
ADA, adenosine deaminase; and PNP, purine-nucleoside phosphorylase.
Materials and Methods
H-NMR spectroscopy in cerebrospinal fluid (CSF) and in
plasma was performed on 500 and 600 MHz Bruker
spectrometers essentially as described previously (5, 7 ).
NMR spectra of urine samples were recorded with the
same spectrometers at 298 K using 60° pulses and a
repetition time pulse interval of 6 s (128 averages; 32 768
539
540
Wevers et al.: 1H-NMR Spectroscopy
Fig. 1. The relevant metabolic pathways of purine (A) and pyrimidine (B) metabolism.
Enzymes involved are: adenylosuccinate lyase (1), ADA (2), PNP (3), XDH (4), HGPRT (5), APRT (6), dihydropyrimidine dehydrogenase (7), dihydropyrimidinase (8), and
b-ureidopropionase (9). PRPP, 5-phospho-a-D-ribosyl-1-pyrophosphate; SAICAR, SAICA-ribotide; AICAR, aminoimidazole carboxamide ribotide; FAICAR, formylaminoimidazole carboxamide ribotide; S-AMP, adenylosuccinate; S-adenosine, succinyladenosine.
data points per scan; sweep width, 6605 Hz). For urine
samples, only minimal sample preparation was required.
Derivatization, extraction, and deproteinization were not
necessary. The pH of each sample was adjusted with a
minimal volume of HCl to pH 2.50 6 0.05.
For quantification, trimethylsilyl-2,2,3,3-tetradeuteropropionic acid (TSP; sodium salt) was added as internal
standard in a final concentration of 2.02 mmol/L (50 mL of
TSP in D2O into 500 mL of urine). The sample volume in
the NMR spectrometer was standardized at 500 mL. The
water resonance was presaturated during the relaxation
delay (6 s). For shimming of the magnetic field, the 29Si-1H
long-range coupling of 3 Hz in the TSP resonance was
used.
For data analysis, a Sine-Bell squared filter (SBB 5 2)
was used. Spectra were Fourier transformed after the free
induction decay was zero-filled to 65 536 data points. The
phase and baseline of the spectra were corrected manually. Resonances were fitted semi-automatically to a
Lorentzian line shape model function. Integrals of these
fits were used for metabolite quantification. Metabolites
in urine are expressed per millimole of creatinine. For
analysis of the spectra, 1D WinNMR and WinFit software
were used (Bruker Analytische Messtechnik).
When certain compounds are described in this study as
“NMR invisible”, it means that in the relevant concentration range no characteristic signals could be observed
under the 1H-NMR spectroscopy standard operating conditions used. The detection limit of 1H-NMR spectroscopy
for each metabolite depends on the number of protons
that contribute to a signal and on the multiplicity of the
resonance (5, 7 ). When the molecule contains a methyl
group, the estimated detection limit is in the range between 5 mmol/L for a singlet resonance and 10 mmol/L
for a doublet resonance. Some purines and pyrimidines
have only one proton contributing to the signal. The
estimated detection limit of the technique in such a case is
15 mmol/L for a singlet resonance and 30 mmol/L for a
doublet resonance. Some metabolites have no contributing proton in their chemical structure. For this reason,
metabolites such as uric acid and 2,8-dihydroxyadenine
are NMR invisible. It is conceivable that uric acid can be
detected by 1H-NMR spectroscopy under different experimental conditions (e.g., at a different pH). Obviously, this
541
Clinical Chemistry 45, No. 4, 1999
would be of utmost importance for diagnosing purine
inborn errors. However, because such conditions are
likely to introduce other problems in the interpretation of
the NMR spectra, the standard operating procedure that
we have used for a long time in diagnosing inborn errors
of metabolism was also used in the present study.
Model compounds used in this study are available
from Sigma Chemical Co., except for 5-acetyl-amino-6formyl-amino-3-methyl uracil (kindly provided by Prof.
H.A. Simmonds, Guy’s Hospital, London, UK) and succinylaminoimidazole carboxamide (SAICA)-riboside and
succinyladenosine (kindly provided by Prof. G. van den
Berghe, Université Catholique de Louvain, Brussels, Belgium).
The diagnoses of all patients used in this study originally were made in the period 1981–1998 at various
metabolic screening laboratories in The Netherlands, Germany, and Belgium. Samples were stored at 220 or
280 °C until analysis. These laboratories all used standard HPLC techniques (reversed-phase column, quantification by reading the absorbance at 260 nm) for measurement of purines and pyrimidines in body fluids.
Results
quantitative performance of nmr spectroscopy
The quantitative data found with NMR spectroscopy are
compared with the HPLC results for relevant metabolites
in Table 1. The HPLC results are the quantitative data on
the sample that were available from the laboratory that
originally diagnosed the patient. For some samples, this
means that .10 years may have elapsed between the
HPLC and NMR measurements. Unfortunately, the limited volumes of most of the samples made repeating the
HPLC measurement simultaneously with NMR spectroscopy impossible. The results of both techniques compare
rather well for most of the metabolites, although rather
large discrepancies were observed for uracil, thymine,
inosine, deoxyinosine, and deoxyguanosine.
interpretation of the 1h-nmr spectrum
It is advantageous that many of the relevant compounds
in purine and pyrimidine metabolism have a chemical
shift between 5.5 and 9.0 ppm, a region of the spectrum
where only few metabolites present in body fluids have
resonances. Other known metabolites in this part of the
spectrum are tyrosine, phenylalanine, formic acid, indoxyl sulfate, histidine, and hippuric acid. A list of
1
H-NMR the resonances of these compounds is available
(5, 7 ). Table 2 shows the resonances of the purine and
pyrimidine metabolites that are relevant under nondiseased physiological conditions or in patients with inborn
errors of metabolism in these pathways. The most important resonances of a molecule for its identification in the
1
H-NMR spectrum have been listed in Table 2. Some
molecules have only one resonance, whereas others posses multiple resonances. These can be recognized by their
complete “fingerprints” in the spectrum.
Table 1. Comparison of NMR and HPLC measurementsa of
urine samples of patients with various defects in purine
and pyrimidine metabolism.
Concentration in urine,
mmol/mmol creatinine
NMR
HPLC
Metabolic defect
Allopurinol
630
1007
Deoxyadenosine
Deoxyguanosine
Deoxyinosine
Guanosine
Hypoxanthine
205
322
522
588
271
227
22
185/167b
177
333
490
194
171
14/47b
1022
1110
214
1048
523
298
162
126
188
283
101
157
Thymine
335
54
Uracil
626
342
Xanthine
229
843
431
182
794
286/643b
358
957
Molybdenum
cofactor
ADA
PNP
PNP
PNP
Xanthine oxidase
Xanthine oxidase
Molybdenum
cofactor
HGPRT
PNP
Adenylosuccinate
lyase
Adenylosuccinate
lyase
Adenylosuccinate
lyase
Adenylosuccinate
lyase
Dihydropyrimidine
dehydrogenase
Dihydropyrimidine
dehydrogenase
XDH
XDH
Molybdenum
cofactor
HGPRT
Metabolite
Inosine
SAICA-riboside
Succinyladenosine
a
HPLC values were not measured simultaneously with the NMR values (see
Results).
b
Values from separate screening laboratories.
Generally there is only minimal intersample variation
in the chemical shift of a certain resonance (,0.005 ppm),
which allows reliable identification of the various purines
and pyrimidines in urine, plasma, and CSF. The only
exception was the extremely variable chemical shift of the
singlet resonance from the proton on carbon-2 of the
imidazole ring of SAICA-riboside, which varied between
8.05 and 8.26 ppm between samples. The SAICA-riboside
doublet resonance at 5.77 ppm has a stable resonance
position, thus allowing reliable identification of this metabolite.
inborn errors of pyrimidine metabolism
Dihydropyrimidine dehydrogenase (EC 1.3.1.2) deficiency. Dihydropyrimidine dehydrogenase is the initial enzymatic
step in the degradative pathway of the pyrimidine bases
uracil and thymine (Fig. 1B). The enzyme converts both
compounds into their respective 5,6-dihydro derivatives.
1
H-NMR spectra were obtained from urine samples of
four unrelated cases with dihydropyrimidine dehydroge-
Wevers et al.: 1H-NMR Spectroscopy
542
Table 2. Most important 1H-NMR resonances from relevant
purine and pyrimidine metabolites.
Metabolite
Adenine
Adenosine
Adenylosuccinic acid
b-Alanine
Allopurinol
Allopurinol-1-riboside
5-AMEUb
3-Aminoisobutyric acid
Caffeine (1,3,7trimethylxanthine)
Cytidine
Cytosine
Deoxyadenosine
Deoxyguanosine
Deoxyinosine
Deoxythymidine
(Thymidine)
4,5-Dihydroorotic acid
5,6-Dihydrothymine
5,6-Dihydrouracil
5,6-Dihydrouridine
2,8-Dihydroxyadenine
Guanine
Guanosine
2-Hydroxyadenine
8-Hydroxyadenine
5-Hydroxymethyluracil
Hypoxanthine
Inosine
Orotic acid
Orotidine
Oxypurinol
Oxypurinol-1-riboside
Oxypurinol-7-riboside
Pseudouridine
Ribose
SAICA-riboside
Succinyladenosine
(S-adenosine)
S-Sulfocysteine
Theophylline
Thymine
Trigonelline
Uracil
3-Ureidoisobutyric acid
3-Ureidopropionic acid
Uric acid
Uridine
Xanthine
Xanthosine
Chemical shift (multiplicity)
8.38
6.16
6.19
2.78
8.16
6.29
2.15
1.26
3.32
(s); 8.43 (s)a
(d); 8.45 (s); 8.55 (s)
(d); 8.44 (s); 8.60 (s)
(t); 3.27 (t)
(s); 8.31 (s)
(d); 8.24 (s); 8.26 (d)
(s); 3.19 (s)
(d); 2.87 (m); 3.13 (m); 3.23 (m)
(s); 3.50 (s); 3.93 (s); 7.88 (s)
5.88
6.13
6.54
6.34
6.49
1.89
(d); 6.23 (d); 8.12 (d)
(d); 7.72 (d)
(t); 8.43 (s); 8.51 (s)
(t); 8.39 (s)
(t); 8.22 (s); 8.36 (s)
(s); 2.26 (t); 6.28 (t); 7.56 (s)
2.97 (AB); 4.35 (m)
1.19 (d); 2.80 (m); 3.19 (dd); 3.50 (dd)
2.68 (t); 3.50 (t)
2.74 (t); 5.83 (d)
1
H-NMR invisible
8.59 (s)
5.95 (d); 8.39 (s)
Model compound not available
Model compound not available
4.34 (s); 7.56 (s)
8.21 (s); 8.39 (s)
6.09 (d); 8.22 (s); 8.36 (s)
6.22 (s)
5.54 (d); 5.77 (s)
8.26 (s)
Model compound not available
Model compound not available
7.66 (s)
3.46–4.22 (various resonances), 4.92
(d), 5.24 (d), 5.37 (d)
5.77 (d); 8.05–8.26 (s)
6.12 (d); 8.40 (s); 8.45 (s)
3.61 (AB); 4.26 (dd)
3.35 (s); 3.55 (s); 8.01 (s)
1.85 (d); 7.36 (q)
4.44 (s); 8.10 (t); 8.87 (q); 9.19 (s)
5.79 (d); 7.41 (d)
Model compound not available
2.57 (t); 3.36 (t)
1
H-NMR invisible
5.58 (d); 7.85 (d)
7.99 (s)
3.93 (d); 5.89 (d); 7.93 (s)
a
s, singlet; d, doublet; t, triplet; m, multiplet; AB, AB system; dd, doubletdoublet; q, quartet.
b
5-Acetyl-amino-6-formyl-amino-3-methyl uracil (metabolite of caffeine).
nase deficiency. An example of a 1H-NMR spectrum of a
urine sample of one of these cases has been published
previously (12 ). In all samples, uracil and thymine were
clearly increased (uracil, 95– 626 mmol/mmol creatinine;
thymine, 270 – 470 mmol/mmol creatinine). In nondiseased urine samples, thymine cannot be detected with
NMR spectroscopy, whereas uracil is observed in trace
amounts in some samples (,10 mmol/mmol creatinine).
5-Hydroxymethyluracil, a metabolite of thymine, was not
detectable in the samples.
Dihydropyrimidinase (EC 3.5.2.2) deficiency. Dihydropyrimidinase (5,6-dihydropyrimidine amidohydrolase) is the
second enzyme in the breakdown of the pyrimidine bases
uracil and thymine. It catalyzes the conversion of 5,6dihydrouracil to 3-ureidopropionic acid and of 5,6-dihydrothymine to 3-ureidoisobutyric acid (Fig. 1B). To our
knowledge, only seven cases of this enzyme defect have
been described. We measured urine, plasma, and CSF in
three patients with this defect. Two of these patients have
been described clinically [case 1 by Putman et al. (9 ); case
2 by Assmann et al. (10 )], whereas case 3 was diagnosed
recently. Examples of 1H-NMR spectra of urine and CSF
of these patients have been published previously (9, 10 )
and, therefore, are not shown here. Diagnostically high
concentrations of uracil, thymine, and dihydropyrimidines were found with 1H-NMR spectroscopy in the urine
of all three patients (concentrations of uracil, thymine,
5,6-dihydrouracil, and 5,6-dihydrothymine: 9 –144, 44 –
230, 178 –760, and 132– 490 mmol/mmol creatinine, respectively). The CSF of cases 1 and 2 showed obviously
increased concentrations for the dihydropyrimidines, but
only slightly increased concentrations for uracil and thymine (concentrations of uracil, thymine, 5,6-dihydrouracil, and 5,6-dihydrothymine: not detectable, trace to 10
mmol/L, 46 –117 mmol/L, and 79 –179 mmol/L, respectively). The CSF of the third patient was not available. For
all four compounds, the concentrations in plasma were
obviously lower than in CSF (9 ). The absence of high
concentrations of 3-ureidopropionic acid (N-carbamyl-balanine; model compound, Sigma cat. no. C3750) in the
urine or CSF of these patients excluded b-ureidopropionase deficiency as a possible defect. NMR findings were
compatible with dihydropyrimidinase deficiency, which
was later confirmed enzymatically in the liver of cases 1
and 2 (10, 13, 14 ). A liver biopsy of case 3 was not
available.
b-Ureidopropionase (EC 3.5.1.6) deficiency. b-Ureidopropionase catalyzes two reactions: the conversion of 3-ureidopropionic acid and 3-ureidoisobutyric acid into b-alanine
and 3-aminoisobutyric acid, respectively. A deficiency of
the enzyme has not yet been recognized in humans.
However, resonances of both substrates should be visible
in the 1H-NMR spectra of body fluids of a patient with
this defect. Table 2 gives the characteristic resonances of
3-ureidopropionic acid.
Clinical Chemistry 45, No. 4, 1999
inborn errors of purine metabolism
Adenine phosphoribosyltransferase (APRT; EC 2.4.2.7) deficiency. APRT is a purine salvage enzyme that catalyzes the
conversion of adenine to AMP (Fig. 1A). When the
enzyme is deficient, adenine is not salvaged to AMP, but
instead is oxidized to 8-hydroxyadenine and 2,8-dihydroxyadenine by xanthine dehydrogenase (XDH). APRT
deficiency is characterized by 2,8-dihydroxyadenine urolithiasis. Patients have increased concentrations of 2,8dihydroxyadenine in the urine. The compound, however,
is 1H-NMR invisible under the conditions used. We have
measured a urine sample of an adult case with this inborn
error and found no unusual resonances. The patient did
not use allopurinol medication. The lactic acid concentration in the sample was very high (7 mmol/mmol creatinine). Adenine, 2-hydroxyadenine, and 8-hydroxyadenine may be increased in this disease. Detection of
2-hydroxyadenine and 8-hydroxyadenine is hampered by
the fact that they are not available as model compounds.
They were not observed in the NMR spectrum of the
urine of our patient. Similar results were obtained when
the sample was measured at its native pH (pH 8.2). A
plasma sample of the patient showed high creatinine,
indicating renal dysfunction. However, no abnormal resonances that may be indicative for the underlying metabolic defect could be observed.
Lesch-Nyhan disease/hypoxanthine-guanine phosphoribosyltransferase (HGPRT; EC 2.4.2.8) deficiency. HGPRT also is a
purine salvage enzyme, converting hypoxanthine and
guanine into IMP and GMP, respectively (Fig. 1A). In the
deficiency state, guanine and hypoxanthine are instead
converted into xanthine and finally into uric acid. Uric
acid overproduction is a characteristic of Lesch-Nyhan
disease. Uric acid is a trioxypurine without protons on its
carbon atoms; therefore, it is 1H-NMR invisible. Increased
hypoxanthine and xanthine are expected to be the only
characteristic abnormalities in the NMR spectrum in body
fluids of untreated Lesch-Nyhan patients. Urine samples
of three classical Lesch-Nyhan cases were measured. Two
of these patients were on allopurinol medication. In the
1
H-NMR spectrum of the patient without allopurinol,
xanthine and hypoxanthine were clearly increased (xanthine, 37 mmol/mmol creatinine; hypoxanthine, 198
mmol/mmol creatinine; reference values for both compounds, ,4.1 mmol/mmol creatinine). In other cases of
Lesch-Nyhan disease, xanthine in urine may be within the
reference interval, but hypoxanthine will always be increased. In the patients that used medication, the excretion was higher (xanthine, 312– 827 mmol/mmol creatinine; hypoxanthine, 855-1363 mmol/mmol creatinine)
because feedback inhibition of de novo purine synthesis
by IMP and GMP does not take place because of the
HGPRT deficiency. Guanine was not detectable, whereas
orotidine was clearly increased in both samples (112 and
63 mmol/mmol creatinine). This can be explained by the
inhibitory effect of oxypurinol, which is generated from
543
allopurinol by the enzymatic action of XDH or aldehyde
oxidase (AO; Fig. 2) on orotate phosphoribosyl transferase. Resonances from allopurinol-1-ribonucleotide and
allopurinol-1-riboside were not observed because the conversion of allopurinol into these metabolites requires the
action of HGPRT, which is deficient in these patients (Fig.
2). In the CSF of the untreated patient, xanthine and
hypoxanthine were both increased (22 and 94 mmol/L
respectively). In most CSF samples, both compounds are
not observed in the NMR spectrum because the reference
intervals for both are below the detection limit of the
technique under the conditions used.
Adenosine deaminase (ADA; EC 3.5.4.4) deficiency. ADA
catalyzes the conversion of adenosine into inosine. The
enzyme also accepts deoxyadenosine as a substrate, converting it to deoxyinosine (Fig. 1A). Deoxyadenosine and
adenosine therefore are the metabolites that are of diagnostic importance. In urine, deoxyadenosine generally is
present in higher concentrations than adenosine in patients with ADA deficiency. We investigated the urine of
one patient with this disease. The NMR spectrum showed
the presence of high concentrations of deoxyadenosine
(205 mmol/mmol creatinine; reference value, ,1 mmol/
mmol creatinine). Adenosine was not detectable with
NMR spectroscopy, but was within the reference interval
when the sample was analyzed with HPLC. Orotic acid
was not observed in the NMR spectrum. A complication
in the NMR spectrum was the presence of an unknown
doublet resonance at 6.54 ppm, which interfered with the
6.54 triplet resonance from deoxyadenosine. Determining
whether the doublet resonance is a characteristic feature
for this metabolic defect requires measurement of additional samples from ADA-deficient cases.
Purine-nucleoside phosphorylase (PNP; EC 2.4.2.1) deficiency.
PNP catalyzes the reversible phosphorylation of inosine
and deoxyinosine or guanosine and deoxyguanosine to
give either hypoxanthine or guanine plus the corresponding ribose-1-phosphate (Fig. 1A). Therefore, inosine,
Fig. 2. Allopurinol metabolism.
Enzymes involved are XDH, AO, HGPRT, orotate phosphoribosyl transferase
(OPRT), and pyrimidine-59-nucleotidase (PY59N).
544
Wevers et al.: 1H-NMR Spectroscopy
guanosine, and their deoxy forms are increased in body
fluids in the PNP-deficient state. Fig. 3A shows the
1
H-NMR spectrum of a urine of a patient with this
deficiency. Urine samples from two cases were available.
Both patients had the typical clinical signs and symptoms
of PNP deficiency (15, 16 ). The enzyme defect was shown
in erythrocytes and fibroblasts in both cases. We found
increased concentrations for inosine (case 1, 49 mmol/
Fig. 3. 600 MHz 1H-NMR spectra of urine samples of patients with PNP
deficiency (case 2; A), xanthinuria during allopurinol treatment (B), or
adenylosuccinate lyase deficiency (C).
mmol creatinine; case 2, 1110 mmol/mmol creatinine),
guanosine (case 1, 35 mmol/mmol creatinine; case 2, 588
mmol/mmol creatinine) and deoxyinosine [case 1, trace
(,5 mmol/mmol creatinine); case 2, 522 mmol/mmol
creatinine]. In controls, these compounds are found in the
urine in very low concentrations (,1 mmol/mmol creatinine). Deoxyguanosine was not observed in the urine of
case 1, but could clearly be observed in the urine of case
2 (322 mmol/mmol creatinine; reference, ,1 mmol/mmol
creatinine). Xanthosine was not found in the urine samples of the two patients. The metabolite concentrations
between the two patients showed remarkable differences.
In addition, other differences between the patients were
observed. Increased orotic acid (95 mmol/mmol creatinine; reference, ,10 mmol/mmol creatinine) was observed in case 1, but not in case 2. Orotidine was not
detectable in these samples. The patients had not used
allopurinol medication. Very high, as yet unexplained,
resonances (doublet resonances at 5.33 and 5.40 ppm and
singlet resonances at 5.60, 5.63, 6.08, and 6.63 ppm) were
found in the sample of case 1. In case 2, the 5.40 ppm
doublet resonance and the 6.63 ppm singlet resonance
were also observed, although these resonances were considerably lower than in the urine of case 1. The 6.08 ppm
resonance may be present in the urine of case 2, but this
resonance would overlap with the doublet resonance
deriving from the very high concentration of inosine in
this sample. The urine sample of case 2 contained very
high additional unknown resonances (singlet resonance at
2.26 ppm and doublet resonance at 2.96 ppm) that were
not present in the urine of case 1. A plasma sample from
another patient had clearly increased inosine (case 3, 40
mmol/L; reference, ,0.1 mmol/L), whereas guanosine
and both deoxy compounds were not detectable. Except
for the 6.08 ppm singlet resonance, the other unknowns
observed in the urine of the other two patients were not
detectable in this blood sample. It would require measurement of urine samples from other patients with this defect
to determine whether these unknown resonances are
caused by the metabolic block.
Xanthinuria. In classical xanthinuria, a deficiency of XDH
(EC 1.1.1.204) exists. The enzyme converts hypoxanthine
into xanthine and xanthine into uric acid (Fig. 1A).
Furthermore, a combined XDH/AO deficiency is known.
Patients with the combined defect have an inability to
convert allopurinol to oxypurinol (Fig. 2). In both diseases, xanthine is the predominant purine excreted. Urine
samples of three patients from two families with isolated
XDH deficiency could be measured both before and
during allopurinol therapy. In one patient, isolated XDH
deficiency was confirmed enzymatically. In urines from
three patients not on medication, the increased xanthine
concentration was the only characteristic feature in the
1
H-NMR spectrum (154 – 890 mmol/mmol creatinine; Fig.
3B), and hypoxanthine was slightly increased (37–217
mmol/mmol creatinine). In four additional urine samples
545
Clinical Chemistry 45, No. 4, 1999
of two of these patients while on allopurinol treatment, a
similar picture was observed. The xanthine concentrations tended to be even higher than before treatment
(507–941 mmol/mmol creatinine). This may relate to the
competitive inhibition of oxypurinol deriving from allopurinol on the residual XDH activity. Resonances from
allopurinol (14 –1176 mmol/mmol creatinine) and oxypurinol (255–367 mmol/mmol creatinine) were observed
in the 1H-NMR spectrum of these four samples (Fig. 3B).
The presence of oxypurinol illustrates that the patients
have isolated XDH deficiency because oxypurinol cannot
be formed in the combined XDH/AO deficiency because
this requires AO enzymatic activity (Fig. 2). Allopurinol1-riboside could also be detected (31–161 mmol/mmol
creatinine). In the urine of one patient while on allopurinol, the orotidine concentration was 30 mmol/mmol creatinine. We were unable to demonstrate the presence of
oxypurinol-1-riboside or oxypurinol-7-riboside while the
patients were on allopurinol. Their detection with 1HNMR spectroscopy is hampered by the fact that they are
not available as model compounds.
Molybdenum cofactor deficiency. Patients with molybdenum
cofactor deficiency have an inability to synthesize the
pteridyl moiety of the cofactor essential for the enzymatic
activity of XDH, sulfite oxidase, and AO. Xanthinuria,
hypouricemia, and increased urinary sulfite, S-sulfocysteine, and thiosulfate concentrations are characteristic for
the disease. Xanthinuria can be demonstrated easily in the
1
H-NMR spectrum. Serum and CSF samples were available from one patient with this defect. Xanthine was
increased in both body fluids (case 1: serum xanthine, 136
mmol/L; CSF xanthine, 24 mmol/L). Hypoxanthine was
not clearly increased in these samples. Urine was available of four patients with this defect (cases 2–5). In three
of these patients without medication (cases 2– 4), xanthine
was clearly increased (450 – 498 mmol/mmol creatinine),
whereas hypoxanthine was only slightly increased (47–56
mmol/mmol creatinine). Xanthosine was detected in only
one of these urine samples, in a low concentration (80
mmol/mmol creatinine). An additional patient (case 5)
described by van Gennip et al. (17 ) was on allopurinol
treatment because of the existence of renal xanthine
stones. In this case, clearly increased xanthine was found
in three urine samples during allopurinol treatment (xanthine, 275–726 mmol/mmol creatinine), whereas hypoxanthine was only slightly increased (22– 67 mmol/mmol
creatinine). Allopurinol was found in two of the three
samples, whereas all three samples contained allopurinol1-riboside (314 – 834 mmol/mmol creatinine). No oxypurinol was found because allopurinol cannot be converted
into oxypurinol in this disease (Fig. 2). Sulfite and thiosulfate are 1H-NMR invisible, and S-sulfocysteine could
not be demonstrated in any of the urines of the patients
with molybdenum cofactor deficiency. For S-sulfocysteine, this may relate to its instability under certain pH
conditions or to the fact that NMR spectroscopy under the
conditions used is too insensitive for its detection. This
relative insensitivity is caused by the multiplicity of the
resonance (4.26 doublet-doublet resonance, four peaks)
and by the fact that only one proton contributes to this
signal. Orotidine was not increased in any of these samples, as expected.
Adenylosuccinate lyase (EC 4.3.2.2) deficiency. Adenylosuccinate lyase, or adenylosuccinase, catalyzes two reactions
in the biosynthesis of purine nucleotides. This concerns
the conversion of SAICA-ribotide into aminoimidazole
carboxamide ribotide and of adenylosuccinate into AMP
(Fig. 1A). A defect in the enzyme leads to accumulation of
two usually undetectable compounds: SAICA-riboside,
deriving from SAICA-ribotide; and succinyladenosine,
deriving from adenylosuccinate. Both compounds can be
detected readily in the urine of affected patients (Fig. 3C).
The 1H-NMR spectrum of urine from two unrelated
patients showed increased succinyladenosine (101 and
188 mmol/mmol creatinine) and SAICA-riboside (162 and
214 mmol/mmol creatinine). Recognition of SAICA-riboside in urine with 1H-NMR spectroscopy may be hampered by the use of the antiepileptic drug Sabril® (Vigabatrin), which produces a characteristic resonance profile
that overlaps with the doublet resonance of SAICAriboside. In such a case, the compound can still be found
in the spectrum through its singlet resonance. In the CSF
of one of the patients, both SAICA-riboside and succinyladenosine were present in detectable amounts (198 and
216 mmol/L, respectively).
Discussion
This study illustrates that many defects in purine and
pyrimidine metabolism can be demonstrated with NMR
spectroscopy of body fluids by the presence of abnormally high concentrations of specific metabolites. In an
earlier study (12 ), we have found unacceptable quantitative differences between HPLC, gas chromatography, and
NMR results for uracil and thymine in urine samples. In
Table 1 of the present study, we have compared HPLC
and NMR results for several metabolites in urine that are
of diagnostic importance in inborn errors of purine and
pyrimidine metabolism. The sometimes very long time
interval (up to 17 years) between the HPLC and the NMR
measurements of course is a serious drawback of this
approach. Table 1 is no more than a rough indication that
for most metabolites the two techniques will give similar
quantitative results. The larger discrepancies found for
uracil, thymine, inosine, deoxyinosine, and deoxyguanosine in part are in line with the differences found
in our earlier study (12 ). The addition of uracil and
thymine to urine samples gave excellent recovery values
with NMR spectroscopy (12 ). Currently we are preparing
a separate validation study on NMR spectroscopy in urine
samples where NMR technical aspects (length of relaxation time and other factors) and NMR performance
(within- and between-run coefficients of variation, accu-
546
Wevers et al.: 1H-NMR Spectroscopy
racy, recovery, and correlation with other techniques) will
be addressed.
NMR spectroscopy measurements in urine samples
were informative in all patients with inborn errors in
pyrimidine metabolism. The 1H-NMR spectra of urine
samples from patients with dihydropyrimidine dehydrogenase and dihydropyrimidinase deficiency led directly
to the diagnosis in all cases. Blood plasma and CSF were
not investigated in dihydropyrimidine dehydrogenase
deficiency; however, in dihydropyrimidinase deficiency,
the diagnosis could be found with 1H-NMR spectroscopy
in both body fluids.
The absence of 3-ureidopropionic acid in the urine
spectra of the patients with dihydropyrimidinase deficiency excluded the theoretical possibility of b-ureidopropionase deficiency. In this case, 1H-NMR spectroscopy
helped to pinpoint the enzyme defect. The defect in these
patients was later confirmed at the enzyme level in a liver
biopsy (10, 13, 14 ). In future cases the use of 1H-NMR
spectroscopy of urine may be considered to confirm this
particular enzyme defect at the metabolite level, omitting
the liver biopsy and the confirmation at enzyme level and
going directly to the DNA level for mutation analysis.
Primary b-ureidopropionase deficiency has as yet not
been found in humans. Secondary b-ureidopropionase
deficiency has been reported in propionic acidemia (18 ).
Characteristic resonances of 3-ureidopropionic acid, one
of the expected metabolites in this deficiency, can be
observed with 1H-NMR spectroscopy. In this way, the
technique may help in diagnosing patients with as yet
unknown inborn errors in purine or pyrimidine metabolism. Whether such resonances in a spectrum would be
properly assigned depends only on the quality of the
available model compound 1H-NMR database. This is
unlike the situation with conventional techniques, where
special and often elaborate sample preparation in combination with thin-layer chromatography or HPLC is required to identify 3-ureidopropionic acid (19, 20 ). To find
as yet unknown inborn errors of metabolism with 1HNMR spectroscopy, the spectra of the model compounds
of as many intermediates relevant in human metabolism
as possible should be measured. This is a topic for further
research.
In the patients with inborn errors of purine metabolism, 1H-NMR spectroscopy of urine samples led to the
diagnosis in all patients with PNP deficiency, xanthinuria,
molybdenum cofactor deficiency, Lesch-Nyhan disease,
ADA deficiency, and adenylosuccinate lyase deficiency.
After allopurinol loading, the NMR spectra of the urine
could discriminate between isolated XDH deficiency and
combined XDH/AO deficiency through the presence of
oxypurinol. Increased urinary xanthine and/or hypoxanthine formed the only characteristic in the 1H-NMR spectrum in all samples from patients with XDH deficiency,
molybdenum cofactor deficiency, and Lesch-Nyhan disease. It, therefore, was impossible to discriminate on the
basis of the 1H-NMR spectrum between these diagnoses
without the use of other additional tests. In this respect it
is a major disadvantage that uric acid is 1H-NMR invisible
and therefore cannot aid discrimination between these
inborn errors of purine metabolism. APRT deficiency was
the only enzyme defect in purine metabolism where no
abnormality could be observed in the urinary 1H-NMR
spectrum. This is mainly because the characteristic metabolite 2,8- dihydroxyadenine is NMR invisible. The diagnosis of this patient would have been missed with 1HNMR spectroscopy of urine.
For some of the inborn errors in purine and pyrimidine
metabolism, no 1H-NMR spectra were recorded. Patients
with myoadenylate deaminase deficiency and patients
with pyrimidine-59-nucleotidase deficiency have no characteristic metabolite profile in body fluids, although increased concentrations of pyrimidine nucleotides have
been demonstrated in erythrocytes with 1H-NMR spectroscopy of pyrimidine-59-nucleotidase-deficient patients
(11 ). No samples were available of patients with orotic
aciduria type I or II. These defects can be diagnosed on
basis of the increased concentrations of orotic acid in
urine. Orotic acid can be detected easily in the 1H-NMR
spectrum (singlet resonance at 6.22 ppm); therefore, these
conditions cannot be missed with 1H-NMR spectroscopy.
In addition, no sample was available from patients with
phosphoribosyl pyrophosphate synthase superactivity.
Because hypoxanthine in urine is characteristically increased in this disease, we expect no difficulties in diagnosing this defect with 1H-NMR spectroscopy.
In PNP deficiency, as yet unidentified resonances were
demonstrated in all body fluid samples investigated. We
never have observed these unknowns in .400 other body
fluid samples (urine, plasma, and CSF) of controls or
patients suspected to have inborn errors of metabolism.
The resonances are likely to derive from several compounds. No other metabolites are known to occur in
increased concentrations in body fluids of patients with
PNP deficiency. Furthermore, there are to our knowledge
no published reports indicating high concentrations of
unknown metabolites in this disease. The urines of cases
1 and 2 were stored at 280 °C for several years; therefore,
storage artifacts should be considered. However, because
some of these unknowns were also found in the serum of
case 3, we consider this option unlikely because this
sample was measured with NMR spectroscopy immediately after venipuncture. There were obvious differences
in the NMR spectrum of the urine of cases 1 and 2. The
concentrations of the four characteristic metabolites in this
disease varied considerably between the two patients, and
orotic aciduria was present only in case 1. Some of the
unknown resonances were present in the urine of both
patients, but again obvious concentration differences
were present. Other unknowns were present in only one
of the two patients. These data allow various explanations. Theoretically, the unknowns can be caused by
medication, derive from the food, or may actually be a
characteristic of the underlying enzyme deficiency. It
547
Clinical Chemistry 45, No. 4, 1999
would be conceivable that the patients may suffer from
different metabolic diseases with partially overlapping
metabolite profiles. However, this seems unlikely because
the enzyme deficiency was shown convincingly in the
patients (15, 16 ). The occurrence of unknown resonances
should be checked in urine samples from other patients
with the disease. Because PNP deficiency is very rare, we
would appreciate if urine, plasma, or CSF samples of
affected patients would be made available for 1H-NMR
spectroscopy. In case our findings can be confirmed in
other cases, additional NMR techniques (COSY, J-resolved, and long-range spectra) may help to solve the
chemical structure of the unknown metabolites involved.
Orotic aciduria is a more or less controversial finding
in PNP deficiency. It was described for the first time by
Cohen et al. (21 ) in two unrelated patients with the
disease and confirmed by van Gennip et al. (22 ). Simmonds et al. (23 ), however, could not confirm it with
three independent methods. The mechanism of the orotic
aciduria is unknown. It could be caused by the inhibition
of orotate phosphoribosyltransferase by accumulating abnormal metabolites. The enzyme converts orotic acid and
5-phospho-a-d-ribosyl-1-pyrophosphate into orotidine 5phosphate. It is known that inosine does not inhibit the
enzyme in vitro. Perhaps one of the unknown metabolites
found in our patients can cause this inhibitory effect.
Another explanation for the presence of orotic acid is the
higher availability of 5-phospho-a-d-ribosyl-1-pyrophosphate, which leads to increased pyrimidine synthesis de
novo and overflow at orotate phosphoribosyl transferase,
the rate-limiting step in this pathway. NMR spectroscopy
of urine of affected patients may uncover the role that the
as yet unknown metabolites play in causing orotic aciduria in these patients.
The overall view on metabolism that 1H-NMR spectroscopy provides is nicely illustrated by the various
inborn errors of metabolism discussed here. The technique allows investigation of a body fluid sample from a
patient clinically suspected to suffer from an inborn error
of metabolism without focusing on a specific group of
metabolites. Conventional techniques in metabolic screening provide information on specific groups of metabolites
(amino acids, organic acids, polyols, and purines). Each of
these requires specific methodology such as derivatization, extraction, chromatography steps, or detection techniques. The 1H-NMR spectrum provides quantitative information on representatives from all of these metabolite
groups without the use of derivatization or extraction
steps. This is the major advantage of 1H-NMR spectroscopy. Unfortunately, the spectrometer required for the
technique is still very expensive. Another disadvantage is
that the software required for the interpretation of the
spectra cannot yet automatically recognize, assign, and
quantify the many resonances present in a spectrum of
complex matrices such as body fluids. Therefore, the
interpretation of the spectra remains the most laborious
part of the analysis. This aspect together with the high
price of high-field strength spectrometers severely discourages the routine application of the technique in
clinical chemistry.
We thank Dr. H.A. Simmonds (Guy’s Hospital, London,
UK) for the kind gift of 5-acetyl-amino-6-formyl-amino-3methyl uracil, and Prof. G. van den Berghe and Dr. C.
Vermylen (both from the University of Louvain Medical
School, Brussels, Belgium) and Dr. B. Poorthuis (University Hospital Leiden, The Netherlands) for making available samples of affected patients.
References
1. Nicholson JK, Wilson ID. High resolution proton magnetic resonance spectroscopy of biological fluids. Prog Nucl Magn Reson
Spectrosc 1989;21:449 –501.
2. Bell JD, Brown JCC, Sadler PJ. NMR studies of body fluids. Nucl
Magn Reson Biomed 1989;2:246 –55.
3. Wevers RA, Engelke U, Heerschap A. High-resolution 1H-NMR
spectroscopy of blood plasma for metabolic studies. Clin Chem
1995;41:1054 – 6.
4. Lehnert W, Hunkler D. Possibilities for selective screening for
inborn errors of metabolism using high resolution 1H-FT-NMR
spectrometry. Eur J Pediatr 1986;145:260 – 6.
5. Wevers RA, Engelke U, Wendel U, de Jong JGN, Gabreëls FJM,
Heerschap A. Standardized method for high-resolution 1H-NMR of
cerebrospinal fluid. Clin Chem 1995;41:744 –51.
6. Abeling NGGM, van Gennip AH, Bakker HD, Heerschap A, Engelke
U, Wevers RA. Diagnosis of a new case of trimethylaminuria using
direct proton NMR spectroscopy of urine. J Inherit Metab Dis
1995;18:182– 4.
7. Wevers RA, Engelke U, Heerschap A. High resolution 1H-NMR
spectroscopy of blood plasma for metabolic studies. Clin Chem
1994;40:1245–50.
8. Iles RA, Chalmers RA. Nuclear magnetic resonance spectroscopy
in the study of inborn errors of metabolism. Clin Sci 1988;74:
1–10.
9. Putman CWMM, Rotteveel JJ, Wevers RA, van Gennip AH, Bakkeren
JAJM, de Abreu RA. Dihydropyrimidinase deficiency: a progressive
neurological disorder? Neuropediatrics 1997;28:106–10.
10. Assmann B, Hoffmann GF, Wagner L, Bräutigam C, Seyberth HW,
Duran M, et al. Dihydropyrimidinase deficiency and congenital
microvillous atrophy: coincidence or genetic relation? J Inherit
Metab Dis 1997;20:681– 8.
11. Kagimoto T, Shirono K, Higaki T, Oda T, Matsuzaki H, Nagata K, et
al. Detection of pyrimidine 59-nucleotidase deficiency using 1H or
31
P nuclear magnetic resonance. Experientia 1986;42:69 –72.
12. Wevers RA, Engelke U, Rotteveel JJ, Heerschap A, de Jong JGN,
Abeling NGGM, et al. 1H NMR spectroscopy of body fluids in
patients with inborn errors of purine and pyrimidine metabolism.
J Inherit Metab Dis 1997;20:345–50.
13. van Gennip AH, de Abreu RA, Van Lenthe H, Bakkeren J, Rotteveel
J, Vreken P, van Kuilenburg ABP. Dihydropyrimidinase deficiency:
confirmation of the enzyme defect in dihydropyrimidinuria. J Inherit
Metab Dis 1997;20:339 – 42.
14. van Gennip AH, de Abreu RA, Vreken P, van Kuilenburg ABP.
Clinical and biochemical aspects of dihydropyrimidinase deficiency. Adv Exp Biol 1998;431:125– 8.
548
Wevers et al.: 1H-NMR Spectroscopy
15. Rijksen G, Kuis W, Wadman SK, Spaapen LJM, Duran M, Voorbrood BS, et al. A new case of purine nucleoside phosphorylase
deficiency: enzymological, clinical and immunological characteristics. Pediatr Res 1987;21:137– 41.
16. Stoop JW, Zegers BJM, Hendrickx GFM, Siegenbeek van Heukelom LH, Staal GEJ, de Bree PK, et al. Purine nucleoside phosphorylase deficiency associated with selective cellular immunodeficiency. New Engl J Med 1977;296:651–5.
17. van Gennip AH, Mandel H, Stroomer LEM, van Cruchten AG. Effect
of allopurinol on the xanthinuria in a patient with molybdenum
cofactor deficiency. Adv Exp Med Biol 1995;370:375– 8.
18. van Gennip AH, Van Lenthe H, Abeling NGGM, Scholten EG, van
Kuilenburg ABP. Inhibition of b-ureidopropionase by propionate
may contribute to the neurological complications in patients with
propionic acidaemia. J Inherit Metab Dis 1997;20:379 – 82.
19. van Gennip AH, Busch S, Elzinga L, Stroomer AEM, van Cruchten
A, Scholten EG, Abeling NGGM. Application of simple chromato-
20.
21.
22.
23.
graphic methods for the diagnosis of defects in pyrimidine degradation. Clin Chem 1993;39:380 –5.
van Gennip AH, Busch S, Scholten EG, Stroomer AEM, Abeling NG.
Simple method for the quantitative analysis of dihydropyrimidines
and N-carbamyl-b-amino acids in urine. Adv Exp Med Biol 1991;
309B:15–9.
Cohen A, Staal GE, Ammann AJ, Martin DW. Orotic aciduria in two
unrelated patients with inherited deficiencies of purine nucleoside
phosphorylase. J Clin Investig 1977;60:491– 4.
van Gennip AH, Grift J, de Bree PK, Zegers BJM, Stoop JW,
Wadman SK. Urinary excretion of orotic acid, orotidine and other
pyrimidines in a patient with purine nucleotide phosphorylase
deficiency. Clin Chim Acta 1979;93:419 –28.
Simmonds HA, Potter CF, Sahota A, Cameron JS, Webster DR,
Becroft DM. Absence of orotic aciduria in adenosine deaminase
deficiency and purine nucleoside phosphorylase deficiency. Clin
Exp Immunol 1978;34:42–5.