Clinica Chimica Acta 365 (2006) 264 – 269

www.elsevier.com/locate/clinchim

Validation of 1H NMR spectroscopy as an analytical tool

for methylamine metabolites in urine
Martin B. Lee a,b, Malina K. Storer a,c, John W. Blunt b, Michael Lever a,b,*
a
Biochemistry Unit, Canterbury Health Laboratories, P.O. Box 151, Christchurch, New Zealand
b
Department of Chemistry, University of Canterbury, Christchurch, New Zealand
c
Department of Pathology, Christchurch School of Medicine and Health Sciences, University of Otago, Christchurch, New Zealand

Received 28 July 2005; received in revised form 4 September 2005; accepted 5 September 2005
Available online 28 October 2005

Abstract

Background: Methylamines have many metabolic roles and there is an increasing demand for their measurement. Glycine betaine is an
important osmolyte, and a reservoir for methyl groups. Proline betaine and trigonelline are important dietary betaines. Trimethylamine,
derived from gut flora, is normally converted to trimethylamine oxide but in Ffish odour syndrome_ is excreted as TMA. These compounds
are all suitable for quantification by 1H NMR spectroscopy as they all have methyl protons.
Method: Urine samples are acidified and 1H NMR spectra are obtained using presaturation for water suppression. Peak integrals or heights
are compared to an internal standard of acetonitrile.
Results: Inter- and intra-assay CV’s were < 5% for TMAO and creatinine, and < 10% for the other analytes. Responses were linear from 50 to
1000 AM for all metabolites, and recoveries were  97%. Limits of detection using NMR are slightly higher than alternative HPLC assays
(15 – 25 AM). However, sensitivity is adequate for the detection of raised levels in urine, and sample analysis was complete in less than 5 min.
Conclusion: 1H NMR spectroscopy is a convenient, rapid and economical option for the determination of betaines and related compounds in
urine in a single analysis.
D 2005 Elsevier B.V. All rights reserved.

Keywords: 1H NMR; Methylamine; Betaine; Trimethylamine

1. Introduction diagnostic for this condition, and useful for diagnosing

We investigated alternative methods for measuring
methylamine metabolites in urine in response to two clinical
demands. Firstly, there has been increasing recent interest in
betaines and betaine metabolites because of their association
with homocysteine metabolism [1 – 5] and hence with
vascular disease [1,6]. Secondly, trimethylamine (TMA)
excretion is increased in patients suffering from a genetic
disorder known as Ffish odour syndrome’ [7,8]. An increase
in the ratio of TMA to trimethylamine N-oxide (TMAO) is
* Corresponding author. Biochemistry Unit, Canterbury Health Labora-
tories, P.O. Box 151, Christchurch, New Zealand. Tel.: +643 364 0319; fax:
+643 364 0750.
E-mail address: michael.lever@chmeds.ac.nz (M. Lever).
0009-8981/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.cca.2005.09.004
heterozygotes. While this is not a fatal condition, it inflicts a
characteristic odour which can be socially debilitating.
In previous work, we have used the HPLC separation of
aracyl derivatives to measure these metabolites [6,9 – 12].
This is satisfactory for plasma but with urine samples there
are a large number of peaks, and to get adequate resolution
it is necessary to resort to long run-times (> 60 min). Even
so, it is difficult to resolve all the metabolites of interest in
one chromatogram. Additionally, TMA and TMAO give the
same derivative and are not separately quantified. In an
animal study [4] we found that 1H NMR spectroscopy could
be used to measure all the metabolites of interest quickly
and at the same time, and subsequent experience [13]
suggested that this technique could be used to provide a
clinical diagnostic service.
 M.B. Lee et al. / Clinica Chimica Acta 365 (2006) 264 – 269
NMR spectroscopy has been used for the identification
and quantification of a number of metabolites in urine and
plasma. In the early 1980’s, 1H NMR spectroscopy
established itself as a versatile tool for the analysis of
whole urine. The early work of Nicholson et al. and Bales et
al. showed that useful clinical information could be derived
from whole urine NMR work [14,15]. More recently,
sophisticated techniques have been developed to suppress
the water and protein signals which complicated those early
spectra [16]. The use of NMR spectroscopy in clinical
applications is becoming more commonplace, given its
unique ability to identify and quantify multiple metabolites
in a single run. Wide ranging clinical studies using NMR
spectroscopy have been undertaken, including disturbed
metabolism of TMA [8,17], supplementation in athletes
[18], determination of novel metabolic disorders [19], and
observation of renal clearance [20]. Urine betaine and
TMAO analysis by 1H NMR spectroscopy has previously
been problematic since they can easily be confused [21]. We
demonstrate that, for the group of methylamine metabolites
studied, these problems can be resolved and a rapid and
versatile diagnostic assay can be developed.
2. Method
2.1. Materials
Glycine betaine, trigonelline, dimethylglycine (DMG),
trimethylamine, trimethylamine N-oxide, acetonitrile, deu-
terium oxide and HCl were purchased from Sigma-Aldrich
(St Louis, MO). Proline betaine was synthesised using the
method of Cornforth and Henry [22].
2.2. Sample preparation
Four hundred microliters of urine was added to 100 AL of
1 M HCl solution containing acetonitrile at 25 mM as an
internal standard.
2.3. NMR assay
All 1H NMR spectra were recorded on a Varian INOVA
500 instrument, at 23 -C in 5 mm NMR tubes with a 3 mm
D2O lock insert.
For the measurements, a 90- radiofrequency pulse, with a
duration of 8.1 As was used. The delay between pulses was
5 s, acquisition time was 1.982 s. Eight transients were
recorded for each sample, with 30,272 data points for each
transient. Sweep width was set to 8000 Hz. Water
suppression was achieved using a presaturating irradiation
on the water resonance of 3 s during the relaxation delay
period. All spectra were zero filled to 128 k points, and then
Fourier transformed. Phasing and baseline corrections were
completed manually. The software used for these procedures
was VNMR version 6.1C (Varian), which also provides a
265
signal to noise calculation facility for the determination of
the limits of detection.
Quantification of the metabolites was performed using an
internal standard of acetonitrile (5 mM). Peak integrals and
peak heights were measured after baseline and drift
correction and then expressed as a ratio to the acetonitrile
integral.
Trigonelline was subject to the same analysis as the other
betaines, but its proximity to the suppressed water resonance
led to baseline distortions. Peak integrals proved to be
unreliable, and a more reproducible measure was obtained
from the peak height ratio to acetonitrile in a method similar
to Fulton et al. [23].
2.4. Precision and recovery study
To make urine samples with elevated levels of betaines,
100 AL of aqueous standards of the betaines were added
(spiked) into 10 mL of urine. This resulted in urine samples
with elevated concentrations of the methylamines in
question (spikes were: 100 AM trigonelline, 750 AM TMAO,
75 AM glycine betaine and TMA, 140 AM DMG and 2.5
mM creatinine). The unspiked urine sample with low
betaine levels was also used in this precision study. Six
batches with four replicates of the spiked and non-spiked
urine were analysed.
2.5. Linearity
Standards of trigonelline, glycine betaine, proline
betaine, and DMG, were prepared from 50 to 700 AM.
Because of differences in biological concentrations TMAO
was prepared at 100– 7000 AM, TMA at 10 – 700 AM, and
creatinine at 0.25 – 17.5 mM. These standards were
prepared by serial dilution of a 10 mM aqueous stock
standard. Samples were analysed and the linearity of the
method was determined using linear regression analysis
over 6 data points.
3. Results
3.1. Internal standards
We used acetonitrile as an internal standard because its
resonance is removed from the spectral region of interest.
The resonance positions of many metabolites are dependent
on the solution pH, so the use of HCl was required to
ensure that the pH of the samples was consistent between
analyses. At this low pH, the TMA peak was split into a
doublet with coupling observed to the ammonium proton
[24], however the methyl resonances were sufficiently
removed from other metabolites for integration of both
peaks to be achieved. Lowering the pH to ¨1 was required
for separation of TMAO and glycine betaine. A typical
spiked spectrum is shown in Fig. 1. Additionally, carnitine
266 M.B. Lee et al. / Clinica Chimica Acta 365 (2006) 264 – 269

Proline betaine Creatinine

N
O NH2
O-
O- N+
Trigonelline N
O
O Glycine betaine
O-
N+ N+

O
Dimethylglycine
O-
+HN
Trimethylamine N oxide
O
O

N Trimethylamine
H

4.4 4.0 3.6 3.2 2.8 2.4 2.0 nppm

Fig. 1. Spiked (300 AM) urine with methyl resonances for the metabolites studied.

was observed at 3.07 ppm, and does not interfere with any
of the species we measured.
3.2. Spectrum analysis
Baseline distortions, which were introduced after the
water suppression was applied, were problematic for the
methyl resonance for trigonelline which was near the
residual water resonance. The use of peak height data,
which can be extracted after smoothing the baseline, was far
more reliable than peak integral data when applied to
trigonelline [23].
3.4. Limits of detection
Limits of detection (S / N = 3) for each compound are
shown in Table 1. The precision of this NMR method is
good at both low levels (unspiked sample) and high levels
(100 AM).
3.5. Accuracy
Recoveries of betaines were over 95% for all betaines
in urine. The recovery values given in Table 2 show that
all of the betaines of interest can be measured accurately.

3.3. Precision 3.6. Linearity

The coefficient of variation (CV) was below 10% for The NMR assay is linear over the range of 50– 1000
within batch and between batch variation when using the AM for trigonelline, glycine betaine, proline betaine, and
internal standard calibrated data (Table 2). A brief
comparison of peak height and peak integral data was
performed. Table 1
The peak integral data was used for most of the Chemical shift (relative to acetonitrile at 1.9 ppm) and limit of detection for
the metabolites analysed
metabolites presented here, because the baseline was flat
and the peaks were well resolved. In the case of Analyte Chemical shift (ppm) LOD (AM)
trigonelline, the residual water resonance caused a Trigonelline* 4.30 25
significant baseline distortion. Peak height proved to be TMAO 3.37 15
Proline betaine 2.99 25
more reliable, with artificial baseline correction performed Glycine betaine 3.19 15
using the Varian VNMR software. Peak heights were DMG 2.82 20
difficult to obtain for the other metabolites, as the TMA 2.73, 2.72 15
neighboring signals made the same baseline corrections Creatinine 2.97 25
complicated. * Trigonelline results from peak height data.
 M.B. Lee et al. / Clinica Chimica Acta 365 (2006) 264 – 269 267

Table 2 Table 3
Results of precision and recovery study of betaines in urine with betaine TMA and TMAO measurements of two suspected cases of heterozygous
spike and unspiked sample (four batches, n = 6) trimethylaminuria and a normal control
Analyte Mean Within Between Recoveries TMA TMAO Ratio Percent Percent
(AM) batch CV batch CV (%) TMA / TMA / TMA + TMAO / TMA +
TMAO TMAO TMAO
Trigonelline*
Low urine 75 4.8 4.1 Patient 1 30.01 603.8 0.05 4.7 95
Spike 178 6.9 2.3 98 Patient 2 56.22 1102.27 0.04 4.8 95
TMAO Control 2.09 236.05 0.009 0.9 99
Low urine 976 4.3 2.0
Spike 1764 2.5 8.5 105
Proline betaine comparison, a normal control has about 1% clearance of
Low urine 67.0 4.7 2.3
Spike 165 2.9 2.8 98
TMA (Table 3).
Glycine betaine
Low urine 63.1 5.8 6.2
Spike 138 4.5 3.0 100 4. Discussion
Dimethylglycine
Low urine 60.5 6.0 5.7
Spike 197 4.1 3.1 98
Measurement of methylamine metabolites by 1H NMR
TMA spectroscopy is a quick, simple, non-destructive and
Low urine 8.4 7.4 6.6 effective quantitative method, involving minimal sample
Spike 83 4.2 5.1 100 preparation and the capacity to analyse many samples in a
Creatinine short time (2– 3 min/sample). As well as the metabolites
Low urine 3949 0.9 0.7
Spike 6371 0.6 3 97
presented here, the method could be adapted to measure
other methylated substances, for example carnitine (which
* Trigonelline results obtained from peak height data.
does not interfere with the urine components that were the
subject of the present studies). Moreover, when an
DMG, 100 – 7000 AM for TMAO, 10 –700 AM for TMA unknown metabolite occurs in a urine sample, the NMR
and 0.250– 17.5 mM for creatinine. Standard curves for spectrum can provide substantial information about its
these metabolites are shown in Fig. 2. nature. If required 2D NMR spectroscopy can also be run
on the sample with no further preparation and structural
3.7. TMA analysis information gathered.
Previous workers have used NMR spectroscopy for
Two patients with suspected cases of trimethylaminuria measuring these methylamines. Lundberg et al. quantified
were given a load of 600 g of fish and had a urine sample DMG and glycine betaine in the urine of patients with
taken 12 h later. Their TMA and TMAO concentrations premature vascular disease [25]. A new inborn error of
were determined and the percentage clearance of TMA and metabolism giving elevated urine excretion of DMG was
TMAO are given at about 5% for both people. As a identified similarly [19], and we used this technique to

1.6
Metabolite/Acetonitrile integral ratio

1.4

1.2

1.0

0.8

0.6 DMG Standard Curve

Glycine Betaine Standard Curve
Proline Betaine Standard Curve
0.4
Trimethylamine Standard Curve
Trigonelline Standard Curve
0.2

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Metabolite concentration (mmol/L)
1
Fig. 2. Standard curves for the metabolites measured by H NMR spectroscopy. TMAO and creatinine are omitted because of differences in scale, however all
methylamines were linear over the specified ranges.
268 M.B. Lee et al. / Clinica Chimica Acta 365 (2006) 264 – 269
observe changes in DMG excretion in a DMG supplemen-
tation study [26]. Often water is evaporated off and replaced
with deuterium oxide [8], however, it is shown here that the
water signal can be suppressed without this time-consuming
step. Alternative methods that have been used for water
suppression in 1H NMR spectroscopy include Double
Pulsed Field Gradient Spin Echo (DPFGSE), WATERGATE
and WET pulse sequences [4,16]. Presaturation was chosen
here, because none of the signals in question were in rapid
exchange with water. It is essential to lower the pH of the
urine to pH ¨1 to correctly assign the signals. Previously
the resonance caused by TMAO (pK a = 4.65) in the urine of
renal failure patients has been incorrectly attributed to
glycine betaine (pK a = 1.83) [21]. These are resolved after
acidifying the samples, and this is necessary since both are
likely to be present in substantial amounts in normal urine.
The methods described here are being used for studies of
urinary betaines and vascular disease [6,13] which require
simultaneous measurement of glycine betaine and its
metabolite DMG, along with the potentially confounding
dietary betaines [27,28], proline betaine and trigonelline. It
is also being used for the diagnosis of TMAuria and an
example of results with presumed heterozygous subjects is
illustrated (Table 3). However, the approach is far more
general, for example the presence of clearly resolved
carnitine peaks in normal urine samples suggests that
carnitine and acylcarnitines could also be measured. Most
groups have measured methylamine metabolites by combi-
nations of chromatographic procedures with UV, fluores-
cence or mass-spectroscopic detection: 1 H NMR
spectroscopy is a rapid non-destructive alternative.
Acknowledgements
MKS acknowledges the financial assistance of a Tech-
nology for Industry Fellowship from the Foundation of
Research Science and Technology (FRST) and a Todd
Foundation Award for Excellence. MBL acknowledges the
financial support of an Enterprise scholarship from FRST.
We thank Professors Stephen T. Chambers and Peter M.
George for their interest and continued support in the
development of all stages of these assays. We also
appreciate the useful discussions with Dr. Bill Bubb at the
University of Sydney.
References
[1] Olthof MR, Verhoef P. Effects of betaine intake on plasma
homocysteine concentrations and consequences for health. Curr Drug
Metab 2005;6:15 – 22.
[2] Finkelstein J, Martin J. Methionine metabolism in mammals.
Distribution of homocysteine between competing pathways. J Biol
Chem 1984;259:9508 – 13.
[3] Craig SAS. Betaine in human nutrition. Am J Clin Nutr 2004;80:
539 – 49.
[4] Slow S, Lever M, Lee MB, George PM, Chambers ST. Betaine
analogues alter homocysteine metabolism in rats. Int J Biochem Cell
Biol 2004;36:870 – 80.
[5] Melse-Boonstra A, Holm PI, Ueland PM, Olthof M, Clarke R, Verhoef
P. Betaine concentration as a determinant of fasting total homocysteine
concentrations and the effect of folic acid supplementation on betaine
concentrations. Am J Clin Nutr 2005;81:1378 – 82.
[6] Lever M, George PM, Dellow WJ, Scott RS, Chambers ST.
Homocysteine, glycine betaine, and N,N-dimethylglycine in patients
attending a lipid clinic. Metab Clin Exp 2005;54:1 – 14.
[7] Mitchell SC, Smith RL. Trimethylaminuria: the fish malodor
syndrome. Drug Metab Dispos 2001;29:517 – 21.
[8] Maschke S, Wahl W, Azaroual N, et al. 1H NMR analysis of
trimethylamine in urine for the diagnosis of fish-odour syndrome.
Clin Chim Acta 1997;263:139 – 46.
[9] Lever M, Sizeland PCM, Bason L, Hayman CM, Chambers ST.
Glycine betaine and proline betaine in human blood and urine.
Biochim Biophys Acta 1994;1200:259 – 64.
[10] Lever M, Sizeland PCM, Frampton CM, Chambers ST. Short and
long-term variation of plasma glycine betaine concentrations in
humans. Clin Biochem 2004;37:184 – 90.
[11] Dellow WJ, Lever M, Chambers ST, Lunt H, Robson RA. Elevated
glycine betaine excretion in diabetes mellitus patients is associated
with proximal tubular dysfunction and hyperglycaemia. Diabetes Res
Clin Pract 1999;43:91 – 9.
[12] Eisert R, Oftedal OT, Lever M, Ramdohr S, Breier BH, Barrell GK.
Detection of food intake in a marine mammal using marine
osmolytes and their analogues as dietary biomarkers. Mar Ecol Prog
Ser 2005;300:213 – 28.
[13] Slow S, Miller WE, McGregor DO, et al. Trigonelline is not
responsible for the acute increase in plasma homocysteine following
ingestion of instant coffee. Eur J Clin Nutr 2004;58:1253 – 6.
[14] Nicholson JK, Oflynn MP, Sadler PJ, Macleod AF, Juul SM, Sonksen
PH. Proton-nuclear-magnetic-resonance studies of serum, plasma and
urine from fasting normal and diabetic subjects. Biochem J 1984;
217:365 – 75.
[15] Bales JR, Higham DP, Howe I, Nicholson JK, Sadler PJ. Use of high-
resolution proton nuclear magnetic-resonance spectroscopy for rapid
multi-component analysis of urine. Clin Chem 1984;30:426 – 32.
[16] Van QN, Chmurny GN, Veenstra TD. The depletion of protein signals
in metabonomics analysis with the WET-CPMG pulse sequence.
Biochem Biophys Res Commun 2003;301:952 – 9.
[17] Podadera P, Areas JAG, Lanfer-Marquez UM. Diagnosis of suspected
trimethylaminuria by NMR spectroscopy. Clin Chim Acta 2005;351:
149 – 54.
[18] Cartigny B, Azaroual N, Mille-Hamard L, et al. H-1 NMR urine
analysis as an effective tool to detect creatine supplementation. J Anal
Toxicol 2002;26:355 – 9.
[19] Moolenaar SH, Poggi-Bach J, Engelke UFH, et al. Defect in
dimethylgycine dehydrogenase, a new inborn error of metabolism:
NMR spectroscopy study. Clin Chem 1999;45:459 – 64.
[20] Hauet T, Gibelin H, Godart C, Eugene M, Carretier M. Kidney
retrieval conditions influence damage to renal medulla: evaluation by
proton nuclear magnetic resonance (NMR) spectroscopy. Clin Chem
Lab Med 2000;38:1085 – 92.
[21] Grasdalen H, Belton PS, Pryor JS, Rich GT. Quantitative proton
magnetic-resonance of plasma from uremic patients during dialysis.
Magn Reson Chem 1987;25:811 – 6.
[22] Cornforth JW, Henry AJ. The isolation of l-stachydrine from the fruit
of Capparis tomentosa. J Chem Soc [Perkin 1] 1952;1:601 – 2.
[23] Fulton DB, Sayer BG, Bain AD, Malle HV. Detection and determi-
nation of dilute, low-molecular-weight organic-compounds in water
by 500-Mhz proton nuclear-magnetic-resonance spectroscopy. Anal
Chem 1992;64:349 – 53.
[24] Leyden DE, Morgan WR. Kinetics of proton exchange of trimethy-
lammonium ion by NMR — a laboratory experiment. J Chem Educ
1969;46 [169- &].
 M.B. Lee et al. / Clinica Chimica Acta 365 (2006) 264 – 269 269

[25] Lundberg P, Dudman NPB, Kuchel PW, Wilcken DEL. 1H NMR
determination of urinary betaine in patients with premature vascular
disease and mild homocysteinemia. Clin Chem 1995;41:275 – 83.
[26] Slow S, McGregor DO, Lever M, Lee MB, George PM, Chambers ST.
Dimethylglycine supplementation does not affect plasma homocys-
teine concentrations in pre-dialysis chronic renal failure patients. Clin
Biochem 2004;37:974 – 6.
[27] de Zwart FJ, Slow S, Payne RJ, et al. Glycine betaine and glycine
betaine analogues in common foods. Food Chem 2003;83:197 – 204.
[28] Slow S, Donaggio M, Cressey PJ, Lever M, George PM, Chambers
ST. The betaine content of New Zealand foods and estimated intake in
the New Zealand diet. J Food Compos Anal 2005;18:473 – 85.