ANALYTICAL BIOCHEMISTRY 187,234-239 (1990)

The Measurement of Dimethylamine, Trimethylamine,

and Trimethylamine A/-Oxide Using Capillary Gas
Chromatography-Mass Spectrometty
Kerry-Ann daCosta,* J. James Vrbanac,? and Steven H. Zeisel*
*Nutrient Metabolism Laboratory, Departments of Pathology & Pediatrics, Boston University School of Medicine,
85 East Newton Street, Room M1002, Boston, Massachusetts 02118, and TDrug Metabolism Research,
The Upjohn Company, Kalamazoo, Michigan 49001

Received December 5,1989

(l-4), which is a potent carcinogen in a wide variety of

We have developed a method for measuring dimethyl- animal species (5). These two aliphatic amines are found
amine (DMA), trimethylamine (TMA), and trimethyl- in human and rat urine, blood, and gastric fluid (1,6,7).
amine N-oxide (TMAO) in biological samples using gas DMA is minimally metabolized in mammals and is ex-
chromatography with mass spectrometric detection. creted intact in the urine (8). TMA is normally oxidized
DMA, TMA, and TMAO were extracted from biological within liver and is excreted in the urine as trimethyl-
samples into acid after internal standards (labeled with amine N-oxide (TMAO) (9). The TMA content of sea-
stable isotopes) were added-p-Toluenesulfonyl chloride foods is commonly used to assessfreshness, as TMA is
was used to form the tosylamide derivative of DMA. formed by bacteria from TMAO in fish muscle and im-
2,2,2-Trichloroethyl chloroformate was used to form
parts a characteristic “fishy” odor (10-14).
the carbamate derivative of TMA. TMAO was reduced
Various procedures exist for the analysis of these
with titanium(II1) chloride to form TMA, which was
then analyzed. The derivatives were chromatographed methylamines, including thin-layer chromatography
using capillary gas chromatography and were detected (9,15), ion chromatography (16,17), calorimetric assays
and quantitated using electron ionization mass spec- (18,19), gas chromatography (12,20-27), and high-per-
trometry (GC/MS). Derivative yield, reproducibility, formance liquid chromatography (13,28,29). These ex-
linearity, and sensitivity of the assay are described. isting methods either lack sensitivity, specificity, or re-
The amounts of DMA, TMA, and TMAO in blood, urine, producibility. DMA and TMA are gases at room
liver, and kidney from rats and humans, as well as in temperature and are easily lost during sample storage.
muscle from fishes, were determined. We also report the In addition, these amines tend to adhere to many sur-
use of this method in a pilot study characterizing di- faces and many chromatographic procedures are subject
methylamine appearance and disappearance from to severe “tailing” and “ghosting” (6). We have devel-
blood in five human subjects after ingesting [13C]di- oped a method for the measurement of DMA, TMA, and
methylamine (0.5 wmol/kg body wt). The method we de- TMAO in tissues and biological fluids using capillary gas
scribe was much more reproducible than existing gas chromatography with mass spectrometric detection.
chromatographic methods and it had equivalent sensi- This assay is preferable to existing methods because the
tivity (detected 1 pmol). The derivatized amines were
derivatized amines are stable during storage, there is no
much more stable and less likely to be lost as gases when
ghosting or tailing, and internal standards labeled with
samples were stored. Because we used GC/MS, it was
possible to use stable isotopic labels in studies of methyl- stable isotopes can be used to correct for variations in
amine metabolism in humans. 0 1990 Academic Press, Inc. recovery and for metabolic tracer studies.

MATERIALS AND METHODS

Dimethylamine (DMA)l and trimethylamine (TMA)
are important precursors of N-nitrosodimethylamine Reagents. All reagents were obtained from Fisher
Chemicals (Springfield, NJ) unless otherwise noted.
1 Abbreviations used: DMA, dimethylamine; TMA, trimethylamine; Stock solutions of DMA and TMA (HCl salts, Sigma
TMAO, trimethylamine N-oxide; FID, flame ionization detector. Chemical Co., St. Louis, MO), TMAO (Sigma), di-

234 0003-2697/90 $3.00

Copyright 0 1990 by Academic Press, Inc.
All rights of reproduction in any form reserved.
 CHROMATOGRAPHIC MEASUREMENT OF METHYLAMINES 235

W,),-N Cd-0-CH, -Ccl, (CH,),-NH cl-O-4 oCH3

trlmethylamlne 2,2,2-trlchloro~thylchloroformate dlmsthylamlne+chlo rt de

electron
loniratlon
(CH& -N-C-0-CH, -CCl,
N,Kdtmathyl-2,2,2-trlchloroethylcarbsmate

electron
Ionization 0+
(CH,),-N -O-@H,
o+
N,N-dimethyl-ptOluen0 sultonamlde
ii
(CH,),-N-C (m/z 72) (m/z 199)
FIG. 1. Derivatization of TMA and DMA. DMA and TMA were derivatized so as to form N,N-dimethyl-p-toluene sulfonamide and N,N-
dimethyl-2,2,2-trichloroethyl carbamate. These were isolated using gas chromatography and fragmented using electron impact ionization.

methyl[‘HG]amine HCl (DMA-ds, 98 at.% D, Aldrich Control samples of urine (24-h collections) and blood
Chemical Co., Milwaukee, WI), dimethyl[13C,]amine were collected from humans eating a normal diet. Five
HCl ([13C]DMA, 99.2 at.% 13C, MSD Isotopes, Mon- normal humans (male, ages 24-36 years) were also given
treal, Canada), trimethyl[2HG]amine HCl {TMA-ds, an oral dose of [13C]DMA (0.5 hmol/kg body wt), and
(CD,H),N; 99 at.% D, MSD Isotopes}, and TMAO-dG urine was collected at timed intervals after the dose.
(prepared from TMA-d, with hydrogen peroxide (30)) Urine was collected and stored as described above for rat
were made up in 0.1 N HCl and stored at 4°C. These urine.
solutions were stable over several months. The derivati- Fresh fish was purchased locally and analyzed that
zation agents, p-toluenesulfonyl chloride (2 M solution day. A sample was also stored at -20°C for several
diluted in toluene) and 2,2,2-trichloroethyl chlorofor- months before analysis.
mate, were purchased from Aldrich Chemical. Metha- Sample preparation and extraction of amines. Deu-
nolic alkali was prepared by dissolving 2.8 g KOH in 100 terated internal standards (DMA-ds, TMA-d,, TMAO-
ml methanol/water (3/l, v/v). The reducing reagent was $; as appropriate) were added to all samples to be
prepared with titanium(II1) chloride (Aldrich), 1 N HCl, analyzed. Before urine samples were assayed, they were
and 10% formalin (l/1/2, v/v; must be prepared fresh filtered with a Sep Pak (Waters Associates, Milford,
under a nitrogen atmosphere). MA; 92-94% of DMA, TMA, or TMAO were recovered).
Experimental. Male Sprague-Dawley rats (300-400 Tissue samples were sonicated (setting 5, 50% pulse;
g, Charles River Breeders, Wilmington, MA) were Heat Systems Ultrasonics, Inc., Model W-225R, Plain-
housed in individual metabolic cages in a controlled en- view, NY) in ice-cold 0.9% NaCl (l/5, w/v). An aliquot
vironment (24”C, 12 h of light from 6 AM to 6 PM). They of sonicate containing 100 mg tissue was added to 0.6 ml
were offered food (Ralston Purina rat chow No. 5001, 0.1 N HCl and 3 ml chloroform to precipitate protein.
Farmers Exchange, Framingham, MA) and water ad Zibi- Aliquots of whole blood (1 ml) were added to 0.6 ml 0.1
turn. Urine was collected, over a 24-h period, into acid N HCl and 3 ml chloroform to precipitate protein. Fish
(so that the final concentration was approximately 0.1 (500 mg) was added to 5 ml 0.1 N HCl and mixed thor-
N HCl). The volumes were recorded, and aliquots were oughly (sonication was not required due to the soft con-
stored at -20°C. After urine collection was completed, sistency of fish). Samples were mixed and then subjected
the rats were anesthetized with diethyl ether, and blood to centrifugation at 2000g for 20 min at 4°C. The super-
was drawn by intracardiac puncture. Blood samples natants were aspirated and used for analyses.
were collected into tubes containing heparin and kept on Derivatization and analysis of DMA, TMA, and
ice. Liver and kidney samples were collected by freeze- TMAO (see Fig. 1). For DMA analysis, N,N-dimethyl-
clamping the specimen between tongs that had been pre- p-toluene sulfonamide was formed (10). A portion of the
cooled in liquid nitrogen. They were stored at -95°C un- acidified biological fluid or extract was placed into a glass
til used. test tube sealed with a Teflon-lined septum (WISP sep-
236 DACOSTA, VRBANAC, AND ZEISEL

Jon 72 amu 4 turn, Waters Associates, Milford, MA). Toluene (1 ml

t
3.OES- 1.2ES- 0’ for urine and fish, 0.5 ml for all other samples) and p-
5 2.5E5- l.OES-
(CH,) 2-N-! toluenesulfonyl chloride (38 mg for urine and fish; 3.8
2 2.0ES- 6.OE4 - mg for liver, kidney, and blood) were added. KOH (1 ml;
a l.SES- B.OE4- -12 65% in water, w/v) was injected into the tube, bringing
1 .OES- 4.OE4- the pH above 9. Tubes were incubated at 90°C for 2.5 h,
5.OE4- 2.OE4-
II I 111 I
vigorously shaken, and then subjected to centrifugation
IL
1 1 I i I I I 1 at 1OOOg for 5 min. An aliquot (1~1) of the toluene layer
3.1 4.5 60 80 100 120 140 160 180 200 220
was injected onto the gas chromatograph/mass spec-
trometer (GC/MS).
In order to measure TMA, N,N-dimethyl-2,2,2-tri-
Q)
2
3.OES
2SE5 (C’H,H),-N-C
71’ chloroethyl carbamate was formed from DMA and TMA
(31). A portion of the acidified biological fluid or extract
; 2.OE5
was placed into a glass test tube sealed with a Teflon-
5 1.5E5
lined septum (Waters Associates). Toluene (1 ml for
n l.OE5
u urine samples, 0.5 ml for all other samples) was then
5.OE4
added. KOH (1 ml; 65% in water, w/v) was injected into
3.1 4.5 60 80 100 120 140 160 180 200 220
the tube. Tubes were incubated at 90°C for 1 h, vigor-
ously shaken, and then subjected to centrifugation at
a Ion 199 amu 4
1OOOg for 5 min at 4°C. An aliquot (0.5 ml) of the cooled
0+
toluene layer was added to a sealed vial containing 50 ~1
1.8E5- (CH,),-N -O-&H,
b - 9’\
199
2,2,2-trichloroethyl chloroformate and 10 mg anhydrous
5 1.5E5- 4.5E5-
sodium carbonate. Tubes were incubated at 110°C for 30
o 1.2E5-
min. Methanolic alkali (1 ml) was added after cooling,
5 #.OE4- 3.OE5-
and the reaction mixture was mixed. Water (2 ml) was
n b.OE4-
a
I\ then added, and the sample mixed and subjected to cen-
3.OE4- 1.5E5-
I. - I- IL trifugation at 1OOOg for 5 min at 4°C. An aliquot (1 ~1)
, I I r-l 1 I I +
5.2 6.7 60 80 100 120 140 160 180 200
of the toluene layer was injected onto the GC/MS.
TMAO in the acidified samples was reduced to TMA
0' (32). A portion of the acidified biological fluid or extract
was placed into a glass test tube sealed with a Teflon-
Q) 1.6E5
lined septum (Waters Associates). An equal volume of
1.5E5
ti reducing reagent was added under an atmosphere of ni-
g 1.2E5
trogen and the tube was sealed and incubated at room
5 0.OE4
temperature for 1 h. The TMA formed was analyzed as
9 &OE4 described earlier.
3.OE4
Gas chromatography. We used a Hewlett-Packard
6.2 6.7 60 80 100 120 140 160 180 200 5890/5970 GC/MS equipped with an automatic liquid
sampler. This operates in the electron ionization mode.
,lon 205 amu
A fused silica capillary column (12.5 m X 0.2 mm i.d.
d 0’
dimethyl silicone bonded-phase HP-l; film thickness,
1.6E5- (C%JiN-O-!+H,
8
9’1
205
0.33 pm; Hewlett-Packard, Andover, MA) was used in
5 1.5E5- 4.5E5-
these analyses. GC conditions were as follows: 0.75 min,
w 1.2E5-
splitless mode; He, carrier gas; temperature gradient,
5 9.OE4- 3.OE5-
from 80 to 280°C at 25”C/min. Fragments formed by
s 6-oE4e
[\ electron impact ionization were quantitated using se-
3.OE4 1.5E5-
!A .I I lected ion monitoring. The molecular ions at m/z 199
I I I 1 , I I IJ
5.2 6.7 60 80 100 120 140 160 180 200
(DMA), m/z 201 (formed from [13C]DMA, and m/z 205
Time (min) Mass/Charge (formed from DMA-de) and the fragments m/z 72
(formed from TMA), m/z 76 (formed from TMA-dG),
FIG. 2. Mass spectra obtained from fragmentation of DMA and and m/z 78 (formed from DMA-&) were monitored (see
TMA. Methylamines were isolated and derivatized (TMAO and its
deuterated standard were first reduced to TMA) as described in the
Figs. 1 and 2).
text. They were separated by GC/MS using a silica-bonded capillary Calculations. DMA concentrations were determined
column and a temperature gradient program. The compounds were directly from analysis of the tosylamide derivatives.
fragmented by electron ionization mass spectrometry. TMA concentrations were calculated using
 CHROMATOGRAPHIC MEASUREMENT OF METHYLAMINES 237

TABLE 2

Concentrations of DMA, TMA, and

TMAO in Various Tissues

Tissue DMA TMA TMAO

nmol/g wet wt

Rat blood 4 40 92
Theoretical pmoles Rat liver 136 437 633
Rat kidney 161 531 0
FIG. 3. Linearity of assays for DMA, TMA, and TMAO. Authentic 4 40 21
Human blood
standards of DMA, TMA, and TMAO were derivatized and quanti-
tated using GC/MS as described under Materials and Methods. Corre-
rmol/kg body wt/24 h
lation coefficients were calculated using linear regression analysis and
were equal to 0.9960 for all three compounds. Rat urine 27 2 14
Human urine 6 1 10

TMA (nmol/ml)
fimol/lOO g wet wt
- { DM&,, x area(DMA) * ml/nmol}
= are%rbamate) 9 Shrimp 10 9 171
area(TMA) * ml/nmd Tuna (canned) 84 247 57
Cod 97 44 3400
where Cod (frozen) 381 218 1958

arqcarbamate)= area at mass 72 Note. DMA, TMA, and TMAO were measured as described in text.
DMAtos = concentration (nmol/ml) of DMA in sample Data are expressed as means of triplicate determinations.
as determined using the toluenesulfonyl chloride
method
area(DMA) = area generated by 1 nmol/ml internal
(Fig. 3). Coefficients of variation for intrasample repro-
standard of isotopically labeled DMA that was con- ducibility were 1.26, 1.96, and 1.26% for DMA, TMA,
verted to the carbamate derivative
and TMAO, respectively. Coefficients of variation for in-
=alTMA) = area generated by 1 nmol/ml internal tersample reproducibility were 3.30, 1.31, and 1.78% for
standard of isotopically labeled TMA that was converted
DMA, TMA, and TMAO, respectively. When authentic
to the carbamate derivative. standards of DMA, TMA, and TMAO were added to
TMAO concentrations were calculated as the differ- samples of urine, blood, and rat liver, recoveries of 93-
ences in TMA content after and before reduction. 103% were observed (Table 1).
The peak contours from derivatized DMA, TMA, and
RESULTS AND DISCUSSION TMAO exhibited modest tailing and no ghosting (Fig. 2;
The assays as described were able to detect as little as common problems in other gas chromatography assays
1 pmol of each amine and remained linear up to 250 pmol

TABLE 1

Recovery of Methylamine Standards from

Biological Fluids and Tissue

Percentage recovery

Urine Blood Liver

DMA 100 f 2.1 99 t 1.8 100 t 1.7

TMA 95 f 1.7 103 + 1.7 101 f 0.4
TMAO 100 f 0.01 101 f 0.3 93 * 1.0 10 20 30 40 50
Time After Dose (hr)
Note. Authentic standards were added to biological samples (250 FIG. 4. Kinetics of excretion of [‘%]DMA in urine of humans. An
nmol of DMA or TMAO was added to 1 ml blood or urine or 100 mg oral dose (0.5 Gmol/kg body wt) of [‘%]DMA was administered to five
liver; 100 nmol of TMA was added to 1 ml of blood or urine; 400 nmol normal humans. Urine was collected into acid in blocks corresponding
of TMA was added to 100 mg liver). Methylamines were determined to the intervals noted in the figure. [‘aCIDMA was analyzed using the
as described in text. Data are expressed as the mean recoveries f stan- methods described in the text. Data are expressed as means f stan-
dard deviation of triplicate determinations. dard error of the mean.
238 DACOSTA, VRBANAC, AND ZEISEL

for these amines). Once the derivatives were formed, after the dose, and most of the isotopic label was ex-
they were stable at room temperature or at -20°C for 2 creted in urine over a 48-h period.
weeks. Ninety-nine percent of the DMA was converted In summary we have developed a GC/MS assay for
to the tosylamide, and 95% of the TMA was converted DMA, TMA, and TMAO which is sensitive, specific, and
to NJ-dimethyl-2,2,2-trichloroethyl carbamate. The reproducible. Because we used mass spectrometry, isoto-
reducing reagent used in TMAO analyses did not alter pically labeled internal standards could be used to cor-
the efficiency of conversion of TMA to the carbamate rect for losses during extraction, derivatization, and gas
derivative. chromatography. In addition, the assay is suitable for
Betaine, ethylamine, phosphatidylcholine, and am- metabolic studies in humans using isotopically labeled
monia did not interfere with the assay. However, at con- methylamines.
centrations 10 times higher than those of the amines of
interest, betaine aldehyde, phosphocholine, and choline ACKNOWLEDGMENTS
formed some dimethyltrichloroethyl carbamate (co.1 %
We thank Drs. John Otis and Daniel R. Knapp for their advice. This
conversion), while 0.5% of carnitine was converted. work was supported by grants from the National Institutes of Health
Dimethylglycine was the only compound that formed (CA26731, RR-00533).
some N,N-dimethyl-p-toluene sulfonamide (~0.2% con-
verted). Choline, which is also present in biological fluids REFERENCES
and tissues, will break down to form TMA when heated
above 110°C. Therefore, it is important that this temper- 1. Zeisel, S. H., daCosta, K., and LaMont, J. T. (1988) Carcinogenesis
9,179-181.
ature limit is not exceeded during derivatization.
2. Lijinsky, W., Keefer, L., Conrad, E., and Van de Bogart, R. (1972)
DMA, TMA, and TMAO concentrations were mea- J. Natl. Cancer Inst. 49,1239-1249.
sured in human urine and blood and in rat urine, blood, 3. Lijinsky, W., and Taylor, H. W. (1977) Food Cosmet. Z’onicol. 15,
liver, and kidney (Table 2). Our values for DMA, TMA, 269-274.
and TMAO in human and in rat urine were consistent 4. Ohshima, H., and Kawabata, T. (1978) ZARC Sci. Publ. 19,143-
with those published in the literature (7,11,18,33-35). 153.
Some of the tissues analyzed had not been previously 5. Bartsch, H., and Montesano, R. (1984) Curcinogenesis 5, 1381-
examined for methylamines or TMAO. Our DMA values 1398.
were lower than those which we and others have pre- 6. Zeisel, S. H., dacosta, K., and Fox, J. G. (1985) Biochem. J. 232,
403-408.
viously reported in blood (6,36), and our TMA values
7. Al-Waiz, M., Mitchell, S. C., Idle, J. R., and Smith, R. L. (1987)
were higher (1). We believe these differences reflect im-
Toxicology 43,117-121.
proved calculation of recovery with our method. We
8. Asatoor, A. M., and Simenhoff, M. L. (1965) Biochim. Biophys.
compared an existing gas chromatography method (6) Acta 111,384-392.
which uses a packed column and FID detection with the 9. Al-Waiz, M., Mitchell, S. C., Idle, J. R., and Smith, R. L. (1987)
GC/MS method. The two methods gave comparable re- Xenobiotica 17,551-558.
sults for DMA and TMA, but the packed column method 10. Singer, G. M., and Lijinsky, W. (1976) J. Agric. Food Chem. 24,
was much more variable for TMAO determinations 550-553.
(data not shown). This occurred because, in the GC-FID 11. Zeisel, S. H., and dacosta, K. (1986) Cancer Res. 46, 6136-
method, there was a variable loss of internal standard 6138.
(isopropylamine) during the reduction of TMAO. In the 12. Lundstrom, R. C., and Racicot, L. D. (1983) J. Assoc. Off Anal.
Chem. 66,1158-1163.
GC/MS method we used TMAO-dG as an internal stan-
13. Lin, J-K., and Lai, C-C. (1980) Anal. Chem. 52,630-635.
dard. Seafoods are a rich source of dietary amines, and
14. Lundstrom, R. C., Correia, F. F., and Wilhelm, K. A. (1982) J.
we analyzed several species for their DMA, TMA, and
Food Biochem. 6,229-241.
TMAO contents (Table 2). We found that the DMA and
15. Churacek, J., Pechova, H., Tocksteinova, D., andzikova, Z. (1972)
TMA concentrations in fresh fish were two to three J. Chromatogr. 72,145-152.
times lower than those present in processed (canned) or 16. Bagnasco, S. (1985) Anal. Biochem. 149,572-574.
frozen fish. As would be expected, fresh fish had higher
17. Bouyoucos, S. A. (1977) Anal. Chem. 49,401-403.
concentrations of TMAO than did canned or frozen fish.
18. Blau, K. (1961) Biochem. J. 80,193-200.
Our methods for measuring DMA, TMA, and TMAO
19. Dyer, W. J. (1945) J. Fish Res. Board Canad 6,351-358.
are suitable for studying the metabolism of these amines
20. Miller, A., III, Scanlan, R. A., Lee, J. S., and Libbey, L. M. (1972)
in humans because we can use a nonradioactive isotopic J. Agric. Food Chem. 20,709-711.
label. In a pilot experiment, in which five humans in- 21. Brewster, M. A., and Schedewie, H. (1983) Ann. Clin. Lab. Sci. 13,
gested [13C]DMA, we observed that DMA was rapidly 20-24.
absorbed from the intestine and excreted into urine (Fig. 22. Lowis, S., Eastwood, M. A., and Brydon, W. G. (1983) J. Chroma-
4). Peak enrichment of urine was observed within 30 min togr. 278,139-143.
 CHROMATOGRAPHIC MEASUREMENT OF METHYLAMINES 239

23. Dalene, M., Mathiasson, L., and Jonsson, J. A. (1981) J. Chroma- 30. Hickinbottom, W. J. (1957) Reactions of Organic Compounds, p.
togr. 207,37-46. 420, Longmans, Green, New York.
24. Brand, J. M., and Galask, R. P. (1986) Obstet. Gynecol. 68, 682- 31. Hartvig, P., Ahnfelt, N. O., and Karlsson, K. E. (1976) Acta
685. Pharm. Suet. 13,181-189.
25. Dunn, S. R., Simenhoff, M. L., and Wesson, L. G., Jr. (1976) Anal. 32. Cohen, J., Krupp, M., and Chidsey, C. (1958) Amer. J. Physiol.
Chem. 48,41-44. 194,229-235.
33. Al-Waiz, M., Ayesh, R., Mitchell, S. C., Idle, J. R., and Smith,
26. Zeisel, S. H., daCosta, K., Edrise, B. M., and Fox, J. G. (1986)
R. L. (1987) Clin. Pharmacol. Ther. 42,608-612.
Carcinogenesis 7, 775-778.
34. Lowis, S., Eastwood, M. A., and Brydon, W. G. (1985) &it. J.
27. DiCorcia, A., and Samperi, R. (1974) Anal. Chem. 46,977-981. Nutr. 54,43-51.
28. Elskamp, C. J., and Schultz, G. R. (1986) Amer. Znd. Hyg. Assoc. 35. Zeisel, S. H., Gettner, S., and Youssef, M. (1989) Food Chem. Toni-
J. 47,41-49. col. 27,31-34.
29. Baba, S., and Watanabe, Y. (1988) Anal. Biochem. 175, 252- 36. Ishiwata, H., Iwata, R., and Tanimura, A. (1984) Food Chem. Tox-
257. icol. 22.649-653.