Sie sind auf Seite 1von 3



Potential Sources of Errors in Cation-Exchange Chromatographic

Measurement of Plasma Taurine
Brian M. Connolly and Harold 0. Goodman

We examined the potential sources of error in automated Materials and Methods

cation-exchange chromatographic quantitation of plasma
taurine, both in sample preparation and in the analysis. Subjects
Principal sources of error include: use of serum instead Volunteers consisted of 10 individuals (six men and four
of plasma, which produces gross overestimates; use of women) who considered themselves to be in good health. In
tripotassium ethylenediaminetetraacetate (EDTA) as addition, 34 patients (23 men, 11 women) participating in a
anticoagulant in systems involving ninhydrin detection (a study of renal stone disease were included. Stone disease is
ninhydrin-positive contaminant of EDTA emerges coinci- not known to affect taurine plasma values.
dent with taurine); contamination with platelets; and placing
volumes exceeding 20 tL on the cartridge used in the Collection of Plasma and Serum
Technicon TSM Amino Acid Analyzer. We arrived at a All blood samples were collected by antecubital venipunc-
simple technique in which we use EDTA as anticoagulant, ture into evacuated blood-collection tubes containing either
micropore filtration to produce platelet-free plasma, and tripotassium ethylenediaminetetraacetate (EDTA), sodium
o-phthalaldehyde as the detection reagent for the sensi- citrate, heparin, or no anticoagulant (serum). The samples
tivity required to measure accurately the low concentration were centrifuged at 650 X g for 15 mm. If the experimental
of taurine in plasma. protocol required further centrifugation, samples were
transferred to plastic centrifuge tubes with a siliconized
Keyphrases:analytical error variation,source Pasteur pipet and centrifuged at 1000 or 10 000 X g for 15 mm.
of disorders of the nervous system
#{149} fluorometry .
Samples requiring filtration were drawn up in a 3-mL plastic
taurine content of platelets reference intervals
syringe and filtered through a 0.45-nm Millex disposable filter
unit (Millipore Co., Bedford, MA 01730). Samples were de-
platelet aggregation and taurine release
protemnized with an equal volume of a 100 g/L sulfosalicylic
acid solution, then centrifuged at 1000 X g for 15 mm. The
Although taurine is one of the most abundant and ubiqui-
supernatant fluid was aspirated and stored at -20 #{176}C until
tous free amino acids in human tissues (1), until recently its
analyzed. Platelets were counted with a “Coulter S-plus”
only well-documented metabolic function has been its con-
counter (Coulter Electronics, Hialeah, FL 33010) and the
jugation to form taurocholic acid. During the past few years,
accuracy was confirmed by microscopic counts on selected
studies of its neuropharmacologic properties (2) have indi-
cated a potentially more significant role in neurotransmission.
In spite of uncertainty as to whether taurine is a neuro-
modulator or inhibitory neurotransmitter (3,4), the consensus Chromatography
is that taurine is a neuroeffector in both the central and pe- Samples were analyzed with a TSM amino acid analyzer
ripheral nervous systems (2). (Technicon Instruments Corp., Tarrytown, NY 10591). We
Abnormal taurine concentrations have been reported in modified the standard manifold to permit simultaneous
tissue and body fluids for several pathological conditions, analysis of duplicate samples by utilizing two standard cat-
including epilepsy, Friedreich’s ataxia, mental depression with ion-exchange columns in parallel. Taurine was eluted at a
or without parkinsonism, Down’s syndrome, bacterial men- column temperature of 45 #{176}C
with a single lithium citrate
ingitis, and leukemia (1, 5, 6). The role of taurine in the buffer (0.3 mol/L, pH 2.70) and emerged 20 mm after the
pathogenesis of these conditions is not now fully under- front. The entire analytical program took 70 mm, including
stood. column wash with 0.3 mol/L lithium hydroxide (16 mm) and
As knowledge and interest in taurine increase, the need column re-equilibration with the eluting 2.70 buffer (34 mm).
arises for accurate measurement of taurine in tissues and body For sample detection we used either the standard ninhydrin
fluids. Cation-exchange chromatography has been the method reagent and the system’s dual-channel colorimeter with
of choice for quantitative taurine analysis. Several method- 570-nm filters, or an o-phthalaldehyde reagent (ii) and de-
ological problems in the preparation of samples (7, 8) and in tection with a filter fluorometer with 360-nm excitation,
chromatographic analysis (9, 10) can lead to inaccurate taurine 455-nm emission (Spectra/Gb; Gilson Med. Electronics,
determinations. The great disparities in normal values for Middleton, WI 53562). A multipen recorder was used for
blood taurine reported in the literature (e.g., 1) suggest that fluorometric analysis, with maximum deflection set at 100 mV.
all measurements reported cannot be valid. Unless otherwise stated, 20.0 fLL of sample or standards was
We have examined various aspects of measurement of cir- absorbed to the cartridge. An external taurine standard (10
culating taurine and report some of our own experiences. We tmol/L) was analyzed every fifth sample and the unknowns
also describe a modification of the Technicon Sequential were quantified by using the average peak height of standards
Multisample (TSM) amino acid analyzer for fluorometric bracketing the unknown. Both detection systems described
analyses of amino acids. give linear signals and, owing to the “spike-like” nature of the
taurine peak, we find that measurement of peak height pro-
vides a reliable estimate of the concentration of taurine in
Section on Medical Genetics, Department of Pediatrics, Bowman samples.
Gray School of Medicine of Wake Forest University, Winston-Salem,
NC 27103. Results
Received Oct. 4, 1979; accepted Dec. 11, 1979. The sampling method used with the TSM can lead to seri-

508 CLINICALCHEMISTRY, Vol. 26, No. 3, 1980

Table 1. Influence of Sample Volume on Peak Height for Taurine
Concn of std., Sample vol., Peak height, % of 1O-L
mmol/L liLa mean (and SEM) n sample SIgnlflcancet
1 10 32.5 (0.5) 12 -

0.5 20 32.1 (0.6) 13 99 p> 0.6

0.2 50 26.5 (1.8) 6 82 p< 0.00 1
0.1 100 14.3 (0.5) 8 44 p< 0.001
0.05 200 7.7 (0.8) 8 24 p < 0.001
a In each case, the amount of taurine in the sample was 0.01 pmol.
b The mealt for each volume was compared to the mean for the 10-iL samples

ous quantitative errors in taurine assay if large sample col- if platelet contamination were contributing to taurine values
umns are used (9, 10). We undertook to determine what of those specimens centrifuged at 1000 and 10 000 X g. Counts
sample volume can be accurately analyzed, and whether the revealed a mean of 4.9 (SD 2.6) X 10 platelets per milliliter
observed limitation of the TSM system was a function of remaining in the supernate of plasmas centrifuged at 1q00 X
sample volume or of sample concentration. g as compared with 1.4 (SD 0.4) X 10 platelets per milliliter
Different concentrations of aqueous taurine standards were in those centrifuged at 10 000 X g. From the difference in
absorbed to the sample cartridge in volumes such that the taurine content between aliquots centrifuged and filtered, one
same amount of taurine was applied to all cartridges. The can estimate the taurine content of platelets. Our mean of 0.43
results (Table 1) show that sample volumes exceeding 20 .iL nmol/106 platelets agrees with published values for platelets
showed a significant decrease in taurine peak height. The peak (18), indicating that the differences between filtered speci-
heights for the 10- and 20-liL samples did not differ signifi- mens and those centrifuged at 1000 and 10 000 )( g can indeed
cantly. Not surprisingly, we observed the same effect of ap- be accounted for by platelets remaining in the supernate.
plied volume on results for taurine in urine and plasma sam- In most studies reporting plasma taurine concentrations,
ples. Clearly, use of sample volumes of 50 .tL or more can lead the investigators have used either EDTA or heparin as anti-
to significant taurine loss from the cartridge, and use of vol- coagulant. When samples are collected in EDTA, use of a
umes exceeding 20 tL will probably result in losses that are ninhydrin detection system can lead to erroneous values for
directly proportional to the volume applied to the cartridge taurine (7). Contaminants in EDTA emerge coincident with
in the TSM system. Presumably, some early-emerging acidic taurine in several cation-exchange chromatographic systems
and neutral amino acids are so loosely bound to the cartridge and, if taurine concentrations are low, this may contribute as
resin that they are eluted by water or the acidic species and much as 50% of the total detected peak (7). We find this to be
are lost when larger volumes are aspirated into the cartridge true for the TSM. On examining several commercially pre-
resin. This problem would not arise with instruments in which pared EDTA Vacutainer Tubes (Becton, Dickinson & Co.,
fluid to be assayed is placed directly onto the column. Orangeburg, NY 10932) and several lots of EDTA from dif-
The foregoing 20-zL restriction on sample volume renders ferent manufacturers, we invariably found this ninhydrin-
impossible the accurate quantitation of the low concentrations positive contaminant. The use of o-phthalaldehyde as a de-
of taurine in plasma with ninhydrin as the detection reagent. tection agent alleviates this problem because o-phthalal-
With use of o-phthalaldehyde conjugation and fluorometric dehyde does not react with the contaminant.
detection, sensitivity was increased almost 100-fold. Analysis Although values for serum taurine significantly exceed
of 20 samples of the same taurine standard demonstrated those for plasma (15), no one has compared the common
acceptable reproducibility (CV = 4.5%). Duplicate analysis anticoagulants for their potential effects on results of plasma
of 20 plasma samples yielded a mean difference of 4.6% (SEM taurine assays. Using aliquots of blood from five subjects, we
0.7%). As little as 100 pmol of taurine can be accurately treated specimens with EDTA, sodium citrate, heparin, or no
quantified. anticoagulant (serum). Platelets were removed by filtration,
The largest potential source of error in the quantitation of as verified by microscopic inspection. The respective mean
plasma taurine is thought to be inadequate removal of plate- taurine values for the various treatments were 39.5 (SD 5.4),
lets (which are rich in taurine) from samples (12). The method 37.5 (SD 5.9), 39.9 (SD 5.3), and 85.6 (SD 18.1) mol/L for
used by most investigators to remove platelets is high-speed those anticoagulated with EDTA, sodium citrate, and heparin,
centrifugation. We obtained the same result by using a Millex and for serum. Note that the variation in sera was dispro-
filter (O.45-lim pore size). Plasma samples thus filtered, ex- portionately and significantly higher than that in any plasma
amined microscopically, were found to be free of platelets. aliquot. Two-way analysis of variance revealed a significant
It is, of course, possible that filtration might have ruptured difference (F = 18.81; 3, 12 d/f; p <0.001). However, if the
platelets or induced taurine release through (e.g.) surface ef- obviously higher serum values are excluded, the differences
fects. To test for these possibilities, we first centrifuged plasma between pairs of anticoagulants are not significant (F = 2.9;
samples from 10 individuals at 650 X g; then aliquots of the 2, 8 d/f;p > 0.05).
supernates were either filtered or centrifuged at 1000 or at Heparin causes platelets to aggregate, and this aggregation
10 000 x g. Taurine analyses of these plasma aliquots bed to results in some release of intracellular constituents of platelets
means of 42.9 (SEM 2.1), 41.1 (SEM 1.9), and 39.7 (SEM 2.0) (13). In earlier studies in which heparin was used and the
mol/L for those centrifuged at 1000 X g, 10 000 X g, and fil- samples were centrifuged at speeds sufficient to remove most
tered, respectively. Two-way analysis of variance revealed platelets, the reported values were high by comparison with
significant differences between treatments (F = 5.83; 2, 8 d/f; our data. Microscopically, platelet aggregation could be seen
p <0.05). Applying paired t-tests, we found that the filtered in all heparinized aliquots before filtration, but not in those
aliquots had a significantly lower mean than those centrifuged treated with sodium citrate or EDTA. However, the amount
at 1000 X g (t = 4.40, p <0.01), but no other comparisons of taurine released by aggregation must be small, as evidenced
reached conventional significance levels. by our data above. From these date we conclude that taurine
However, the differences were in the direction anticipated release by platelet aggregation is small relative to intra-indi-

CLINICALCHEMISTRY,Vol.26,No. 3, 1980 509

vidual variation and experimental error. of taurine and taurine derivatives. Physiol. Rev. 48, 424-511
Using EDTA or sodium citrate as anticoagulant, we found (1968).
2. Huxtable, R. J., and Barbeau, A., Section introduction: Neuro-
the mean for plasma taurine to be 45.0 (SEM 2.0) zmol/L
pharmacology and electrophysiology of taurine. In Taurine and
among our 44 control subjects. The range was 25-93 imol/L. Neurological Disorders, A. Barbeau and R. J. Huxtable, Eds., Raven
A comparison of mean plasma taurine concentrations in the Press, New York, NY, 1978, pp 179-180.
stone-disease patients with that of normal subjects revealed 3. Oja, S. S., and Kontro, P. Neurotransmitter actions of taurine in
no significant difference (t = 1.03, 42 d/f, p > 0.3). the central nervous system. In Ibid., pp 181-200.
4. Phillis, J. W., Overview of neurochemical and neurophysiological
Discussion actions of taurine. In Ibid., pp 289-303.
Our adaptation of the TSM amino acid analyzer to fluoro- 5. Huxtable, R., and Barbeau, A. Taurine. Raven Press, New York,
metric analysis was prompted by a previous report in which NY, 1976.
the Beckman amino acid analyzer was so used (16). f-Amino 6. Taurine and Neurological Disorders. A. Barbeau and R. J. Hux-
acids react well with o-phthalaldehyde to give a good fluoro- table, Raven Press, New York, NY, 1978.
metric yield (17). The increased sensitivity obtained with this 7. Perry, T. L., and Hansen, S., Technical pitfalls in the quantitation
system allows analysis in biological fluids and tissues for of plasma amino acids. Clin. Chim. Acta 25, 53-58(1969).
several amino acids that cannot be quantified with the con- 8. Scriver, C. R., Clow, C. L., and Lamm, P., Plasma amino acids:
ventional TSM system. With further electrical expansion, we Screening, quantitation and interpretation. Am. J. Clin. Nutr. 24,
876-890 (1971).
estimate that as little as 10 pmol of amino acids could be
9. Ersser, R. S. A fast automated cation-exchange chromatographic
method for the analysis of plasma amino acids. Med. Lab. Sci. 33,
The self-elution of taurine from cartridges by volumes much 257-263 (1976).
exceeding 20 tL, as demonstrated above, agrees with Ersser’s 10. Lemieux, B., Giguere, R., Barbeau, A., et al., Taurine in cere-
(9) estimate of maximal permissible volume. Lemieux et al. brospinal fluid in Friedreich’s ataxia. Can. J. Neurol. Sci. 5, 125-129
(10) reported that 25 tL could be used without error but they (1978).
presented no data on volumes less than 25 liL. This source of 11. Benson, J. R., and Hare, P. E. o-Phthalaldehyde fluorogenic de-
error is, of course, common to several neutral or strongly acidic tection of primary amines in the picomole range. Comparison with
amino acids when the TSM system is used. fluorescamine and ninhydrin. Proc. Natl. Acad. Sci. USA 72,619-622
Our calculations of the taurine content of platelets agree (1975).
well with the values reported by Perry (18), who used a slightly 12. Perry, T. L., Taurine in dominantly inherited cerebellar atrophies
and other human neurological disorders. In ref. 6., pp 441-451.
different method. Using a method quite similar to ours, Ta-
13. Eika, C., On the mechanism of platelet aggregation induced by
chiki et al. (19) reported a value of 294 (SD 34) nmol of taurine
heparin, protamine and polybrene. Scand. J. Haematol. 9, 248-257
per 100 000 platelets. Recalculating data from their Table I,
we found that the reported value should have been 29.4 (SD 14. White, J. G., Ultrastructural physiology and cytochemistry of
3.4) nmol of taurine per 100 000 platelets, a value that still is blood platelets. In The Platelet. K. M. Brinkhous, R. W. Shermer,
exceedingly high and might reflect error from a variety of and F. K. Mostofi, Eds., Williams & Wilkins, Baltimore, MD, 1971,
sources, or subject differences. pp 83-115.
In papers reporting inaccurately high values for plasma 15. Armstrong, M. D., and Stave, U., A study of plasma free amino
taurine, the cause almost invariably appears to be contami- acid levels. I. Study of factors affecting validity of amino acid analyses.
nation with platelet taurine. To our knowledge, there are only Metabolism 22, 549-569 (1973).
four reports in the literature that do not differ significantly 16. Lund, E., Thomsen, J., and Brunfeldt, K., The use of o-phthal-
from our control values. Three of these papers were based on aldehyde for fluorescence detection in conventional amino acid ana-
lyzers. Sub-nanomole sensitivity in the analysis of phenylthiohy-
data from five or fewer individuals (1, 19, 21). Barbeau et al. dantoin-amino acids. J. Chromatogr. 130,51-54 (1977).
(22) reported plasma taurine values for 37 individuals used
17. Cronin, J. R., Pizzarello, S., and Gandy, W. E., Amino acid anal-
as controls in their study of amino acids in Friedreich’s ataxia. ysis with o-phthalaldehyde detection: Effects of reaction temperature
Their mean (44, SD 29 tmol/L) for control plasma taurine is and thiol on fluorescence yields. Anal. Biochem. 93, 174-179
essentially the same as ours (45, SD 13 lzmol/L) but their (1979).
variation around the mean significantly exceeded ours (F = 18. Perry, T. L., Hereditary mental depression with taurine defi-
4.55; 36, 43 d/f, p <0.01). We cannot explain this fact. ciency: Further studies, including a therapeutic trial of taurine ad-
ministration. In ref. 5, pp 365-374.
We conclude that plasma taurine is most conveniently and 19. Tachiki, K. H., Hendrie, H. C., Kellams, J., and Aprison, M. H.,
accurately measured by using EDTA as the anticoagulant, A rapid column chromatographic procedure for the routine mea-
surement of taurine in plasma of normals and depressed patients.
followed by low-speed centrifugation and micropore filtration
Clin. Chim. Acta 75, 455-465 (1977).
to yield platelet-free plasma. Use of o-phthalaldehyde for
20. Christensen, P. J., Data, J. W., Schonheyder, F., and Volqvartz,
detection is necessary for accurate quantitation of the low K., Amino acids in blood plasma and urine during pregnancy. Scand.
concentrations of several amino acids in plasma, including J. Clin. Lab. Invest. 9, 54-61 (1957).
taurine, and should also prove very useful in measuring tau- 21. Stein, W. H., and Moore, S., The free amino acids of human blood
rine in tissue. plasma. J. Biol. Chem. 211, 915-926 (1957).
22. Lemieux, B., Barbeau, A., Beroniade, V., et al., Amino acid me-
References tabolism in Friedreich’s ataxia. Can. J. Neurol. Sci. 3, 373-378
1. Jacobsen, J. G., and Smith, L. H., Biochemistry and physiology (1976).

510 CLINICAL CHEMISTRY, Vol. 26, No. 3, 1980