- Email: [email protected]
Normal plasma free amino acid values in adults: The influence of some common physiological variables
Normal
Plasma
Free Amino Acid Values in Adults: The Influence Some Common Physiological Variables
Charles R. Striver,
Diana M. Gregory,
David Sovetts,
of
and Gerry Tissenbaum
We measured plasma free amino acids in ten healthy adults (five males, five females, ages 26 to 64 years), at four times of day, under two dietary protocols. The design allowed us to measure the effects of some common physiological variables on the metrical distributions (mean, SD, skewness, kurtosisj of 80 values fin most cases), for each of 20 amino acids. Analysis of variance indicates that individuals in a “public” population possess “private” phenotypes for 18 plasma amino acids under typical external experiences; the exceptions were aspartate and taurine. The collective infradiem variation rarely exceeded 60% of the nadir value for any amino acid. We conclude that genotype is an important determinant of plasma amino acid phenotype in normal persons, and that there is striking homeostasis of plasma amino acid values in human adults adapted to normal daily activity. m 1985 by Grune & Stratton, Inc.
C milieu interieur is the condition of a free and independent existence.“’ The metrical distribution of LAUDE BERNARD said, “The constancy of the
values for concentration (mass/vol) of amino acids in plasma observes a certain constancy (homeostasis) in humans that supports this statement. The finite distributions allow us to know “normal” values and to recognize “abnormal” values that may be signals of disease (dishomeostasis).’ It has been customary in clinical medicine to collect blood samples for amino acid analysis in the morning after an overnight fast. This is not necessarily the normal physiological circumstance for the person. The practice reflects an administrative convenience as much as a necessary condition for defining the significance of the value. In an earlier study we collected blood samples in late morning before the noon meal in order to classify adults for heterozygosity at the phenylalanine hydroxylase 10cus.~ This practice has now lead us to investigate the metrical distribution of plasma amino acid values in healthy adult persons under the usual conditions of daily life. In addition to describing the effects of sex, time of day, and dietary practice on individual free amino acids, we have made the observation that interindividual variation is greater than intraindividual variation for most amino acids. We did not study the bound amino acid fraction in plasma. From the Departments of Biology and Pediatrics, McGill University; the Medical Research Council Genetics Group, Center for Human Geneties; and the Quebec Network of Genetic Medicine. Address reprint requests to The deBelle Laboratoryfor Biochemical Genetics, McGill University-Montreal ChildrenS Hospital Research Institute, 2300 Tupper Street, Montreal, Quebec, Canada H3H IP3. Supported by the Medical Research Council and the Quebec Network of Genetic Medicine. D. Gregory and G. Tissenbaum were sponsored by the McGill University Department of Biology Independent Studies Program. 0 1985 by Grune & Stratton, Inc. 0026-0495/85/3409-0013$3.00/0 868
MATERIALS
AND
METHODS
Subjects Ten Caucasian adult subjects, all members of the laboratory staff, participated; there were five males and five females. Their age range was 26 to 54 years.
Blood Samples Blood was drawn from the antecubital vein into a heparinized syringe with minimal venous stasis. The plasma was separated immediately by centrifugation and withdrawn through a Pasteur pipette without perturbing the cell-plasma interface. Plasma was deproteinized immediately by the addition of 10 pL of 50% sulfosalycilic acid, to which was added 10 IL internal standard (norleucine) and 10 jtL 7% lithium hydroxide; protein was removed immediately by centrifugation. The supernatant was stored at -20°C until analysis. Storage times were mimimal to avoid reciprocal changes in glutamine and glutamate content of the sample.’
Protocols Four blood samples were obtained at 90-minute intervals beginning at 9 AM. This protocol allowed us to examine a portion of the normal diurnal variation in plasma amino acid concentrations. Two dietary protocols were used. In the fasting version, the 9 AM sample was obtained before or without the breakfast meal; it was taken after the subject’s normal breakfast in the nonfasting version. All subjects ate lunch between 12 noon and 1 PM; each person followed their own dietary preferences which were not greatly different one from another. These protocols allowed us to examine the influence of meal habits on plasma amino acids. All subjects were ambulatory and pursued their normal activities before and during the protocols.
Analytical Deproteinized supernatant (40 rL) was analyzed in a Durrum Amino Acid Analyzer (model DSOO). No more than five samples were analyzed sequentially in a batch. A standard amino acid mixture was run before and after each batch. The analytical coefficient of variation (norleucine standard) was 5.8% for the 80 analyses. The computerized integration of peak areas was unreliable for the following ten amino acids: alanine, arginine, histidine, isoleucine, leucine, lysine, methionine, ornithine, phenylalanine and tyrosine; the calculations were performed by hand for these amino acids. The fault is in the automated reading of baselines in the relevant regions of the elution chromatogram. Proline, hydroxyproline and citrulline values are not reported because of additional technical difficulties with their measurements; nor did we undertake the procedures in Metabolism,
Vol
34, No 9 (September),
1985
PLASMA FREE AMINO ACID VALUES IN ADULTS: VARIABLES
869
sample preparation necessary for accurate analysis of free tryptophan. The data set is incomplete for asparagine, cystine and glutamic acid because imperfect separation of their peaks prevented accurate measurements in some samples.
Statistical The principal moment statistics [mean, SD, kurtosis
(g&l
were
calculated
by standard
skewness
method?
(g,)
and
using
a Basic Statistics program and Hewlett Packard-HP 85 programable calculator. Amino acid values were analyzed by the two-tailed Student’s t test for effects of sex and diet; and by one-way ANOVA for interand intraindividual variation.
RESULTS Distribution of Plasma Amino Acid Values We obtained descriptors (mean, SD, skewness and kurtosis) for the distributions of 20 amino acids in plasma of the human adult (Table 1); several amino acids of interest (eg proline, hydroxyproline, citrulline and tryptophan) are not included for technical reasons. We show the rank order of values by magnitude in Fig 1 and the frequency distribution for each amino acid in Fig 2. Analysis of variance was performed on 17 amino acids for which we obtained the full set of 80 measurements; asparagine, cystine and glutamate were excluded because the sets of values were incomplete. The ANOVA (F9,r,,) indicates that interindividual variation is significantly greater (at P < 0.001) than intraindividual variation for the individual amino acids except aspartate (P = 0.087) and taurine (P = 0.229). These findings indicate that individuals within a “public” population possess a “private” plasma phenotype for most amino acids; the phenotype is expressed within the normal physiological experience which includes variance derived from sex of subjects, dietary practice, and time of day. To corroborate the evidence of chemical individuality, we examined the interindividual distribution of values for representative amino acids by calculating z scores. The z score is the difference in SD units between the mean of eight values for an individual and the mean of 80 values for the total sample; a plus score indicates that the individual’s values are distributed above the population mean; a negative score, below the mean; magnitude of the score indicates distance from the population mean. The z scores for lysine, which has a normal distribution in the total sample (n = 80, Table 2), were as follows: -1.1861, -0.9214, -0.6309, -0.5676, -0.0958, +0.1919, +0.4709, f0.6205, +0.9974, and + 1.1211. As expected, there were ten independent z scores, five positive and five negative, We found the same for threonine whose distribution of values is also normal but influenced by sex (see following text). Valine had a positively skewed
GLN ALA
0
100
I
I1
-
/-
VAL*
-
LYS
-
THR
-
-
TAU
-
ARG
-
HIS
-
ORN
-
ASN” CYS PHE GLU MET ASP’
_--
__/---
__---
17 IT -
LEUSER
TYR
300 400 500 600 700 800 900 1000 I I1 I I I I1
-
GLY
ILE
200
,+ / ‘w” ;w .w
_w - ‘$
_ d - ;+ - ,y ; &V
rl
I
1
1
0
100
200
’ 300
1 400
SO
I
I
I
I
500
600
700
800
I
900
4
1000
Plasma Free Amino Acid Concentrations @moles/l plasma) Fig 1. Plasma amino acid values [mean (canter of circle), f 1 SD solid bar (or diam. of circle): f 2 SD dotted boundaries] for adult subjects (n - 10) arranged in rank order by magnitude. The distributions encompass variation due to sax of individual (males 5, females 5). dietary protocol (n = 2). end times of day (n - 4) from 9 AMto 1:30 PM;total number measurements is 80 for all amino acids except esparagina and glutamate (n - 27). and oystina (n = 34). Asterisk indicates that distribution of values is not normal (g, or & statistic > 1.01.
distribution (Table 1) with seven negative and three positive z scores. Effect of Physiological Variables Sex of the individual did not influence the distribution of values for most amino acids; the exceptions were alanine, arginine, glutamine, leucine, threonine, and tyrosine (Table 2). We found no significant differences (at P = 0.05) attributable to sex for any amino acid in the 9 AM sample after an overnight fast; however, the measurements were few in this case (five females and five males v 40 females and 40 males for the total sample). In all previous studies of adult subjects, amino acids were measured in plasma obtained in the morning after an overnight fast. We compared the 9 AM fasting values (n = 10) with the corresponding nonfasting values (n = 10) in the same subjects and also with all other values (n = 70). The fasting values were significantly lower only for valine; and significantly higher
SCRIVER ET AL
870
Table 1. Plasma Amino Acid Values Encompassing Physidogical Variation in Adults pmol/L Aminoacid ” - 80
Meanf
SD
Skewness
360
69
0.053
Arginine
94
20
A&ragine+
62
19
Alanine
Aspartic acid Cystine*
-0.03
Kurtosis
1
1.211
7
4
61
13
-0.243 -0.25
P
P,?,S
- 0.800
4.9 1
0.0000
-0.598
4.23
0.0002
1.78
0.087 1
8.00
0.0000
5.53
o.oooo
5.91
o.oooo
0.988
1.175
1.205 0.296
34
13
Glutamine
656
146
0.025
Glycine
232
44
0.320
Histidine
94
14
0.325
lsoleucine
64
23
0.974
0.785
4.31
0.0002
Leucine
133
39
0.696
1.160
5.78
0.0000
Lysine
Glutamic acid*
1
-0.274 -0.057 0.276 -0.298
192
44
0.490
0.804
11.45
o.oooC
Methionine
24
8
0.887
0.847
7.33
0.0000
Ornithine
66
17
0.130
0.068
11.42
0.0000
Phenylalanine
58
14
0.234
9.45
0.0000
Serine
113
30
0.582
0.45 1
Taurine
94
48
1.345
1.428
145
39
64
19
0.713
264
79
1.333
2658
368
0.035
Threonine Tyrosine Valine Zaa§
-0.240
-0.172
10.82
o.oCoo
1.35
0.2291
-0.594
11.49
0.0000
-0.290
8.76
0.0000
2.468 -0.467
7.94
0.0000
6.99
0.0000
+Asn, n = 27; Glu, n = 27; Cys, n = 34. TStatistical moments of distribution (mean. SD, skewness, kurtosis) calculated by the BASIC STATISTICS program on an HP-85 programmable calculation (Hewlett Packard). $One way ANOVA for interindividual variation. §Sum of plasma amino acid values excludes Asn. Cys, and Glu.
0
0
27 55 82110 137
0
0
47 94 141188235
0
203406610813 1016
10 20 29 39 49
0
73 145 218 290 363
0
21 42 62 83 104
0
114 229 343 456 572
20 15 10 5 0
A. b n B.
9
2
6
0"
22
15
14
10
8 0
C. 5
24 46 71 95 119
0
4
8
12 16 20
0
51 102 152 203 254
(M icromoles per Lit re) Frequency distribution for plasma amino acid values in adults. Values (Amol/L) on abscissa are arranged in arbitrary intervals; Fig 2. superimposed curves are for normal distributions; deviation from a normal distribution is indicated by shape of histogram. Values for skewness (g, statistic) and kurtosis 1% statistic) are given, along with mean and SD, in Table 1. Group A has normal distribution; group 6 has skewness value > 1.O; group C has values for skewness and kurtosis >‘I .O.
PLASMA FREE AMINO ACID VALUES IN ADULTS: VARIABLES
Table 2. Effect of Physiological Variables on Plasma Amino Acid Values in Adults pmol/L Condition A
Condition B
Amino Acid
Mean
Mean
Sex of subjects
Alanine
378
male (Al Y
Arginine
101
18
female (8)*
Glutamine
726
121
Leucine
140
33
Threonine
135
Tyrosine
69
Valine Taurine
Condition
9 AM sample fasting (A) Y
SD
60
P
SD
342
73
0.0089
87
19
0.0006
586
135
0.0001
126
44
0.0433
38
154
39
0.0128
20
60
17
0.0166
223
48
270
82
0.0400
134
64
88
42
0.0017
nonfasting f8jT Dietary Protocols
Arginine
101
17
88
20
0.0019
fasting (A) v
Glutamine
710
125
602
146
0.0003
nonfasting fEi)$
Glycine
241
31
222
53
0.0274
l40 values in male group (condition A); 40 values in female group (Condition BI. TFasting samples (Condition A: n - 10). taken at 9
AM
after overnight
fast were compared with all other samples (Condition B; n - 70). $All samples in fasting protocol (Condition A: n - 40: see Materials and Methods) were compared with all samples in nonfasting protocol (Condition 8; n = 40).
only for taurine (Table 2). All values were similar in the fasting protocol (n = 40) when compared with the nonfasting protocol (n = 40) with the exception of arginine, glutamine, and glycine (lower in the nonfasting protocol; Table 2). The design of our study encompassed some of the diurnal variation in amino acid values. This wellknown physiological variable did not extinguish the chemical individuality of subjects in the population (see previous text). Infradiem variation in amino acid values has its own intrinsic interest. Accordingly, values at each time interval are shown (Fig 3) for the
Fig 3. lnfradiem variation in plasma amino acid values in adults. Scale on ordinate encompasses the lowest values end twice thst value, in the set of meesurements for each amino acid fn - 80 for all except: cys, 34: asn and glu, n - 27 each) in the fasting protocol (e-01, non-fasting protocol (O--O) and pooled values (A-4). Percent deviation from the nadir measure8 infradiem variation.
871
fasting protocol (n = lo), the nonfasting protocol (n = lo), and for the pooled samples (n = 20). Because there are particular circadian variations for individual amino acids3s4 we did not attempt to correlate the values of one amino acid with another; this can be done only at a specific time and would be valid only for that time. Infradiem oscillations above the nadir did not exceed 50% of the minimum concentration value with the exceptions of aspartate (+ 100%) glutamate (+ 63%) methionine (+ 58%), isoleucine (+ 75%) and taurine (+ 87%). This finding reveals, in yet another way, the striking homeostasis of plasma amino acid values in humans. DISCUSSION
We have determined plasma amino acid values in a small number of adults (n = 10) of both sexes under eight different physiological conditions representing typical experiences for healthy individuals. Accordingly, we obtained 80 measurements for most amino acids; technical artifacts during analysis prevented us from obtaining complete measurements for some amino acids. Since the measurements for any one amino acid were not all independent of each other, we could not estimate confidence limits of the metrical distributions. However, our primary interest was to obtain a statement about the physiological distribution of values and in this regard we succeeded. Our values for means and standard deviations of plasma amino acid distributions in adults (Table l), in general, correspond to those reported for 194 fasted adults by Armstrong and Stave.6 Ours would be more precisely representative with a larger sample of sub-
Time of Day
072
jects, but the present number of values is larger than reported in many previous studies; and unlike those in any other study known to us, they encompass common physiological variables (eg sex of subject, diet, and time of day). Values for skewness (g, statistic) with few exceptions (arginine, cystine, glutamate and threonine) were slightly positive; those for kurtosis (g, statistic) were both positive and negative. We did not attempt to determine significance of the g, and g, statistics. Of particular interest is our evidence that intraindividual variation is significantly less (at P < 0.001) than interindividual variation for all amino acids except aspartate and taurine. Since the environmental experience shared by the individuals was quite similar, the finding implies that genotype is an important determinant of the plasma amino acid phenotype. Many genes are involved in the homeostasis of each amino acid.4 Current theory ascribes interindividual heterogeneity to genetic polymorphism.’ Therefore, it is highly unlikely that any two individuals will have identical homeostatic parameters for plasma amino acids. There is individuality of phenotype, in the rat, an outbred species, even within a single component (net renal reabsorption) of the homeostatic network controlling plasma amino acid concentration*; so it is not surprising to find phenotypic individuality in a more outbred species such as human beings, in whom many more components of the network contribute to the phenotypic value. Others’ have identified intraindividual constancy in plasma amino acid levels measured on four occasions at yearly intervals under similar external conditions. These and our own findings simply outline further the theme of biochemical individuality advanced much earlier by Garrod” and Williams.” Our study of the effects of some physiological variables on plasma amino acid values was by no means exhaustive; others have studied them before. In one study,6 the sex of the subject influenced fasting values, male values were higher than female values, except for glycine, serine and threonine; and essential amino acids reflected the difference more than nonessential amino acids. When we found significant differences related to the sex of the subject (Table 2) male values were again higher than female values with the exception of threonine; our study, in contrast to the other,6 found nonessential amino acids more often affected. One would expect fasting values to be lower than nonfasting values. Tikanoja and Sirnell’* found smaller oscillations in plasma acids due to feeding in children and adult subject relative to infants implying that homeostatic mechanisms mature with age, if the finding in the infant is not simply a reflection of the larger protein intake relative to body mass. Our values for
SCRIVER ET AL
adults, which are higher than those reported for younger subjects, correspond to earlier reports4’13 We found only minor differences between fasting and nonfasting 9 AM plasma samples: valine was lower and taurine higher in the fasting sample (Table 2). On the other hand, values were lower in the nonfasting infradiem samples when there was a difference between fasting and nonfasting protocols (Table 2); and the lowest values were at 12 noon (Fig 3). This finding implies that the homeostatic mechanisms called into play vary at different times of the day. We found taurine and aspartate to be the most variable amino acids, an observation also made by Armstrong and Stave.13 This finding might be attributed to contamination of plasma by platelet or cellular elements’4*‘5 but we believe it is unlikely here. Others34,‘6 noted that freshly separated and deproteinized plasma and serum samples gave similar results; serum gave values different from plasma only when the coagulated blood sample had stood for several hours before the serum was drawn off the sample. Armstrong and Stave” studied correlations between plasma values of different amino acids; corresponding behavior in various combinations of amino acids could imply corresponding networks of homeostatic mechanisms. We did not attempt to study correlations because we found that even closely related amino acids (eg phenylalanine and tyrosine) had independent correlations in their plasma values at different times of the day.3 Since our protocol encompassed several hours of infradiem variation, it was inappropriate to pursue the analysis of ratios between mean plasma values for various amino acids. We calculated the g, and g, statistics for plasma amino acid values because our protocol was designed to amplify the effect of physiological variables on frequency distributions. Accordingly, skewness and kurtosis might reveal facts of homeostatic mechanisms, for example, extreme skewness might reveal interindividual genetic heterogeneity due to rare or polymorphic alleles, and extreme kurtosis might indicate “brittle” homeostasis.‘8 We did not find such evidence. This could be a reflection of the powerful homeostatic mechanisms that control plasma amino acid values in human beings; it is equally likely to be a reflection of the small number of individuals in our sample. A larger population might reveal the elements of brittleness and heterogeneity that surely characterize the chemical individuality of some human beings. ACKNOWLEDGEMENT We thank our colleagues who participated in this study; Professor Kurt Sittmann for valuable discussion and advice, and Dr Jack Neal for assistance with statistical analysis in the early stages. Huguette RizzZro and Lynne Prevost patiently prepared the typescript.
PLASMA
FREE AMINO
ACID VALUES IN ADULTS:
873
VARIABLES
REFERENCES 1. Bernard C: Les Phinomtnes de la Vie, vol 1. Paris, Libraire J-B Bailliere et Fils, 1878, p 879 2. Murphy EA: A scientific viewpoint on normalcy. Perspect Biol Med 9:333-348, 1966 3. Rosenblatt D and &river CR: Heterogeneity in genetic control of phenylalanine metabolism in man. Nature 218:677-678, 1968 4. Striver CR and Rosenberg LE: Distributions of amino acids in body fluids, in Amino Acid Metabolism and Its Disorders. Philadelphia, Saunders, 1973, pp 39-60 5. Sokal RR and Rohlf FJ: Biometry. San Francisco, Freeman, 1969 6. Armstrong MD and Stave U: A study of plasma free amino acid levels. II. Normal values for children and adults. Metabolism 22561-569.1973 7. Harris H: Molecular aspects of genetic heterogeneity in Ccckburn F and Gitzelmann R (eds): Inborn Errors of Metabolism in Humans. Lancaster, England, 1982, pp 3-l 1 8. McInnes RR and Striver CR: Net reabsorption of alphaaminoisobutyric acid by rat kidney in vivo. Am J Physiol 237:F274F284,1979 9. Armstrong MD and Stave U: A study of plasma free amino acid levels. IV. Characteristic individual levels of the amino acids. Metabolism 22:821-825, 1973 IO. Garrod AE: The incidence of alkaptonuria. A study in chemical individuality. Lancet 2: 1616-l 620, 1902
11. Williams RJ: Biochemical Individuality. The Basis for the Genetotrophic Concept. New York, John W&y and Sons, Science Editians, pp 97-105,1963 12. Tikanaja T and Simell 0: Plasma amino acids after a feed of human milk or formula at three months of age. J Pediatr Gastro Nutr 2:252-255, 1983 13. Armstrong MD and Stave U: A study of plasma free amino acid levels. III. Variations during growth and aging. Metabolism 22:571-578, 1973 14. Armstrong MD and Stave U: A study of factors affecting validity of amino acid analyses. Metabolism 22:549-560, 1973 15. Striver CR, Clow CL and Lamm P: On the screening, diagnosis and investigation of hereditary aminoacidopathies. Clin Biochem 6:142-188, 1973 16. Katz NR and Keck M: Comparison of amino acids in human serum and plasma. J Clin Chem Clin Biochem 15:89-91, 1977 17. Armstrong MD and Stave U: A study of plasma free amino acid levels. V. Correlations among the amino acids and between amino acids and some other blood constituents. Metabolism 22:827833,1973 18. Murphy EA: Only authorized persons admitted: The quantitative genetics of health and disease. The Johns Hopkins Med J 148:11&122,1981
Plasma
Free Amino Acid Values in Adults: The Influence Some Common Physiological Variables
Charles R. Striver,
Diana M. Gregory,
David Sovetts,
of
and Gerry Tissenbaum
We measured plasma free amino acids in ten healthy adults (five males, five females, ages 26 to 64 years), at four times of day, under two dietary protocols. The design allowed us to measure the effects of some common physiological variables on the metrical distributions (mean, SD, skewness, kurtosisj of 80 values fin most cases), for each of 20 amino acids. Analysis of variance indicates that individuals in a “public” population possess “private” phenotypes for 18 plasma amino acids under typical external experiences; the exceptions were aspartate and taurine. The collective infradiem variation rarely exceeded 60% of the nadir value for any amino acid. We conclude that genotype is an important determinant of plasma amino acid phenotype in normal persons, and that there is striking homeostasis of plasma amino acid values in human adults adapted to normal daily activity. m 1985 by Grune & Stratton, Inc.
C milieu interieur is the condition of a free and independent existence.“’ The metrical distribution of LAUDE BERNARD said, “The constancy of the
values for concentration (mass/vol) of amino acids in plasma observes a certain constancy (homeostasis) in humans that supports this statement. The finite distributions allow us to know “normal” values and to recognize “abnormal” values that may be signals of disease (dishomeostasis).’ It has been customary in clinical medicine to collect blood samples for amino acid analysis in the morning after an overnight fast. This is not necessarily the normal physiological circumstance for the person. The practice reflects an administrative convenience as much as a necessary condition for defining the significance of the value. In an earlier study we collected blood samples in late morning before the noon meal in order to classify adults for heterozygosity at the phenylalanine hydroxylase 10cus.~ This practice has now lead us to investigate the metrical distribution of plasma amino acid values in healthy adult persons under the usual conditions of daily life. In addition to describing the effects of sex, time of day, and dietary practice on individual free amino acids, we have made the observation that interindividual variation is greater than intraindividual variation for most amino acids. We did not study the bound amino acid fraction in plasma. From the Departments of Biology and Pediatrics, McGill University; the Medical Research Council Genetics Group, Center for Human Geneties; and the Quebec Network of Genetic Medicine. Address reprint requests to The deBelle Laboratoryfor Biochemical Genetics, McGill University-Montreal ChildrenS Hospital Research Institute, 2300 Tupper Street, Montreal, Quebec, Canada H3H IP3. Supported by the Medical Research Council and the Quebec Network of Genetic Medicine. D. Gregory and G. Tissenbaum were sponsored by the McGill University Department of Biology Independent Studies Program. 0 1985 by Grune & Stratton, Inc. 0026-0495/85/3409-0013$3.00/0 868
MATERIALS
AND
METHODS
Subjects Ten Caucasian adult subjects, all members of the laboratory staff, participated; there were five males and five females. Their age range was 26 to 54 years.
Blood Samples Blood was drawn from the antecubital vein into a heparinized syringe with minimal venous stasis. The plasma was separated immediately by centrifugation and withdrawn through a Pasteur pipette without perturbing the cell-plasma interface. Plasma was deproteinized immediately by the addition of 10 pL of 50% sulfosalycilic acid, to which was added 10 IL internal standard (norleucine) and 10 jtL 7% lithium hydroxide; protein was removed immediately by centrifugation. The supernatant was stored at -20°C until analysis. Storage times were mimimal to avoid reciprocal changes in glutamine and glutamate content of the sample.’
Protocols Four blood samples were obtained at 90-minute intervals beginning at 9 AM. This protocol allowed us to examine a portion of the normal diurnal variation in plasma amino acid concentrations. Two dietary protocols were used. In the fasting version, the 9 AM sample was obtained before or without the breakfast meal; it was taken after the subject’s normal breakfast in the nonfasting version. All subjects ate lunch between 12 noon and 1 PM; each person followed their own dietary preferences which were not greatly different one from another. These protocols allowed us to examine the influence of meal habits on plasma amino acids. All subjects were ambulatory and pursued their normal activities before and during the protocols.
Analytical Deproteinized supernatant (40 rL) was analyzed in a Durrum Amino Acid Analyzer (model DSOO). No more than five samples were analyzed sequentially in a batch. A standard amino acid mixture was run before and after each batch. The analytical coefficient of variation (norleucine standard) was 5.8% for the 80 analyses. The computerized integration of peak areas was unreliable for the following ten amino acids: alanine, arginine, histidine, isoleucine, leucine, lysine, methionine, ornithine, phenylalanine and tyrosine; the calculations were performed by hand for these amino acids. The fault is in the automated reading of baselines in the relevant regions of the elution chromatogram. Proline, hydroxyproline and citrulline values are not reported because of additional technical difficulties with their measurements; nor did we undertake the procedures in Metabolism,
Vol
34, No 9 (September),
1985
PLASMA FREE AMINO ACID VALUES IN ADULTS: VARIABLES
869
sample preparation necessary for accurate analysis of free tryptophan. The data set is incomplete for asparagine, cystine and glutamic acid because imperfect separation of their peaks prevented accurate measurements in some samples.
Statistical The principal moment statistics [mean, SD, kurtosis
(g&l
were
calculated
by standard
skewness
method?
(g,)
and
using
a Basic Statistics program and Hewlett Packard-HP 85 programable calculator. Amino acid values were analyzed by the two-tailed Student’s t test for effects of sex and diet; and by one-way ANOVA for interand intraindividual variation.
RESULTS Distribution of Plasma Amino Acid Values We obtained descriptors (mean, SD, skewness and kurtosis) for the distributions of 20 amino acids in plasma of the human adult (Table 1); several amino acids of interest (eg proline, hydroxyproline, citrulline and tryptophan) are not included for technical reasons. We show the rank order of values by magnitude in Fig 1 and the frequency distribution for each amino acid in Fig 2. Analysis of variance was performed on 17 amino acids for which we obtained the full set of 80 measurements; asparagine, cystine and glutamate were excluded because the sets of values were incomplete. The ANOVA (F9,r,,) indicates that interindividual variation is significantly greater (at P < 0.001) than intraindividual variation for the individual amino acids except aspartate (P = 0.087) and taurine (P = 0.229). These findings indicate that individuals within a “public” population possess a “private” plasma phenotype for most amino acids; the phenotype is expressed within the normal physiological experience which includes variance derived from sex of subjects, dietary practice, and time of day. To corroborate the evidence of chemical individuality, we examined the interindividual distribution of values for representative amino acids by calculating z scores. The z score is the difference in SD units between the mean of eight values for an individual and the mean of 80 values for the total sample; a plus score indicates that the individual’s values are distributed above the population mean; a negative score, below the mean; magnitude of the score indicates distance from the population mean. The z scores for lysine, which has a normal distribution in the total sample (n = 80, Table 2), were as follows: -1.1861, -0.9214, -0.6309, -0.5676, -0.0958, +0.1919, +0.4709, f0.6205, +0.9974, and + 1.1211. As expected, there were ten independent z scores, five positive and five negative, We found the same for threonine whose distribution of values is also normal but influenced by sex (see following text). Valine had a positively skewed
GLN ALA
0
100
I
I1
-
/-
VAL*
-
LYS
-
THR
-
-
TAU
-
ARG
-
HIS
-
ORN
-
ASN” CYS PHE GLU MET ASP’
_--
__/---
__---
17 IT -
LEUSER
TYR
300 400 500 600 700 800 900 1000 I I1 I I I I1
-
GLY
ILE
200
,+ / ‘w” ;w .w
_w - ‘$
_ d - ;+ - ,y ; &V
rl
I
1
1
0
100
200
’ 300
1 400
SO
I
I
I
I
500
600
700
800
I
900
4
1000
Plasma Free Amino Acid Concentrations @moles/l plasma) Fig 1. Plasma amino acid values [mean (canter of circle), f 1 SD solid bar (or diam. of circle): f 2 SD dotted boundaries] for adult subjects (n - 10) arranged in rank order by magnitude. The distributions encompass variation due to sax of individual (males 5, females 5). dietary protocol (n = 2). end times of day (n - 4) from 9 AMto 1:30 PM;total number measurements is 80 for all amino acids except esparagina and glutamate (n - 27). and oystina (n = 34). Asterisk indicates that distribution of values is not normal (g, or & statistic > 1.01.
distribution (Table 1) with seven negative and three positive z scores. Effect of Physiological Variables Sex of the individual did not influence the distribution of values for most amino acids; the exceptions were alanine, arginine, glutamine, leucine, threonine, and tyrosine (Table 2). We found no significant differences (at P = 0.05) attributable to sex for any amino acid in the 9 AM sample after an overnight fast; however, the measurements were few in this case (five females and five males v 40 females and 40 males for the total sample). In all previous studies of adult subjects, amino acids were measured in plasma obtained in the morning after an overnight fast. We compared the 9 AM fasting values (n = 10) with the corresponding nonfasting values (n = 10) in the same subjects and also with all other values (n = 70). The fasting values were significantly lower only for valine; and significantly higher
SCRIVER ET AL
870
Table 1. Plasma Amino Acid Values Encompassing Physidogical Variation in Adults pmol/L Aminoacid ” - 80
Meanf
SD
Skewness
360
69
0.053
Arginine
94
20
A&ragine+
62
19
Alanine
Aspartic acid Cystine*
-0.03
Kurtosis
1
1.211
7
4
61
13
-0.243 -0.25
P
P,?,S
- 0.800
4.9 1
0.0000
-0.598
4.23
0.0002
1.78
0.087 1
8.00
0.0000
5.53
o.oooo
5.91
o.oooo
0.988
1.175
1.205 0.296
34
13
Glutamine
656
146
0.025
Glycine
232
44
0.320
Histidine
94
14
0.325
lsoleucine
64
23
0.974
0.785
4.31
0.0002
Leucine
133
39
0.696
1.160
5.78
0.0000
Lysine
Glutamic acid*
1
-0.274 -0.057 0.276 -0.298
192
44
0.490
0.804
11.45
o.oooC
Methionine
24
8
0.887
0.847
7.33
0.0000
Ornithine
66
17
0.130
0.068
11.42
0.0000
Phenylalanine
58
14
0.234
9.45
0.0000
Serine
113
30
0.582
0.45 1
Taurine
94
48
1.345
1.428
145
39
64
19
0.713
264
79
1.333
2658
368
0.035
Threonine Tyrosine Valine Zaa§
-0.240
-0.172
10.82
o.oCoo
1.35
0.2291
-0.594
11.49
0.0000
-0.290
8.76
0.0000
2.468 -0.467
7.94
0.0000
6.99
0.0000
+Asn, n = 27; Glu, n = 27; Cys, n = 34. TStatistical moments of distribution (mean. SD, skewness, kurtosis) calculated by the BASIC STATISTICS program on an HP-85 programmable calculation (Hewlett Packard). $One way ANOVA for interindividual variation. §Sum of plasma amino acid values excludes Asn. Cys, and Glu.
0
0
27 55 82110 137
0
0
47 94 141188235
0
203406610813 1016
10 20 29 39 49
0
73 145 218 290 363
0
21 42 62 83 104
0
114 229 343 456 572
20 15 10 5 0
A. b n B.
9
2
6
0"
22
15
14
10
8 0
C. 5
24 46 71 95 119
0
4
8
12 16 20
0
51 102 152 203 254
(M icromoles per Lit re) Frequency distribution for plasma amino acid values in adults. Values (Amol/L) on abscissa are arranged in arbitrary intervals; Fig 2. superimposed curves are for normal distributions; deviation from a normal distribution is indicated by shape of histogram. Values for skewness (g, statistic) and kurtosis 1% statistic) are given, along with mean and SD, in Table 1. Group A has normal distribution; group 6 has skewness value > 1.O; group C has values for skewness and kurtosis >‘I .O.
PLASMA FREE AMINO ACID VALUES IN ADULTS: VARIABLES
Table 2. Effect of Physiological Variables on Plasma Amino Acid Values in Adults pmol/L Condition A
Condition B
Amino Acid
Mean
Mean
Sex of subjects
Alanine
378
male (Al Y
Arginine
101
18
female (8)*
Glutamine
726
121
Leucine
140
33
Threonine
135
Tyrosine
69
Valine Taurine
Condition
9 AM sample fasting (A) Y
SD
60
P
SD
342
73
0.0089
87
19
0.0006
586
135
0.0001
126
44
0.0433
38
154
39
0.0128
20
60
17
0.0166
223
48
270
82
0.0400
134
64
88
42
0.0017
nonfasting f8jT Dietary Protocols
Arginine
101
17
88
20
0.0019
fasting (A) v
Glutamine
710
125
602
146
0.0003
nonfasting fEi)$
Glycine
241
31
222
53
0.0274
l40 values in male group (condition A); 40 values in female group (Condition BI. TFasting samples (Condition A: n - 10). taken at 9
AM
after overnight
fast were compared with all other samples (Condition B; n - 70). $All samples in fasting protocol (Condition A: n - 40: see Materials and Methods) were compared with all samples in nonfasting protocol (Condition 8; n = 40).
only for taurine (Table 2). All values were similar in the fasting protocol (n = 40) when compared with the nonfasting protocol (n = 40) with the exception of arginine, glutamine, and glycine (lower in the nonfasting protocol; Table 2). The design of our study encompassed some of the diurnal variation in amino acid values. This wellknown physiological variable did not extinguish the chemical individuality of subjects in the population (see previous text). Infradiem variation in amino acid values has its own intrinsic interest. Accordingly, values at each time interval are shown (Fig 3) for the
Fig 3. lnfradiem variation in plasma amino acid values in adults. Scale on ordinate encompasses the lowest values end twice thst value, in the set of meesurements for each amino acid fn - 80 for all except: cys, 34: asn and glu, n - 27 each) in the fasting protocol (e-01, non-fasting protocol (O--O) and pooled values (A-4). Percent deviation from the nadir measure8 infradiem variation.
871
fasting protocol (n = lo), the nonfasting protocol (n = lo), and for the pooled samples (n = 20). Because there are particular circadian variations for individual amino acids3s4 we did not attempt to correlate the values of one amino acid with another; this can be done only at a specific time and would be valid only for that time. Infradiem oscillations above the nadir did not exceed 50% of the minimum concentration value with the exceptions of aspartate (+ 100%) glutamate (+ 63%) methionine (+ 58%), isoleucine (+ 75%) and taurine (+ 87%). This finding reveals, in yet another way, the striking homeostasis of plasma amino acid values in humans. DISCUSSION
We have determined plasma amino acid values in a small number of adults (n = 10) of both sexes under eight different physiological conditions representing typical experiences for healthy individuals. Accordingly, we obtained 80 measurements for most amino acids; technical artifacts during analysis prevented us from obtaining complete measurements for some amino acids. Since the measurements for any one amino acid were not all independent of each other, we could not estimate confidence limits of the metrical distributions. However, our primary interest was to obtain a statement about the physiological distribution of values and in this regard we succeeded. Our values for means and standard deviations of plasma amino acid distributions in adults (Table l), in general, correspond to those reported for 194 fasted adults by Armstrong and Stave.6 Ours would be more precisely representative with a larger sample of sub-
Time of Day
072
jects, but the present number of values is larger than reported in many previous studies; and unlike those in any other study known to us, they encompass common physiological variables (eg sex of subject, diet, and time of day). Values for skewness (g, statistic) with few exceptions (arginine, cystine, glutamate and threonine) were slightly positive; those for kurtosis (g, statistic) were both positive and negative. We did not attempt to determine significance of the g, and g, statistics. Of particular interest is our evidence that intraindividual variation is significantly less (at P < 0.001) than interindividual variation for all amino acids except aspartate and taurine. Since the environmental experience shared by the individuals was quite similar, the finding implies that genotype is an important determinant of the plasma amino acid phenotype. Many genes are involved in the homeostasis of each amino acid.4 Current theory ascribes interindividual heterogeneity to genetic polymorphism.’ Therefore, it is highly unlikely that any two individuals will have identical homeostatic parameters for plasma amino acids. There is individuality of phenotype, in the rat, an outbred species, even within a single component (net renal reabsorption) of the homeostatic network controlling plasma amino acid concentration*; so it is not surprising to find phenotypic individuality in a more outbred species such as human beings, in whom many more components of the network contribute to the phenotypic value. Others’ have identified intraindividual constancy in plasma amino acid levels measured on four occasions at yearly intervals under similar external conditions. These and our own findings simply outline further the theme of biochemical individuality advanced much earlier by Garrod” and Williams.” Our study of the effects of some physiological variables on plasma amino acid values was by no means exhaustive; others have studied them before. In one study,6 the sex of the subject influenced fasting values, male values were higher than female values, except for glycine, serine and threonine; and essential amino acids reflected the difference more than nonessential amino acids. When we found significant differences related to the sex of the subject (Table 2) male values were again higher than female values with the exception of threonine; our study, in contrast to the other,6 found nonessential amino acids more often affected. One would expect fasting values to be lower than nonfasting values. Tikanoja and Sirnell’* found smaller oscillations in plasma acids due to feeding in children and adult subject relative to infants implying that homeostatic mechanisms mature with age, if the finding in the infant is not simply a reflection of the larger protein intake relative to body mass. Our values for
SCRIVER ET AL
adults, which are higher than those reported for younger subjects, correspond to earlier reports4’13 We found only minor differences between fasting and nonfasting 9 AM plasma samples: valine was lower and taurine higher in the fasting sample (Table 2). On the other hand, values were lower in the nonfasting infradiem samples when there was a difference between fasting and nonfasting protocols (Table 2); and the lowest values were at 12 noon (Fig 3). This finding implies that the homeostatic mechanisms called into play vary at different times of the day. We found taurine and aspartate to be the most variable amino acids, an observation also made by Armstrong and Stave.13 This finding might be attributed to contamination of plasma by platelet or cellular elements’4*‘5 but we believe it is unlikely here. Others34,‘6 noted that freshly separated and deproteinized plasma and serum samples gave similar results; serum gave values different from plasma only when the coagulated blood sample had stood for several hours before the serum was drawn off the sample. Armstrong and Stave” studied correlations between plasma values of different amino acids; corresponding behavior in various combinations of amino acids could imply corresponding networks of homeostatic mechanisms. We did not attempt to study correlations because we found that even closely related amino acids (eg phenylalanine and tyrosine) had independent correlations in their plasma values at different times of the day.3 Since our protocol encompassed several hours of infradiem variation, it was inappropriate to pursue the analysis of ratios between mean plasma values for various amino acids. We calculated the g, and g, statistics for plasma amino acid values because our protocol was designed to amplify the effect of physiological variables on frequency distributions. Accordingly, skewness and kurtosis might reveal facts of homeostatic mechanisms, for example, extreme skewness might reveal interindividual genetic heterogeneity due to rare or polymorphic alleles, and extreme kurtosis might indicate “brittle” homeostasis.‘8 We did not find such evidence. This could be a reflection of the powerful homeostatic mechanisms that control plasma amino acid values in human beings; it is equally likely to be a reflection of the small number of individuals in our sample. A larger population might reveal the elements of brittleness and heterogeneity that surely characterize the chemical individuality of some human beings. ACKNOWLEDGEMENT We thank our colleagues who participated in this study; Professor Kurt Sittmann for valuable discussion and advice, and Dr Jack Neal for assistance with statistical analysis in the early stages. Huguette RizzZro and Lynne Prevost patiently prepared the typescript.
PLASMA
FREE AMINO
ACID VALUES IN ADULTS:
873
VARIABLES
REFERENCES 1. Bernard C: Les Phinomtnes de la Vie, vol 1. Paris, Libraire J-B Bailliere et Fils, 1878, p 879 2. Murphy EA: A scientific viewpoint on normalcy. Perspect Biol Med 9:333-348, 1966 3. Rosenblatt D and &river CR: Heterogeneity in genetic control of phenylalanine metabolism in man. Nature 218:677-678, 1968 4. Striver CR and Rosenberg LE: Distributions of amino acids in body fluids, in Amino Acid Metabolism and Its Disorders. Philadelphia, Saunders, 1973, pp 39-60 5. Sokal RR and Rohlf FJ: Biometry. San Francisco, Freeman, 1969 6. Armstrong MD and Stave U: A study of plasma free amino acid levels. II. Normal values for children and adults. Metabolism 22561-569.1973 7. Harris H: Molecular aspects of genetic heterogeneity in Ccckburn F and Gitzelmann R (eds): Inborn Errors of Metabolism in Humans. Lancaster, England, 1982, pp 3-l 1 8. McInnes RR and Striver CR: Net reabsorption of alphaaminoisobutyric acid by rat kidney in vivo. Am J Physiol 237:F274F284,1979 9. Armstrong MD and Stave U: A study of plasma free amino acid levels. IV. Characteristic individual levels of the amino acids. Metabolism 22:821-825, 1973 IO. Garrod AE: The incidence of alkaptonuria. A study in chemical individuality. Lancet 2: 1616-l 620, 1902
11. Williams RJ: Biochemical Individuality. The Basis for the Genetotrophic Concept. New York, John W&y and Sons, Science Editians, pp 97-105,1963 12. Tikanaja T and Simell 0: Plasma amino acids after a feed of human milk or formula at three months of age. J Pediatr Gastro Nutr 2:252-255, 1983 13. Armstrong MD and Stave U: A study of plasma free amino acid levels. III. Variations during growth and aging. Metabolism 22:571-578, 1973 14. Armstrong MD and Stave U: A study of factors affecting validity of amino acid analyses. Metabolism 22:549-560, 1973 15. Striver CR, Clow CL and Lamm P: On the screening, diagnosis and investigation of hereditary aminoacidopathies. Clin Biochem 6:142-188, 1973 16. Katz NR and Keck M: Comparison of amino acids in human serum and plasma. J Clin Chem Clin Biochem 15:89-91, 1977 17. Armstrong MD and Stave U: A study of plasma free amino acid levels. V. Correlations among the amino acids and between amino acids and some other blood constituents. Metabolism 22:827833,1973 18. Murphy EA: Only authorized persons admitted: The quantitative genetics of health and disease. The Johns Hopkins Med J 148:11&122,1981