Journol oJNeurochemisby, 1975. Vol. 24, pp. 587-589. Pergamon Press. Printed in Great Britain
SHORT COMMUNICATION CSF amino acids and plasma-CSF amino acid ratios in adults' (Receiued 16 July 1974. Accepted 14 August 1974)
SEVERAL groups of investigators have reported mean concentrations of free amino acids in the CSF of adult subjects, using automated ion exchange chromatographic techniques (PERRY& JONES,1961; DICKINSON & HAMILTON,1966; PERRY et al., 1969; VAN SANDEet al., 1970; GJESSING et al., 1972). In all of these studies except one (GJESSINGet al., 1972), the number of subjects examined was either small, or values were not reported for some of the measurable amino compounds. We present here mean values for all 28 readily measurable amino compounds found in the CSF of 43 adult subjects. Data are also given for the mean concentrations of amino acids in fasting plasma from 77 healthy adults, and we call attention to some anomalies in the plasma-CSF ratios of certain amino compounds. MATERIALS AND METHODS Subjects
Venous blood was obtained after an overnight fast from 77 healthy persons (37 male and 40 female) aged 14-56y (mean 32 y). CSF was obtained by lumbar puncture from 43 persons (31 male and 12 female) aged 14-74 y (mean 46 y). Seven of these control subjects were neurologically normal individuals from whom CSF was obtained during induction of spinal anesthesia prior to elective surgery. The remaining 36 subjects suffered from a variety of neurological and psychiatric disorders, none ofwhich is known to involve abnormalities in the concentrations of any amino acids in either CSF or plasma. These patients underwent lumbar puncture as part of the diagnostic study of their illnesses. Specifically excluded from the pool of control subjects were patients with phenylketonuria, where phenylalanine is markedly elevated in CSF, and patients with Huntington's chorea, where levels of 7 amino acids are decreased in CSF (PERRYet a!., 1969; PERRY et al., 1973). No CSF specimens were used which contained blood or abnormal numbers of leucocytes. Although ideally it would have been preferable to have obtained data from CSF of healthy subjects only, ethical considerations make it virtually impossible to secure such data. METHODS Blood specimens were collected in heparinized tubes and immediately centrifuged at 21,000g for 10 min. The super-
' This work was supported by a grant from the Medical Research Council of Canada. 5x7
natant plasma was removed with a Pasteur pipette, taking care not to aspirate any of the buffy coat. Plasma was then deproteinized within 30 min of the venepuncture by adding 30 mg of solid sulphosalicylic acid per ml, shaking to mix, and centrifuging at 21,000 g for 10 min. Deproteinized plasma supernatants were stored at -70" until analysed. CSF specimens were collected in tubes containing dry sulphosalicylic acid (5 mg per ml CSF), and were centrifuged at 21,000 g for 10 min. The deproteinized CSF supernatants were stored a t -70" until analysed. Amino acid analyses were performed on a Technicon automatic amino acid analyser, using a single long column with a lithium citrate elution puffer system (PERRY et a!., 1968a). This technique gives an optimal resolution of the wide variety of amino compounds present in physiological fluids, and in particular makes possible accurate determinations of aspartic acid and glutamic acid, and their corresponding arnides. Analyses were performed on 0 5 m l of deproteinized plasma, or 2.0 ml of deproteinized CSF. The concentrations of amino acids recorded on chromatograms were calculated with a Technicon integrator-calculator. RESULTS AND DISCUSSION Table 1 gives the mean concentrations of free amino acids and related ninhydrin-positive compounds found in the CSF of 43 adult patients, and in the fasting plasma of 77 healthy adults. Plasma-CSF ratios are shown for the compounds which are present in both fluids in measurable amounts. Concentrations of most of the amino compounds in CSF in Table 1 agree well with values reported by VAN SANDE et al. (1970) and by GJESSINGet al. (1972). However, our mean values for glutamine and asparagine are considerably higher than reported by these investigators, and our mean values for aspartic and glutamic acids are much lower. The values we present for thew 4 amino acids are more likely to be accurate, since our CSF specimens were deproteinized immediately after collection by lumbar puncture, and since they were stored at -70°C. rather than at -20"C, prior to analysis. Enzymes present in CSF may slowly hydrolyse the 2 amides if deproteinization is not promptly carried out, and glutamine concentrations gradually decrease even after deproteinization when plasma or CSF is stored a t -20°C (PERRY & HANSEN, 1969).Table 1 also lists for the first time values for c-N-methyllysine and for y-aminobutyryl-lysine in CSF. The mean values shown for amino compounds in CSF of adults may not in all cases be the same as those which pre-
Short communication
588
TABLE1. FREEAMINO COMPOUNDS I N CSF AND FASTING PLASMA OF ADULTS Amino compound Taurine Phosphoethdnolamine Aspartic acid Threonine Serine Asparagine Glutamic acid Glutamine Proline Glycine Alanine Citrulline a-Amino-n-butyric acid Valine Cystine Methionine lsoleucine Leucine Tyrosine Phenylalanine Tryptophan Ethanolamine Ornithine Lysine 6-N-Methyl1y sine Histidine 1 -Methylhistidine 3-Me thylhistidine Homocarnosine y-Aminobutyr yl-1ysine Arginine
CSF' (43 subjects)
Fasting plasma' (77 subjects)
6 6 f 1.7 4.9 f 1.6 0 3 0.3 31.4 8.3 26.8 6.0 8.5 & 2.5 1.8 0.7 602.1 & 118.6 ND 6.4 1.7 30.0 & 6.5 2.1 & 0.8 3.1 2.0 18.1 4.7 0.1 & 0.2 3.0 0.9 4.8 & 1.5 12.9 3.5 8.0 2.1 8.7 f 3.5 1.0 f 0.7 15.4 f 4.8 5.5 f 1.9 27.5 5 5.1 1.5 f 2'0 12.7 & 2.7 ND ND 1.8 1.8 0.1 & 0.2 19.7 f 4.3
56 13 2*1 2+1 140+ 28 106 f 26 58+ 11 22* 11 614 f 94 116 f 59 219 f 54 354 i 84 33 8 22 f 7 229 & 42 46 8 22 f 4 59rf: 12 115 23 52 f 12 50 f 9 39 & 10 ND 5 4 + 15 188 30 7+6 9 0 + 13 4+8 3+1 ND ND 84 22
+ + + + +
+ + + + +
+
+
+ +
+
+
+
Plasma-CSF ratio 85 0.4 61 4.6 4.0 6.8 12.2 1 .o
34.2 11.8 15.7 5.9 12.7 7.4 12.3 8.9 6.5 5.8 39.0 9.8 6.8 4.7 7.I
43
* Results (mean and s.D.)are expressed in pmol/l. ND = not detectable.
vail in CSF of normal infants and young children. In a limited number of infants, we have found glutamine concentrations in CSF somewhat lower than in adults, whereas homocarnosine concentrations in CSF are much higher in infants and children than in adults (PERRYet al., 1968b). Small amounts of proline are found in the CSF of some infants, but proline is not detectable in the CSF of adults. Five amino compounds which are present in large amounts in living human brain (PERRYet al., 1971) are consistently absent from CSF. These are glycerophosphoethanolamine, GSH, GSSG, cystathionine and GABA. The mean amino acid concentrations in fasting plasma shown in Table 1 are in most instances similar to those recently reported for a large series of normal adults by ARMSTRONG& STAVE (1973). However, these investigators' values for taurine and glutamic acid may be artefactually elevated, since they used platelet-rich plasma for their analyses. Blood must be centrifuged at sufficiently high speed to avoid any contamination of plasma by platelets and leucocytes, if accurate values for plasma taurine, phosphoethanolamine, aspartic acid, or glutamic acid are to be obtained. These 4 compounds are present inside cells in enormous con-
centrations compared to those prevailing naturally in plasma (PERRY & HANSEN,1969). Table 1 also lists mean normal values for 5 amino compounds infrequently given in published reports of plasma amino acids: phosphoethanolamine, aspartic acid, c-N-methyllysine, and the 1- and 3methylhistidines. For most amino compounds present in both fluids, the concentration in plasma is 4-16 times greater than that in CSF. However, concentrations of glutamine are virtually identical in both plasma and CSF. The same holds true for urea (values not shown in Table 1). CSF levels of phosphoethanolamine are more than twice as high as those in fasting plasma. Ethanolamine, homocarnosine, and yaminobutyryl-lysine are present in CSF, but are undetectable in plasma, while cystine and proline are present in far higher concentrations in plasma than in CSF. Plasma concentrations of tryptophan are almost 40 times higher than those found in CSF. This may well be due to the much greater protein content of plasma, with tryptophan being largely bound to albumin, but recoverable as a free amino acid after denaturation of plasma proteins with sulphosalicylic acid.
Short communication The mean concentration of glycine is 34 times higher in fasting plasma than that found in CSF. The surprisingly low levels of glycine normally observed in CSF could be the result of the high affinity re-uptake mechanism for glycine in spinal neurons, where this amino acid may act as an inhibitory synaptic transmitter. The high plasma-CSF glycine ratio might also be explained by a relatively high rate of transport of glycine from CSF into plasma like that found in the rat by DUDZINSKI & CUTLER (1974).It is obvious that complex differences occur among various amino compounds in their transport between brain and CSF, between CSF and spinal cord, and between CSF and plasma. The values given in Table 1 for amino compounds in the CSF of adults with various neurological and psychiatric diseases presumably correspond reasonably with levels present in the CSF of healthy persons. The data may prove of value to investigators who have occasion to examine the CSF of patients with known disorders of amino acid metabolism, but who are unable to secure CSF from large numbers of control subjects. Individual amino acids have already been shown to be greatly increased in the CSF of patients with phenylketonuria, histidinemia, hyperprolinemia and argininemia (VANSANDE et al., 1970), in other urea cycle disorders (SKIH & EFRON,1972),and in hyperlysinemia (GHADIMI, 1972).We have found taurine markedly decreased in the CSF of a patient with an unusual hereditary neuropsychiatric disorder, whose brain content of taurine was subsequently shownatautopsy tobedeficient(PERRYe t d . , 1974).We have also found that glycine concentrations are enormously increased in the CSF of children with non-ketotic hyperglycinemia,but not in the CSF of children with other types of hyperglycinemia. In some of these metabolic disorders, discovery of a greater abnormality in the level of an amino acid in a patient’s CSF than in his plasma may indicate a grossly abnormal biochemical milieu in brain, which may in turn account for the patient’s neurological and mental symptoms. Department of Pharmacology, University of British Columbia, Vancouver, Canada V6T 1W 5
T. L. PERRY SHIRLEY HANSEN JANETKENNEDY
589
REFERENCES ARMSTRONGM. D. & STAVEU. (1973)Metabolism 22, 561569. DICKINSON J. c.& HAMILTQNP. B. (1966)J. Neurochem. 13, 1 1 79-1187. DUDZINSKI D.S. & CUTLERR. W. P. (1974)J. Neurochem. 22,355-361. GHADIMI H. (1972)in The Metabolic Basis oflnherited Disease (STANBURY J. B., WYNGMRDEN J. B. & FREDERICKSON D. S., eds.) 3rd ed., pp. 393-403. McGraw-Hill, New York. P. & SJAASTAD 0.(1972)J. NeurGJESINGL. R.,GJESDAHL ochem. 19,1807-1808. PERRY T. L.& JONESR. T. (1961)J . din. Invest. 40, 1363 1372. P E R ~T.YL.,STEDMAN D. & HANSENS.(1968a)J . Chromatogr. 38,460-466. PERRY T. L.,HANSENS., STEDMAN D. & LOVED. (1968b)J. Neurochem. 15, 1203-1206. PERRY T. L. & HANSENS.(1969)Cfin.Chim. Acta 25,5358. PERRY T. L.,HANSENS., DIAMOND S. & STEDMAN D. (1969) Lancet i, 806-808. PERRY T. L., HANSENS., BERRYK., MOK C. & LESKD. (1971)J. Neurochem. 18,521-528. PERRY T. L., HANSENS., LESKD. & KLOSTERM. (1973)in Advances in Neurology (BARWAUA., CHASET. N. & PAULSON G. W., eds.) Vol. 1, pp. 609-618. Raven Press New Y ork. PERRYT. L., BRAITYP. J. A., HANSENS., KENNEDY J., URQUHART N. & DOLMAN C. L. (1974)Archs. Neurol. in press. SHIHV. E. & EFRONM.L.(1972)in The Metabolic Basis of Inherited Disease (STANBURY J. B.,WYNGAARDEN J. B. & FREDERICKSON D. S.. eds.) 3rd ed., pp. 370-392.McGrawHill, New York. VANSANDEM., MARDENS Y.,ADRLAENSSENS K. & LOWENTHAL A. (1970)J. Neurochem. 17,125-135.
SHORT COMMUNICATION CSF amino acids and plasma-CSF amino acid ratios in adults' (Receiued 16 July 1974. Accepted 14 August 1974)
SEVERAL groups of investigators have reported mean concentrations of free amino acids in the CSF of adult subjects, using automated ion exchange chromatographic techniques (PERRY& JONES,1961; DICKINSON & HAMILTON,1966; PERRY et al., 1969; VAN SANDEet al., 1970; GJESSING et al., 1972). In all of these studies except one (GJESSINGet al., 1972), the number of subjects examined was either small, or values were not reported for some of the measurable amino compounds. We present here mean values for all 28 readily measurable amino compounds found in the CSF of 43 adult subjects. Data are also given for the mean concentrations of amino acids in fasting plasma from 77 healthy adults, and we call attention to some anomalies in the plasma-CSF ratios of certain amino compounds. MATERIALS AND METHODS Subjects
Venous blood was obtained after an overnight fast from 77 healthy persons (37 male and 40 female) aged 14-56y (mean 32 y). CSF was obtained by lumbar puncture from 43 persons (31 male and 12 female) aged 14-74 y (mean 46 y). Seven of these control subjects were neurologically normal individuals from whom CSF was obtained during induction of spinal anesthesia prior to elective surgery. The remaining 36 subjects suffered from a variety of neurological and psychiatric disorders, none ofwhich is known to involve abnormalities in the concentrations of any amino acids in either CSF or plasma. These patients underwent lumbar puncture as part of the diagnostic study of their illnesses. Specifically excluded from the pool of control subjects were patients with phenylketonuria, where phenylalanine is markedly elevated in CSF, and patients with Huntington's chorea, where levels of 7 amino acids are decreased in CSF (PERRYet a!., 1969; PERRY et al., 1973). No CSF specimens were used which contained blood or abnormal numbers of leucocytes. Although ideally it would have been preferable to have obtained data from CSF of healthy subjects only, ethical considerations make it virtually impossible to secure such data. METHODS Blood specimens were collected in heparinized tubes and immediately centrifuged at 21,000g for 10 min. The super-
' This work was supported by a grant from the Medical Research Council of Canada. 5x7
natant plasma was removed with a Pasteur pipette, taking care not to aspirate any of the buffy coat. Plasma was then deproteinized within 30 min of the venepuncture by adding 30 mg of solid sulphosalicylic acid per ml, shaking to mix, and centrifuging at 21,000 g for 10 min. Deproteinized plasma supernatants were stored at -70" until analysed. CSF specimens were collected in tubes containing dry sulphosalicylic acid (5 mg per ml CSF), and were centrifuged at 21,000 g for 10 min. The deproteinized CSF supernatants were stored a t -70" until analysed. Amino acid analyses were performed on a Technicon automatic amino acid analyser, using a single long column with a lithium citrate elution puffer system (PERRY et a!., 1968a). This technique gives an optimal resolution of the wide variety of amino compounds present in physiological fluids, and in particular makes possible accurate determinations of aspartic acid and glutamic acid, and their corresponding arnides. Analyses were performed on 0 5 m l of deproteinized plasma, or 2.0 ml of deproteinized CSF. The concentrations of amino acids recorded on chromatograms were calculated with a Technicon integrator-calculator. RESULTS AND DISCUSSION Table 1 gives the mean concentrations of free amino acids and related ninhydrin-positive compounds found in the CSF of 43 adult patients, and in the fasting plasma of 77 healthy adults. Plasma-CSF ratios are shown for the compounds which are present in both fluids in measurable amounts. Concentrations of most of the amino compounds in CSF in Table 1 agree well with values reported by VAN SANDE et al. (1970) and by GJESSINGet al. (1972). However, our mean values for glutamine and asparagine are considerably higher than reported by these investigators, and our mean values for aspartic and glutamic acids are much lower. The values we present for thew 4 amino acids are more likely to be accurate, since our CSF specimens were deproteinized immediately after collection by lumbar puncture, and since they were stored at -70°C. rather than at -20"C, prior to analysis. Enzymes present in CSF may slowly hydrolyse the 2 amides if deproteinization is not promptly carried out, and glutamine concentrations gradually decrease even after deproteinization when plasma or CSF is stored a t -20°C (PERRY & HANSEN, 1969).Table 1 also lists for the first time values for c-N-methyllysine and for y-aminobutyryl-lysine in CSF. The mean values shown for amino compounds in CSF of adults may not in all cases be the same as those which pre-
Short communication
588
TABLE1. FREEAMINO COMPOUNDS I N CSF AND FASTING PLASMA OF ADULTS Amino compound Taurine Phosphoethdnolamine Aspartic acid Threonine Serine Asparagine Glutamic acid Glutamine Proline Glycine Alanine Citrulline a-Amino-n-butyric acid Valine Cystine Methionine lsoleucine Leucine Tyrosine Phenylalanine Tryptophan Ethanolamine Ornithine Lysine 6-N-Methyl1y sine Histidine 1 -Methylhistidine 3-Me thylhistidine Homocarnosine y-Aminobutyr yl-1ysine Arginine
CSF' (43 subjects)
Fasting plasma' (77 subjects)
6 6 f 1.7 4.9 f 1.6 0 3 0.3 31.4 8.3 26.8 6.0 8.5 & 2.5 1.8 0.7 602.1 & 118.6 ND 6.4 1.7 30.0 & 6.5 2.1 & 0.8 3.1 2.0 18.1 4.7 0.1 & 0.2 3.0 0.9 4.8 & 1.5 12.9 3.5 8.0 2.1 8.7 f 3.5 1.0 f 0.7 15.4 f 4.8 5.5 f 1.9 27.5 5 5.1 1.5 f 2'0 12.7 & 2.7 ND ND 1.8 1.8 0.1 & 0.2 19.7 f 4.3
56 13 2*1 2+1 140+ 28 106 f 26 58+ 11 22* 11 614 f 94 116 f 59 219 f 54 354 i 84 33 8 22 f 7 229 & 42 46 8 22 f 4 59rf: 12 115 23 52 f 12 50 f 9 39 & 10 ND 5 4 + 15 188 30 7+6 9 0 + 13 4+8 3+1 ND ND 84 22
+ + + + +
+ + + + +
+
+
+ +
+
+
+
Plasma-CSF ratio 85 0.4 61 4.6 4.0 6.8 12.2 1 .o
34.2 11.8 15.7 5.9 12.7 7.4 12.3 8.9 6.5 5.8 39.0 9.8 6.8 4.7 7.I
43
* Results (mean and s.D.)are expressed in pmol/l. ND = not detectable.
vail in CSF of normal infants and young children. In a limited number of infants, we have found glutamine concentrations in CSF somewhat lower than in adults, whereas homocarnosine concentrations in CSF are much higher in infants and children than in adults (PERRYet al., 1968b). Small amounts of proline are found in the CSF of some infants, but proline is not detectable in the CSF of adults. Five amino compounds which are present in large amounts in living human brain (PERRYet al., 1971) are consistently absent from CSF. These are glycerophosphoethanolamine, GSH, GSSG, cystathionine and GABA. The mean amino acid concentrations in fasting plasma shown in Table 1 are in most instances similar to those recently reported for a large series of normal adults by ARMSTRONG& STAVE (1973). However, these investigators' values for taurine and glutamic acid may be artefactually elevated, since they used platelet-rich plasma for their analyses. Blood must be centrifuged at sufficiently high speed to avoid any contamination of plasma by platelets and leucocytes, if accurate values for plasma taurine, phosphoethanolamine, aspartic acid, or glutamic acid are to be obtained. These 4 compounds are present inside cells in enormous con-
centrations compared to those prevailing naturally in plasma (PERRY & HANSEN,1969). Table 1 also lists mean normal values for 5 amino compounds infrequently given in published reports of plasma amino acids: phosphoethanolamine, aspartic acid, c-N-methyllysine, and the 1- and 3methylhistidines. For most amino compounds present in both fluids, the concentration in plasma is 4-16 times greater than that in CSF. However, concentrations of glutamine are virtually identical in both plasma and CSF. The same holds true for urea (values not shown in Table 1). CSF levels of phosphoethanolamine are more than twice as high as those in fasting plasma. Ethanolamine, homocarnosine, and yaminobutyryl-lysine are present in CSF, but are undetectable in plasma, while cystine and proline are present in far higher concentrations in plasma than in CSF. Plasma concentrations of tryptophan are almost 40 times higher than those found in CSF. This may well be due to the much greater protein content of plasma, with tryptophan being largely bound to albumin, but recoverable as a free amino acid after denaturation of plasma proteins with sulphosalicylic acid.
Short communication The mean concentration of glycine is 34 times higher in fasting plasma than that found in CSF. The surprisingly low levels of glycine normally observed in CSF could be the result of the high affinity re-uptake mechanism for glycine in spinal neurons, where this amino acid may act as an inhibitory synaptic transmitter. The high plasma-CSF glycine ratio might also be explained by a relatively high rate of transport of glycine from CSF into plasma like that found in the rat by DUDZINSKI & CUTLER (1974).It is obvious that complex differences occur among various amino compounds in their transport between brain and CSF, between CSF and spinal cord, and between CSF and plasma. The values given in Table 1 for amino compounds in the CSF of adults with various neurological and psychiatric diseases presumably correspond reasonably with levels present in the CSF of healthy persons. The data may prove of value to investigators who have occasion to examine the CSF of patients with known disorders of amino acid metabolism, but who are unable to secure CSF from large numbers of control subjects. Individual amino acids have already been shown to be greatly increased in the CSF of patients with phenylketonuria, histidinemia, hyperprolinemia and argininemia (VANSANDE et al., 1970), in other urea cycle disorders (SKIH & EFRON,1972),and in hyperlysinemia (GHADIMI, 1972).We have found taurine markedly decreased in the CSF of a patient with an unusual hereditary neuropsychiatric disorder, whose brain content of taurine was subsequently shownatautopsy tobedeficient(PERRYe t d . , 1974).We have also found that glycine concentrations are enormously increased in the CSF of children with non-ketotic hyperglycinemia,but not in the CSF of children with other types of hyperglycinemia. In some of these metabolic disorders, discovery of a greater abnormality in the level of an amino acid in a patient’s CSF than in his plasma may indicate a grossly abnormal biochemical milieu in brain, which may in turn account for the patient’s neurological and mental symptoms. Department of Pharmacology, University of British Columbia, Vancouver, Canada V6T 1W 5
T. L. PERRY SHIRLEY HANSEN JANETKENNEDY
589
REFERENCES ARMSTRONGM. D. & STAVEU. (1973)Metabolism 22, 561569. DICKINSON J. c.& HAMILTQNP. B. (1966)J. Neurochem. 13, 1 1 79-1187. DUDZINSKI D.S. & CUTLERR. W. P. (1974)J. Neurochem. 22,355-361. GHADIMI H. (1972)in The Metabolic Basis oflnherited Disease (STANBURY J. B., WYNGMRDEN J. B. & FREDERICKSON D. S., eds.) 3rd ed., pp. 393-403. McGraw-Hill, New York. P. & SJAASTAD 0.(1972)J. NeurGJESINGL. R.,GJESDAHL ochem. 19,1807-1808. PERRY T. L.& JONESR. T. (1961)J . din. Invest. 40, 1363 1372. P E R ~T.YL.,STEDMAN D. & HANSENS.(1968a)J . Chromatogr. 38,460-466. PERRY T. L.,HANSENS., STEDMAN D. & LOVED. (1968b)J. Neurochem. 15, 1203-1206. PERRY T. L. & HANSENS.(1969)Cfin.Chim. Acta 25,5358. PERRY T. L.,HANSENS., DIAMOND S. & STEDMAN D. (1969) Lancet i, 806-808. PERRY T. L., HANSENS., BERRYK., MOK C. & LESKD. (1971)J. Neurochem. 18,521-528. PERRY T. L., HANSENS., LESKD. & KLOSTERM. (1973)in Advances in Neurology (BARWAUA., CHASET. N. & PAULSON G. W., eds.) Vol. 1, pp. 609-618. Raven Press New Y ork. PERRYT. L., BRAITYP. J. A., HANSENS., KENNEDY J., URQUHART N. & DOLMAN C. L. (1974)Archs. Neurol. in press. SHIHV. E. & EFRONM.L.(1972)in The Metabolic Basis of Inherited Disease (STANBURY J. B.,WYNGAARDEN J. B. & FREDERICKSON D. S.. eds.) 3rd ed., pp. 370-392.McGrawHill, New York. VANSANDEM., MARDENS Y.,ADRLAENSSENS K. & LOWENTHAL A. (1970)J. Neurochem. 17,125-135.
