tournol ot Newt,( hernistrY Lippincott-Raven Publishers, Philadelphia n 1995 International Society for Neurochemistry

y-Glutamylglutamine and Taurine Concentrations Are Decreased in the Cerebrospinal Fluid of Drug-Naive Patients with Schizophrenic Disorders

K . Q. Do, *C. J. Lauer, *W. Schreiber, M . Zollinger, U. Gutteck-Amsler, M. Cuénod, and *F. Holsboer

Brain Research Institute, University ofZürich, Zürich, Switzerland; and *Max Planck Institute of Psychiatry, Clinical Institute, Munich, Germany

Abstract: HPLC and gas chromatography-mass spec- monoaminergic and, in particular, the dopaminergic trometry analyses of 18 amino acids, N-acetylaspartate, systems; whereas the other claims a role for the excit- N-acetylaspartylglutamate, and 5-hydroxyindoleacetic atory amino acids, especially glutamate. acid, derived from serotonin, and homovanillic acid, de- Based on the psychotomimetic properties of amphet- rived from dopamine, were performed in CSF collected amine, which is dopamine agonist, and observations a with either a from group of patients schizophrenia who that almost all antipsychotic medications counteract had been drug free for at least 1 year (n = 5) or were drug naive for psychotropic drugs (n = 21) and in 15 dopamine activity in the brain, Matthysse ( 1973) pro- control subjects . Significant differences were found only posed that schizophrenic illness is a manifestation of for taurine (15% lower in the patients) and isoleucine a hyperdopaminergic state. But this initial view was (7% higher) . A number of unidentified substances were not unequivocally supported by subsequent studies, be- detected, one of which proved to be markedly reduced cause conflicting observations were reported regarding (16%) among the schizophrenic patients. Liquid chroma- homovanillic acid (HVA) concentrations in CSF, tography-mass spectrometry with continuous flow-fast plasma, and postmortem brain tissue of individuals atom bombardment interface allowed us to identify this with schizophrenia (for an overview, see Davis et al ., substance as y-glutamylglutamine. The decreased level 1991) . During the past few years, however, evidence of y-glutamylglutamine may reflect a deficiency in the y- glutamyltransferase system, a system probably involved has been accumulated that schizophrenic disorders in glutamate uptake, or a deficiency in glutamine, an im- may be characterized by both hypodopaminergia in the portant precursor of releasable glutamate. Although glu- mesocortical dopamine neurons and hyperdopaminer- tamate was nonsignificantly reduced in the patients, it gia in the mesolimbic dopamine neurons (Davis et al., was one of the five substances (including -/-glutamylglu- 1991) . D2-like receptors were reported to be elevated tamine) that were necessary for the best discrimination in postmortem brain tissue of schizophrenic patients between the schizophrenic patients and the controls . (Seeman et al., 1990), which supports the hypothesis These findings support the notion that the glutamatergic that dopaminergic neurotransmission in schizophrenia system is affected in schizophrenic disorders . In addition, they underscore the need to apply rigid bioanalytical techniques and use drug-naive patients to gain in-depth Received April 7, 1995 ; final revised manuscript received July 3, information on the pathophysiology of brain disorders 1995 ; accepted July 3, 1995 . such as schizophrenia. Key Words: Amino acids-N- Address correspondence and reprint requests to Dr. K. Q. Do at Brain Research Institute, University of Zürich, August-Forel-Strasse Acetylaspartate-N-Acetylaspartylglutamate-5-Hy- 1, CH-8029 Zurich, Switzerland . droxyindoleacetic acid- Homovanillic acid-Cerebro- Abbreviations used: Abbreviations for amino acids are according spinal fluid-Schizophrenia . to the IUPAC-IUB Commission on Biochemical Nomenclature, ce- J. Neurochem. 65, 2652-2662 (1995) . AAA, a-arninoadipic acid ; ANCOVA, analysis of covariance; CF- FAB, continuous flow-fast atom bombardment ; DSM-111-R, Diag- nostic and Statistical Manual of Mental Disorders, third edition, revised ; FMOC, N-9-fluorenylmethyloxycarbonyl ; FMOC-CI, 9- fluorenylmethylchloroformate ; GC, gas chromatography ; y-Glu- Schizophrenia is a nosological concept comprising Gln, y-glutamylglutamine ; FICA, homocysteic acid ; HCSA, homo- a large variety of signs and symptoms that make a cysteinesulfinic acid ; 5-HIAA, 5-hydroxyindoleacetic acid ; HVA, unifactorial explanation of causality unlikely . In psy- homovanillic acid ; LC-MS, liquid chromatography-mass spectrom- etry; NAA, N-acetylaspartate; NAAG, N-acetylaspartylglutamate; chiatric research, two major theories on the pathophys- NMDA, N-methyl-D-aspartate ; OPA, o-phthalaldehyde; OPA-1`15 .5, iology of this disorder have been developed during OPA derivative of 1`15 .5; PCP, ; SCID, Structured the past decades . One assumes a dysfunction of the Clinical Interview for DSM-111-R .

2652 REDUCED y-Glu-Gln AND TAURINE IN SCHIZOPHRENIA 2653 is enhanced through overactive response by D2 recep- MATERIALS AND METHODS tors. This concept has been criticized because clozap- Subjects ine, which has only low affinity for D2 receptors, is Twenty-six inpatients ( 15 women, l l men ; age range, clinically a highly effective neuroleptic. Recently, 21-53 years) with a diagnosis of a schizophreniform or however, the group led by Seeman demonstrated the a schizophrenic disorder according to the Diagnostic and existence of a polymorphic D4 receptor to which clo- Statistical Manual (#'Mental Disorders, third edition, revised zapine binds with an affinity 10 times higher than to (DSM-III-R) (American Psychiatric Association, 1987; any other yet-characterized dopamine receptors (Van schizophreniform disorder, n = 9 ; schizophrenic disorder, Tol et al., 1992) . It is interesting that the same group paranoid subtype, n = 13 ; schizophrenic disorder, disorga- found D4 receptors to be elevated in the brain of pa- nized subtype, n = 4), participated in the study . The diagno- tients with schizophrenia (Seeman et al ., 1993) . ses were verified by the Structured Clinical Interview for DSM-III-R (SCID; German version : Wittchen et al., 1990) . The second major hypothesis, as originally formu- None of the participants fulfilled DSM-111-R diagnostic crite- lated by Kim and colleagues (1980), claimed a role ria for any other psychiatric disorders. According to the crite- for excitatory amino acids, i .e., glutamate, in the patho- ria established by the SCID, 15 patients were rated to be physiology of schizophrenia. On the basis of decreased moderately ill and the remaining 11 patients severely ill . levels of glutamate in the CSF of individuals with Four patients had been free of psychoactive medication for schizophrenia, these authors postulated that either a at least I year and one patient for 8 years. The remaining hyperactivity of dopaminergic neurons leads to an en- 21 patients were drug naive in this regard . Furthermore, none of the patients had received any medication during the hanced inhibition of glutamate release or a hypofunc- 2 months before lumbar puncture . tion of the glutamate receptor causes the decreased Fifteen age- and sex-matched subjects (nine women, six glutamate release. Indeed, in rodents, the release of men; age range, 19-58 years) who had been admitted to glutamate from the striatal terminals is inhibited by the neurological ward for diagnostic clarification of health dopamine (see Carlsson and Carlsson, 1990) . The re- problems [headache, hypertension, PRIND (prolonged re- formulation of the glutamatergic hypothesis is based versible ischemic neurologic deficit) occasional spasms, low on the observation that phencyclidine (PCP) elicits back pain] served as the control group. In these subjects, both productive and deficit symptoms of schizophre- internal, neurological, and psychiatric disorders were ex- nia. Although PCP exerts its effects in part via central cluded after a thorough examination. None of these subjects had received any medication during the 2 months before dopamine systems, it is a potent and noncompetitive lumbar puncture . antagonist at the N-methyl-D-aspartate (NMDA) re- The experimental protocol was approved by the Ethics ceptor that has distinct recognition sites for glutamate, Committee for Human Experiments of the Max Planck Insti- aspartate, and . When PCP binds to its receptor tute of Psychiatry . sites within the NMDA ion channels, these voltage- dependent channels are blocked. As a result, the flow CSF sampling procedure of ions into the channel is blocked and the effects After 1 night of bed rest and fasting, subjects were placed in a sitting position at 0800 h. After administration of an of' NMDA agonists such as glutamate or glycine are intradermal xylocaine anesthesia, an 18-G 32, or a 19-G 3'-, antagonized (see Heresco-Levy et al., 1993) . Thus, Yale spinal needle was inserted into the lumbar subarachnoid the current hypoglutamatergic theory of schizophrenia space through the L3-L4 or L4-L5 interspace. After collec- proposes diminished glutamatergic neurotransmission tion of CSF, 5 ml of the fluid was transferred into an ice- at the level of the NMDA-receptor site (Deutsch et cooled test tube. The CSF was vortex mixed and aliquoted al ., 1989 ; Squires and Saederup, 1991) . into five 1-ml vials. These vials were then immediately fro- Although both the dopaminergic and glutamatergic zen and stored at -80°C until analysis . hypotheses of schizophrenia have stimulated intense Materials work in this field, there has been neither unequivocal Reference amino acid solutions were prepared with crys- support for nor clear rejection of either theory, mainly talline amino acids dissolved in water at a standard concen- because in vivo studies in humans are limited to brain tration of 0.1 pM and frozen in aliquots at -20°C. AGI- imaging studies and bioanalysis of CSF content. As- X8 anion exchanger (200-400 mesh, chloride form) was suming that changes in CSF composition are a direct obtained from Bio-Rad (Richmond, CA, U.S.A.) and con- verted to the formate form . HCl (0 M) or indirect reflection of changes in neurotransmitter- .5 in methanol was prepared by bubbling HCl gas through methanol and was receptor action in brain areas involved in the pathology stored in ampules at -30°C. N-Acetyl-t-glutamyl-t-gluta- of schizophrenia, we applied a high-powered bioana- mate was synthesized from i_-glutamyl-t-glutamate lytical method [HPLC, gas chromatography (GC)], (Bachem; Bubendorf, Switzerland) by treatment with acetic which enables determination of amino acids, mono- anhydride and was purified by ion-exchange chromatogra- amine metabolites, N-acetylaspartate (NAA), and N- phy. All other solvents and chemicals used were ofcommer- acetylaspartylglutamate (NAAG) levels and identifi- cial analytical grade from Merck (Darmstadt, Germany) or cation of heretofore unknown compounds. As a study Fluka (Buchs, Switzerland) . sample, we used a group ofpatients with schizophrenia CSF deproteinization in whom long-term changes secondary to previous an- The samples were thawed slowly on ice. Internal standard tipsychotic treatment could be excluded. (norvaline ; final concentration, 2 pM) was then added and

J. Neurochen. Vn/. 65. No . 6, 1995 2654 K. Q. DO ET AL. the samples were deproteinated by ultrafiltration at 18,300 graph from Varian (Walnut Creek, CA, U.S .A.) with a nitro- g (4°C) in Millipore Ultrafree-MC filter units (10,000 mo- gen/phosphorus-specific detector. The components were lecular weight cutoff) . This method allowed deproteiniza- separated on an SE-54 Permabond column (Macherey-Na- tion without using sulfosalicylic acid, which, at concentra- gel, Düren, Germany) with helium as carrier gas. For calibra- tions > 1 %, has been shown to increase CSF GABA concen- tion, artificial samples containing known amounts of NAA, trations and lead to falsely high levels of aspartic and NAAG, and the internal standard in water were treated and glutamic acids (Ferraro et al., 1983) . Care was also taken measured the same way as the CSF samples. For both NAA to maintain CSF at ice-bath temperature throughout depro- and NAAG, the calibrations were linear in the relevant range teinization and filtering procedures to avoid enzyme-medi- and relative standard deviations were estimated to be below ated breakdown of bound to free amino acids, as otherwise ± 15% for the largest values measured in the biological sam- occurs in the case of GABA (Grossman et al., 1980) . ples and below ±30% for the smallest ones. o-Phthalaldehyde (OPA)-precolumn Isolation and characterization of P15.5 derivatization and HPLC analysis of CSF In addition to aforementioned amino acids, several un- The HPLC system consisted of a Hewlett-Packard chro- identified substances were observed and quantified in CSF matograph (Model 1090), combined with a variable autoin- ofpatients and control subjects by HPLC after OPA derivati- jector, a cooled (4°C) autosampler, and a programmable zation. One of these compounds, named P15.5, showed a fluorescence detector (HP 1046A) . Analytical Hyperchrome significantly lower concentration in CSF of the patients than HPLC columns, 250 X 4.6 mm (Bischoff, Leonberg, Ger- in that of the controls (see Results) . Several different meth- many), packed with Hypersil ODS 5 pen (Shandon, Chesh- ods were used to identify P15 .5 chemically. .), in column . ire, U.K were used combination with a guard The elution time of the OPA derivative of P15.5 (OPA- The samples were analyzed by OPA precolumn derivatiza- P15.5) was compared with that of OPA derivatives of vari- tion, followed by reversed-phase HPLC (Lindroth and Mop- ous commercially available polar compounds [cysteic acid, per, 1979) . An automated on-line derivatization technique cysteine sulfinic acid, homocysteic acid (HCA), homocys- (Klancnik ., 1992) was used. A linear gradient elution et al teine sulfinic acid (HCSA), ,(3-aspartylglycine, 6-aspartyl- of 6-100% of mobile phase B [acetonitrile/0.I M sodium alanine, y-glutamylglutamate, y-glutamylglutamine (y-Glu- acetate (8:2)] in mobile phase A [0.03 M sodium acetate Gln), y-glutamylglycine, y-glutamylalanine, and y-gluta- (pH 7 .0)/0.5% tetrahydrofuran] was performed for 85 min myltaurine] . Usually, 10 pl of CSF was spiked with 2 pmol at a flow rate of 0.5 ml/min. Fluorescent compounds were of reference compound in 2 pl of water. Three different pH monitored using an excitation wavelength of 230 nm and an values (6.0, 7.0, and 8.0) of the mobile phase A and a less emission wavelength of 455 urn The detection limit was . steep gradient than in the standard analysis were used. - 10-50 fmol and the relative standard deviation for peak Among all the substances tested, only the OPA derivative areas based on 10 injections was 1-4%. The quantitation of the dipeptide y-Glu-Gln coeluted with OPA-P15 .5 under was based on area measurements and the internal standard all three conditions . method. As the OPA derivatives are not stable enough to be iso- lated for further characterization, CSF was treated with 9- Determination of monoamine metabolites In a subsample of 17 patients [nine women, eight men; fluorenylmethylchloroformate (FMOC-Cl) to produce stable mean age, 30.2 -} 8.1 (SD) years; age range, 21-53 years; and fluorescent N-9-fluorenylmethyloxycarbonyl (FMOC) schizophreniform disorder, n = 6 ; schizophrenic disorder, derivatives according to the procedure of Einarsson et al. paranoid subtype, n = 8 ; schizophrenic disorder, disorga- (1983) . A component ("FMOC-P15.5") eluting at the re- nized subtype, n = 3] and of nine control subjects (six tention time of the FMOC derivative of y-Glu-Gln was col- women, three men; mean age, 31 .4 ± 11 .2 years; age range, lected, purified, and investigated subsequently by liquid 19-47 years), HVA and 5-hydroxyindoleacetic acid (5- chromatography-mass spectrometry (LC-MS), as follows: HIAA) were determined. After centrifugation, 10 y1 of CSF Starting with a total of 2 ml CSF, three consecutive steps of was analyzed by reversed-phase HPLC (LiChrospher C18, purification by HPLC on different reversed-phase columns 4 X 120 mm) using isocratic elution with 0.1 M sodium and with different mobile phases were performed. The finally acetate, 0.1 M citric acid, 0.1 mM EDTA, 2.0 mM heptane- collected fraction contained -200 pmol FMOC-P15.5 (as- sulfonate, and 2% methanol in water at a flow rate of 0.8 suming the same response factor as FMOC-y-Glu-Gln) and ml/min. The compounds were monitored using a Gyncothec was subjected to micro-LC coupled to MS. (Germering, Germany) electrochemical detector operated at The LC-MS instrumentation consisted of the following 0.77 V . Calculations of concentrations were performed rela- components : a dual syringe gradient system Model 140A tive to external standards. (Applied Biosystems; Foster City, CA, U.S.A.), a precolumn splitting device, an injector, a 150 X 0.32-mm fused-silica Determination of NAA and NAAG column packed with 3 yen Hypersil RP-18 stationary phase To aliquots of 40 pl deproteinated CSF, 600 pmol N- (LC Packings ; Amsterdam, The Netherlands) connected in acetyl-L-glutamyl-L-glutamate was added as the internal series to a UV detector (Model 785A, Applied Biosystems) standard. In unspiked CSF, this compound was not detect- and then to a Finnigan MAT (Bremen, Germany) continuous able. NAA, NAAG, and the standard were purified by an flow-fast atom bombardment (CF-FAB) interface, coupled ion-exchange procedure and esterified with HCl in methanol, to an MAT 90 (Finnigan MAT) double-focusing mass spec- and the derivatives were then extracted and taken up in trometer . Compounds were eluted with a gradient of water /ethyl acetate (9 :1, vol/vol) as described by Zol- against acetonitrile/water (9 :1, vol/vol), both containing 2 linger et al . (1990) . These samples were analyzed by GC. vol-% glycerol and 0.15 M acetic acid. The flow rate through Injection was performed in the on-column mode. The GC the column and onto the FAB target was 3-5 p.1/min . The instrumentation consisted of a Model 3700 gas chromato- FAB gun was operated with xenon at -7 kV. The ion source

1 . Neurochem_ Vol. 65, No. 6, 1995 REDUCED y-Glu-Gln AND TAURINE IN SCHIZOPHRENIA 2655

temperature was 50°C . Positive and negative ion spectra were taken at a resolution of -900 . To show that the compounds termed FMOC-P15 .5 and OPA-P 15 .5 were indeed derivatives of the same substance, 40 pinol of purified FMOC-P15 was treated with 20 pl of triethylamine for I h at 50°C to cleave the FMOC group . The sample was evaporated to dryness, redissolved in 60 pl of water, and analyzed by OPA-precolumn derivatization HPLC as described above . As expected, a peak at the reten- tion time of OPA-P15 .5 resulted .

Statistical methods In addition to calculation of descriptive statistics (mean ± SD), an analysis of covariance (ANCOVA) was per- formed, with age and sex as the covariates . When more than two groups were compared, an ANOVA was performed ; in case of significant main group effects, group-by-group comparisons were calculated using Student's t test (method : "modified least significant difference tests" to control for the increased type I error rate) . In addition, linear canonical discriminant analysis (method : minimizing Wilks' lambda ; validation method : "jack knife") was performed to analyze the pattern of the several compounds in the patients and the control subjects . Pearson product-moment correlation coefficients were calculated to assess correlations (level of significance : 5%, two-tailed) .

RESULTS

Analytical aspects Figure l shows typical HPLC chromatograms of a CSF sample and of reference amino acids . The OPA derivatives of most amino acids [including aspartate FIG. 1 . HPLC analysis of CSF samples . A : Representative chro- (Asp), glutamate (Glu), a-aminoadipic acid (a- matogram of OPA derivatives from CSF sample . B: Separation AAA), phosphoethanolamine (Pea), asparagine of a mixture of reference amino acids . Derivatization and chro- matographic conditions are described in Materials and Methods . (Asn), serine (Set), citrulline (Cit), glycine (Gly), IS, internal standard . threonine (Thr), arginine (Arg), alanine (Ala), tau- rine (Tau), tyrosine (Tyr), (Met), trypto- phan (Trp), phenylalanine (Phe), isoleucine (lie), and leucine (Leu) ] were well resolved . However, coelution of the [M+H] + and [M-H] - ions, respectively, as of Gln with histidine (His), of GABA with homocar- well as numerous fragment ions (Fig . 2) . These spectra nosine, and of HCA with HCSA occurred . Therefore, were indistinguishable from the corresponding spectra these compounds were quantified together . of synthetic FMOC-y-Glu-Gln . The chromatographic and mass spectral data provided solid evidence that Isolation of P15.5 and its identification as the two compounds are identical . y-Glu-Gln The chromatographic retention times of OPA-P15 .5 Clinical aspects under three different mobile phase conditions sug- After controlling for possible effects of age and sex, gested strongly that the compound was identical with the CSF levels of NAA and NAAG as well as of 5- OPA-y-Glu-Gln, but because of the complexity of the HIAA and HVA were within the same range in the composition of CSF, additional procedures were per- patients and the control subjects (Table 1) . With re- formed to confirm the structure of P15 .5 . In all of gard to the amino acids, Ile was significantly higher the chromatographic systems used for isolation and and Tau was significantly lower in the patients than in purification of FMOC-P15 .5 from the large excess of the controls . Furthermore, Glu was lower in the pa- other derivatized and underivatized endogenous com- tients ; however, this difference did not quite reach the ponents, this compound eluted at the same retention level of significance (p < 0.10) . The CSF concentra- time as FMOC-y-Glu-Gln . In several systems, coelu- tions of the remaining amino acids investigated were tion of FMOC-P15 .5 with synthetic FMOC-y-Glu-Gln similar in the patients and the control subjects . Among was substantiated by coinjection of the two com- the unidentified compounds, P15 .5 (newly identified pounds . The positive and negative ion CF-FAB mass as the dipeptide y-Glu-Gln) was significantly de- spectra from FMOC-P15 .5 showed abundant signals creased in the patients .

t. Neuro,henr . Vol. 6.5, No. 6, 1995 2656 K. Q . DO ET AL.

FIG . 2 . Negative ion mass spectrum of FMOC- P15 .5 from a micro-lC-CF-FAB-MS run ; back- ground substracted . Most of the peaks not labeled with m/z values appeared to belong to residual background, as judged from their mass chromato- grams. Compound abbreviations as in Fig . 1 .

Age effects Course of illness Significant age effects were found for Gly only . Cal- To control for the duration of illness, we divided culation of Pearson correlation coefficients showed a the patient sample into the following three groups positive and significant association of age and Gly in (DSM-III-R definition) : "acute" (schizophreniform), the total sample (r = 0 .39, p < 0.05) but neither for duration < 6 months (n = 9) ; subchronic, between 6 the patient group (r = 0 .35, p > 0.10) nor for the months and 2 years (n = 9) ; and chronic, >2 years control subjects alone (r = 0 .41, p > 0.10) . The corre- (n = 8) . ANOVA yielded significant main group ef- lation coefficients between age and NAA, NAAG, fects for Ile, Tau, and y-Glu-Gln (Table 4) . Compared the amino acids, y-Glu-Gln, and the monoamine me- with the control subjects, y-Glu-Gin was decreased in tabolites were far from being of significance (r < 0 .20, the acute patients, Ile was increased in the acute and p > 0.25) . subchronic patients, and Tau was decreased in all three subgroups of patients . Sex effects ANCOVA revealed significant effects of sex on Acute exacerbation of illness Asp, Cit, and y-Glu-Gln . Although Asp was, on aver- To determine which of the compounds might be age, elevated in women compared with men, no sig- altered, dependent we nificant main group effect was found (Table 2) . The on the state of the illness, com- values of both Cit and y-Glu-Gln showed a significant pared the patients with an acute exacerbation of the main group effect, which in the case of Cit was due disorder at the time of investigation (n = 20 ; 11 to elevated concentrations of this compound in the women, nine men ; mean age, 30 .9 -- 6 .0 years) and male controls compared with the female controls . Re- those without (n = 6 ; four women, two men, mean garding the y-Glu-Gin values, the main group effect age, 36 .8 ± 11 .4 years) . The only compound differing was due to significantly diminished values in the fe- significantly between the two subgroups was Asp, male patients compared with the male patients and which was higher in the patients without an acute exac- female control subjects . Although on average the male erbation [326 _- 89 and 221 ± 99 pmol/ml ; F(1,24) patients had lower y-Glu-Gln values than the male = 5 .30, p < 0 .05 J . controls, the difference was not statistically significant. Discriminant analysis Severity of illness All of the compounds (except HVA and 5-HIAA) To test for influences of severity of illness, we com- were used as discriminating variables, and the group- pared moderately ill patients (n = 15), severely ill ing variables were control subjects versus schizo- patients (n = l I), and control subjects (Table 3) . phrenic patients . The following compounds were nec- There were significant main group effects for NAA, essary for the best discrimination : Tau, y-Glu-Gln, Ile, Tau, Thr, and y-Glu-Gln . NAA was higher and Thr Glu, and Asp . With the discriminant function obtained, was lower in the severely ill patients than in the moder- -83% of the subjects were classified correctly (Table ately ill patients . Furthermore, Thr was lower in the 5) . The combined use of the five compounds differen- severely ill patients than in the control subjects . Re- tiated schizophrenic patients from control subjects with garding Tau and y-Glu-Gln, the two patient subsam- a specificity of 92.3% (percentage of correctly classi- ples had similar values, both of which were signifi- fied schizophrenic patients) and a sensitivity of 66.7% cantly lower than that of the control group . (percentage of correctly classified control subjects) .

J. Neurochem ., Vol . 65, No . 6, 1995 REDUCED y-Glu-Gln AND TAURINE IN SCHIZOPHRENIA 2657

TABLE 1 . Concentrations (pmollml) of various compounds of the CSF of schizophrenic patients and control subjects

ANCOVA

Covariates

Schizophrenic Control Group Age Sex patients subjects effect effects effects (n = 26) (n = 15) tF(1,39)i F(1) F(1)

Age (yr) 32.2 ± 7.8 36 .0 - 12 .5 1 .43 - - Women/men 15111 9/6 X2 = 0.02 - - NAA 800 { 620 810 t 380 0.02 0.06 0.76 NAAG 2,110 ! 900 2,220 ± 870 0.03 0.94 0.28 NAA/NAAG (ratio) 0.38 ± 0.19 0.41 ± 0.21 0.32 1 .77 1 .72 Ala 4,044 ± 579 4,235 - 646 0.67 0.84 1 .37 Arg 9,640 ± 1,244 9,329 ± 809 0.61 0.23 0.00 Asn 6,013 ± 657 5,993 } 923 0.00 0.30 0.70 Asp 245 ± 106 232 !- 182 0.17 0.07 6.47" Cit 5,381 ± 940 5,812 ± 1,230 1 .29 2.28 9 .38" GABA + homocamosine 7,326 ± 642 7,578 ± 573 1 .39 0.33 1 .06 Gln + His 11,727 ± 1,388 11,810 ± 1,040 0.01 0.45 0.00 Glu 323 ± 90 443 ± 298 2.76 0.02 2.32 Gly 4,478 ± 751 4,630 ± 746 0.00 6.01' 2.04 HCA + HCSA 53 ± 15 63 ± 21 1 .87 0.91 1 .70 Ile 5,996 ± 650 5,520 ± 645 5 .80" 0.45 0.16 Met 2,735 ± 677 2,879 ± 691 0.72 0.74 0.73 Phe 4,677 ± 1,042 4,286 - 819 1 .60 0.02 0.01 Ser 7,524 { 873 7,676 ± 670 0.36 0.00 1 .78 Tau 4,988 865 5,871 - 630 15 .28' 1 .61 2.70 Thr 9,683 ~- 2,225 10,359 ± 885 1 .25 0.01 0.32 Trp 2,671 ± 595 2,648 t 524 0.01 0.52 0.91 Tyr 7,807 ± 1,383 7,976 - 1060 0.27 0.07 4.05 y-Glu-Gln 2,077 ± 522 2,469 ± 473 7 .59" 0.10 11 .60° 5-HIAA " 19.9 ± 9.4 25 .7 ± 11 .8 1 .37 0.06 3.59 HVA" 38.6 ± 16 .0 47 .7 ± 27.3 0.87 0.37 2.47

All data are mean ± SD values . "p < 0.01 ;''p < 0.05 ; `p < 0.001, by ANCOVA . Schizophrenic patients, n = 17; control subjects, n = 9.

DISCUSSION Analytical considerations We took special care to maintain CSF at ice-bath or In the present study, we investigated NAA, NAAG, lower temperature throughout the sampling, deprotein- amino acids, monoamine metabolites, and a compo- ization, and filtering procedures. These precautions nent (P15.5) that was identified as y-Glu-Gln in the prevent enzyme-mediated breakdown of bound to free CSF of 26 schizophrenic patients who either were drug amino acids, the occurrence of which has been demon- naive (81%) or had been off psychoactive medication strated in the case of GABA (Grossman et al., 1980) . for between 1 and 8 years (19%) . Compared with the Moreover, we avoided using acid for deproteinization control subjects, the patients had significantly higher because it could have led to falsely high levels of Asp concentrations of Ile and significantly lower concentra- and Glu (Ferraro et al., 1983) . Beyond this, we found tions of Tau and y-Glu-Gin. The concentration of Glu, y-Glu-Gin to be unstable when we used strong acid on average, was lower in the patients (this difference for deproteinization . approximated significance) . For the remaining amino Particular difficulties were encountered in the recog- acids measured, as well as NAA, NAAG, HVA, and nition of y-Glu-Gln at low levels in physiological sam- 5-HIAA, the groups did not differ. For the best discrim- ples using ion-exchange chromatography, a technique ination between schizophrenic patients and controls, used in dedicated amino acid analysis, because of vari- the following compounds were necessary : Glu, y-Glu- ations in peak shape and retention time (Hammond et Gln, Asp, Tau, and Ile. The discriminant function com- al ., 1990) . With OPA-precolumn derivatization re- puted with these compounds classified 83% of the sub- versed-phase HPLC, depending on the pH of the mo- jects correctly. The specificity was 92% and the sensi- bile phase, the risk of coelution of y-Glu-Gin with Glu tivity was 67%. is high, especially with a steep gradient of the mobile

J. Neurochem., Vo/. 65, No. 6, 1995

2658 K. Q. DO ET AL.

TABLE 2. Significant effècts of .sex on concentrations (pmollml) of various coml)ounds of the CSF of schizophrenic patients and control subjects

Schizophrenic patients Control subjects MLSD tests

Men Women Men Women ANOVA Patients Control, Men Women (n = 11) (n = 15) (n = 6) (n = 9) I F(3.37)1 (W vs . M) (W vs. M) (SP vs . CS) (SP vs . CS)

Age (yr) 30 .1 ± 7.1 33 .8 ± 8.1 38 .2 ± 12 .5 33 .8 ± 12.8 1 .13 - - - - Asp 187 ± 113 287 ± 80 163 ± 80 279 ' 218 2.22 - - - - Cit 5,722 ± 809 5,137 } 978 6,708 ± 1,049 5,215 ± 978 435" NS c NS NS y-(jlu-Gln 2,379 ± 460 1,854 ± 458 2,733 ± 444 2,293 ± 425 6.43` c NS NS c

SP, schizophrenic patients ; CS, control subjects ; W, women; M, men; MLSD, modified least significance difference test ; (-), not tested ; NS, not significant. " p < 0.05; " 1) < 0.01, by ANOVA. ' Significant group differences.

phase. Coelution probably occurred in the case of some amino acids Asp and Glu . Whereas the functions of reports where high levels of Gin were observed with NAA are still unclear, NAAG is regarded as a potential this method. At pH 7.0 and with a shallow gradient, neurotransmitter or neuromodulator that acts selec- the OPA derivative of y-Glu-Gin eluted -1 min after tively on the NNIDA-type of ionotropic glutamate re- that of Glu (Fig. 1) and its averaged level in control ceptor (Blakely and Coyle, 1988; Trombley and West- CSF was about five times higher than that of Glu. brook, 1990) . Although present at a higher level than Glu, y-Glu- In the schizophrenic patients, the CSF concentration Gln has not been measured routinely in CSF, probably of NAA and NAAG were well comparable with those due to the aforementioned analytical difficulties . observed in the control subjects . Similar results re- For identification of P15 .5 as y-Glu-Gln, we found cently were reported by Swahn (1990) . Furthermore, it essential to use the LC-MS coupling via the CF- no associations were obtained between these com- FAB interface to obtain high-quality spectra of the pounds and the demographic and clinical characteris- small amounts of compound isolated. Traces of resid- tics of the subjects (i .e., age, sex, and course of the ual salts, which often reduce considerably the sensitiv- disorder) . However, the moderately ill patients had ity of FAB-MS, and other remaining contaminants significantly lower levels of NAA than the severely ill were well separated from the analyte by the on-line patients, with that of the control subjects in between. micro-HPLC, and spectra taken immediately before In addition, Nasrallah et al . (1991), using the in vivo and after the elution of the analyte allowed an optimal magnetic resonance spectroscopic method, found de- subtraction of the high background from the FAB ma- creased NAA concentrations in the right (but not in trix in the low mass region of the spectra. the left) hippocampus of schizophrenic patients. Therefore, one might suggest a possible modulatory Pathophysiological considerations involvement of NAA at least in the severity of this We investigated the CSF levels of NAA and NAAG disorder. No definite conclusion can be drawn about because of the metabolic and physicochemical rela- whether NAAG represents a primary candidate in the tionship of these two compounds to the excitatory pathophysiology of schizophrenia, because in the pres-

TABLE 3. Significant effects of the .severity of'illness on concentrations (pmollml) of various compounds of the CSF of .schizophrenic patients and control subjects

Schizophrenic patients MLSD tests Control Moderately ill Severely ill subjects ANOVA Moderately ill Moderately ill Severely ill (n = 15) i n = 11) (n = 15) IF(2,38)l vs . severely ill vs . controls vs . controls

Age (yr) 32 .4 ± 9.0 32 .0 - 6.0 36 .0 ± 12.5 0.70 Women/men 9/6 6/5 9/6 X2(2) = 0.10

NAA 580 ± 170 1,120 ± 880 810 ± 380 3.40" c NS NS Tau 5,191 ± 788 4,709 ± 923 5,871 ± 630 7.42' NS c c Thr 10,282 ± 1,495 8,385 ± 3,082 10,359 ± 885 3.38" c NS c y-Glu-Gln 2,110 ± 552 2,031 -! 501 2,469 ! 473 3.29" NS c c

MLSD, modified least significance difference test ; (-), not tested ; NS, not significant. "/) < 0.05; °p < (1 .01, by ANOVA. ' Significant group differences.

.1. Nruro ( hein ., Vo/. 65. No. 6, 1995

REDUCED y-Glu-Gln AND TAURINE IN SCHIZOPHRENIA 2659

TABLE 4 . Significant effects of the course of illness on concentrations (pmol/ml) of various compounds of the CSF of schizophrenic patients and control subjects

MLSD tests Schizophrenic patients Control Acute Subchronic Chronic Acute Subchronic Chronic subjects ANOVA vs . vs . vs . (n = 9) (n - 9) (n = 8) (n = 15) IF(3,37)] controls controls controls

Age (yr) 33 .3 ± 5 .7 34 .9 ± 9 .5 28 .0 } 6.6 36 .0 ± 12 .5 1 .25 Women/men 7/2 6/3 6/2 9/6 X2(3) = 5 .34 Ile 6,108 ± 663 6,162 ± 694 5,683 r- 541 5,520 - 645 2 .93" b b NS Tau 4,997 !- 1,052 5,149 ± 1,002 4,794 -- 422 5,871 ± 630 4 .13" b b b -y-Glu-Gin 1,891 - 498 2,280 ± 468 2,056 ± 584 2,469 ! 473 2 .87" b NS NS

MLSD, modified least significance difference test; (-), not tested ; NS, not significant . "p < 0 .05 by ANOVA . " Significant group differences.

ent study its CSF level was found to be similar in the paminergic state . Even when considering the proposal patients and the control subjects ; but Tsai et al . (1991) of Davis et al . (1991), that schizophrenia may be char- reported preliminary data showing an elevated post- acterized by both a hyper- and a hypodopaminergic mortem tissue level of NAAG in the frontal and pari- state, our findings do not support the dopamine theory etal cortex of schizophrenic patients . of schizophrenia, because the CSF HVA concentration The CSF concentrations of the monoamine metabo- was not increased or decreased in either the total pa- lites HVA and 5-HIAA were very similar in the schizo- tient sample or various subgroups of the sample . phrenic patients and the control subjects . Furthermore, The analysis of the amino acid concentrations re- no clear associations between these metabolites and vealed significantly diminished concentrations of Tau age, sex, severity of the illness, or an acute exacerba- and significantly elevated levels of lie in the patients tion of the disorder emerged . In general, these findings with schizophrenia compared with the controls . Fur- are in accordance with those of numerous other CSF thermore, in the patients, the dipeptide y-Glu-Gln was studies (see Widerl6v, 1988 ; Pickar et al ., 1990) . In diminished highly significantly and the Glu concentra- the meantime, it has become clear that the CSF sam- tion tended to be decreased . The remaining amino acids pling technique for determination of HVA concentra- investigated, including Asp, Gly, Ser, and GABA, did tions bears several limitations ; e .g ., there is a substan- not differ between the two samples . Thus, the changes tial gradient along the spinal cord, acute neuroleptic observed for a few amino acids appear to be selective treatment can elevate HVA but this effect may be abol- against the background of a majority of unaffected ished during chronic neuroleptic administration, and, amino acids . finally, the degree of motor activity before lumbar In the present study, Tau, an inhibitory amino acid puncture affects substantially the CSF HVA concentra- that is also involved in osmoregulation (for an over- tions . In the present study, some (but not all) of these view, see Huxtable, 1992), was found to be reduced limitations were controlled [e .g ., motor activity before significantly in the CSF of the schizophrenic patients lumbar puncture, drug status, and possible circadian regardless of the patient's age and sex as well as the influences (see Materials and Methods)] ; but never- course and severity of the illness . These observations theless the results obtained do not fit with the hypothe- are at odds with reports in the literature (Korpi et al ., sis that schizophrenia is a manifestation of a hyperdo- 1987 ; Macciardi et al ., 1990 ; Rao et al ., 1990 ; Prieto- Rinc6n et al ., 1991) . An adequate explanation of these discrepancies is difficult . One might argue that most TABLE 5 . Classification table of schizophrenic patients of our patients were drug naive, whereas in the other and control subjects using Tau, y-Glu-Gln, Ile, studies the patients were on neuroleptics (Macciardi Glu, and Asp as discriminating variables et al ., 1990), they were off drugs for 13-45 days Predicted group (Korpi et al ., 1987), or they were a mixed sample membership" in this regard (Prieto-Rinc6n et al ., 1991 ) . To our knowledge, no study has focused on the effects of Actual group Number of Schizophrenics Controls neuroleptics on the CSF concentration of Tau . How- membership subjects (%) (%a) ever, at least the serum concentrations of Tau were Schizophrenic patients 26 24 (92.3) 2 (7 .7) reported to be similar in drug naive and off-drug and Control subjects 15 5 (33 .3) 10 (66 .7) on-drug schizophrenic patients (Rao et al ., 1990), ren- dering pharmacological influences unlikely . One might Correctly classified subjects : 82.9% . further speculate that the differing methods used could

J. Neurochern ., Vo/ . 65, No . 6, 1995 2660 K. Q. DO ET AL.

be responsible for the discrepancies between our obser- decreased in those patients with a relatively short dura- vation and those reported by others . However, the tion of the disorder (<6 months), indicating a func- decrease in Tau observed in the present study is consis- tional role of this dipeptide predominantly during the tent with findings of others that the enzymes methio- initial course of schizophrenic disorders. nine adenosyltransferase and serine hydroxymethyl- Up to now, no consideration has been given to a transferase, which are responsible for the synthesis of possible involvement of y-Glu-Gln in the pathophysi- the Tau precursors S-adenosylmethionine and Ser, ology of schizophrenia. This dipeptide was first iso- have decreased activity in schizophrenia (Carl et al ., lated from bovine brain (Kakimoto et al., 1964) and 1978 ; Kelsoe et al ., 1982) . has been reported to be widely distributed in various As in the present study, Macciardi et al . (1990) mammalian tissues and human brain areas (Kanazawa observed elevated Ile levels in the plasma of schizo- and Sano, 1967 ; Sano, 1970) . It has also been identi- phrenic patients. Furthermore, in the present study Ile fied in human plasma and CSF (Hammond et al., was one of five amino acids necessary for the best 1990) . The y-Glu dipeptides are most probably syn- discrimination between the patients and the controls. thetized by the enzyme y-glutamyltransferase, a mem- However, the significance of these observations re- brane-bound enzyme that is involved in transport pro- mains unclear, not only because other investigators cesses across the cell membrane of glutathione, y- reported quite undisturbed levels of this compound in glutamyl dipeptides, and amino acids in the y-glutamyl both drug-naive and drug-free schizophrenics (Rao et cycle (Meister and Anderson, 1983) . Indeed, a sig- al ., 1990 ; Pietro-Rinc6n et al., 1991), but also because, nificant pathway of glutathione breakdown involves a so far, no solid hypothesis can be formulated about reaction, catalyzed by y-glutamyltransferase, in which a possible involvement of Ile in the etiology or the the y-glutamyl moiety is transferred to an amino acid pathophysiology of schizophrenia . acceptor, with L-Gin and L-Met being the best ac- In the schizophrenic patients, the mean CSF concen- ceptors (Tate and Meister, 1974) . Furthermore, y-glu- tration of Gin was diminished compared with control tamyltransferase has been proposed to be involved in subjects. Although this difference failed to reach statis- the transport of some amino acids, including Gin and tical significance (0.05 < p < 0.10), it is sufficiently Gin into various brain compartments (Emmuels et al., high (-25%) to warrant a comment here. Several au- 1978; Lisy et al., 1983 ; Kvamme et al., 1985) . This thors did not find any changes in the Glu concentra- is supported by the fact that y-glutamyltransferase is tions in serum, plasma, or CSF of schizophrenic pa- associated with glutamatergic structures in the hippo- tients (Perry, 1982; Gattaz et al., 1985; Korpi et al., campus (Stastny et al., 1988), one of the brain regions 1987) . On the other hand, the present observation of thought to be involved primarily in schizophrenia (e .g., decreased Glu is in line with that reported by Alfreds- Altshuler et al ., 1987), and that the Glu uptake in brain son et al . (1985 ), Rao et al. (1990), and Prieto-Rinc6n slices and cell cultures is reduced in the presence of et al . (1991) . Thus, there is some evidence for both a y-glutamyltransferase inhibitor (Lisy et al ., 1983 ; the initial "glutamatergic theory" (Kim et al., 1980) Jankaskova et al., 1992) . Thus, the decreased level of and its modification, in which a dysfunction of the y-Glu-Gln may reflect a deficiency in the y-glutamyl- PCP/NMDA-receptor complex is postulated as being transferase system (either in its level or its activity) primarily involved in schizophrenic disorders or a deficiency in Gin, an important precursor of releas- (Deutsch et al., 1989; Heresco-Levy et al., 1993) . able Gin . In both cases, our finding of diminished y- Moreover, the present investigation provides additional Glu-Gin concentrations is related to Gin and would evidence supporting rather than refuting the "gluta- be consistent with the hypothesis that this excitatory mate theory" ; i.e., Glu was one of the five amino acids neurotransmitter and possibly the NMDA receptor are necessary to best discriminate between schizophrenic involved in the pathophysiology of schizophrenia . In patients and controls, a finding similar to that noted this context, it is worth noting that Met given per os by Macciardi et al. (1990), and y-Glu-Gln was sig- has been shown to be the only amino acid that exacer- nificantly decreased in the patients, a finding that is bates the psychotic symptoms in schizophrenic patients also consistent with a general perturbation of Glu me- (Park et al., 1965) . This observation might be related tabolism (for discussion, see below) . to a deficit in the y-glutamyltransferase system in The CSF concentration of the dipeptide y-Glu-Gln which Met competes with Glu. An excess of Met was diminished highly significantly in the schizo- would impede the already deficient transport of Gin. phrenic patients compared with the control subjects. However, the assumption that a decreased level of y- This finding was independent of the severity of the Glu-Gln reflects a deficiency in the T-glutamyltransfer- disorder and whether the patients were investigated ase system, which would be in line with the gluta- during an acute exacerbation of the disorder. Although matergic and the PCP/NMDA-receptor hypothesis, re- a significant sex effect was observed, in that women mains speculative . showed lower values than men, subsequent analyses In summary, our findings obtained in patients with revealed that this effect was apparently independent of schizophrenic disorders who either had been drug free whether the subjects were patients or controls. On the for > 1 year or were drug naive provide little support other hand, the level of y-Glu-Gln was most markedly for the dopaminergic theory of schizophrenia ; i .e., the

J. Neunhem ., Vol. 65, No . 6, 1995 REDUCED y-Glu-Gln AND TAURINE IN SCHIZOPHRENIA 2661

CSF concentration of the monoamine metabolites in- Gamma-glutamylglutamine identified in plasma and cerebrospi- nal fluid from hyperammonaemic patients . Chn. Chim. Acta vestigated did not differ between the patients and the 194, 173- 184 . control subjects . However, there were several observa- Heresco-Levy U ., Javitt D. C., and Zukin S . R. (1993) The phency- tions that fit well with the glutamatergic hypothesis of clidine/N-methyl-t)-aspartate (PCP/NMDA) model of schizo- schizophrenia. Although the CSF concentration of Gin phrenia: theoretical and clinical implications . Psychiatr. Anna. showed a tendency to be decreased only in the patients, 23, 135-143 . Huxtable R. J . ( 1992) Physiological actions of taurine . Physiol . Rev. this excitatory amino acid was one of the five amino 72, 101-163 . acids necessary for the best discrimination between the Jankaskova B., Lisy V., and Stastny F. ( 1992) Effect of gamma- patients and the controls. Furthermore, the compound glutamyl transpeptidase inhibitors on the transport of glutamate that was identified as y-Glu-Gln was diminished highly into neuronal and glial primary cultures . Int. J. Der . Neurosci. significantly in the patients. Inasmuch as this dipeptide 10, 225-230 . Kakimoto Y., Nakajima T., Kanazawa A., Takesada M., and Sano appears to be involved in transport processes of both I. ( 1964) Isolation of gamma-r, glutanryl-r. and Gin and Gin, the latter being the precursor of releasable gamma r.-glutamyl-L-glutamine from bovine brain. Biochim . Glu, this finding provides strong support for the gluta- Biophys. Acta 93, 333-338 . mate hypothesis. Moreover, the decreased CSF con- Kanazawa A . and Sano 1. (1967) The distribution of gamma-t,- centration of y-Glu-Gln also points toward a deficiency glutamyl-t-glutamine in mammalian tissue. J. Neurochem . 14, 596-598 . in the y-glutamyltransferase system, which could play Kelsoe .1 . R., Tolbert L. C., Crews E. L., and Smythies J. R. ( 1982) a significant role in the pathophysiology of schizo- Kinetic evidences for decreased methionine adenosyl-trans- phrenia. ferase activity in erythrocytes from schizophrenics . .I. Neurosci . Res . 8, 99-103 . Kim J . S., Kornhuber H . H., Schmid-Burgk W., and Holzrnüller REFERENCES B. ( 1980) Low cerebrospinal fluid glutamate in schizophrenic patients and a new hypothesis of schizophrenia. Neurosci. Len. Alfredsson G., Lindberg M., and Sedvall G. (1985) Relationship 20, 379-382. between glutamate and glutamine in cerebrospinal fluid. Klancnik J . M., Cuénod M., Gähwiler B. H., Jiang Z. B., and Do (Abstr .) .I. Neurochem. 44 (Suppl .), 195 . K. Q. ( 1992) Release of endogenous amino acids, including Allshuler L. 1.., Conrad A., Kovelman J. A., and Scheibel A. (1987) homocysteic acid and cysteine aulphinic acid, from rat hippo Hippocampal pyramidal cell orientation in schizophrenia: a con- campal slices evoked by electrical stimulation of Schaffer col- trolled neurohistologie study of the Yakovlev collection. Arch. lateral-commissural fibres . Neuroscience 49, 557-570 . Gen. Psychiatry 44, 1094-1098 . Korpi E. R., Kaufmann C. A., Marnela K. M., and Weinberger D. R. American Psychiatric Association ( 1987) Diagnostic and Statistical ( 1987) Cerebrospinal fluid amino acid concentrations in chronic Manual of Mental Disorders, 3rd edit. revised . American Psy- schizophrenia . Psychiatry Res. 20, 337-345 . chiatric Association, Washington, D .C. Kvamrne E., Schousboe E., Hertz L., Torgner I. A., and Svenneby Blakely R. D. and Coyle J. 'f. (1988) The neurobiology of N- Developmental changes of endogenous glutamate acetylaspartylglutarnate, in of Neurobiol- G. (1985) International Review and gamma-glutamyl transferase in cultured cerebral cortical ogy, Vol. 30 (Smythies J. R. and Bradley R. J., eds), pp. 39- interneurons and cerebellar granule cells, and in mouse cerebral I00. Academic Press, San Diego. cortex and cerebellum in vivo . Neurochem. Res . 7, 993-1008. Carl G. F., Crews E. L., Carmichael S. M., Benesh F. L., and Smy- thies J. C. ( 1978) Four enzymes of one-carbon metabolism in Lindroth P. and Mopper K. (1979) High performance liquid chro- blood cells in schizophrenia . Biol. Psychiatry 13, 773-776 . matographic determination of subpicomole amounts of amino Carlsson M. and Carlsson A. ( 1990) Interactions between gluta- acids by precolumn fluorescence derivatization with o-phthalal- matergic and monoaminergic systems within the basal gan- dehyde . Anal. Chem. 51, 1667-1674 . glia-implications for schizophrenia and Parkinson's disease . Lisy V., Stasmy F., Murphy S ., and Hajkova B. (1983) Glutamate Trends Neurosci. 13, 272-276. uptake into cerebral cortex slices is reduced in the presence Davis K. L., Kahn R. S., Ko G., and Davidson M . (1991 ) Dopamine of a gamma-glutamyl transpeptidase inhibitor . Experientia 39, in schizophrenia : a review and reconceptualization . Am. J. Psy- III . chiatrv 148, 1474-1486 . Macciardi F., Lucca A., Catalano M., Marino C., Zanardi R., and Deutsch S. 1 ., Mastropaolo J., Schwartz B. L., Rosse R. B., and Smeraldi E. (1990) Amino acid patterns in schizophrenia : some Morihisa J. M. ( 1989) A "glutamatergic hypothesis" of schizo- new findings . Psychiatry Res. 32, 63-70. phrenia. Clin. Neuropharneacol. 12, 1-13. Matthysse S. (1973) Antipsychotic drug actions : a clue to the netu-o- Einarsson S ., Josefsson B ., and Lagerkvist S . (1983 ) Determination pathology of schizophrenia? Fed. Proc . 32, 200-205 . of amino acids with 9-fluorenylmethyl chloroformate and re- Meister A. and Anderson M. E. (1983) Gluthadione . Anna . Rev. versed-phase high-performance liquid chromatography . ,/ . Biochem. 52, 711 -760 . Chroinatogr. 282, 609-618 . Nasrallah H. A., Skinner T. S., Schmalbrock P., Obringer A., and I'.nunuels S ., Fish I., and Freedman L. S. (1978) Effect of y-glutamyl Robitaille P. M. L. ( 1991 ) In vivo H 1 magnetic resonance cycle inhibitors on brain amino acid transport and utilization. spectroscopy (MRS) of the hippocampus/amygdala complex Neurochem. Res . 3, 619-631 . in schizophrenia. Soc . Neurosci. Abstr. 17, 1455 . Ferraro T. N., Manyam B. V., and Hare T. A. (1983) Further charac- Park L. C., Baldessarini R . J., and Kety S. S. (1965) Methionine terization of in vitro conditions appropriate for GABA determi- effects on chronic schizophrenics . Arch. Gen. Psychiatry 12, nation in human CSF : impact of acid deproteinization and 346-351 . freeze/thaw . J. Neurochem . 41, 1057-1064 . Perry T. L. (1982) Normal cerebrospinal fluid and brain glutamate Gattaz W . F., Gasser T., and Beckmann H. (1985) Multidimensional levels in schizophrenia do not support the hypothesis of gluta- analysis of the concentrations of 17 substances in the CSF of matergic neuronal dysfunction . Neurosci. Lett. 28, 81-85. schizophrenics and controls . Biol. Psychiatry 20, 360-366 . Piekar D., Breier A., Hsiao J . K ., Doran A. R., Wolkowitz O. M., Grossman M. H., Hare T. A., and Manyam N. V. B . ( 1980) Stability Pato C. N., Konicki P. E., and Potter W. Z. ( 1990) Cerebrospi- of GABA levels in CSF under various conditions of storage . nal fluid and plasma monoamine metabolites and their relation- Brain Res. 182, 99-106 . ship to psychosis . Arch. Gen. Psychiatry 47, 641-648 . Hammond J. W., Potter M., Tsucott R., and Wilcken B. (1990) Prieto-Rinc6n D., Piner6a-Shuhaibar L., Bonilla E., Prasad A., and

t. Neuroehein ., Vol. 65, No . 6, 1995 2662 K. Q. DO ET AL.

Arrieta A. (1991) Esquizofrenia paranoide : aminoácidos libres Tate S. S. and Meister A. ( 1974) Interaction of gamma-glutamyl en et liquido cefalorraquideo . Invest. Clin. 32, 149-155 . transpeptidase with amino acids, dipeptides, and derivates and Rao M. L., Gross G., Strebel B., Br5uning P., Huber G., and Kloster- analogs of glutathione. .l. Biol. Chem. 249, 7593-7602 . k6tter J . ( 199(1) Serum amino acids, central monoamines, and Trombley P. Q. and Westbrook G. L. ( 1990) Excitatory synaptic hormones in drug-naive, drubfree, and neuroleptic treated transmission in cultures of rat olfactory bulb . J. Neurophvsiol. schizophrenic patients and healthy subjects . PsYchiatrv Res . 34, 64, 598-606 . 243-257. Tsai G., Slusher B. S., Carter R. E., Passani L., Casanova F., Sano 1. ( 1970) Simple peptides in brain, in International Revieu, of Kleinman J., and Coyle J. T. ( 1991 ) Abnormalities in NAAG Neurobiologv, Vol. 12 (Pfeiffer C. X. and Symthies J. R., eds), and NAALADase levels in schizophrenic brain. Soc . Neurosci. pp. 235-263. Academic Press, New York. Abstr. 17, 1455 . Seeman P., Niznik H . B., and Guan H. C. ( 1990) Elevation of Van Tol H. H. M., Wu C. M., Guan H. C., Ohara K., Bunzow J . R., D2 receptors in schizophrenia is underestimated by radioactive Civelli O., Kennedy J., Seeman P., Niznik H. B ., and Jovanovic raclopride . Arch. Gen. Psychiatrv 47, 1170-1172 . V. (1992) Multiple dopamine D4 receptor variants in the human Seeman P., Guan H. C., and Van Tol H. H . M. (1993) Dopamine D4 population . Nature 358, 149-152 . receptors elevated in schizophrenia . Nature 365, 441-445 . Widerl6v A appraisal monoamine metabo- Squires R . F. and Saederup E. (1991) A review of evidence for E. ( 1988) critical of CSF GABergic predominance/glutamatergic deficit as a common lite studies in schizophrenia . Ann . NY Acad . Sc i. 537, 309-323 . etiological factor in both schizophrenia and affective psychoses : Wittchen H.-U ., Zaudig M., Schramm E., Spengler P., Mombour more support for a continuum hypothesis of "functional" psy- W., Klug J ., and Horn R., eds ( 1990) Strukturiertes Klinisches chosis . Neurochem. Res. 16, 1099-1111 . Interview für DSM-111-R. Eine deutschsprachige, erweiterte Stasmy F., Rothe F., Schmidt W., Wolf G., Keilhoff G., and Lisy Bearbeitung der amerikanischen Originalversion des SCID von V. ( 1988) Changes in the activity of gamma-glutamyl transpep- R. I,. Spitzer, .l. B. W. Williams, M. Gibbon. Beltz, Weinheim. tidase induced by and surgical lesions of the hippo- Zollinger M., Amsler U ., and Brauchli J . ( 1990) Quantification of campal formation in young rats . Brain Res . 462, 56-61 . N-acetylaspartylglutamate, an N-terminal blocked dipeptide Swahn C.-G. ( 1990) Determination of N-acetylaspartic acid in hu- neurotransmitter candidate, in brain slices superfused by gas man cerebrospinal fluid by gas chromatography-mass spec- chromatography-mass spectrometry . .I. Chromatogr. 532, trometry . .1 . Neurochem. 54, 1584-1588 . 27-36.

1 . Neuro, hon ., Vo/. 65. No. 6, 1995