Molecular Psychiatry (1998) 3, 227–237  1998 Stockton Press All rights reserved 1359–4184/98 $12.00

ORIGINAL RESEARCH ARTICLE Superoxide, neuroleptics and the ubiquinone and in brain and lymphocytes from normals and schizophrenic patients SA Whatley1, D Curti2, F Das Gupta1, IN Ferrier3, S Jones1, C Taylor4 and RM Marchbanks1,5

1Department of Neuroscience, Institute of Psychiatry, Denmark Hill, London SE5 8AF, UK; 2Istituto di Farmacologia, Universita di Pavia, 11 Piazza Botta, 27100 Pavia, Italy; 3MRC Neurochemical Pathology Unit, Newcastle General Hospital, NE4 6BE, UK; 4Department of Forensic Psychiatry, Institute of Psychiatry, Denmark Hill, London SE5 8AF, UK

The effects of the neuroleptic flupenthixol on the expression of the genes coding for the mito-

chondrial ubiquinone and cytochrome b5 reductases have been studied because of the impor- tance of these in energy , oxidative stress and also because similar but oppositely directed changes have been previously observed in the cerebral cortex from schizophrenics. The neuroleptic flupenthixol reduces the expression in rats of the gene

coding for NADH-cytochrome b5 as measured by in situ hybridisation and its enzy- mic manifestation. Flupenthixol also reduces the enzymic activity of the mitochondrial NADH- ubiquinone reductase, and it has been previously shown that mRNA from the mitochondrially coded parts of the are reduced by the drug. Both the cis- and therapeutically less active trans-flupenthixol were found to produce these changes in rats. Post-mortem brain tissue from schizophrenics who have received neuroleptic medication have reduced levels of both reductases as measured enzymically, Lymphocyte samples from schizophrenics also − have reduced levels of both reductases compared with normals. The superoxide anion O2 is the principle agent of oxidative stress and both the cytochrome b5 and the ubiquinone reductase enzymes were semi-purified from sheep liver and shown to produce appreciable amounts of superoxide. Superoxide production is reduced in brain homogenates from rats treated with flupenthixol. Its production is also reduced in brain tissue and lymphocytes from schizophrenics receiving neuroleptic medication. We conclude that neuroleptic medication

reduces the expression of both the ubiquinone and and among the effects of this reduction is a decrease in the production of neurotoxic superoxide. Keywords: oxidative stress; flupenthixol; psychosis; Complex I; mitochondria; gene expression; in situ hybridisation

Introduction a disjunction between the timescale of biological effect and therapeutic efficacy.1 Schizophrenia is characterised by delusional experi- Liability to schizophrenia is partially inherited. Cal- ences and thought disorder with a lifetime morbidity culations from twin and family studies indicate that of the order of 1%. Studies on the biological basis of approximately 60–70% of the risk is inherited. A fairly the psychoses have turned up many sometimes contra- intensive study of those parts of the human genome dictory clues but the substantive aetiological theories that code for dopamine receptors and related enzymes result from the effects of drugs which afford some mea- has not so far revealed a linkage that is reproducible sure of symptomatic relief. Hence the ‘dopaminergic’ under extended scrutiny.2,3 A linkage study covering hypothesis of schizophrenia because neuroleptic medi- the whole nuclear coded genome at an average 9 cM cations are found to affect various parts of the dopa- intervals has revealed about 25 loci linked to schizo- mine neurotransmitter system. Despite intensive inves- phrenia4 with probabilities greater than P Ͻ 0.05, but tigation the ‘dopamine’ hypothesis has not been found there does not appear to be any functional connection to be particularly predictive and in particular there is between these loci. An alternative approach is to try and find out which Correspondence: SA Whatley, Dept of Neuroscience, Institute of genes are expressed differently in patients suffering Psychiatry, Denmark Hill, London SE5 8AF, UK. E-mail from psychosis. We characterised the mRNA from post- ϽspbcsawȰiop.bpmf.ac.ukϾ mortem brain of schizophrenics and normals by 2-D 5Current address: Wessex Human Genetics Institute, Duthie translational analysis to establish that it was not so Building (Mailpoint 808), Southampton General Hospital, SO16 5–7 6YD, UK degraded as to confound the whole approach. The Received 21 July 1997; revised 19 November 1997; accepted 10 mRNA populations in post-mortem brain of schizo- December 1997 phrenics were then compared to normals by differen- Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 228 tial screening, a protocol that allows the identification samples 24 were from the MRC Brain Bank, Newcastle of genes whose relative transcript levels are altered General Hospital and a further three schizophrenic and between two tissue types.8 Five clones were selected three normal samples were provided from the Institute as being differentially expressed in schizophrenics and of Psychiatry brain bank. The sample set is described were then found to be derived from mitochondrial in Table 1. Because of varying tissue availability the (mt)DNA; four of these were for ribosomal RNA but an same samples were not always used in each study. upregulated clone coding for cytochrome sub- Diagnosis of the patients was carried out by means of unit II was also found.9 DSM-III. Death was at average age of 65.7 × 16 years Neuroleptic medication is known to cause changes (normals) and 70.6 ± 10 years (schizophrenics) and was in gene expression in brain,10 and in parallel studies it from various causes. The post-mortem delay ranged was found that cis(␣)-flupenthixol by comparison with from 3 to 84 hours with averages of 25.6 ± 13 (normals) its therapeutically inactive isomer trans(␤)-fluphen- and 69.4 ± 3.3 (schizophrenics) and 41.2 ± 16.8 thixol caused the differential expression in rat brain of (unmedicated schizophrenics). Because of the rather 29 clones, 14 of which were of mitochondrial origin (S higher post-mortem delay for the schizophrenic cohort Jones, personal communication and Whatley et al11). a sub-group with shorter post-mortem delay (39.2 ± 6.7 Predominant among the mitochondrial sequences were n = 5) was identified and the results of enzyme assays those coding for cytochrome oxidase subunits I and II and superoxide production calculated separately. In no and the NADH-ubiquinone reductase, all of which case was there a significant difference between the were downregulated by cis-flupenthixol. Expression of results from the short and long post-mortem delay sub-

the nuclear gene coding for NADH-cytochrome b5 groups. reductase was also found to be downregulated by cis- flupenthixol. Human lymphocytes These were obtained from nor- There is therefore an apparent involvement of the mals and male patients admitted to the hospital wing processes of mitochondrial oxidative energy metab- of Broadmoor. Lymphocytes were isolated from an 8- olism in the aetiology of schizophrenia and an effect ml whole blood sample by the technique of Boyum.13 of neuroleptic medication on the same processes. An Diagnosis of these patients was carried out by means

additional issue is that both the ubiquinone and the b5 of DSM-III and the sample set is described in Table 2. reductase enzymes are flavoproteins that while All tissue was stored at −70°C until use. mediating the transfer of electrons from NAD(P)H are known to produce small amounts of superoxide which Enzyme preparations NADH ubiquinone and cyto-

has been invoked as an aggravating or even causal fac- chrome b5 reductases and cytochrome b5 itself were tor in neurotoxicity and mental illness.12 To study this, made from 66 g fresh sheep liver following the general a simple method of measuring superoxide production procedure described by Yang and Cederbaum.14 The has been devised and applied to the brain samples. tissue was homogenised in 600 ml 0.25 M sucrose buff- In this investigation we seek to relate the amounts of ered with 10 mM Tris-Cl pH 7.4 and 1.0 mM EDTA. the reductases in brain and lymphocyte tissues from The nuclear fraction was removed as a pellet after cen- schizophrenics, the extent to which any changes might trifugation twice at 500 g for 10 min. Mitochondria be due to neuroleptic medication and the role both were sedimented at 11 000 g for 16 min and washed play in the production of toxic superoxide. twice with 20 volumes sucrose buffer. The mitochon- Preliminary studies11 of the activity of the two drial fraction was frozen in aliquots and used without reductase enzymes suggested that in the brains of further treatment except for dilution to 0.85 ␮g schizophrenic patients either the respiratory chain ␮l−1 as a source of NADH-ubiquinone reductase. Each (ubiquinone) reductase activity was decreased with a microgram of protein in the preparation was derived

compensating increase in the outer membrane (b5) from 0.26 mg tissue. reductase or that there was a change in the properties Polyethylene glycol (PEG average molecular wt of ubiquinone reductase in schizophrenics such that it 6000) was added to the post-mitochondrial supernatant is less sensitive to rotenone. In order to distinguish and to a final concentration of 8.5% and spun at 13 000 g measure the two reductases correctly in small amounts for 20 min. The microsomal pellet was washed twice of tissue we have investigated the properties of the with 60 ml PEG-sucrose. The washed pellet was resus- semi-purified enzymes and applied improved methods pended in 100 ml (ie approximately 8 mg protein ml−1) so that rotenone sensitivity of the ubiquinone of 0.1 M Tris-acetate pH 8.1 containing 1 mM EDTA, reductase from schizophrenics can be independently 1 mM phenylmethyl sulphonyl fluoride and 2% final monitored to resolve this question. concentration Triton X-100 and stirred for 1 h. After centrifugation at 13 000 g for 25 min, the supernatant was taken for chromatography onto 20 g DEAE cellu- Methods lose in a 400 mm × 20 mm column equilibrated with 80 mM Tris-acetate buffer pH 8.1 containing 1 mM Human brain tissue This was obtained from frontal EDTA and 2% Triton X-100. After charging the column cortex obtained post-mortem from 10 normals, 15 it was eluted with the Tris-acetate buffer. NADH-ferri- schizophrenic patients of which five had not received cyanide reductase activity was assayed in the eluate medication within 4 weeks of death. Of the brain and the peak fractions (282–561 ml) pooled to consti- Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 229 Table 1 Details of post-mortem frontal cortex samples from normals and schizophrenics

Group Sex Age Age PM Cause Medication of onset of death delay (h) of death Psychotropic Other

Normal M – 64 8 IHD N – F – 59 28 CA N – M – 85 22 IHD N – F – 90 60 IHD N OP F – 81 25 CA N OP F – 78 18 CA N – M – 63 31 IHD N – M – 51 15 COAD N – M – 51 14 S N – M – 63 27 IHD N – M – 38 34 IHD N – Schizophrenic M 32 66 68 IHD Y – M 20 60 41 IHD Y – F 25 71 48 BP Y BZP, Li M398429BPY – M 35 47 36 IHD Y BZP F317164CAY – M 30 72 50 BP,CA Y OP M 28 78 60 IHD Y – M ? 70 48 BP Y – F247697BPY – Unmedicated schizophrenic F307778BPN – M23669 BPN – M 39 71 28 SEPT N – M 30 63 84 GITB N – F30877 BPN –

KEY: IHD, Ischaemic heart disease. CA, carcinomatosis. COAD, chronic obstructive airways disease. S, sudden death. BP, bronchopneumonia. SEPT, septicaemia. OP, optiates. BZP, benzodiazepines. Li, Lithium. GITB, gastrointestinal tract bleeding. Psychotropic medication was either flupenthixol, fluphenazine or thioridazine in the therapeutic range.

tute the source of NADH-cytochrome b5 reductase. drated in 70%, 80% and 90% ethanol (v/v). After dehy- After a further 100 ml of elution with the Tris-acetate dration, delipidation in chloroform for 1 min was per- buffer NASCN (90 mM) was added to it, the Tris being formed followed by rehydration in 100% and 95% correspondingly reduced to 10 mM and elution con- ethanol (v/v). The sections were stored at −20°C. Four tinued with a further 360 ml. This eluate constituted picomoles of specific oligonucleotide probes were ␮ −1 the source of cytochrome b5. Both eluates from the col- end-labelled using terminal (25 U l , umn were dialysed with three changes against 2 L of Promega, UK) at 37°C for 1 h with 40 pmol 35S dATP water. Triton X-100 was removed by treatment with (1200 Ci mmol−1, Dupont, UK) and unincorporated 100 g Amberlite XAD previously washed with two bases removed using Nunctrap push columns changes of methanol and four changes of water. The (Dupont). The probes were then resuspended in solutions were lyophilised and reconstituted as hybridisation buffer (4 × SSC, 50% (v/v) formamide, 1× required to give suspensions (20-fold concentration) Denhart’s solution, 20 mM sodium phosphate, 10% containing 0.12 ␮g protein ␮l−1 of the reductase and (w/v) dextran sulphate, 100 ␮gml−1 salmon sperm 0.3 ␮g ␮l−1 of the cytochrome. Each microgram of the DNA, yeast tRNA, polyadenylic acid and 1% (w/v) reductase was derived from 40.2 mg of tissue and 1 ␮g dithiothreitol) to an activity of 1 × 107 cpm per 90 ␮l. of the cytochrome was derived from 1.46 mg of tissue. Sections were then incubated overnight at 40°Cina All preparations were kept at −70°C. chamber humidified with hydridisation buffer. Follow- ing hybridisation, sections were washed in 1 × SSC In situ hybridisation Rat brain sections (20 ␮m) were (150 mM NaCl, 15 mM sodium citrate) for 4 × 15 min fixed for 1 min in 4% paraformaldehyde and acetylated at 35°C, followed by two 60-min washes at room tem- for 1 min in 0.1 M triethanolamine containing 0.9% perature, and were allowed to dry in air. Exposure to (v/v) NaCl, pH 8.0, and 0.25% (v/v) acetic anhydride. autoradiographic film (Hyperfilm-3H, Amersham, UK) Sections were then washed for 1 min in PBS and dehy- was between 10–15 days and the resulting autoradio- Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al

230 ␮ Table 2 Details of lymphocyte samples from schizophrenic of a final concentration of 100 M of CoQ1 ina20mM patients KPi buffer pH 8.1 containing 1 mM KCN. Extinction coefficient values of 1.02 (ferricyanide), 12.0 (DCIP), Group Sex Age Medication 12.7 (cytochrome c) and 4.7 (NAD(P)H) mM−1 cm−1 Psychotropic Other were adopted and the results are expressed throughout as nmol min−1 mg prot−1. Normal For ease of comparison the activities of cytochrome 015 F 30 N – b5 reductase are reported as though the electron donor 016 M 36 N – was NADH. This is done by multiplying the activities 017 F 44 N – measured with NADPH as donor by the ratio of activi- 018 M 30 N – ties of the two donors reported in Table 3 019 M 43 N – = 020 M 26 N – (671/270 2.48). 021 F 40 N – 022 M 30 N – Superoxide This was measured by its ability to Schizophrenic reduce nitro-tetrazolium blue to an insoluble blue- 001 M 42 Y – brown formazan using a modification of the method of 002 M 33 Y – 15 003 M 50 Y – Beauchamp and Fridovich. The formazan was filtered 004 M 40 Y – off in the wells of a ‘dot-blot’ manifold onto glass-fibre 005 M 47 Y – filter paper and after drying was quantitated with a 008 M 61 Y – Logitech hand-held scanner. The darkness of the 009 M 41 Y – deposits was estimated using Quantiscan software. 010 M 34 Y – Superoxide decomposes rapidly in aqueous solution 011 M 30 Y – but is slightly soluble and relatively stable in dimethyl- 012 M 46 Y – sulphoxide (DMSO). Solid potassium superoxide 014 M 38 Y – (approx 100 mg) was introduced into 1 ml BDH Analar grade DMSO triturated and then spun down in a Medication: all patients had received doses of flupenthixol or microfuge to clear particulate matter from the super- its congeners in the therapeutic range and in addition some had received risperidone (5–10 mg), chlorpromazine (100– natant. Twenty microlitres of the clear supernatant 2000 mg) or sulpiride (200–2000 mg). reproducibly reduced 2.7 nmoles of cytochrome c and gave a blue-brown precipitate from nitro-tetrazolium blue having an integrated grey value (IGV) of 660. A standard curve is shown in Figure 1. graphs were hard-developed using Kodak D19 devel- The detailed procedure for measurement of superox- oper, fixed and then dried. The level of non-specific ide production in biological samples was as follows; hybridisation was assessed by pretreatment with − enzyme or biological material was introduced into RNAses (60 mg ml 1)at37°C overnight, prior to hybrid- 200 ␮l of Tris buffer pH 7.4 containing 1 mM EDTA, isation and competition with 200-fold excess of unlab- − 1mMCN and 100 ␮M NAD(P)H. Two microlitres of a elled oligonucleotide probe. The relative integrated stock 10 mM solution of nitro-tetrazolium in dimethyl- grey values (IGV) were determined by comparison in formamide was added and a 20-␮l sample was taken the linear range with a 14C autoradiograph standard for the initial reading followed by others at 1-min inter- supplied by Amersham UK. Background IGV values vals. These were pipetted into the wells of the filter were taken immediately adjacent to the tissue section. manifold and immediately washed with 200 ␮l phos- phate buffered saline. After drying the filters were Ubiquinone and cytochrome b reductase 5 stable for about 12 h, the integrated grey value (IGV) of assay These were usually carried out at 25°Cin the spot was determined by scanning and the whole 20 mM KPi buffer pH 7.4 containing 1 mM EDTA using procedure was found to be sufficiently reproducible to NAD(P)H at a final concentration of 100 ␮M. The − apply the factor 0.24 IGV = 1 pmole O2 . change in absorbance with time at 420 nm (ferricyanide), 600 nm (dichlorindophenol DCIP) and 550 nm (cytochrome c) were monitored. The rate of Lipid peroxides Lipid peroxides were measured by decrease of NAD(P)H absorbance changes at 340 nm the thiobarbituric acid procedure of Aust.16 Thiobarbit- was also used to verify the activity of the ferricyanide, uric acid (0.375%) and 15% trichloracetic acid were DCIP and cytochrome c reductases. The concentrations dissolved in 0.25 M HCl. One hundred microlitres of of acceptors were ferricyanide 1.0 mM, DCIP 100 ␮M this solution and 10 ␮l of a 10% homogenate of frontal ␮ and cytochrome c 80 M. The b5 reductase does not cortex were heated in a microfuge tube for 10 min in a directly transfer electrons to cytochrome c so cyto- boiling water bath. Four hundred microlitres of water

chrome b5 was added to a final concentration of were added and this solution was centrifuged and the 0.6 ␮gml−1 to act as a catalyst when cytochrome c was optical density of the supernatant measured at 532 nm. used as an acceptor. NADH-ubiquinone reductase The procedure was standardised with malondial- activity was measured by following the rate of decrease dehyde and found to be linear in the range 0.1– in absorbance change at 340 nm following the addition 10 nmoles with a blank value about 50 pmol. Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 231 Table 3 Properties of the reductases

Donor Acceptor Activity (nmoles min−1mg prot−1)

Inhibitor/activator Ubiquinone reductase Cytochrome b5 reductase

NADH cyt c – – 350 + ± NADH cyt c cyt b5 171 671 18 + + NADH cyt c b5 rotenone 80% ndi + ± ± NADPH cyt c b5 – 115 30 270 34 ± ± NADH FeCl3 – 364 87 4198 288 + ± ± NADH FeCl3 rotenone 60 13% 92 8% ± ± NADPH FeCl3 – 309 92 763 110 + NADPH FeCl3 rotenone 117% 77% NADH DCIP – 1282 ± 241 2401 ± 414 NADH DCIP + rotenone 66 ± 6% 80 ± 5% NADPH DCIP – 155 ± 51 896 ± 328 NADPH DCIP + rotenone 74 ± 11% ndi ± ± NADH CoQ1 – 337 45 68 23 + ± NADH CoQ1 rotenone 65 8% ndi + ± NADH FeCl3 SOD – 75 6% + ± NADPH FeCl3 SOD – 77 10% NADH DCIP + SOD – 74 ± 2% NADPH DCIP + SOD – 50 ± 8% + NADH CoQ1 SOD 71% – + + NADPH cyt c b5 SOD – 25% + NADH CoQ1 chlorpromazine 83% – + + NADPH cyt c b5 chlorpromazine – 89% + NADH CoQ1 cis-flupenthixol 96% – NADH DCIP + cis-flupenthixol – 82 ± 5%

Statistics given are SEM with n = between 3 and 5. The action of inhibitors is expressed as a percentage of matched controls. ndi = no detectable inhibition. Concentrations of inhibitors were: rotenone 20 ␮M, SOD (superoxide dismutase) 100 units ml−1, chlorpromazine 100 ␮M and cis-flupenthixol 20 ␮M.

Protein Protein was measured by the method of Brad- and for the b5 reductase was NADPH-DCIP and ford.17 NADPH-cyt c; these donor-acceptor couples were used in subsequent studies of the distribution of the two reductases. Results Rotenone showed a consistent inhibition of the

Properties of the ubiquinone and cytochrome b5 ubiquinone reductase at about 60–70% of control and reductases a somewhat variable effect on the b5 reductase showing In order to select optimum conditions for measurement hardly any inhibition with cyt c, FeCl3 or CoQ1 but a of the two reductases in samples containing mixtures significant inhibition with DCIP as acceptor. of both which had also been frozen and thawed, the Chlorpromazine as has been reported previously two enzymes were semi-purified from a sample of fresh showed a weak inhibitory effect on both reduc- sheep’s liver and then frozen and thawed to mimic the tases.19,20 Cis-flupenthixol which is a more powerful treatment of the brain fractions. The ubiquinone neuroleptic also had a pronounced inhibitory effect on

reductase being essentially a mitochondrial fraction the b5 reductase. will contain that component of the b5 reductase known A number of other substances variously reported as 18 to be bound to mitochondria but the b5 reductase reductase inhibitors were investigated to see if any should not be appreciably contaminated by the ubiqui- specificity they had could be usefully employed to dis- none reductase. Table 3 gives the results of a compari- tinguish between the two enzymes. The following

son of the two preparations with various donors, inhibited the b5 reductase at the concentrations men- acceptors and inhibitors. As has been reported by Yang tioned but did not show any specificity: rhein (9,10-

and Cederbaum, cytochrome b5 is required for the opti- dihydro-9,10-dioxo-2-anthracene carboxylic acid; ␮ mum transfer of electrons to cytochrome c by the b5 20 M, 62% of control); spiperone (8-[4-(4- reductase, similarly the activity with NADH as electron fluorophenyl)-4-oxobutyl]-1-phenyl-1,3,8-triazaspiro- donor is greater than with NADPH and the activity [4,5]decan-4-one hydrochloride, 10 ␮M, 76% of ␮ with FeCl3 as acceptor is greater than with DCIP or control); aurine tricarboxylic acid (20 M, 66% of cytochrome c. One molecule of NADH reduced only control). The following at the concentrations shown 3+ one Fe ion whereas two molecules of DCIP were did not inhibit the b5 reductase: proadifen (N,N diethy- reduced. The most discriminatory donor-acceptor com- laminoethyl-2,2-diphenylvalerate hydroxide 100 ␮M); ␮ bination for the ubiquinone reductase was NADH-CoQ1 metyrapone (2-methyl,2,dipyridyl 100 M); phenol- Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 232 phthalein 100 ␮M; chrysophanic acid (1,8-dihydroxy- 3-methyl-9,10-anthracenedione 100 ␮M); dicoumarol 100 ␮M; propylthiouracil 100 ␮M. In connection with a study of superoxide production by these enzymes the effect of superoxide dismutase on

the utilisation of NAD(P)H and the reduction of FeCl3 reduction of DCIP was investigated. As Table 3 shows the dismutase caused a significant reduction in the

amount of product formed by the b5 reductase.

Superoxide production The formazan method for detecting superoxide pro- duction is more sensitive and therefore requires less tissue, also since the rate constant of the reaction of − O2 is faster with tetrazolium than cytochrome c the fraction of superoxide captured before its spontaneous dissipation is greater, therefore the method is in prin- ciple more accurate. In Figure 1 it can be seen that while superoxide dismutase (10 units ml−1) readily pre- vents the reduction of cytochrome c21 it hardly affects the production of formazan. The explanation for this is that whereas the capture of the superoxide ion by the dismutase is faster than its reaction with cytochrome c is not so fast as the reaction with nitro-tetrazolium in a homogeneous system. In experiments in which superoxide was formed slowly by an enzymic reaction (see Table 3), formazan production was variously attenuated by superoxide dismutase. It is recognised that this attempt at quantitation assumes that the capture of superoxide by nitro-tetra- zolium is sufficiently fast that competing processes do not cause appreciable losses and that NAD(P)H does not produce tetrazolium reducing materials other than by donating electrons to . The latter assumption has been checked (see Table 4) by carrying out the reac- tion anaerobically; it can be seen that oxygen is neces- sary for the reduction of nitro-tetrazolium by NADH

and the b5 reductase. Also shown in Figure 1(c) is a timing curve for the production of formazan by reductase, it can be seen that the reaction rate flattens off appreciably after about 5 min. The rates quoted in Table 4 are taken over the first minute of the reaction and therefore approximate to the initial rate in the sense used in enzymic kinetics.

The substrates of the b5 reductase (cytochrome, FeCl3 and DCIP) all reduce the amount of formazan product. Since these substances are not inhibitors of the enzy- mic activity per se (Table 3) they must act by either being an alternative to oxygen as a sink for electrons − from NADH or by being directly reduced by O2. Super- oxide directly reduces cytochrome c and FeCl3 but it Figure 1 Quantitation of superoxide production. (a) does not reduce DCIP in the absence of enzyme Relationship between superoxide concentration and cyto- (Table 4), so one must conclude that in this case super- bJJb ᭺JJ᭺ chrome c reduction. ( ) Control; ( ) in the pres- oxide is an intermediate in the electron transfer pro- ence of superoxide dismutase 100 units per ml. (b) Relation- cesses mediated by the NAD(P)H oxido-reductases. ship between superoxide concentration and formazan production from nitro-tetrazolium blue. (bJJb) Control; This would also be consistent with the inhibitory effect (᭺JJ᭺) in the presence of superoxide dismutase 100 units of superoxide dismutase on the reductases noted in per ml. (c) Time course of superoxide production by NADH- Table 3. ubiquinone reductase. (bJJb) Enzyme; (᭺JJ᭺) water blank. Neuroleptic effects on the expression of NADH-cytoch-

rome b5 reductase These were investigated by submit- Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 233 Table 4 Superoxide production

Preparation Inhibitors Amount (nmoles min−1) cytochrome c formazan

Water blank – 0.025 0.116 ␮ H2O2 100 M – 1.88 0.112 ␮ −1 H2O2 100 M HRP (4 units ml ) 2.65 0.033

NADH 100 ␮M– Ͻ0.1 0.24

− a O2 20 nmoles – 8.25 7.6 − a −1 O2 20 nmoles SOD (100 units ml ) 0.1 1.21 − a ± O2 20 nmoles FeCl3 –147% − a ± O2 20 nmoles DCIP – 86 8% 100 ␮M NADH as electron donor and formazan as detector: Rate (nmol min−1 mg prot−1)

Cytochrome b5 reductase 297 ± Cytochrome b5 reductase anaerobic 36 9% ± Cytochrome b5 reductase SOD 83 11% + ± Cytochrome b5 reductase SOD HRP 66 19% ␮ −1 ± Cytochrome b5 reductase cytochrome b5 (0.6 gml )7812% ± Cytochrome b5 reductase FeCl3 (0.4 mM) 65 1% ± Cytochrome b5 reductase DCIP (0.1 mM) 56 10% ± Cytochrome b5 reductase chlorpromazine 70 10% ± Cytochrome b5 reductase cis-flupenthixol 71 5% ± Cytochrome b5 reductase NADPH as electron donor 61 13% Ubiquinone reductase 201 Ubiquinone reductase anaerobic 47 ± 20% Ubiquinone reductase SOD 88 ± 9% Ubiquinone reductase SOD + HRP 87 ± 7% Ubiquinone reductase chlorpromazine 96 ± 12% Ubiquinone reductase cis-flupenthixol 88 ± 11%

Values given are in units shown or as % matched control ± SEM (n = 2–4). a − − ␮ Approximate amount only of O2 because in these preliminary experiments excess solid KO2 was added to water and 50 l used after 1 min. Amount is calculated from rate constants of dismutation given by McCord and Fridovich.21

ting laboratory rats to a protocol of flupenthixol admin- administered cause decreases in both the ubiquinone

istration designed to mimic the regimen administered and cytochrome b5 reductase in rat frontal cortex to psychotic patients (1 mg kg−1 ip daily for 3 weeks). although with only the latter enzyme was the effect The animals were then killed within 24 h of the last statistically significant. The flupenthixols also caused injection and fresh frozen slices of brain cortex taken a significant decrease in the capacity of the tissue to for in situ hybridisation with a cDNA probe that ident- produce superoxide anion. These decreases are

ified either the b5 reductase or COX mRNAs. The unlikely to be caused by residual flupenthixol in the remaining tissue was retained for measurements of brain homogenates (estimated maximum conc. 1 ␮M)

superoxide production activity and ubiquinone and b5 because its effects in vitro (see Tables 3 and 4) would reductase activity. The results of in situ hybridisation not be sufficient to account for the effects shown in with several probes in the frontal cortex are given in Table 6. Table 5. There were highly significant reductions in The subcellular distribution of these activities is also

expression of the mRNA coding for cytochrome b5 shown in Table 6. As would be expected the ubiqui- reductase in both cis and trans-FPT treated animals in none reductase is principally in the mitochondrial frac-

the frontal cortex. The specificity of this effect is shown tion whereas the b5 reductase is found about equally by the data for the expression of the nuclear coded gene in both mitochrondrial and supernatant fractions. for cytochrome oxidase subunit IV and for the mito- Superoxide production is found mainly in the mito- chondrially coded gene for cytochrome oxidase sub- chondria. unit II. The expression of neither was much affected by cis-ortrans-flupenthixol. Reductase activity in brain samples from normal and The enzymic activity of the reductases was measured schizophrenic patients In Table 7 are shown the in the same samples with results shown in Table 6. The results of measurements of the reductase enzymes in changes paralleled those found in the mRNA distri- homogenates of frontal cortex tissue obtained post- bution. Both cis- and trans-flupenthixol chronically mortem from normals and schizophrenics, some of Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 234 Table 5 Effect of flupenthixol on cytochrome b5 reductase mRNA concentrations measured by in situ hybridisation in fronto- parietal cortex of rat brain

Region cis-FPT Trans-FPT Control

L Motor area 31.8*** ± 4.1 25.9*** ± 4.6 57.5 ± 3.6 R Motor area 31.2*** ± 3.1 25.4*** ± 3.2 56.7 ± 2.3 L Somatosensory 25.9*** ± 3.2 17.1*** ± 3.4 54.5 ± 4.9 R Somatosensory 27.2*** ± 3.3 20.9*** ± 2.4 54.3 ± 4.1 Probed with COX IV mRNA: L Motor 42.3 ± 2.6 48.1 ± 0.73 46.3 ± 0.74 Probed with Cox II mRNA (heavy chain): L Motor 26.5 ± 9.7 22.9 ± 9.4 27.0 ± 7.4 Probed with COX II mRNA (light chain): L Motor 28.7 ± 4.8 21.9 ± 5.8 38.8 ± 5.6

Results given are the mean integrated grey value (IGV) ± SEM (n = 5) after background subtraction. ***P Ͻ 0.001 different from control.

Oligonucleotide probes used were for NADH cytochrome b5 reductase: GTA AAT GTG GGA GGA GGA GAA GGA TGC TGA GAG AT; for COX II heavy chain: CAA AGC ATA GGT CTT CAT AGT CAG TAT ATT CAT; for COX II light chain: CGC CCT CCC TTC CCT ACG AAT TTT ATA CAT AAT AGC; and for COX IV: GGT TTC CAG TAA ATA GGC ATG GAC.

Table 6 Effect of fluphenthixol on the activity of reductases and superoxide in rat frontal cortex

Condition Ubiquinone Cytochrome b5 Superoxide reductase reductase production (nmol min−1 mg prot−1) (pmol min−1 mg prot−1)

Saline control 9.2 ± 2.2 10.7 ± 2.6 28 ± 4.7 cis-Flupenthixol 4.6 ± 0.7 4.6 ± 0.9* 13 ± 0.74* trans-Flupenthixol 4.7 ± 1.1 4.2 ± 0.8* 10 ± 0.45** Percentage distribution between mitochondrial and supernatant fractions Mitochondrial 94 ± 348± 385± 3 Supernatant 6 ± 352± 314± 7

Enzyme activities and superoxide production were measured in a 10% homogenate of brain tissue. Statistic given is SEM (n = 4), *P Ͻ 0.05, **P Ͻ 0.01. Dosage of flupenthixol was 1 mg kg−1 given daily for 3 weeks; estimated brain concentration approximately 10 ␮M. 20 ␮M concentration is sufficient to inhibit purified enzyme about 10–20% (see Table 2). Washed mito- chondrial and supernatant fractions were made as described in the Methods section except that the post-mitochondrial super- natant was taken as ‘supernatant’ without precipitation with PEG. Ubiquinone reductase was measured as the rate of reduction in optical density at 340 nm of an 0.1 mM NADH solution in a KPi buffer solution pH 8.3 at room temperature with 0.1 mM

CoQ1 as acceptor. Cytochrome b5 reductase was measured as the rate of reduction in optical density at 340 nm of an 0.1 mM ␮ ␮ −1 NADPH solution in KPi buffer solution pH 8.3 at room temperature using cytochrome b5 (concn 0.6 g l ) as the acceptor in the presence of 8.3 ␮M cytochrome c.

whom had not received medication. Also shown as of Table 7), the changes approximate to those found in data in parentheses are the results when the short delay our previous study11 in that the activity of NADH

post-mortem group of schizophrenics is calculated sep- cytochrome b5 reductase appeared to be higher in arately. The top half of the table shows the activities samples from schizophrenic brain compared with nor- measured using donor-acceptor couples specific for the mals. The inhibitory effect of rotenone is thought to two enzymes. There was a significant decrease in the reside wihin the electron transport chain at Complex I

cytochrome b5 reductase in the tissue from medicated whose enzymic manifestation is the NADH-ubiquinone schizophrenics. Unmedicated schizophrenics showed reductase. Reductase activity that was not inhibited by

an intermediate value so we are inclined to attribute rotenone was ascribed to the NAD(P)H-cytochrome b5 the decrease to the effects of neuroleptic medication reductase. However as was noted at the time, the sensi- which was usually a flupenthixol congener. tivity of the ubiquinone reductase to inhibition by When the distinction between the enzymes was rotenone might itself be changed in the schizophrenic made on the basis of rotenone sensitivity (bottom half condition. The results of the bottom and top half of Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 235 Table 7 Ubiquinone and cytochrome b5 reductase activity in thiazine medication have elevated concentrations of frontal cortex from schizophrenics and normals lipid peroxides in the cerebrospinal fluid.22

Ubiquinone Cytochrome b5 Reductase activity and superoxide production in reductase reductase lymphocytes − − (nmol min 1 mg prot 1) In order to see whether the pattern of these changes was specific to brain or could be found elsewhere in Donor: NADH NADPH the body, lymphocyte samples were obtained from a + Acceptor: CoQ1 cytochrome c (b5) set of male schizophrenic patients and compared with ± ± lymphocytes from normal males and females. Normal 3.62 0.26 1.76 0.20 Reductase activity and superoxide production in lym- Schizophrenic 4.46 ± 0.76 (3.94) 1.05 ± 0.14** (1.00) Unmedicated 3.38 ± 0.59 1.35 ± 0.32 phocytes from these patients and normals are shown schizophrenic in Table 9. The general pattern in lymphocytes is simi- Donor: NADH NADH lar to brain; those from schizophrenic patients show a Acceptor: cytochrome c cytochrome c much reduced ubiquinone and b5 reductase activity. Rotenone-sensitive Rotenone-insensitive Superoxide production is also much reduced in the Normal 3.59 ± 0.76 1.75 ± 0.80 schizophrenic patients all of whom were being treated Schizophrenic 2.51 ± 0.79 (2.40) 3.00 ± 0.48 (2.59) with neuroleptics. Unmedicated 2.57 ± 0.83 2.16 ± 0.96 schizophrenic Discussion Statistic given is SEM with n = 10 for normals and schizo- = In this study we have shown that neuroleptics as exem- phrenics and n 5 for unmedicated schizophrenics. plified by flupenthixol repress the expression of the **P Ͻ 0.01 different from normals. Values in parentheses apply to the short post-mortem delay group. nuclear gene that codes for the enzyme NADH-cyto- chrome b5 reductase. The related enzyme NADH- ubiquinone reductase which contains both mitochon- Table 7 taken together suggest that in schizophrenics drially and nuclear coded subunits is known to be the ubiquinone reductase is indeed less sensitive to decreased by neuroleptics,23,24 and is also decreased in 25 inhibition by rotenone. The short post-mortem delay Parkinson’s disease. The decrease in cytochrome b5 group of schizophrenics did not show values appreci- reductase has been shown at the mRNA level in rats ably different from the whole cohort. by in situ hybridisation (Table 5) and at the enzymic level also in rat brain (Table 6). In humans we find that Superoxide formation and lipoperoxidation post-mortem brain tissue (Table 7) and lymphocytes Superoxide formation and lipid peroxidation in homo- (Table 9) from schizophrenics have reduced levels and genates of post-mortem frontal cortex from normals since these patients have received neuroleptic medi- and schizophrenics is shown in Table 8. Superoxide cation (often fluphenthixol itself), we believe the production is slightly but significantly lower in schizo- reduction is an effect of the drug rather than the illness. phrenics than normals whereas lipid peroxide concen- The data we present from the small group of unmedi- trations are higher. The reduced capacity to produce cated patients available to us approximates more to the superoxide would be expected to be the result of neur- normals than the schizophrenics supporting this con- oleptic-induced decrease in the amounts of the clusion. There are differences in the post-mortem reductase enzymes. The amounts of lipid peroxide will delays experienced by our normal and schizophrenic result from the immediate pre-history of oxidative samples but no significant differences in enzyme stress which is not easily assessable in our sample, activity or superoxide production were found when the though it has been reported that patients taking pheno- schizophrenic group was subdivided into those experi- encing long and short post-mortem delays. Therefore we believe that the human results are not a post-mor- Table 8 Superoxide formation and lipoperoxidation in fron- tem artefact and that as in animals flupenthixol reduces tal cortex from normal and schizophrenic patients the expression of the cytochrome b5 and ubiquinone reductase. Superoxide production Lipid peroxidation −1 −1 −1 The major problem encountered in this study was (pmol min mg prot ) (pmol mg prot ) discriminating between the two reductases when measurements have to be made (as here) on crude Normal 2.05 ± 0.12 2.55 ± 0.17 tissue extracts. Using semi-purified enzymes, an Schizophrenic 1.73 ± 0.07* (1.94) 3.30 ± 0.26* Unmedicated 2.26 ± 0.21 2.61 ± 0.64 attempt was made to find more specific or effective schizophrenic inhibitors than rotenone, or combinations of electron donors and acceptors that distinguish the reductases. Statistic given is the mean and SEM of 10 normals, 10 schizo- The inhibitory activity of rotenone is sensitive to vari- phrenics and five unmedicated schizophrenics measured in ous factors including freezing and thawing and other duplicate. *P Ͻ 0.05. Values in parentheses relate to the short treatments expected to derange the lipid environment post-mortem delay sub group. of the enzyme.26 No very satisfactory differential Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 236 Table 9 Reductase activity and superoxide production in lymphocytes from normal and schizophrenic patients

Ubiquinone Cytochrome b5 Superoxide reductase reductase production (nmol min−1mg prot−1) (nmol min−1 mg prot−1)

Donor: NADH NADPH + Acceptor: CoQl cytochrome c (b5) Normals (M & F)a 65.1 ± 11.3 185 ± 41 5.73 ± 0.49 Schizophrenic (M) 16.7 ± 4.4** 102 ± 13* 2.53 ± 0.40***

Statistic given is mean ± SEM where n = 8 for the normals and n = 12 for the schizophrenics. aNo significant differences were found between normal males and females so pooled figures are given. *P Ͻ 0.05; ** P Ͻ 0.01; ***P Ͻ 0.001.

inhibitors were found and the best strategy that involved in drug metabolism. It is deficient in heredi- emerged was to identify the ubiquinone reductase with tary methaemoglobinurea32 causing mental and neuro-

the NADH-CoQ1 couple and the b5 reductase with the logical impairment. NADPH-cytochrome c couple with additions of trace The ubiquinone and b5 reductase are both flavopro- amounts of cytochrome b5 as catalyst. teins which as a class are well known under certain The criteria of sensitivity to rotenone inhibition to conditions to transfer univalent electrons to oxygen to distinguish ubiquinone reductase or Complex I activity yield superoxide (reviewed in33). It is evident that the is widely used but does not correlate well with distinc- oxygen accepts electrons from the reduced flavin group tions based on donor and acceptor specificity in brain directly to form superoxide. Oxygen can be viewed as (Table 7). The discrepancy between the two methods but one of many alternative electron acceptors. Super- of measurement invites the overall interpretation that oxide being extremely reactive has the capability of

there is a real decrease in the cytochrome b5 reductase transferring electrons to other acceptors or undergoing activity in schizophrenics due to their neuroleptic self dismutation. This explains the inhibition of both medication as would be expected from the animal reductases by superoxide dismutase when their studies and that the apparent decrease in ubiquinone activity is measured with the usual electron acceptors

reductase activity is due to a decrease in rotenone sen- ie FeCl3 or DCIP (Table 3) and the inhibitory effect of − sitivity either arising from the illness itself or a differ- these substances when O2 production is measured ential response to the freezing and thawing that neces- (Table 4). The inhibitory effect of the neuroleptic chlor-

sarily arises from the use of post-mortem tissue. promazine on the b5 reductase was noted in an earlier The mechanism by which flupenthixol causes its study and is extended here by our observations on flu- effects on the expression of reductases is not known penthixol. Our results suggest that these substances but we have no evidence to exclude its anti-dopamine can accept electrons in competition with oxygen and receptor activity as being a stage in the causal sequence therefore reduce the production of superoxide and probably mediated via cyclic AMP effects on glucose inhibit reductase activity measured with other electron metabolism at the 6-phosphofructo-1-kinase control acceptors. There is also the possibility that neurolep- point.27,28 In the animal studies we have not observed tics react directly with superoxide. differences between cis- and trans-flupenthixol in The decrease in superoxide production caused by experiments on mRNA, reductase activity or superox- neuroleptics in animal and human brain and lympho- ide production (Tables 5 and 6). It is claimed that cyte tissue in vitro is very marked (Tables 6, 8 and 9) therapeutic efficiency of flupenthixol in humans is and can reach 50%. From Table 3 and Table 7 it can restricted to the cis isomer though the concentration be inferred that the overall ability to produce superox- differential is very variable (compare the studies of ide is approximately the same for both the ubiquinone 1 29 Johnstone et al, Miller et al with Jenner and and the cytochrome b5 reductase. Since both enzymes Marsden30). Because we have not observed stereospec- are downregulated by neuroleptics it seems likely that ificity in rats it might be argued that the effects we have they are the causal link connecting the decrease in observed bear no relation to the therapeutic modus superoxide production by the drugs. Their effect on operandi of neuroleptics in humans. However in view superoxide production is quantitatively not negligible of the uncertainties in concentration dependence and and considering the toxicity of superoxide12 we suggest species sensitivity we believe an agnostic stance to be that this result reveals an important facet of neuroleptic more appropriate. medication. It has been reported34 that neuroleptics

The NADH-cytochrome b5 reductase is present on increase the amounts of manganese-superoxide dismu- the outer membrane of mitochondria and also on the tase in the brain of psychotics. It is possible that this endoplasmic reticulum. In the former situation it is effect contributes to the diminution of superoxide that thought to mediate an alternative pathway of electrons we have observed on flupenthixol treatment although

from extra-mitochondrial NADH via cytochrome b5 to it would not account for the changes in reductase cytochrome c31 while in the latter position it is activity. Superoxide, neuroleptics and reductases in schizophrenia SA Whatley et al 237 In conclusion we have observed that the neuroleptic 12 Haliwell B. Reactive oxygen species and the central nervous sys- flupenthixol causes a downregulation of the NADH- tem. J Neurochem 1992; 59: 1609–1623. 13 Boyum A. Separation of leucocytes from blood and bone marrow. cytochrome b5 reductase as well as the mitochondrial Scand J Clin Invest 1968; 21: Suppl 97. ubiquinone reductase and as a consequence a 14 Yang M-X, Cederbaum AL. Fractionation of liver microsomes with reduction in the brain production of neurotoxic super- polyethylene glycol and purification of NADH-cytochrome b5 oxi- oxide. In our earlier studies we found that the changes doreductase and cytochrome b5. Arch Biochem Biophys 1994; 315: 438–444. in gene expression wrought by flupenthixol are in the 15 Beauchamp C, Fridovich I. Superoxide dismutase: improved assays opposite direction to those found in schizophrenia. and an assay applicable to polyacrylamide gels. Anal Biochem This suggests that the action of neuroleptics in revers- 1971; 44: 276–281. ing oxidative stress might also play a part in reversing 16 Ausst SD. Lipid peroxidation. In: Greenwald RA (ed). CRC Hand- the pathogenesis of schizophrenia. Studies currently book of Methods for Oxygen Radical Research. CRC Press, 1985, pp 203–207. underway with a larger group of unmedicated schizo- 17 Bradford MM. A rapid and sensitive method for the quantitation phrenics might illuminate this possibility. of microgram quantities of protein utilizing the principle of protein dye binding. Analyt Biochem 1977; 78: 248–254. 18 Borgese N, Pietrini G. Distribution of the integral membrane protein

Acknowledgements NADH-cytochrome b5 reductase in rat liver cells studied with a quantitative radio immunoblotting assay. Biochem J 1986; 239: We thank Professor IN Ferrier of the MRC Newcastle 393–403. and Dr Nigel Cairns of the Institute of Psychiatry Brain 19 Dawkins MJR, Judah JD, Rees KR. Action of chlorpromazine. Bank for their help in the collection of post-mortem Biochem J 1960; 76: 200–205. brain samples. DC and RMM thank the European 20 Grebing C, Crane FL, Low H, Hall K. Transmembranous NADH- Science Foundation for a travel grant that made the in human erythrocyte membranes. J Bioenergetics and Biomembranes 1984; 16: 517–533. biochemical studies possible. CoQ1 was the generous 21 McCord JM, Fridovich I. Superoxide dismutase. J Biol Chem 1969; gift of the Eisai Pharmaceutical Co, Tokyo, Japan. 244: 6049–6055. 22 Pall HS, Williams AC, Blake DR, Lucas JR. Evidence of enhanced lipid peroxidation in the cerebrospinal fluid of patients taking References phenothiazines. Lancet 1987; II: 596–599. 1 Johnston EC, Crow TJ, Frith CD, Carney MWP, Price JS. Mechanism 23 Burkhardt C, Kelly JP, Lin YH, Filley CM, Parker WD. Neuroleptic of the antipsychotic in the treatment of acute schizophrenia. Lancet medications inhibit Complex I of the electron transport chain. Ann 1978; 1: 848–851. Neurol 1993; 33: 512–517. 2 Coon H, Byerley W, Holik J, Hoff M, Myles-Worsley M, Lannfelt L 24 Prince JA, Yassin MS, Oreland L. Neuroleptic induced mitochon- et al. Linkage analysis of schizophrenia with five dopamine recep- drial enzyme alterations in the rat brain. J Pharmacol Exp Therap tor genes in nine pedigrees. Am J Hum Genet 1993; 52: 327–334. 1997; 280: 261–267. 3 Mulcrone J, Whatley SA, Marchbanks RM, Wildenauer D, Altmark 25 Schapira AHV, Cooper JM, Dexter D, Clark JB, Jenner P, Marsden D, Daoud H et al. Genetic linkage analysis of schizophrenia using CD. Mitochondrial complex I deficiency in Parkinson’s disease. J chromosome 11q13–24 markers in Israeli pedigrees. Am J Med Neurochem 1990; 54: 823–826. Genet 1995; 60: 103–108. 26 Finel M, Skehel JM, Albracht SPJ, Fearnley IM, Walker JE. Subcom- 4 Moises HW, Yang L, Ktistbjarnarson H, Wiese C, Byerley W, Macci- plexes of NADH: ubiquinone . Biochemistry 1992; ardi F et al. An international two-stage genome wide search for 31: 11425–11434. schizophrenia susceptibility genes. Nature Genet 1995; 11: 321– 27 Lazareno S, Mainot DB, Nahorski S. Differential effects of selec- 324. tive–non selective neuroleptics on intra and extra cellular cyclic 5 Perrett CW, Marchbanks RM, Whatley SA. Characterisation of AMP accumulation in rat striatal slices. Brain Res 1985; 361: 91– messenger RNA extracted post-mortem from the brains of schizo- 98. phrenic, depressed and control subjects. J Neurol Neurosurg Psych- 28 Marchbanks RM, Mulcrone J, Whatley SA. Editorial: Aspects of oxi- iatr 1988; 51: 325–331. dative metabolism in schizophrenia. Br J Psychiatry 1995; 167: 6 Perrett CW, Marchbanks RM, Ferrier IN, Whatley SA. An analysis 293–298. of post-mortem brain gene expression in schizophrenic and control 29 Miller RJ, Horn AS, Iversen LL. The action of neuroleptic drugs on subjects. Schizophrenia Res 1990; 3: 15–16. dopamine-stimulated adenosine cyclic 3′,5′-monophosphate pro- 7 Perrett CW, Whatley SA, Ferrier IN, Marchbanks RM. Changes in duction in rat neostriatum and limbic forebrain. Mol Pharmacol relative levels of specific brain mRNA species associated with 1974; 10: 759–766. schizophrenia and depression. Mol Brain Res 1992; 12: 163–171. 30 Jenner P, Marsden CD. Substituted benzamide drugs as selective 8 Mulcrone J, Whatley SA. Editorial: The application of differential neuroleptic agents. Neuropharmacology 1990; 20: 1285–1281. cloning techniques in post-mortem studies of neuropsychiatric dis- 31 Marzutti D, Cafagno L, La-Piana G, Cipriani T. NADH-b5 reductase orders: a way ahead. Psychol Med 1993; 23: 825–829. as an electron shunt for exogenous NADH. Arch Biochem Biophys 9 Mulcrone J, Whatley SA, Ferrier IN, Marchbanks RM. A study of 1991; 288: 293–301. altered gene expression in frontal cortex from schizophrenic 32 Shirabe K, Fujimoto Y, Yubisui T, Takeshita M. An in-frame patients using differential screening. Schizophr Res 1995; 14: deletion of codon 298 of the NADH cytochrome b5 reductase gene 203–213. results in hereditary methemoglobinaemia type II (generalised). J Biol Chem 1994; 269: 5952–5957. 10 Perrett´ C, Owen M, Marchbanks R, Whatley S. The effect of chronic O- and ␤- flupenthixol treatment upon mRNA populations in the 33 Dixon M, Webb EC. The Enzymes (3rd edition). Longman: London, rat brain. Mol Neuropharmacol 1992; 2: 289–296. 1979, pp 281–282. 11 Whatley SA, Curti D, Marchbanks RM. Mitochondrial involvement 34 Loven DP, James JF, Biggs L, Little KY. Increased manganese-super- in schizophrenia and other functional psychoses. Neurochem Res oxide dismutase activity in postmortem brain from neuroleptic- 1996; 21: 995–1004. treated psychotic patients. Biol Psychiatry 1996; 40: 230–232.