VGR

BefpeMhRipVtCe

Kt

BDaembcomch(i(e21a2a

i

F

A

R

0d

alproate Decreases Inositol Biosynthesisalit Shaltiel, Alon Shamir, Joseph Shapiro, Daobin Ding, Emma Dalton, Meir Bialer, Adrian J. Harwood,obert H. Belmaker, Miriam L. Greenberg, and Galila Agam

ackground: Lithium and valproate (VPA) are used for treating bipolar disorder. The mechanism of mood stabilization has not beenlucidated, but the role of inositol has gained substantial support. Lithium inhibition of inositol monophosphatase, an enzyme requiredor inositol recycling and de novo synthesis, suggested the hypothesis that lithium depletes brain inositol and attenuateshosphoinositide signaling. Valproate also depletes inositol in yeast, Dictyostelium, and rat neurons. This raised the possibility that theffect is the result of myo-inositol-1-phosphate (MIP) synthase inhibition.ethods: Inositol was measured by gas chromatography. Human prefrontal cortex MIP synthase activity was assayed in crude

omogenate. INO1 was assessed by Northern blotting. Growth cones morphology was evaluated in cultured rat neurons.esults: We found a 20% in vivo reduction of inositol in mouse frontal cortex after acute VPA administration. As hypothesized,

nositol reduction resulted from decreased MIP synthase activity: .21–.28 mmol/L VPA reduced the activity by 50%. Amongsychotropic drugs, the effect is specific to VPA. Accordingly, only VPA upregulates the yeast INO1 gene coding for MIP synthase. ThePA derivative N-methyl-2,2,3,3,-tetramethyl-cyclopropane carboxamide reduces MIP synthase activity and has an affect similar to

hat of VPA on rat neurons, whereas another VPA derivative, valpromide, poorly affects the activity and has no affect on neurons.onclusions: The rate-limiting step of inositol biosynthesis, catalyzed by MIP synthase, is inhibited by VPA; inositol depletion is a first

vent shown to be common to lithium and VPA.

ey Words: Valproic acid, inositol, myo-inositol-1-phosphate syn-hase, brain, dorsal root ganglia growth cones, psychotropic drugs

ipolar disorder is a severe, chronic, and disabling illnessthat affects 1% to 2% of the population (Weissman et al1988). Two drugs are approved by the U.S. Food and

rug Administration for treatment of bipolar disorder: lithiumnd valproate (VPA). Yet, 20%–40% of patients fail to respond toither drug. Some patients respond better to single-drug treat-ent, whereas others benefit from combined treatment withoth drugs. Although there is a plethora of biochemical andellular effects of lithium (Jope 1999), its therapeutic mechanismf action has not been elucidated, but the role of inositoletabolism has gained substantial support. For VPA, a branched-

hain monocarboxylic fatty acid, a variety of biological targetsave been found within its therapeutic concentration range.35–.7 mmol/L) (for review, see Loscher 1999, 2002). Thesenclude an increase in GRP78 messenger ribonucleic acidmRNA) levels (Bown et al 2002) and bcl-2 protein levels (Chent al 1999), inhibition of histone deacetylase activity (Phiel et al001), and stimulation of glutamate release (Dixon and Hokin997). Very recently, it has been reported that VPA enhances thectivity of the �-aminobutyric acid uptake system (Whitlow et al003); however, the molecular mechanism of VPA therapeuticction in bipolar illness is also still unknown.

A heuristic hypothesis for mood stabilization by lithium is thenositol depletion hypothesis proposed by Berridge (1985),

rom the Stanley Research Center and Zlotowski Center for Neuroscience(GS, AS, JS, RHB, GA), Ben-Gurion University of the Negev, and MentalHealth Center, Beersheva; Department of Pharmaceutics and David R.Bloome Center for Pharmacy (MB), School of Pharmacy, Faculty of Med-icine, The Hebrew University of Jerusalem, Jerusalem, Israel; Departmentof Biological Sciences (DD, MLG), Wayne State University, Detroit, Mich-igan; and Intracellular Signalling Group (ED, AJH), Medical ResearchCouncil Laboratory for Molecular Cell Biology, University College Lon-don, London, United Kingdom.

ddress reprint requests to Galila Agam, Ph.D., Ben-Gurion University of theNegev, Mental Health Center, Psychiatry Research Unit, P.O. Box 4600,Beersheva 84170, Israel; E-mail: galila@bgumail.bgu.ac.il.

eceived February 17, 2004; revised July 20, 2004; accepted August 28, 2004.

006-3223/04/$30.00oi:10.1016/j.biopsych.2004.08.027

based on Hallcher and Sherman’s finding of uncompetitiveinhibition of inositol monophosphatase (IMPase) by therapeuti-cally relevant lithium concentrations (Hallcher and Sherman1980). The hypothesis suggests that inhibition of IMPase bylithium results in depletion of brain inositol, an increase ininositol monophosphate, downregulation of the phosphoinosi-tide (PI) signaling cycle, and subsequent dampening of overac-tive neurotransmission. Support for this hypothesis comes fromevidence that chronic lithium administration reduces agonist-stimulated PI hydrolysis in rat brain (Casebolt and Jope 1989;Godfrey et al 1989; Kendall and Nahorski 1987) and exertsmultiple effects on the PI second messenger generating system(Jope and Williams 1994; Manji and Lenox 1994). Furthermore,the administration of inositol attenuates many of the effects ofchronic lithium administration (Busa and Gimlich 1989; Kofmanand Belmaker 1993; Kofman et al 1991; Manji et al 1995).

Intriguingly, it has recently been reported that VPA, likelithium, causes inositol depletion in yeast (Vaden et al 2001).Vaden et al (2001) found a reduction in intracellular inositol andan increase in the levels of the substrate of IMPase, inositol-1-phosphate, after treatment with lithium; however, treatment withVPA led to reduction in both inositol and inositol-1-phosphatelevels, consistent with a decrease in activity of myo-inositol-1-phosphate (MIP) synthase, which catalyzes the conversion ofglucose-6-phosphate to inositol-1-phosphate, the rate-limitingstep in inositol de novo synthesis.

Myo-inositol-1-phosphate synthase is a ubiquitous enzymefound in all eukaryotes and conserved through evolution. Theactive form of the purified enzyme in a wide range of organismsis a multimer of identical subunits ranging in molecular weightfrom 58 kd to 67 kd (Majumder et al 1997). The enzyme requiresthe presence of divalent ions and nicotineamide adenine dinu-cleotide (NAD�) for catalytic activity (Maeda and Eisenberg1980). In a comparative study, Eisenberg (1967) showed that MIPsynthase activity in rat brain is of the same order of magnitude asin liver, lung, and spleen. In bovine brain, MIP synthase wasfound to be highly concentrated in the cerebral vascular bed(Wong et al 1987). In yeast, MIP synthase is encoded by the INO1gene. There is 56% amino acid similarity between the INO1product and the human MIP synthase (Shaldubina et al 2002).

Expression of INO1 is repressed during growth in the presence of

BIOL PSYCHIATRY 2004;56:868–874© 2004 Society of Biological Psychiatry

ei

enri2

msrmsanrtl

M

A

(Ucta(dwammtotPavdhmpsatmmwIId43tcc

G

s

G. Shaltiel et al BIOL PSYCHIATRY 2004;56:868–874 869

xogenous inositol and dramatically increased in inositol-deplet-ng conditions (Vaden et al 2001).

It has recently been found that lithium, VPA, and carbamaz-pine all cause increased spreading of the growth cones ofewborn rat dorsal root ganglia (DRG) cells. These effects areeversed by addition of myo-inositol, supporting the notion of annositol depletion mechanism for these drugs (Williams et al002).

To further elucidate the role of inositol in the therapeuticechanism of VPA, it is necessary to demonstrate that MIP

ynthase is active in the adult human brain, that therapeuticallyelevant concentrations of VPA cause inositol depletion in theammalian brain, and that this occurs via inhibition of MIP

ynthase. The present study is the first showing MIP synthasectivity in crude postmortem human prefrontal cortex homoge-ate and the first to show that human brain MIP synthase iseduced in the presence of therapeutically relevant VPA concen-rations. Our results indicate that inositol depletion is common toithium and VPA, albeit through inhibition of different enzymes.

ethods and Materials

nimalsThe study was approved by the Ben-Gurion University

Beer Sheva, Israel) Institutional Review Committee for these of Animals, and the procedures were carried out inompliance with the National Institutes of Health’s Guide forhe Care and Use of Laboratory Animals (Council 1996). Forcute VPA administration, male Institute of Cancer ResearchICR) mice (20–25 g) (Harlan, Indianapolis, Indiana) wereivided into five groups (n � 13 in each group). Two groupsere injected (IP) with a single dose of saline and three withsingle injection of different doses of VPA (300 mg/kg, 600g/kg, and 800 mg/kg). Higher VPA doses are required inice to attain plasma levels similar to those in humans, owing

o the 10-fold lower half-life of VPA in mice. A valproate dosef 390 mg/kg/day IP has been shown to be effective to reachhe medium effective dose of the drug in mice (Loscher 1999).lasma VPA levels in these mice ranged between .4 mmol/Lnd 1.0 mmol/L, without a significant difference between thearious dosages. One saline-injected group was killed imme-iately after the injection. The other four groups were killed 1our after injection. The frontal cortex was removed from allice immediately after killing and frozen (�70°C). Thisrocedure has previously been established by us in multipletudies of brain inositol levels (Agam et al 1994; Bersudsky etl 1994; Shapiro et al 2000). For chronic VPA administration,wo independent experiments with two groups of male ICRice (20–25 g) (Harlan) were carried out. In experiment 1,ice of group I (n � 29) were treated with VPA in drinkingater (12.5 g/L) for 11 days until being killed; mice of group

I (n � 30, control mice) drank the same water without VPA.n experiment 2, mice of group I (n � 15) were injected twiceaily (IP) with varying doses of VPA (2 days at 400 mg/kg/day,days at 200 mg/kg/day, 2 days at 250 mg/kg/day, 4 days at

00 mg/kg/day, and 3 days at 400 mg/kg/day) and killed onhe morning of day 15, resulting in mean (�SD) plasma VPAoncentration of .50 � .08 mmol/L. Mice of group II (n � 13,ontrol mice) were injected with saline (IP) twice daily.

as Chromatographic Measurement of Brain Inositol LevelsMouse brain free inositol levels were analyzed as trimethyl-

ilyl derivatives by gas-liquid chromatography, as previously

described (Shapiro et al 2000), with a capillary column. Thecorrelation coefficients of the daily standard curves were alwaysgreater than .987. Replicate reliability was tested in rat brain.When two different specimens from four rat brains were sam-pled, the correlation coefficient was .80. When a single brainextract was divided into 10 separate samples, each assayedindividually, the coefficient of variation was 23.5%. Each sampleextract was assayed at least in duplicate. The results presentedare the average of the replicates. Assays were performed in ablind and balanced design, such that each run included samplesfrom all comparison groups.

Human Brain TissueRight hemisphere postmortem prefrontal human brain sam-

ples from subjects with no history of psychiatric disordersderived from a collection previously described (Gross-Isseroff etal 1990) were used to measure MIP synthase activity. The use ofthe human brain collection was approved by the Israeli Ministryof Health Helsinki Committee.

Human Brain MIP Synthase ActivityMyo-inositol-1-phosphate synthase activity was measured by

a modification of the procedures of Wong et al (1987) and Novaket al (1999). Supernatant fraction (5 �L) obtained after sonication(Ultrasonic Processor, Vibracell, Newtown, Connecticut; 15 sec at.1 output watts at 4°C) and centrifugation for 20 min at 9000 g at4°C of postmortem human prefrontal cortex (1 mg wet weight in.5 mL of 50 mmol/L Tris HCl, pH 7.4) was added to 20 �L ofreaction mixture containing 4 mmol/L D-glucose-6-phosphate,1.25 �Ci D-glucose-6-phosphate [14C], 1.6 mmol/L NAD�, .45mmol/L KCl, 5.4 mmol/L MgCl2, and .9 mmol/L Tris HCl, pH 7.6,with or without 5 mU IMPase (Sigma, St. Louis, Missouri).Incubation was carried out for 3 hours at 37°C (this wasdetermined experimentally to be within the linear range ofactivity). The reaction was stopped by the addition of 50 �L coldddH20. Of the 80 �L, 70 �L was added to test tubes containing1.25 g strong basic anion resin (Amberjet 4200; Rohm and Haas,Philadelphia, Pennsylvania) in 1 mL ddH20. The mixtures werevortexed for 10 min, centrifuged for 10 min at 10,000 g at roomtemperature, and 200 �L supernatant taken for 14C counting(Liquid Scintillation �-Counter; Kontron, Basel, Switzerland). Theenzymatic activity was calculated by subtracting the valuesobtained with IMPase from the values obtained without IMPase.The use of excess IMPase ensures that the decrement representsonly the product of MIP synthase. The measurements werecarried out in triplicate. To verify that inositol is the reactionproduct under these conditions, the following procedures wereundertaken in two separate experiments. The supernatant of theanion exchanger was concentrated (Speed HetoVac, Heto-Hol-ten, Allerod, Denmark), 20 �g/mL unlabeled inositol added, andthe mixture spotted onto a silica thin layer plate (Alguram SILG/UV, Marchering-Nagel, Duren, Germany). In parallel, standardamounts of 3H-inositol (.25 �Ci and 20 �g/mL unlabeled inosi-tol), 14C-glucose-6-P (.05 �Ci and 20 �g/mL unlabeled glucose-6-P), 14C-glucose (.05 �Ci and 20 �g/mL unlabeled glucose), andunlabeled inositol-1-P (20 �g/mL) were spotted. The plate wasrun in an equilibrated tank containing 120 mL of isopropanol/pyridine/water (3:1:1, by volume) for 8 hours. After drying, theplate was exposed to iodine crystals overnight, spots weremarked, and the plate was exposed to Kodak Biomax MS film(Kodak, Rochester, New York) in a cassette containing KodakBioMax TranScreen-LE screen for 4 days for autoradiography.

Inositol identified in this manner was scraped from the silica

www.elsevier.com/locate/biopsych

pNtIgoIstMs

R

opd(c2(Df2ia(

M

SttlaDcpAacWaIeRt

870 BIOL PSYCHIATRY 2004;56:868–874 G. Shaltiel et al

w

lates, and radioactivity was counted in a �-scintillation counter.et cpm in 14C-inositol (cpm in the presence of IMPase minus in

he absence of IMPase) was 400–650 cpm in the control samples.n the presence of 1 mmol/L VPA, no cpm were observed. Theross counts of the glucose spots were on the order of magnitudef 5000 cpm and did not differ in the presence or absence ofMPase. Similar results were obtained by direct counting of theupernatant of the anion exchanger. Table 1 details the drugsested for possible reduction of crude human brain homogenateIP synthase activity. Each drug was tested twice with triplicate

amples.

at Neuron DRG ExplantsDorsal root ganglia were dissected from the spinal cord area

f newborn Sprague-Dawley rats and cultured individually onoly-ornithine and laminin-coated coverslips in serum-free me-ium supplemented with mouse 7S form nerve growth factor25 ng/mL) at 37°C with 5% CO2. After 24 hours for attachment,ytosine �-arabinofuranoside (ara-C, 10 �mol/L) was added for4 hours to kill nonneuronal cells. Valproate and its derivatives1 mmol/L) in fresh serum-free medium were then added to theRG explants for 24-hour exposure. The explants were then

ixed in 4% paraformaldehyde in phosphate-buffered saline for0 min at room temperature. Growth cones were analyzed on annverted microscope (Zeiss Axiovert; Carl Zeiss, Jena, Germany)nd measured with National Institutes of Health Image softwareBethesda, Maryland).

easurement of Yeast INO1 ExpressionThe Saccharomyces cerevisiae strain used in this study was

H302, a derivative of PMY168, genotype his3-�200, leu2-�1,rp1-�63 and ura3–52. Minimal synthetic defined medium con-ained glucose (2% wt/vol), amino acids (histidine [10 mg/L],eucine [60 mg/L], methionine [10 mg/L], tryptophan [10 mg/L])nd uracil (10 mg/mL), and the salts and vitamin components ofifco Vitamin Free Yeast Base (Difco, Detroit, Michigan). Liquidultures were supplemented with the indicated concentration ofsychotropic drug (Table 2), inoculated with yeast cells to an

550 of .04, incubated at 30°C with constant shaking for 20 hours,nd harvested in the late logarithmic phase of growth. Ribonu-leic acid (RNA) was isolated by hot phenol extraction (Elion andarner 1984), fractionated on an agarose gel, and transferred tonylon membrane. The blots were hybridized with 32P-labeled

NO1 riboprobe, followed by a riboprobe for the constitutivelyxpressed ribosomal protein gene TCM1 to normalize for totalNA. Ribonucleic acid probes for Northern analysis were syn-

Table 1. Psychotropic Drugs Screened for Possible InhiMIP Synthase Activity

Type of Drug Drug

Anticonvulsant Mood Stabilizers CarbamazepinePhenytoinLamotrigine

Typical Antipsychotics ChlorpromazineHaloperidol

Atypical Antipsychotics ClozapineTricyclic Antidepressants Clomipramine

Imipramine

MIP, myo-inositol-1-phosphate.

hesized with the Gemini II core system (Promega, Madison,

ww.elsevier.com/locate/biopsych

Wisconsin) from plasmids linearized with restriction enzymes asfollows (plasmid, restriction enzyme, RNA polymerase): pJH310,HindIII, T7 (INO1); pAB309, EcoRI, SP6 (TCM1) (Ashburner andLopes 1995). The results were visualized and quantitated byphosphorimaging.

Statistical AnalysisStudent’s t test was used to compare brain inositol levels of

mice treated chronically with VPA with levels in control mice.Analysis of variance was used to compare brain inositol levels ofmice treated acutely with the three different doses of VPA withlevels in control mice.

Results

Effect of Acute and Chronic VPA Administration on MouseBrain Inositol Levels

Frontal cortex inositol levels were measured in mice injectedwith a single dose of VPA. Acute VPA administration resulted ina statistically significant 20% reduction in frontal cortex inositol atall concentrations tested (Figure 1). Because both control groups(animals killed immediately after the saline injection or 1 hourafter the injection) produced exactly the same mean inositollevels, the two groups were pooled. Inositol levels of two mice(one from the control group and one from the 800 mg/kg VPAgroup) were excluded because they exceeded 2 SD of eachgroup’s mean.

In contrast to the acute response, brain inositol was notdecreased after chronic administration of VPA to mice for 11 daysin drinking water or 14 days IP (Table 3).

of Human Postmortem Prefrontal Brain Homogenate

herapeuticRange Concentration Studied

mL ng/mL �g/mL (mmol/L) ng/mL (�mol/L)

12 20 (.08)15 20 (.08)20 30 (.12)

25–150 250 (.78)2–15 30 (.08)

350 750 (2.29)200–500 750 (2.63)

50–350 500 (1.57)

Table 2. Valproate But Not Other Psychotropic Drugs UpregulatesINO1 Expression

Treatment INO1/TCM1

Control 1.00Inositol (40 �mol/L) .13Valproate (.6 mmol/L) 15.50Carbamazepine (50 �mol/L) 1.57Phenytoin (70 �mol/L) 1.83Lamotrigine (78 �mol/L) 1.04Haloperidol (.032 �mol/L) 1.17Imipramine (.79 �mol/L) 1.40

bition

T

�g/

4–10–

8–

Data represent the average of two independent experiments.

VH

httcrpcaspTfinpVi

RC

parr9T

Fip(38

TW

3(2(

G. Shaltiel et al BIOL PSYCHIATRY 2004;56:868–874 871

PA Reduces MIP Synthase Activity in Crude Homogenate ofuman Prefrontal Cortex

We characterized the effect of VPA on MIP synthase activity inuman prefrontal cortex, the tissue that is implicated in theherapeutic effect of the drug. A Dixon’s plot (Figure 2) showshat human brain MIP synthase activity is reduced by therapeuticoncentrations of VPA (plasma range .35–.7 mmol/L); 50%eduction is obtained at .21 mmol/L VPA. Myo-inositol-1-phos-hate synthase activity at substrate (glucose-6-phosphate) con-entrations in the range of .48–2.48 mmol/L in the presence andbsence of .525 mmol/L VPA was measured. The results pre-ented as a Lineweaver-Burk plot revealed an apparent noncom-etitive mode of inhibition of MIP synthase by VPA (Figure 3).he apparent Michaelis-Menten constant (Km) of MIP synthaseor glucose-6-phosphate and the apparent maximal initial veloc-ty (Vmax) derived from the figure are .625 mmol/L and .02mol/min mg protein, respectively. The apparent Km in theresence of .525 mmol/L VPA does not change, and the apparentmax decreases to .006 nmol/min mg protein; 50% reduction

s obtained at .28 mmol/L VPA.

eduction of MIP Synthase Activity by VPA Derivativesorrelates with Their Effect on Rat Neurons

Valpromide (VPD) and N-methyl-2,2,3,3,-tetramethyl-cyclo-ropane carboxamide (M-TMCD) are central nervous system–ctive derivatives of VPA (Bialer 1991; Bialer et al 1996; Isoher-anen et al 2002). At comparable concentrations, M-TMCDeduces human brain crude homogenate MIP synthase activity by3%, whereas VPD reduces the activity by only 34% (Table 4).he effect of these VPA derivatives on MIP synthase is consistent

igure 1. Acute valproate (VPA) treatment depletes mice frontal cortexnositol levels. Results are means � SE. Analysis of variance: F(3,59) � 3.14,� .031, n � 25 (control), 13 (300 mg/kg VPA), 13 (600 mg/kg VPA), and 12

800 mg/kg VPA). *Post hoc least significant difference tests: control vs.00 mg/kg VPA, p � .038; control vs. 600 mg/kg VPA, p � .018; control vs.00 mg/kg VPA, p � .024. W.W, wet weight.

able 3. Chronic Valproate (VPA) Administration Does Not Affect Micehole Brain Inositol Levels

Valproate Control(mmol/kg

wet weight)(mmol/kg

wet weight) Treatment

0.1 � 1.7 32.5 � 2.0 VPA in drinking watern � 29) (n � 30)7.9 � 3.2 29.0 � 2.7 IP injection of VPA

n � 15) (n � 13)

Data are presented as mean � SEM.

Student’s t test between VPA-treated mice and control mice: p � ns.

with their effects on the DRG growth cone morphology. Valpro-mide showed no increased growth cone spreading, whereas theeffect of M-TMCD was similar to that of VPA, namely increasinggrowth cone spreading (Figure 4).

Specificity of the Effect of VPA on Human Brain MIP SynthaseActivity

We examined the effect of other psychotropic drugs onhuman brain MIP synthase activity. None of the other anticon-vulsant mood stabilizers, including carbamazepine, lamotrigine,or phenytoin; the typical antipsychotics haloperidol and chlor-promazine; the atypical antipsychotic clozapine; or the tricyclicantidepressants clomipramine and imipramine affected MIP syn-thase activity at concentrations 1.5 times greater than the uppertherapeutic dose (Table 1). Consistent with these observations,only VPA upregulated the yeast INO1 gene (Table 2).

Discussion

The inositol depletion hypothesis proposed by Berridge (1985)is based on the observed inhibition of inositol biosynthesis bylithium. In this report, we show that the mood-stabilizing drug VPAalso inhibits inositol biosynthesis in human brain. The fact that thetwo structurally dissimilar drugs reduce inositol synthesis lendssupport to the hypothesis that inositol depletion is a first event in themolecular mechanism of these anti-bipolar drugs. In the presentstudy, we determined that acute VPA administration to mice results

Figure 2. Human brain myo-inositol-1-phosphate synthase activity is reducedby therapeutically relevant valproate (VPA) concentrations (Dixon’s plot). Re-sults are means � SE. The intersect of the best fit line with the x axis � � (con-centration leading to 50% reduced activity). n � 10 (0 mmol/L), 5 (.35 mmol/L),7 (.525 mmol/L) and 7 (.7 mmol/L). V0, activity in the absence of VPA; VVPA,activity in the presence of the appropriate VPA concentration.

Figure 3. Valproate (VPA) reduces myo-inositol-1-phosphate synthase ac-tivity by an apparent noncompetitive mode (a Lineweaver-Burk plot). Theintersect of the best fit line with the x axis � �1/apparent Km. The intersect

of the best fit line with the y axis � 1/apparent Vmax.

www.elsevier.com/locate/biopsych

itosac1tncrrrc1

sbraMtciagrrIc

diIebpwdelieiImmyt

T

T

VVM

pM2

872 BIOL PSYCHIATRY 2004;56:868–874 G. Shaltiel et al

w

n a 20% reduction in frontal cortex inositol concentrations, even athe lowest dose tested (300 mg/kg). The decrease in inositol is notbserved after chronic treatment with VPA. These findings areimilar to the observations that intracellular inositol was reducedfter acute treatment with lithium, whereas reports of reduction afterhronic lithium administration were less consistent (Belmaker et al998). Reports concerning brain inositol levels after VPA adminis-ration to mammals are also inconsistent. Moore et al (2000) foundo clinical- or VPA-related changes in cingulate cortex inositol/reatine-phosphocreatine ratio, whereas O’Donnell et al (2000)eported reduction in rat brain inositol levels measured by magneticesonance spectroscopy after chronic VPA. It is noticeable that ouresults are in mice that exhibit a VPA half-life time of .8 hours,ompared with 2–5 hours in rat and 9–18 hours in humans (Loscher999).

We also show that VPA causes a decrease in activity of MIPynthase in crude homogenate human prefrontal cortex. It haseen previously suggested that VPA causes inositol depletion inat cultured neurons as well (Williams et al 2002). These resultsre consistent with the hypothesis that VPA leads to decreasedIP synthase activity and inositol depletion in the acute phase of

reatment, after which metabolism is adapted to restore adequateoncentrations of inositol. The acute decrease in intracellularnositol might further lead to a cascade of secondary changesssociated with mood stabilization, such as altered expression ofenes that respond to inositol levels (Vaden et al 2001). Ourecent finding that treatment of mice with lithium for 10 daysesults in 30% upregulation of hippocampal MIP synthase andMPA1 mRNA levels (Shamir et al 2003) might represent such aonsequence.

Although regulation of inositol biosynthesis is not well un-erstood in mammalian cells, this pathway has been character-zed at the molecular level in the yeast Saccharomyces cerevisiae.n yeast, inositol depletion leads to a dramatic increase inxpression of INO1, the gene encoding MIP synthase (reviewedy Carman and Henry 1999; Greenberg and Lopes 1996). Val-roate causes a decrease in synthesis of inositol-1-P, consistentith decreased MIP synthase activity. This is followed by aecrease in intracellular inositol and a dramatic increase in INO1xpression (Vaden et al 2001). Inhibition of IMPase by lithiumeads to a transient decrease in intracellular inositol, as well as anncrease in INO1 expression (Murray and Greenberg 1997; Vadent al 2001). Thus, in yeast and mammalian cells, the decrease inntracellular inositol by both drugs, due to reduced activity ofMPase by lithium and MIP synthase by VPA, is conserved. Theechanism of the reduction of MIP synthase activity by VPA isost likely indirect, given that VPA does not inhibit the purified

east or human enzymes (Ju et al 2004). Therefore, we propose

able 4. Effect of VPA Derivatives on Human Brain MIP Synthase Activity

reatmentReduction of Human BrainMIP Synthase Activity (%)

PA (1.0 mmol/L) 100 � 0PD (1.4 mmol/L) 34 � 11-TMCD (1.3 mmol/L) 93 � 7

Data are presented as mean � SEM.MIP synthase activity in the control assays was .14 � .06 nmol/min mg

rotein. n � 5 (control), 10 (VPA), 4 (VPD), and 2 (M-TMCD). VPA, valproate;IP, myo-inositol-1-phosphate; VPD, valpromide; M-TMCD, N-methyl-

,2,3,3,-tetramethyl-cyclopropane carboxamide.

hat tissue component(s) induce a conformation of MIP synthase

ww.elsevier.com/locate/biopsych

that is susceptible to VPA or that VPA is metabolized to a productthat directly inhibits MIP synthase.

The apparent Vmax for human MIP synthase obtained fromall control experiments (no drug) in the present study is in therange of .02–.2 nmol/min mg protein. This is comparable withthe Vmax reported by Barnett et al (1970) in rat testis(.104 nmol/min mg protein) and differs from that reported byEisenberg (1967). O’Donnell et al (2000) measured a basalglucose-6-phosphate concentration of .226 �mol/g in rat brain.Assuming similar human and rat brain glucose-6-phosphatelevels, and given the apparent Km (.625 mmol/L) measured inthe present study, in vivo MIP synthase activity is less than half ofits Vmax in the human brain, enabling a significant range forregulatory manipulation. The VPA concentrations that lead to a50% reduction of crude homogenate postmortem human brainMIP synthase activity reported here, .21 mmol/L from the Dixon’splot and .28 mmol/L from the Lineweaver-Burk plot, are in goodagreement. Because brain levels of VPA are approximately 20%of blood levels (Loscher 1999), therapeutic plasma levels of VPAwould be sufficient for approximately a 50% reduction of MIPsynthase activity in this tissue. In human brain tissue andneuronal cell cultures, the pathways that supply intracellular freeinositol exhibit the following comparable Vmax values: foruptake via the sodium-myo-inositol transporter: .06–.10 nmol/min mg protein (Hertz et al 1997); and for inositol-1-phosphatedephosphorylation via IMPase: .2–2.0 nmol/min mg protein(Shaltiel et al 2001). This makes the calculated value of .02–.20nmol/min mg protein for MIP synthase a meaningful contri-bution to cellular brain myo-inositol.

Lithium (.1–1.0 mmol/L) has been reported to have no effecton rat testis purified MIP synthase activity (Maeda and Eisenberg1980). We now show that reduction of crude homogenatepostmortem human brain MIP synthase activity is specific forVPA and is not demonstrated by other anti-bipolar drugs, typicaland atypical antipsychotics and tricyclic antidepressants. Simi-

Figure 4. The effects of valproate (VPA) derivatives on spreading of ratneuron dorsal root ganglia (DRG). (a) Control, (B) 1 mmol/L VPA, (C) 1mmol/L valpromide (VPD), (D) 1 mmol/L N-methyl-2,2,3,3,-tetramethyl-cy-clopropane carboxamide (M-TMCD). For each condition, two DRG weretested, 30 growth cones per DRG, so that overall 60 growth cones per eachcondition were measured. Histograms show frequency distribution of thearea of the growth cones. Gray bars represent contracted growth cones

(spread area �50 �m2).

lbebb111rtdgmmsem

Vtudngr

SM

A

A

B

B

B

B

B

B

B

B

C

C

G. Shaltiel et al BIOL PSYCHIATRY 2004;56:868–874 873

arly, these other psychotropic drugs do not affect yeast inositoliosynthesis, because only VPA led to an increase in yeast INO1xpression. It is unlikely that the drugs are not taken up by yeast,ecause similar compounds in yeast and other cell cultures haveeen shown to exhibit effects on sodium channels (Lang et al993), calcium influx and H�-adenosine triphosphatase (Eilam983; Eilam et al 1985), and ergosterol content (Moebius et al996). Interestingly, two VPA derivatives, VPD and M-TMCD,educed postmortem human brain crude homogenate MIP syn-hase activity. This reduction correlates with the effect of theserugs on the growth cones of rat DRGs. The morphology ofrowth cones of these cultured neurons is an additional relevantodel for studying drug effects. Valproate and its derivativesight also affect the growth cones by additional non-MIP

ynthase mechanisms; however, the previous results of Williamst al (2002) strongly suggest the involvement of inositol in thisodel.The discovery that human brain MIP synthase is affected by

PA identifies a new target for the development of drugs for thereatment of bipolar disorder. Carbamazepine, another widelysed mood stabilizer, is also reported to act via an inositolepletion mechanism (Williams et al 2002); however, this doesot involve reduction of IMPase or MIP synthase activity, sug-esting that there might be additional drug targets involved in theegulation of inositol concentration.

This study was supported by United States–Israel Binationalcience Foundation grant 2001035 (MLG and GA) and by grantH56220 from the National Institutes of Health (MLG).

gam G, Shapiro Y, Bersudsky Y, Kofman O, Belmaker RH (1994): High-doseperipheral inositol raises brain inositol levels and reverses behavioraleffects of inositol depletion by lithium. Pharmacol Biochem Behav 49:341–343.

shburner BP, Lopes JM (1995): Regulation of yeast phospholipid biosyn-thetic gene expression in response to inositol involves two superim-posed mechanisms. Proc Natl Acad Sci U S A 92:9722–9726.

arnett JE, Brice RE, Corina DL (1970): A colorimetric determination of inosi-tol monophosphates as an assay for D-glucose 6-phosphate-1L-myo-inositol 1-phosphate cyclase. Biochem J 119:183–186.

elmaker RH, Agam G, van Calker D, Richards MH, Kofman O (1998): Behav-ioral reversal of lithium effects by four inositol isomers correlates per-fectly with biochemical effects on the PI cycle: Depletion by chroniclithium of brain inositol is specific to hypothalamus and inositol levelsmay be abnormal in postmortem brain from bipolar patients. Neuropsy-chopharmacology 19:220 –232.

erridge MJ (1985): Phosphoinositides and signal transductions. Rev ClinBasic Pharm 5(suppl):5S–13S.

ersudsky Y, Kaplan Z, Shapiro Y, Agam G, Kofman O, Belmaker RH (1994):Behavioral evidence for the existence of two pools of cellular inositol. EurNeuropsychopharmacol 4:463– 467.

ialer M (1991): Clinical pharmacology of valpromide. Clin Pharmacokinet20:114 –122.

ialer M, Hadad S, Kadry B, Abdul-Hai A, Haj-Yehia A, Sterling J, et al (1996):Pharmacokinetic analysis and antiepileptic activity of tetra-methylcyclo-propane analogues of valpromide. Pharm Res 13:284 –289.

own CD, Wang JF, Chen B, Young LT (2002): Regulation of ER stress proteinsby valproate: Therapeutic implications. Bipolar Disord 4:145–151.

usa WB, Gimlich RL (1989): Lithium-induced teratogenesis in frog embryosprevented by a polyphosphoinositide cycle intermediate or a diacylglyc-erol analog. Dev Biol 132:315–324.

arman GM, Henry SA (1999): Phospholipid biosynthesis in the yeast Sac-charomyces cerevisiae and interrelationship with other metabolic pro-cesses. Prog Lipid Res 38:361–399.

asebolt TL, Jope RS (1989): Long-term lithium treatment selectively re-duces receptor-coupled inositol phospholipid hydrolysis in rat brain.

Biol Psychiatry 25:329 –340.

Chen G, Zeng WZ, Yuan PX, Huang LD, Jiang YM, Zhao ZH, et al (1999): Themood-stabilizing agents lithium and valproate robustly increase thelevels of the neuroprotective protein bcl-2 in the CNS. J Neurochem72:879 – 882.

Council NR (1996): Guide for the Care and Use of Laboratory Animals.Washington, DC: National Academy Press.

Dixon JF, Hokin LE (1997): The anti-bipolar drug valproate mimics lithium instimulating glutamate release and inositol 1,4,5-trisphosphate accumu-lation in brain cortex slices but not accumulation of inositol monophos-phates and bisphosphates. Proc Natl Acad Sci U S A 94:4757– 4760.

Eilam Y (1983): Membrane effects of phenothiazines in yeasts. I. Stimulationof calcium and potassium fluxes. Biochim Biophys Acta 733:242–248.

Eilam Y, Lavi H, Grossowicz N (1985): Active extrusion of potassium in theyeast Saccharomyces cerevisiae induced by low concentrations of triflu-operazine. J Gen Microbiol 131:2555–2564.

Eisenberg F Jr (1967): D-myoinositol 1-phosphate as product of cyclizationof glucose 6-phosphate and substrate for a specific phosphatase in rattestis. J Biol Chem 242:1375–1382.

Elion EA, Warner JR (1984): The major promoter element of rRNA transcrip-tion in yeast lies 2 kb upstream. Cell 39:663– 673.

Godfrey PP, McClue SJ, White AM, Wood AJ, Grahame-Smith DG (1989):Subacute and chronic in vivo lithium treatment inhibits agonist- andsodium fluoride-stimulated inositol phosphate production in rat cortex.J Neurochem 52:498 –506.

Greenberg ML, Lopes JM (1996): Genetic regulation of phospholipid biosyn-thesis in Saccharomyces cerevisiae. Microbiol Rev 60:1–20.

Gross-Isseroff R, Salama D, Israeli M, Biegon A (1990): Autoradiographicanalysis of [3H]ketanserin binding in the human brain postmortem:Effect of suicide. Brain Res 507:208 –215.

Hallcher LM, Sherman WR (1980): The effects of lithium ion and other agentson the activity of myo-inositol-1-phosphatase from bovine brain. J BiolChem 255:10896 –10901.

Hertz L, Wolfson M, Hertz E, Agam G, Richards M, Belmaker RH (1997):Lithium-inositol interactions: Synthesis, uptake, turnover. In: Honig A,Van Praag HM, editors. Depression: Neurobiological, Psychopathologicaland Therapeutic Advances. Chichester, United Kingdom: John Wiley &Sons 519 –534.

Isoherranen N, White HS, Finnell RH, Yagen B, Woodhead JH, Bennett GD, etal (2002): Anticonvulsant profile and teratogenicity of N-methyl-tetra-methylcyclopropyl carboxamide: A new antiepileptic drug. Epilepsia 43:115–126.

Jope RS (1999): Anti-bipolar therapy: Mechanism of action of lithium. MolPsychiatry 4:117–128.

Jope RS, Williams MB (1994): Lithium and brain signal transduction systems.Biochem Pharmacol 47:429 – 441.

Ju S, Shaltiel G, Shamir A, Agam G, Greenberg ML (2004): Human 1-D-myo-inositol-3-phosphate synthase is functional in yeast. J Biol Chem 279:21759 –21765.

Kendall DA, Nahorski SR (1987): Acute and chronic lithium treatments influ-ence agonist and depolarization-stimulated inositol phospholipid hy-drolysis in rat cerebral cortex. J Pharmacol Exp Ther 241:1023–1027.

Kofman O, Belmaker RH (1993): Ziskind-Somerfeld Research Award 1993.Biochemical, behavioral and clinical studies of the role of inositol inlithium treatment and depression. Biol Psychiatry 34:839 – 852.

Kofman O, Belmaker RH, Grisaru N, Alpert C, Fuchs I, Katz V, et al (1991):Myo-inositol attenuates two specific behavioral effects of acute lithiumin rats. Psychopharmocol Bull 27:185–190.

Lang DG, Wang CM, Cooper BR (1993): Lamotrigine, phenytoin and carba-mazepine interactions on the sodium current present in N4TG1 mouseneuroblastoma cells. J Pharmacol Exp Ther 266:829 – 835.

Loscher W (1999): Valproate: A reappraisal of its pharmacodynamic proper-ties and mechanisms of action. Prog Neurobiol 58:31–59.

Loscher W (2002): Basic pharmacology of valproate: A review after 35 yearsof clinical use for the treatment of epilepsy. CNS Drugs 16:669 – 694.

Maeda T, Eisenberg F Jr (1980): Purification, structure and catalytic proper-ties of L-myo-inositol-1-phosphate synthase from rat testis. J Biol Chem255:8458 – 8464.

Majumder AL, Johnson MD, Henry SA (1997): 1L-myo-inositol-1-phosphatesynthase. Biochim Biophys Acta 1348:245–256.

Manji HK, Lenox RH (1994): Long-term action of lithium: A role for transcrip-tional and posttranscriptional factors regulated by protein kinase C.

Synapse 16:11–28.

www.elsevier.com/locate/biopsych

M

M

M

M

N

O

P

S

874 BIOL PSYCHIATRY 2004;56:868–874 G. Shaltiel et al

w

anji HK, Potter WZ, Lenox RH (1995): Signal transduction pathways. Molec-ular targets for lithium’s actions. Arch Gen Psychiatry 52:531–543.

oebius FF, Bermoser K, Reiter RJ, Hanner M, Glossmann H (1996): Yeaststerol C8-C7 isomerase: Identification and characterization of a high-affinity binding site for enzyme inhibitors. Biochemistry 35:16871–16878.

oore CM, Breeze JL, Gruber SA, Babb SM, Frederick BB, Villafuerte RA, et al(2000): Choline, myo-inositol and mood in bipolar disorder: A protonmagnetic resonance spectroscopic imaging study of the anterior cingu-late cortex. Bipolar Disord 2:207–216.

urray M, Greenberg ML (1997): Regulation of inositol monophosphatase inSaccharomyces cerevisiae. Mol Microbiol 25:541–546.

ovak JE, Turner RS, Agranoff BW, Fisher SK (1999): Differentiated humanNT2-N neurons possess a high intracellular content of myo-inositol.J Neurochem 72:1431–1440.

’Donnell T, Rotzinger S, Nakashima TT, Hanstock CC, Ulrich M, SilverstonePH (2000): Chronic lithium and sodium valproate both decrease theconcentration of myo-inositol and increase the concentration of inositolmonophosphates in rat brain. Brain Res 880:84 –91.

hiel CJ, Zhang F, Huang EY, Guenther MG, Lazar MA, Klein PS (2001):Histone deacetylase is a direct target of valproic acid, a potent anti-convulsant, mood stabilizer and teratogen. J Biol Chem 276:36734 –36741.

haldubina A, Ju S, Vaden D, Ding D, Belmaker RH, Greenberg ML (2002):

Epi-inositol regulates expression of the yeast INO1 gene encoding ino-sitol-1-P synthase. Mol Psychiatry 7:174 –180.

ww.elsevier.com/locate/biopsych

Shaltiel G, Shamir A, Nemanov L, Yaroslavsky Y, Nemets B, Ebstein RP, et al(2001): Inositol monophosphatase activity in brain and lymphocyte-derived cell lines of bipolar, schizophrenic and unipolar patients. WorldJ Biol Psychiatry 2:95–98.

Shamir A, Shaltiel G, Greenberg ML, Belmaker RH, Agam G (2003): Lithium’seffect on gene expression of inositol synthesis enzymes in mouse hip-pocampus: Comparison with the yeast model. Mol Brain Res 115:104 –110.

Shapiro J, Belmaker RH, Beigon A, Seker A, Agam G (2000): Scyllo-inositol inpost-mortem brain of suicide victims, bipolar, unipolar and schizo-phrenic patients. J Neural Transm 107:603– 607.

Vaden DL, Ding D, Peterson B, Greenberg ML (2001): Lithium and valproatedecrease inositol mass and increase expression of the yeast INO1 andINO2 genes for inositol biosynthesis. J Biol Chem 276:15466 –15471.

Weissman MM, Leaf PJ, Tischler GL, Blazer DG, Karno M, Bruce ML, Florio LP(1988): Affective disorders in five United States communities. PsycholMed 18:141–153.

Whitlow RD, Sacher A, Loo DD, Nelson N, Eskandari S (2003): The anticonvul-sant valproate increases the turnover rate of gamma -aminobutyric acidtransporters. J Biol Chem 278:17716 –17726.

Williams RS, Cheng L, Mudge AW, Harwood AJ (2002): A common mecha-nism of action for three mood-stabilizing drugs. Nature 417:292–295.

Wong YH, Kalmbach SJ, Hartman BK, Sherman WR (1987): Immunohisto-chemical staining and enzyme activity measurements show myo-inosi-

tol-1-phosphate synthase to be localized in the vasculature of brain.J Neurochem 48:1434 –1442.