Journal of NeurochemistryLippincott—RavenPublishers,Philadelphia© 1997 InternationalSociety for Neurochemistry

Reduced [3H] Cyclic AMP Binding in Postmortem Brainfrom Subjects with Bipolar Affective Disorder

*shafiqur Rahman,*t~peterP. Li, §L. Trevor Young, ~OraKofman, 1~~StephenJ. Kish,

and *t1~JerryJ. Warsh

* Sectionof Biochemical Psychiatry and ¶Human NeurochemicalPathologyLaboratory, Clarke Institute of Psychiatry;

Departmentsof ~Psvchiatry and ~:Pharmacology and 5lnstituteof Medical Sciences,University of Toronto, Toronto;§Departmentof Psychiatry,McMaster University, Hamilton, Ontario, Canada;and ~Department of Behavioral Sciences,

Ben Gurion University of the Neget’, BeerSheva,Israel

Abstract: Findings of increased G5cr levels and forskolin-

stimulated adenylyl cyclase activity in selective cerebralcortical postmortem brain regions in bipolar affective dis-order (BD) implicate increased cyclic AMP (cAMP)-medi-ated signaling in this illness. Accumulating evidence sug-gests that intracellular levels of cAMP modulate the abun-dance and disposition of the regulatory subunits ofcAMP-dependent protein kinase (cAMP-dPK). Thus, inthe present study, we tested further whether hyperfunc-tional G5a-linked cAMP signaling occurs in BD by de-termining [

3H]cAMPbinding, a measure of the levels ofregulatory subunits of cAMP-dPK, in cytosolic and mem-brane fractions from discrete brain regionsof postmortemBD brain. Specific [3H]cAMP(5 nM) binding was deter-mined in autopsied brain obtained from 10 patients withDSM-Ill-R diagnoses of BD compared with age- andpostmortem delay-matched controls. [3H]cAMPbindingwas significantly reduced across all brain regions in cyto-solic fractions of BD frontal (—22%), temporal (—23%),occipital (—22%) and parietal (— 15%) cortex, cerebellum(—36%), and thalamus (—13%) compared with controls,but there were no differences in [3H]cAMPbinding inthe membrane fractions from these same regions. Theseresults suggest that changes occur in the cAMP-dPK reg-ulatory subunits in BD brain, possibly resulting from in-creased cAMP signaling. The possibility that antemortemlithium and/or other mood stabilizer treatment may con-tribute to the above changes, however, cannot be ruledout. Key Words: Cyclic AMP-dependent protein kinase—Postmortem brain—Lithium— Bipolar affective disorder.J. Neurochem.68, 297—304 (1997).

Avissar and Schreiber,1992; Young et al., 1993). Inthis regard,findingsof elevatedcerebralcorticalstimu-latory G proteina subunit(Gsa)levels in postmortemBD brain, which were also associatedwith increasedforskolin-stimulated adenylyl cyclase activity, areamongthemostdirect evidenceof postreceptordistur-bancein this disorder(Young et al., 1991, 1993).

Although the mechanism(s)underlyingthe G pro-tein changesin brain from BD subjectsis yet to beelucidated,an equally important questionis that ofwhethersuchelevationsare attendedby increasedsig-naling throughthe G

5a-mediatedcyclic AMP (cAMP)signalingcascade.Aside from the aboveobservationof increasedforskolin-stimulatedadenylyl cyclaseac-tivity in temporalandoccipital cerebralcorticalregionsfrom BD postmortembrain (Young et al., 1993),thereis a paucity of data to support this notion. Unfortu-nately, the rapid changesin brain cAMP levels post-mortemprecludethe useof measurementsof this sec-ondmessengerto assessthestateof cAMP signalinginbrain (Jonesand Stavinoha,1979). Thereare severalimportantprotein targetsin the cAMP signalingcas-cade,however, the function of which is modulatedby this secondmessenger.Thus, measurementof theconcentrationsand/or function of such downstreamtargetsmay provide further evidenceof alteredcAMPsignaling in postmortembrain in neuropathologicalconditions(Nishino et al., 1993).

cAMP-dependentprotein kinase (cAMP-dPK), atetramercomposedof a dimeric regulatory (R) andtwo monomericcatalytic (C) subunits,is the primary

The pathogenesisof bipolaraffectivedisorder(BD)has beenlinked to disturbancesin neuronalfunctionsecondaryto alteredactionof oneor moreintracellularsecondmessengers(Manji, 1992;Hudsonet al., 1993;WarshandLi, 1996).Substantialevidenceimplicatesalteredheterotrimericguaninenucleotidebinding (G)proteinfunction asthebasisfor postulatedsignaltrans-duction disturbancesin BD (Schreiberet al., 1991;

ReceivedMay 23, 1996; revised manuscriptreceivedAugust 21,1996; acceptedAugust 22, 1996.

Addresscorrespondenceandreprint requeststo Dr. J. J. WarshatSection of BiochemicalPsychiatry,Clarke Institute of Psychiatry,250 CollegeStreet,Toronto,Ontario, CanadaMST 1R8.

Abbreviotic,osused: AEBSF, 4-( 2-aminoethyl)benzenesulfonylfluoride; BD, bipolaraffectivedisorder;cAMP, cyclic AMP; cAMP-dPK, cyclic AMP-dependentprotein kinase; cGMP, cyclic GMP;IBMX, 3-isobutyl-I -methylxanthine.

297

298 S. RAHMAN ET AL.

target of intracellular cAMP signaling (WalaasandGreengard,1991;Cho-ChungandClair, 1993;Spauld-ing, 1993; Francisand Corbin, 1994). Multiple iso-forms of both R (Ria, Rl1@, and RITa, RH,@) and C(Ca,C~,andCy) subunitshavebeenidentified (Wa-laas and Greengard, 1991; Doskelandet al., 1993;Francis and Corbin, 1994). EachR subunit has twocAMP bindingsitesthat showpositivecooperativityincAMP binding andactivation(Ogreid andDoskeland,1981; Ogreid et al., 1983). On binding of cAMP toeachR subunit, the inactive holoenzymedissociatesinto a dimeric R-subunitcomplexand two monomericactive C subunits (Walaas and Greengard, 1991;Spaulding, 1993). In addition to modulatingcatalyticactivity, recentstudiessuggestedthat sustainedeleva-tions in intracellular cAMP levels cause adaptivechangesin the levels of cAMP-dPK R subunits (forreview, see Spaulding, 1993; Francis and Corbin,1994).Notwithstandingthepossiblecell type-specificnatureof such changesin RI and Rh subunit abun-dance(PrashadandRosenberg,1978;Liu et al., 1981;Budillon et al., 1995; Garrel et al., 1995), alterationin levels of one, the other, or both R subunits in re-sponseto sustainedelevationsin intracellularcAMPlevels may afford a potential markerof modificationsin cAMP signaling. We reasonedthereforethat mea-surementof R subunitlevelsin membraneand/orcyto-solic fractions from postmortemBD brain samples,estimatedby I

3H1cAMP binding, might provide a“trace” of the increasedcAMP signalingpositedtooccur in this disorder.

In the presentstudy, we examinedspecific [3H1-cAMP binding in four cerebralcorticalregionsof post-mortem BD brainsin which G

5a protein levels werepreviouslyshownto behigherthan in matchedcontrols(Young et al., 1993), as well as in several regionsfrom thesesamebrains that showed no G5a contentchanges.We report here significantly lower specific[3HIcAMP binding in cytosolic but not membrane

fractionsin postmortembrain regionsfrom BD com-pared with a nonpsychiatricnonneurologicalcontrolgroup.

MATERIALS AND METHODS

Materials[3H]cAMP (28.2—31.2 Ci/mmol) was obtained from

New England Nuclear-Du Pont (Boston, MA, U.S.A.).cAMP, adenosine,cyclic GMP (cGMP), S ‘-AMP, ATP,EDTA, leupeptin, and 2-mercaptoethanolwere purchasedfrom Sigma (St. Louis, MO, U.S.A.). 4-(2-Amino-ethyl)benzenesulfonyl fluoride (AEBSF), 3-isobutyl-1-methylxanthine(IBMX), and tris(hydroxymethyl)amino-methanewereobtainedfrom Calbiochem(SanDiego, CA,U.S.A.). All otherchemicalsused wereof analyticalgrade.

Postmortem brainAutopsiedbrainswere obtained from 10 subjects(five

malesandfive females)with a verified DSMIII-R (AmericanPsychiatricAssociation,1987)diagnosisof BD andno neu-rological disorder as previously described(Young et al.,

1991. 1993). A comparisongroup, with no history of neuro-logic, psychiatric,or substanceabusedisorder,was matchedasclosely aspossibleon age, gender,andpostmortemdelay.Details of patient histories, including causeof death andantemortemdrug treatment,were obtainedfrom availablemedical recordsandare summarizedin Table I. Autopsiedbrainswere frozen(—70°C)within 24 h after deathexceptfor two subjectsfor which the interval from deathto freezingof the brain was33 and 39 h, respectively.Conditionsfordissectionanddeterminationof brainpH wereas previouslydescribed(Younget al., 1991, 1993).Brain regionsassayedincludedfrontal (middlegyrus, area10, Brodmann’snomen-clature),temporal(middletemporalgyrus,area21), parietal(area7), andoccipital (lips of the calcarinesulcus,area17)cortex, thalamus(mediodorsal),andcerebellarcortex.

[3HIcAMP binding assaySpecific I 5H]cAMP bindingwasperformedas previously

described(Nishino et al., 1993) with minor modifications.In brief, brain samples(20—30 mg) were homogenizedbyultrasonication(sonicatorduty cycle setting 30%,micro tiplimit at 3, 10 s; SonicsandMaterials,Danbury,CT, U.S.A.)in 10 volumesof ice-coldbufferA containing20 mM Tris-HCI (pH 7.4at 25°C),2 mMEDTA, 25 mM2-mercaptoeth-anol, 0.5 mM AEBSF,and10 pg/mI leupeptin,followedbycentrifugationat 48,000g for 30 mm. Thesupernatantwasseparatedand recentrifugedat 48,000g for 30 mm. andthefinal supernatant(cytosolic fraction) was used for [3HJ-cAMP bindingassay.Similarly, theretainedpelletsfrom thetwo centrifugationswerepooledfor eachsampleandwashedtwice by resuspensionin 200 p1 of bufferA andcentrifuga-tion (48,000gfor 15 mm). Thefinal pellets,i.e., membranefractions,were resuspendedin ISO p1 of buffer A. Proteinconcentrationwas determinedby the method of Bradford(1976)usingbovineserumalbumin asthestandard.Aliquotsof the ~.ytosolicand membranefractions were stored at—70°Cuntil assay.

[3HJcAMP bindingassayswere performedin triplicate inan incubationbuffer containing20 mM phosphatebuffer(pH 7.4 at25°C),2 mM EDTA, and15 mM 2-mercaptoetha-nol (PEM buffer), ItHIcAMP (0.125—10nM). membraneor cytosolicprotein (~—25pg), bovine serumalbumin (0.25mg), and1.5 mM IBMX in a total volume of 500p1. Afterincubatingfor 60 mm at roomtemperature,incuhationswereterminatedby rapid filtration undervacuum throughglassfiber receptorbinding filtermat with the Skatron cell har-vester(SkatronInstrument,Lier, Norway) followed by twowasheswith 2 ml of cold PEM buffer. The radioactivityretainedon the filters wasmeasuredby liquid scintillationspectrometry(countingefficiency,42%). Nonspecificbind-ing was definedas the radioactivity bound in the presenceof 5 p.M cAMP. For competition experiments,assayswereperformedwith variousconcentrationsof cAMP (0.01—300nM), cGMP (0.001—30pM), 5’-AMP or ATP (10 pM—ImM), and [3H]cAMP (2.5 nM) in a final volume of500 pl.

Animal treatmentMale Sprague—Dawleyrats (weighing 150—250 g:

CharlesRiver, Montreal,Quebec.Canada)werehousedindi-vidually in a light- (12-h light/dark cycle) andtemperature-controlled environmentand fed either rat chow (controlgroup)or lithium carbonate(0.22%) in rat chow (BioserveLtd.) and waterad libitum for 24 days.Animals werekilledby decapitation,the brainswere rapidly removedfrom the

J. Neurochein.,Vol. 68, No. I, 1997

[3H]cAMP BINDING IN BIPOLAR AFFECTIVE DISORDER 299

TABLE 1. Subjectcharacteristics

Diagnosis

Age

(years)/sex

PM(h) Causeof death Psychotropic drugs

Temporal cortexlithium (mM)°

ControlBipolar”

48/M40/M

1412

CarcinomaHepaticfailure

NoneLithium, haloperidol 0.079

ControlBipolar

66/F70/F

1817

Myocardial infarctionCarcinomawith erosioninto the aorta

NoneLithium. benztropine,perphenazine 0.066

ControlBipolar’

92/F80/M

17.517.4

Rectal carcinomaPneumonia

NoneThioridazine 0.129

ControlBipolar

74/F70/F

1739

Myocardial infarctionPneumonia

3Haloperidol, lorazepam,maprotiline 0.083

ControlBipolar

71/M71/M

1023.5

PneumoniaPneumonia

NoneLithium 0.396

ControlBipolar

51/M37/M

21)23.5

Myocardial infarctionSuicide, overdoseof lithium, codeine

NoneLithium, amitriptyline, codeine.

chlordiazepoxide0.127

ControlBipolar

30/M27/F

2116

Rupturedinferior venacavaSuicide, overdoseof doxepin,

amitriptyline

3Doxepin, amitriptyline 0.154

ControlBipolar

28/F28/F

2316

IntraabdominalhemorrhageSuicide, overdoseof lithium

NoneLithium, thioridazine 0.166

ControlBipolar”

58/F42/F

II33

Rectal carcinomaSuicide, hanging

NoneLithium, prochlorperazine,

isocarboxazid,alprazolamND

ControlBipolar

86/M94/F

98

ArrhythmiaCerebrovascularaccident?

None? 0.080

PM, postmortemdelay; ND, not detected.“Minimal detectablelevel, 0.066 mM.“Coexistent alcohol abuse.‘Differential diagnosisof schizoaffectivedisorder.“Bipolar type 11.

cranium,andtheprefrontalcortexwasdissectedover ice andstoredat —70°Cuntil use.Brain sampleswerehomogenized,centrifuged,and assayedas for humanpostmortembrain.Plasmalithium levelsin ratsweredeterminedon heparinizedtrunk bloodsamplesby a standardclinical procedure(Samp-son Ct aI., 1994) using an ion-selectiveelectrode (NovaBiomedical electrolyte analyzer). All animal procedureswere performedin strict accordancewith the guidelinesofthe CanadianCouncil on Animal Careandwere approvedby the local institutional Animal Care Committee.

Determination of brain lithium levelsLithium concentrationsweredeterminedin BD temporal

cortex by inductively coupledargon plasmaemissionspec-trometry (Joneset al., 1987). In brief, —30 mg of braintissuewas digestedin 1 ml of concentratedultrapurenitricacid in a sealedTeflon vesselusing a pressure-programmedmicrowaveheatingcycle. After digestionwascomplete(22mm), the vesselswere cooled and opened,and digestateswere rinsed out and diluted to 10.0 ml with deionizedultrafiltered water(18 MO). Sampleswere injected into aJarrelAsh model 6lE, 34-channelsimultaneousinductivelycoupledargonplasmaemissionspectrometerusing an ultra-sonic nebulizer to achieve maximum detection sensitivity.Lithium wasdetectedandquantifiedat an emissionwave-lengthof 670.784nm. Sensitivity andcoefficientof variationwere 1.5 ng/ml digestateand 10%, respectively.

Data analysisAll datawereexpressedasmean±SEM values.Analysis

of binding data was performedusing nonlinearreiterativecurve-fittingtechniqueswith the RADLIG softwarepackage(McPherson, 1985). Statistical analysesof the datawereperformedby two-way ANOVA followed by posthoccon-trastsfor simple effects or Tukey’s test using the SPSS6.0(Chicago, IL, U.S.A.) statistical software package.Simplegroupcomparisonsof age,postmortemdelay,andpH wereanalyzedby Student’st tests. Relationshipsbetweenlithiumlevels, age, or postmortemdelay and [‘H]cAMP bindingwereassessedusingthePearsonProductMomentcorrelationanalysis.Values of p < 0.05 were consideredstatisticallysignificant.

RESULTS

Subject characteristicsBD and comparisongroups did not differ signifi-

cantly (p > 0.1) in meanage (56 ± 8 vs. 60 ± 7years)or postmortemdelay (21 ± 3 vs. 16 ±2 h).Six of the BD patientshad a history of maintenancelithium treatmentwithin the 6 months before death;four of thesehad evidenceof continued lithium usewithin 4 weeks antemortem.The remaining four pa-

.1. Neurochem.,Vol. 68. No. 1, 1997

300 S. RAHMANET AL.

otidestested, cAMPwas the most potent(displacementoccurredat nanomolarconcentrations).In comparison,cGMP, 5 ‘-AMP, and ATP showedonly marginaldis-placementevenat micromolarconcentrations(datanotshown).

FIG. 1. Saturation experiment with [3H]cAMPin cytosolic frac-tion of postmortem control temporal cortex. Cytosolic fractions(—-25 pg of protein) were incubated with varying concentrationsof[3H]cAMP in the absence (0; total binding) and presence (LI;nonspecific binding) of 5 pM cAMP as described in Materialsand Methods. Specifically bound [3H]cAMPis also indicated(.). Inset: Scatchard plot of specific [3H]cAMPbinding data.Each point represents the mean of triplicate determinations. Thedata were best fit for a one-site model. B/F, bound/free.

tients had no documented historyof lithium use inthe 6-month interval before death, althoughpreviouslithium therapy could not becompletely ruled outbasedon available medicalrecords (Table 1). BrainpH, which affords an index of agonal status, wasslightly but significantly higher (4%) in BD (6.56±0.05) compared with control (6.30 ± 0.08; p<0.05) subjects.

Characterization of I 3H] cAMP bindingIn preliminary experiments, specific binding of

[3H1cAMP (5 nM) was measuredin cytosolic andmembranefractionsof postmortem frontal cortexfroma control subject.Maximum [3HJcAMP binding wasreached at 60mm, remainedstableup to 120 mm,and waslinear with respectto protein concentrationbetween0.01 and 0.08 mg ofprotein(datanotshown).As demonstratedin Fig. I, specific [3H1 cAMP bindingin cytosolic fractionsof control postmortemtemporalcortexwassaturableandexhibiteda singlehigh-affin-ity binding site with a dissociationconstant(Ks) of0.73 ±0.04 nM and maximalbinding density(Br,,ax)of 232 ±7 fmol/mg of protein (n = 3) as estimatedby nonlinear reiterative curvefitting and Scatchardanalyses.Nonspecificbinding determinedin the pres-ence of5 fLM cAMP was 8—10% of thetotal binding.In parallelexperimentswith autopsied temporalcorti-cal membranes,nonlinear regressionanalysis of thedatawas best fit by a two-site model: ahigh-affinitysite with KD of 0.9 :t 0.1 nM and B,,,,

5 of 154 ±28fmol/mg of protein and alow-affinity site with KD of3.6 ± 0.4nM andBma. of 283 ±36 fmol/mgof protein(n = 3).

To characterizefurther the binding sites labeledby[3H]cAMP, theability of severaldifferent nucleotides

to displacespecific [3H]cAMP binding wasanalyzedin competitionexperiments.Among the variousnude-

Distribution of 113H]cAMP binding in postmortemhuman brain

Cytosolic [3H]cAMP binding varied acrossthebrain regions from 409 to 629fmol/mg of protein incontrol and from298 to 488 fmol/mg of protein in BDsubjects(Table2). [3H]cAMP binding in membranefractionsvariedmore widely, ranging from185 to 51 8fmol/mg of protein in control and from 167 to 430fmol/mgof proteinin BD subjects.Forboth cytosolicand membranefractions,therewas asignificant maineffect of brainregionon [3HI cAMP bindingidentifiedby two-factor ANOVA [cytosolic fraction, F = 3.08,df(5,80); p = 0.013; membranefraction, F = 8.39,df(5,80); p < 0.001] without a significant interactionwith subject group. Pairwise post hoc analysis(Tu-key’s test)of themeandifferencesfor thebrainregionsacrossthe control and BD groups demonstratedsig-nificantly (p = 0.05) higher cytosolic [3H]cAMPbinding only in frontal and occipital cortexcomparedwith temporal cortex. In contrast, membrane [3H]-cAMP binding was significantly(p < 0.05)higher inoccipital and parietal corticalareas(390—518fmol/mg of protein)comparedwith temporal cortex,thala-mus, andcerebellum(170—271 fmol/mg of protein).Forbothcontrol andBD groups, the ratio ofcytosolicto membrane[3H]cAMP binding variedfrom 1.4 to2.6 acrossthe brainregions examinedexcept forpan-etal cortex, for which the ratio ofcytosohic to mem-brane [3H]cAMP binding was —-~0.8.

Specific [3H]cAMP binding in BD brainMeanspecific [3HI cAMP bindingin cytosolicfrac-

tions was significantly lower [F = 6.55, df(l,80); p= 0.012] acrossall brain regions in BD comparedwithmatchedcontrol subjects,with reductionsof 15—23%in the cerebralcortical regions,36% in cerebellarcor-tex, and13% in thahamus(Fig. 2A). As notedabove,therewasno interaction,however, betweendiagnosticgroup and brain region. Post hoc analysis ofsimpleeffectsof diagnosisrevealedthat only the largestre-duction in [3H]cAMP binding in BD cerebellumap-proached statisticalsignificance(p = 0.06)comparedwith controls, whereasthe 23%decrementin the tem-poral cortex did not reachstatistical significance(p= 0.132). In contrast,therewere no significantdiffer-ences [F = 1.41, df(l,80); p > 0.1] in [3HJcAMPbinding in membrane fractions between BD andmatched controlsfor the six brain regionsexamined(Fig. 2B). In bothcontrol and BD groups, therewereno significant correlationsbetween age,postmortemdelay, orbrainpH and [3H]cAMPbindingin cytosolicor membranefractionsfrom any of the brainregionsexamined.

j. Neurochein.,Vol. 68, No. 1, 1997

[3HJcAMP BINDING IN BIPOLAR AFFECTIVE DISORDER 30]

TABLE 2. Regionaldistribution of specific [3HJcAMPbinding in cytosolic andmembranefractionsfrom control andBD postmortembrain

Brain region

Specific [3H]cAMPbinding (fmol/mg of protein)

Cytosolic fraction Membranefraction

Control BD Control BD

Frontal cortexTemporalcortexParietalcortexOccipital cortexCerebellumThalamus

613 ± 62 478 ± 60 (8)”417 ± 48 323 ± 46 (10)409 ± 46 340± 51(5)629 ± 110 488 ± 59 (10)”435 ± 60 298 ± 45 (7)476 ± 92 415 ±25 (6)

439 ± 79 326 ±54 (8)276± 38 272 ±51(10)518 ±15 390 ±34 (5)”430± 5t) 430 ± 60 (10)”240±59 167 ±25 (7)185 ±30 200 ±20 (6)

Dataare mean± SEM valuesfor thenumbersof subjectsindicated in parentheses.[‘HIcAMP(5 nM)was incubatedin triplicatewith cytosolic or membranefractions of various brain regionsasdescribedinMaterialsand Methods,using 5 pM cAMP to definenonspecificbinding.

Statisticaldifferencesbetweenbrain regionswere assessedacrossBD and controlsusing pairwise posthoc analyses(Tukey’s test): °p< 0.05 comparedwith the temporal cortex; 5p < 0.05 comparedwithtemporalcortex,thalamus,andcerebellum.

Relationship of brain lithium levels to [3H IcAMPbinding

In rats receiving chronic lithium treatment, {3H]-cAMP binding showeda trend toward a significantreduction(84± 5% of controlvalues,n = 6;p = 0.06)in cytosolic (controls, 910 ± 42 fmol/mg of protein;lithium-treated,762 ± 47 fmol/mg of protein)butnotmembrane(control, 1,018 ± 68 fmol/mg of protein;lithium-treated,861 ± 63 fmol/mg of protein) frac-tions from prefrontalcortex.Althoughthe plasmalith-ium concentrationsin theseanimals (0.85 ± 0.09mmol/L) did not correlatesignificantly with eithercy-tosolic or membrane[3H]cAMP binding, the reducedcytosohic [1H]cAMP binding in rat prefrontal cortexsuggestedthat antemortemlithium exposuremayhavealtered the levels of cAMP-dPK R subunits.As thiseffect might havecontributed to the lower cytosolic[3H]cAMP binding in BD brain, we determinedwhetherarelationshipexistedbetweenlithium concen-trations and specific [3H]cAMP binding in cytosohicandmembranefractionsfromBD brain.Lithium levelsmeasuredin the BD temporalcortexvariedfrom 0.066to 0.396 mM (Table 1). Furthermore,no significantcorrelationwas observedbetweenlithium concentra-tion andmembraneor cytosolic (Fig. 3) [3H]cAMPbinding in the temporalcortex.

DISCUSSION

The presentstudy demonstrates,for the first time,significantly reduced[3HI cAMP binding in the cyto-solic but not membranefractions from postmortembrain of BD comparedwith control subjectsmatchedon ageand postmortemdelay. As [3HIcAMP bindspredominantly to the R subunits of cAMP-dPK inbrain, the observeddecrementslikely reflect changesinvolving oneor morecAMP-dPK R-subunitisoformsin BD brain comparedwith controls.Thedifferencesin

[3H]cAMP binding betweenBD andcontrol subjectscouldnot be explainedby differencesin age.postmor-tem delayin removaland freezingof brain tissue,oragonalstatusas estimatedby brain pH, nor wasthereany relationship with history of lithium treatmentortissuelithium concentrations.Accordingly, theseob-servationsraisethepossibility that reduced[3H]cAMPbinding in postmortemBD brain may reflect changesin intracellularsignalinglinked eitherdirectly or indi-rectly to the pathophysiologyof this disorder.

The decreasein [3H]cAMP binding in the cytosolicfractions, although small, was evident acrossall BDbrain regionsas supportedby the significant main ef-fectof diagnosticgroupwithoutaninteractionbetweenthe latter factorandbrain region on the ANOVA. De-spitethe significant maineffect, posthoc analysisper-formed to identify specific regions that accountedforthis effect only revealedmarginally statistically sig-nificant reductionsin [3H]cAMP binding for cerebel-lum and temporal cortex, regions that showed thelargestdifferencesin BD comparedwith control sub-jects.The inability to detectstatisticallysignificantre-ductionsacrossa broaderrangeof brain regions,giventhe significant main effect and lack of interactiononANOVA, is likely attributableto the smallsamplesizeandrelatively large varianceof the estimatesof [3H1-cAMP binding obtained.Of note,however,is that themanifestationof the trend to diminished [3H]cAMPbinding also in the cerebellum suggeststhe pro-cess(es)that accountsfor this reductionoccurswidelyin BD brain.

It is unlikely that the reduced[3H]cAMP bindingfound in BD brain is attributableto differencesin ageor postmortemdelay,as BD andcontrol subjectswerematchedon thesevariables,and [3H]cAMP bindingdid notcorrelatesignificantlywith thesefactors.Therewerealso no apparentdifferencesin [3H]cAMP bind-ing betweenBD subjectswho died by suicideversus

.1. “l,’uroch,’,n. , t’ol. 68. No. 1, /997

302 S. RAHMAN ET AL.

FIG. 2. Scatter plots of specific [3H]cAMP binding in (A) cyto-

solic and (B) membrane fractions from postmortem BD brain(.) and matched controls (0): temporal cortex (TCX), frontal(FCX), parietal (PCX), and occipital (OCX) cortices, cerebellarcortex (CBX), and thalamus (THAL). Horizontal bars indicatethe mean values. Two-way ANOVA of cytosolic [3H]cAMP bind-ing data revealed significant main effects of group [F = 6.55,df(1,80); p = 0.012] and region [F = 3.10, df(5,80); p = 0.013]but no interaction [F = 0.31, df(5,80); p > 0.1]. ANOVA ofmembrane [3HJcAMP binding revealed a significant effect of re-gion [F = 8.39, df(5,80); p < 0.0011 but not group [F = 1.41,df(1 80); p > 0.01] and no interaction [F = 0.30, df(5,80); p>0.1].

those who did not. Thus, itis unlikely that suicidalitycontributedto the observeddifferences.AlthoughbrainpH, an indexof the severityof the agonal state(Har-rison et al., 1995), was slightly higher (4%) in BDcomparedwith control postmortembrain, thisparame-ter did notcorrelatewith eithercytosolicor membraneI ~H1cAMPbinding, thus making itunlikely thatdiffer-ences inagonal statusaccount for theobservedreduc-tion in [3H]cAMP bindingin BD brain. Moreover,thedecreasedoes notappearto be related to a generallossof nerve cellsbecause[‘H] cAMP binding in themembranefractionswas notdifferent betweenBD andcontrol subjects.

In the presentstudy, saturationexperimentswerenotperformedowing tolimited availability of postmor-tem BD brain samples.Thus, it wasnot possibletodiscern whetherthe lower [31-I]cAMP binding in BDtemporalcortexis duesolelyto changesin B,,,,,, or alsoinvolves alterationsin K[).

TheBD subjects in thepresentstudyhadtakendrugssuch asneurolepticsand antidepressants,in additionto lithium (seeTable 1). As [3H1cAMP binding is

not affectedby neuroleptics(Nishino et al., 1993) orlithium (up to 1 mM) in vitro (S. Rahrnan etal..unpublisheddata), it is unlikely that residual drug inpostmortem brain accounts for the observedchanges.butthe contributionof long-termdrug treatmentto thereductionsin [3HIcAMP binding in BD brain cannotbe completelyruled out. However,chronicantidepres-sant administration in rats actually increasedmem-brane cerebral cortex cAMP binding (Perezet al..1989, 1991). Although [3H1cAMP binding, in thisstudy, was moderatelyreducedin the cytosolic butnotmembranefractions fromprefrontal cortexin rats thatreceivedlithium chronically, thedecreasesdid notcor-relate significantly with plasmalithium levels. Further-more, we did not observe a significant relationshipbetweencytosolic [3HI cAMP binding and lithiumcon-centrationin BD temporalcortex, and the lithiumlev-els weresubtherapeuticbasedon in vivo estimatesinbrain (0.3—0.9 mM) of BD patientson maintenancelithium therapy(Kato et al., 1992;Sachset al.. 1995).Takentogether, thesefindings suggest reduced [3H]-cAMP binding in BD brain is not likely related toantemortern lithium or antidepressanttreatment, al-though effectsof other maintenancemedicationscan-not be ruled out.

Theextent to which [‘H]cAMP bindingreflectsspe-cific labeling of the R subunits ofcAMP-dPK in post-mortem humanbrain fractions is a crucial questionunderpinningthe attribution of the observed differ-ences in binding to changes in the levels of specificcAMP-dPK R-subunitisoforms. In line with previousobservations(Nishino et al., 1993), [3H]cAMP la-beled a homogeneous populationof sites in cytosolicfractionsat least in therepresentativetemporal cortexregion in which the binding kinetics were examined.Furthermore,the results ofcompetition experimentswith nucleotide analoguessupport the specificity of[3H]cAMP binding to a cAMP binding protein(s).That this binding occurs primarily to cAMP-dPK Rsubunitsin brain issupportedby the findingthat cAMP

FIG. 3. Scatter plot ofcytosolic [3H]cAMP binding versus lithiumlevels in BD temporal cortex. The hatched line indicates thecalculated regression line. No significant relationship was foundbetween cytosolic [3H]cAMP binding and lithium concentrationsin this region (r = 0.01 ‘p > 0.1; Pearson Product Momentcorre-lation analysis).

J. N,’urochem., Vol. 68, No. I, 1997

[3HJcAMP BINDING IN BIPOLAR AFFECTIVE DISORDER 303

reduced the photoactivatedincorporation of 8-N3-

[32PIcAMP into RI and RH subunits in rat brain by90 and94%, respectively(Walter et al., 1978).Fur-thermore,Stein et al. (1987)also showeda highper-centageof cAMP binding (56—84%)to RIl subunitsin membraneandcytosol from neurons,astrocytes,andoligodendrogliafractionatedfrom rat brain.

In the membranefractions from temporal cortex,[3H]cAMP binding wasbest describedby a two-sitemodel with estimatedKD values of 0.9 and 3.6 nM,respectively. Other investigatorshave also reportedheterogeneous[1H]cAMP binding sites in membranefractions of bovine myocardiumand skeletal muscle(DoskelandandOgreid, 1981;OgreidandDoskeland,1981; Ogreid et al., 1983; Doskelandet al., 1993).WhereascAMP bindsnonselectivelyto typeI or typeII cAMP-dPK, it is unlikely that differential localiza-tion of R-subunit subtypes in cellular fractions ac-counts for the observeddifference in binding kineticsbetweenthe two fractions.Both RI and Rh subtypesareexpressedin cytosolicfractionsof ratbrain (Walteret al., 1978),yet cytosolic [3HIcAMP binding in hu-man temporal cortex is characterizedby single-sitebinding kinetics as shown in this study and reportedby others (Nishino et al., 1993). A more plausibleexplanationfor the two classesof [3HIcAMP bindingsitecharacterizedin themembranefractions,however,is that they may representstepwisebinding of cAMPto the R subunits in a positive cooperativemanner(Doskelandand Ogreid, 1981; Ogreidet al., 1989).

The distribution of [3H 1 cAMP binding in cytosolicand membranefractionsvariedacrossthe brain areasexamined,althoughtheregionaldifferencesweremoremarkedin the latter fractions. Cytosolic [3H]cAMPbinding was highestin occipital and frontal cortices,in agreementwith thefindingsof Nishinoet al. (1993),whereasin themembranefractionsbindingwashighestin occipital andparietalcortices,moderatein temporalcortexandcerebellum,andlowestin thalamus.Greater[3H]cAMP binding in the cytosolic fractions (1.4—2.6-fold) than in correspondingmembranefractionsalso concurswith earlier findings of enrichmentofcAMP-dPKR subunitsin the cytosolicfractionsof rat,guineapig, and bovine cerebralcortex (Walter et al.,1978).

Among possiblemechanismsthatmight accountforthereducedcytosolic [3H]cAMP binding in BD brain,altered synthesisor degradationof cAMP-dPK R sub-units merits further consideration.It is unlikely, how-ever, that differential redistributionof cAMP-dPK Rsubunits occurs between cytosolic and membranepools in BD in contrastto control brain, becausespe-cific [3H]cAMP bindingwasnotsignificantlydifferentin the membranefractionsof BD comparedwith con-trol brain.

Someevidencesuggeststhat the loss of R subunitsmay impact on regulatoryfunctionsin the cAMP sig-naling cascade.In Aplysia, reducedlevels of R sub-units, found after treatmentsthat producelong-term

sensitization,may result in increasedkinase activityandenhancedprotein phosphorylationat subsaturatingcAMP concentrations(Greenberget al., 1987)conse-quentto a decreasedR to C subunit ratio. Consonantwith this notion, Perezet al. (1995) recentlyreportedincreasedcAMP-dependentendogenousphosphoryla-tion inplateletsfrom euthymicbipolarpatients.Furtherstudies arerequiredto determinewhethersucha rela-tive reductionin R subunits,accompaniedby alteredcAMP-dPK activity andcAMP-dependentendogenousphosphorylation,occurs in BD brain. Moreover, fromapathophysiologicalperspective,it remainsto bedem-onstratedwhetherchangesin cAMP-dPK function areprimary disease-relatedchangesspecific to BD or re-flect secondaryadaptiveresponsesconsequentto up-streamalterationsin processesmodulatingcAMP lev-elsand,if so, whethertheyafford anyprotectiveadvan-tage in this disorder. Finally, in light of the resultsshowingmoderatereductionof cytosolic [3H]cAMPbinding in prefrontal cortex of rats receiving lithiumchronically, the possibility that antemortemlithiumtreatmentmay, in part, contributeto theabovechangesalso cannotbe excluded.

Regardlessof the mechanismunderlyingthe lossofR subunitof cAMP-dPK in BD brain, the presentre-sults, along with the growing body of evidenceof al-teredsecondmessengerand signal transductionpro-cessesin brain and peripheraltissuesof BD patients(Young et al., 1991, 1993, 1994; Perezet al., 1995;Mathewset al., 1996;Jopeet al., 1996;WarshandLi,1996), support the hypothesisthat abnormalitiesinsignaltransductionmechanismsplay an importantrolein the pathophysiologyof this major psychiatricdisor-der.

Acknowledgment: This work was supportedby a grantfrom the Medical ResearchCouncil of Canada(to J.J.W.and L.T.Y.). S.R. is a postdoctoralfellow supportedby theClarke Instituteof PsychiatryResearchFoundation.We aregrateful to Dr. J. Forresterof Novamann International forbrain lithium content determination,Ms. Kathleen ShanakandMs. Kin Po Siu for autopsiedbrain preparation,andMs.CathySpegg for her advice on statisticalanalyses.

REFERENCES

AmericanPsychiatricAssociation(1987)DiagnosticandStatisticalManualof MentalDisorder,,,3rd edit., revised.American Psy-chiatric Press,Washington,D.C.

Avissar S. and SchreiberG. (1992) The involvement of guaninenucleotide binding proteinsin the pathogenesisand treatmentof affectivedisorders.Biol. Psychiatry31, 435—459.

BradfordM. M. (1976)A rapid andsensitivemethodfor thequanti-tationof microgramquantitiesof proteinutilizing the principleof protein—dyebinding.Anal. Biochem.72, 248—252.

Budillon A., CresetoA., Kondrashin A., NesterovaM., Merlo G.,Clair T., and Cho-ChungY. S. (1995) Point mutation of theautophosphorylationsite or in thenuclearlocationsignal causesproteinkinase A Rll

3 regulatorysubunit to lose its ability torevert transformedfibroblasts.Proc. NatI. Acad. Sd. USA 92,10634—10638.

Cho-Chung Y. S. and Clair T. (1993) The regulatory subunit ofcAMP-dependentprotein kinaseas a target for chemotherapy

J. Neurochem.,Vol. 68, No. 1, 1997

304 S. RAHMANET AL.

of cancerandothercellulardysfunctional-relateddisease.Phar-macal. Ther. 60, 265—288.

DtskelandS. 0. and Ogreid D. (1981)Binding proteinsfor cyclicAMP in mammaliantissues.ml. J. Biochem. 13, 1— 19.

Doskcland5. 0., MarondeE., andGjertsenB. T. (1993)Thegeneticsubtypesof cAMP-dependentproteinkinase—functionallydif-ferent or redundant.Biochim. Biophy.c.Acta 1178,249—258.

Francis S. H. and Corbin J. D. (1994) Structureand function ofcyclic nucleotide-dependentprotein kinases.Annu. Rev. Phys-iol. 56, 237—272.

Garrel G., Delahaye R.. Hemmings B. A.. and Counis R. (1995)Modulation of regulatoryandcatalyticsubunitlevelsof cAMP-dependentproteinkinaseA in anteriorpituitary cells in responseto direct activationof protein kinasesA andC or afterGnRHstimulation.Neuroendocrinology62, 514—522.

GreenbergS. M., CastellucciV. F., Bayley H., and SchwartzJ. H.I 987) A molecularmechanismfor long—term sensitizationin

Aplvsia. Nature 329, 62—65.HarrisonP. J.,HeathP. R., EastwoodS. L.,BurnetP. W. J.,McDon-

ald B., and PearsonR. C. A. (1995)The relativeimportanceofpremortemacidosisand postmorteminterval for humanbraingene expression studies: selective mRNA vulnerability andcomparisonwith their encodedproteins. Neuro,sci. Let!. 200,151— 154.

HudsonC.J., YoungL. T., Li P. P., and Warsh J. J. (1993)CNSsignal transductionin the pathophysiologyand pharmacother-apy of affectivedisordersand schizophrenia.Synapse13, 278—293.

JonesC.L., Hodge V. F.,SchoengoldD. M., BiesiadaH., andStarksT. H. ( I 987)An Inierlahorators’ Studyof Inductively CoupledPlasmaAtomicEmissionSpectro.cc’ops’Method6010and Dige.v-lion Method 3050. Publication no. EPA-600/4-87-032,U.S.EnvironmentalProtectionAgency,Las Vegas.Nevada.

JonesD. J. and StavinohaW. B. (1979) Microwaveinactivationasatool for studyingtheneuropharmacologyofcyclic nucleotides,in Neuropharmadologyof Cyclic Nucleotides (Palmer G. C.,ed), pp. 253—281. Urbanand Schwarzenbcrg,Munich.

JopeR. S.,SongL., Li P. P., YoungL. T., Kish S. J., PachecoM. A.,and WarshJ.J. (1996) The phosphoinositidesignal transduc-tion systemis impaired in bipolar affective disorderbrain. J.Neurochem.66, 2402—2409.

Kato T., TakahashiS.,andInubushiT. (1992)Brain lithium concen-trationby ‘Li-and- ‘H-magneticresonancespectroscopyin bipo-lar disorder.Psychiatry Re,,. 45, 53—63.

Licameli V., MattiaceL. A., Erlichman J., Davies P., Dickson D.,andShalit-ZagardoB. (1992)Regionallocalizationof theregu-latory subunit (R%8) of the type II cAMP-dependentproteinkinase in humanbrain. Brain Res. 578,61—68.

Liu A. Y.-C., Chan T., and Chen K. Y. (1981) Induction of theregulatory subunitof type I adenosinecyclic 3 ‘,S ‘-monophos-phate-dependentprotein kinasein differentiatedN- 18 mouseneuroblastomacells. CancerRes.41, 4579—4587.

Manji H. K. (1992)G proteins: implicationsfor psychiatry.Am. J.Psychiatry 149, 746-760.

MathewsR., Li P. P., YoungL. T., Kish S. J., andWarshJ. J. (1996)IncreasedGa,,,,

1immunoreactivityin postmortemoccipitalcor-tex from patientswith bipolaraffectivedisorder.Biol. P.svchia-try (in press).

McPhersonG. A. (1985) Analysis of radioligandbinding experi-ments:a collection of computerprogramsfor the IBM PC. I.Pharomacol.Methods 14, 213—218.

Nishino N., Kitamura N., HashimotoT., Kajimoto Y., Shirai Y.,Murakami N., Nakai T., Osamu K., Shirakawa0., Mita T.,and Nakai H. (1993)Increasein [‘HI cAMP binding sitesanddecreasein Ga and G,,a immunoreactivitiesin left temporal

corticesfrom patientswith schizophrenia.Brain Rca. 615, 41 —

49.Ogreid D. and DoskelandS. 0. (1981)The kinetics of association

of cyclic AMP to thetwo typesof binding sitesassociatedwithprotein kinase 11 from bovine myocardium. FEBS Lelt. 129,287—292.

Ogreid D., DoskelandS. 0., and Miller J. P. (1983) Evidencethatcyclic nucleotidesactivatingrabbit muscleprotein kinaseI in-teractwith both typesof cAMP binding sites associatedwiththe enzyme.J. Biol. Chem. 258, 1041—1049.

OgreidD., EkangerR., SuvaR. H.. Miller J. P., andDoskelandS. 0.(1989) Comparisonof the two classesof binding sites (A andB) of typeI and type II cyclic-AMP dependentprotein kinascsby using cyclic nucleotide analogues.Furl. Biocheoi. 181,19—31.

PrashadN. and RosenbergR. N. (1978)Induction of cyclic AMP-binding proteins by dibutyryl cyclic AMP in mouseneuro-blastomacells. Biochim. Biophy.s.Ac/a 539, 459—469.

PerezJ., Tinelli D., Brunello N.. and Racagni G. (1989) cAMPdependentphosphorylationof solubleandmicrotubulefractionafter prolongedDM1 treatmentin rat cerebral cortex.Liii. I.Phar,nacol. 172, 305—316.

PerezJ., Tinelli D., BianchiE.. Brunello N., andRacagniG. (19911cAMP binding proteinsin the rat cerebralcortexafteradminis-trationof selective5-HT andNE reuptakeblockerswith antide-pressantactivity. Neurop.vvchophormacology4, 57—64.

PerezJ.,ZanardiR.,Mon S.,GasperiniM., SmeraldiE., andRacagniG. (1995) Abnormalities of cAMP-dependentendogenousphosphorylationin plateletsfrom patientswith bipolardisorder.Am. J. Psychiatry 152, 1204—1206.

SachsG. S., RenshawP. F., Lafer B ., .Stoll A. L., GuimaraesA. R..RosenbaumJ. F., and Gonzalez R. G. (1995) Variability of’brain lithium levels duringmaintenancetreatment:a magneticresonancespectroscopystudy. Biol. P.sychiatr,’ 38, 422—428.

SampsonM., RuddelM., and Elm R. J. (1994) Lithium determina-tions evaluatedin eight analyzers.Clin. Chem.40, 869—872.

SchreiberG., Avissar S., Danon A., and Belmaker R. H. (1991HyperfunctionalG proteinsin mononuclearleukocytesof pa-tientswith mania.Biol. Psychiatry29, 273—280.

SpauldingS. W. (1993) The waysin which hormoneschangecyclicadenosine3 ‘,S ‘-monophosphate-dependentprotein kinasesub-units,and how suchchangesaffect cell behavior.Endocr. Re,’.14, 632—650.

SteinJ. C., FarooqM., NortonW. T., andRubinC. S. (1987)Differ-ential expressionof isoforms of theregulatorysubunit of typeII cAMP-dependentprotein kinasein rat neurons,astrocytesand oligodendrocytes.J. Biol. Chemn.262,3002—3006.

WalaasS. I. and GreengardP. (1991)Protein phosphorylationandneuronalfunction. Pharmacol.Rev. 43, 299—349.

WalterU., KanofP.,SchulmanI-I., andGreengardP. (1978)Adeno-sine 3.5 ‘-monophosphatereceptor proteins in mammalianbrain. J. Biol. Chem. 253,6275—6280.

WarshJ. J. andLi P. P. (1996)Secondmessengersystemsandmooddisorders.Curr. Opin. Psychiatry 9, 23—29.

Young L. T.. Li P. P., Kish S. J., Siu K. P., and WarshJ. J. (1991)PostmortemcerebralcortexG, a subunitlevels areelevatedinbipolaraffectivedisorder.Brain Re,,. 553, 323—326.

Young L. T., Li P. P., Kish S. J., Siu K. P., Kamble A., Hornykie-wicz 0.. and WarshJ. J. (1993) Cerebral cortex G,a proteinlevelsand forskolin-stimulatedcyclic AMP formation are in-creasedin bipolar affectivedisorder. J. Neurochem.61, 890—898.

YoungL. T., Li P. P., Kamble A., Siu K. P., andWarshJ.J. (1994)Mononuclearleukocyte levels of G proteinsin depressedpa-tientswith bipolaror majordepressivedisorder.Am. J. P,s’s’chia-try 151, 594—596.

J, N,’,,,’o,’hcn,., Vol. 68, No. I, /997