Proc. NatL Acad. Sci. USAVol. 80, pp. 3531-3535, June 1983Neurobiology

Isolation, characterization, and purification to homogeneity of anendogenous polypeptide with agonistic action onbenzodiazepine receptors

(endogenous benzodiazepine ligand/GABA receptor/anxiety/partial peptide sequence)

ALESSANDRO GUIDOTTI, CONCETTA M. FORCHETTI, MARIA G. CORDA, DENNIS KONKEL,CARL D. BENNETT*, AND ERMINIO COSTANational Institute of Mental Health, Saint Elizabeths Hospital, Washington, D.C. 20032; and *Department of Medicinal Chemistry, Merck Sharp & Dohme ResearchLaboratories, West Point, Pennsylvania 19486

Contributed by Erminio Costa, February 28, 1983

ABSTRACT A brain polypeptide termed diazepam-bindinginhibitor (DBI) and thought to be chemically and functionally re-lated to the endogenous effector of the benzodiazepine recogni-tion site was purified to homogeneity. This peptide gives a singleband of protein on NaDodSO4 and acidic urea gel electrophoresis.A single UV-absorbing peak was obtained by HPLC using threedifferent columns and solvent systems. DBI has a molecular massof 411,000 daltons. Carboxyl-terminus analysis shows that tyro-sine is the only residue while the amino-terminus was blocked.Cyanogen bromide treatment of DBI yields three polypeptidefragments, and the sequences of two of them have been deter-mined for a total of 45 amino acids. DBI is a competitive inhibitorfor the binding of [3H]diazepam, [3H]flunitrazepam, 1-[3HIcar-boline propyl esters, and 3H-labeled Ro 15-1788. The K, for [3H]-diazepam and 3[3H]carboline binding were 4 and 1 ,uM, re-spectively. Doses of DBI that inhibited[3H]diazepam binding by>50% fail to change [3H]etorphine, y-amino[3H]butyricacid [3H]-quinuclidinyl benzilate, pH]dihydroalprenolol, [3H]adenosine, and[3H]imipramine binding tested at their respective Kd values. DBIinjected intraventricularly at doses of 5-10 nmol completely re-versed the anticonflict action of diazepam on unpunished drinkingand, similar to the anxiety-inducing 1-carboline derivative FG 7142(13-carboline-3-carboxylic acid methyl ester), facilitated the shock-induced suppression of drinking by lowering the threshold for thisresponse.

Specific high-affinity binding sites for benzodiazepines are lo-cated in synaptic membranes prepared from various regions ofmammalian central nervous system (1-3). In various species,man included, the occupation of benzodiazepine recognitionsites by specific ligands such as diazepam or ,B-carbolines in-duces behavioral changes that oscillate from the relief to theinduction of anxiety states characterized by fear of punishment(4-6). For instance, in rats diazepam elicits anticonflict andanxiety-abating responses (4, 5) whereas a number of [-car-boline derivatives elicit proconflict and anxiety-inducing effects(7). In line with reports (8) proposing that ligands of benzo-diazepine recognition sites modulate y-aminobutyric acid (GABA)receptors, it has been shown that, in vitro, some ligands downregulate and others up regulate GABA recognition sites locatedin crude synaptic membrane preparations (6). The down reg-ulation has predictive value for the in vivo proconflict action,whereas the up regulation predicts anticonflict action (7).

Thus, it can be proposed that benzodiazepine recognitionsites are modulatory sites of GABA receptors and it can be an-ticipated that the benzodiazepine sites are physiologically oc-

cupied by endogenous modulators that, by tuning GABA re-ceptors, elicit a spectrum of behavioral states ranging fromanxiety to sedation. Several brain components have been sug-gested to function as physiological ligands for the benzodiaze-pine-GABA receptor system (9-16). However, for most of thesesubstances, the evidence that they may act as neuromodulatorsof GABA receptors is lacking. This report describes the iso-lation and purification to homogeneity of a rat brain peptidethat competitively inhibits benzodiazepine binding to their rec-ognition sites. This peptide was termed DBI (diazepam-bindinginhibitor). When injected intraventricularly, this peptide blocksthe anticonflict effect of benzodiazepines and facilitates the be-havioral suppression induced by punishment. This action of DBIis blocked by previous treatment with Ro 15-1788, a ligand ofbenzodiazepine recognition sites devoid of intrinsic activity (17).DBI antagonizes the benzodiazepine-elicited increase in themaximal binding value of GABA to its high-affinity recognitionsite.

MATERIALS AND METHODSAnimals. Male Sprague-Dawley rats (175-200 g) were ob-

tained from Zivic-Miller (Allison Park, PA).Materials. Sephadex G-100 and G-75 were purchased from

Pharmacia. Bio-Gel P-2 and a Bio-Sil ODS-10 reversed-phaseHPLC column were obtained from Bio-Rad. The Zorbax CNHPLC column was from DuPont and the Synchropak AX300anion exchange HPLC column was from Synchrome (Linden,IN); trypsin (ketone treated) and chymotrypsin were from Wor-thington; Pronase and papaine were from Sigma; FG 7142 (,-carboline-3-carboxylic acid methyl ester), ,B-carboline-3-car-boxylic ethyl ester, and ,B-[3H]carboline-3-carboxylic propylester were a gift from C. Braestrup (Ferrosan, Copenhagen,Denmark); diazepam, Ro 15-1788 (ethyl 8-fluoro-5,6-dihydro-5-methyl-6-oxo-4H-imidazol[1,5-a]-[1,4] benzodiazepine-3-carboxylate), and 3H-labeled Ro 15-1788 were gifts from H.Mohler (Hoffman-LaRoche). Pyroglutamate aminopeptidase wasobtained from Boehringer Mannheim. All the radioactive re-ceptor ligands were purchased from New England Nuclear.HPLC. Analysis of peptides was carried out by using a Spec-

trophysics liquid chromatography system equipped with aSpectro-Physics model 8700 or 8000B solvent delivery system.Peptides were detected by monitoring absorbance at 210 nm orby colorimetric reaction using the Lowry method (18).

Polyacrylamide Slab Gel Electrophoresis. An apparatusmanufactured by Bio-Rad (model 220) was used in these ex-

Abbreviations: GABA, y-aminobutyric acid; DBI, diazapam-binding in-hibitor; QNB, quinuclidinyl benzilate.

3531

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertise-ment" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

3532 Neurobiology: Guidotti et aL

periments with polyacrylamide gels 2.8 mm thick. Urea/acidpH (pH 3.2) gels were prepared according to the procedure ofPanyim and Chalkley (19). NaDodSO4 gels were run accordingto Laemmli and Favre (20) with minor modifications. TheNaDodSO4 gel was calibrated using the "low molecular weight"protein calibration kit from Pharmacia. Proteins were stainedwith Coomassie brilliant blue G.

Bioassay for DBI-Like Activity. This assay measures the abilityof tissue extracts and purified proteins to competitively inhibitthe binding of [3H]diazepam or other 3H-labeled benzodiaze-pine ligands to Triton X-100 (0.05%)-treated crude synapticmembranes (13). Triton X-100 was washed from the mem-branes before the assay. The assay (final vol, 1 ml) was carriedout in 20 mM potassium phosphate buffer (pH 7.0). Compo-nents of the mixture were added to the reaction tubes in thefollowing order: suspension of synaptic membranes (0.3-0.4mg of protein), tissue extract to be tested for DBI-like activityor control buffer (10-100 1ul), and [3H]benzodiazepine at var-ious concentrations (in 50 A.l of buffer). Nonspecific binding(usually 10% of total binding) was determined by adding 5 UMdiazepam. After 30 min of incubation at 2-4TC, the reactionwas terminated by rapid filtration on GFC filters (13). The ef-fect of DBI on [3H]etorphine (21), [3H]adenosine (22). [3H]imip-ramine (23), 3-[3H]quinucidinyl benzilate (24),-[ H]dihydro-alprenolol (25), and [3H]GABA (13) binding was studied by pre-viously published methods using a concentration of the ligandin the Kd range.

Stimulation of [3H]GABA Binding by Diazepam. The di-azepam-induced increase of [3H]GABA binding was studied infrozen-thawed and three-times-washed cortical synaptic mem-branes using 20 nM [3H]GABA in 20 nM KPO4/50 mM KC1,pH 7. Diazepam was incubated for 10 min at 2-4°C with themembranes prior to addition of the radioactive ligand. The re-action was stopped by centrifugation 10 min later (13).DBI Iodination. DBI was iodinated with '2I using the Bol-

ton-Hunter reagent. Briefly, 10 ,ug of protein was treated for1 hr with 250 ,uCi of Bolton-Hunter reagent (1 Ci = 37 GBq)at 0°C. The reaction was stopped by the addition of 200 /4 of1 M acetic acid, and the sample was applied to a 0.7 x 10 cmBio-Gel P-2 column equilibrated with 0.1 M acetic acid. Thepurity of the iodinated material was tested by both NaDodSO4/polyacrylamide gel electrophoresis and reversed-phase HPLC.

Analysis of Amino Acid Composition of Peptides. The pro-tein was hydrolyzed with 6 M HCl for 20 hr at 110°C at reducedpressure. Analysis was performed by a Beckman 121MB single-column system with ninhydrin detection (26).

Peptide Cleavage by CNBr and Separation of Peptide Frag-ments. The cleavage of 100 ,ug of the protein by CNBr was car-ried out in 70% formic acid. The sequence of the resulting mix-ture was determined (see below). An additional 150 ,ug wascleaved with CNBr and the mixture was separated by HPLCon a 4.5 X 30 cm Watersc,8 column. The initial solvent was0.02% trifluoroacetic acid in water and the peptides were elutedwith a linear gradient from 100% of 0.02% trifluoroacetic acidin water to 50% acetonitrile in water over 30 min. Peptides wereidentified by their absorption at 210 nm, collected, and lyophi-lized. The sequence of the only single peptide (see Discussion),which eluted at 15.6 min, was determined.Amino Acid Sequence Determination. Sequences were de-

termined by using a Beckman model 890C Sequencer (spinningcup), slightly modified, with 4 mg of Polybrene in the cup (27).The phenylthiohydantoin amino acid derivatives were sepa-rated and quantitated by HPLC using a Zorbax CN column.

Pyroglutamate Aminopeptidase Digestion. The method ofCastel et aL (28) was used. After 24 hr of digestion, no NH2-terminal amino acid could be detected in the Sequencer.

Evaluation of Proconflict and Anticonflict Activities. Theseexperiments were conducted in rats kept without water for 72hr prior to the conflict session. Apparatus and other experi-mental conditions were identical to those reported previously(7). A chronic intraventricular polyethylene cannula was im-planted in the left lateral brain ventricle (29) 3 days before theexperiment.To study the "anticonflict" action, the intensity of the cur-

rent delivered through the drinking tube was 1 mA for a du-ration of 0.5 sec. To study the "proconflict" action, the shockduration was left constant but the intensity was decreased to0.25 mA. To study the effect of the tested compound on thirst(unpunished drinking), the shock was omitted.

At the time of the test, each rat was allowed to habituate tothe chamber without drinking for 15 min. At the end of thisperiod, the animals received an intracerebroventricular injec-tion of DBI or vehicle. After 3 min, the drinking tube was in-serted into the cage and the animals were allowed to lick for 3sec (one "licking period') before the first shock was delivered.A 3-min test period started at the end of the first shock.

RESULTSExtraction and Purification of DBI. Typically, 20 rat brains

were used for each preparation. After decapitation of the rat,the brain was rapidly (15-30 sec) removed and homogenized in10 vol of hot (800C) 1 M acetic acid. The homogenate was cen-trifuged at 48,000 X g for 20 min at 2-40C, and the supernatantwas adjusted to pH 5 and centrifuged under identical condi-tions. This second supernatant was Iyophilized, resuspended in10 ml of 0.1 M acetic acid and centrifuged again at 48,000 xg, and this supernatant was applied to a 2.5 x 70 cm SephadexG-100 column previously equilibrated with 0.1 M acetic acid.The column was eluted with 0.1 M acetic acid at a flow rate of0.2 ml/min, and 5-ml fractions were collected. Each fractionwas Iyophilized and resuspended in 1 ml of water, and aliquotsof 10-100 1A were used for monitoring DBI-like activity andprotein content. Proteins with DBI-like activity emerged fromthis column in the eluate included between the elution posi-tions of ribonuclease A (Mr, 13,700) and 2-mercaptoethanol (Mr,78). The fractions containing DBI activity were pooled, Iyophi-lized, resuspended in 10 ml of 0.1 M acetic acid, and rechro-matographed on a 2.5 x 70 cm Sephadex G-75 column. Thiscolumn was eluted with 0.1 M acetic acid at a flow rate of 0.2ml/min, and fractions of 5 ml were collected. Again, DBI ac-tivity emerged with the eluate included between markers of100 and 14,000 daltons. The fractions containing DBI were ly-ophilized to dryness, resuspended in 10 ml of 0.1 M acetic acid,and treated with 60% saturated ammonium sulfate. The 48,000X g supernatant was desalted on a 0.7 X 10 cm Bio-Gel P-2column previously equilibrated with 0.1 M acetic acid. DBIemerged from this column with the void volume well separatedfrom the bulk of salts.

After lyophilization, the residue was taken up in 1-2 ml of1 M acetic acid. Aliquots containing 100-200 1g of protein wereapplied to a Bio-Sil ODS-10 (Bio-Rad) reversed-phase HPLCcolumn previously equilibrated with 0.1 M NaH2PO4/0. 1%H3PO4, pH 2.5, and eluted with an acetonitrile gradient of 0%to 60% (Fig. 1A) and then applied to a Bio-Sil ODS-10 re-versed-phase HPLC column previously equilibrated with 0.1%trifluoroacetic acid in H20 and eluted with a 0-70% gradientof 0.1% trifluoroacetic acid/acetonitrile (Fig. 1B). DBI-likematerial eluted from the second HPLC step (Fig. 1B) was ex-amined for purity using different HPLC and acrylamide gelelectrophoretic systems.

As shown in Fig. 2, 50 Aug of protein gave a single UV-ab-

Proc. Nad Acad. Sci. USA 80 (1983)

Proc. Natd Acad. Sci. USA 80 (1983) 3533

c-

a

A

0 15 30 0 15 30Volume, ml

FIG. 1. HPLC procedure for DBI purification. (A) Reversed-phaseHPLC of250 .g ofprotein with DBI-like activity. The sample was firstpurified by chromatography on Sephadex G-100 and G-75, ammoniumsulfate precipitation, and Bio-Gel P-2 chromatography (see Table 1).Conditions were column, Bio-Sil ODS-10 (4 x 250 mm); flow rate, 1 ml/min; temperature, 20°C. After sample application, the column waswashed for 15 min with the starting buffer (0.1 M Na H2P04/0.1% MH3PO4, pH 2.5). Peptides were eluted with a gradient from 100% start-ing buffer to 60% acetonitrile/40% starting buffer. Fractions of 1 mlwere collected, neutralized, lyophilized, and analyzed for DBI activity.(B) DBI obtained fromA was reapplied to the Bio-Sil ODS-10 column,previously equilibrated with 0.1% trifluoroacetic acid in H20, and de-veloped with a 35-min linear gradient of0.1% trifluoroacetic acid in 0-70% acetonitrile. The flow was 1 ml/min. After lyophilization, frac-tions were analyzed forDBI activity, DBI activity is associated with theUV-absorbing peak.

+

B

100 50 25 10 GM

few go

G)C.

100 50 20 Mr

- 9467

- 43- 30

20.1

ADM A*as 14.4

4 6+

FIG. 3. Polyacrylamide gel electrophoresis of DBI obtained fromreversed-phase HPLC (Fig. 1B). (A) Urea (6.25 M)/15% polyacryla-mide gel, pH 3.2. Amounts ofprotein applied are given in ug. Lane GM:GABA-modulin (30) was run for comparison. Protein moved toward thecathode. Coomassie brilliant blue G was used to stain the proteins. (B)NaDodSO4/17% polyacrylamide gel. Lane MAr: molecular weight mark-ers (x 10-3) were phosphorylase B, 94,000; bovine serum albumin, 67,000;ovalbumin, 43,000; carbonic anhydrase, 30,000; soybean trypsin in-hibitor, 21,000; a-lactalbumin, 14,000; insulin, 6,000.

a 15% acidic urea gel (Fig. 3) produced a single band stainingwith Coomassie blue.An average preparation from 20 brains yielded 20-25 jug of

DBI. By adding 2 ,ug of 125I-labeled DBI (specific activity, 5X 103 cpm/ng) to the initial homogenate, we established thatthe recovery was 1.5-2%.DBI Characterization. The amino acid composition of DBI

is shown in Table 1. The protein contains 105 amino acid res-idues and basic residues, particularly lysine, are relativelyabundant. As expected, DBI biological activity was abolishedby incubation with trypsin, chymotrypsin, or papaine.

Initial sequence determination of DBI showed that the aminoterminus is blocked. The blocking group was not removed bydigestion with pyroglutamite aminopeptidase.

Treatment of the protein with CNBr produced two amino

sorbing peak under the HPLC conditions selected. Further-more, the application of various amounts of protein (10-100,ug) to a 12% NaDodSO4 gel, a 17% NaDodSO4 gel (Fig. 3), or

;a

060 =

-40S040

20 tR

Volume, ml

FIG. 2. HPLC analysis of purified DBI. DBI obtained from re-versed-phase HPLC (Fig. 1B) was tested for purity under three con-ditions. (A) Column, Synchropak AX300 oxiranes anion exchange resinequilibrated with 0.02 M Na2HPO4, pH 8.3; flow rate 1 ml/min. DBIeluted with the void volume. No other proteins were eluted when thecolumn (at 60 min) was eluted with a linear gradient from 0.02 MNa2HPO4 to 0.2 M NaHPO4/1 M NaClO4, pH 8.3 (----). (B) Column,Bio-Sil ODS-10 equilibrated with 0.1 M NaH2PO4/0.1% H3PO4/30%acetonitrile under isocratic conditions; flow rate, 1 ml/min. (C) Col-umn, Zorbax CN equilibrated with 0.1 M NaH2PO4/0.1% H3PO4, pH2.5; elution, 60-min linear gradient (----) of 0-60% acetonitrile. Eachcolumn was charged with 50 pg of DBI.

Table 1. Amino acid composition of DBI purified from rat brain

ResidueAsparagine/aspartateThreonineSerineGlutamate/glutamineProlineGlycineAlanineCysteineValineMethionineIsoleucineLeucineTyrosineHistidinePhenylanineLysineArginineTryptophan

Total

Relative amount,number of residues

1071012368042374331931

105

The NH2 terminus is blocked. No sequence could be determined usingthe native preparation. CNBr treatment produced three fragments: 1,(Met)-Leu-Phe-Ile-Tyr-Ser-His-Phe-Lys-Gln-Ala-Thr-Val-Gly-Asp-Val-Asn-Thr-Asp-Arg-Pro-Gly-Leu-Leu-Asp-Leu-Lys-Gly-Lys-X-Ile; 2,(Met)-Lys-Thr-Tyr-Val-Glu-Lys-Val-Glu-Glu-Leu-Lys-Lys-Lys-Tyr.Fragment 2 is the carboxyl-terminal fragment.

Neurobiology: Guidotti et aL

I0

:

3534 Neurobiology: Guidotti etalP

termini. Sequence determination of the mixture proceeded for30 steps with one sequence terminating abruptly at step 14. Asecond sample was cleaved with CNBr and separated on HPLC.Under the conditions of separation,- peptides containing ho-moserine elute as pairs of peptides 4 min-apart, due to partialhomoserine lactone formation. The sequence of the only singlepeptide eluted was determined. It contained 14 amino acidsand had a COOH-terminal tyrosine. Since it did not containhomoserine, it must be the carboxyl-terminal peptide in theprotein. With this information, a partial sequence could thenbe assigned to the other (middle) peptide (Table 1).The molecular weight of DBI was estimated to be approx-

imately 10,000 by NaDodSO4 gel electrophoresis and was cal-culated on the basis of amino acid composition to be 11,700.

Binding Studies. DBI is a competitive inhibitor for the bind-ing of [3H]diazepam, [3H]flunitrazepan, f3-[3H]carboline ethylester, and 3H-labeled Ro 15-1788. The Ki for [3H]diazepambinding was approximately 4 AtM and that for ,&[3H]carbolineethyl ester, 1 AM. Doses of DBI that inhibited by more than50% [3H]diazepam and /3-[3H]carboline binding failed to in-hibit [3H]etorphine, [3H]GABA, [3H]quinuclidinyl benzilate,[3H]dihydroalprenolol, [3H]adenosine, and [3H]imipraminebinding when tested at their respective Kd values. Because DBIinhibits the binding of agonists and antagonists of benzodiaze-pine receptors, we tested whether DBI has binding character-istics similar to those of benzodiazepines or of f-carbolines.

"0

.-

-0.

.-4

14

A B

t

20-t

15-

10

0 VVehicle DRI Vehicle DBI FG 7142

FIG. 4. Effect of intraventricular injection ofDBI on shock-inducedsuppression of water drinking by thirsty rats. DBI obtained from theHPLC step of Fig. 1B was lyophilized and resuspended in saline, andthe pH was adjusted to 7; samples of 50 or 100 jig in 10 1.l were injectedintraventricularly. Vehicle was an equivalent volume of HPLC eluatecollected during a control gradient. (A) Anticonflict test. The intensityof the current delivered through the drinking tube was 1.0 mA. Di-azepam at 0.5 mg/kg was injected intravenously 10 min before the test,and DBI (100 lug) or vehicle was injected 5 min before the test. DBI (50Ilg) gave results comparable with those obtained with 100 ,ug of DBI.El, Control; F, diazepam treated. (B) Proconflict test. The intensity ofcurrent was adjusted to 0.25 mA. Ro 15-1788 at 1 mg/kg was injectedintravenously 15 min before the test, DBI (100 Ag) or vehicle was in-jected 3 min before the test, and FG 7142 at 1 mg/kg was injected in-travenously 15 min before the test. Rats that did not receive electricshock totaled 27 ± 1 (n = 10) licking periods (data not shown). DBI andFG 7142 failed to change the number of licking periods in the absenceof shock: DBI (100 fig), 25 3 (n = 6); DBI (50 ,ug), 28 + 3 (n = 3); FG7142 (1 mg/kg), 26 ± 2 (n = 6). El, Control; 1, Ro 15-1788 treated.*P < 0.01; controls vs. diazepam-treated rats.tRb 15-1788 in doses that failed to alter the number of licking periodsin the absence of shock prevented (P < 0.01) the inhibitory effects ofDBI and FG 7142.*DBI (100 mg) and FG 7142 (1mg/kg) decreased the number of lickingperiods in rats receiving electric shock (P < 0.01; vehicle-treated ratsvs. DBI- or FG 7142-treated rats). DBI (50 ug) decreased the numberof licking periods to 13 ± 3 (n = 3) (P < 0.01) (data not shown).

In a first group of experiments, we studied the inhibition of3H-labeled Ro 15-1788 binding by DBI, 83-carboline-3-carbox-ylic acid ethyl ester, and diazepam in the presence and absenceof 0.1 mM GABA. The inhibition of 3H-labeled Ro 15-1788binding by diazepam was potentiated by GABA. In contrast,the inhibition by DBI and f3-carbolines was not.

In a second group of experiments, the effect of DBI on thediazepam-induced increase in [3H]GABA binding was com-pared with that of 13-carbolines. DBI, like (3-carbolines, abol-ished the increase of [3H]GABA binding induced by diazepam.

Behavioral Studies. Intraventricular injection to rats de-prived of water for 3 days and implanted with a cerebral ven-tricle cannula of DBI (50 or 100 ,ug) failed to alter unpunisheddrinking or shock (1 mA)-suppressed drinking behavior. How-ever, 50-100 ,ug of DBI injected intraventricularly almost com-pletely reversed the anticonflict action of diazepam on pun-ished drinking (Fig. 4). When the shock delivered through thedrinking tube was reduced from 1 to 0.25 mA, intraventricularinjection of 50-100 .ug of DBI elicited a proconflict effect; thatis, it facilitated the shock-induced suppression of drinking bylowering the threshold for this response. This facilitation wasblocked by Ro 15-1788 (Fig. 4), and it was similar to that ob-served for the ,B-carboline derivative FG 7142.

DISCUSSIONSeveral reports (see ref. 8) have suggested that the benzodi-azepine recognition sites are part of GABA receptors and mayfunction as the recognition sites for a second synaptic signal thatregulates the gain at which GABA receptors operate. Evidencethat brain contains a constituent that displaces [3H]benzodi-azepine from its binding sites includes the fact that the Kd for[3H]diazepam binding to GABA-free crude synaptic mem-branes is reduced by approximately 50% by repeated washingof the membranes with low concentrations of detergent andthat the extract obtained produces a dose-related reduction ofthe Kd of [3Hldiazepam binding without affecting the maximalbinding capacity (13, 31).

In this report, we identify a peptide of Mr =11,000 that couldbe operative in eliciting changes in the Kd of [3H]diazepambinding in brain homogenates.

This peptide was termed DBI, and its extraction was rou-tinely carried out using as a -starting material rat brain homog-enized in hot (80°C) 1 M acetic acid. Important steps in the pu-rification of DBI are the use of 1 M acetic acid at 80°C andtreatment with 60% saturated ammonium sulfate of the ma-terial eluted from Sephadex G-75. This treatment precipitatesapproximately 80% of the proteins, leaving the DBI in solutionin the 48,000 x g supernatant. Among the proteins precipitatedby the ammonium sulfate treatment is GABA-modulin (30). Re-moval of GABA-modulin at this stage facilitates the purificationof DBI because GABA-modulin is eluted in the HPLC re-versed-phase procedure (Fig. 1) very close to DBI and becauseGABA-modulin also may inhibit [3H]diazepam binding by vir-tue of its role on the modulation of GABA recognition sites (30).

Brain DBI was purified to homogeneity, as indicated by thepresence of a single band of protein on NaDodSO4 and acidicurea gel electrophoresis and of a single protein peak on HPLCusing three different column and solvent systems. In addition,the peptide present in the HPLC peak has DBI activity andcontains tyrosine as a single carboxyl terminus.

Experiments carried out to establish the amino acid se-quence of this purified peptide showed that the NH2-terminalamino acid is blocked. This blockade was not removed with py-roglutamate aminopeptidase; however, the presence of twomethionine residues in the molecule has now allowed the gen-

Proc. Natl. Acad. Sci. USA 80 (1983)

Proc. Natl. Acad. Sci. USA 80 (1983) 3535

eration of three fragments by cyanogen bromide treatment. Theentire sequence of the carboxyl-terminal fragment was deter-mined, and a partial sequence of the middle fragment was de-termined. These sequences do not resemble any known mam-malian peptide sequence (32).An important characteristic of DBI is that it is present at high

concentrations (10-25 AtM) in rat brain and only traces (less than1 ,tM) were found in peripheral organs (liver, kidney, and spleen).DBI appears to interact with benzodiazepine binding sites,competitively inhibiting the specific binding of [3H]benzodi-azepine or ,3[3H]carboline derivatives; however, it fails to blockthe binding of a number of ligands to other transmitter rec-ognition sites. DBI functionally interacts with the benzodiaze-pine recognition site in a specific manner that resembles anx-iety-inducing (-carbolines such as FG 7142 (6, 7). It differs fromanxiety-abating benzodiazepines and from benzodiazepines that,like Ro 15-1788, bind to the benzodiazepine recognition sitewith high affinity but are devoid of intrinsic activity (17).We considered whether a (3-carboline structure formed dur-

ing the DBI purification was associated with DBI. Experimentswith trypsin, Pronase, and papaine digestion, which resulted incomplete loss of DBI activity, indicated that this is unlikely.

However, the similarity in the action of DBI with that of theanxiety-inducing (3-carboline FG 7142 suggests that DBI isprobably the precursor of a lower molecular weight endoge-nous effector of the benzodiazepine recognition site. This ef-fector may modulate a specific set of GABA receptors and, whenthese receptors are down regulated, the onset of anxiety maybe facilitated. Further studies of DBI and its active fragmentsmay lead to the identification of endogenous physiologicallyimportant ligands for the benzodiazepine receptor. These li-gands may provide a fine tuning for the GABAergic transmis-sion, thereby modulating anxiety, excitability, and onset of sleep.We thank Mr. J. A. Rodkey for sequence determination and Mrs. S.

S. Fitzpatrick for amino acid analyses.

1. Braestrup, C. & Squires, R. F. (1977) Proc. NatL Acad. Sci. USA74, 3805-3809.

2. Mohler, H. & Okada, T. (1978) Science 198, 849-851.3. Mohler, H. & Okada, T. (1978) Br. J. Psychiatry 133, 261-268.4. Cook, L. & Sepinwall, J. (1975) in Mechanism of Action of Ben-

zodiazepines, eds. Costa, E. & Greengard, P. (Raven, New York),pp. 1-28.

5. Stein, L., Wise, C. D. & Belluzzi, J. D. (1975) in Mechanism ofAction of Benzodiazepines, eds. Costa, E. & Greengard, P. (Raven,New York), pp. 28-44.

6. Braestrup, C., Schmiechen, R., Neef, G., Nielsen, G. & Peter-sen, E. W (1982) Science 216, 1241-2143.

7. Corda, M. G., Blaker, W. M., Mendelson, W. B., Guidotti, A. &Costa, E. (1983) Proc. Natl Acad. Sci. USA 80, 2072-2076.

8. Costa, E. & Guidotti, A. (1979) in Annual Review of Pharmacol-ogy and Toxicology, eds. Okum, G. R. & Cho, A. K. (Annu. Rev.,Palo Alto, CA), Vol. 19, pp. 531-545.

9. Asano, T. & Spector, S. (1979) Proc. Natl. Acad. Sci. USA 76, 977-981.

10. Davis, L. & Cohen, R. K. (1980) Biochem. Biophys. Res. Com-mun. 92, 141-148.

11. Colello, -G. D., Hockenbery, D. M., Bosmann, H. B., Fuchs, S.& Folkers, K. (1978) Proc. Natl, Acad. Sci. USA 75, 6319-6323.

12. Karobath, M., Sperk, G. & Schonbeck, G. (1978) Eur. J. Phar-macol 49, 323-326.

13. Massotti, M., Guidotti, A. & Costa, E. (1981)J. Neurosci. 1, 409-418.

14. Mohler, H., PoIc, P., Cumin, R., Pieri, L. & Ketter, R. (1979)Nature (London) 278, 563-565.

15. Skolnick, P., Marangos, P. J. & Goodwin, F. K. (1978) Life Sci.23, 1473-1480.

16. Woolf, J. H. & Nixon, J. (1981) Biochemistry 20, 4263-4269.17. Hunkeler, W., Mohler, H., Pieri, L., Polc, P., Bonetti, E. P.,

Cumin, R., Schaffer, R. & Haefely, W. (1981) Nature (London) 514-516.

18. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J.(1951) J. Biol Chem. 193, 265-275.

19. Panyim, S. & Chalkley, R. (1966) Arch. Biochem. Biophys. 130,337-346.

20. Laemmli, U. K. & Favre, M. (1973)1. Mol Biol 80, 575-599.21. Simon, E. J., Hiller, J. M. & Edelman, I. (1973) Proc. Natl Acad.

Sci. USA 70, 1947-1949.22. Snyder, S. H., Katims, J. J., Annau, Z., Bruns, R. F. & Daly, J.

W. (1981) Proc. Nati Acad. Sci. USA 78, 3260-3264.23. Raisman, R., Briley, M. & Langer, S. Z. (1979) Nature (London)

281, 148-150.24. Yamamura, H. I. & Snyder, J. H. (1974) Proc. NatL Acad. Sci. USA

71, 1725-1729.25. Chuang, D. M. & Costa, E. (1979) Proc. Natl Acad. Sci. USA 76,

3024-3028.26. Bennett, C. D., Rodkey, J. A., Sondey, J. M. & Hirschmann, R.

(1978) Biochemistry 17, 1328-1337.27. Spiess, J., Rivier, J. E., Rodkey, J. A., Bennett, C. D. & Vale,

W. (1979) Proc. Natl Acad. Sci. USA 76, 2974-2978.28. Castel, A., Kraal, B., DeGraff, J. M. & Bosch, L. (1979) Eur. J.

Biochem. 102, 125-138.29. Robinson, C. A., Hengerald, C. A. & DeBalbian-Vester, F. (1969)

Physiol Behav. 4, 123-124.30. Guidotti, A., Konkel, D. R., Ebstein, B., Corda, M. G., Krutzsch,

H., Meek, J. L. & Costa, E. (1982) Proc. NatL Acad. Sci. USA 79,6084-6088.

31. Massotti, M. & Guidotti, A. (1980) Life Sci. 27, 847-854.32. Dayhoff, M. O., Hunt, L. T., Barker, W. C., Schwartz, R. M. &

Orcutt, B. C. (1978) in Protein Segment Dictionary (National Bio-chemical Research Foundation, Washington, DC), Vol. 78.

Neurobiology: Guidotti et aL