demonstration benzodiazepine-like molecules the mammalian1 ,umchlordiazepoxide(c);...

5
Proc. Nad. Acad. Sci. USA Vol. 82, pp. 5560-5564, August 1985 Neurobiology Demonstration of benzodiazepine-like molecules in the mammalian brain with a monoclonal antibody to benzodiazepines (agonist/benzodiazepine receptors/localization/characterization) LAKSHMI SANGAMESWARAN AND ANGEL L. DE BLAS* Department of Neurobiology and Behavior, State University of New York, Stony Brook, NY 11794 Communicated by Severo Ochoa, April 5, 1985 ABSTRACT An anti-benzodiazepine monoclonal anti- body has been used to demonstrate the existence of benzodi- azepine-like molecules in the brain. Immunocytochemical ex- periments show that these molecules are neuronal and not glial and that they are ubiquitously distributed throughout the brain. Immunoblots indicate the presence of benzodiazepine- like epitopes in several brain peptides. Small benzodiazepine- like molecules were isolated from the brain soluble fraction by immunoaffinity chromatography. They block the binding of agonists, inverse agonists, and antagonists to the neuronal-type benzodiazepine receptor. The neurotransmitter ry-amino- butyric acid increases the affinity of the benzodiazepine recep- tor for the purified endogenous molecules. The results indicate that the immunoafIinity-purified molecules behave like the neuronal-type benzodiazepine receptor agonists. The purified molecules, however, do not inhibit the binding of tritiated Ro 5-4864 to the "peripheral-type" benzodiazepine receptor. The results demonstrate the existence of benzodiazepine-like mol- ecules in the brain that bind to the benzodiazepine receptor. These molecules are different from the endogenous benzodi- azepine receptor ligands reported by others. The mammalian brain has been shown to have receptors for benzodiazepines (BZDs) (1, 2). One could ask, why has evolution provided the brain with receptors for these syn- thetic drugs (e.g., Valium, Librium, etc.)? Possibly the brain could produce endogenous BZD-like molecules that bind to the BZD receptors (BZDRs), modulating y-aminobutyric acid (GABA) neurotransmission. This notion is supported by the existence in the brain of endogenous opiate peptides, which were discovered after the discovery of the brain opiate receptors. A number of substances extracted from brain and other tissues bind to the BZDR-e.g., purines and purine nucleosides (3, 4), nicotinamide (5), ,3-carbolines (6), and a number of peptides and proteins of Mr 1500-70,000 (7-11). However, there is no real evidence that these compounds play any physiological role as ligands for BZDR (12, 13). Thus, purines, purine nucleosides, and nicotinamide displace diazepam binding to the receptor with an IC50 of -1 mM, whereas diazepam has a Kd in the nanomolar range. Although the ,B-carbolines have high affinity (Kd in the nanomolar range) and high specificity for BZDR, they are not natural brain constituents but are rather artifacts formed during tissue extraction (12, 13). The best-characterized brain pep- tide that binds to BZDR is the diazepam-binding inhibitor (DBI), Mr 11,000. DBI has been purified and the amino acid sequence, including a 18 amino acid active fragment, is known (11, 14, 15). DBI inhibits the binding of p-carbolines and BZDs to BZDR with Ki = 1-4 ,uM. The characterization of other proteins and peptides that displace [3H]diazepam binding to the brain BZDR is incomplete. In this paper we describe the use of an anti-benzodiazepine monoclonal antibody (mAb) for the identification and pre- liminary characterization of BZD-like molecules that bind to the brain BZDR. Some of these results have been commu- nicated in preliminary form (16). MATERIALS AND METHODS The mAb to BZD, 21-7F9, was obtained after immunizing BALB/c mice with 3-hemisuccinyloxyclonazepam coupled to bovine serum albumin (BZD-BSA). The production of the hybridomas and the comparative study of the BZD binding properties of this and other antibodies will be reported elsewhere (38). The hybridomas secreting mAbs to BZD were identified by assaying the conditioned culture media in three ways: (i) binding of 3H-labeled flunitrazepam (FNZ) tp the mAbs, (ii) antibody competition with tie [3H]FNZ binding to the brain membrane BZDR, and (iii) antibody binding to the 3-hemisuccinyloxyclonazepam conjugate with keyhole lim- pet hemocyanin (BZD-KLH) in a standard ELISA assay. The hybridomas were recloned by limiting dilution. The mAb 21-7F9 (an IgM) was purified from ascites fluid by ammonium sulfate precipitation followed by gel exclusion chromatography on Sephacryl S300 (Pharmacia). The puri- fied mAb 21-7F9, coupled to Bio-Rad Affi-Gel 10 (3.5 mg of antibody protein per ml of gel), was used for the isolation of BZD-like compounds. Brains of 20 Sprague-Dawley rats (30 days old; Taconic Farms, Germantown, NY) were homoge- nized in 60 ml of 5 mM Tris HCl, pH 8.1/0.3 mM phenylmethylsulfonyl fluoride/0.1 mM EGTA/20 ,g of tryp- sin inhibitor per ml/1 ,ug of pepstatin per ml. The homogenate was centrifuged at 100,000 x g for 90 min at 4°C and the supernatant was centrifuged at the same speed for 30 min. The final supernatant was adjusted to pH 7.4 by adding 4 ml of 1 M Tris HCl, pH 7.4. The extract was passed through a 5-ml immunoaffinity column (containing immobilized mAb 21-7F9) at a flow rate of 6 ml/hr. The column was washed first with 50 ml of 50 mM TrisHCl, pH 7.4, containing the aforementioned protease inhibitors, followed by 25 ml of 50 mM Tris HCl, pH 7.4, containing 0.1 M NaCl and the protease inhibitors, and finally with 50 ml of H20. The retained material was eluted with 75 ml of 0.2 M acetic acid at a flow rate of 50 ml/hr. The eluate was dried uhder reduced pressure and dissolved in 5 ml of 50 mM Tris-HCI, pH 7.4. The final pH of the solution was 7.4. Materials. [3H]FNZ (85 Ci/mmol; 1 Ci = 37 GBq), [3H]Ro 5-4864 (76.6 Ci/mmol), and [3H]Ro 15-1788 (87 Ci/mmol) Abbreviations: 83CCE, f8-carboline-3-carboxylate ethyl ester; BZD, benzodiazepine; BZD-BSA, 3-hemisuccinyloxyclonazepam conju- gate with bovine serum albumin; BZD-KLH, 3-hemisuccinyloxyclo- nazepam conjugate with keyhole limpet hemocyanin; BZDR, ben- zodiazepine receptor; DBI, diazepam-binding inhibitor; FNZ, flunitrazepam; GABA, y-apiinobutyric acid; mAb, monoclonal an- tibody. *To whom reprint requests should be addressed. 5560 The publication costs of this article were defrayed. in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Downloaded by guest on October 28, 2020

Upload: others

Post on 07-Aug-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Demonstration benzodiazepine-like molecules the mammalian1 ,uMchlordiazepoxide(C); 10AM,3CCE(D),and10,uMRo15-1788 (E) on the 21-7F9 binding to the cerebellum. (F) Negative control

Proc. Nad. Acad. Sci. USAVol. 82, pp. 5560-5564, August 1985Neurobiology

Demonstration of benzodiazepine-like molecules in the mammalianbrain with a monoclonal antibody to benzodiazepines

(agonist/benzodiazepine receptors/localization/characterization)

LAKSHMI SANGAMESWARAN AND ANGEL L. DE BLAS*Department of Neurobiology and Behavior, State University of New York, Stony Brook, NY 11794

Communicated by Severo Ochoa, April 5, 1985

ABSTRACT An anti-benzodiazepine monoclonal anti-body has been used to demonstrate the existence of benzodi-azepine-like molecules in the brain. Immunocytochemical ex-periments show that these molecules are neuronal and not glialand that they are ubiquitously distributed throughout thebrain. Immunoblots indicate the presence of benzodiazepine-like epitopes in several brain peptides. Small benzodiazepine-like molecules were isolated from the brain soluble fraction byimmunoaffinity chromatography. They block the binding ofagonists, inverse agonists, and antagonists to the neuronal-typebenzodiazepine receptor. The neurotransmitter ry-amino-butyric acid increases the affinity of the benzodiazepine recep-tor for the purified endogenous molecules. The results indicatethat the immunoafIinity-purified molecules behave like theneuronal-type benzodiazepine receptor agonists. The purifiedmolecules, however, do not inhibit the binding of tritiated Ro5-4864 to the "peripheral-type" benzodiazepine receptor. Theresults demonstrate the existence of benzodiazepine-like mol-ecules in the brain that bind to the benzodiazepine receptor.These molecules are different from the endogenous benzodi-azepine receptor ligands reported by others.

The mammalian brain has been shown to have receptors forbenzodiazepines (BZDs) (1, 2). One could ask, why hasevolution provided the brain with receptors for these syn-thetic drugs (e.g., Valium, Librium, etc.)? Possibly the braincould produce endogenous BZD-like molecules that bind tothe BZD receptors (BZDRs), modulating y-aminobutyricacid (GABA) neurotransmission. This notion is supported bythe existence in the brain of endogenous opiate peptides,which were discovered after the discovery of the brain opiatereceptors. A number of substances extracted from brain andother tissues bind to the BZDR-e.g., purines and purinenucleosides (3, 4), nicotinamide (5), ,3-carbolines (6), and anumber of peptides and proteins of Mr 1500-70,000 (7-11).However, there is no real evidence that these compoundsplay any physiological role as ligands for BZDR (12, 13).Thus, purines, purine nucleosides, and nicotinamide displacediazepam binding to the receptor with an IC50 of -1 mM,whereas diazepam has a Kd in the nanomolar range. Althoughthe ,B-carbolines have high affinity (Kd in the nanomolarrange) and high specificity for BZDR, they are not naturalbrain constituents but are rather artifacts formed duringtissue extraction (12, 13). The best-characterized brain pep-tide that binds to BZDR is the diazepam-binding inhibitor(DBI), Mr 11,000. DBI has been purified and the amino acidsequence, including a 18 amino acid active fragment, isknown (11, 14, 15). DBI inhibits the binding of p-carbolinesand BZDs to BZDR with Ki = 1-4 ,uM. The characterizationof other proteins and peptides that displace [3H]diazepambinding to the brain BZDR is incomplete.

In this paper we describe the use of an anti-benzodiazepinemonoclonal antibody (mAb) for the identification and pre-liminary characterization of BZD-like molecules that bind tothe brain BZDR. Some of these results have been commu-nicated in preliminary form (16).

MATERIALS AND METHODSThe mAb to BZD, 21-7F9, was obtained after immunizingBALB/c mice with 3-hemisuccinyloxyclonazepam coupledto bovine serum albumin (BZD-BSA). The production of thehybridomas and the comparative study of the BZD bindingproperties of this and other antibodies will be reportedelsewhere (38). The hybridomas secreting mAbs to BZD wereidentified by assaying the conditioned culture media in threeways: (i) binding of 3H-labeled flunitrazepam (FNZ) tp themAbs, (ii) antibody competition with tie [3H]FNZ binding tothe brain membrane BZDR, and (iii) antibody binding to the3-hemisuccinyloxyclonazepam conjugate with keyhole lim-pet hemocyanin (BZD-KLH) in a standard ELISA assay. Thehybridomas were recloned by limiting dilution.The mAb 21-7F9 (an IgM) was purified from ascites fluid

by ammonium sulfate precipitation followed by gel exclusionchromatography on Sephacryl S300 (Pharmacia). The puri-fied mAb 21-7F9, coupled to Bio-Rad Affi-Gel 10 (3.5 mg ofantibody protein per ml of gel), was used for the isolation ofBZD-like compounds. Brains of 20 Sprague-Dawley rats (30days old; Taconic Farms, Germantown, NY) were homoge-nized in 60 ml of 5 mM Tris HCl, pH 8.1/0.3 mMphenylmethylsulfonyl fluoride/0.1 mM EGTA/20 ,g of tryp-sin inhibitor per ml/1 ,ug ofpepstatin per ml. The homogenatewas centrifuged at 100,000 x g for 90 min at 4°C and thesupernatant was centrifuged at the same speed for 30 min.The final supernatant was adjusted to pH 7.4 by adding 4 mlof 1 M Tris HCl, pH 7.4. The extract was passed through a5-ml immunoaffinity column (containing immobilized mAb21-7F9) at a flow rate of6 ml/hr. The column was washed firstwith 50 ml of 50 mM TrisHCl, pH 7.4, containing theaforementioned protease inhibitors, followed by 25 ml of 50mM Tris HCl, pH 7.4, containing 0.1 M NaCl and theprotease inhibitors, and finally with 50 ml of H20. Theretained material was eluted with 75 ml of 0.2 M acetic acidat a flow rate of50 ml/hr. The eluate was dried uhder reducedpressure and dissolved in 5 ml of 50 mM Tris-HCI, pH 7.4.The final pH of the solution was 7.4.

Materials. [3H]FNZ (85 Ci/mmol; 1 Ci = 37 GBq), [3H]Ro5-4864 (76.6 Ci/mmol), and [3H]Ro 15-1788 (87 Ci/mmol)

Abbreviations: 83CCE, f8-carboline-3-carboxylate ethyl ester; BZD,benzodiazepine; BZD-BSA, 3-hemisuccinyloxyclonazepam conju-gate with bovine serum albumin; BZD-KLH, 3-hemisuccinyloxyclo-nazepam conjugate with keyhole limpet hemocyanin; BZDR, ben-zodiazepine receptor; DBI, diazepam-binding inhibitor; FNZ,flunitrazepam; GABA, y-apiinobutyric acid; mAb, monoclonal an-tibody.*To whom reprint requests should be addressed.

5560

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

Dow

nloa

ded

by g

uest

on

Oct

ober

28,

202

0

Page 2: Demonstration benzodiazepine-like molecules the mammalian1 ,uMchlordiazepoxide(C); 10AM,3CCE(D),and10,uMRo15-1788 (E) on the 21-7F9 binding to the cerebellum. (F) Negative control

Proc. Natl. Acad. Sci. USA 82 (1985) 5561

Table 1. Specific binding of BZDR ligands to mAb 21-7F9

Conc., BindingLigand Nature nM cpm fmol

[3H]FNZ Agonist 0.66 13,706 ± 178 154 ± 2[3H]Diazepam Agonist 0.46 14,364 ± 266 108 ± 2[3H]Ro 5-4864 Peripheral 0.86 12,4% ± 923 176 ± 13[3H]Ro 15-1788 Antagonist 1.15 0 0[3H]8CCE Inverse agonist 2.0 0 0

K CO2CH2CH3F

JW N. J ~~C02C2H5F

0 CH3 H

FIG. 1. Structures of some BZbR ligands. From left to right areFNZ, diazepam, and Ro 5-4864 in the upper row and Ro 15-1788 and,3CCE in the lower row.

were from New England Nuclear. [3H]Diazepam (72.5Ci/mmol) and 83-[6-3H]carboline-3-carboxylate ethyl ester([3H],3CCE; 29 Ci/mmol) were from Amersham. All theunlabeled BZDs used in this study were the kind gifts of R. I.Fryer, W. E. Scott, and P. F. Sorter ofHoffmann-La Roche.

RESULTS

mAbs with High Affinity for Free I1ZDs. Four mAbs withhigh binding affinity for the BZDR agonists [3H]FNZ and[3H]diazepam were generated, They did not bind the BZDRinverse agonist [3H]3CCE or the antagonist [3H]Ro 15-1788.t(See Fig. 1 for structures.) No binding of these mAbs toligands for other transmitter receptors was found. Only oneof the anti-BZD antibodies (21-7F9) showed binding to brainmolecules. Some of the BZD binding specificities of the mAb21-7F9 are shown in Table 1.The binding affinity of 21-7F9 for [3H]FNZ was calculated

by Scatchard analysis, which showed a straight line with Kd= 0.55 x 10-9 M. The iffinities of 21-7F9 for other BZDRligands were calculated by determining their blocking poten-cies on the binding of [3H]FNZ to the antibodies. The resultsshow (38) that whereas 21-7F9 binds with very high affinityto FNZ, Ro 11-6896, Ro 5-4864, diazepam, and Ro 11-6893,it binds with very low affinity to j8CCE, ,B-carboline-3-carboxylate methyl ester, Ro 15-1788, and CL218-872. Theseresults and Table 1 show that 21-7F9 has high affinity formany of the BZDR agonists but low affinity for BZDRantagonists (e.g., Ro 15-1788) and for inverse agonists (e.g.,P-carbolines). Another noteworthy feature of 21-7F9 is itshigh binding affinity for both FNZ and Ro 5-4864. However,the affinity of the neuronal-type BZDR for Ro 5-4864 is verylow, which shows that the binding sites of the mAb and theBZDR are similar but not identical.mAb 21-7F9 Binds to Neuronal Antigens with BZD-like

Epitopes. Immunocytochemistry. Fig. 2 shows the distribu-tion of the 21-7F9 binding to some selected brain areas andcell types. The antibody binds to molecules distributedthroughout the brain that are present in neurons but not inglial cells. These molecules are in the perikarya and pro-cesses of many but not all neurons. Fig. 3 shows that thebinding of mAb 21-7F9 to the brain molecules (Fig. 3A) canbe blocked by 0.1 ,uM FNZ (Fig. 3B) but not by 10 ,uM Ro15-1788 (Fig. 3E) or 1 ,uM ,BCCE (Fig. 3D). However, 1 AMchlordiazepoxide (Fig. 3C) has an intermediate blocking

tThe B3ZDR agonists potentiate the effect of GABA on the openingof the chloride channel. The BZDR inverse agonists decrease theeffect of GABA on the opening of the channel and have pharma-cological actions opposite to BZDs. The BZDR antagonists blockthe binding of both agonists and inverse agonists to the BZDR buthave no major biological activity.

Hybridoma culture medium (50 ,l) was added to 450 ,ul of theradioligand in 50 mM Tris HCl, pH 7.4, and incubated for 1 hr at 4°C.The nonspecific binding was determined by using 1 ,uM unlabeledligand. The specific binding is the difference between total andnonspecific binding. No radioligand binding to control mAbs wasdetected. The antibody-BZD complexes were precipitated with 7.5%(vol/vol) polyethylene glycol 8000, using 0.125% bovine gammaglobulin as a carrier. The complexes were recovered by filtrationthrough Whatman GF/B filters and the radioactivity was determinedas described elsewhere (17). Values are means and standard devia-tions of two experiments.

potency. Other effective blockers of 21-7F9 binding to braintissue were Ro 11-6896, Ro 5-4864, diazepam, and Ro 11-6893(not shown).Immunoblotting. Fig. 4 shows that 21-7F9 binds to a

number of peptides present in the homogenates of variousbrain regions. No qualitative differences between theseregions were found. The molecular weights of the reactivepeptides range between 20,000 and 300,000. The most abun-dant peptide with BZD-like epitopes is the Mr 100,000 one.Some, peptides (including the Mr 100,000 species) are asso-ciated with a synaptosomal membrane fraction. Fig. 4 alsoshows that the binding of 21-7F9 to the blotted peptides isblocked by FNZ and diazepam but not by the antagonist Ro

Ae.0Bv--K ,

I CA

A~~~~~~

4iPNM*S.

FIG. 2. Immunocytochemistry with the anti-BZD mAb 21-7F9:Cerebellum (A), cerebral cortex (B), hippocampus (C), and dentategyrus (D) from an adult Sprague-Dawley rat. (E-H) High magnifi-cation of two cerebellar Purkinje cells (E and F), a pyramidal neuronof the layer V of the cerebral cortex (G), and cerebellar granular cells(H). Differential interference-contrast optics. The bar inA represents100 ,um; the bar in E represents 10 gm. The peroxidase-antiperoxidase immunocytochemistry was done as explained else-where, using hybridoma culture medium (18, 19).

Neurobiology: Sangarneswaran and de Blas

Dow

nloa

ded

by g

uest

on

Oct

ober

28,

202

0

Page 3: Demonstration benzodiazepine-like molecules the mammalian1 ,uMchlordiazepoxide(C); 10AM,3CCE(D),and10,uMRo15-1788 (E) on the 21-7F9 binding to the cerebellum. (F) Negative control

5562 Neurobiology: Sangameswaran and de Blasaid as wj Ad BHe .w

..

u Wt .f a. s.toss;,., .* *) vr:FW.

..A. WS'-

;; of. Set .,, :, r,, 9.4 me

_ .,, ¢ * < .;

A; 8 ,.h

S. {S I> ..

At .. w

C

....

~~~~~~~~~~..

F .s

Iw w

Ft

-_ p.o A:.by_ inv - 2,.

!g$ F

_i p.,.

As. A- f

_ *: F^} f . IrA,21-I

FIG. 3. Displacement of the antibody 21-7F9 binding to thecerebelluni by BZDs and other BZDR ligands. (A) Positive controlwith 21-7F9. The other panels show the effect of 0.1 uM FNZ (B);1 ,uM chlordiazepoxide (C); 10 AM ,3CCE (D), and 10 ,uM Ro 15-1788(E) on the 21-7F9 binding to the cerebellum. (F) Negative controlusing the culture medium from the parental myeloma lineP3X63Ag8.6.5.3. The bar represents 100 ,um.

15-1788. Other effective blockers of the mAb binding are Ro11-6896, Ro 11-6893, and Ro 5-4864 (not shown). In contrast,f3CCE and 3-carboline-3-carboxylate methyl ester were in-effective blockers (not shown). Fig. 4 shows further that FNZdoes not block the binding of other mAbs (8-2H5, 8-7A5, and4-4C3) to synaptosomal membrane antigens. The antigensrecognized by these antibodies are unrelated to the BZDsystem and have been described elsewhere (18). Theseantibodies were used as negative controls.There is a very good correlation between the effectiveness

of various BZDs for blocking [3H]FNZ binding to mAb21-7F9 and their potency for blocking the binding of 21-7F9to the brain molecules in both the immunocytochemistry andthe immunoblot assays. These results suggest very closestructural similarities between BZDs and the endogenousbrain BZD-like epitopes recognized by 21-7F9.

Brain Molecules Purified by Immunoaffinity Chromatogra-phy on Immobilized mAb 21-7F9 Specifically Bind to the BZDBinding Site of the Neuronal-Type BZDRs. Table 2 shows thatthe immunoaffinity-purified BZD-like brain molecules blockthe binding of the agonists [3H]FNZ and [3H]diazepam, theinverse agonist [3H]p3CCE, and the antagonist [3H]Ro 15-1788to the neuronal-type BZDR, as is expected of a ligand for thisreceptor (it is worth noting the difference between the BZDRand the mAb in that the former has binding sites for the threeclasses of ligands while the latter has only the agonist bindingsite). However, the purified molecules do not block [3H]Ro5-4864 binding to the "peripheral-type" or "glial-type" brainBZDR to which this ligand bindst (22-25). The purifiedmolecules do not'block the binding of the ligands for othertransmitter receptors (/3-adrenergic, GABA, and adenosinereceptors). These results show the very high binding speci-ficity of the purified BZD-like molecules for the neuronal-type BZDR. It is worth noting, however, that both [3H]FNZand [3H]Ro 5-4864 bind with high affinity to the anti-BZDmAb 21-7F9 as seen above. The results show that the

tThe brain "peripheral-type" BZDRs are also found on the neuronsof certain brain regions (26).

SPM

7F9 i h

.c zDc 4 7

x< 10

300260

16080

40

20

FIG. 4. Immunoblots of extracts of various brain regions withmAb 21-7F9 and displacements of 21-7F9 binding to brain peptidesby BZDR ligands. The samples were taken from homogenate of ratcerebellum (CBL), olfactory bulb (OB), and cerebral cortex (CTX).SPM represents synaptosomal plasma membranes from cerebralcortex. From left to right the strips in CBL and CTX show the bindingof 21-7F9 without and with 1 ,iM FNZ. OB also shows the generalprotein stain with amido black on the left side. The four strips at theleft ofSPM show the binding of21-7F9 to the peptides in the presenceof (left to right) 1 ,uM FNZ, 1 /iM diazepam, 1 /iM Ro 15-1788, andno BZD, respectively. The SPM blots were also stained with themAbs to synaptosomal membranes 8-2H5, 8-7A5, and 44C3 asmolecular weight markers (18, 20). The binding of these antibodieswas also done without and with (left and right strips) 1 ILM FNZ. Theparental myeloma P3X63Ag8.6.5.3 culture medium was used as anegative control (extreme right strip). The immunoblotting proce-dure and the preparation of the SPM fraction were describedelsewhere (20, 21).

immobilized mAb does not bind any endogenous ligands,which might be present in the brain, for the peripheral-typeBZDR (27).The blocking activity of the purified molecules for

[3H]FNZ binding was retained by a second, identical, im-munoaffinity chromatography column (not shown). Further-more, in experiments in which only the buffer and not thebrain extract was passed through the column, no activity wasdetected in the eluate. These experiments indicate that theactivity blocking the binding of [3H]FNZ originated from thebrain extract and was not an artifact introduced by thecolumn. The possibility that mAb 21-7F9 was leaking fromthe column was ruled out not only by the previous experi-ments but also because the immunoaffinity-purified mole-cules did not bind [3H]FNZ.Table 3 shows that the inhibition of the binding of [3H]Ro

15-1788 to BZDR by the purified molecules is potentiated byGABA. In this regard, the behavior of the purified moleculesis similar to that of the BZDs (receptor agonists)-e.g.,diazepam (Table 3)-but different from that of the fB-carbo-lines (inverse agonists) and antagonists (28). Furthermore,the binding of [3H]FNZ to BZDR was competitively inhibitedby the purified molecules (not shown). These results indicatethat the purified molecules specifically bind to the BZDbinding site but not to the f3-carboline or the antagonistbinding sites of the neuronal-type BZDR. Table 3 also showsthat, under similar binding conditions, GABA did not poten-tiate the nicotinamide, hypoxanthine, and purine nucleosideinhibition of [3H]Ro 15-1788 binding to BZDR. Inhibition wasobserved only at very high concentrations of these com-pounds. Table 3 shows that these compounds can also inhibitto a similar extent the binding of both [3H]FNZ and [3H]Ro

Proc. Natl. Acad. Sci. USA 82 (1985)

I

zt2T

Dow

nloa

ded

by g

uest

on

Oct

ober

28,

202

0

Page 4: Demonstration benzodiazepine-like molecules the mammalian1 ,uMchlordiazepoxide(C); 10AM,3CCE(D),and10,uMRo15-1788 (E) on the 21-7F9 binding to the cerebellum. (F) Negative control

Proc. Natl. Acad. Sci. USA 82 (1985) 5563

Table 2. Effect of the immunoaffinity-purified BZD-like substances on the binding of ligands to BZDR

Conc., Ligand bound, fmolReceptor Ligand nM Without eluate With eluate % of control

Neuronal-type BZDR [3H]FNZ 0.1 10.1 + 2.6 0.7 ± 0.4 7[3H]Diazepam 0.46 8.7 ± 0.9 1.0 ± 0.1 11[3H]pCCE 2.0 53.9 ± 0.6 12.4 ± 5.3 23[3H]Ro 15-1788 1.15 78.2 ± 0.3 5.3 ± 0.2 7

Peripheral-type BZDR [3H]Ro 5-4864 0.86 23.3 ± 3.7 22.3 ± 1.5 9GABA receptor [3H]Muscimol 2.4 15.0 ± 2.3 16.2 ± 2.8 108The reaction mixture was 410 Ag of rat brain membranes, the radioligand, and 100,u1 of the immunoaffinity column eluate

in a total volume of 500 A.l of 50 mM Tris HCl, pH 7.4 at 40C. The membrane-bound radioligand was determined afterfiltration through a Whatman GF/B filter as described elsewhere (17). Values are means and standard deviations of twoexperiments.

5-4864 to the neuronal- and glial-type receptors, respectively.Inosine can distinguish between the two receptor types.Nevertheless, the distinction is not as dramatic as the onemade by the purified molecules (Tables 2 and 3).

In a preliminary attempt to characterize the immunoaf-finity-purified molecules, we found that their activity([3H]FNZ binding inhibition) (i) is resistant to proteases(Pronase, leucine aminopeptidase, proteinase K, trypsin, andchymotrypsin); (ii) is resistant to heating for 5 min at 100TC;and (iii) has the mobility of a Mr 1000 molecule (tentative), asdetermined by gel exclusion chromatography.

DISCUSSION

We have shown in this study that (i) the brain has a family ofneuronal molecules with BZD-like epitopes that can berecognized by the anti-BZD mAb 21-7F9 and (ii) the brainmolecules purified by immunoaffinity chromatography on theimmobilized mAb 21-7F9 block [3H]FNZ binding to the brainBZDR. The purified molecules specifically bind to the BZD-binding site of the neuronal-type BZDR and behave asagonists.The immunocytochemical experiments indicate that the

endogenous molecules are localized in many neurons but notin glial cells. Distribution throughout the brain is consistentwith their possible role as modulators of GABA neurotrans-mission: it has been estimated that up to 40%o of the brainsynapses are mediated by GABA (29), and GABA and theGABA-synthesizing enzyme glutamate decarboxylase are

ubiquitously distributed in the brain. Some neurons that show21-7F9 immunoreactivity (Fig. 2 B, C, and D) do not contain

glutamate decarboxylase-e.g., the pyramidal neurons of thecerebral cortex (30) and hippocampus (31) and the granulecells of the dentate gyrus (32). However, other GABA-ergicneurons such as the Purkinje, basket, stellate, and Golgi cellsin the cerebellum showed reactivity with mAb 21-7F9 (Fig.2A). It is worth noting that almost identical distributions ofimmunoreactivity in the cerebellum were observed withanti-BZD mAb and an antiserum to GABA (33).Our present strategy is to purify the smallest BZD-like

molecules, which should be the easiest to characterizechemically. They are found in the soluble fraction andtherefore they are the ones purified by immunoaffinitychromatography. In contrast, the larger peptides with BZD-like epitopes are associated to the membrane fraction. Theactivity of the immunoaffinity-purified BZD-like molecules isprotease-resistant and thermostable, but these properties donot rule out a peptide, for there are examples of smallpeptides with similar properties (e.g., thyrotropin-releasinghormone, TRH). The relationship between the purified mol-ecules and the peptides that are recognized by the antibody21-7F9 in the immunoblots is unknown. We do not know yetwhether the former are derived from the latter and whetherthe immunoblotted peptides are related to each other suchthat the larger are precursors of the smaller. Enkephalins,endorphins, and dynorphins have Mr 30,000 precursor pep-tides (34), and the epidermal growth factor precursor is aprotein of 130,000 Mr (35). It has recently been reported thatantibodies to neuropeptide-related antigens bind to aneuronal membrane glycoprotein of 100,000 Mr, and it hasbeen suggested that this glycoprotein could be a precursor ofsmaller neuropeptides (36). This resembles our finding thatmAb 21-7F9 recognizes a Mr 100,000 synaptosomal mem-

Table 3. GABA increases the inhibitory potency of the purified BZD-like molecules on [3H]Ro15-1788 binding to BZDRs

Radioligand bound, %

[3H]Ro 15-1788 [3H]-Ligand Without GABA With 0.1 mM GABA [3H]FNZ Ro 5-4864

None 100.0 100.0 100.0 100.0Column eluate 1 53.0 ± 2.0 36.5 ± 2.3Column eluate 2 44.2 ± 0.8 31.7 ± 3.1 36 ± 7 101 ± 1Diazepam (10 nM) 33.0 ± 0.1 19.6 ± 1.3f3CCE (2 nM) 50.4 ± 2.3 55.6 ± 3.2Nicotinamide (10 mM) 42.9 ± 0.3 41.9 ± 2.6 39 ± 1 43 ± 10Hypoxanthine (4 mM) 49.8 ± 5.0 47.9 ± 0.2 37 ± 1 22 ± 2Adenosine (4 mM) 75.2 ± 0.2 75.6 ± 1.2 62 ± 6 77 ± 9Inosine (4 mM) 42.0 ± 1.8 40.6 ± 2.3 32 ± 3 60 ± 5Guanosine (4 mM) 67.7 ± 2.0 69.4 ± 1.0

The values are the percent of controls, in which no ligand was added. The binding assays were asfor Table 2. All the inhibitors (including the column eluates) were used at the concentrations thatproduced 30-70% inhibition of the radioligand binding. Column eluates 1 and 2 were obtained from twodifferent preparations. A - indicates not determined.

Neurobiology: Sangarneswaran and de Blas

Dow

nloa

ded

by g

uest

on

Oct

ober

28,

202

0

Page 5: Demonstration benzodiazepine-like molecules the mammalian1 ,uMchlordiazepoxide(C); 10AM,3CCE(D),and10,uMRo15-1788 (E) on the 21-7F9 binding to the cerebellum. (F) Negative control

5564 Neurobiology: Sangameswaran and de Blas

brane protein besides smaller peptides. If the BZD-likeepitopes present on the immunoblotted peptides are formedby the amino acids and not by other covalently associatedmolecules, then our finds would differ from the ones obtainedwith opiates in that no antibody crossreactivity has beendemonstrated yet between morphine and endogenous opiatepeptides. The smallest peptide that we can detect by immu-noblotting is 20,000 Mr, but we cannot exclude the possibilitythat the brain might contain even smaller BZD-like peptides.Smaller peptides do not bind well to the nitrocellulose paperduring the electrotransfer (37).The immunological approach has a number of advantages,

which include not only the characterization of the peptidessize but also the study of the cellular and subcellularlocalization of the BZD-like molecules and their purificationby immunoafflinity chromatography. This approach has al-lowed us to identify a family of endogenous BZD-likemolecules different from all the other putative endogenousBZDs reported in the literature. Thus, the pharmacologicalproperties of our immunoaffinity-purified molecules differfrom those of the DBI peptide (11): (i) whereas DBI binds tothe ,3-carboline site ofthe BZDR, the immunoaffinity purifiedBZD-like molecules bind to the BZD (agonist) binding site,and (ii) the activity of our preparation, unlike that of DBI, isPronase resistant (14). Thus, unlike DBI and p-carbolines,the molecules recognized by mAb 21-7F9 probably potentiatethe inhibitory effect of GABA on the GABA synapses.The results presented in Tables 2 and 3 also show that the

purified molecules are different from nicotinamide, hypoxan-thine, and purine nucleosides, which have been suggested tobe endogenous BZDs (3-5). Moreover, whereas the com-bined concentrations of these substances in our brain extractwould not exceed 50 ,uM (3-5), their binding affinities formAb 21-7F9 are very low (Kd = 2-10 mM). Even assuming100% recovery after immunoaffinity chromatography, theircombined concentration in the [3H]FNZ binding assay wouldbe less than 100 ,iM. At this concentration, we observed noeffect of the aforementioned substances on the [3H]FNZbinding to the membranes (not shown). In contrast, ourpreparations of endogenous BZD-like molecules still retained>50% of the initial inhibitory activity after a 1:10 dilution.The results indicate that we have identified a class of

endogenous BZD-like neuronal molecules that bind to theneuronal BZDR and that this binding is potentiated byGABA. We cannot yet rule out the possibility that the activesubstance is a BZD that might be present in the controlled ratdiet and accumulates in the brain. However, if these sub-stances are synthesized in the brain, then we predict thathypofunction of the brain endogenous BZD system (i.e.,decreased synthesis or release, increased degradation, ordecreased affinity for the receptor) might lead to pathologicalforms of anxiety and to the forms of epilepsy that respond toBZD treatment.

We thank Holly M. Cherwinski for her excellent technical help.We also thank Drs. Harvey Karten and Jeffrey F. McKelvy for theircomments and discussions and for the use of some of their laboratoryfacilities. We are grateful to Dr. Phil Skolnick of the NationalInstitutes of Health for his generous gift of [3H]diazepam, [3H]Ro5-4864, and [3H],3CCE. We thank Dr. Severo Ochoa for the sugges-tions on the manuscript and Linda Cerracchio and Pam Mackey fortheir help in the preparation of the manuscript. This research wassupported by Grant NS 17708 from the National Institute of Neu-

rological and Communicative Disorders and Stroke and GrantBNS-8401466 from the National Science Foundation.

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

2. Mohler, H. and Okada, T. (1977) Science 198, 849-851.3. Skolnick, P., Marangos, P. J., Goodwin, F. K., Edwards, M.

& Paul, S. M. (1978) Life Sci. 23, 1473-1480.4. Asano, T. & Spector, S. (1979) Proc. Natl. Acad. Sci. USA 76,

977-981.5. Mohler, H., Polc, P., Cumin, R., Pieri, L. & Kettler, R. (1979)

Nature (London) 278, 563-565.6. Braestrup, C., Nielsen, M. & Olsen, C. E. (1980) Proc. Natl.

Acad. Sci. USA 77, 2288-2292.7. Collello, G. D., Hockenbery, D. M., Bosmann, H. M., Fuch,

S. & Folkers, K. (1978) Proc. Natl. Acad. Sci. USA 75,6319-6323.

8. Davis, L. G. & Cohen, R. K. (1980) Biochem. Biophys. Res.Commun. 92, 141-148.

9. Poddar, M. K., Urguhart, D. & Sinha, A. K. (1980) Brain Res.193, 519-528.

10. Woolf, J. & Nixon, J. C. (1981) Biochemistry 20, 4263-4269.11. Guidotti, A., Forchetti, C. M., Corda, M. G., Konkel, D.,

Bennet, C. D. & Costa, E. (1983) Proc. Natl. Acad. Sci. USA80, 3531-3535.

12. Martin, I. L. (1980) Trends Neurosci. 3, 299-301.13. Mohler, H. (1981) Trends Pharmacol. Sci. 1, 116-119.14. Costa, E., Ferrari, M., Ferrero, P. & Guidotti, A. (1984)

Neuropharmacology 23, 989-991.15. Ferrero, P., Guidotti, A., Conti-Tronconi, B. & Costa, E.

(1984) Neuropharmacology 23, 1359-1362.16. Sangameswaran, L., Cherwinski, H. M. & de Blas, A. L.

(1984) Soc. Neurosci. Abstr. 10, 644.17. Mernoff, S. T., Cherwinski, H. M., Becker, J. W. & de Blas,

A. L. (1983) J. Neurochem. 41, 752-758.18. de Blas, A. L., Kuljis, R. 0. & Cherwinski, H. M. (1984)

Brain Res. 322, 277-287.19. de Blas, A. L. (1984) J. Neurosci. 4, 265-273.20. de Blas, A. L. & Cherwinski, H. M. (1983) Anal. Biochem.

133, 214-219.21. de Blas, A. L., Wang, Y.-Y., Sorensen, R. & Mahler, H. R.

(1979) J. Neurochem. 33, 647-659.22. Marangos, P. J., Patel, J., Boulenger, J.-P. &

Clark-Rosenberg, R. (1982) Mol. Pharmacol. 22, 26-32.23. Schoemaker, H., Boles, R., Horst, W. D. & Yamamura, H.

(1983) J. Pharmacol. Exp. Ther. 275, 61-64.24. Gallager, D. W., Mallorga, P., Oertel, W., Henneberry, R. &

Tallman, J. (1981) J. Neurosci. 1, 218-225.25. Bender, A. S. & Hertz, L. (1984) J. Neurochem. 43,

1319-1327.26. Anholt, R. R. H., Murphy, K. M. M., Mack, G. E. & Snyder,

S. H. (1984) J. Neurosci. 4, 593-603.27. Mantione, C. R., Weissman, B. A., Goldman, M. E., Paul,

S. M. & Skolnick, P. (1984) FEBS Lett. 176, 69-74.28. Mohler, H. & Richards, J. G. (1981) Nature (London) 294,

763-765.29. Bloom, F. & Iversen, L. (1971) Nature (London) 229, 628-630.30. Houser, C. R., Hendry, S. H. C., Jones, E. G. & Vaughn,

J. E. (1983) J. Neurocytol. 12, 617-638.31. Somogyi, P., Smith, A. D., Nunzi, M. G., Gorio, A., Takagi,

H. & Wu, J. Y. (1983) J. Neurosci. 3, 1450-1468.32. Seress, L. & Ribak, C. E. (1983) Exp. Brain Res. 50, 173-182.33. Seguela, P., Geffard, M., Buijs, R. M. & LeMoal, M. (1984)

Proc. Natl. Acad. Sci. USA 81, 3888-3892.34. Nakanishi, S., Inoue, A., Kita, T., Nakamura, M., Chang,

A. C. Y., Cohen, S. N. & Numa, S. (1979) Nature (London)278, 423-427.

35. Gray, A., Dull, T. J. & Ullrich, A. (1983) Nature (London)303, 722-725.

36. Sutcliffe, J. C., Milner, R. J. & Bloom, F. E. (1983) ColdSpring Harbor Symp. Quant. Biol. 48, 477-484.

37. Burnette, W. N. (1981) Anal. Biochem. 112, 195-203.38. de Blas, A. L., Sangameswaran, L., Haney, S. A., Abraham,

C. J., Jr., & Rayner, C. A. (1985) J. Neurochem., in press.

Proc. Natl. Acad. Sci. USA 82 (1985)

Dow

nloa

ded

by g

uest

on

Oct

ober

28,

202

0