Proc. Nati. Acad. Sci. USA Vol. 76, No. 2, pp. 977-981, February 1979
Neurobiology
Identification of inosine and hypoxanthine as endogenous ligands for the brain benzodiazepine-binding sites (benzodiazepine receptor/cyclic nucleotide phosphodiesterase/diazepam)
TOMIKo ASANO* AND SYDNEY SPECTORt Roche Institute of Molecular Biology, Nutley, New Jersey 07110
Communicated by John J. Burns, November 1, 1978
Two endogenous ligands for the brain benzodiazepine-binding sites were isolated from bovine brain through gel filtration, paper electrophoresis, and paper chromatography. These ligands were identified as inosine and hypoxanthine, and both had a higher affinity for the brain benzodiazepine-binding sites than for benzodiazepine sites in some peripheral tissues. They did not bind to any other receptors tested, such as the opiate, muscarinic cholinergic, y-aminobutyric acid, and f,-adrenergic receptors. Both inosine and hypoxanthine competitively inhibited the binding of [3H]diazepam to the brain binding site. The benzodiazepine compounds, of which diazepam is a representative, have specific receptors in the central nervous system (1-4). These compounds exert anxiolytic, muscle relaxation, hypnotic, and anticonvulsive effects (5-6). The fact that specific receptors for these compounds exist in the brain has led to the assumption that an endogenous ligand for these receptors must also exist in the brain. The endogenous ligand is assayed by its ability to competitively inhibit the binding of [3H]diazepam to the brain receptors. We have pointed out (7) that the approach we used to isolate an endogenous nonpeptide morphinelike compound with antibodies developed for morphine is generally applicable. That is, antibodies to a drug can be used as a surrogate receptor to isolate an endogenous ligand(s) for receptors to such a drug. Thus, one can use antibodies developed to bind diazepam for isolating substances that can compete with diazepam for binding to available sites on the antibody. This paper will show that inosine and hypoxanthine bind to the benzodiazepine-binding sites in brain and also to diazepam antibodies. By the criteria of activity in binding assays, these two purines can be considered endogenous ligands. ABSTRACT
METHODS Tissue extracts were assayed for their ability to inhibit binding of [3H]diazepam (New England Nuclear, 69.7 Ci/mmol; 1 Ci = 3.7 X 101 becquerels) to a crude membrane preparation of rat cerebral cortex (receptor assay) or to an antibody generated in rabbit against a diazepam immunogen (radioimmunoassay, RIA), as previously described (8). The method of Mohler and Okada (3) was modified for the receptor assay. Rat cerebral cortex was homogenized in 50 vol of iced 50 mM Tris-HCI (pH 7.4) and centrifuged for 10 min at 4000 X g. The pellet was washed once with 50 vol of the buffer and then resuspended in 50 mM Tris-HCI (pH 7.4) so that the final total weight was 10 times the original tissue weight. The suspended pellet was then used for the receptor assay. The assay mixture contained 20 ,ul of membrane fraction (corresponding to 2 mg of original tissue), 50 mM Tris-HCl (pH 7.4), 0.74 nM [3H]diazepam, and
an appropriate amount of tissue extract to give a final volume of 100 til. The nonspecific binding was determined in the presence of 5 ,gM unlabeled diazepam, and was 4-7% of the total binding. The assay mixture was incubated at 00C for 15 min. The incubation was terminated by filtration through a Whatman GF/B glass fiber filter under reduced pressure. The filters were washed with 10 ml of iced buffer and transferred to counting vials that contained 1 ml of NCS solubilizer (Amersham). After 1 hr at 250C, the solubilized filters were shaken vigorously with 10 ml of scintillation cocktail [4 g of Omnifluor (New England Nuclear)/liter of toluene] in counting vials. Binding studies with other 3H-labeled ligands were performed according to published procedures (9-12) with minor modification; 6.2 nM [3H]dihydromorphine, 3.1 nM [3H]quinuclidinyl benzilate, 41.3 nM [3H]muscimol, and 4.0 nM [3H]dihydroalprenolol were used as ligands for the binding to opiate, muscarinic cholinergic, y-aminobutyric acid, and fl-adrenergic receptors, respectively. The membrane fractions of rat kidney and liver were prepared as above. The final pellet was suspended in 50 mM Tris.HCI (pH 7.4) so that the final total weight was 10 times (for kidney) or 5 times (for liver) the original tissue weight. Forty microliters of membrane fraction (corresponding to 4 mg of kidney protein or 8 mg of liver protein) was added into the assay mixture, whose composition was that mentioned previously for the brain. Rabbit antiserum to 5-[3-(4-aminophenylazo)-4-hydroxyphenyl]-7-chloro-1, 3-dihydro-1-methyl-2H-1, 4-benzodiazepine-2-one (Ro 20-9748, Hoffmann-La Roche) (8) was used for the RIA. The assay mixture contained 10 ,tl of normal rabbit serum, 0.02 ,p of antiserum, 0.15 nM [3H]diazepam, 10 mM phosphate-buffered saline (pH 7.4), and an appropriate amount. of tissue extract in a final volume of 0.5 ml. Nonspecific binding was determined by omitting the antiserum and was about 10% of the total binding. The reaction mixture was incubated for 1 hr at 40C. The antibody-bound [3H]diazepam was separated from free [3H]diazepam by the addition of an equal volume of a saturated ammonium sulfate solution (pH 7.4). After 30 min at 40C, the reaction mixture was centrifuged at 2500 X g for 30 min and the supernatant was discarded. The precipitate was washed once with 1 ml of a 50% srte ammonium sulfate solution (pH 7.4), dissolved in 0.5 iml of distilled water, and transferred to a counting vial containing 5 ml of Riafluor (New England Nuclear). The incubation tube was washed twice with 3 ml of Riafluor, the washings being added to the counting vial. Radioactivity was measured in a Beckman LS-250 refrigerated liquid scintillation system. Standard curves were generated for each membrane fraction (receptor assay) or antibody (RIA) by incorporating increasing amounts of unlabeled diazepam into the above assay system. Abbreviation: RIA, radioimmunoassay. * Present address: Institute for Developmental Research, Aichi Prefecture Colony, Kasugai, Aichi, Japan. t To whom correspondence should be addressed.
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.
977
978
Neurobiology: Asano and Spector
Proc. Natl. Acad. Sci. USA 76 (1979)
Linear semilogarithmic plots were obtained in the range of 0.05-5 pmol (receptor assay) or 0.01-2 pmol (RIA) when percent inhibition of [3Hldiazepam binding was plotted versus the concentration of unlabeled diazepam. The endogenous ligands can then be expressed as pmol of diazepam equivalents. The procedure for purification of the endogenous ligands for the benzodiazepine receptor is summarized in Fig. 1. Bovine brain was homogenized in 3 vol of acidified methanol (0.2 ml of concentrated HCI/liter) by using a Waring blender and centrifuged at 8000 X g for 30 min. The supernatant was evaporated under reduced pressure and the residue was dissolved in 0.1 M HCl (1.2 ml/10 g of tissue) and centrifuged again at 40,000 X g for 30 min. The clear supernatant was then lyophilized. The resulting residue was dissolved in methanol (1 ml/5 g of tissue) and centrifuged at 40,000 X g for 20 min. The supernatant was evaporated under reduced pressure and the resulting residue was dissolved in distilled water (1 ml/15 g of tissue). This solution was extracted with an equal volume of diethyl ether and the aqueous phase was lyophilized. The residue was dissolved in distilled water (1 ml/80 g of tissue) and referred to as the "crude extract." The crude extract was applied to a Sephadex G-10 column (2.6 X 50 cm), which was run at a flow rate of 30 ml/hr in distilled water. Gel filtration showed two peaks (peak I and peak II, Fig. 2, as measured by both receptor assay and RIA). The fractions comprising each peak were pooled, lyophilized, and then applied to paper electrophoresis. Paper electrophoresis was carried out on Whatman 3 MM paper at 50 V/cm for 80 min in 5% acetic acid/pyridine (pH 3.5). After drying, the paper was cut into 1-cm strips and extracted in a test tube with distilled water overnight, and the water extracts were assayed by the receptor assay. The extract that contained inhibitory activity Bovine Brain + 3 vol. of acidified methanol hompgenized centrifuged
pellet
supernatant
evaporated |+ 0.1 M HCU centrifuged
supe1natant
pellet
lyophilized |+ methanol
was lyophilized. Peak I after electrophoresis was chromatographed on Whatman 3 MM paper in isobutyric acid/0.5 M NH40H (10:6, vol/vol). After chromatography, the material with inhibitory activity was eluted from the paper with water by the same procedure used after paper electrophoresis and then rechromatographed on the same paper in n-butanol that had been saturated with water. The material having inhibitory activity was extracted from paper with water and used for studies of UV absorption, mass spectrum, and binding to other receptors. Peak II, after electrophoresis, was applied on Whatman 3 MM paper and developed with n-butanol saturated with water, and the inhibitory activity was eluted from the paper with water. This fraction was then used for UV, mass spectrum, and binding studies. Thin-layer chromatography was performed on 20 X 20 cm cellulose plates (Avicel, Analtech, Inc., Newark, DE) developed in n-butanol that had been saturated with distilled water. The plates were scraped in 1-cm bands and the scrapings were eluted with 1 ml of 50 mM Tris-HCl (pH 7.4). The extract was assayed for its ability to inhibit [3H]diazepam binding to both the receptor and the antibody. RESULTS Bovine brain crude extract contained about 70% of the inhibitory activity obtained from the initial brain methanol extraction, as measured by using the receptor assay. On Sephadex G-10 chromatography the crude extract was resolved into two peaks of activity capable of inhibiting [3H]diazepam binding in both the receptor assay and the RIA (Fig. 2). In quantitative terms, peak I represents approximately 15 pmol equivalents/ml of extract by the receptor assay and about 0.3 pmol equivalents/ml by RIA. Peak II had about 40 pmol equivalents/ml as determined by the receptor assay and about 0.2 pmol equivalents/ml as determined by RIA. When peak I was chromatographed on a cellulose thin-layer plate, the inhibitory activities in both receptor assay and the RIA were obtained at a RF of 0.20 (Fig. 3). When peak II was chromatographed on a thin-layer plate, inhibitory activity in the receptor assay was obtained at RF 0.28, but the inhibitory activity in the RIA was too low to detect. Treatment of peak I with 1 M HCI, at 100°C for 1 hr resulted in a marked reduction of peak I (RF 0.20) with a concomitant elevation of peak II (RF 0.28), indicating the conversion of peak I to II.
centrifuged pellet
supernatant
evaporated + H20 + Diethyl ether H20 phase
ether phase
evaporated +
H20
"crude extract"
Sephadex G-10 column Peak I l
paper electrophoresis
I
Peak II l
paper electrophoresis
I
paper chromatography paper chromatography (isobutyric acid - NH40H)(n-butanol - H20) paper chromatography
(n-butanol
-
(Hypoxanthine)
H20)
(Inosine) FIG. 1. Purification of the endogenous ligand for the benzodiazepine receptor.
20 Fraction
FIG. 2. Sephadex G-10 column chromatography of bovine brain extract. Six milliliters of bovine brain extract was chromatographed on a 2.6 X 50 cm column at a flow rate of 30 ml/hr in distilled water. Ten-milliliter fractions were collected and assayed for inhibition of [3H]diazepam binding to both receptor (0) and antibody (0). Bars indicate the fractions in which inosine or hypoxanthine was eluted under the same conditions.
Neurobiology:
Asano and Spector
Proc. Natl. Acad. Sc. USA 76 (1979)
979
Inosine
o 4.,
o
40~~~~~~~~~~~ ._ E ~~~~~~~~~~1.0
CL
200
~~~~~~~~~~~~0.5~
(U
10
5
Solvent front
Origin
0.1 1/diazepam, nM-1
Fraction
FIG. 3. Thin-layer chromatography of peak I from Sephadex G-10. Chromatography was performed on 20 x 20 cm cellulose plates developed in n-butanol saturated with distilled water. The plates were scraped in 1-cm bands and the scrapings were eluted with 1 ml of 50 mM Tris-HCI (pH 7.4). The extract was assayed for inhibition of [3H]diazepam binding to both receptor (0) and antibody (0). Bar indicates the site for inosine.
Peaks I and II remained at the origin during paper electrophoresis. In screening various compounds, it was found that both inosine and hypoxanthine remained at the origin under the same conditions. When peak I was chromatographed on Whatman 3 MM paper in isobutyric acid/0.5 M NH40H (10:6, vol/vol), all the activity in the receptor assay was found at RF 0.5. Inosine had an identical RF. In another solvent system, n-butanol saturated with water, all the activity could be localized at RF 0.12. Again, inosine exhibited identical characteristics. Peak II showed a RF of 0.24 on paper chromatography in n-butanol saturated with water; hypoxanthine had the same RF value. Fig. 4 shows the UV absorption spectra of peaks I and II after elution from the Whatman paper with water. The spectra of
240
260
280
FIG. 5. Double reciprocal analysis of the inhibition of [3H]diazepam binding to brain receptor by inosine. 0, Control; 0, 1 mM inosine; 0, 2 mM inosine.
peak I at pH 2 and 14 are similar to those of inosine, while the spectra of peak II resembled those of hypoxanthine. Mass spectrometric analysis indicated that peaks I and II had mass fragments consistent with an identification as inosine and hypoxanthine. To determine the binding specificities of peaks I and II, extracts after paper electrophoresis and paper chromatography were tested for their abilities to bind to various brain receptors. Table 1 indicates that peaks I and II bound only to the benzodiazepine receptor and not to the opiate, muscarinic cholinergic, y-aminobutyric acid, or ,B-adrenergic receptors in rat brain. Inosine and hypoxanthine bound only to the brain benzodiazepine receptor and did not inhibit ligand binding to the other receptors tested. We knew that inosine and hypoxanthine inhibited about 50% of the binding of [3H]diazepam to the benzodiazepine receptor at a concentration of 1.5 mM; if peaks I and II were indeed inosine and hypoxanthine, they should have given the same degree of inhibition at the same concentration.
300 Wavelength, nm
240
260
280
FIG. 4. UV absorption spectra of peak I and inosine (A) and peak II and hypoxanthine (B).
980
Neurobiology: Asano and Spector
Proc. Natl. Acad. Sci. USA 76 (1979)
Table 1. Effect of the endogenous ligands for the benzodiazepine receptor on other receptors in rat brain % inhibition of specific receptor binding
Substance PeakI PeakII Inosine Hypoxanthine
Concentration, mM
1.5 1.5
A25.*
Benzodiazepine
Muscarinic cholinergic
-y-Aminobutyric acid
fl-Adren-
Opiate
15.3 15.5 16.0 16.7
49 53 53 52
4 3 0 0
2 1 2 2
0 0 0 0
0 3 0 4
ergic
* Concentrations of peaks I and II were adjusted so as to give the same A250 as inosine and hypoxanthine.
Peaks I and II extracted after paper electrophoresis and paper chromatography were quantified on the basis of their absorption at 250 nm, and, as shown in Table 1, the degree of inhibition of binding per mole was the same as that of authentic inosine and hypoxanthine. In order to determine tissue specificity of these compounds, the receptors from rat kidney and liver were tested and compared to those from brain. Table 2 shows that peak I, peak II, inosine, and hypoxanthine had a much lower affinity for the receptors of the peripheral tissues than for those of the brain. A double reciprocal analysis of the inhibition of [3H]diazepam binding in the presence of two concentrations of inosine is shown in Fig. 5. The apparent Km increased with increasing amounts of inosine while maximum binding was unchanged, suggesting that the inhibition was competitive and that the Ki value was 1.0 mM. Hypoxanthine also showed competitive inhibition of binding, with a Ki of 1.1 mM. Table 3 lists other purine and pyrimidine derivatives and their abilities to interact -with the benzodiazepine receptor of rat brain. The addition of a phosphate group to inosine, adenosine, or guanosine reduces the affinity of the nucleoside for the benzodiazepine receptor. Although adenosine had a lower affinity for the receptor than inosine, the adenosine analogues 2-chloroadenosine and 8-methylaminoadenosine had greater affinities. Other purine derivatives such as theophylline, caffeine, and 3-isobutyl-1-methylxanthine have higher affinities for the benzodiazepine receptor than does inosine or hypoxanthine. The pyrimidines uracil and cytosine do not interact with the receptor even at high concentrations. DISCUSSION Since the discovery of benzodiazepine-binding sites in mammalian brain, there has been considerable speculation on the possibility of an endogenous ligand for those sites. Marangos et al. (13) and Karobath et al. (14) have recently presented evidence for endogenous factors that compete with [3H]diazepam binding to brain membranes. The present studies show that brain extracts do indeed contain material that specifically binds to the brain benzodiazepine-binding sites, and two major components have been identified as inosine and hypoxanthine. H. Mohler and T. Okada (personal communication) have also found that this purine and its nucleoside can bind to the brain Table 2. Effect of endogenous ligands on the benzodiazepine receptors from different tissues
Substance Peak I Peak II Inosine Hypoxanthine *
% inhibition of specific receptor binding
Conc., mM
A250*
Brain
Kidney
Liver
1.5 1.5
15.3 15.5 16.0 16.7
54 54 56 56
1 9 2 9
0 10 2 5
Concentrations of peaks I and II were adjusted so as to give the same A250 as inosine and hypoxanthine.
benzodiazepine-binding sites. Braestrup and Squires (2) have shown that, although there are benzodiazepine receptors in some peripheral tissues, they differ from the brain receptors with respect to the pattern of binding of specific benzodiazepine drugs. Inosine and hypoxanthine have binding characteristics similar to clonazepam and diazepam, which are bound to the brain benzodiazepine sites with a much higher affinity than to the kidney or liver binding sites. Having said that inosine and hypoxanthine bind to brain receptors with a higher affinity than they do to peripheral receptors, we must emphasize that the affinity of these substances for brain receptors is low. However, if the concentrations of inosine and hypoxanthine in the brain-are relatively high, then it may not be critical to have a high affinity for the receptors. What is required under such circumstances is that the ligand be specific. The data on the binding of inosine and hypoxanthine to other receptors in the brain indicate that they are specific for the benzodiazepine receptor. Although both inosine and hypoxanthine bind to the benzodiazepine-binding sites, it should be noted (Table 3) that adenosine and guanosine also interact with these sites. It is known that most purines in brain exist as nucleotides and that the adenine nucleotides are the major ones. Many investigators (15-17) have shown that the sum of ATP, ADP, and AMP is about 2.5 ,umol/g wet weight in various mammalian brains. Inosine and hypoxanthine are formed mainly from these adenine nucleotides, normally at low levels. It has been shown that Table 3. Inhibition of [3H]diazepam binding to rat brain membrane receptor by various compounds
Compound Inosine
Hypoxanthine IMP
Adenosine 2-Chloroadenosine Deoxyadenosine 8-Methylaminoadenosine AMP Cyclic AMP ADP ATP NAD Guanosine Isoguanosine GMP GTP Theophylline Caffeine
3-Isobutyl-1-methylxanthine Uracil
Cytosine
IC50, mM* 1.3 1.3 >5 5.2 0.75 0.95 0.86 >5 >5 >5 >5 >5 1.0 2.3 5.6 5.6 0.66 0.47 0.27 >5 >5
* Serial dilutions were used to estimate IC5o values (concentrations causing 50% inhibition of specific [3H]diazepam binding).
Neurobiology:
Asano and Spector
the sum of inosine and hypoxanthine in cat brain is about 0.05 mM (17). However, it is possible that the levels of these compounds can increase during certain physiological states. Kleihues et al. (17) have reported that the levels of adenine nucleotides can decrease dramatically while levels of adenosine, inosine, and hypoxanthine increase markedly in cat brain after ischemia. The content of inosine and hypoxanthine after an ischemia was found to be 0.73 Amol/g wet weight (17). This concentration is close to the Ki values of inosine and hypoxanthine. Dickman et al. (18) have also reported that the ATP level in brain decreases as a consequence of a stress caused by electroshock. They have not determined the levels of inosine and hypoxanthine, but their results suggest that during stress there is an increased formation of these compounds, which may then act on the brain benzodiazepine receptors in a compensatory mechanism. Table 3 shows that methylxanthines also have the capacity to bind to the brain benzodiazepine-binding sites. The methylxanthines theophylline, caffeine, and 3-isobutyl-1methylxanthine are also known to inhibit cyclic nucleotide phosphodiesterase. Beer et al. (19) have shown that the benzodiazepines also inhibit brain cyclic nucleotide phosphodiesterase. These results suggest that the benzodiazepine receptors are the membrane-bound phosphodiesterase. Therefore, comparison was made of the affinities of diazepam, methylxanthines, papaverine, inosine, and hypoxanthine to the brain benzodiazepine-binding site and the membrane-bound cyclic nucleotide phosphodiesterase. All of the aforementioned compounds inhibited both the binding of [3H]diazepam to the binding site and the activity of the phosphodiesterase. However, there was no relationship between the inhibition of [3H]diazepam binding and the inhibition of phosphodiesterase activity; i.e., diazepam had high affinity for the binding sites and very low affinity for the phosphodiesterase, while papaverine and 3-isobutyl-1-methylxanthine inhibited phosphodiesterase much more effectively than they bound to on the receptor (data are not shown). These studies have demonstrated that both inosine and hypoxanthine can interact with the brain benzodiazepine-binding sites. Under certain physiological conditions these compounds can be found in brain in concentrations that inhibit benzodi-
Proc. Natl. Acad. Sci. USA 76 (1979)
981
azepine binding. It is now necessary to ask whether the binding of these substances correlates with demonstrable pharmacologic effects. We thank Dr. Y. Furuichi of the Roche Institute of Molecular
Biology for his valuable advice in the identification of the endogenous ligands. 1. Squires, R. F. & Braestrup, C. (1977) Nature (London) 266, 732-734. 2. Braestrup, C. & Squires, R. F. (1977) Proc. Nati. Acad. Sci. USA
74,3805-3809. 3. Mohler, H. & Okada, T. (1977) Life Sci. 20,2101-2110. 4. Mohler, H. & Okada, T. (1977) Science 198,849-851. 5. Zbinden, G. & Randall, L. 0. (1967) Adv. Pharmacol. 5, 213291. 6. Randall, L. O., Schallek, W., Sternbach, L. H. & Ning, R. Y. (1974) in Psychopharmacological Agents, ed. Gordon, M. (Academic, New York), Vol. 3, pp. 175-281. 7. Gintzler, A. R., Levy, A. & Spector, S. (1976) Proc. Natl. Acad. Sci. USA 73,2132-2136. 8. Peskar, B. & Spector, S. (1973) J. Pharmacol. Exp. Ther. -186, 167-172. 9. Pert, C. B. & Snyder, S. H. (1973) Proc. Natl. Acad. Sci. USA 70, 2243-2247. 10. Yamamura, H. I. & Snyder, S. H. (1974) Proc. Nati. Acad. Sci. USA 71, 1725-1729. 11. Enna, S. J. & Snyder, S. H. (1977) J. Neurochem. 28, 857-860. 12. Bylund, D. B. & Snyder, S. H. (1976) Mol. Pharmacol. 12, 568-580. 13. Marangos, P. J., Paul, S. M., Greenlaw, P., Goodwin, F. K. & Skolnick, P. (1978) Life Sci. 22, 1893-1900. 14. Karobath, M., Sperk, G. & Schonbeck, G. (1978) Eur. J. Pharmacol. 49, 323-326. 15. Minard, F. N. & Davis, R. V. (1962) J. Biol. Chem. 237, 12831289. 16. Drewes, L. R. & Gilboe, D. D. (1973) J. Biol. Chem. 248, 2489-2496. 17. Kleihues, P., Kobayashi, K. & Hossmann, K. A. (1974) J. Neurochem. 23, 417-425. 18. Dickman, S. R., Harrison, J. F. & Grosser, B. I. (1973) Brain Res.
53,483-487.
19. Beer, B., Chasin, M., Clody, D. E., Vogel, J. R. & Horovitz, Z. P. (1972) Science 176, 428-430.
Neurobiology
Identification of inosine and hypoxanthine as endogenous ligands for the brain benzodiazepine-binding sites (benzodiazepine receptor/cyclic nucleotide phosphodiesterase/diazepam)
TOMIKo ASANO* AND SYDNEY SPECTORt Roche Institute of Molecular Biology, Nutley, New Jersey 07110
Communicated by John J. Burns, November 1, 1978
Two endogenous ligands for the brain benzodiazepine-binding sites were isolated from bovine brain through gel filtration, paper electrophoresis, and paper chromatography. These ligands were identified as inosine and hypoxanthine, and both had a higher affinity for the brain benzodiazepine-binding sites than for benzodiazepine sites in some peripheral tissues. They did not bind to any other receptors tested, such as the opiate, muscarinic cholinergic, y-aminobutyric acid, and f,-adrenergic receptors. Both inosine and hypoxanthine competitively inhibited the binding of [3H]diazepam to the brain binding site. The benzodiazepine compounds, of which diazepam is a representative, have specific receptors in the central nervous system (1-4). These compounds exert anxiolytic, muscle relaxation, hypnotic, and anticonvulsive effects (5-6). The fact that specific receptors for these compounds exist in the brain has led to the assumption that an endogenous ligand for these receptors must also exist in the brain. The endogenous ligand is assayed by its ability to competitively inhibit the binding of [3H]diazepam to the brain receptors. We have pointed out (7) that the approach we used to isolate an endogenous nonpeptide morphinelike compound with antibodies developed for morphine is generally applicable. That is, antibodies to a drug can be used as a surrogate receptor to isolate an endogenous ligand(s) for receptors to such a drug. Thus, one can use antibodies developed to bind diazepam for isolating substances that can compete with diazepam for binding to available sites on the antibody. This paper will show that inosine and hypoxanthine bind to the benzodiazepine-binding sites in brain and also to diazepam antibodies. By the criteria of activity in binding assays, these two purines can be considered endogenous ligands. ABSTRACT
METHODS Tissue extracts were assayed for their ability to inhibit binding of [3H]diazepam (New England Nuclear, 69.7 Ci/mmol; 1 Ci = 3.7 X 101 becquerels) to a crude membrane preparation of rat cerebral cortex (receptor assay) or to an antibody generated in rabbit against a diazepam immunogen (radioimmunoassay, RIA), as previously described (8). The method of Mohler and Okada (3) was modified for the receptor assay. Rat cerebral cortex was homogenized in 50 vol of iced 50 mM Tris-HCI (pH 7.4) and centrifuged for 10 min at 4000 X g. The pellet was washed once with 50 vol of the buffer and then resuspended in 50 mM Tris-HCI (pH 7.4) so that the final total weight was 10 times the original tissue weight. The suspended pellet was then used for the receptor assay. The assay mixture contained 20 ,ul of membrane fraction (corresponding to 2 mg of original tissue), 50 mM Tris-HCl (pH 7.4), 0.74 nM [3H]diazepam, and
an appropriate amount of tissue extract to give a final volume of 100 til. The nonspecific binding was determined in the presence of 5 ,gM unlabeled diazepam, and was 4-7% of the total binding. The assay mixture was incubated at 00C for 15 min. The incubation was terminated by filtration through a Whatman GF/B glass fiber filter under reduced pressure. The filters were washed with 10 ml of iced buffer and transferred to counting vials that contained 1 ml of NCS solubilizer (Amersham). After 1 hr at 250C, the solubilized filters were shaken vigorously with 10 ml of scintillation cocktail [4 g of Omnifluor (New England Nuclear)/liter of toluene] in counting vials. Binding studies with other 3H-labeled ligands were performed according to published procedures (9-12) with minor modification; 6.2 nM [3H]dihydromorphine, 3.1 nM [3H]quinuclidinyl benzilate, 41.3 nM [3H]muscimol, and 4.0 nM [3H]dihydroalprenolol were used as ligands for the binding to opiate, muscarinic cholinergic, y-aminobutyric acid, and fl-adrenergic receptors, respectively. The membrane fractions of rat kidney and liver were prepared as above. The final pellet was suspended in 50 mM Tris.HCI (pH 7.4) so that the final total weight was 10 times (for kidney) or 5 times (for liver) the original tissue weight. Forty microliters of membrane fraction (corresponding to 4 mg of kidney protein or 8 mg of liver protein) was added into the assay mixture, whose composition was that mentioned previously for the brain. Rabbit antiserum to 5-[3-(4-aminophenylazo)-4-hydroxyphenyl]-7-chloro-1, 3-dihydro-1-methyl-2H-1, 4-benzodiazepine-2-one (Ro 20-9748, Hoffmann-La Roche) (8) was used for the RIA. The assay mixture contained 10 ,tl of normal rabbit serum, 0.02 ,p of antiserum, 0.15 nM [3H]diazepam, 10 mM phosphate-buffered saline (pH 7.4), and an appropriate amount. of tissue extract in a final volume of 0.5 ml. Nonspecific binding was determined by omitting the antiserum and was about 10% of the total binding. The reaction mixture was incubated for 1 hr at 40C. The antibody-bound [3H]diazepam was separated from free [3H]diazepam by the addition of an equal volume of a saturated ammonium sulfate solution (pH 7.4). After 30 min at 40C, the reaction mixture was centrifuged at 2500 X g for 30 min and the supernatant was discarded. The precipitate was washed once with 1 ml of a 50% srte ammonium sulfate solution (pH 7.4), dissolved in 0.5 iml of distilled water, and transferred to a counting vial containing 5 ml of Riafluor (New England Nuclear). The incubation tube was washed twice with 3 ml of Riafluor, the washings being added to the counting vial. Radioactivity was measured in a Beckman LS-250 refrigerated liquid scintillation system. Standard curves were generated for each membrane fraction (receptor assay) or antibody (RIA) by incorporating increasing amounts of unlabeled diazepam into the above assay system. Abbreviation: RIA, radioimmunoassay. * Present address: Institute for Developmental Research, Aichi Prefecture Colony, Kasugai, Aichi, Japan. t To whom correspondence should be addressed.
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.
977
978
Neurobiology: Asano and Spector
Proc. Natl. Acad. Sci. USA 76 (1979)
Linear semilogarithmic plots were obtained in the range of 0.05-5 pmol (receptor assay) or 0.01-2 pmol (RIA) when percent inhibition of [3Hldiazepam binding was plotted versus the concentration of unlabeled diazepam. The endogenous ligands can then be expressed as pmol of diazepam equivalents. The procedure for purification of the endogenous ligands for the benzodiazepine receptor is summarized in Fig. 1. Bovine brain was homogenized in 3 vol of acidified methanol (0.2 ml of concentrated HCI/liter) by using a Waring blender and centrifuged at 8000 X g for 30 min. The supernatant was evaporated under reduced pressure and the residue was dissolved in 0.1 M HCl (1.2 ml/10 g of tissue) and centrifuged again at 40,000 X g for 30 min. The clear supernatant was then lyophilized. The resulting residue was dissolved in methanol (1 ml/5 g of tissue) and centrifuged at 40,000 X g for 20 min. The supernatant was evaporated under reduced pressure and the resulting residue was dissolved in distilled water (1 ml/15 g of tissue). This solution was extracted with an equal volume of diethyl ether and the aqueous phase was lyophilized. The residue was dissolved in distilled water (1 ml/80 g of tissue) and referred to as the "crude extract." The crude extract was applied to a Sephadex G-10 column (2.6 X 50 cm), which was run at a flow rate of 30 ml/hr in distilled water. Gel filtration showed two peaks (peak I and peak II, Fig. 2, as measured by both receptor assay and RIA). The fractions comprising each peak were pooled, lyophilized, and then applied to paper electrophoresis. Paper electrophoresis was carried out on Whatman 3 MM paper at 50 V/cm for 80 min in 5% acetic acid/pyridine (pH 3.5). After drying, the paper was cut into 1-cm strips and extracted in a test tube with distilled water overnight, and the water extracts were assayed by the receptor assay. The extract that contained inhibitory activity Bovine Brain + 3 vol. of acidified methanol hompgenized centrifuged
pellet
supernatant
evaporated |+ 0.1 M HCU centrifuged
supe1natant
pellet
lyophilized |+ methanol
was lyophilized. Peak I after electrophoresis was chromatographed on Whatman 3 MM paper in isobutyric acid/0.5 M NH40H (10:6, vol/vol). After chromatography, the material with inhibitory activity was eluted from the paper with water by the same procedure used after paper electrophoresis and then rechromatographed on the same paper in n-butanol that had been saturated with water. The material having inhibitory activity was extracted from paper with water and used for studies of UV absorption, mass spectrum, and binding to other receptors. Peak II, after electrophoresis, was applied on Whatman 3 MM paper and developed with n-butanol saturated with water, and the inhibitory activity was eluted from the paper with water. This fraction was then used for UV, mass spectrum, and binding studies. Thin-layer chromatography was performed on 20 X 20 cm cellulose plates (Avicel, Analtech, Inc., Newark, DE) developed in n-butanol that had been saturated with distilled water. The plates were scraped in 1-cm bands and the scrapings were eluted with 1 ml of 50 mM Tris-HCl (pH 7.4). The extract was assayed for its ability to inhibit [3H]diazepam binding to both the receptor and the antibody. RESULTS Bovine brain crude extract contained about 70% of the inhibitory activity obtained from the initial brain methanol extraction, as measured by using the receptor assay. On Sephadex G-10 chromatography the crude extract was resolved into two peaks of activity capable of inhibiting [3H]diazepam binding in both the receptor assay and the RIA (Fig. 2). In quantitative terms, peak I represents approximately 15 pmol equivalents/ml of extract by the receptor assay and about 0.3 pmol equivalents/ml by RIA. Peak II had about 40 pmol equivalents/ml as determined by the receptor assay and about 0.2 pmol equivalents/ml as determined by RIA. When peak I was chromatographed on a cellulose thin-layer plate, the inhibitory activities in both receptor assay and the RIA were obtained at a RF of 0.20 (Fig. 3). When peak II was chromatographed on a thin-layer plate, inhibitory activity in the receptor assay was obtained at RF 0.28, but the inhibitory activity in the RIA was too low to detect. Treatment of peak I with 1 M HCI, at 100°C for 1 hr resulted in a marked reduction of peak I (RF 0.20) with a concomitant elevation of peak II (RF 0.28), indicating the conversion of peak I to II.
centrifuged pellet
supernatant
evaporated + H20 + Diethyl ether H20 phase
ether phase
evaporated +
H20
"crude extract"
Sephadex G-10 column Peak I l
paper electrophoresis
I
Peak II l
paper electrophoresis
I
paper chromatography paper chromatography (isobutyric acid - NH40H)(n-butanol - H20) paper chromatography
(n-butanol
-
(Hypoxanthine)
H20)
(Inosine) FIG. 1. Purification of the endogenous ligand for the benzodiazepine receptor.
20 Fraction
FIG. 2. Sephadex G-10 column chromatography of bovine brain extract. Six milliliters of bovine brain extract was chromatographed on a 2.6 X 50 cm column at a flow rate of 30 ml/hr in distilled water. Ten-milliliter fractions were collected and assayed for inhibition of [3H]diazepam binding to both receptor (0) and antibody (0). Bars indicate the fractions in which inosine or hypoxanthine was eluted under the same conditions.
Neurobiology:
Asano and Spector
Proc. Natl. Acad. Sc. USA 76 (1979)
979
Inosine
o 4.,
o
40~~~~~~~~~~~ ._ E ~~~~~~~~~~1.0
CL
200
~~~~~~~~~~~~0.5~
(U
10
5
Solvent front
Origin
0.1 1/diazepam, nM-1
Fraction
FIG. 3. Thin-layer chromatography of peak I from Sephadex G-10. Chromatography was performed on 20 x 20 cm cellulose plates developed in n-butanol saturated with distilled water. The plates were scraped in 1-cm bands and the scrapings were eluted with 1 ml of 50 mM Tris-HCI (pH 7.4). The extract was assayed for inhibition of [3H]diazepam binding to both receptor (0) and antibody (0). Bar indicates the site for inosine.
Peaks I and II remained at the origin during paper electrophoresis. In screening various compounds, it was found that both inosine and hypoxanthine remained at the origin under the same conditions. When peak I was chromatographed on Whatman 3 MM paper in isobutyric acid/0.5 M NH40H (10:6, vol/vol), all the activity in the receptor assay was found at RF 0.5. Inosine had an identical RF. In another solvent system, n-butanol saturated with water, all the activity could be localized at RF 0.12. Again, inosine exhibited identical characteristics. Peak II showed a RF of 0.24 on paper chromatography in n-butanol saturated with water; hypoxanthine had the same RF value. Fig. 4 shows the UV absorption spectra of peaks I and II after elution from the Whatman paper with water. The spectra of
240
260
280
FIG. 5. Double reciprocal analysis of the inhibition of [3H]diazepam binding to brain receptor by inosine. 0, Control; 0, 1 mM inosine; 0, 2 mM inosine.
peak I at pH 2 and 14 are similar to those of inosine, while the spectra of peak II resembled those of hypoxanthine. Mass spectrometric analysis indicated that peaks I and II had mass fragments consistent with an identification as inosine and hypoxanthine. To determine the binding specificities of peaks I and II, extracts after paper electrophoresis and paper chromatography were tested for their abilities to bind to various brain receptors. Table 1 indicates that peaks I and II bound only to the benzodiazepine receptor and not to the opiate, muscarinic cholinergic, y-aminobutyric acid, or ,B-adrenergic receptors in rat brain. Inosine and hypoxanthine bound only to the brain benzodiazepine receptor and did not inhibit ligand binding to the other receptors tested. We knew that inosine and hypoxanthine inhibited about 50% of the binding of [3H]diazepam to the benzodiazepine receptor at a concentration of 1.5 mM; if peaks I and II were indeed inosine and hypoxanthine, they should have given the same degree of inhibition at the same concentration.
300 Wavelength, nm
240
260
280
FIG. 4. UV absorption spectra of peak I and inosine (A) and peak II and hypoxanthine (B).
980
Neurobiology: Asano and Spector
Proc. Natl. Acad. Sci. USA 76 (1979)
Table 1. Effect of the endogenous ligands for the benzodiazepine receptor on other receptors in rat brain % inhibition of specific receptor binding
Substance PeakI PeakII Inosine Hypoxanthine
Concentration, mM
1.5 1.5
A25.*
Benzodiazepine
Muscarinic cholinergic
-y-Aminobutyric acid
fl-Adren-
Opiate
15.3 15.5 16.0 16.7
49 53 53 52
4 3 0 0
2 1 2 2
0 0 0 0
0 3 0 4
ergic
* Concentrations of peaks I and II were adjusted so as to give the same A250 as inosine and hypoxanthine.
Peaks I and II extracted after paper electrophoresis and paper chromatography were quantified on the basis of their absorption at 250 nm, and, as shown in Table 1, the degree of inhibition of binding per mole was the same as that of authentic inosine and hypoxanthine. In order to determine tissue specificity of these compounds, the receptors from rat kidney and liver were tested and compared to those from brain. Table 2 shows that peak I, peak II, inosine, and hypoxanthine had a much lower affinity for the receptors of the peripheral tissues than for those of the brain. A double reciprocal analysis of the inhibition of [3H]diazepam binding in the presence of two concentrations of inosine is shown in Fig. 5. The apparent Km increased with increasing amounts of inosine while maximum binding was unchanged, suggesting that the inhibition was competitive and that the Ki value was 1.0 mM. Hypoxanthine also showed competitive inhibition of binding, with a Ki of 1.1 mM. Table 3 lists other purine and pyrimidine derivatives and their abilities to interact -with the benzodiazepine receptor of rat brain. The addition of a phosphate group to inosine, adenosine, or guanosine reduces the affinity of the nucleoside for the benzodiazepine receptor. Although adenosine had a lower affinity for the receptor than inosine, the adenosine analogues 2-chloroadenosine and 8-methylaminoadenosine had greater affinities. Other purine derivatives such as theophylline, caffeine, and 3-isobutyl-1-methylxanthine have higher affinities for the benzodiazepine receptor than does inosine or hypoxanthine. The pyrimidines uracil and cytosine do not interact with the receptor even at high concentrations. DISCUSSION Since the discovery of benzodiazepine-binding sites in mammalian brain, there has been considerable speculation on the possibility of an endogenous ligand for those sites. Marangos et al. (13) and Karobath et al. (14) have recently presented evidence for endogenous factors that compete with [3H]diazepam binding to brain membranes. The present studies show that brain extracts do indeed contain material that specifically binds to the brain benzodiazepine-binding sites, and two major components have been identified as inosine and hypoxanthine. H. Mohler and T. Okada (personal communication) have also found that this purine and its nucleoside can bind to the brain Table 2. Effect of endogenous ligands on the benzodiazepine receptors from different tissues
Substance Peak I Peak II Inosine Hypoxanthine *
% inhibition of specific receptor binding
Conc., mM
A250*
Brain
Kidney
Liver
1.5 1.5
15.3 15.5 16.0 16.7
54 54 56 56
1 9 2 9
0 10 2 5
Concentrations of peaks I and II were adjusted so as to give the same A250 as inosine and hypoxanthine.
benzodiazepine-binding sites. Braestrup and Squires (2) have shown that, although there are benzodiazepine receptors in some peripheral tissues, they differ from the brain receptors with respect to the pattern of binding of specific benzodiazepine drugs. Inosine and hypoxanthine have binding characteristics similar to clonazepam and diazepam, which are bound to the brain benzodiazepine sites with a much higher affinity than to the kidney or liver binding sites. Having said that inosine and hypoxanthine bind to brain receptors with a higher affinity than they do to peripheral receptors, we must emphasize that the affinity of these substances for brain receptors is low. However, if the concentrations of inosine and hypoxanthine in the brain-are relatively high, then it may not be critical to have a high affinity for the receptors. What is required under such circumstances is that the ligand be specific. The data on the binding of inosine and hypoxanthine to other receptors in the brain indicate that they are specific for the benzodiazepine receptor. Although both inosine and hypoxanthine bind to the benzodiazepine-binding sites, it should be noted (Table 3) that adenosine and guanosine also interact with these sites. It is known that most purines in brain exist as nucleotides and that the adenine nucleotides are the major ones. Many investigators (15-17) have shown that the sum of ATP, ADP, and AMP is about 2.5 ,umol/g wet weight in various mammalian brains. Inosine and hypoxanthine are formed mainly from these adenine nucleotides, normally at low levels. It has been shown that Table 3. Inhibition of [3H]diazepam binding to rat brain membrane receptor by various compounds
Compound Inosine
Hypoxanthine IMP
Adenosine 2-Chloroadenosine Deoxyadenosine 8-Methylaminoadenosine AMP Cyclic AMP ADP ATP NAD Guanosine Isoguanosine GMP GTP Theophylline Caffeine
3-Isobutyl-1-methylxanthine Uracil
Cytosine
IC50, mM* 1.3 1.3 >5 5.2 0.75 0.95 0.86 >5 >5 >5 >5 >5 1.0 2.3 5.6 5.6 0.66 0.47 0.27 >5 >5
* Serial dilutions were used to estimate IC5o values (concentrations causing 50% inhibition of specific [3H]diazepam binding).
Neurobiology:
Asano and Spector
the sum of inosine and hypoxanthine in cat brain is about 0.05 mM (17). However, it is possible that the levels of these compounds can increase during certain physiological states. Kleihues et al. (17) have reported that the levels of adenine nucleotides can decrease dramatically while levels of adenosine, inosine, and hypoxanthine increase markedly in cat brain after ischemia. The content of inosine and hypoxanthine after an ischemia was found to be 0.73 Amol/g wet weight (17). This concentration is close to the Ki values of inosine and hypoxanthine. Dickman et al. (18) have also reported that the ATP level in brain decreases as a consequence of a stress caused by electroshock. They have not determined the levels of inosine and hypoxanthine, but their results suggest that during stress there is an increased formation of these compounds, which may then act on the brain benzodiazepine receptors in a compensatory mechanism. Table 3 shows that methylxanthines also have the capacity to bind to the brain benzodiazepine-binding sites. The methylxanthines theophylline, caffeine, and 3-isobutyl-1methylxanthine are also known to inhibit cyclic nucleotide phosphodiesterase. Beer et al. (19) have shown that the benzodiazepines also inhibit brain cyclic nucleotide phosphodiesterase. These results suggest that the benzodiazepine receptors are the membrane-bound phosphodiesterase. Therefore, comparison was made of the affinities of diazepam, methylxanthines, papaverine, inosine, and hypoxanthine to the brain benzodiazepine-binding site and the membrane-bound cyclic nucleotide phosphodiesterase. All of the aforementioned compounds inhibited both the binding of [3H]diazepam to the binding site and the activity of the phosphodiesterase. However, there was no relationship between the inhibition of [3H]diazepam binding and the inhibition of phosphodiesterase activity; i.e., diazepam had high affinity for the binding sites and very low affinity for the phosphodiesterase, while papaverine and 3-isobutyl-1-methylxanthine inhibited phosphodiesterase much more effectively than they bound to on the receptor (data are not shown). These studies have demonstrated that both inosine and hypoxanthine can interact with the brain benzodiazepine-binding sites. Under certain physiological conditions these compounds can be found in brain in concentrations that inhibit benzodi-
Proc. Natl. Acad. Sci. USA 76 (1979)
981
azepine binding. It is now necessary to ask whether the binding of these substances correlates with demonstrable pharmacologic effects. We thank Dr. Y. Furuichi of the Roche Institute of Molecular
Biology for his valuable advice in the identification of the endogenous ligands. 1. Squires, R. F. & Braestrup, C. (1977) Nature (London) 266, 732-734. 2. Braestrup, C. & Squires, R. F. (1977) Proc. Nati. Acad. Sci. USA
74,3805-3809. 3. Mohler, H. & Okada, T. (1977) Life Sci. 20,2101-2110. 4. Mohler, H. & Okada, T. (1977) Science 198,849-851. 5. Zbinden, G. & Randall, L. 0. (1967) Adv. Pharmacol. 5, 213291. 6. Randall, L. O., Schallek, W., Sternbach, L. H. & Ning, R. Y. (1974) in Psychopharmacological Agents, ed. Gordon, M. (Academic, New York), Vol. 3, pp. 175-281. 7. Gintzler, A. R., Levy, A. & Spector, S. (1976) Proc. Natl. Acad. Sci. USA 73,2132-2136. 8. Peskar, B. & Spector, S. (1973) J. Pharmacol. Exp. Ther. -186, 167-172. 9. Pert, C. B. & Snyder, S. H. (1973) Proc. Natl. Acad. Sci. USA 70, 2243-2247. 10. Yamamura, H. I. & Snyder, S. H. (1974) Proc. Nati. Acad. Sci. USA 71, 1725-1729. 11. Enna, S. J. & Snyder, S. H. (1977) J. Neurochem. 28, 857-860. 12. Bylund, D. B. & Snyder, S. H. (1976) Mol. Pharmacol. 12, 568-580. 13. Marangos, P. J., Paul, S. M., Greenlaw, P., Goodwin, F. K. & Skolnick, P. (1978) Life Sci. 22, 1893-1900. 14. Karobath, M., Sperk, G. & Schonbeck, G. (1978) Eur. J. Pharmacol. 49, 323-326. 15. Minard, F. N. & Davis, R. V. (1962) J. Biol. Chem. 237, 12831289. 16. Drewes, L. R. & Gilboe, D. D. (1973) J. Biol. Chem. 248, 2489-2496. 17. Kleihues, P., Kobayashi, K. & Hossmann, K. A. (1974) J. Neurochem. 23, 417-425. 18. Dickman, S. R., Harrison, J. F. & Grosser, B. I. (1973) Brain Res.
53,483-487.
19. Beer, B., Chasin, M., Clody, D. E., Vogel, J. R. & Horovitz, Z. P. (1972) Science 176, 428-430.
