Properties and Roles of the Three Subclasses of Histamine Receptors in Brain

Historically, drugs interacting with histamine (HA) receptors have been, by chance, at the origin of the two major classes of agents presently used in psychiatry and which were introduced before the realization that HA receptors were present in brain.

HA does not readily cross the ‘blood-brain barrier’ and, therefore, the function of HA receptors in the CNS is to mediate the actions of the locally synthesized, stored and released amine. The two types of brain cells (neurones and mast cells) in which these processes occur were identified more than 10 years ago, mainly from indirect approaches, i.e neurochemical and lesion studies (Schwartz, 1975). However, it is only in the last few years that immunohistochemical methods, developed in several laboratories, have allowed them to be visualized, largely confirming and extending these previous studies (reviewed by Watanabe et al. 1985; Pollard et al. 1985a; Steinbusch & Mulder, 1985).

Most histaminergic neurones arise from several magnocellular mamillary nuclei in the ventral part of the posterior hypothalamus. Their varicose axons constitute long, ascending fibre tracts passing through the lateral hypothalamic area, projecting, mainly in an ipsilateral fashion, to almost all regions of di- and tele-encephalon. A minor descending pathway, also emanating from the hypothalamus, projects to mesencephalic and, presumably, more posterior structures. Hence, the general disposition of the histaminergic neuronal system (i.e. a compact cell group and a widespread distribution of long fibres) resembles that of catecholaminergic and serotoninergic systems in brain. As described below, these similarities extend to receptor-mediated intracellular events and, most probably, to functional roles (reviewed by Schwartz et al. 1980a, 1982; Schwartz, Pollard & Quach, 1980b;Schwartz, Garbarg & Pollard, 1986; Hough & Green, 1984).

However, an original feature of cerebral HA is the existence of a non-neuronal, slowly turning over pool, which contributes about 50% of the total amine content, mainly held in mast cells. These connective tissue cells are scarce in brain (but their HA content is relatively high) and seem closely associated with vascular elements; in addition, cerebral microvessels themselves may also store HA. HA might be released from these non-neuronal stores, for example during inflammatory processes, and thereby may control blood flow and vascular permeability in brain. In any case, the multiple cellular localizations of HA in brain suggest multiple messenger actions, thus accounting for the presence of specific receptors.

Two major breakthroughs have dominated the history of HA pharmacology: the discovery of ‘antihistamines’ (now designated H₁-receptor antagonists) by Bovet & Staub (1937) and that of H₂-receptor antagonists by Black et al. (1972). These two classes of compounds, as well as more or less selective HA receptor agonists (Ganellin, 1982), have not only established that HA actions are mediated by two distinct molecular entities but also provided physiologists with invaluable tools to delineate the functions of the amine: for example the presence of both H₁ and H₂-receptors in brain and their participation in the stimulation of cyclic AMP formation was shown in 1975 (Baudry, Martres & Schwartz, 1975). Furthermore, in the same way as selective Hpreceptor antagonists, like mepyramine, have paved the way for the identification of the H₂-receptor, the utilization of various HA agonists, or antagonists, has facilitated the characterization of the H₃-receptor when their potencies at H₁ and H₂-receptors (expressed by relative potencies of agonists and apparent dissociation constants of antagonists) were taken into consideration (Arrang, Garbarg & Schwartz, 1983).

Fig. 1 shows the structures of a few characteristic agents acting at HA receptors (see also reviews by Ganellin, 1982, 1985; Durant et al. 1985) and their potencies on reference systems (i.e. guinea-pig ileum contractions, atrium rate and inhibition of HA release from brain slices for H₁-, H₂- and H₃-receptors, respectively). It is apparent that no highly selective Hi-receptor agonist, and H₃-receptor agonist or antagonist is as yet available. From a practical point of view it should be emphasized that Hi-receptor antagonists are, in general, highly lipophilic molecules which easily penetrate the blood-brain barrier. However, this is not the case for any other agents which act at HA receptors. Hence pharmacological tools which modify cerebral histaminergic transmission in vivo, are still largely lacking for either experimental or therapeutic purposes.

Binding techniques (which have been so fruitful in recent years in investigating the properties of a variety of central nervous receptors) have also been applied to HA receptors.

Hill, Young & Marrian (1977) demonstrated that [³H]mepyramine can be used as a selective ligand of H preceptors in homogenates of smooth muscle of the guinea-pig ileum. In this preparation, there is very good agreement between the affinity constants, determined for a range of ligands from the inhibition of [³H]mepyramine binding, and the values derived from competitive antagonism of the contractile response to HA.

Specific binding (defined as that inhibited by a micromolar concentration of a non-radioactive H₁ antihistamine, promethazine or triprolidine), occurs with the expected high affinity (K_D less than 1 nmol I⁻¹) in guinea-pig brain but with a lower affinity (K_D more than 5 nmol I⁻¹) in rat brain (Chang, Tran & Snyder, 1978a; Hill & Young, 1980). The pharmacological specificity of the binding sites (assessed by establishing the inhibitory potencies of a variety of histaminergic and non-hista-minergic antagonists) leaves little doubt that they represent the recognition moiety of Hpreceptors in the guinea-pig. There are, however, some discrepancies for rat brain. Stereospecificity is shown by the large difference in inhibitory potency of the two chlorpheniramine enantiomers, although the active one is an order of magnitude less potent in the rat than in the guinea-pig. It is likely therefore, that [³H]mepy-ramine labels, with high affinity, a heterogenous population of sites in rat, which may mainly, but not only, represent H₁receptors. HA itself presents a relatively low inhibitory potency towards [³H]mepyramine binding. However, in the micromolar range, its estimated affinity is roughly consistent with its median effective concentration (EC50) for H₁preceptor-mediated stimulation of either cyclic AMP accumulation (Palacios, Garbarg, Barbin & Schwartz, 1978a) or [³H]glycogen hydrolysis in brain slices (Quach, Duchemin, Rose & Schwartz, 1980a).

The mean density of [³H]mepyramine binding sites in the brain of several species is about 100 fmol mg⁻¹ protein, but there are considerable differences in their regional distribution: for instance, although the cerebellum shows the highest density in the guinea-pig brain (Hill, Emson & Young, 1978), it exhibits the lowest in the rat brain (Chang et al. 1978b). This rather unusual situation remains to be clarified.

Studies of receptors at the molecular level commenced with the advent of selective radioligands. These not only facilitate purification of macromolecules but even allow biochemical studies to be carried out when they are still in their membrane-bound state. For example, covalent chemical modifications of receptor proteins have been useful in obtaining information about functional groups. The most widely used are reagents modifying sulphhydryl groups and disulphide bonds. Agonist and antagonist binding to receptors are often affected differentially, suggesting a role of thiol groups or disulphide bonds in changes of conformation of the receptor protein elicited only by agonists (see review by Strauss, 1984). The reactive groups may play a critical role for ligand binding as well as in transduction. In the case of HA, receptor alkylation of sulphhydryl groups by N’-ethylmaleimide (NEM) inhibits the Hi-receptor-mediated contractions of rabbit aorta, whereas reduction of disulphide bonds by dithiothreitol potentiates both this response (Fleisch, Krzan & Titus, 1973, 1974) and the guinea-pig ileum contraction (Glover, 1979). However, the specificity of the dithiothreitol effects was recently questioned (Fontaine, Famaey & Reuse, 1984) and the effects of these reagents on intact cell preparations cannot be easily interpreted.

The effect of the thiol alkylating agent NEM on HA Hi-receptors from guineapig cerebellum, labelled with [³H]mepyramine, has been investigated (Yeramian, Garbarg & Schwartz, 1985). The properties of [³H]mepyramine binding (apparent dissociation constant and maximal number of sites) were not modified by prior treatment of the membranes with 2 or 5 mmol I⁻¹ NEM. This treatment did not change the inhibition curves of α-chlorpheniramine or mianserin, two H_rreceptor antagonists. In contrast, treatments of membranes with NEM significantly decreased the IC₅₀ values (concentrations at which binding is inhibited by 50%) of HA and the slope indexes (pseudo-Hill coefficients) of HA inhibition curves, which became less than unity. These effects were irreversible; their extent was related to the duration of NEM treatment and to NEM concentration. A computer analysis of the data indicated that some of the H_rreceptors were converted from a state of low affinity for the amine (IC₅₀ value of 75μmoll⁻¹) to a high affinity state (IC₅₀ value of 2μmoll⁻¹). The change was less marked for partial agonists than for HA. The NEM-induced change was observed in the presence and absence of Na⁺ ions which are known to decrease the affinity of HA for H₁ receptors. Agonists or antagonists did not protect against the modification of HA affinity induced by NEM.

NEM treatment might alkylate a critical thiol group located outside the ligand binding domain of the Hi-receptor and thereby stabilize the latter in a different conformation from that of the activated state.

Some progress has already been reported towards purification of the Hi-receptor: it is easily solubilized and can be retained on a lectin affinity column, indicating its glycoprotein nature (Garbarg, Yeramian, Korner & Schwartz, 1985). Sucrose gradient and gel filtration experiments indicated an M_r of approximately 430000 for the receptor-digitonin complex and the isoelectric point of the receptor is 4·8 (Toll & Snyder, 1982).

In view of the low abundance of neurotransmitter receptors in the mammalian CNS, sensitive probes are important tools for biochemical and localization studies. For this purpose an iodinated, highly sensitive, probe for the H₁-receptor was recently developed (Korner et al. 1986). [¹²⁵I]iodobolpyramine is a mepyramine derivative with increased affinity for H₁-receptors of guinea-pig cerebellum (K_D 0·15 nmol I⁻¹) and a 100-fold higher specific radioactivity. Together with a rather low non-specific binding, these features ensure a 50-to100-fold increase in sensitivity of assays of H₁-receptors in membranes or in solubilized form.

Furthermore, [¹²⁵I]iodobolpyramine allows a detailed autoradiographic mapping of Hi-receptors in guinea-pig brain to be established (Fig. 2). Well-contrasted autoradiographic pictures, with low non-specific binding, were obtained after only 2 days’ exposure. The highest grain densities were found in the molecular layer of the cerebellum. This agrees with both autoradiographic (Palacios, Young & Kuhar, 1979; Palacios, Schwartz & Garbarg, 1981a) and biochemical (Hill et al. 1978) data with [³H]mepyramine. Functional H₁-receptors seem to be present in the cerebellum (Daum, Downes & Young, 1984) which, otherwise, contains a low L-histidine decarboxylase activity (Schwartz, Lampart & Rose, 1970). In the hippocampal formation, a projection field for histaminergic neurones (Barbin, Garbarg, Schwartz & Storm-Mathisen, 1976; Haas et al. 1978), the distribution of [¹²⁵I]iodobol-pyramine autoradiographic grains was apparently similar to that of [³H]mepyramine binding sites (Palacios, Wamsley & Kuhar, 1981b).

Other areas of very high H₁-receptor densities are the nucleus accumbens (not previously reported) and the whole thalamus (Hill et al. 1978). There was more or less uniform labelling of the whole cerebral cortex with some higher densities in superficial layers (laminae H-IH). In rat brain, on the other hand, [³H]mepyramine binding sites were mainly located in lamina IV of the temporal cortex (Palacios et al. 1981b). The widespread occurrence of Hj-receptors in the whole cerebral cortex could conceivably account for the sedating (‘mental clouding’) properties of H_r antihistamines (Quach et al. 1979). Moderate to rather high labelling was found in the colliculi, mesencephalic central grey and ventral tegmental area, as well as in the bed nucleus of the stria terminalis. Very high levels of L-histidine decarboxylase activity have been reported in the three last areas (Pollard et al. 1978; Ben Ari et al. 1977). In the hypothalamus, several mamillary nuclei were distinctly labelled with [³H]mepyramine (Palacios et al. 1981b). Localization of H₁-receptors in this region, where HA cell bodies of the ascending pathway have been localized by lesion (Schwartz et al. 1982) and immunohistochemical studies (Watanabe et al. 1984; Steinbusch & Mulder, 1984; Panula, Yang & Costa, 1984; Pollard et al. 1985a; Pollard, Pachot & Schwartz, 1985b), raises the possibility (but does not prove) that H₁-receptors may be present on HA cell bodies or dendrites.

In general, the localization of Hj-receptors (visualized with [^12SI]iodobolpyramine in the guinea-pig brain) is consistent with the known histaminergic projections and HA cell bodies or dendrites and this agrees well with studies using [³H] mepyramine in rat brain. However, there is also a species difference in regional distribution. The regions with the highest density of H₁receptors in guinea-pig brain, namely cerebellum, thalamus and nucleus accumbens, were hardly labelled with [³H]mepyramine in rat brain (Palacios et al. 1981b).

Furthermore, mepyramine and several other antagonists display a higher affinity for H₁receptors in the guinea-pig than in rat or mouse brain (Chang et al. 1978b, Hill & Young, 1980; Quach et al. 1980a). We consistently found bolpyramine to be about four times more potent in guinea-pig than in rat cerebral membranes and there was no evidence for significant [¹²⁵I]iodobolpyramine-specific binding to rat brain membranes. This negative result can be tentatively explained by the combination of a lower affinity of [¹²⁵I]iodobolpyramine, a relatively high level of non-specific binding and a limited number of binding sites in rat brain membrane preparations. Specific binding was also not seen in tissue sections.

These observations strongly confirm the species differences in Hi-receptor pharmacological specificities while, at the same time, indicating that the new ligand will not be usable for rat tissues.

[³H]Mepyramine can also be used to label H₁-receptors in the brain of the mouse in vivo (Quach et al. 1980b; Schwartz et al. 1980a). Thus, a few minutes following intravenous administration of [³H]mepyramine in low doses, a saturable binding occurs in vivo which presents characteristics of regional heterogeneity and pharmacological specificity paralleling those observed in binding studies in vitro. The density of [³H]mepyramine binding sites in vivo agrees well with their density estimated in vitro, indicating that no substantial loss of Hi-receptors occurs during the preparation of membranes used for the in vitro test. A major interest in the in vivo test is that it shows that systemic administration of most H]-antihistamines in doses currently used in therapeutics to alleviate allergic symptoms results in the occupation of a major fraction of H₁receptors in the brain. This observation strongly suggests that the well-known central effects of H₁-antihistamines (such as sedation ‘mental clouding’, increased sleep duration, EEG changes resembling those elicited by barbiturates) are, indeed, mediated by blockade of the actions of endogenous HA at H₁-redeptors. Interestingly, mequitazine, an H₁-antihistamine that a double-blind study has shown to be devoid of sedative properties (Gervais, Gervais, De Beule & Van Der Bijl, 1975), does not occupy H1-receptors in brain, probably because it does not cross the blood-brain barrier. It should also be noted that several psychotropic compounds, classified as either antidepressants (doxepin, amitriptyline) or neuroleptics (levomepromazine, droperidol), but which are all characterized by sedative properties in patients, significantly inhibit [³H]mepyramine binding (in therapeutic doses) in the brain of the mouse in vivo.

With H₂-receptor labelling by radioligands, the situation seems much less favourable than in the case of H₁ -receptors.

A saturable binding of [³H]cimetidine to homogenates from guinea-pig or rat brain has been described, but it is unlikely that this ligand labels H₂-receptors (Warrander, Norris, Rising & Wood, 1983). This conclusion is based on several pieces of evidence: the of cimetidine differs according to the preparation and to the authors (range 40 – 400 nmol I⁻¹) and was 10–100 times lower than its K₁ regarding HA antagonism at well-defined H₂-receptor systems; the inhibitory potency of various histaminergic agents did not reflect their pharmacological potency: ranitidine, for example, gave no detectable inhibition even at 1 mmol I⁻¹; there was no clear-cut regional distribution of the binding sites. Other attempts to label H₂-receptors have been performed with [³H]ranitidine (Bristow, Hare, Hearn & Martin, 1981), [³H]impromidine or [³H]tiotidine (unpublished observations from our laboratory), also apparently without success. More recently, a successful binding of [³H]tiotidine to H₂-receptors was reported, a difference attributed by the authors to the availability of a chemically more stable batch of the ligand (Gajtkowski, Norris, Rising & Wood, 1983). Although specific [³H]tiotidine binding was relatively low as compared to non-specific binding, its pharmacology was well correlated with that of reference H₂-receptor-mediated responses. The distribution of binding sites between regions of guinea-pig brain (striatum > cerebral cortex > hippocampus, no binding in cerebellum or brainstem) was markedly different from that of H₂-linked adenylate cyclase and inconsistencies were reported in peripheral tissues (Norris, Gajtkowski & Rising, 1984; Rising & Norris, 1985).

Labelling the H₂-receptors is likely to be more successfully undertaken with the availability of new [³H]-labelled ligands that present higher selectivity and affinity than those currently available.

Another approach to HA receptors in brain has used [³H]HA itself as a ligand. In general [³H]-agonists cannot be utilized as easily as [³H]-antagonists for receptor labelling because, in most cases, the antagonists display much higher affinities. However, high-affinity binding sites for catecholamine or serotonin agonists, which are likely to represent true receptors, have been described during the past few years. [³H]HA associates with rather high affinity (K_D 10 nmol I⁻¹) to membrane fractions from rat brain (Palacios et al. 1978b; Barbin et al. 1980). The binding sites have several characteristics which suggest that they might represent HA receptors. For example, many other putative neurotransmitters or HA metabolites fail to displace [³H]HA, while various agonists (N^α,N^α-dimethylHA; dimaprit) are reasonably potent inhibitors of binding. In addition, several other correlative features also suggest that these sites represent postsynaptic receptors: the marked regional heterogeneity, the localization into subcellular fractions enriched in synaptosomal membranes, the developmental pattern during ontogenesis paralleling development of putative histaminergic synapses and the decrease following administration of the neurotoxin kainic acid (Barbin et al. 1980).

However, the [³H]HA binding sites might well correspond to a class of ‘desensitized state’ HA receptors as suggested by the decrease of the maximal capacity of [³H]HA binding sites in the presence of guanyl nucleotides; this effect occurs in particulate fractions as well as in a solubilized preparation.

The pharmacological specificity of [³H]HA binding sites differs from that of typical H₁-and H₂-receptors. This is clearly shown by the relative potencies of agonists: impromidine is 10 times less potent than HA on these sites, although it is 10-50 times more potent on H₂-receptors; 2-thiazolylethylamine is 1000-fold less potent than HA, although it is only four times less potent on H₁-receptors. Moreover, the inhibition of [³H]HA binding by H₁ or H₂-antagonists such as mepyramine or metiamide yielded data indicating that [³H]HA may bind to a unique site with cooperativity, or to more than one site. The latter hypothesis is supported by the observation that [³H]HA binding is only partially decreased in the presence of guanyl nucleotides (Barbin et al. 1980). Recently, a heterogeneous binding has also been suggested, with a fraction occurring at H₂-receptors (Wells et al. 1985), but the identity of the [³H]HA binding sites largely remains to be elucidated.

The occurrence of H₁-receptors in nervous tissues was first demonstrated by pharmacological analysis of a series of cellular responses induced by HA. These responses require intact cells (or vesicular sacs). Whereas activation of the H₁ receptor alone does not stimulate cyclic AMP formation, it potentiates the action of direct activators of adenylate cyclase at either adenosine (Daum, Hill & Young, 1982) or HA H₂-receptors (Palacios et al. 1978a) by a complex mechanism that will be discussed below. H₁-receptors also mediate more directly other effects of HA, like the activation of cyclic GMP formation, as mainly studied in neuroblastoma cells (Richelson, 1978) and lysis of [³H]glycogen in slices from cerebral cortex (Quach et al. 1980a). Although these responses seem quite diverse, they all depend upon the availability of Ca²⁺, suggesting that the primary action of H₁-agonists on all cell types (including those of peripheral smooth muscles that they contract) is to induce increases in the intracellular concentration of Ca²⁺, by which other intracellular events, like activation of guanylate cyclase or glycogen phosphorylase, are triggered.

However, changes in intracellular Ca²⁺ elicited by H₁-receptor stimulation, although likely, have never been directly shown, even when monitored in neuroblastoma cells with the luminescent protein aequorin (Richelson, 1985).

Nevertheless, the theory of a link between H₁-receptor activation and Ca²⁺ mobilization has recently received strong support. Following initial observations that HA stimulates the incorporation of inorganic [³²P] into cerebral phospholipids in vivo, H₁-receptor-mediated stimulation of [³H]phosphatidylinositol 4,5-bisphos-phate breakdown was shown in several laboratories (Daum, Downes & Young, 1983; Brown, Kendall & Nahorski, 1984; Donaldson & Hill, 1985) by monitoring the accumulation of [³H] inositol phosphate in the presence of Li⁺, and inositol phosphatase inhibitor.

The HA-sensitive adenylate cyclase which, in a variety of tissues, appears to be uniformly coupled to H₂-receptors, was first characterized in P. Greengard’s laboratory on membranes from guinea-pig hippocampus (reviewed by Schwartz et al. 1982; Johnson, 1982). Both subcellular fractionation (Kanof, Hegstrand & Green-gard, 1977) and lesion studies (Garbarg, Barbin, Palacios & Schwartz, 1978) are consistent with a postsynaptic localization of the HA-sensitive, cyclic-AMP-generating system in brain. This suggests that at least some synaptic actions of HA are mediated by a rise in the intracellular level of the cyclic nucleotide and the relationship between the two series of cellular events has started to be clarified with the electrophysiological studies of H. Haas in Zurich (reviewed in Haas, 1985).

Stimulation of H₂-receptors only slightly depolarizes pyramidal cells of slices of rat hippocampus, but profoundly potentiates a variety of excitatory signals like depolarizations induced by excitatory amino acids, synaptically-evoked spikes, epileptiform bursts. These potentiations result from a disinhibitory effect of HA, consisting of a blockade of long-lasting hyperpolarizations or of accommodations of firing which, otherwise, restrict the amplitude and duration of excitatory signals (Haas & Konnerth, 1983).

This modulatory effect, attributable to a decrease of calcium-activated potassium conductance, is potentiated by a phosphodiesterase inhibitor and mimicked by intracellular application of cyclic AMP. Similar modulatory effects occur following stimulation of β-adrenoreceptors, also known to be coupled with adenylate cyclase. All these observations strongly suggest that the rise in nucleotide level elicited by HA is responsible for the change in conductance. However the link between the two processes remains unknown since HA-induced protein phosphorylation, via a cyclic-AMP-dependent protein kinase, has never been described.

This mode of action, together with the highly divergent topography of HA neurones, may indicate that the function of the latter is not to transmit discrete information but to increase the sensitivity of large cerebral areas to excitatory inputs. This is consistent with the hypothesis that HA might be a ‘waking amine’ through the activation of H₂-receptors (as well as of H₁-receptors mediating the potentiation of the cyclic AMP response).

An H₂-receptor-mediated stimulation of cyclic AMP formation also occurs in cerebral microvessels (Kamushina et al. 1980) and might be related to the increased capillary permeability and brain oedema elicited by HA (Gross, 1985).

Although HA maximally stimulates (via H₂-receptors linked to adenylate cyclase) cyclic AMP formation by only 2-to 3-fold in a cell-free system, the corresponding response in brain slices (which involves both H₁ and H₂-receptors) can be as high as a 10-to 20-fold stimulation (Palaciosel al. 1978a). The participation of H₁-receptors in cyclic AMP accumulation in intact cells thus results in a striking amplification of the intracellular message generated by HA.

The H₁-receptor-mediated amplification seems to involve a dual mechanism. This is first shown by the influence of Ca²⁺ ions in the H₁-receptor-mediated response in hippocampal slices. In the absence of extracellular Ca²⁺ the H2-receptor-mediated stimulation of cyclic AMP accumulation (evaluated by the response to dimaprit, a selective H2-agonist) is not affected, whereas the H₁-receptor-mediated stimulation (e.g. the response to 2-thiazolylethylamine) is reduced by about 50% (Fig. 3). This suggests that Ca²⁺ translocation is partly involved in the H₁-receptor-mediated amplification.

The dual mechanism is also suggested by the effects of phorbol esters (i.e. protein kinase C activators). As previously discussed, H₁ receptor agonists stimulate phosphatidylinositol hydrolysis (reviewed by Carswell, Daum & Young, 1985), an effect which is known to result in the release of two distinct intracellular messengers (Fig. 4). The first is inositol 1,4,5-trisphosphate (IP3), a messenger mobilizing Ca²⁺ from its intracellular stores, for example in the endoplasmic reticulum (ER). The second is diacylglycerol (DAG). This activates protein kinase C and thereby promotes the phosphorylation of intracellular regulatory proteins. Phorbol esters like 4β-phorbol 12,13-dibutyrate (PDB) readily enter intact cells and mimic the effect of DAG on protein kinase C. PDB strongly potentiated, in a concentration-dependent manner, the effects of dimaprit (Table 1) or HA (Hollingsworth, Sears & Daly, 1985). This suggests that protein kinase C activation results in the phosphorylation of a protein regulating the H₂-receptor-linked adenylate cyclase. The latter has not yet been identified but, by analogy with other systems, could be a guanylnucleotide-binding protein.

It is important to emphasize, however, that, even in the presence of PDB at maximally activating concentration, 2-thiazolylethylamine (or HA) was still effective in inducing its H₁-receptor-mediated amplification (Table 1). This suggests that IP3 (the other messenger released by H₁-receptor-activated phosphatidyl inositol hydrolysis) may also play a role in the amplification process. Alternatively, H₁ receptors might mediate an increased influx of extracellular Ca²⁺, although this effect is not directly documented. In this case the amplification of the cyclic AMP response could result from an action of the Ca²⁺-calmodulin complex at the level of the catalytic unit of cerebral adenylate cyclase (Coussen, Haiech, D’Alayer & Monneron, 1985).

In summary, although the molecular details of the interaction are not yet completely unravelled it appears that the amplification of the H_z-receptor-mediated response involves a dual mechanism triggered by H]-receptor stimulation: protein kinase C activation and Ca²⁺ mobilization.

Elevated extracellular K⁺ concentration has been largely used as a depolarizing stimulus for release of various neurotransmitters and, in the case of either endogenous HA (Taylor & Snyder, 1973) or endogenously synthesized [³H]HA (Ver-dière, Rose & Schwartz, 1975), results in a Ca²⁺-dependent HA release. It is well established that several neurotransmitters may control their own release from cerebral neurones by interacting with presynaptic autoreceptors (Langer, 1977; Starke, 1981). To assess the existence of such a self-regulation process of HA release in brain, we developed a model in which slices of rat cerebral cortex were labelled by preincubation with [³H]-L-histidine (³H-L-His), to allow for [³H]HA synthesis, and subsequently depolarized in a non-superfused system. The cerebral cortex was selected for this study as various approaches have shown that HA synthesis in this region occurs in terminals of histaminergic extrinsic neurones (Baudry et al. 1973; Garbarg, Barbin, Feger & Schwartz, 1974; Barbin, Hirsch, Garbarg & Schwartz, 1975).

Depolarization by 30 mmol I⁻¹ K⁺ of cortical slices in the absence of any exogenous HA added, resulted in the release of about 15% of tissue [³H]HA. However, when non-radioactive exogenous HA was present in the medium, an inhibition of the K⁺-induced [³H]HA release was observed (Arrang et al. 1983). The inhibitory action of HA was concentration-dependent and saturable with a maximal inhibition of 60% (Fig. 5A).

The EC₅₀ value of HA was of 0·13 ± 0·02μmoll⁻¹, indicating that HA in this response is about 100 times more potent than in H₁ or H₂-receptor-mediated responses.

The inhibitory action of HA was only mimicked by its two Nαand2Vα,2Vα-methyl derivatives: two agonists, both of H₁ and H₂-receptors.

In contrast, neither specific H] agonists, such as 2-methyl-HA, nor specific H₂ agonists, such as dimaprit, had any inhibitory effect (Fig. 1). Autoinhibitory receptors show a strong stereoselectivity as demonstrated with two HA derivatives including a chiral centre: α,Nαdimethyl HA and Nα-methyl-α-chloromethyl HA. Both compounds acted as weak agonists and the activity ratio of their respective enantiomers was higher than at H₁ and H₂-receptors. Moreover, at autoinhibitory receptors (+)-isomers of both structures (absolute configuration corresponding to S-configurated L-histidine) possessed greater activity, whereas at H₂-receptors (–)-isomers were more active (Arrang, Schwartz & Schunack, 1985c).

Impromidine, a selective and potent H₂ agonist, not only failed to mimic the inhibitory action of HA, but increased the K⁺-evoked release of [³H]HA. This facilitatory action of impromidine suggested that it might have been acting as an antagonist at autoinhibitory receptors towards endogenous HA released together with [³H]HA by the K⁺ stimulus. Indeed, the inhibitory action of exogenous HA on release was antagonized in a competitive manner by impromidine or burimamide (Arrang et al. 1983).

The antagonism was always surmountable by increasing the HA concentration and was compatible with a competitive nature. Again, a facilitation of [³H]HA release similar to that elicited by impromidine alone was observed in the presence of both agents in high concentrations. In contrast, various H_rantihistamines like mepyr-amine or chlorpheniramine were ineffective in concentrations at which they block H₁-receptors (Fig. 1).

Also the two potent H₂-antihistamines, tiotidine and ranitidine were almost ineffective at autoinhibitory receptors, whereas burimamide and its lower homologue, norburimamide, were about 100 times more potent than at H₂-receptors. Impromidine acted as a potent antagonist at autoinhibitory receptors (K_i value 0·06 ± 0·01 μmoll⁻¹).

Thus, from the potencies of agonists and the apparent dissociation constants of antagonists, as well as from the lack of effect of antagonists of other neurotransmitters, it can be concluded that the autoinhibition of HA release in brain is mediated by a novel class of HA receptor that we recently proposed to term H₃ (Arrang et al. 1983).

H₃-receptors seem to be directly located on HA-synthesizing axon terminals and not on interneurones or nearby axon terminals impinging on the former (Arrang, Garbarg & Schwartz, 1985b).

Hence, the unchanged inhibitory effect of exogenous HA in brain slices, in which the traffic of action potentials was blocked by tetrodotoxin, tends to exclude the possibility of participation of interneurones. This is confirmed by the persistence of H₃-receptor-mediated effects in striatal slices after a kainate treatment. This treatment abolishes the HA-induced cyclic AMP response mediated by H₁ and H₂-receptors. It does not, however, modify the L-histidine decarboxylase activity, a presynaptic marker (Garbarg et al. 1978).

In addition, experiments with rat brain cortex synaptosomes left little doubt that pharmacologically-identified H₃-receptors were directly located on HA-synthesizing terminals. Thus, exogenous HA inhibited [³H]HA release from depolarized synaptosomes, with an EC50 value (0·20 ± 0·05μmol 1⁻¹) similar to that found with slices (Fig. 5A) and, on the two preparations, the effect was inhibited by impromidine, a competitive H₃-receptor antagonist with similar apparent K_i values, e.g. 0·11 ± 0·03/xmoll⁻¹ and 0·06 ± 0·01 μmoll⁻¹, respectively (Fig. 5B).

In conclusion, all these data support the hypothesis that HA modulates its own release from cerebral neurones by interacting with H₃-presynaptic autoreceptors.

As previously shown (Verdière et al. 1975) with hypothalamic slices, depolarization of slices from cerebral cortex in the presence of 30 mmol I⁻¹ K⁺ for 30 min elicited a 100% increase in synthesis of [³H]HA from [³H]-L-His. However, when non-radioactive exogenous HA was present in the medium, an inhibition of this K⁺-induced response occurred. The inhibitory effect of HA was concentration-dependent and saturable. As in the modulation of its release, HA in this response was about 100 times more potent (EC₅₀ value 0·34 ± 0·03 μmol I⁻¹) than in H₁ or H₂-receptor-mediated responses in the brain.

Impromidine and burimamide acted as antagonists of HA with apparent dissociation constants similar to those found at H₃-autoreceptors modulating HA release, whereas mepyramine and tiotidine were again ineffective (Fig. 6A).

Although of limited amplitude, a similar inhibitory effect of exogenous HA (EC₅₀ value 0·10 ± 0·06 mol I⁻¹) was observed in cortical synaptosomes in which a facilitatory effect of impromidine occurred (Fig. 6B).

These data strongly suggest that presynaptic H₃-autoreceptors also modulate HA synthesis in brain (Arrange al. 1985a).

It is not clear at which biochemical level this control occurs. A direct retro-inhibition of the enzyme L-histidine decarboxylase by its product, HA, was previously excluded, whereas it was suggested that control of HA synthesis might occur at the step of L-His uptake into HA-synthesizing cells (Verdière et al. 1975). However, the latter uptake process cannot easily be studied because of the presumably small number of HA neurones in brain as compared with other cells in which active uptake of L-His occurs.

All the current data support the hypothesis that HA modulates its own release from cerebral neurones by interacting with H₃-presynaptic autoreceptors, displaying a highly specific pharmacology and stereoselectivity. The activation of these autoreceptors seems to involve mechanisms bearing many similarities with those by which other neurotransmitters self-regulate their release. H₃-autoreceptors also mediate the modulation of HA biosynthesis in cerebral neurones. The development of compounds able to stimulate or to block these autoreceptors selectively, might provide a useful tool for elucidating the precise functions of histaminergic neurones in brain.