Neuropharmacology of Endogenous Opioid Peptides

John J. Wagner and Charles I. Chavkin

Interest in the physiologic and pharmacologic actions of the endogenous opioid peptides spans the full range of neuroscience research. Investigative approaches ranging from whole-animal/behavioral studies to the molecular study of opioid action are being employed to gather data at the preclinical level. The principal goal of these studies is to define the role of the endogenous opioid neuropeptides in brain function. Ultimately, information concerning the normal actions of this neuropeptide transmitter system can be related to clinical topics such as opiate tolerance and dependence, respiratory and cardiovascular regulation, learning and memory processes, analgesia and nociception, and epileptogenesis and seizure pathology, to name a few.

Following the biochemical purification and characterization of the three endogenous opioid peptide families (the enkephalins, the endorphins, and the dynorphins; see ref. (1), much progress has been made in mapping the specific locations of opioid peptides and their binding site distributions throughout the brain (49, 56). Pharmacological studies have shown that multiple types (mu, delta, kappa) of opioid receptors exist, and they have indicated that subtypes (i.e., mu1, mu2, kappa1, kappa2, delta1, delta2, etc.) are also likely to occur (75). Exogenous application of opiates and opioid peptides has been used to study drug actions down to the molecular level of ion channel modulation and enzymatic regulation (12, 19, 69, 73). As a model system of putative neurotransmitter peptides, the study of endogenous opioids has the advantage of a relatively extensive list of pharmacologic tools (selective agonists and antagonists) compared to the other peptidergic systems. Despite this significant advantage, two general areas of opioid research have lagged behind. First, receptor purification and cloning has been slow, and results leading to the cDNA and amino acid sequence for the opioid receptors has only recently been obtained, some 18 years after the endogenous ligands were discovered (31, 50). Second, elucidation of the specific actions of opioid peptides released from endogenous stores has also been only recently described (11, 95, 97) in the central nervous system (CNS). Recent and continuing advances in these general areas should ensure a prominent place for opioids in future neuropharmacological research.

In this chapter, we review the evidence addressing the properties of opioid peptides as neurotransmitters in the CNS. Opioid peptide processing, opioid receptor heterogeneity, opioid involvement in seizures/excitotoxicity, and mechanisms regarding opiate tolerance and addiction are areas which are covered elsewhere in this volume (see Intracellular Messenger Pathways as Mediators of Neural Plasticity, The Psychopharmacology of Sexual Behavior, Animal Models of Drug Addiction, and Opioids) or in other reviews (1, 39, 86). Here, we will be discussing the pharmacologic actions (i.e., of exogenously applied opiates and opioid peptides) and the physiologic actions (i.e., of endogenously released opioid peptides) of opioids in the mammalian CNS, with an emphasis placed on investigations done using hippocampal tissue when appropriate.

The principal thesis of this chapter is that recent studies of endogenous opioid action in the hippocampus have provided a better understanding of how these peptides normally function as transmitters. The key questions about the nature of the opioid peptide synapse concern the anatomical aspects (where are the sites of release and sites of action?), the physiological aspects (what are the kinetics of opioid action and the effects on the neural circuitry?), and the molecular aspects (which receptors and signal transduction mechanisms are involved?). Our current understanding can be summarized as follows: (a) Opioid peptides are stored in dense-core vesicles in specific neurons which also contain a classical fast-acting transmitter such as glutamate; the opioids thus act as co-transmitters serving to modulate the actions of the primary transmitter. (b) The opioids are released into the extracellular space following prolonged depolarization of the neuronal membrane; thus, opioids modulate synaptic transmission only under conditions of intense afferent input. (c) The structure of the opioid peptide synapse may be larger than that of the fast-acting amine transmitters because the peptides may diffuse some distance from their sites of release to their sites of action; this is evident from the slow kinetics of the opioid peptide action (seconds until onset of effects, minutes until termination of the effect). A wider radius of action from the release site follows from the considerably higher affinities the opioids have for their receptors than the amine transmitters have for theirs. (d) Opioids activate a variety of signal transduction processes; different mechanisms are evident in different cell types. Evidence supporting each of these assertions is presented below. This model of neuropeptide function is clearly consistent with earlier ideas based on actions of neuropeptides in the peripheral nervous system (45, 54), and recent work has provided an increasingly detailed description of the unique properties of neuropeptide transmitters. (Reproductive actions of opioids are described in The Psychopharmacology of Sexual Behavior.)

The modulation of opioid ligand binding by physiologic salts, guanosine phosphates, and regulation of GTPase and adenylate cyclase activities by opioids has been extensively reviewed (see refs. 19, 38, 53, and 75). We wish to briefly summarize the results derived using in vitro assays which suggest that opioid receptors are likely to be members of the G-protein-coupled "superfamily" of receptors. The binding of ligands to all of the three well-accepted types of opioid receptors (mu, delta, and kappa) is modulated by the addition of GTP to radioligand binding assays. GTP significantly decreases the affinity of binding sites for opioid agonists without greatly affecting the binding of antagonists. Similarly, sodium also differentially decreases agonist binding, an effect which is additive with that of GTP. Mu, delta, and kappa agonists have all been shown to stimulate GTPase activity in a naloxone-sensitive manner. Opioid agonists inhibit the activity of adenylate cyclase, an effect which is GTP-dependent and pertussis-toxin-sensitive. The ability of antisera selective for various G proteins to inhibit opioid binding, GTPase activity, and the inhibition of adenylate cyclase activity is also consistent with opioid receptors being coupled to G proteins (see ref. 19).

The question of which G proteins are activated is still extensively debated (see ref. 91). Opioid receptor effects have been shown to be pertussis-toxin-sensitive, implicating members of the Gi or Go type in mediating opioid action (see ref. 19). However, a cholera-toxin-sensitive activation of adenylate cyclase in a neuroblastoma cell line by mu or delta receptors was also recently shown (26), suggesting that activation of Gs may mediate opioid effects in some cells. In addition, whether the alpha subunit of the G-protein complex acts alone or whether the beta/gamma subunits contribute to the signal transduction process has not been resolved.

An elusive goal of many research groups has been to achieve the purification or cloning of an opioid receptor protein (see ref. 86). This information would convincingly determine the validity of the hypothesis that opioid binding sites are G-protein-coupled receptors. The cloning and amino acid sequence has recently been obtained for delta opioid receptors (31, 50), and related mu and kappa clones, isolated by hybridization screening, have also been reported recently (18, 105). The sequences of the cloned opioid receptors supports the hypothesis that these are also members of the G-protein-coupled receptor superfamily.

The list of molecular targets of the activated G proteins is extensive (Fig. 1). Opioids have well-defined effects on adenylate cyclase activity and ion channel conductance mediated by G-protein activation (see below). In addition to the well-known inhibition of adenylate cyclase, some evidence has indicated that an opioid-mediated enhancement of this activity can also be demonstrated at low agonist concentrations (25, 36). Evidence describing opioid regulation of additional second messenger systems has also been reported. In particular, kappa agonists affect the turnover of phosphatidylinositol (PI). Both positive (76) and inhibitory (64) effects on PI turnover have been reported in the hippocampus and cerebellum, respectively. A biphasic modulation of PI turnover has been studied in rat brain cell cultures (3). In 7-day-old cultures, PI formation was inhibited by kappa agonists, whereas in 21-day-old cultures, PI formation was enhanced. It is clear from this description that a unitary mechanism of opioid receptor signal transduction is unlikely. (Also see Cholinergic Transduction, Signal Transduction Pathways for Catecholamine Receptors, and Serotonin Receptors: Signal Transduction Pathways for additional details of G-protein coupled receptors.)

Opioids primarily act as inhibitory agents in the neural circuit (although examples of excitation are noted below). Opiates and opioid peptides have been shown to inhibit the release of a wide range of neurotransmitters from many brain regions in neurochemical release assays (69). These studies usually involve measuring the release either by preloading neuronal stores with radiolabeled transmitter or by specific radioimmunoassays of the perfusate. Electrical or chemical depolarization of the preparation is then performed in the absence and presence of applied opioid compounds. In the hippocampus, mu agonists have been shown to inhibit the release of norepinephrine and acetylcholine (43, 44, 98). Kappa agonists have also been shown to inhibit the release of both norepinephrine and acteylcholine (43, 44, 98). Delta agonists may be effective in inhibiting the release of norepinephrine in guinea pigs (98) but not in rats (43, 44, 69). Using a mossy fiber synaptosomal preparation, very high concentrations of kappa agonist could inhibit both dynorphin and glutamate release (35).

A series of studies describing biphasic effects of low and high concentrations of opioids on enkephalin release from the guinea-pig myenteric plexus illustrates the potential for diversity of opioid actions (36). Low concentrations (<10 nM) of mu, delta, or kappa agonists enhanced stimulated release, whereas higher concentrations (>10 nM) inhibit the stimulated release of enkephalin. Importantly, the inhibitory and enhancing effects could be dissociated based on the class of G protein coupled to the respective effects. Pertussis toxin, which inactivates Gi and Go classes of G proteins, could block the inhibitory effects of high opioid concentrations, leaving the enhancement by low concentrations uneffected. The converse was true when cholera toxin treatment was used to decrease Gs activity. Enhancement of release by low concentrations of opioids was blocked by cholera toxin, whereas the inhibitory effects of higher opioid concentrations was unaffected (36).

These studies have provided clear examples of the effectiveness of exogenously applied opiates and opioid peptides in the various preparations, but only a few examples of an electrophysiological correlate of these actions for endogenously released opioids have been reported (11, 95 97; see below). One recent paper has described the ability of endogenously released opioids to inhibit the release of endogenous NE in hippocampal tissue (85). The methods of the study involved the use of an in vitro competition slice binding assay, which allows the release of endogenous transmitters to be quantified (70). The distinction between the effects of exogenously applied opioids and effects of endogenously released opioid peptides is important, because it is not necessarily true that a given population of receptors occupied by bath-applied drug are relevant targets for opioid peptides released from endogenous stores.

In this section, we discuss the effects of exogenously applied opioids on neuronal activity. The regulation of ion channels by opioids has been intensely studied by several laboratory groups in various neuronal preparations (see refs. 12 and 60). We will be emphasizing the more recent advances in this area. In general, opioids have been found to inhibit neuronal excitability via two mechanisms: inhibition of calcium conductances and enhancement of potassium conductances (4,73). Initially, reports indicated that mu and delta receptors modulated potassium currents, whereas kappa receptors modulated calcium currents. The simplicity of this arrangement was nullified by the findings that delta agonists could inhibit calcium currents in cultured NG-108 cells (90) and that mu agonists could inhibit calcium currents in spinal ganglion neurons (82). Thus far, a clear example of kappa modulation of a potassium current has not been reported; however, it seems likely that this will eventually be shown.

The characterization of the potassium current enhanced by opioids in the locus coeruleus has indicated that the opioid-coupled current is of the inward rectifying type (100). This potassium current is blocked by cesium and is resistant to cadmium and tetrodotoxin. It is also barium-sensitive, and cAMP does not seem to be involved in opioid receptor modulation of the current. Instead, a direct, G-protein-coupled arrangement is suggested by single-channel studies (65, 83). Being an inward rectifier, the opioid-induced conductance is much greater at potentials below EK. The increase in potassium conductance has been shown to directly hyperpolarize and inhibit the firing frequency of locus coeruleus neurons.

Another potassium current, which is kinetically distinct from the inward rectifier, has been characterized in nonpyramidal hippocampal neurons (101). This one is a voltage-gated potassium current activated at positive potentials relative to the resting membrane potential and was insensitive to barium. In addition, cAMP application blocked the opioid-induced current, suggesting that a protein kinase mechanism may be involved in the modulation of this potassium current (102). It is interesting to note that this type of potassium current, unlike an inward rectifier, is not expected to affect firing threshold or frequency. Instead, by activating at depolarized potentials, the delayed rectifier would enhance the repolarization phase and reduce the duration of a firing burst and may thus change the firing pattern of these nonpyramidal cells.

Complex effects of opioids on a third type of potassium conductance (IM) have also been observed by Siggins and co-workers (67) in CA3 pyramidal cells. Kappa agonists [U50488h and micromolar concentrations of dynorphin A(1-17)] increase the amplitude of this potassium current in a norbinaltorphimine sensitive way; whereas, delta agonists and low concentrations (20–100 nM) of dynorphin A(1-17) reduce IM.

In addition to these actions, one group reported concentration-dependent, biphasic effects of opioids on potassium and calcium currents in sensory neurons (25). In these studies, low concentrations of opioid agonist (<10 nM) had the converse effect (i.e., inhibition of potassium current or enhancement of calcium current) with respect to the typical effect of higher concentrations (>10 nM, enhancement of potassium current or inhibition of calcium current). Pertussis-toxin-sensitive G proteins were shown to be involved in the inhibitory actions, whereas a cholera-toxin-sensitive G protein mediated the excitatory actions of low opioid concentrations (25). As with the release studies of Gintzler and Xu (36) discussed above, these results point out the potential for complex variations in the mechanisms of endogenous opioid actions. No analogous, direct excitatory action of opioids in the hippocampus has yet been described.

A novel mechanism of mu opioid action involving regulation of the N-methyl-D-aspartate (NMDA) receptor was reported in trigeminal cells (17). In this study, a mu agonist prolonged and enhanced the NMDA-receptor-mediated current. The opioid effect was blocked by protein kinase C inhibitors and mimicked by protein kinase C activation, suggesting that mu effects are coupled to protein kinase C activity in this preparation. This work provides a clear example of a mechanism by which direct, excitatory actions of mu agonists could occur. Such a mechanism could potentially be important for opioid involvement in synaptic plasticity and excitotoxicity, processes in which NMDA currents are known to be involved. A second example of an interaction between mu opioids and NMDA currents was shown in acutely dissociated rat spinal dorsal horn neurons; here mu receptor activation first depressed then potentiated the effects of applied NMDA (80). The mechanisms of these complex effects were not determined. A similar suppression of NMDA-induced responses was seen in the dentate gyrus of the rat hippocampus (104). In the latter report, application of a mu agonist decreased the synaptic NMDAmediated excitatory postsynaptic current (EPSC) evoked by perforant path stimulation. The nature of the interaction was not characterized, but because non-NMDA EPSCs were not affected by mu agonist, a presynaptic action of inhibition of glutamate release (as discussed for kappa agonists below) did not appear likely. Although the mechanistic basis for the interactions between opioid and NMDA responses is not clear, the observed phenomena suggest that opioid receptors may indirectly modulate responses to other transmitters by activating other, less-well-defined systems. (See Electrophysiology, for a review of these ion channels and their regulation.)

Predominantly inhibitory actions of exogenously applied opioids on neuronal activity and transmitter release are found throughout the CNS (30). The hippocampal region is one exception to this general finding, because pyramidal neuron excitability is increased following opioid application (71). (Fig. 2) It is important to note that no direct effects of opioids on CA1 pyramidal cells themselves have been reported (72, 84). This paradoxical effect was explained via a mechanism of "disinhibition" in which opioids were acting directly to inhibit inhibitory interneurons, resulting in excitation of the pyramidal neurons (55, 107). This basic mechanism of opioid action in the hippocampus has been investigated by a number of researchers, and their findings have been summarized (12, 16). In an extension of this work, recent studies have provided evidence that opioid receptors are specifically present on the terminals of interneurons, acting to inhibit gamma-aminobutyric acid (GABA) release. Lambert et al. (51) monitored pharmacologically isolated, monosynaptic inhibitory postsynaptic potentials (IPSPs) recorded from CA1 pyramidal cells following mu agonist application. Mu agonist application close to the recording site inhibited IPSPs, and application near the site of stimulation in the radiatum was ineffective. This indicated that the relevant mu receptors involved in inhibition of the IPSPs were located very near to the pyramidal cell bodies—that is, on GABAergic terminals. This report also found that barium application was not able to inhibit this opioid effect. This is of interest because evidence from the locus coeruleus region suggested that GABAB and opioid agonists modulated the same potassium conductance (20). The GABAB-coupled potassium current is blocked by extracellular barium, so it would appear that the same potassium conductance is not involved in the mu effects observed by Lambert et al. (51). Two other groups have studied the effects of opioid application on tetrodotoxin (TTX)-insensitive spontaneous miniature IPSPs and found a decrease in the frequency of these events (22, 79). Baclofen, which is known to hyperpolarize interneurons (55), did not affect the frequency of these action potential-independent IPSPs. Therefore, the decrease in IPSP frequency is consistent with the opioid acting at the interneuron terminal, not at the soma (22). This is not to say that opioid actions at the level of the interneuron soma may not be important for disinhibition, but somatic hyperpolarization is unlikely to be the only site of opioid action on interneurons.

In addition to disinhibition resulting from the reduction of GABA release, electrophysiological evidence for opioid inhibition of glutamate release has also been recently described. Kappa opioid agonists were found to inhibit excitatory synaptic transmission at the perforant path–granule cell synapse in guinea-pig hippocampal slices (94, 95). In contrast to the mu and delta agonist effects described above, kappa agonists had no effect on inhibitory synaptic potentials, indicating that the kappa action was not mediated via interneurons. No changes in the membrane properties of the postsynaptic cell were observed, so a direct inhibition of the granule cells was unlikely. Excitatory synaptic potentials elicited by stimulation in the molecular layer were, however, significantly inhibited. Excitatory postsynaptic potentials (EPSPs) elicited from hilar stimulation were not significantly affected, arguing against a change in the postsynaptic response to endogenously released glutamate (95). The sum of these results argues that kappa receptors are located on perforant path terminals and act presynaptically to inhibit glutamate release in the dentate molecular layer of guinea-pig hippocampal slices. A similar presynaptic mechanism for kappa opioid action in the locus coeruleus has been described (60, 77). Pharmacologic activation of mu receptors was similarly shown to inhibit excitatory amino acid release from primary afferent fibers in rat spinal cord (48). Thus, opioid receptor regulation of excitatory transmission may be a very general mechanism of opioid action.

Because of the dense opioid (largely dynorphin) immunoreactivity localized to dentate granule cell axons, the mossy fiber–CA3 pyramidal cell synapse has long been thought to be a likely site for endogenous opioid action. Defining an effect of kappa ligands in the CA3 region has been enigmatic, because excitatory, inhibitory, nonopioid, and opioid-mediated actions have all been reported by various groups (2, 10, 42, 66, 96).

Up to this point, we have only been considering the actions of exogenously applied opioids and opioid peptides in various neurochemical and neurophysiological assays, the rationale being that endogenously released opioid peptides are likely to act via similar mechanisms. A key question is, How closely do the neurotransmitter actions of endogenous opioid peptides resemble pharmacologically applied opioids? Under what physiologic conditions are the peptides released? Once released, given the large variety of actions of opioids described above, what will the "net" effect of opioid action be on neuronal activity? Answering these questions has not been a simple exercise. Much of the anatomical data concerning the relationship between opioid peptide and binding site localization in the hippocampus suggest that endogenous opioids are not likely to act in a classical synaptic manner, and the nonclassical properties are only just being defined. The evidence that opioid peptides can act as neurotransmitters from several studies done using hippocampal tissue and employing a wide range of techniques will be summarized below.

Endogenous opioid peptide distribution in the hippocampal formation has been described in several studies using immunohistochemical techniques (32, 49, 62). The results have shown that enkephalin-immunoreactivity is present in the lateral perforant path and the mossy fiber pathways, as well as in scattered interneurons. Dynorphin-immunoreactivity is restricted to the dentate granule cells (most prominently in their axonal projection to CA3, the mossy fibers) and a few interneurons in the dentate molecular layer. The distribution of opioid binding sites, which represents the locations of putative opioid receptors, is much more complex (56). The anatomical "mismatch" between localization of opioid peptides and their binding sites is contrary to what would be expected for a classical transmitter such as glutamate, suggesting that they may act in a neurohumoral manner (63). Much of the added complexity concerning binding site distributions is due to the large species variations evident when comparing rat and guinea pigs, for example (63). In addition to the endogenous opioid peptides and their binding sites, discrete localization of degradative enzymes thought to possibly be involved in the termination and/or spatial restriction of opioid peptide action has also been described (78).

Enkephalin and dynorphin peptides and peptide-immunoreactivity have been shown to be present in hippocampal tissue as measured with specific radioimmunoassays and high-pressure liquid chromatography (HPLC) assays (13, 14). These assay methods have been used to measure the relative concentrations present and the amount of peptide released in response to chemically induced depolarization of the tissue. As one would expect for a neurotransmitter, the release of opioid peptides was calcium-dependent (13, 14).

A radioligand displacement assay in which endogenously released opioids compete for opioid binding sites provided a means by which endogenous transmitter release occurring in a single in vitro hippocampal slice could be measured (70). This study also provided autoradiographic evidence suggesting that endogenously released opioids could diffuse to distant sites of action. While using mu- or kappa-selective opioid radioligands, focal electrical stimulation of the slice was utilized to demonstrate the release of endogenous opioids from specific opioid-containing fiber tracts in response to physiologic stimulation paradigms (92, 93). High-frequency trains (>1 Hz) of stimulation were found to be most effective in eliciting opioid peptide release (11, 92). This observation is similar to the requirements for the release of neuropeptide Y (NPY) and teleost luteinizing-hormone-releasing hormone (tLHRH) described in the peripheral nervous system and may represent a common characteristic of neuropeptide release (45, 54).

Autoradiography of kappa1 binding sites in the guinea pig revealed a population of putative kappa receptors which were restricted to the molecular layer of the dentate gyrus (93). High-frequency stimulation of regions containing either the perforant path or mossy fibers elicited the release of endogenous kappa opioids able to compete for kappa1 binding. Importantly, application of a glutamate receptor antagonist could block the effects of perforant path stimulation, but not of the commissural/associational fibers. This result was consistent with granule cells being the likely endogenous source of kappa ligands. Experiments in which dynorphin antisera were found to be effective in blocking the effects of mossy fiber stimulation supported the hypothesis that dynorphin peptides released from granule cells were reaching kappa receptors in the molecular layer (93). These observations indicated that dynorphins could act as negative feedback transmitters on the perforant path afferents, modulating incoming excitatory transmission in the dentate gyrus (93, 94).

Synaptic effects of endogenously released opioids were first demonstrated in the CA3 region of the hippocampus (11). Using the stimulation parameters previously characterized in the radioligand slice binding assay (92), the stimulated release of opioid peptides was found to control IPSP amplitudes measured CA3 pyramidal cells (11). The actions of the released opioids were blocked by naloxone [but not the inactive stereoisomer (+)naloxone] and were pathway-specific (i.e., only high-frequency stimulation of the perforant path had an effect on CA3 IPSPs) (11). Interestingly, the response to opioid release was extremely slow and prolonged, and the onset and duration were in the range of minutes following peptide release. The mechanism of the opioid effect was likely mediated by an effect on norepinephrine release (rather than a direct effect on GABA release) because the effects on IPSP amplitudes were sensitive to propranolol, a beta-adrenergic receptor antagonist (11), and the same stimulation parameters were also shown to directly reduce norepinephrine release in guinea-pig hippocampal slices (85). These results provide an example whereby one extrinsic afferent of the hippocampus (the perforant path) could regulate the actions of another extrinsic afferent (the norepinephrine projection from the locus coeruleus).

A more direct mechanism of endogenous opioid action has been characterized in the dentate gyrus of the guinea pig (95). The approach was based on the findings of two previous studies describing the effects of exogenously applied kappa opioids on EPSPs at the dentate granule cell–perforant path synapse (94) and the effects of high-frequency stimulation-induced release of dynorphins as measured in the "in vitro slice" binding assay (93). Therefore, an attempt to electrophysiologically measure the effects of endogenously released dynorphin on perforant path excitation was made by recording both whole-cell voltage clamp or extracellular population spike responses of dentate granule cells (95). Synaptic responses were monitored before and after a high-frequency stimulus train was given in the hilus. Following granule cell activation via antidromic stimulation, excitatory transmission at the perforant path synapse was reduced. This effect of hilar stimulation was blocked by a kappa-selective antagonist or by a cocktail of dynorphin antisera. Thus, a particular family of endogenous opioid peptides (i.e., the dynorphins) was identified, as well as a specific type of opioid receptor (i.e., the kappa receptor; see ref. 95). Although granule cells contain the vast majority of dynorphin-immunoreactivity found in the hippocampus, the specific site of dynorphin release was not determined. Release either from recurrent collaterals or from dendrites of the granule cells is possible. Functionally, as the postsynaptic cell is the source of a molecule acting on the presynaptic terminal, dynorphin is acting as a retrograde transmitter. This inhibitory modulation of synaptic transmission mediated by dynorphin may be important in the pathophysiology of the hippocampus, because kappa opioids have been shown to be neuroprotective in excitotoxicity studies (39). In this role, endogenously released dynorphin could be acting in a compensatory manner to limit hyperexcitability existing during seizure activity (89). Consistent with this hypothesis is the phenomena of mossy fiber sprouting which occurs in human temporal lobe epilepsies (40). Contrary to expectations, the time course of this proliferation of mossy fiber recurrent collaterals into the inner molecular layer following a kindling paradigm could not be correlated with an increase in granule cell excitability (87). Rather, a recovery of inhibition was seen, which is consistent with the inhibitory influence of dynorphin as we have described above. Thus endogenous dynorphin modulation of synaptic activity in the dentate gyrus region may have a prominent role in the pathological and physiological function of the hippocampus.

A recent study (97) has also described an inhibitory presynaptic mechanism for kappa opioid actions in the CA3 region of the guinea-pig hippocampus, similar to the results of Wagner et al. (95) in the dentate gyrus. As well as the results described above for the perforant path synapse, this report also showed that the response to locally applied exogenous glutamate was not affected by kappa agonist, lending further support for the presynaptic site of action. Additionally, paired pulse facilitation, thought to be a presynaptic phenomena in origin, was enhanced by kappa agonists. Thus the net effect of kappa opioid application in both the CA3 and dentate gyrus regions of the guinea-pig hippocampus was a decrease in excitation of the principal neuron, rather than the net excitation of principal cells described previously (95, 97). In this case, the endogenous dynorphins act to inhibit transmitter release from the same mossy fibers releasing the peptide (autoinhibition) as well as from adjacent mossy fibers (heterosynaptic depression).

Thus it would appear that several of the commonly listed criteria for classifying a molecule as an authentic neurotransmitter have been met in the hippocampus specifically and the CNS in general: (a) Opioid peptides are synthesized in, and localized to, specific neuronal populations. (b) Enzymatic means for degradation of the peptide has been described. (c) These peptides are released in a calcium-dependent manner from neuronal tissue. (d) Specific binding sites, representing putative opioid receptors, have been localized throughout the brain. (e) Pharmacologic effects have been described in response to exogenous application of the endogenous opioid peptides. (f) These effects can be mimicked by opioid agonists and reversed by antagonists. The principal differences between the actions of opioid transmitters and the more classical transmitters in this region are the relatively slow kinetics of opioid action and the larger radius of action. The functional implications of endogenous opioid action will be considered next.

We will begin our discussion by reviewing studies which have described the effects of naloxone application on physiologic responses assayed in the hippocampus. The most extensive body of evidence implicating endogenous opioids as having an active part in synaptic physiology concerns the effect of naloxone on the phenomena of long-term potentiation (LTP) in certain hippocampal pathways (7)). LTP can be described simply as a long-lasting enhancement of synaptic response following an appropriate "inducing stimuli" (5, 7, 47). LTP is typically induced following a high-frequency (>10 Hz) tetani of the pathway of interest, conditions which we have described above as also likely to favor the release of endogenous opioids. Consistent with the excitatory effects of pharmacologically applied opioids in the hippocampus, many studies have demonstrated that LTP induction in the mossy fibers and the lateral perforant path (both of which contain endogenous opioids) is blocked in the presence of naloxone (8, 9, 28, 29, 41, 57, 103). The ability of naloxone to alter LTP induction is in marked contrast to the lack of naloxone effects on single-pulse, low-frequency synaptic events measured in the same opioid-containing pathways (15, 28, 29, 52, 103).

In the mossy fiber pathway, previous studies of both the population spike in slice preparations (41, 57) and the field EPSP response in vivo (28, 29) have demonstrated naloxone block of mossy fiber LTP, while LTP induction at a separate pathway (the commissural/associational fibers) is unaffected. The use of selective opioid receptor antagonists has provided evidence indicating that mu receptors are involved (29). Naloxone had no effect on LTP maintenance or expression when applied after LTP induction has occurred (28). Additionally, dynorphin application augmented mossy fiber LTP (81). Therefore, it was surprising that a recent study has failed to reproduce the naloxone-sensitivity of mossy fiber LTP induction, and in fact showed that a kappa-selective antagonist could facilitate LTP induction under their conditions (97). Interestingly, consistent with an inhibitory effect of endogenously released opioids on synaptic transmission in this region, the study also showed that naloxone blocked the phenomena of heterosynaptic depression (6), which accompanies mossy fiber LTP. The work of Weisskopf and co-workers involved measuring field EPSPs in the stratum lucidum in response to stimulation in the granule cell layer to evoke mossy fiber synaptic potentials in guinea-pig hippocampal slices. Although two prior groups had also used guinea pigs, the CA3 population spike response was monitored in the CA3 pyramidal cell layer (41, 57). This seemingly subtle point could potentially be a basis for the discrepancy, because a dissociation of opioid effects on LTP induction monitored in the molecular layer field EPSPs and the granule cell population spike response has been described in the rat dentate gyrus (9, 103; see below). An additional complication is that the dentate gyrus–CA3 area contains rather complex circuitry, making it very important (and difficult) to isolate and characterize the pathway being studied (21).

The LTP of the lateral perforant path (LPP) has been shown to be blocked by naloxone both in vivo (8, 9) and when using the in vitro slice preparation (103), while LTP of the medial perforant path is unaffected. The potential for delta receptor involvement in lateral perforant path LTP was shown with the use of a delta-selective antagonist to selectively block the field EPSP enhancement without affecting the population spike response increase (9). Conversely, mu receptors were implicated by the ability of a mu-selective agonist to facilitate LTP of the population spike response without significantly affecting the field EPSP (103). Thus both studies revealed a dissociation of opioid effects on population spike and field EPSP LTP induced by lateral perforant path stimulation, and each provided evidence for a specific type of opioid receptor as being responsible.

The mechanism by which the facilitation of LTP induction in the mossy fibers by mu receptors and in the LPP by mu and delta receptors has not been determined, but several hypotheses exist. Based on the large amount of evidence indicating that exogenously applied opioids have disinhibitory effects, it is tempting to assume that is the case for LTP induction as well. Indeed it is well known that disinhibition caused by GABAA antagonists facilitates LTP induction (99). Disinhibition has been proposed to be the underlying mechanism by which mu receptors facilitate LTP of the lateral-perforant-path-evoked population spike (103). This mechanism has not been favored in the case of delta-mediated effects on lateral perforant path LTP (7, 9), and a direct excitatory action at perforant path terminals was proposed (7), but a recent report in which naloxone had no effect on lateral perforant path in the presence of GABAA antagonists suggests that a disinhibitory mechanism may be relevant after all (37). Because the underlying locus of change in mossy fiber LTP appears to be presynaptic (106) and due to mu antagonist effects on the presynaptic phenomena of post tetanic potentiation, it has been suggested that a direct effect on mossy fiber presynaptic terminals is the underlying mechanism of action for opioids at this synapse (29). As reviewed above, however, there is no currently known example of opioids exerting a direct, excitatory effect in the hippocampus that this hypothesis would require to explain the enhancement of LTP induction.

In contrast to the mu- and delta-mediated facilitory effects described above, and consistent with the kappa effects described by Weisskopf et al. (97) in the CA3, we have demonstrated that kappa opioids can have an inhibitory effects on LTP induction in the dentate gyrus of guinea pig (88, 95). Either application of a kappa agonist or high-frequency (50 Hz) stimulus trains delivered in the hilus to induce dynorphin release from dentate granule cells was effective in blocking LTP of the population spike response. Blocking either the transmitter (by prior application of dynorphin-selective antisera) or the receptor (by kappa-selective antagonist) was effective in reversing the effects of hilus stimulation on LTP induction. As discussed previously, exogenously applied kappa agonists act in a manner consistent with presynaptic inhibition. Therefore, we hypothesize that a presynaptic inhibition of glutamate release underlies the ability of endogenously released or exogenously applied kappa agonists to block LTP induction at the perforant path–dentate granule cell synapse.

In summary, numerous studies have described opioid-mediated actions on LTP induction in the two opioid-containing pathways in the hippocampal formation. Although discrepancies among reports have occurred, it can be noted once again that endogenous opioid actions are correlated with high-frequency stimulation events. If one assumes that the phenomena of LTP is representative of learning and/or memory mechanisms, then by analogy with their interactions with LTP, endogenous opioids should potentially affect these processes as well.

Opiate antagonists typically facilitate learning and memory, suggesting that endogenous opioids negatively affect these processes (61). Unfortunately, most of the studies have involved systemic application of opioids and opiate compounds, rather than local application to specified regions within the CNS. Thus although it is possible to distinguish between peripheral and CNS-mediated effects by utilizing opiate antagonists which do not cross the blood–brain barrier, the specific site of opioid action remains obscure. Studies done testing both peripheral and central administration of agonists and antagonists indicate that peripheral endogenous opioid systems are important in some forms of conditioning (58). Because we have been emphasizing work done in the hippocampus, the discussion in this section will be limited to the effects of opioid antagonists and the effects of stimulating opioid-containing pathways in experimental paradigms likely to involve hippocampal function.

The hippocampus has been identified as a brain structure necessary for the performance of certain spatial memory tasks (68, 74). Once the maze procedure is learned, the animal relies on spatial cues to make the correct procedural choices in a given trial to obtain the reward. Post-training peripheral naloxone administration has been shown to enhance the acquisition of novel information presented when the spatial cues surrounding the maze apparatus are altered (33, 34). This enhancement appears to be task-specific because post-training naloxone treatment does not enhance acquisition during the initial training of the animal (27, 34). The endogenous opioids do not appear to be implicated in procedural memory (i.e., "learning how" to navigate the maze efficiently), but they do alter declarative memory (i.e., "learning where" the reward is) for a given trial. In addition to studies involving opioid antagonists, direct electrical stimulation of the hippocampal formation also affects spatial memory performance (23, 24). High-frequency stimulation (60 Hz) of the dentate cell layer retrogradely impairs the performance of spatial memory tasks, an effect that is blocked by pretreatment with systemic application of naloxone. Interestingly, stimulation in the CA3 or CA1 layers (cells which do not contain endogenous opioids) did not retrogradely affect declarative memory (24). Given the information concerning opioid peptide localization and release parameters described above, a hypothesis in which dynorphins are released from granule cells following highfrequency stimulation and then act to impair spatial memory is easily supported. This hypothesis gains further support from a study done comparing the spatial memory performance and dynorphin content in young, middle-aged, and aged rats (46). In this study, the dynorphin content in the hippocampal formation was significantly elevated in middle-aged and aged rats compared to young animals. Importantly, animals exhibiting impaired spatial memory performance had significantly higher dynorphin A(1-8) levels than did their unimpaired, age-matched cohorts. Thus elevated dynorphin predicted impaired spatial memory performance in middle-aged and aged animals (46), and electrical stimulation of dynorphin-containing cells in young animals also impaired spatial memory (23, 24).

In the years since their first discovery, extensive progress has been made in the characterization of the transmitter properties of the endogenous opioid peptides. The clearest results are at the cellular and molecular level where the inhibitory mechanisms activated by opioids are best described. Less well defined is the modulatory role of opioids in the neural circuit where questions about the physical dimensions of the "opioid peptide synapse" and the kinetics of endogenous opioid action should be resolved to provide a clearer insight to the role of these neuropeptides in synaptic function. Ultimately, new high-resolution techniques must be developed to define the roles of the opioid peptides in the whole animal. But clearly, considering the prevalence of peptides present in the brain, an understanding of how neuropeptides function in the control of behavior is essential to the description of complex neural function.

published 2000