I. Introduction

Top Next References

Our understanding of chemical signal transmission between neurons and between axon terminals and target cells has advanced significantly since Elliott (1904), Loewi (1921), and Dale (1934) first elaborated on the concept that epinephrine and acetylcholine (ACh)² are released from the neuron and may be able to transmit signals toward target cells. Today, our knowledge of how information is conveyed chemically from one cell to another has been heavily influenced by textbook data on the neuromuscular junction (Katz, 1969), in which the transmitter ACh is stored in vesicles and released into the junctional gap in quanta. This system is adopted for very fast signaling: The information transfer occurs within millisecond time intervals and is able to transmit messages of several hundred impulses per second. Each synaptic vesicle releases a quantum of 7000 to 12,000 ACh molecules into the narrow junctional cleft, raising the local concentration to the millimolar range (Kuffler and Yoshikami, 1975; cf. Van der Kloot and Molgo, 1994). Under this condition, the receptors receiving the chemical messages are of low affinity. As far as the structure for chemical information processing is concerned, since the work of Ramon-y-Cajal (1893) and Sherrington (1906), much of our current knowledge comes from studies based on junctional architecture (cf. Tansey, 1998). The idea that the transmitter is released in quanta on the arrival of the action potential is well established and has been accepted at the neuromuscular junction, but it is not at all clear that this is the case at the autonomic neuroeffector transmission site. In contrast to striatal muscle, autonomic neuroeffector systems are thus not organized in units, but the innervation is quite diffuse. The quantal release is less clearly established in the central nervous system (CNS), although evidence is presented that this is probably the case. Accordingly, the brain was considered a telephone network that receives signals via synapse processing by means of a high concentration of transmitters through receptors. A chemical signal transmission system that primarily uses synaptic transmission in the CNS possesses several characteristics. Because transmitter concentration in the synaptic gap can be high (~0.01-1 mM), the receptors expressed on both presynaptic and postsynaptic sites are of low affinity (MacDermott et al., 1999). Additionally, once released, the transmitter is removed from the cleft either enzymatically or by an uptake system and by diffusion.

One well characterized mechanism by which chemical neurotransmission can be modulated is the presynaptic modulation of transmitter release via presynaptic receptors expressed on axon terminals. Activation of these receptors by endogenous or exogenous ligands results in inhibition or facilitation of the amount of transmitters released into the extracellular space by an action potential (Starke et al., 1977, 1989; Westfall 1977; cf. Starke, 1981; Langer, 1981a; Vizi, 1979; Muscholl, 1980a,b; Kalsner and Westfall, 1990; Wu and Saggau, 1997). It is interesting to note that Koelle (1990) mentioned in his review that the first concrete evidence of presynaptic receptors was published in a classic report by Masland and Wigton (1940). These authors claimed that the fasciculation that follows the intra-arterial injection of ACh or an anticholinesterase drug into a skeletal muscle reflects the firing of the motor units rather than stimulation of the muscle fibers. The other possibility to modulate the extracellular concentration of transmitters, once released, is their removal by one of the Na⁺/Cl-dependent neurotransmitter transporters. The transporter is a plasma membrane protein that operates by reuptake of the released transmitter. The tricyclic antidepressants exert their effect on monoamine transporters by prolonging the time needed for clearance of transmitters from the extracellular space.

Neurochemical (Vizi, 1980, 1984; cf. Vizi and Kiss, 1998) and morphological (Descarries et al., 1987; Oleskevich et al., 1989; Umbriaco et al., 1995) evidence has shown that some neurotransmitters may be released from both synaptic and nonsynaptic sites (Fig. 1) for diffusion to target cells more distant than those observed in synaptic transmission. It has been shown in the gut and brain that in response to activation of the noradrenergic neurons, there was an -adrenoceptor-mediated inhibition of ACh release from neighboring cholinergic terminals (Vizi and Knoll, 1971, 1976; Vizi, 1974, 1980b) without any morphologic (synaptic) contact between them. These findings indicate that there is functional interaction (presynaptic inhibition) between neurons without any morphological contact (cf. Vizi, 1980a, 1984a). This was supported by the fact that matches between release sites and localization of receptors sensitive to the chemical signal are exceptions rather than the rule (Herkenham, 1987). Although the disparities between axon terminals (release sites) and receptors were noted in several reports (cf. Herkenham, 1991), Herkenham was the first who studied this mismatch carefully. Even in the report on substance P receptors (Rothman et al., 1984), they put forth that mismatches were the rule rather than the exception. The conclusion (Herkenham, 1987; McLean et al., 1987) drawn from this "mismatch" problem was that mismatches reflect on the existence of high-affinity nonsynaptic receptors that are able to mediate "parasynaptic" (Schmitt, 1984) signal transmission. The nonsynaptic interactions between neurons would be a form of communication transitional between discrete classic neurotransmission (in Sherrington's synapse) and the relatively nonspecific neuroendocrine secretion. Recent findings indicate that in addition to monoamines [norepinephrine (NE), dopamine (DA), and serotonin(5-hydroxytryptamine, or 5-HT)], other transmitters, such as ACh (Descarries et al., 1997) and nitric oxide (NO; Dawson and Snyder, 1994), also may be involved in these nonsynaptic interactions. NO can influence the function of uptake carrier systems (Cutillas et al., 1998; Kiss et al., 1999), which may be an important factor in the regulation of extracellular concentration of different transmitters (Gainetdinov et al., 1998; Segovia and Mora, 1998).

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Synaptic and nonsynaptic interaction between neurons. The transmitter release from varicosity without synaptic contact diffuses far away from the release site and activates remote receptors of high affinity. The transmitter concentration in the extracellular space is low and drops in a function of 3. In the synapse, the transmitter concentration is very high, and there is a possibility of spillover. Transmitter released is taken up by plasma membrane transporters located in the synapse or extrasynaptically.

This review focuses on the role of nonsynaptic receptors and transporters in presynaptic modulation of chemical transmission in the CNS, and I outline some of the potential points at which we might expect the occurrence of nonsynaptic functional interaction between neurons. It should be clear from this review that nonsynaptic interaction between neurons mediated via receptors and transporters of high affinity not localized in synapses has the potential to be an important contributor to the properties and function of neuronal networks. In addition, receptors and transporters expressed nonsynaptically and of high affinity are the target of drugs taken by the patient.

	II. Modulation of Neurochemical Transmission: Role of Receptors and Plasma Membrane Transporters

Top Previous Next References

Action potential at the nerve terminal results in an increase in Ca²⁺ influx through Ca²⁺ channels, activating Ca²⁺ sensors, which in turn trigger the release machinery to cause vesicle fusion and transmitter release (cf. Llinas, 1977). Extracellular Ca²⁺ concentration ([Ca²⁺]_o)-dependent release of transmitters [glutamate (Glu): Uchihashi et al., 1998; Nakai et al., 1999; DA: Milusheva et al., 1992, 1996; and NE: West and Fillenz, 1980) is vesicular (Katz, 1969)], has a high requirement for energy, and is very sensitive to the intracellular ATP level. The [Ca²⁺]_o-dependent release can be blocked by tetrodotoxin and subjected to presynaptic modulation via activation of different presynaptic receptors (Table 1).

It was not until the late 1960s that the concept of presynaptic receptors (i.e., receptors on nerve endings, as opposed to postsynaptic receptors on effector cells) was proposed and the hypothesis was advanced that presynaptic receptor mechanisms are involved in the modulation of neuronal ACh and NE release via - and muscarinic heteroreceptors (Lindmar et al., 1968; Vizi, 1968; Löffelholz and Muscholl, 1969a,b; Paton and Vizi, 1969).

In 1971, in different laboratories (De Potter and Chubb, 1971; Farnebo and Hamberger, 1971; Kirpekar and Puig, 1971; Starke, 1971), NE was shown to inhibit its own release from noradrenergic terminals via presynaptic -adrenoceptors. This was named "negative feedback" modulation. Later, similar autoreceptor-mediated modulation was described for other transmitters. It was even shown that there is a positive feedback modulation when the transmitter released into the synaptic cleft increases its own release via stimulation of receptors located on terminals from which the transmitter is released.

The release of transmitter is subjected to presynaptic receptor-mediated modulation (Langer, 1977, 1981a; Nicoll and Alger, 1979; Vizi, 1979, 1984; cf. Starke et al., 1989) if the release is of vesicular origin, is [Ca²⁺]_o-dependent (Table 1), and is associated with axonal conduction. The ligand-gated release by nicotinic acetylcholine receptor (nAChR) or P2X receptor stimulation is also [Ca²⁺]_o-dependent (Sershen et al., 1997; cf. Wonnacott, 1997). Because it has been shown that ₂-adrenoceptor activation (Vizi et al., 1995b) inhibits the release of NE evoked by nAChR stimulation, it seems very likely that this type of release is subjected to presynaptic inhibition.

There is a [Ca²⁺]_o-independent release that is not associated with neuronal conduction and not of vesicular origin. It has been reported that transmitters can be released by drugs (e.g., ouabain, indirectly acting sympathomimetic amines) or conditions (e.g., ischemia) in the absence of [Ca²⁺]_o (cf. Ádam-Vizi, 1992; Bernáth, 1992), which has been attributed to an increase in intracellular Na⁺ concentration ([Na⁺]_i; Vizi, 1972; Baker and Crawford, 1975; Erulkar and Rahamimoff, 1978; Schoffelmeer and Mulder, 1983). This type of release is not subject to presynaptic modulation (Vizi, 1984) and is carried out by reversed operation of the plasma membrane transporter mediated via an increase in [Na⁺]_i (Table 1). The carrier-mediated release (Vizi, 1972, 1978; Vizi et al., 1985; Kauppinenen et al., 1988; Pin and Bockaert, 1989; Attwell et al., 1993; Levi and Raiteri, 1993; Milusheva et al., 1994, 1996; Malva et al., 1998a,b; cf. Vizi and Kiss, 1998) does not require energy and is consistent with a drop in intracellular ATP levels and the consequent inhibition of Na⁺,K⁺-activated ATPase activity, which leads to a decline in the Na⁺ electrochemical gradient across the plasma membrane and accumulation of [Na⁺]_i (Nicholls and Attwell, 1990). The excessive transmitter release of nonvesicular origin (Attwell et al., 1993) cannot be modulated via presynaptic receptors (Table 1). A similar mechanism is responsible for [Ca²⁺]_o-independent release of different transmitters during ischemia simulated by oxygen and glucose withdrawal (Kauppinenen et al., 1988; Budd, 1998).

K⁺ excess has been used for a long time to study transmitter release. It is [Ca²⁺]_o-dependent and is likely to be of vesicular origin (Table 1), at least at low concentrations.

The overwhelming majority of the evidence for presynaptic receptor-mediated modulation of transmitter release derives from assays of agonist- and antagonist-induced changes in the release rate from in vitro (slice, synaptosomes) and in vivo (microdialysis, amperometry) preparations. Although assay of the amount of transmitters in the superfusate or in the dialysate is limited in spatial and temporal resolution, electrophysiological methods (e.g., whole-cell recording) in brain slices or in cell culture help us to overcome these problems and to study the effects of ligands on presynaptic terminals recording the postsynaptic responses. However, even the electrophysiological recordings of the frequency of spontaneous postsynaptic currents or the amplitude of the evoked postsynaptic current, without a change in postsynaptic sensitivity to the synaptic transmitter, provide convincing evidence that the receptor in the study is expressed on the terminal. This technique fails to exclude the possibility that a chemical is released from the postsynaptic site that may act retrogradely, affecting the release of transmitter from the presynaptic site. Therefore, to circumvent these difficulties, the final conclusion should be drawn from data obtained from different techniques.

1. Heteroreceptor-Mediated Control of Transmitter Release. An important consequence of the expression of receptors on axon terminals is the capability of modulation (increase or decrease) of transmitter release triggered by the action potentials when they invade presynaptic terminals. Heteroreceptors are located presynaptically; they bind transmitters other than those released by the axon terminal on which they reside. Heteroreceptors can be ionotropic or metabotropic.

The first neurochemical and pharmacological evidence of heteroreceptors was obtained when it was shown in guinea pig ileal longitudinal muscle strip preparation that the stimulation-evoked release of ACh from the Auerbach plexus is tonically controlled by NE released from the neighboring noradrenergic neurons via -adrenoceptors (Vizi, 1968

2. Autoreceptor-Mediated Control of Transmitter Release. The first evidence was provided in the 1970s when Starke (1971) and others (De Potter and Chubb, 1971; Farnebo and Hamberger, 1971a; Kirpekar and Puig, 1971) showed that NE inhibits its own release via -adrenoceptors. Later, it turned out that these receptors are different from those located on the postsynaptic site. Therefore, presynaptic -adrenoceptors have been named ₂-adrenoceptors (Langer, 1977, 1981a,b).

3. Presynaptic Ionotropic Receptors. Recent convergence of data from morphological and functional (pharmacological, neurochemical, and electrophysiological) studies provided new insights into the role of presynaptic ligand-gated ion channels (cf. McGehee and Role, 1996; MacDermott et al., 1999) in modulation of transmitter release, thereby in the efficacy of synaptic and nonsynaptic communication. Activation of ionotropic receptors results in very rapid changes of ion channels.

b. Glutamate Receptors (N-Methyl-D-aspartic Acid, -Amino-3-hydroxy-5-methysoxazole-4-propionic Acid, and Kainate Receptors). The vast majority of excitatory synapses are glutamatergic, in which Glu transmits the signal through postsynaptic ionotropic [N-methyl-D-aspartic acid (NMDA), -amino-3-hydroxy-5-methysoxazole-4-propionic acid (AMPA), and kainate (KA)] and metabotropic receptors (Bettler and Mulle, 1995

c. -Aminobutyric Acid_A Receptors. Frank and Fuortes (1951)

4. Presynaptic Metabotropic Receptors. The activation of membrane receptors expressed in the membrane is a common mechanism through which cellular functions, including neurotransmitter release, can be modulated (Table 2). Metabotropic receptors are coupled to G proteins. The latter are proteins (containing -, -, and -subunits) that are present in membranes of the cell and transduce the receptor activation event to changes in enzyme or ion-channel activity.

a. ₂-Adrenoceptors. It is clear from their structure and pharmacology that ₂-adrenoceptors belong to the G protein-linked family and, in most cell types, are coupled to PTX-sensitive G proteins. It is well established that some receptors inhibit adenylyl cyclase through the G protein G_i. The activation of ₂-adrenoceptor subtype has been shown to inhibit adenylyl cyclase activity, decrease cAMP levels, and inhibit Ca²⁺ channels in many cell types, including neurons. cAMP has a facilitatory effect on many transmitter systems, including the noradrenergic system (Majewski et al., 1990

View this table: [in this window] [in a new window]   TABLE 3 Subtypes of presynaptic auto- and hetero-2-adrenoceptors expressed on axon terminals and on locus ceruleus For expression of mRNA of 2-adrenoceptor subtypes, see Winzer-Sershan et al. (1997a,b).

f. -Aminobutyric Acid_B Receptors. The metabotropic GABA_B receptors are widely distributed in the CNS, where they are located (Bowery, 1993

B. Plasma Membrane Transporters. Na⁺/Cl-dependent transporters, such as DA, 5-HT, NE, GABA, glycine, taurine, proline, and betaine, are members of a large family of Na⁺/Cl-containing putative transmembrane domains (Kanner and Schuldiner, 1987; Amara and Kuhar, 1993; Lester et al., 1994; Nelson, 1998). These ion-coupled transporters are "electrogenic" and lead to conductive properties (cf. Lester et al., 1996). These transporters are different from that expressed in the membrane of vesicles (Henry et al., 1998). It is generally accepted that these high-affinity transporters located on the nerve terminals and surrounding glial cells (cf. Nelson, 1998) control the temporal and spatial concentration of transmitters released into the intrasynaptic and extrasynaptic spaces via rapid uptake into nerve terminals. For example, in tissue with dense noradrenergic and GABAergic innervation, as much as 70 to 80% of released NE (Bönisch and Brüss, 1994) and GABA (Iversen and Kelly, 1975) may be recaptured, indicating that the transporter plays an important role in setting the concentration of transmitter in the extracellular space. Thus, plasma membrane transporters maintain low intrasynaptic and extrasynaptic neurotransmitter concentrations, thereby regulating synaptic and nonsynaptic efficacy; many of the transporters have been implicated as important sites for drug actions.

1. Characteristics of Transporters. a. Protein Kinase C Dependence. It has been suggested that protein kinase C (PKC) plays a role in acute modulation of Na⁺/Cl-coupled transporters, including GABA, DA, and 5-HT (Corey et al., 1994; Osawa et al., 1994; Zhang et al., 1997; Ramamoorthy et al., 1998). Activation of PKC by phorbol ester inhibits the uptake of GABA (Osawa et al., 1994), DA (Zhu et al., 1997), 5-HT (Qian et al., 1997; Ramamoorthy et al., 1998), and NE (Apparsundaram et al., 1998).

/

2. Substrate Selectivity. a. Norepinephrine Transporter. NE transporters (NETs) located in the neuronal plasma membrane mediate the removal of NE from the extracellular space (cf. Graefe and Bönisch, 1988; cf. Trendelenburg, 1991; Nelson, 1998), limiting the activation of auto- and hetero-adrenoceptors expressed on different neurons by reducing the extracellular concentration of NE and thereby the amount of NE available for diffusion. NETs also transport structurally similar molecules, including DA, tyramine, and amphetamine (Bönisch, 1986; Bönisch and Brüss, 1994). Similar to other transporters, NETs use the energy of the transmembrane Na⁺ gradient to take up NE inside the neuron from the intrasynaptic and/or extrasynaptic space. The direction of transport can be reversed by inward transport of any substrate (cf. Chen and Justice, 1998). At a high frequency of stimulation, so much transmitter is being released that the transporter capacity (which has been used up) becomes fully exploited and unable to continue to clear the extracellular space.

/

d. -Aminobutyric Acid Transporter. The role of GABA transporter is to terminate synaptic events evoked by GABA released from GABAergic terminals. GABA uptake inhibitors increase and prolong GABA_B receptor-mediated transmission (Dingledine and Korn, 1985

e. Serotonin Transporter. The serotonin transporter is also a member of a superfamily of Na⁺- and Cl-dependent neurotransmitter transporters, which are important targets for both drugs of abuse and antidepressant compounds (Amara and Kuhar, 1993

	III. Nonsynaptic Varicosities

Top Previous Next References

It is well known that axons both in the CNS and in the autonomic nervous system form varicose (boutons-en-passant) branches. The varicose axon terminals, which in the overwhelming majority do not make synaptic contacts, are the main target of presynaptic modulation. A substance released in or diffusing to the vicinity of the axon terminal can modulate the release of the principal transmitter or that of another modulator provided the axon is equipped with sensitive receptors. Many authors (cf. Langer, 1977; cf. Starke, 1977) support the idea that presynaptic modulation is a question of secretion coupling. However, others (Alberts et al., 1981; Stjärne, 1981) suggest that mechanisms related to the failure of varicosity invasion are responsible for presynaptic inhibition. It was suggested that presynaptic inhibition is the reduction in the safety factor for terminal varicosity invasion by the axonal nerve impulse (Stjärne, 1981, 1989). The release of NE from sympathetic nerve terminals by the invading nerve impulse is a very uncertain process, with a high proportion of failures at any individual varicosity (Blakeley and Cunnane, 1979), so that any procedure marginally reducing the chance of invasion may have profound effects. Alberts et al. (1981) suggested that varicosity hillocks could control the excitability of the varicosity and the invasion of the more distal parts of the branch. The wave of depolarization arriving at the varicosity hillock reaches the firing level and generates a propagating impulse in the next intervaricose section. However, this site of action seems very unlikely because it is based on acceptance that the method of action potential conduction in a varicose terminal is similar to that in a neuron, and this is not the case. The action potential attempts to invade the whole branch without alternating depolarization with action potential generation and intervaricosital axonal conduction. The bouton is small; therefore, the action potential arriving at the first and consecutive boutons tries simply to depolarize and pass them. Although methods for analysis of the mode of impulse conduction in boutons-en-passage terminals are not available, it is tempting to speculate that the difference in size between the cross section of the intervaricosital axonal part and the bouton makes it difficult for impulses to invade the whole branch. The sudden change in size of the varicose branch at the bouton might produce changes in, for example, the length constant (Vizi, 1984) and thereby influence the length of invasion of the rather lengthy arborization.

There is convincing neurochemical evidence, mainly based on studies with synaptosomal preparations and potassium-induced release, that the site of action is on the axon terminals. Therefore, there is general agreement that secretion coupling is affected by the modulators (cf. Langer, 1977; cf. Starke, 1977).

Although the axo-axonic synapse is the anatomical correlate of presynaptic modulation, convincing anatomical evidence is available that noradrenergic (cf. Oleskevich et al., 1989), serotonergic (Descarries et al., 1975; Seguela et al., 1989; Oleskevich et al., 1991), dopaminergic (Descarries et al., 1991), and cholinergic (Descarries et al., 1997) varicosities in the CNS in a rather high percentage do not make synaptic contact. A very low synaptic incidence of monoaminergic innervation has already been documented in the adult rat cortex (Descarries et al., 1975, 1977; Beaudet and Descarries, 1978; Seguela et al., 1990, 1989), hippocampus (Oleskevich et al., 1989; Daszuta et al., 1991; Umbriaco et al., 1995), dorsal horn of the spinal cord (Ridet et al., 1993), and cerebellum (Beaudet and Sotelo, 1981).

Although the incidence of nonsynaptic nerve terminals amounted to 64% for 5-HT and 57% for NE in the dorsal horn of the spinal cord (Ridet et al., 1993), in the ventral horn (Privat et al., 1988) the synaptic contacts were the predominant, indicating that nonsynaptic communication is characteristic of the dorsal horn, and low-affinity 5-HT₂ receptors are numerous in the ventral horn but scarce in the dorsal horn (Pazos et al., 1985, Pazos and Palacios, 1985).

Recent immunoelectromicroscopic studies have revealed a low incidence (14% in the cerebral cortex, 7% in the hippocampus, and 9% in the neostriatum) of synaptic