Main

In the late 1970s, using radioligand binding techniques, it was discovered that antipsychotic drugs interacted not only with dopamine receptors but also with serotonin (5-HT) receptors, particularly in the frontal cortex (Leysen et al. 1978). Shortly thereafter, using radiolabeled 5-HT and the hallucinogen LSD (d-lysergic acid diethylamide) as ligands, at least two distinct populations of 5-HT receptors, termed 5-HT1 and 5-HT2, were described in the brain (Peroutka and Snyder 1979). Based on this classification, it was determined that some antipsychotic drugs, such as clozapine, interacted more potently with the 5-HT2 than D2 receptor; however, clinical potency appeared to correlate only with affinity for the D2 receptor (Peroutka and Snyder 1980). Only later, when antipsychotic drugs were grouped separately according to whether they had “typical” or “atypical” characteristics, did a high affinity for 5-HT2 relative to D2 receptors emerge as a distinguishing feature of the drugs in the atypical category (Altar et al. 1986; Meltzer et al. 1989).

Antipsychotic drugs are generally classified as atypical if they have reduced extrapyramidal side effects and/or an enhanced spectrum of antipsychotic efficacy, particularly with regard to negative symptoms. An early clinical study suggested that a drug, setoperone, with high affinity for 5-HT2 relative to D2 receptors, in addition to having relatively low extrapyramidal side effects, may have enhanced efficacy in the treatment of negative symptoms (e.g., emotional withdrawal, blunted affect) (Ceulemans et al. 1985). It is noteworthy that part of the rationale given in the latter study for testing the antipsychotic properties of 5-HT2-antagonist drugs was their ability to block certain effects of LSD in animal model systems (Colpaert and Janssen 1983). The hypothesis that 5-HT2 receptor blockade contributes to the favorable profile of atypical antipsychotic drugs led to the development of a new generation of agents with a high 5-HT2A to D2 ratio, some of which (e.g., risperidone) have now become accepted for clinical use in the treatment of schizophrenia (for reviews, see: Breier et al. 1997; Lieberman et al. 1998; Meltzer and Nash 1991).

Meanwhile, a large body of evidence has accumulated that 5-HT2 receptors may be involved in the mechanism of action of psychedelic hallucinogens. Originally, this idea was based on the demonstration of a high correlation between hallucinogenic potency in humans and the affinity of both phenethylamine (e.g., mescaline) and indoleamine (e.g. LSD and psilocybin) hallucinogens for 5-HT2 receptors (Glennon et al. 1984; Titeler et al. 1988). Indeed, among all known 5-HT receptor subtypes, affinity for 5-HT2 receptors is the only one shared by both the indoleamine and phenethylamine classes of hallucinogens. There is now abundant evidence from biochemical (Sanders-Bush et al. 1988), electrophysiological (Marek and Aghajanian 1996), and animal behavioral (Glennon 1990) studies that the effects of hallucinogens involve a partial agonist action at 5-HT2 receptors, particularly of the 5-HT2A subtype. Very recently, studies on the role of 5-HT2 receptors in the action of hallucinogens have been extended to human subjects. As predicted by the preclinical studies, the psychotomimetic effects of the indoleamine hallucinogen psilocybin have now been shown to be blocked completely by the preferential 5-HT2A antagonist ketanserin, the atypical antipsychotic risperidone, but not the typical antipsychotic haloperidol (Vollenweider et al. 1998). However, it remains to be determined whether 5-HT2A antagonism alone is sufficient for the treatment of schizophrenia or whether a threshold level of D2 antagonism may be required for full antipsychotic efficacy (Farde et al. 1992; Kapur et al. 1999). The resolution of these issues will require the completion of clinical testing with highly selective 5-HT2A antagonists such as M100907 (also referred to as MDL 100,907).

Because of their possible involvement in the action of psychedelic hallucinogens and atypical antipsychotics there has been much interest in the location and function of 5-HT2A receptors in the central nervous system. Autoradiographic studies show 5-HT2A receptor binding in several regions of the brain, but the bulk of these receptors are found in the cerebral cortex (Lopez-Gimenez et al. 1997; Pazos and Palacios 1985). A correspondingly high density of 5-HT2A receptor mRNA has been found by in situ hybridization in the cerebral cortex (Mengod et al. 1990; Wright et al. 1995). Recent immunocytochemical studies have demonstrated a particularly high density of 5-HT2A receptors within the apical dendrites of cortical pyramidal cells (Hamada et al. 1998; Jakab and Goldman-Rakic 1998; Willins et al. 1997). This postsynaptic localization is consistent with electrophysiological studies on cortical pyramidal cells showing direct depolarizing effects mediated by 5-HT2 receptors (Aghajanian and Marek 1997; Araneda and Andrade 1991; Sheldon and Aghajanian 1990; Tanaka and North 1993).

The focus of this review is on the physiological role of 5-HT2A receptors in the medial prefrontal and anterior cingulate regions of the cerebral cortex, particularly in relation to their ability to enhance certain kinds of glutamatergic transmission by indoleamine and the phenethylamine hallucinogens. These electrophysiological effects are discussed in relation to the NMDA antagonist class of psychotomimetics, which, surprisingly, have now been reported to induce an increase in glutamate release in the cerebral cortex. Finally, potential new targets for antipsychotic drug design based on a hyperglutamatergic model of psychotomimetic drug action are discussed.

5-HT2A RECEPTORS ENHANCE GLUTAMATE RELEASE IN NEOCORTEX

Electrophysiological actions of hallucinogens, acting via 5-HT2 receptors, have been described in several subcortical regions of the rat (Aghajanian 1980; Garratt et al. 1993; McCall and Aghajanian 1980; Rasmussen and Aghajanian 1986). However, because hallucinogens characteristically alter higher-level processes, such as cognition, perception, and mood, studies on their effects in the cerebral cortex are receiving increasing attention. One of the most striking effects of 5-HT is an increase in spontaneous postsynaptic potentials (PSPs) in various cortical regions. Originally, recordings from pyramidal cells in brain slices from rat piriform cortex, a paleocortical region, showed that 5-HT induces mostly inhibitory postsynaptic potentials (IPSPs) (Sheldon and Aghajanian 1990). However, more recent studies have shown that in layer V pyramidal cells of the neocortex, 5-HT-induced synaptic events consist mostly of excitatory postsynaptic potentials/currents (EPSPs/EPSCs) (Figure 1 ), shown in part by the fact that they are blocked by the AMPA/kainate glutamate receptor antagonist LY293558 but not the GABAA antagonist bicuculline (Aghajanian and Marek 1997). The 5-HT-induced EPSCs are antagonized by low concentrations of the highly selective 5-HT2A antagonist M100907 (Figure 2 ), indicating that they are mediated by 5-HT2A receptors (Aghajanian and Marek 1997; Marek and Aghajanian 1999). The 5-HT-induced increase in EPSCs is most pronounced in frontal regions such as the medial prefrontal cortex (Aghajanian and Marek 1997), where the density of 5-HT2A receptors is increased relative to more posterior regions (Lopez-Gimenez et al. 1997; Pazos and Palacios 1985). Norepinephrine (NE), via α1 adrenoceptors, also induces an increase in EPSPs in layer V pyramidal cells, but, at least in the rat, to only a fraction of that produced by 5-HT (Marek and Aghajanian 1999).

Figure 1
figure 1

5-HT-induced increases in the frequency and amplitude of EPSCs in neocortical pyramidal cells. A shows a whole cell recording of the effect produced by bath applied 5-HT (100 μM, 1 min) on spontaneous EPSCs in a layer V pyramidal cell in medial prefrontal cortex; seven consecutive 1-s episodes are shown. B shows a plot of the normalized cumulative distribution of EPSC amplitudes before and during the application of 5-HT for the cell illustrated in A (data taken from 10 1-s episodes; 2 pA cutoff). The shift in distribution to higher amplitudes was significant at p < .001 (Kolmogorov–Smirnov test). C shows the change in mean frequency and amplitude (±SEM) in response to 5-HT (100 μM, 1 min) for the entire sample of cells tested (n = 9). Mean basal frequency, 12.6 ± 2.2 Hz; mean basal amplitude, 7.9 ± 0.9 pA. The increase in frequency was significant at p < .001 (paired t- test) (Aghajanian and Marek 1997)

Full size image
Figure 2
figure 2

Concentration-response curve for 5-HT-induced EPSCs in neocortical layer V pyramidal cells and blockade by the 5-HT2A antagonist M100907 (MDL). A shows a normalized concentration-response curve for randomly selected layer V pyramidal cells (n = 6, 4 medial prefrontal cortex and 2 frontoparietal cortex) where the 5-HT-induced increase in EPSC frequency at each concentration (3, 10, 30, 100, and 300 μM) is expressed as the percentage of the 300 μM 5-HT response for each cell (mean ± SEM). The EC50 was 19.7 μM, and the maximal EPSC frequency was 36.5 Hz. B shows the 5-HT concentration-response curve for a frontoparietal layer V pyramidal cell both before (closed squares) (EC 50 = 19.4 μM) and after (open circles) (EC 50 = 57.0) a 27-min treatment with 3 nm M100907 (M100907 perfusion continued in between the 1-min 5-HT applications). The pA 2 in this cell calculated by the Schild equation was 8.8. Note that M100907 treatment resulted in a rightward parallel shift of the concentration-response curve and that the antagonism by M100907 was surmountable by a higher 5-HT concentration (300 μM) (Aghajanian and Marek 1997)

Full size image

Quantitatively, the most pronounced effect of 5-HT in prefrontal cortex is to increase the frequency of EPSCs (Figure 1). Changes in the frequency of synaptic currents or potentials are generally regarded as being indicative of a modulation of presynaptic function. Accordingly, the relatively nonspecific group II/III metabotropic glutamate receptor agonist (1S, 3S)-ACPD (Aghajanian and Marek 1997) (Figure 3 ) or the selective group II metabotropic agonist LY354740 (Marek et al. 1999), acting at presynaptic inhibitory autoreceptors located on glutamatergic nerve terminals, suppresses the 5-HT-induced increase in the frequency of EPSCs in prefrontal cortex. These findings are consistent with the idea that activation of 5-HT2A receptors increases the release of glutamate onto layer V pyramidal cells through a presynaptic mechanism. However, this presynaptic effect might be mediated indirectly through a retrograde messenger, because, as discussed below, 5-HT2A receptors appear predominantly to have a postsynaptic localization (Hamada et al. 1998; Jakab and Goldman-Rakic 1998). 5-HT also produces a small but significant increase in the amplitude of spontaneous EPSCs, an effect that may involve a postsynaptic amplification mechanism (Aghajanian and Marek 1997). Such a postsynaptic effect is consistent with the finding of a high density of 5-HT2A receptor immunoreactivity in the apical dendrites of cortical pyramidal cells (Jakab and Goldman-Rakic 1998; Willins et al. 1997).

Figure 3
figure 3

The effect of the group II/III metabotropic agonist (1S, 3S)-ACPD on 5-HT-induced EPSCs. A shows a whole-cell recording from a layer V cell in medial prefrontal cortex in which the prior (4 min) and concurrent application of (1S, 3S)-ACPD (200 μM) largely blocks the 5-HT-induced increase in EPSCs; the effect of (1S, 3S)-ACPD reversed rapidly during 4 to 5 min of washout (not shown). B gives normalized mean EPSC amplitude and frequency data for five cells; basal amplitude was 31 ± 4 pA, and basal frequency was 2 ± 0.7 Hz. The predominant effect of (1S, 3S)-ACPD was to reduce the 5-HT-induced increase in EPSC frequency (p < .05, two-tailed t-test). C represents a voltage-clamp recording (K acetate electrode) of inward AMPA currents measured before and during the application of (1S, 3S)-ACPD (200 μM). Note that the inward current produced by bath applied AMPA (5 μM) is not reduced during the application of (1S, 3S)-ACPD. In all five cells tested, (1S, 3S)-ACPD caused an increase rather than a reduction in postsynaptic AMPA currents. Inward currents are downward in direction; input conductance was monitored by periodic (20 s) 1-s negative current pulse, giving rise to brief downward deflections (Aghajanian and Marek 1997)

Full size image

The 5-HT-induced EPSCs are blocked by bath application of the fast sodium channel blocker tetrodotoxin (TTX) or perfusion of the slice with a solution containing no added calcium (”0” calcium) (Aghajanian and Marek 1997). Ordinarily, TTX sensitivity or Ca2+ dependence would suggest that 5-HT had activated glutamatergic cells located within the slice, leading to an impulse-flow dependent release of glutamate. However, several lines of evidence argue against this conventional interpretation. First, rarely were neurons within the confines of the brain slice induced to fire by bath application of 5-HT. Second, none of the recorded pyramidal cells (a potential source of intracortical excitatory inputs) was depolarized sufficiently by 5-HT to reach threshold for firing. Third, EPSCs could be induced by the microiontophoresis of 5-HT onto “hotspots” along the apical dendrites of layer V pyramidal cells, but no cell firing could be detected in the same locations while recording extracellularly through the microiontophoretic electrode (Aghajanian and Marek 1997). Together, these experiments suggest that 5-HT induces EPSCs in neocortical cells through a focal action involving a Ca2+-dependent mechanism that does not require impulse flow.

5-HT2A RECEPTORS AND ASYNCHRONOUS TRANSMISSION IN THE CEREBRAL CORTEX

Recently, we have begun to investigate possible atypical mechanisms by which 5-HT, in the absence of an increase in afferent impulse flow, could induce a focal, TTX-sensitive/Ca2+-dependent release of glutamate. Previously, it has been reported that mildly depolarizing agents, such as the K+ channel blocker 4-aminopyridine (4-AP), can induce a TTX-sensitive release of glutamate from isolated cortical synaptosomes; whereas, the increase in glutamate release produced by strongly depolarizing concentrations of KCl (sufficient to activate voltage-gated Ca2+ channels) is not TTX-sensitive (Tibbs et al. 1989). Notably, 4-AP preferentially enhances the “slow” rather than the “fast” component of glutamate release from cortical synaptosomes; this slow component has been hypothesized to represent the asynchronous mode of release, because it is supported by Sr2+ in the absence of Ca2+ (Herrero et al. 1996). Sr2+ substitutes for Ca2+ at the high-affinity Ca2+ sensor synaptotagmin III, which is believed to be responsible for slow, asynchronous transmitter release; in contrast, Sr2+ is ineffective at the low-affinity Ca2+ sensor synaptotagmin I thought to be essential for fast, synchronous transmitter release (Figure 4 ) (Geppert et al. 1994; Goda and Stevens 1994; Li et al. 1995). It has been proposed that the formation of a multi-molecular complex, consisting of synaptotagmin I together with synaptic core complex proteins, interacts with the segment II/III intracellular loop of Ca2+ channels (Tobi et al. 1998), forming the basis for rapid Ca2+-triggered transmitter release (for review, see Stanley 1997). Thus, if coupling to the synaptic protein interaction site of Ca2+ channels is prevented, fast synchronous transmitter release is impaired; whereas, late asynchronous release remains intact (Mochida et al. 1996).

Figure 4
figure 4

A schematic of how a synaptic vesicle associated with the low-affinity Ca2+ sensor synaptotagmin I (Stg I) is coupled to a voltage-gated Ca2+ channel (arrow), while a synaptic vesicle associated with synaptotagmin III (Stg III) is at a distance from the Ca2+ channel. With the arrival of a nerve impulse, the entry of Ca2+ is able to trigger rapid, synchronous release of transmitter via Stg I, because there is a high, localized concentration of this ion (∼200 μm); Sr2+ can also enter but cannot trigger transmitter release via Stg I. As Ca2+ or Sr2+ diffuse away from the Ca2+ channel, their concentration declines (e.g., to ∼10 μm), but both ions can trigger late or asynchronous release via the high-affinity Ca2+ sensor Stg III

Full size image

In various regions of brain, including cerebral cortex (Araneda and Andrade 1991; Sheldon and Aghajanian 1990; Tanaka and North 1993), the effects of 5-HT2A-receptor activation resemble those of 4-AP in producing a slow depolarization through a reduction in K+ channel conductance. By analogy with 4-AP, it is possible that 5-HT induces an a TTX/Ca2+-sensitive focal increase in spontaneous EPSCs through an increase in the Sr2+-supported asynchronous mode of glutamate release. To test this possibility, we have examined the ability of Sr2+ to substitute for Ca2+ in supporting 5-HT-induced spontaneous EPSCs. We have found that the 5-HT-induced increase in the frequency of spontaneous EPSCs is fully supported by Sr2+ in the absence of added Ca2+, suggesting mediation by the asynchronous rather than synchronous mode of transmitter release (Figure 5 ). In contrast, the amplitude of spontaneous EPSCs is not fully supported by Sr2+ in the absence of Ca2+, consistent with the importance of Ca2+ in the postsynaptic amplification of glutamatergic inputs to apical dendrites of layer V pyramidal cells (Schiller et al. 1997).

Figure 5
figure 5

Effect of Sr2+ on 5-HT-induced spontaneous EPSCs in layer V pyramidal cells of medial prefrontal cortex. (A1-6) Whole cell patch-clamp recording showing responses of an individual pyramidal cell to a 1-min application of 5-HT (100 μM) in normal ACSF, after 9 to 10 minutes of ACSF with “0” added Ca2+ (2 mM Mg2+ substituted for Ca2+), or after 9 to 10 min in Sr2+ (1 mM) substituted for Ca2+. (B) Cumulative probability plot for the amplitude distribution of spontaneous EPSCs for the cell illustrated in (A); 5-HT induces a significant shift to higher amplitudes both in the normal ACSF and Sr2+ conditions (p < .001, Kolmogorov–Smirnov test). (C) Summary of results for five cells showing increased frequency of spontaneous EPSCs induced by 5-HT in normal ACSF or Sr2+/“0” Ca2+ conditions (p < .004, paired t-test); spontaneous EPSCs were not increased over the normal ACSF baseline by 5-HT in the “0” Ca2+ condition (Aghajanian and Marek 1999)

Full size image

In layer V pyramidal cells of medial prefrontal cortex, when extracellular Ca2+ is removed, there is a loss of synchronous electrically evoked EPSCs; the subsequent addition of Sr2+ leads to the appearance of late asynchronous EPSCs in the continued absence of synchronous EPSCs (Figure 6 ) (Aghajanian and Marek 1999). A similar increase in late, nonsynchronous component electrically evoked EPSCs was observed during 5-HT washout (when 5-HT2A receptors are unopposed by non-5-HT2A actions of 5-HT) or application of the 5-HT2A/2C partial agonist DOI (Figure 7 ). M100907 is highly effective in reversing the DOI enhancement of the late component of the evoked EPSC, confirming that this effect is mediated by 5-HT2A receptors (Figure 8 ). The conventional interpretation of this late component would be that it represents polysynaptic or epileptiform activity. However, in several respects (e.g., refractory period) this late component of the evoked response is distinct from conventional polysynaptic evoked EPSCs that occur, for example, in the presence of the GABAA antagonist bicuculline (Aghajanian and Marek 1999). Therefore, we hypothesize that the late component of the EPSC enhanced by 5-HT2A receptors is mediated by the asynchronous mode of glutamate release at the nerve terminal rather than through a polysynaptic pathway.

Figure 6
figure 6

Evoked EPSCs in a layer V pyramidal cell of medial prefrontal cortex: effects of Sr 2+ substitution for Ca2+. Electrical stimuli (0.2 ms; arrowheads) were applied at 10-s intervals to the underlying white matter of the forceps minor (fmi), medial to the corpus callosum (cc; see inset). (A) shows 10 consecutive responses in normal ACSF where only a short latency (∼2 ms) synchronous EPSCs were evoked. (B) All EPSCs, both synchronous and asynchronous, are suppressed markedly in ACSF with “0” added Ca2+ applied for 6 to 7 min. (C) The inclusion of Sr2+ in “0” Ca2+ ACSF results in sweeps of asynchronous evoked EPSCs following each stimulus in the absence of synchronous EPSCs. Downward deflections indicate inward currents (Aghajanian and Marek 1999)

Full size image
Figure 7
figure 7

DOI increases the occurrence of late components of evoked EPSCs in a layer V pyramidal cell. (A) Under basal conditions, there is an all-or-none sweep of a late component of the evoked EPSC only after the first in a series of 10 stimuli. (B) Following DOI (3 μM) applied for 10 to 12 minutes, there is a progressive increase in the proportion of sweeps with a persistent late component of EPSCs; at the time point shown (∼10 min following the end of DOI application), all sweeps show a late component following the synchronous response. Note that DOI, because of its partial agonist properties relative to 5-HT, induces only a minor increase in spontaneous EPSC (Aghajanian and Marek 1999). (C–D) During the application of 5-HT (100 μM) for 1 min, there is a suppression of this late component; the latter return after approximately 10 min of washout (Aghajanian and Marek 1999)

Full size image
Figure 8
figure 8

The 5-HT2A antagonist M100907 (MDL) reverses the increase in evoked EPSCs by DOI. (A–C) Illustrate in a single layer V pyramidal cell the increase in evoked EPSCs induced by DOI (3 μm) and the reversal of this effect (p < .001 compared to basal or DOI/M100907) after a 35-min application of M100907 (100 nm). (B) Bar graph shows a summary of data for five cells (Aghajanian and Marek 1999)

Full size image

PROPOSED MECHANISMS FOR 5-HT2A RECEPTOR-INDUCED ASYNCHRONOUS GLUTAMATE RELEASE

A mechanism by which 5-HT2A receptors, either directly or indirectly, could promote asynchronous EPSCs in layer V pyramidal cells would be by increasing residual Ca2+ in excitatory nerve terminals; thereby activating of synaptotagmin III (see Figure 4). There are a number of possible mechanisms through which the activation of 5-HT2A receptors could augment levels of residual Ca2+ (Figure 9 ). It is known that 5-HT2A receptors are coupled via the Gq family of G proteins to the phospholipase C/phosphoinositide second messenger pathway, one limb of which leads to the formation of inositol trisphosphate (IP3), a releasor of Ca2+ from intracellular stores (Conn and Sanders-Bush 1986). This has been demonstrated directly in C6 glioma cells where a 5-HT2A-receptor mediated release of intracellular Ca2+ has been shown to be mimicked by an increase in intracellular IP3 (Bartrup and Newberry 1994). Nevertheless, although an elevation of intraterminal Ca2+ levels may contribute to residual Ca2+ levels, the observed dependence on extracellular Ca2+ (or Sr2+) indicates that entry of extracellular Ca2+ is required for 5-HT2A-induced asynchronous EPSCs. In the case of electrically evoked EPSCs, Ca2+ entry through voltage-gated Ca2+ channels could synergize with Ca2+ released from intracellular stores. However, in the case of spontaneous EPSCs, it is more difficult to account for the entry of extracellular Ca2+ because, in the absence of an increase in impulse flow, voltage-gated Ca2+ channels would not be activated. Earlier, we reported that 5-HT enhances a subtheshold, TTX-sensitive persistent Na+ current in layer V pyramidal cells (Aghajanian and Marek 1997). We have proposed elsewhere, that, if 5-HT acts to induce an increase in a persistent subthreshold Na+ current in excitatory nerve terminals, the resulting increase in intracellular Na+ could activate reverse Na+/Ca2+ exchange (NCXr), resulting in increased Ca2+ influx (Marek and Aghajanian 1998). Figure 9 depicts how Ca2+ entry through voltage-gated channels, in conjunction with reverse Na+/Ca2+ exchange and IP3-induced release from internal stores, may increase residual Ca2+ levels thereby, enhancing asynchronous release.

Figure 9
figure 9

Schematic representation of how Ca2+ entry through voltage-gated channels, combined with reverse Na+/Ca2+ exchange (NCXr) and inositol trisphosphate (IP3)-induced release from internal stores, may synergize to increase residual Ca2+ levels, thereby enhancing asynchronous release via synaptotagmin III (Stg III). Activation of the 5-HT2A receptor is shown as increasing the formation of IP3 via the G protein Gq coupled to phospholipase C (PLC) hydrolysis

Full size image

The fact that 5-HT produces a marked increase in the frequency of spontaneous EPSCs points to the involvement of a presynaptic site of action. However, a recent electron immunocytochemical study has shown only a scattering of 5-HT2A-labeled nerve terminals in prefrontal cortex; whereas, the bulk of these receptors are located postsynaptically on pyramidal cell apical dendrites (Jakab and Goldman-Rakic 1998). Thus, although postsynaptic 5-HT2A receptors may serve to increase the amplitude of spontaneous EPSCs (Aghajanian and Marek 1997), there is no simple way to explain how such a postsynaptic action could result in a marked increase in the frequency of EPSCs. Given the evidence for a predominantly postsynaptic location of 5-HT2A receptors, there is a need to explain how interactions with presynaptic modulators may occur. One possibility would be that a retrograde messenger is generated through the action of 5-HT on postsynaptic 5-HT2A receptors, which then could have a presynaptic effect on excitatory nerve terminals. Inhibitory modulators could then suppress this retrograde effect through their separate presynaptic site of action. Nevertheless, it is still possible that the effect of 5-HT on asynchronous EPSCs is mediated through a direct presynaptic action upon the apparently small subset of 5-HT2A-positive terminals that have been demonstrated to exist by electron immunocytochemistry (Jakab and Goldman-Rakic 1998).

EFFECTS OF PSYCHOTOMIMETIC NMDA ANTAGONISTS ON GLUTAMATE RELEASE: COMPARISON WITH PSYCHEDELIC HALLUCINOGENS

Recently, it has been reported that many effects of psychotomimetic NMDA antagonists, such as phencyclidine and ketamine, may be mediated through excess glutamate acting at non-NMDA (i.e., AMPA/kainate) receptors. Microdialysis techniques show that systemic administration of ketamine enhances the release of glutamate in prefrontal cortex (Moghaddam et al. 1997). Parallel behavioral studies have shown that the systemically active AMPA/kainate antagonist LY293558 ameliorates the cognitive deficits produced by ketamine, suggesting that NMDA antagonists may disrupt cognitive function by increasing the release of glutamate thereby, stimulating non-NMDA receptors (Moghaddam and Adams 1998). Moreover, a group II metabotropic glutamate agonist, LY354740, which prevented excessive release of glutamate, reduced the cognitive and motor effects of phencyclidine. The mechanism by which NMDA antagonists induce an increase in glutamate release appears to be distinct from that of 5-HT2A agonists, because bath application of phencyclidine to brain slices does not result in an increase in EPSCs in layer V pyramidal cells of prefrontal cortex (Marek and Aghajanian unpublished data). This lack of a direct effect suggests a requirement for the activation of intact afferent systems for NMDA antagonists to induce an increase in glutamate release. In any case, the precise mechanisms by which psychedelic hallucinogens and NMDA antagonists cause an increase in glutamate release are likely to differ. The evidence for an increase in glutamate transmission for both the psychedelic hallucinogens and the NMDA antagonists raises the possibility that there is a convergent or common final glutamatergic pathway that may account for overlapping aspects of their psychotomimetic effects. On the other hand, the effects of the two classes of drugs would not be expected to be identical, because the psychedelic hallucinogens do not block NMDA receptors.

Brain-imaging studies in human subjects also reveal similarities in the effects of psychedelic hallucinogens and NMDA antagonists. Representatives of the two major classes of psychedelic hallucinogens, the indoleamine psilocybin (Vollenweider et al. 1997b) and the phenethylamine mescaline (Hermle et al. 1992), have been shown to produce metabolic hyperfrontality in the anterior cingulate and other frontal regions. Interestingly, psychotomimetic doses of ketamine have been shown to produce a similar pattern of hyperfrontality, both in healthy volunteers (Breier et al. 1997; Vollenweider et al. 1997a) and schizophrenic patients (Lahti et al. 1995). In rat studies, the metabolic activation produced by ketamine in prefrontal cortex and other regions is blocked by clozapine but not haloperidol, perhaps because of the 5-HT2A antagonist properties of clozapine, which are not shared by haloperidol (Duncan et al. 1998). Regardless of differences in specific mechanisms, it is possible that an increased release of glutamate underlies the hyperfrontality seen with both psychedelic hallucinogens and ketamine.

In contrast to the hyperfrontality seen in the drug studies, the results of brain-imaging studies in schizophrenic patients have been mixed. Some studies, particularly in acutely psychotic, unmedicated, or drug-naive patients, have reported a hyperfrontal pattern similar to that produced by psychotomimetic drugs (Cleghorn et al. 1989; Ebmeier et al. 1993; Parellada et al. 1994). However, other studies, even those conducted in unmedicated or drug-naive schizophrenic patients (Andreasen et al. 1997; Haznedar et al. 1997), did not find a hyperfrontal pattern. A more consistent finding in schizophrenic subjects, regardless of whether they display hypofrontality (Andreasen et al. 1992; Buchsbaum et al. 1992; Weinberger et al. 1986) or hyperfrontality (Parellada et al. 1994) under resting conditions, is a diminished ability to respond with prefrontal activation when challenged with cognitive tasks. Interestingly, this deficit in task-related prefrontal activation is mimicked by the administration of psilocybin to normal human volunteers (Gouzoulis-Mayfrank et al. 1999).

NEW TARGETS FOR ANTIPSYCHOTIC DRUGS BASED ON PSYCHOTOMIMETIC DRUG MODELS

If hyperglutamatergic states play a role in the pathogenesis of schizophrenia, as they do in the psychotomimetic drug models, then treatments that limit or suppress glutamate release may be therapeutic or prophylactic in this disease. However, because glutamate is the main excitatory transmitter in the central nervous system, a generalized block of glutamatergic transmission would not be useful. Our studies on 5-HT2A-receptor mediated glutamate release in the cerebral cortex reveal the highly localized nature of this process, which involves only a subset of glutamatergic afferents innervating the apical dendrites of layer V pyramidal cells. Thus, the selective 5-HT2A antagonist M100907 is able to achieve a high degree of specificity by blocking glutamate release at a restricted set of glutamatergic terminals regulated by this receptor. We also have found an increased release of glutamate mediated by α1 receptors, which, like 5-HT2A receptors, are coupled through the Gq/11/phosphoinositide pathway. It is possible that antagonists of other receptors that are coupled through the Gq/11/phosphoinositide pathway also have therapeutic potential.

An alternative way to normalize glutamatergic transmission would be to suppress glutamate release at sites downstream from the initial 5-HT2A or other Gq-coupled receptor. Such an approach would allow for the possibility that the site of pathology in naturally occurring psychoses may not be at the initial receptor but may be at one or more of the steps involved in the complex process of regulating glutamate release. The hallucinogen model suggests that one such target is the regulation of residual Ca2+ levels, which underlie the process of asynchronous release. Residual Ca2+ can be regulated both positively (e.g., through release from internal stores via the phosphoinositide/inositol trisphosphate pathway and reversal of Na+/Ca2+ exchange) and negatively (e.g., by presynaptic autoreceptors). In preliminary studies, we have shown that low concentrations of the group II metabotropic agonist LY354740 can reduce the late asynchronous EPSP while sparing the early synchronous EPSP (Marek et al. 1999). One possible basis for this selectivity is that metabotropic subtypes are differentially expressed in different afferents to layer V pyramidal cells. For example, mGluR2 receptor mRNA is expressed strongly by relay cells in the anterior and midline nuclei of thalamus (Ohishi et al. 1993), possibly contributing to the band of mGluR2 receptor protein corresponding to the terminal regions of the thalamocortical pathway, which can be seen in the mid-layer of the cerebral cortex (Ohishi et al. 1998). Interestingly, this is the same layer that has been implicated in 5-HT2A-induced EPSCs (Marek and Aghajanian 1998).

As described above, pharmacological manipulations of glutamate transmission provide unexpected parallels between the hallucinogen and NMDA antagonists drug models of psychosis. Thus, the group II/III metabotropic agonist (1S, 3S)-ACPD and the preferential group II metabotropic agonist LY354740, which reduce the release of glutamate by acting upon presynaptic inhibitory autoreceptors, are able to block EPSCs induced by activation of 5-HT2A receptors in vitro (Aghajanian and Marek 1997; Marek et al. 1999). Similarly, LY354740 (which is active by the systemic route of administration [Schoepp et al. 1997]) has been shown to ameliorate certain cognitive deficits in rats produced by the NMDA antagonist phencyclidine in vivo (Moghaddam and Adams 1998). Taken together, these results suggest that metabotropic agonists would be useful in normalizing excesses in glutamate release, regardless of the cause. The availability of orally active metabotropic glutamate receptor agonists makes it feasible to test the hypothesis that excessive glutamate release, particularly in such critical sites as the prefrontal cortex, plays a role in the positive and/or negative symptoms of schizophrenia.

Clearly, there is need to explore new treatment approaches in schizophrenia, because even the best of therapeutic responses obtained with existing typical or atypical antipsychotic drugs are often delayed and not fully restorative (Tamminga 1998). As suggested above, a possible reason for this lack of full efficacy may be that the primary site of pathology in schizophrenia may lie downstream from the receptors (D2, 5-HT2A, etc.) that are targeted by currently available drugs.