Elsevier

Biological Psychiatry

Volume 46, Issue 10, 15 November 1999, Pages 1388-1395
Biological Psychiatry

Schizophrenia Research Series: Treatment
Neurotransmitter interactions in schizophrenia—therapeutic implications

https://doi.org/10.1016/S0006-3223(99)00117-1Get rights and content

Abstract

The search for new and improved antipsychotic agents has escalated during the past five years. The era of searching for non-toxic copies of clozapine has been followed by several different lines of research, some of which pursue the traditional dopamine track, although at a higher level of sophistication, whereas others focus on other neurotransmitters, such as serotonin and glutamate. Emerging knowledge about the interactions between different neurotransmitters in complex neurocircuits opens up possibilities for achieving antipsychotic activity by interfering with many different neurotransmitters. Most intriguing is the finding in animal experimental models, indicating that it should be possible to alleviate psychotic conditions by stabilizing rather than paralyzing neurocircuits, thus avoiding the risk of motor and mental side effects of the currently used drugs. Among these new classes dopaminergic stabilizers and 5-HT2A receptor antagonists seem to offer most promise at present. In a longer perspective, drugs interfering with glutamate function via different mechanisms may also turn out to be useful, especially in the control of negative symptoms.

Introduction

In recent years important progress has been made in basic schizophrenia research. The dopamine hypothesis of schizophrenia, that postulates a dopaminergic dysfunction in this disorder but has for a long time been supported only by indirect pharmacologic evidence (Carlsson A 1995), has now received more direct support by two different lines of research, both using imaging techniques. First, it has been shown that the synthesis of labeled dopamine or fluorodopamine in the brain, measured by means of PET after administration of radiolabeled dopa or fluorodopa, is increased in drug-naive schizophrenic patients, compared to age-matched controls Hietala et al 1994, Dao-Costellana et al 1997, Lindstrom et al 1997. Second, SPECT and PET studies, using a sophisticated technique to measure the release of dopamine in the basal ganglia in vivo, have shown that after an amphetamine challenge this release is elevated in drug-naive schizophrenic patients compared to age-matched controls, and that this elevation correlates to the induction of positive psychotic symptoms Laruelle et al 1996, Breier et al 1997, Abi-Dargham et al 1998.

Although these novel data are impressive, a number of caveats should be remembered. First of all, although aberrations of dopamine synthesis and release have been demonstrated in schizophrenic patients and shown to be statistically significant, the data show a considerable scatter, and thus the values observed are within a normal range in a certain proportion of the patients. Thus a dysfunction of dopamine may only occur in a subpopulation of patients suffering from this probably heterogeneous disorder (a few observations suggest that the aberration of dopamine synthesis may actually go in the opposite direction in catatonia, compared to other cases of schizophrenia). Second, it must be remembered that the abnormal values were obtained in patients who were challenged by amphetamine or exposed to the stress inevitably caused by the imaging procedures. Whether a dopamine dysfunction occurs also under minimal stress thus remains an open question.

Third, the patients examined were in acute episodes, and the situation may be different in chronic schizophrenic patients between episodes. In fact, recent observations by Laruelle et al (personal communication) indicate that the amphetamine-induced release of dopamine in schizophrenic patients in remission is within the normal range. This tallies with the frequently observed phenomenon that patients in remission complain more about the side effects of antidopaminergic drugs than during an exacerbation. If a normal level of dopaminergic function in patients in remission can be corroborated, the practical implications are obvious. All agents used today to prevent relapse in schizophrenia are antidopaminergic and should induce a state of hypodopaminergia, provided that the baseline level of dopamine function is normal. Hypodopaminergia is a most unpleasant and incapacitating condition leading to extrapyramidal side effects and, perhaps more importantly, to a failure of the reward system, resulting in dysphoria and anhedonia. To develop drugs capable of preventing relapse without these side effects should be an urgent task. In fact, as will be discussed below, such agents may already be underway.

An additional point deals with the interpretation of the data obtained with labeled dopa. The increased synthesis rate of labeled dopamine observed in the schizophrenic patients does not necessarily mean that the rate of endogenous dopamine synthesis is increased. It should be remembered that the rate-limiting step in the synthesis of dopamine is generally assumed to be the hydroxylation of tyrosine rather than the decarboxylation of dopa. A cautious interpretation of these interesting observations would thus be that there seems to exist in central dopaminergic neurons of schizophrenic patients a metabolic aberration involving the rate of dopa decarboxylation. The functional significance of this aberration has, however, not yet been fully clarified.

In addition to these caveats, the question must be raised whether dopamine is the only neurotransmitter showing dysfunction in schizophrenia. In view of the close interaction between neurotransmitters in the brain, it seems likely that other neurotransmitters show aberrations as well. The next question to be raised is whether the change in dopaminergic function is primary or secondary to aberrations elsewhere. As mentioned, the dopaminergic dysfunction in schizophrenia may even be a compensatory phenomenon.

In any event there is an obvious need to study the function of several other neurotransmitters in schizophrenia, e.g., noradrenaline, serotonin, acetylcholine, glutamate and gaba. These neurotransmitters are more difficult to study in the living intact brain than dopamine. Least difficult would perhaps be serotonin, because it seems possible to study it using the same kind of approach as for dopamine, that is to administer radiolabeled precursor (5-hydroxytryptophan) and measure the turnover of serotonin. Such a study has actually been carried out in depressed patients, in whom an abnormal serotonin turnover was demonstrated (Ågren et al 1991).

In recent years considerable interest has focussed on the possible role of glutamate in schizophrenia Kim et al 1980, Garland Bunney 1995. One reason for this is the discovery that phencyclidine (PCP, “angel dust”), that can induce a psychotic condition mimicking schizophrenia, perhaps even more faithfully than the amphetamines, is a powerful antagonist on one of the glutamate receptor subtypes, namely the NMDA receptor. This receptor is equipped with an ion channel regulating the penetration of calcium and other cations into the neuron. PCP binds to a specific site in this channel, thereby blocking the function of the receptor. Recently a number of other NMDA antagonists have been identified, some of them binding to the “PCP site,” e.g., MK-801 and ketamine, others binding competitively to the same site as glutamate, e.g., AP5, d-CPPene and CGS 19755, and still others to a site on the NMDA receptor, where glycine functions as an additional agonist. An example of antagonists acting at the glycine site is d-cycloserine. All these different NMDA antagonists seem to be psychostimulants, at least in rodents, and are psychotogenic in humans (review by Lodge and Johnson 1989).

Thus a deficiency of glutamate function in schizophrenia must be considered. At an early stage it was suggested that the psychotogenic action of glutamate antagonists could be mediated by an increased catecholaminergic activity. Dopamine neurons, like other monoaminergic brainstem neurons, seem to be controlled by corticofugal glutamatergic neurons either directly or via gabaergic interneurons, acting as accelerators and brakes, respectively (Figure 1). An enhanced dopaminergic activity, mediated via hypoglutamatergia, could then be induced by a failure of the brake. Normally there seems to be a balance between the accelerator and the brake, but if for example the dopamine function is enhanced, e.g., by a releasing agent such as amphetamine, then a negative feedback regulation is probably activated, leading to a strong overweight of the brake. This can be demonstrated in experimental animals by superimposing an NMDA antagonist upon amphetamine. Then the release of dopamine is dramatically enhanced (Miller and Abercrombie 1996). This phenomenon is of immediate clinical interest, because it opens up a possibility to explain the previously mentioned, enhanced amphetamine-induced release in schizophrenic patients. This enhancement could be due to a glutamate deficiency, leading to a weakened negative feedback control. In fact, co-treatment with the NMDA antagonist ketamine has been found to cause enhancement of the amphetamine-induced dopamine release in humans, as demonstrated by means of SPECT (Laruelle, personal communication).

These findings can of course not be extrapolated to mean that hypoglutamatergia, in the absence of a challenge induced by, e.g., amphetamine, should also lead to enhanced dopamine release. Actually, treatment of experimental animals with NMDA antagonists has given ambiguous results. For example, in rats Miller and Abercrombie (1996) found but a slight, not dose-dependent release of dopamine, studied by microdialysis, after treatment with MK-801. Using the same technique other laboratories have found similar, more or less impressive effects of this agent. As to the competitive NMDA antagonists, the available evidence suggests that these agents, if anything, inhibit dopamine release, and this decrease is concomitant with behavioral stimulation (Waters et al 1996). Thus we have to look for a mechanism other than increased dopamine release to account for at least an important part of the psychostimulant action of NMDA antagonists.

NMDA receptor antagonists seem to stimulate 5-HT turnover and release more consistently than dopaminergic activity. This is of special interest in view of the striking effect of the selective 5-HT2A antagonist M100,907 (Schmidt et al 1995) on the behavioral stimulation induced by NMDA-receptor antagonism (Martin et al 1998). This effect can be seen after doses of M100,907 that are unable to influence the activity of normal mice. In fact, hyperserotonergia seems to be a prerequisite for this antagonism (Martin et al 1998). This remarkable profile of M100,907 may have important therapeutic implications.

PCP, that is a somewhat less selective NMDA antagonist than MK-801, does indeed cause a more pronounced release of dopamine, probably due to a concomitant blockade of the dopamine transporter. The psychostimulation caused by PCP does not seem to depend so much on this enhanced release, because it can be nearly abolished by LY354740, a group II metabotropic glutamate receptor agonist, despite the fact that this agonist leaves the enhanced dopamine release unchanged (Moghaddam and Adams 1998). Interestingly, in this study PCP was found to enhance the release of glutamate, and this effect was antagonized by LY354740. This phenomenon will be commented below.

That the behavioral stimulation induced by NMDA receptor blockade can occur independently of dopamine was demonstrated 10 years ago, when Carlsson and Carlsson (1989) reported that MK-801 is capable of inducing motor activity in mice completely depleted of dopamine and noradrenaline by pretreatment with reserpine plus α-methyltyrosine. Subsequent work showed that competitive NMDA-receptor antagonists were also active under these conditions, and that not only systemic but also local treatment with NMDA antagonists in the nucleus accumbens was able to induce movements despite virtually complete monoamine depletion (Svensson and Carlsson 1992).

Whereas the local administration of NMDA receptor antagonists in the nucleus accumbens of monoamine-depleted mice induced a fairly normal motility pattern, the systemic treatment with these drugs caused a highly abnormal motility, i.e., compulsory forward locomotion with apparently total loss of the ability to switch between different behavioral patterns. A possible explanation of this difference may be that systemic treatment will lead to antagonism of NMDA receptors not only in the basal ganglia but also in the cerebral cortex, where the failure of glutamatergic association pathways could lead to loss of important functions, such as the ability to select appropriate behavioral programs. If glutamatergic deficiency is a relevant pathogenetic mechanism in schizophrenia and if this includes the cerebral association pathways, it is not far-fetched to propose that this could lead to important consequences, involving cognitive disturbances, loss of flexibility, ambivalence and other behavioral aberrations, perhaps mainly belonging to the sphere of negative schizophrenic symptomatology. Hypofrontality could also be a result of failure of cortical association pathways, and these could be especially vulnerable in so far as they engage chains of glutamatergic pathways.

Our subsequent work revealed a dramatic synergism between a variety of monoaminergic agonists and MK-801 or other NMDA receptor antagonists Carlsson and Carlsson 1990, Carlsson and Svensson 1990, Carlsson 1995. This was true, for example, of apomorphine, a mixed D1/D2 agonist, SKF 38393, a selective D1 agonist, clonidine, an α2-adrenergic agonist, and LSD, a 5HT2 agonist. A synergy between muscarinic and NMDA receptor antagonists was also demonstrated. Because these phenomena could be demonstrated in the absence of monoamines, the synergism must be assumed to occur postsynaptically in the ventral striatum. The exact mechanism of this synergism is not clear. It may occur locally or involve some kind of loop-mediated regulation.

Based on these observations, we have proposed a hypothetical scheme, illustrating the interaction between several neurotransmitters to form a network of psychotogenic pathways (Figure 2).

These observations lend support to a previously advanced hypothesis (Carlsson 1988), that psychomotor activity and psychotogenesis depend on an interplay between dopamine and glutamate projecting to the striatum from the lower brainstem and cortex, respectively (Figure 3). These neurotransmitters are predominantly antagonistic to each other when acting on striatal gabaergic projection neurons, the former being inhibitory and the latter stimulating. These gabaergic projection neurons belong to so-called indirect striatothalamic pathways, that exert an inhibitory action on thalamocortical glutamatergic neurons, thereby filtering off part of the sensory input to the thalamus to protect the cortex from a sensory overload and hyperarousal. Hyperactivity of dopamine or hypofunction of the cortico striatal glutamate pathway should reduce this protective influence and could thus lead to confusion or psychosis.

The above-mentioned hypothesis (Carlsson 1988) focussed on the indirect striatothalamic pathways, that have an inhibitory influence on the thalamus. The corresponding direct pathways exert an opposite, excitatory influence. Both pathways are controlled by glutamatergic cortico striatal fibers, enabling the cortex to regulate the thalamic gating in opposite directions. In other words, they seem to serve as brakes and accelerators, respectively, in analogy to the regulation of monoaminergic brainstem neurons mentioned above. Normally, the inhibitory, indirect pathways seem to dominate over the direct pathways. Thus, NMDA receptor inhibitors are behavioral stimulants, at least in rodents. The balance between the direct and indirect pathways may vary, depending on the state of the system. Failure of the direct pathway, induced, e.g., by glutamatergic deficiency, might contribute to the so-called negative symptomatology of schizophrenia. It has been suggested that the activity of the direct pathways is predominantly phasic, whereas that of the indirect pathways is mainly tonic (Alexander and Crutcher 1990). This difference could have important consequences for a differential responsiveness of the direct and indirect pathways to drugs.

Needless to say, the postulated existence of a thalamic filter would not exclude a gating function located in other parts of the brain. The impressive sophistication of the gating function, enabling a focussed attention to relevance and novelty at the expense of trivial sensory inputs, would actually speak in favor of a more widely spread location.

Little is known about the role of different receptor subtypes in the respective pathways. As to the glutamatergic receptors, NMDA receptor antagonists, as mentioned, are behavioral stimulants, and this has been interpreted as the result of a failure of the indirect, inhibitory pathways. AMPA receptor antagonists have been studied less intensely but have been found to act in the same direction as NMDA antagonists in some experiments, whereas they act as antagonists to NMDA antagonists in other experiments. As to the metabotropic receptors, the recent observations of Moghaddam and Adams (1998) briefly referred to above, are most interesting. They found that the behavioral stimulation caused by PCP could be antagonized by a metabotropic receptor agonist, and at the same time the PCP-induced elevation of glutamate release was antagonized. The question arises if these data can be accommodated to the model of direct and indirect pathways outline above. Glutamate release is, generally speaking, much more difficult to measure and interpret than, e.g., dopamine release. For example, glutamate plays an important role in general cell metabolism, in addition to serving as a neurotransmitter. Perhaps glutamate release is predominantly indicative of the activity of the direct pathways, because they seem to be mainly phasic, and release by burst firing may be more likely to show up in microdialysis. Thus the PCP-induced elevation of glutamate release, as measured by microdialysis, is perhaps indicative of an increased activity of the direct pathway; possibly the metabotropic receptor agonist antagonized this release by stimulating glutamatergic autoreceptors. Of course, such a mechanism is speculative.

At this point the choice between the two major pharmacological models of schizophrenia is difficult. It may have to await the comparison between different pharmacological treatments. Animal data suggest that such a strategy is feasible. Whereas haloperidol turned out to be superior to M100,907 in antagonizing amphetamine-induced hyperactivity, the reverse was true of MK-801-induced hyperactivity (Carlsson et al 1999). As yet it is not clear if M100,907 does in fact possess antipsychotic activity; it will have to await the outcome of ongoing phase-3 trials. If the outcome is positive, the interesting question arises if some patients respond better to one type of drugs than to the other and if the reverse is true for other patients. The possibility may thus open up to distinguish between patient populations with presumably different pathogenesis. Another possibility will be that combined treatment will be superior to either treatment alone. The latter alternative may be suggested by the apparently superior efficacy of clozapine and other “atypical” antipsychotic agents possessing both antidopaminergic and antiserotonergic activity, but the existence of a number of additional properties of the “atypical” group of agents makes this prediction uncertain.

The advent of a number of agents interacting in different ways with the glutamatergic system, that are now in different stages of development, is eagerly awaited. Examples of this group of drugs are the glycine agonists, glycine reuptake inhibitors, AMPA agonists and antagonists, and ampakines. In the case of AMPA ligands it seems at present uncertain if agonists, antagonists or partial agonists/modulators will be most successful. Finally, drugs acting on different subtypes of metabotrophic glutamate receptors seem to offer some promise.

Is the therapeutic potential of dopaminergic agents exhausted? Several reasons support the view that this is far from the case. First of all, the role of the various subtypes of dopamine receptors need to be further explored. But perhaps even more important will be some ongoing attempts to reach a deeper understanding of the function of these receptors. This may open up entirely new ways to improve this function and to optimize the receptors’ ability to cope with aberrations in neural circuits. In support of this prediction some recent observations in our research group will be briefly mentioned.

We have developed a series of compounds that are capable of stabilizing the dopaminergic system in different ways without inducing the hypodopaminergia, so characteristic of the currently used antipsychotic drugs. Some of these new drugs are partial dopamine receptor agonists, acting on the D2 family of receptors. A number of partial dopamine receptor agonists, developed by us and by others are now in clinical trials and seem to offer promise (for recent clinical data on (−)-3PPP, see Lahti et al 1988; several of the partial dopamine receptor agonists studied so far are less suitable as probes because of their poor selectivity, e.g., the ergolines terguride and SDZ 208-912). Others are pure antagonists, again acting on the D2 family of receptors, and can thus antagonize signs of elevated dopamine functions, but in contrast to the currently used antipsychotic agents they do not cause hypodopaminergia. In fact, they rather tend to antagonize subnormal dopamine function. The reason for this aberrant pharmacological profile seems to be that their action on different subpopulations of dopamine receptors differs from that of the currently used drugs. Thus, whereas they exert a strong action on dopaminergic autoreceptors, they have a weaker effect postsynaptically and seem unable to reach a subpopulation of postsynaptic dopamine receptors Svensson et al 1986, Sonesson et al 1994, Hansson et al 1995. In subhuman primates, where Parkinsonism had been induced by MPTP, one member of this class, named (−)-OSU6162, has been found capable of preventing l-dopa-induced dyskinesias without interfering with the therapeutic movement response, and in subsequent trials on Parkinsonian patients the same kind of response was observed (J. Tedroff et al, unpublished data). Subsequent trials on patients with Huntington’s disease showed a marked reduction of choreatic movements, considerably outlasting the presence of the drug in the blood. These observations support the view that drugs of this class are capable of stabilizing dopaminergic tone, that is, they are able to alleviate signs of hyperdopaminergia without inducing any signs of reduced dopaminergic function. If these findings can be extrapolated from neurology to psychiatry, these agents should possess antipsychotic activity without any concomitant signs of hypodopaminergia. Forthcoming trials with such agents in schizophrenia will answer this question.

Section snippets

Conclusion

Although remarkable progress has been made regarding the role of neural networks and neurotransmitters in schizophrenia, the available data can offer themselves to a variety of interpretations. In summarizing the data presented above we would like to present the following, tentative interpretation. Whereas a number of subpopulations with different pathogenesis may exist among schizophrenic patients, a mechanism involving a glutamatergic deficiency seems to deserve special attention. This

Acknowledgements

This work was presented at the conference, “Schizophrenia: From Molecule to Public Policy” held in Santa Fe, New Mexico in October 1998. The conference was sponsored by the Society of Biological Psychiatry through an unrestricted educational grant provided by Eli Lilly and Company.

References (30)

  • A. Carlsson

    The dopamine theory revisited

  • M.L. Carlsson

    The selective 5-HT2A receptor antagonist MDL 100,907 counteracts the psychomotor stimulation ensuing manipulations with monoaminergic, glutamatergic or muscarinic neurotransmission in the mouse—implications for psychosis

    J Neural Transm

    (1995)
  • M.L. Carlsson et al.

    The NMDA antagonist MK-801 causes marked locomotor stimulation in monoamine-depleted mice

    J Neural Transm

    (1989)
  • L.M. Carlsson et al.

    Interfering with glutamatergic neurotransmission by means of MK-801 administration discloses the locomotor stimulatory potential of other transmitter systems in rats and mice

    Pharmacol Biochem Behav

    (1990)
  • M.L. Carlsson et al.

    The 5-HT2A receptor antagonist M100907 is more effective in counteracting NMDA antagonist- than dopamine agonist-induced hyperactivity in mice

    J Neural Transm

    (1999)
  • Cited by (0)

    View full text