Sir

Presenting two clinical cases of schizophrenic patients, Cassey et al seek to back up their previously presented argument that the therapeutic action of valproic acid (VPA) could result from a modulation of dopamine (DA) signaling via GABAergic mechanisms (Casey et al, 2003; Wassef et al, 2003). The authors now describe that they obtained evidence from treating two patients concomitantly with VPA and neuroleptic drugs that VPA withdrawal is dampening postsynaptic DA signaling in the nigrostriatal circuit and tuberoinfundibular–pituitary axis (ie extrapyramidal symptoms and galactorrhea emerged) while increasing DA transmission in the mesolimbic system (ie psychotic symptoms increased). In the opinion of the authors, these effects of VPA would further strengthen the view that VPA impacts DA transmission via an increase of brain GABA concentrations, although in a differential way, depending on the target system. The notion, however, that VPA affects DA signaling via GABAergic mechanisms was recently put into question by us in pointing out that electrophysiological animal experiments did not find any effect of VPA administration on GABA-receptor-mediated neuronal activity at doses that are considered to be in the ‘therapeutic’ range (Winterer and Herrmann, 2000; Winterer, 2003).

The VPA effects on DA signaling that were suggested on the basis of the two clinical cases certainly cannot prove that GABAergic mechanisms account for the observed clinical changes that are generally related to DA transmission. VPA is a ‘dirty’ drug and affects many neurotransmission systems as reviewed elsewhere (Winterer and Herrmann, 2000). It is therefore conceivable that any effect of VPA on DA signaling is indirectly mediated by mechanisms that are independent of GABA. Also, it is uncertain as to whether the emergence of extrapyramidal symptoms or galactorrhea on discontinuation of VPA treatment during neuroleptic treatment can be generalized to larger patient populations. Likewise, it is currently not really clear as to whether VPA is possessing ‘antipsychotic’ treatment effects or whether it is rather ‘antiaggressive’, which may confound psychopathological assessments of psychotic symptoms (Winterer and Herrmann, 2000). In other words, it still remains to be shown whether VPA has any impact on GABAergic transmission and indirectly on DA signaling and whether these effects are the relevant mechanisms of VPA with regard to the treatment of schizophrenic patients. Nevertheless, the notion that VPA affects DA signaling via GABAergic mechanisms remains an attractive idea – simply because VPA increases GABA brain levels (Goden et al, 1969; Sawaya et al, 1975) and because VPA is effective in augmenting antipsychotic treatment of schizophrenic patients.

Under the assumption that VPA affects DA signaling through GABAergic mechanisms in schizophrenia illness, it appears to be pertinent for the illustration of the complexity of the problem to summarize briefly our current knowledge on the interaction of the two neurotransmission systems. During the past few years, it has become increasingly apparent that altered GABAergic function may critically contribute to schizophrenia pathophysiology. In particular, fast-spiking parvalbumin immunoreactive interneurons, that is, Chandelier cells, which seemingly play a central role in cortical recurrent inhibition and signal-to-noise ratio during information processing (Volk et al, 2002), have been implicated. For instance, several post-mortem studies of schizophrenic patients have consistently found decreased mRNA expression of glutamic acid decarboxylase GAD67, which is responsible for presynaptic GABA synthesis (Akbarian et al, 1995; Volk et al, 2000; Heckers et al, 2002) and these molecular changes do not appear to result from antipsychotic drug treatment (Lipska et al, 2003). Consequently, the question came up whether diminished GABAergic function interacts with impaired DA signaling in schizophrenia, since DA is considered another key player in schizophrenia illness (Kapur and Lecrubier, 2003).

In fact, there is currently emerging evidence—mainly from basic animal experiments—that an interaction between GABA and DA exists in both directions and at least on three levels, that is, (1) on the synaptic level, (2) on the level of neuronal microcircuits, and (3) on the level of neuronal macrocircuits. On the synaptic level, a crosstalk between DA D5 and GABA-A receptors resulting from direct protein–protein coupling has been recently reported (Liu et al, 2000). It was found that agonist-dependent protein–protein complex formation between the C-terminal tail domain of DA D5 receptors and the intracellular domain (IL2) of GABA-A receptors γ2 subunits permits the rapid, reciprocal modulation of GABA-A α1β2γ2 and DA D5-receptor-mediated events independent of their respective classically defined signal transduction cascades. The authors proposed that this mechanism might be potentially relevant in schizophrenia pathophysiology. On the microcircuit level, another GABA–DA interaction has been described. Using whole-cell patch-clamp recordings in vitro, and receptor-subtype-specific agonists and antagonists, Seamans et al (2001) demonstrated that DA D2-receptor agonists induce an early and brief (ie phasic) decrease in GABA release probability of cortical interneurons and a reduction of the response to a GABA-A agonist, whereas DA D1-receptor agonists cause a delayed and long-lasting (ie tonic) increase of the intrinsic excitability of interneurons. This D1-receptor-mediated action of DA appears to be specific for parvalbumin immunoreactive interneurons (Gorelova et al, 2002) and the relevance of these DA effects on GABAergic transmission with regard to schizophrenia pathophysiology are currently a matter of discussion (Winterer and Weinberger, 2004). Most investigations on the interaction between DA and GABA, however, have been conducted on the macrocircuit level. Thus, it is known already for some considerable time that GABA is influencing the firing of the nigrostriatal, mesolimbic, and mesocortical DA tracts. For instance, an inhibitory effect on DA neuronal firing has been demonstrated by iontophoretic application of GABA and GABA antagonists into the substantia nigra (SA) and ventral tegmental area (VTA) (Wolf et al, 1978; Waszczak and Walters, 1979). In addition, electrical stimulation of the targets of these tracts, that is, the nucleus caudatus and nucleus accumbens, exerts a reciprocal inhibition of the SA and VTA, which is blocked by the GABA antagonist bucculline (Yoshida and Precht, 1971; Wolf et al, 1978). On the other hand, there are also reports in conflict with these findings. Waszczak and Walters (1979) found that systemic application of the GABA agonist muscimol produced increased firing in DA neurons in pars compacta of SN, whereas non-DA neurons in the adjacent pars reticulata were inhibited. Similar findings were obtained by Grace et al (1980) and with regard to the DA cell firing in the VTA by Waszczak ans Walters (1979) or more recently by Cruz et al (2004). Therefore, the conclusion drawn by Garbutt and Van Kammen (1983), that is, that the effect of GABA on DA signaling depend on the brain region assayed, the duration of stimulation, and whether DA antagonists are administered, is probably correct.

The chain of arguments of Casey et al, that is, that VPA affects DA signaling via GABAergic mechanisms essentially rests on the early findings of a macrocircuit interaction of GABA and DA and the notion that VPA is increasing brain GABA concentrations via synthesis, reuptake, and metabolism. On the other hand, there is our argument that VPA does not affect GABA-mediated postsynaptic signaling in the therapeutic range in animal experiments. Given our current knowledge on the multilevel complexity of GABA–DA interactions, there are several theoretical possibilities that could reconcile this apparent contradiction. First of all, there is the very fundamental argument that animal findings cannot necessarily be generalized to humans. By extension, it also conceivable that the conditions in schizophrenic patients differ substantially from healthy humans. For instance, the above-cited evidence from post-mortem studies that GABAergic interneurons are functionally downregulated in schizophrenia could imply that VPA helps to restore interneuron function by increasing GABA synthesis. In other words, VPA may be functionally more relevant at pathologically low GABA baseline levels, whereas comparatively little effects are seen under normal conditions. Another question is whether VPA exerts its effects primarily on GABAergic and DA transmission on the level of macrocircuits as postulated by Casey et al. Thus, on the level of microcircuits, DA may have a differential impact via DA D1 and D2 receptors on interneurons depending on the – pathologically changed – GABA synthesis rate in these neurons. In principle, similar thoughts may also apply to the functional coupling of DA and GABA receptors on the synaptic level.

Taken together, it is currently still difficult to say whether VPA has an effect on DA signaling via altering GABA neurotransmission. If so, it is hard to tell on what level of neuronal processing this effect actually occurs and whether this mechanism is relevant for the VPA treatment of schizophrenia. Additional research is required to clarify these important and exciting questions.