Introduction

The hippocampus has been implicated in the neuropathology of psychiatric disorders. Magnetic resonance imaging (MRI) studies have consistently demonstrated a reduction in the hippocampus volume in patients with psychiatric disorders [1–3] and functional MRI studies have shown changes in the hippocampal activity that are associated with symptoms of psychiatric disorders [4, 5]. Histological studies on the postmortem brain have revealed changes in expression of various genes in the hippocampus of the patients (see Table 1). These studies in humans have provided valuable information as to potential molecular targets for the treatment of the diseases and would also help develop an objective way to elucidate effects of ongoing treatments on the hippocampus. However, it is still poorly understood how functions of the hippocampal neurons are actually altered in the brain of patients with psychiatric disorders. In order to investigate cellular mechanisms underlying a particular disease, detailed studies using animal models would be indispensable. Despite fundamental difficulties in modeling abnormalities of human mind using experimental animals, biological approaches to psychiatric disorders that utilize animal models are getting generally accepted. In conventional pharmacological approaches, psychotropic drugs used in humans are administered to experimental animals assuming that drug-induced cellular and/or molecular changes in animal brains would be comparable to those in human brains. Accumulating evidence suggests that mice with the targeted mutation in disease-associated genes would be valuable models for psychiatric disorders [6]. The validity of the mutant mouse model is best exemplified by the finding of association of the gene encoding a subunit of calcineurin with schizophrenia. In this case, the association was predicted based on behavioral characterization of calcineurin mutant mice [7, 8] and confirmed by human genetic studies [9–11]. Recent studies using experimental animals have proposed the critical importance of the dentate gyrus and its output, the mossy fiber, in the pathophysiology and treatment for psychiatric disorders. Especially, there is an emerging possibility that abnormalities in the functional maturation of the dentate gyrus and mossy fiber synaptic transmission are critically involved in schizophrenia and other related psychiatric disorders. It should be noted that this idea is partly based on observations in electrophysiological studies that revealed unique properties of the mossy fiber synapse and marked dysfunction of this synapse relevant to psychiatric disorders. In this paper, I review recent studies addressing the involvement of the dentate gyrus and mossy fiber in psychiatric disorders, mainly focusing on evidence derived from experimental animals. I start with a summary of physiological properties of the mossy fiber synapse, and discuss why dysfunction of mossy fiber can substantially affect hippocampal functions and how pharmacological treatments can modify the mossy fiber synaptic transmission.

Table 1 Abnormalities in the dentate gyrus and mossy fiber observed in the postmortem brain of patients with schizophrenia
Full size table

Unique Role for Mossy Fibers in Hippocampal Neuronal Circuit

The hippocampal neuronal circuit is comprised of anatomically distinct three major areas connected by glutamatergic excitatory synapses. The dentate gyrus is positioned at the entrance of this so-called trisynaptic circuit and has been implicated in pattern separation or disambiguation of similar patterns of sensory inputs [12–15]. The principal neurons of the dentate gyrus, the granule cells, have relatively low in vivo firing rates [16], which may be at least partly due to very polarized resting membrane potentials of these cells [17–20]. The axons of the granule cells, the mossy fibers, form large synaptic terminals, typically 3 to 5 μm in diameter, preferentially at the proximal apical dendrite of pyramidal cells in the CA3 region [21]. Based on the anatomical feature, the mossy fiber synapse has been theoretically hypothesized to be a “detonator” or “teacher” synapse in the CA3 network [22, 23]. This theoretical prediction was confirmed by later experimental studies, but in a slightly modified form (see below).

The mossy fiber-CA3 pyramidal cell synapse has unique physiological properties. The mossy fiber synaptic transmission has a very wide dynamic range and is strongly dependent on the presynaptic firing rate [24–26]. For example, a rise in the stimulus frequency from 0.1 or 0.05 to 1 Hz leads to an increase in the synaptic efficacy by 500% to 700% [20, 24, 27]. Activity-dependent short-term modifications of synaptic efficacy are called short-term synaptic plasticity and are mostly dependent on presynaptic mechanisms [28]. Although the short-term plasticity is generally observed in a form of either facilitation (enhancement) or depression, the magnitude of the mossy fiber synaptic facilitation, especially, the frequency facilitation described above, is exceptionally large. Due to this frequency dependence of the transmission efficacy, the mossy fiber synapse can work as a high-pass filter. Indeed, Henze et al. showed that high-frequency stimulation of the granule cells in vivo reliably evokes spiking in postsynaptic CA3 neurons, but single spikes or low-frequency stimulation is much less effective [29]. Importantly, high-frequency activation of even single granule cells is sufficient for firing postsynaptic cells [29]. Furthermore, in slice preparations, activation of putative single granule cells in high-frequency bursts provides effective postsynaptic activity for the induction of long-term potentiation (LTP) at converging non-mossy fiber inputs [30]. Therefore, the mossy fiber input can serve as a “conditional detonator” that has strong impact on activity of postsynaptic cells only during high-frequency transmission [29]. Collaterals of the mossy fibers make synaptic contacts on neurons in the hilar region [31, 32]. The hilar neurons include inhibitory interneurons and the excitatory mossy cells, which innervate the granule cells and interneurons [33, 34]. The mossy fiber-mossy cell synapse also shows large frequency facilitation and is generally similar to the mossy fiber-CA3 pyramidal cell synapse [35]. The magnitude of the frequency facilitation at the mossy fiber-CA3 synapse increases with postnatal development and reaches the adult level in 3 to 4 weeks [27]. The somatic properties of the granule cells also change with development. Immature granule cells are more easily excited by somatic current injection than mature granule cells. However, repetitive spiking can be evoked in mature granule cell, but not in immature cells [18, 19]. Therefore, the dentate-to-CA3 neuronal system develops into the mature state that is suitable for transmitting high-frequency signaling (Fig. 1).

Fig. 1
figure 1

Developmental changes in physiological properties of the granule cell soma and mossy fiber synapse. Immature granule cells are more easily excited than mature cells by somatic current injection (top). However, sustained current injection can evoke repetitive action potentials in mature granule cell, but not in immature cells (middle). The frequency facilitation of the mossy fiber synaptic transmission increases with development (bottom). Therefore, mature granule cells can much more effectively transmit high-frequency signaling than immature granule cells

Full size image

Long-term plasticity has also been demonstrated at the mossy fiber synapse. In most excitatory synapses in the hippocampus, the induction of long-term synaptic plasticity requires coincident pre- and postsynaptic activities and the resultant activation of N-methyl-d-aspartate (NMDA) subtypes of glutamate receptors [36]. However, at the mossy fiber synapse, neither NMDA receptors nor postsynaptic depolarization is required for LTP [37–40; but see 41–43] or long-term depression [24, 44; but see 45]. Therefore, presynaptic firing patterns primarily regulate the efficacy of the mossy fiber synapse in both short and long terms.

The mossy fiber synaptic transmission is also regulated by various neurotransmitters. Glutamate autoreceptors play a variety of roles in both short- and long-term plasticity [46–48; reviewed by 49]. Endogenous opioids can be released from mossy fiber terminals during high-frequency firing and regulate the induction of mossy fiber LTP [50]. This opioid-mediated modulation is impaired after chronic morphine treatments [51]. Ambient adenosine has been shown to continuously and strongly inhibit the mossy fiber synaptic transmission. In the absence of the adenosine-mediated tonic inhibition, the synaptic efficacy is greatly increased and the short-term plasticity is suppressed [52; but see 53]. Therefore, a change in extracellular adenosine concentrations in pathological conditions such as ischemia and epilepsy [54] would substantially affect the mossy fiber synaptic transmission and the CA3 circuit activity. The mossy fibers transmit sub-threshold “analog” signals as well as action potentials [55]. Since the sub-threshold depolarization in the granule cell soma can enhance the action potential-mediated transmitter release from the mossy fiber terminals [55], modulations of the somatodendritic membrane conductance, such as tonic inhibition by γ-aminobutylic acid [56], would be also important in regulating the mossy fiber synaptic transmission. Recently, it has been demonstrated that the mossy fiber synaptic transmission is potentiated by dopamine [57, 58] and serotonin [59]. Noradrenaline has been shown to modulate the induction of LTP without affecting the basal transmission [60]. The serotonergic modulation of the mossy fiber synapse is a plausible target for antidepressant drugs (see below). It should be noted that all these monoaminergic modulations of the mossy fiber synapse are mediated by activation of Gs-coupled receptors. The mossy fiber synapse is highly sensitive to intracellular cAMP concentrations. An adenylate cyclase activator, forskolin, induces marked potentiation of the synaptic transmission [58, 59, 61], and an inhibitor of phosphodiesterase that catalyzes cAMP hydrolysis augments the synaptic transmission and the monoaminergic modulations [58, 59]. Cyclic AMP signaling pathways have been implicated in antidepressant and ntipsychotic effects. Phosphodiesterase inhibitors have been shown to have antidepressant-like [62] and antipsychotic-like effects [63, 64]. The intracellular pathways as well as the receptors involved in the mossy fiber synaptic modulations would be potential targets for treatments of psychiatric disorders.

Effects of Antidepressants on Dentate Gyrus and Mossy Fiber Synapse

It is generally accepted that new neurons are continuously generated in the dentate gyrus of the adult brain. The adult neurogenesis has been demonstrated in various mammalian species including humans [reviewed by 65]. Newly generated neurons can be functionally integrated into the existing neuronal circuits [66–68]. In experimental animals, the adult neurogenesis in the dentate gyrus is increased by chronic administration of antidepressant drugs including tricyclic antidepressant drugs, selective serotonin reuptake inhibitors (SSRIs), and monoamine oxidase inhibitors [69–71]. Electroconvulsive stimulation (ECS), an experimental model of electroconvulsive therapy for depression, also increases the adult neurogenesis [69]. Chronic treatment with fluoxetine, a widely used SSRI, can stimulate the maturation of newly generated neurons [72]. Since various forms of stress can depress the adult neurogenesis, it has been hypothesized that the adult neurogenesis is involved both in the pathogenesis of depression and in antidepressant effects [reviewed by 73]. Indeed, ablation of the proliferation of neuronal progenitor cells in the adult hippocampus by x-irradiation prevents behavioral effects of antidepressant drugs [70, 71, 74]. However, x-irradiation itself neither increased depression-related behaviors nor affected stress-induced changes in behaviors. Furthermore, x-irradiation does not prevent antidepressant-like effects of a corticotropin-releasing factor 1 antagonist or a vasopressin 1b antagonist [74]. Therefore, the intact cytogenesis in the dentate gyrus does not seem to be essential for depression-related behaviors themselves, but is likely required for some of the behavioral effects of antidepressant drugs. Although the facilitatory effects of antidepressants on the adult neurogenesis have been consistently demonstrated in animal studies, so far there has been no clear evidence for the involvement of the neurogenesis in either depression or antidepressant actions in humans [75].

Antidepressant treatments generally increase mRNA levels of brain-derived neurotrophic factor (BDNF) in the hippocampus [reviewed by 76]. BDNF is abundantly expressed in the dentate gyrus and along the mossy fiber pathway [77, 78]. A recent study demonstrated that the selective deletion of BDNF in the dentate gyrus suppresses behavioral effects of subchronic antidepressant treatments [79], further supporting the importance of the dentate gyrus in antidepressant effects. In the postmortem human brain, an increase in BDNF protein levels was found in the dentate gyrus of subjects treated with antidepressant medications at the time of death, compared with antidepressant-untreated subjects [80]. In experimental animals, ECS consistently increases BDNF protein as well as mRNA levels [81–83]. However, after pharmacological antidepressant treatments, hippocampal BDNF protein levels are not significantly changed [81, 83, 84], decreased [82], or increased only in the hippocampus proper, but not in the dentate gyrus [85]. Interpretation of these results are further complicated by the fact that proBDNF, from which mature BDNF was processed, can also be secreted and work as a signaling molecule with distinct actions [86–88]. In order to understand a role for BDNF in antidepressant actions, changes in protein levels of both mature and proBDNF after antidepressant treatments have to be elucidated.

Given the importance of the dentate gyrus in antidepressants actions, one might expect that antidepressant treatments also affect the mossy fiber system. However, so far only a few studies have addressed this issue. Repeated ECS induces sprouting of mossy fibers [89–91], but antidepressant drug treatments do not [91]. The ability of ECS to induce mossy fiber sprouting might be related to the superiority of the electroconvulsive therapy in effectiveness against depression as compared with antidepressant drugs [92]. However, the clinical significance of the ECS-induced sprouting remains to be elucidated. Recently, Kobayashi et al. showed that chronic administration of fluoxetine causes robust changes in the serotonergic modulation of the mossy fiber synaptic transmission in mice [59]. Serotonin (5-hydroxytriptamine (5-HT)) potentiates the mossy fiber synaptic transmission via activation of 5-HT4 receptors. Chronic fluoxetine reduced the synaptic potentiation induced by higher concentrations of 5-HT but enhanced that induced at lower concentrations, which represents the stabilization of the serotonergic modulation (see Fig. 2). The reduction and enhancement of the serotonergic modulation by chronic fluoxetine could be explained by downregulation of the 5-HT transporter and by desensitization/downregulation of 5-HT4 receptors or downstream effectors, respectively. Chronic antidepressant treatments reduce effects of a synthetic 5-HT4 agonist on excitability of hippocampal pyramidal cells [93, 94]. The desensitization or downregulation of 5-HT receptors by antidepressant treatments has been demonstrated in other brain regions both in experimental animals and humans [reviewed by 95, 96]. The downregulation of the serotonin transporter after chronic SSRI treatments can be seen in whole brain regions [97, 98]. Therefore, it is possible that the bidirectional modulation or stabilization of the serotonergic modulation can be generally induced by chronic SSRI treatments. While the clinical significance of the stabilizing action is unknown, the effect of chronic fluoxetine on the mossy fiber synapse is associated with reduced locomotor activity of mice in novel environments [59], suggesting that the stabilization can contribute to the suppression of excess reactivity to external stimuli.

Fig. 2
figure 2

Possible mechanisms underlying stabilizing effects of chronic fluoxetine on serotonergic modulations. Assume that 5-HT facilitates glutamate release by activating presynaptic 5-HT receptors. Downregulation and/or block of 5-HT transporter by fluoxetine increase extracellular 5-HT concentrations. Downregulation of 5-HT receptors and/or modulation of downstream signaling reduce effects of 5-HT on the glutamate release. These changes can lead to stabilization of the serotonergic modulation

Full size image

The serotonergic stabilizing effects of chronic fluoxetine described above may not be due to either facilitation of the adult neurogenesis or changes in BDNF expression. The facilitation of the adult neurogenesis may increase the proportion and the number of young granule cells, which could then change functions of the dentate gyrus. Chronic SSRI treatments have been shown to reduce expression of calbindin, a marker for mature granule cells, in the hippocampus [99, 100]. Newly generated granule cells exhibit enhanced LTP only between 1 and 1.5 months of the cell age [101]. The chronic fluoxetine treatment indeed causes the enhancement of LTP at the perforant path-granule cell synapse that is blocked by x-irradiation [72; but see 102]. The enhanced neurogenesis may also affect the mossy fiber synaptic transmission, since the functional properties of the mossy fiber synapse change with development [26, 27]. BDNF is required for the development of the dentate gyrus [103] and has been shown to acutely affect CA3 population spikes evoked by mossy fiber stimulation [104]. Therefore, chronic antidepressant treatments may also affect intrinsic properties of the mossy fiber synaptic transmission in addition to the serotonergic modulation. A preliminary study suggests that chronic fluoxetine at relatively high doses reduces the mossy fiber synaptic facilitation in adult mice [105]. Although this result is consistent with the idea that chronic fluoxetine increases the proportion of young granule cells, this issue largely remains to be elucidated.

Involvement of Dentate Gyrus and Mossy Fiber in Neuropathology and Pathophysiology of Schizophrenia

The hippocampus has been implicated in schizophrenia [reviewed by 106, 107]. MRI studies have shown reduced hippocampal size in patients with schizophrenia [1, 3]. Since the hippocampus plays an important role in working memory [13] and prepulse inhibition [108], which are impaired in patients with schizophrenia [109, 110], it is highly likely that hippocampal dysfunction can cause some of symptoms seen in schizophrenia. In experimental animals, selective lesions of the dentate gyrus by colchicine have been shown to impair working memory [111, 112]. Studies on the postmortem brain revealed substantial abnormalities in the dentate gyrus in patients with schizophrenia (Table 1). Those include changes in expression of various transmitter receptors [113–122], other synaptic proteins [123, 124], and proteasome, ubiquitin, and mitochondrial genes [125], suggesting profound functional defects of the dentate gyrus in schizophrenia. The mossy fiber system also has some abnormalities in synaptic structure [126], density of synaptic terminals [127], and expression of presynaptic proteins [128]. It is worth noting that some of these changes are relatively specific to the dentate gyrus, but not observed in CA1 (see Table 1), as represented by a decease in expression of the dysbindin-1 (DTNBP1) gene [129]. DTNBP1 is one of the most promising schizophrenia susceptibility genes [130] and mice with a deletion in the DTNBP1 gene show impairment in working memory [131]. Dysbindin-1 has been shown to regulate release of glutamate neurotransmitter [132, 133]. Intense dysbindin-1-like immunoreactivity is seen in the mossy fiber terminal area and the dentate inner molecular layer in humans [134, 135]. This dysbindin-1-like immunoreactivity is strongly reduced in schizophrenia [135], raising the possibility that the mossy fiber synaptic transmission is impaired in patients with schizophrenia. It should also be noted that there are abnormalities related to neuronal development and maturation such as reduced neural progenitor cell proliferation [75], reduced density of reelin positive cells [136, 137], increased expression of retinoic acid receptor α [138], and reduced expression of the mature granule cell marker calbindin [125]. Abnormal neurodevelopment has been implicated in the pathogenesis of schizophrenia [139]. Therefore, it is possible that the dentate gyrus of schizophrenic patients fails to develop into the proper mature state.

Studies in experimental animals support this hypothesis. The NPAS3 (neuronal PAS domain protein 3) gene has been shown to be disrupted by chromosomal translocation in a family with schizophrenia [140]. Mice lacking NPAS3 exhibit hyperactivity [141] and greatly reduced adult neurogenesis in the dentate gyrus [142]. Disrupted-in-Schizophrenia 1 (DISC1) is one of the most plausible schizophrenia susceptibility genes [130]. DISC1 is abundantly expressed in the hippocampus throughout the embryonic and postnatal development [143, 144]. In the adult forebrain, DISC1 expression is more localized than in embryo, with the most prominent expression in the dentate granule cells [143–146]. DISC1 is also strongly expressed in the dentate granule cells of the adult human brain [147]. Virus-mediated knockdown of DISC1 in newly generated granule cells in adult mice causes mispositioning of the cell, accelerated dendritic development, and formation of basal dendrites that are not usually seen in rodents [148]. The formation of basal dendrites resembles morphological changes observed in patients with schizophrenia [149]. The DISC1 knockdown also impairs targeting of the mossy fibers, accelerates the development of mossy fiber synaptic boutons, and hampers full morphological maturation in some boutons [150]. Furthermore, a mutation in DISC1 that models a schizophrenic risk allele impairs dendritic growth during the postnatal development of the granule cells [151]. These results are consistent with the hypothesis that the development or maturation of the dentate gyrus is impaired in schizophrenia. As described above, the remarkable frequency facilitation at the mossy fiber synapse is established during the postnatal development [27] and somatic properties of the granule cells also strongly depend on the developmental stage of the cell [17–19] (see Fig. 1). Therefore, these synaptic and somatic properties are supposed to be altered in the dentate gyrus with abnormalities in development and possibly in the brain of patients with schizophrenia. A recent study on mice heterozygous for alpha-calcium and calmodulin-dependent protein kinase II (alpha-CaMKII +/− mice) provided a strong support for this idea [20]. Alpha-CaMKII +/− mice show profound behavioral abnormalities including hyperactivity, a severe working memory deficit, and an exaggerated infradian rhythm, which are related to symptoms in schizophrenia and other psychiatric disorders. In the dentate gyrus of the mutant mice, the granule cells have less developed dendritic arborization. The expression of calbindin is largely suppressed, while markers for immature granule cells are increased. As expected, the physiological properties of the mutant granule cells are strikingly similar to those of immature granule cells. Thus, the magnitude of mossy fiber synaptic facilitation is strongly reduced, and somatic spikes are easily evoked, but reduced in number during sustained depolarization. Analysis of the gene expression of the postmortem human hippocampus using biomarkers that characterize the mutant hippocampus suggested similarities in gene expression patterns between alpha-CaMKII +/− mice and patients with schizophrenia, which include the reduction of calbindin expression [20]. Another piece of evidence for the present hypothesis was provided by studies on mice lacking the kainite receptor subunit GluR6. GluR6 is essential for the large facilitation of the mature mossy fiber synapse [27, 152, 153]. The GluR6-deficient mice show severe behavioral abnormalities including hyperactivity and increased responsiveness to psychostimulants, which are related to mania or psychosis [154]. Consistently, in the postmortem brain of schizophrenic patients, the expression of GluR6 mRNA is reduced in the dentate gyrus and CA3 region [117]. Taken together, these lines of evidence suggest that dysfunction of the mossy fiber synapse associated with abnormalities in the granule cell maturation is involved in the pathophysiology of schizophrenia and/or other related psychiatric disorders. The abnormalities in the dentate gyrus and mossy fiber may be part of the neuropathology of schizophrenia. However, the dentate gyrus may be particularly susceptible to developmental problems due to the continuous postnatal neurogenesis, and the developmental defects of the dentate gyrus could be functionally expressed in an exaggerated form as changes in the large synaptic facilitation at the mossy fiber synapse, thereby substantially contributing to the pathophysiology of schizophrenia.

Most of the animal studies described above have been carried out using rodents. In addition to difficulties in modeling psychiatric disorders using experimental animals in general, marked differences in the brain anatomy, such as the relative hippocampal size, between rodents and primates would pose a problem in extrapolating results of rodent studies to humans. There are some important differences in the anatomy of the dentate gyrus between rodents and primates. A subpopulation of the mature granule cells has basal dendrites in primates, but not in rodents [155]. The distribution of the mossy fibers is significantly different among species [156]. It is not known whether these anatomical differences are accompanied by functional differences. Mossy fiber LTP in cynomolgus monkeys has been shown to be independent of NMDA receptors as in rodents [157]. Some other studies also report similarities in synaptic and somatic membrane properties of the granule cells between rodents and nonhuman primates [158, 159]. Further cellular physiological studies in nonhuman primates would be important to clarify the involvement of the dentate gyrus and mossy fibers in psychiatric disorders.

Effects of Antipsychotic Drugs on Dentate Gyrus and Mossy Fiber Synapse

In contrast to a lot of literature regarding antidepressant effects, effects of antipsychotic drugs on the dentate gyrus in experimental animals have not been addressed in detail. Typical and atypical antipsychotic drugs cause a decrease and increase in BDNF mRNA levels, respectively [160]. Typical antipsychotic drugs increase levels of the presynaptic protein SNAP-25 in hippocampal synaptic regions with the strongest effect at the mossy fiber pathway [161]. Some antipsychotic drugs can increase cell proliferation in the dentate gyrus [reviewed by 162]. Atypical antipsychotic drugs have been shown to facilitate the adult neurogenesis in the dentate gyrus in some studies [163, 164] but have no significant effects in others [165–168]. The discrepancies in reported effects might be due to methodological differences. Haloperidol, a typical antipsychotic drug, has no effects on the neurogenesis in rats [69, 163, 165, 167–169] but facilitates the neurogenesis in gerbil [170]. Thus, the adult neurogenesis in the dentate gyrus might be involved in some actions of antipsychotic drugs. Since the adult neurogenesis is modulated by atypical, but not by typical, antipsychotic drugs in rats, it may be associated with negative symptoms of schizophrenia, on which atypical antipsychotic drugs generally have superior therapeutic effects as compared with typical antipsychotics [171]. Ameliorating effects of antipsychotic drugs on positive or psychotic symptoms have been ascribed to their antagonistic actions on dopamine D2 or D3 receptors [172]. Activation of D2-like receptors by exogenous agonists increases the adult neurogenesis in the hippocampus [173, 174], which may be mediated by ciliary neurotrophic factor released from astrocytes [174]. The lack of effects of the highly potent D2 antagonist haloperidol on the neurogenesis suggests that endogenous dopamine does not activate this D2-dependent pathway for the regulation of the adult neurogenesis in a normal condition. Since over-activation of astrocytes has been suggested in schizophrenia [175], this pathway might be endogenously activated in the pathological condition. In the postmortem brain of patients with schizophrenia, the neural progenitor cell proliferation in the dentate gyrus has been shown to be reduced [75]. However, most subjects in this study received antipsychotic drugs, which may complicate interpretation of the results.

Although effects of currently utilized antipsychotic drugs on the mossy fiber synapse have not been well investigated, this synapse can be a potential target of recently developed candidate antipsychotic drugs. A metabotropic glutamate 2/3 receptor (mGluR2/3) agonist has been shown to have antipsychotic actions both in animal studies [176] and in a clinical trial [177]. Activation of mGluR2/3 strongly suppresses the mossy fiber synaptic transmission via presynaptic mechanisms [46, 178], but other hippocampal synapses are either insensitive or less sensitive to mGluR2/3 agonists [178, 179]. Therefore, mGluR2/3 agonists at optimal doses would preferentially inhibit signaling via the dentate gyrus-mossy fiber pathway in the hippocampal circuit. The synaptic facilitation increases along with the synaptic inhibition caused by mGluR2/3 activation [178]. Therefore, mGluR2/3 agonists may be effective in treating the reduced mossy fiber synaptic facilitation associated with abnormalities in the dentate gyrus maturation.

A partial agonist of α7 neuronal nicotinic acetylcholine receptors has also been shown to ameliorate some symptoms of schizophrenia [180, 181]. The α7 nicotinic receptor gene, CHRNA7, is one of the candidate genes for schizophrenia [182]. Alpha7 nicotinic receptors are highly expressed in the hippocampus, especially in the hilar region [183]. In the dentate gyrus of patients with schizophrenia, binding of α-bungarotoxin, a blocker of α7 neuronal nicotinic receptors, is reduced [121]. Mice lacking α7 nicotinic receptors exhibit impairment in attention [184] and working or episodic-like memory [185]. Application of nicotine induces a rise in calcium concentrations in the mossy fiber terminals and increases action potential-independent release of neurotransmitter probably via activation of α7 nicotinic receptors [186–188; but see 189]. It is possible that modulation of mossy fiber synaptic transmission by α7 agonists can contribute to improvement of cognitive functions.

Conclusions: Mossy Fiber Synapse as a Target for Treatments of Psychiatric Disorders

Since the mossy fiber synapse plays critical roles in regulating excitability and plasticity in the CA3 circuit, dysfunction of this synapse would substantially affect functions of the hippocampus. Various neurotransmitter receptors at the mossy fiber synapse can be potential targets for pharmacological treatments of psychiatric disorders. Abnormal development or maturation of the dentate granule cells likely contributes to the pathophysiology of schizophrenia and other related psychiatric disorders. Dysfunction of the mossy fiber synaptic transmission could be secondary to the abnormalities in the granule cells maturation. In this case, drugs that directly act on the mossy fiber synapse may ameliorate or control a part of symptoms but may not afford a complete cure of the disorder. Instead, drugs that are expected to affect neuronal maturation or differentiation such as retinoic acid analogs might be effective in treating such defects [190]. It would also be important to reevaluate currently utilized antipsychotic drugs with respect to possible efficacy in alleviating the dysfunction associated with maturational or developmental defects of the dentate gyrus, thereby indirectly modulating the mossy fiber synaptic transmission. Genetically manipulated mice such as alpha-CaMKII +/− mice as well as conventional schizophrenia model animals would be valuable tools for re-evaluation of current psychotropic drugs and screening of candidate drugs.