|Home | About | Journals | Submit | Contact Us | Français|
The dopamine system is under multiple forms of regulation, and in turn provides effective modulation of system responses. Dopamine neurons are known to exist in several states of activity. The population activity, or the proportion of dopamine neurons firing spontaneously, is controlled by the ventral subiculum of the hippocampus. In contrast, burst firing, which is proposed to be the behaviorally salient output of the dopamine system, is driven by the brainstem pedunculopontine tegmentum. When an animal is exposed to a behaviorally salient stimulus, the pedunculopontine tegmentum elicits a burst of action potentials in the dopamine neurons. However, this bursting only occurs in the portion of the dopamine neuron population that is firing spontaneously. This proportion is regulated by the ventral subiculum. Therefore, the ventral subiculum provides the gain, or the amplification factor, for the behaviorally salient stimulus. The ventral subiculum itself is proposed to carry information related to the environmental context. Thus, the ventral subiculum will adjust the responsivity of the dopamine system based on the needs of the organism and the characteristics of the environment. However, this finely tuned system can be disrupted in disease states. In schizophrenia, a disruption of interneuronal regulation of the ventral subiculum is proposed to lead to an overdrive of the dopamine system, rendering the system in a constant hypervigilant state. Moreover, amphetamine sensitization and stressors also appear to cause an abnormal dopaminergic drive. Such an interaction could underlie the risk factors of drug abuse and stress in the precipitation of a psychotic event. On the other hand, this could point to the ventral subiculum as an effective site of therapeutic intervention in the treatment or even the prevention of schizophrenia.
Schizophrenia is a devastating disorder that arises during late adolescence/early adulthood. This disorder is characterized by a spectrum of disruptions, including disruption of perceptions and hallucinations, thought disorder, and cognitive disturbances. Schizophrenia has been shown to be genetically linked, in that the propensity for an individual to develop schizophrenia is related to the proportion of shared genetic material (Gottesman and Shields, 1971, 1976). Nonetheless, it is clear that this disorder is not completely genetically determined. Thus, investigators have found a number of risk factors that predispose an individual to schizophrenia, or alternately can lead to exacerbation of psychosis or relapse in those individuals in which the schizophrenia is in remission, including stress, perinatal factors, environmental factors, and drug abuse (Jones et al., 1994; Hultman et al., 1999; McDonald and Murray, 2000). However, how such risk factors can interact with a genetic predisposition to facilitate transition to psychosis, or what common neurobiological substrates underlie this comorbidity, is unknown. Using studies in a rat model of schizophrenia, we found that risk factors such as stress and psychostimulant abuse affect the mesolimbic dopamine (DA) system in a similar fashion. Moreover, preliminary data suggest that interruption of the cycle of system disruption set in motion as a consequence of uncontrolled stress early in life may serve to circumvent the transition to psychosis (Fox and Grace, 2009).
Substantial evidence exists implicating the dopamine system in schizophrenia. Thus, drugs that increase DA release will exacerbate psychosis (Angrist et al., 1975) and may cause a schizophrenia-like psychosis in normal individuals (Angrist et al., 1974; Friedman and Sienkiewicz, 1991; Giladi et al., 2000). Moreover, all antipsychotic drugs in current use attenuate DA transmission (Carlsson and Lindqvist, 1963; Creese et al., 1976; Seeman et al., 1976; Kapur and Remington, 2001). Finally, studies using in vivo imaging in humans demonstrate that amphetamine-induced raclopride displacement (which is an index of phasic DA release;(Grace, 1991; Laruelle, 1998; Wong et al., 2006) is significantly higher in the brains of schizophrenia patients, and moreover the magnitude of the increase above control levels is proportional to the ability of the amphetamine to exacerbate the psychosis of the individual tested (Laruelle and Abi-Dargham, 1999; Laruelle et al., 1999). Thus, there is substantial correlative evidence that the DA system is involved at least in the psychotic features of schizophrenia. Nonetheless, there is little direct evidence that it is the DA system that is abnormal in schizophrenia (Post et al., 1975; van Kammen et al., 1986; Beuger et al., 1996). Instead, evidence has pointed to a dysfunction in the regulation of the DA system – i.e., that the DA system is in a hyper-responsive state due to alterations in its afferent regulation.
We have performed extensive studies into the mechanisms that control DA neuron activity states. First, it is known that the mesolimbic DA neurons can exist in several activity states. First, the neurons can either be spontaneously firing or in a quiescent, non-firing “in reserve” state (Bunney and Grace, 1978; Grace and Bunney, 1984). The proportion of neurons firing spontaneously has been termed the “population activity” of the DA neurons (West and Grace, 2000; Floresco et al., 2001b; Moore et al., 2001a). Once the DA neurons are firing spontaneously, they can demonstrate different patterns of activity. In the baseline, unstimulated state, DA neurons fire in a slow, irregular patter, of about 3–5 Hz. However, if the DA system is activated, either by antagonist administration (Bunney and Grace, 1978; Grace et al., 1997), glutamate activation (Grace and Bunney, 1984; Lodge and Grace, 2006b), or afferent stimulation (Smith and Grace, 1992; Floresco et al., 2003b; Lodge and Grace, 2006a, b), the DA neurons will begin to fire in bursts of action potentials. These bursts are also emitted by DA neurons in awake behaving animals in response to a behaviorally activating stimulus, such as reward (Schultz, 1998; Anstrom and Woodward, 2005; Zweifel et al., 2009). We have found that a brainstem nucleus known as the pedunculopontine tegmentum (PPTg) is the most potent activator of mesolimbic DA neuron burst firing in the rodent brain (Floresco et al., 2003a; Lodge and Grace, 2006b), with an associated region, the laterodorsal tegmentum, gating the ability of the PPTg to drive bursting (Lodge and Grace, 2006a) (see Fig. 1). Burst firing in DA neurons occurs via stimulation of N-methyl-D-aspartate (NMDA) receptors on the DA neuron somatodendritic tree (Chergui et al., 1993; Zweifel et al., 2009). It is known that a neuron must be in a depolarized state for the NMDA receptor to be capable of activation; otherwise, there is a magnesium blockade of the NMDA channel (Mayer et al., 1984). As a result, only DA neurons that are spontaneously firing can be driven to burst fire (Lodge and Grace, 2006c).
Our studies show that these two activity patterns; i.e., population activity and burst firing, are driven by distinct processes. With respect to population activity, in vivo intracellular recording studies have revealed that DA neurons are constantly bombarded with high-amplitude GABAergic inhibitory postsynaptic potentials (IPSPs) arriving at a rapid pace (Grace and Bunney, 1985). These IPSPs presumably arise from GABAergic afferents derived from the ventral pallidum, a region containing neurons that are GABAergic, fire at rapid rates (Tsai et al., 1989), and project to the midbrain DA neuron population. We have posited that the ventral pallidum, via high discharge rates and direct GABAergic projections to midbrain DA neurons, holds subsets of these neurons in a hyperpolarized, non-firing state (Floresco et al., 2001a). The primary source of input to the ventral pallidum arises from the nucleus accumbens which, in turn, is potently driven by glutamatergic afferents from the ventral subiculum of the hippocampus. We have found that the most potent control of DA neuron population activity arises from this ventral subicular input (Floresco et al., 2001a; Floresco et al., 2003a). Thus, when the ventral subiculum is activated, it provides a powerful glutamatergic drive over the nucleus accumbens, which in turn inhibits the ventral pallidum and releases the DA neuron population activity from inhibitory control (Fig. 1).
What is the purpose of modulating DA neuron population activity? We have shown that the population activity of the DA system will have a minor influence on DA transmission in the striatum; i.e., increases in the number of neurons firing appears to cause a linear increase in the low level, tonic DA levels present in the striatum (Floresco et al., 2003a). However, controlling the population activity is likely to have a much more important consequence when it comes to regulating the responsivity of the DA system. Thus, as outlined above, burst firing is believed to be the behaviorally salient output of the DA system. Furthermore, burst firing is driven directly by glutamatergic afferents arising from the PPTg; a region that is known to respond to conditioned responses and behaviorally salient cues (Pan and Hyland, 2005).
Thus, a salient stimulus will activate burst firing in spontaneously firing DA neurons via activation of the PPTg. However, the number of DA neurons that are driven to burst fire will depend on the population activity. Therefore, the more DA neurons firing, the larger the amplitude of the phasic response (Lodge and Grace, 2006b). As a result, the ventral subiculum, by controlling the number of DA neurons firing, will set the amplification factor, or the gain, of the phasic burst firing response driven by the PPTg.
What is the role of the ventral subiculum, beyond its influence over DA neuron activity? Studies have shown that the ventral subiculum (which is the rodent homolog of the anterior hippocampus in humans) receives afferent input from a number of structures, including the dorsal hippocampus CA1 region (Amaral et al., 1991) as well as areas related to limbic system information processing (French et al., 2003; Herman and Mueller, 2006). It is well-known that dorsal regions of the hippocampus are involved in place learning, and contain place cells believed to code one’s location in space (O'Keefe, 1976). However, as one moves ventrally through the rodent hippocampus, the place information appears to be more integrated with limbic input. I propose that this represents a layering of affective input on top of the location, which would transform information from “where am I in space” to “what is the emotional significance of my location?” The emotional salience of a location would, by definition, be its behaviorally salient context. This is highly consistent with studies showing that the ventral subiculum is involved in context-dependent fear conditioning (Fanselow, 2000) as well as other context-related processes (Jarrard, 1995; Maren, 1999; Sharp, 1999). This would also be consistent with its role in determining DA neuron population activity; i.e., the amplitude of the stimulus-driven phasic response of the DA system should be a function of the behavioral context of the event. Thus, in conditions in which the context is safe and innocuous, stimuli should cause little activation of the DA system and promote minor behavioral modifications. However, in conditions in which the organism is in an environment in which rapid and massive responses are required to obtain reward (e.g., identifying prey) or reacting to threat, the context should maintain the DA system at a high state of responsivity (Fig 2).
As outlined above, the gain of the DA system is regulated by drive from the ventral subiculum in a context-dependent manner. In other words, the system is modulated to be highly vigilant and reactive to stimuli under conditions of high threat/high reward, but to return to a baseline state when in a more safe context. Such modulation would be necessary in order to maintain proper perspective over situations, and to modulate responses appropriately. It would not be behaviorally advantageous to be in a constant low-active state, particularly when a rapid reaction to a threat is required. However, on the other hand, being in a maintained high-vigilance state at all times would be disadvantageous for other reasons – one would be constantly over-reacting to non-threatening or benign/non-salient stimuli, with no ability to screen out events. In other words, the normal filter that tells the brain what is important and screens out everything else is turned off, as has been described by schizophrenia patients (Saks, 2007). Such a condition appears to be present in the brain of schizophrenia patients, as well as in animal models of schizophrenia. Thus, as reviewed above, in the schizophrenia patient the DA system is in a hyper-responsive state, with abnormal levels of DA release occurring to stimuli when contrasted with control conditions. Moreover, there is increasing evidence that the anterior hippocampus (the human homolog of the ventral subiculum) is overactive in schizophrenia patients (Malaspina et al., 1999; Kegeles et al., 2000; Heckers, 2001; Medoff et al., 2001; Meyer-Lindenberg et al., 2005).
A similar condition appears to exist in animal models of schizophrenia. One model which was developed in my laboratory (Moore and Grace, 1997; Grace and Moore, 1998) and has been extensively characterized (Moore et al., 2001b; Flagstad et al., 2004; Gourevitch et al., 2004; Le Pen et al., 2006; Moore et al., 2006) is based on administration of the mitotoxin methylazoxymethanol acetate (MAM). We have found that administration of MAM to pregnant rats during gestational day 17 creates in the adult offspring a set of characteristics consistent with what one would predict for an animal model of schizophrenia, including limbic cortical thinning with increased cell packing density, disruption of prepulse inhibition of startle and latent inhibition, deficits in social interactions, reversal learning and hyper-responsivity to PCP and amphetamine, all characteristics shared with the MAM rat model and human schizophrenia patients (Talamini et al., 1999; Flagstad et al., 2004; Gourevitch et al., 2004; Le Pen et al., 2006; Moore et al., 2006; Lodge et al., 2009). In these animals, we found that there is a substantial increase in the population activity of the DA neurons, being nearly 2.5× higher than that observed in controls. Moreover, this occurs in concert with hyperactivity of ventral subicular neurons (Lodge and Grace, 2007). In these animals, both the increased DA neuron population activity and the hyper-responsivity to amphetamine can be restored to baseline upon inactivation of the ventral subiculum (Lodge and Grace, 2007). The source of this hyper-activity in the ventral subiculum appears to correlate with a loss of parvalbumin GABAergic interneuron staining observed in the ventral subiculum and in the prefrontal cortex of MAM rats (Lodge et al., 2009) as well as in schizophrenia patients (Lewis et al., 2001; Zhang and Reynolds, 2002). Furthermore, the diminished parvalbumin staining corresponds to a disruption of gamma band rhythmic activity in these regions in the MAM-treated rat (Lodge et al., 2009); a characteristic also found in schizophrenia patients (Gonzalez-Hernandez et al., 2003; Gallinat et al., 2004; Cho et al., 2005; Basar-Eroglu et al., 2007). Thus, these studies suggest that, in both animal models of schizophrenia and in the human schizophrenia patient, a loss of parvalbumin-GABAergic regulation of hippocampal activity leads to hyper-responsivity of the DA system and loss of control over appropriate responses to stimuli.
Schizophrenia is a genetically linked disorder. However, there are a number of environmental factors that can increase the probability that a particular person will convert to psychosis. Studies have shown that substance abuse can play a significant role in susceptibility to schizophrenia. Thus, studies by Murray and collaborators (Arseneault et al., 2002) have shown that abuse of cannabis, particularly during adolescence, can render an individual more susceptible to schizophrenia, particularly when this is combined with other risk factors such as a polymorphism at the site for catechol-O-methyl-transferase enzyme responsible for metabolism of DA in the frontal cortex (Caspi et al., 2005). Cannabinoids themselves are known to be endogenous transmitters that modulate transmission, particularly when it involves GABA and glutamate synapses (Schlicker and Kathmann, 2001). GABA interneurons receive potent afferent drive from long-loop projections, such as those arising from the amygdala. Studies have shown that cannabinoids will potently modulate emotional learning mediated by amygdala afferents to the prefrontal cortex (Laviolette and Grace, 2006a, b). Thus, cannabinoid agonists will potentiate emotional learning to the point where it will enable learning and conditioning to stimuli that otherwise would be below threshold for mediating a learned response (Laviolette and Grace, 2006b). Interestingly, although GABAergic neurons that contain parvalbumin do not express cannabinoid receptors (Katona et al., 1999; Marsicano and Lutz, 1999; Hajos et al., 2000), these neurons are potently modulated by cannabinoid-receptor containing cholecystokinin (CCK)-expressing (Marsicano and Lutz, 1999; Tsou et al., 1999) GABAergic interneurons (Karson et al., 2009). Indeed, studies show that cannabinoid agonists can disrupt the ability of parvalbumin interneuron-driven gamma rhythmicity (Hajos et al., 2000; Robbe et al., 2006). Therefore, cannabinoids are positioned to interfere with the same neuronal subtype that was shown above to be disrupted in schizophrenia frontal cortical regions.
In addition to cannabinoids, other factors are known to play a prominent role in susceptibility to schizophrenia. Thus, other drugs of abuse, such as psychostimulants, are known to exacerbate the symptoms of schizophrenia (Angrist et al., 1975), and abuse of these drugs can render an individual more susceptible to transition to psychosis (Chen et al., 2003). Another factor that has been associated with schizophrenia susceptibility is stress. Stress is known to lead to relapse in schizophrenia patients that are in remission (Benes, 1997), and early adolescent stress is known to be a risk factor in the development of schizophrenia (Benes, 1997; Tsuang, 2000). Indeed, studies by Johnstone and colleagues have shown that in children at genetic risk for developing schizophrenia, those that show abnormally high responses to stressors tend to be those that convert to psychosis (Johnstone et al., 2002). Thus, there is an interesting relationship between stress and schizophrenia susceptibility.
Stress is a complex variable that is mediated via a number of brain circuits. Interestingly, the ventral subiculum is posited as a primary central integrator of the stress response (O'Mara, 2005; Herman and Mueller, 2006). Moreover, the prefrontal cortex has been shown to be a potent regulator of the stress response, in part via attenuation of responses in the amygdala (Rosenkranz and Grace, 2002a, b; Rosenkranz et al., 2003), which is the brain region in which fear and anxiety are expressed (LeDoux, 2000). We have posited that a potential etiological factor in schizophrenia may relate to an ineffective prefrontal cortical suppression of the stress response (Grace, 2004; Thompson et al., 2004). Thus, an initial deficit in prefrontal cortical function could interfere with the normal suppression of stress responses, leading to the heightened activation seen in those that transition to schizophrenia. A brain region that has been shown to be particularly susceptible to the deleterious effects of maintained stress is the hippocampus (Sapolsky et al., 1990; Sapolsky, 2000; Lee et al., 2002). Thus, it may be that deficits in prefrontal cortical modulation of the stress response could lead to a condition of overdrive of stress circuits, resulting in hippocampal interneuron damage and hyperactivity leading to psychosis (Grace, 2004; Thompson et al., 2004). If such a condition were indeed present, one would posit that limiting the stress response during the periadolescent period, when the at-risk individuals are showing heightened stress responses, may indeed circumvent this transition and prevent the onset of schizophrenia. In preliminary experiments, we have found that if the anti-anxiety drug diazepam is administered in the peri-pubertal period of MAM-treated rats, we prevent the increase in DA neuron population activity observed in the adult that we posit is related to psychosis (Fox and Grace, 2009). It may be that such an intervention in humans, in which the abnormal stress response in children at risk is treated, may actually prevent schizophrenia from emerging later in life.
Stress and drug abuse have been shown to be risk factors in schizophrenia; moreover, studies show that stress is a risk factor in the onset or relapse to drug-taking behavior (O'Doherty, 1991; Koob and Le Moal, 2001). This can occur via long-term stress-induced hippocampal damage, as mentioned above. However, there is also evidence that both stress and psychostimulant abuse exert common actions on the DA system.
In the animal model of schizophrenia, we found that the increased behavioral response to amphetamine was correlated with an increase in the DA neuron population activity, or the number of DA neurons firing spontaneously (Lodge and Grace, 2008). There is another condition in which animals exhibit abnormally heightened responses to psychostimulants, and this is behavioral sensitization. Thus, when a rat is administered amphetamine or cocaine for a period of 5 days and then is withdrawn from the amphetamine for an equivalent period, the rat will show an abnormally large, or sensitized, response when tested with a single dose of amphetamine (Post, 1980; Robinson and Berridge, 2000; Lodge and Grace, 2008). We found that animals treated in this manner exhibited a substantial increase in DA neuron population activity (Lodge and Grace, 2008), in a similar manner (but less magnitude) than was observed in the schizophrenia model (Lodge and Grace, 2007; Fig. 3).
What is the source of the increased DA neuron population activity that is present following sensitization? An observation that may yield a clue to this is the finding that the sensitization is strongest when the behavior is tested in the same cage in which the sensitization occurred (Vezina et al., 1989; Badiani et al., 1995; Crombag et al., 2000). Thus, sensitization is a context-dependent phenomenon, and context is mediated via the ventral subiculum. Inactivation of the ventral subiculum in the amphetamine-sensitized rats was found to reverse the increase in DA neuron population activity, bringing the number of DA neurons firing spontaneously back to control levels. Moreover, following ventral subicular inactivation, the sensitized behavioral response to amphetamine was reversed (Lodge and Grace, 2008). Note that the rats still responded with behavioral activation to amphetamine administration, but the magnitude of the response was now identical to that observed in controls (Fig. 3).
Stress is also a phenomenon that demonstrates a number of characteristics that are similar in nature to what is observed in these other models. Thus, stress has been shown to cross-sensitize with amphetamine, in that an animal exposed to a stressor will show a heightened behavioral response to subsequent administration of amphetamine (Antelman et al., 1980; Pacchioni et al., 2002). Moreover, stress is a context-dependent phenomenon, in that a rat will exhibit anxiety-like behaviors when returned to an environmental context in which it had been previously stressed (Fanselow, 2000). The ventral subiculum, as mentioned above, is a region that is proposed to mediate central effects of stress (O'Mara, 2005; Herman and Mueller, 2006), and has afferent input from numerous stress-related structures, such as the amygdala and the noradrenergic locus coeruleus (Oleskevich et al., 1989; Schroeter et al., 2000; French et al., 2003), each of which we found will activate ventral subicular firing (Lipski and Grace, 2008). Stress itself has been shown to increase levels of dopamine in postsynaptic targets (Abercrombie et al., 1989; Finlay et al., 1995; Castro and Zigmond, 2001). What is the state of the DA system following an acute stressor? To examine this, we employed a 2-hour restraint protocol, which has been reported to lead to an increased behavioral response to amphetamine (Pacchioni et al., 2002). We found that, following 2 hours of restraint, there was a significant increase in DA neuron population activity. Moreover, as shown in the other paradigms, inactivation of the ventral subiculum reversed the increase in DA neuron population activity and normalized the behavioral response to amphetamine (Valenti and Grace, 2008).
The DA system plays a critical role in regulating responses to environmental stimuli. In particular, the DA system adjusts depending on the needs of the organism and the demands of the environment to modulate how an organism responds to stimuli. Thus, the responsivity of the DA system itself must be adjusted in order to enable a contextually-appropriate response to occur. Behaviorally salient stimuli can evoke different responses in different environmental contexts. Provided that the system is functioning in a normal fashion, the organism can efficiently interact with the environment and other organisms, yet be ready to alter their response pattern as the demands change. Being in a low responsive state when confronted with threat, or being in a highly vigilant state even in a very benign setting will lead to inappropriate and perhaps deleterious responses. For this reason, a complex system is in place to adjust the amplitude of the DA response based on a number of mood- and location-related stimuli. However, if this system is disrupted, such as is proposed to occur with conditions of ventral subiculum hyperactivity, then the responses will not match the environmental contingencies, leading the organism to over-attribute salience to benign stimuli, to be overwhelmed with attention-demanding stimuli without the ability to prioritize their responses, or to behave in a manner not consistent with effectively achieving their goals. The ability to pharmacologically restore normal activity levels within a damaged hippocampal circuit could provide an effective means of alleviating this deficit state at the source of disruption.