DOI 10.1515/revneuro-2014-0004 Rev. Neurosci. 2014; 25(4): 605–619
Shengyuan Ding and Fu-Ming Zhou*
Serotonin regulation of subthalamic neurons Abstract: The subthalamic nucleus (STN) is a key component of the basal ganglia. As the only basal ganglia nucleus comprised of mostly glutamatergic neurons, STN neurons provide a key driving force to their target neurons. Thus, regulation of STN neuron activity is important. One STN regulator is the serotonin (5-HT) system. The STN receives a dense 5-HT innervation. 5-HT1A, 5-HT1B, 5-HT2C, and 5-HT4 receptors are expressed in the STN. 5-HT may regulate the STN via several mechanisms. First, 5-HT may affect STN neuron excitability directly by either inhibiting a subpopulation of STN neurons via activation of 5-HT1A receptors or exciting STN neurons through activation of 5-HT2C and 5-HT4 receptors. Second, 5-HT may affect synaptic inputs to the STN. Via activation of 5-HT1B receptors on the afferent terminals, 5-HT inhibits glutamatergic input to the STN, but the inhibitory effect on GABAergic input is smaller. Third, 5-HT may regulate the STN glutamatergic output by activating presynaptic 5-HT1B receptors, thus reducing burst firing in target neurons. Last, 5-HT may affect glutamate release at the intra-STN axon collaterals and regulate the recurrent excitation. These mechanisms may work in concert to fine-tune the intensity and pattern of STN activity and reduce STN output bursts. Keywords: 5-HT (5-hydroxytryptamine); basal ganglia; burst firing; Parkinson’s disease; presynaptic inhibition; recurrent excitation; serotonin receptor; substantia nigra; subthalamic nucleus. *Corresponding author: Fu-Ming Zhou, Department of Pharmacology, University of Tennessee College of Medicine, Memphis, TN 38163, USA, e-mail: [email protected] Shengyuan Ding: Department of Pharmacology, University of Tennessee College of Medicine, Memphis, TN 38163, USA
Introduction The subthalamic nucleus (STN) is a key component of the basal ganglia (BG). As the only BG nucleus comprised primarily of glutamatergic neurons (Smith and Parent, 1988; Lévesque and Parent, 2005; Gerfen and Bolam, 2010), the STN critically affects the activity of its target neurons, exerting its normal physiological function and also its dysfunction under pathophysiological
conditions, as demonstrated by the profound therapeutic effects of STN high-frequency stimulation in Parkinson’s disease (Perlmutter and Mink, 2006). Although intrinsically active, STN neurons are also under the regulation of several neurotransmitter systems. In this review, we discuss the serotonergic regulatory effects on STN neuron intrinsic activity and on the synaptic inputs to and synaptic output from STN neurons. Readers are also referred to several recent reviews that cover other aspects of the STN (Bevan et al., 2007; Charpier et al., 2010; Baunez et al., 2011; Wilson and Bevan, 2011).
Connectivity of the subthalamic nucleus As illustrated in Figure 1A and B, the STN is a critical component of the BG, a group of interconnected subcortical nuclei that also include the striatum (caudate-putamen), external and internal segments of the globus pallidus (GPe and GPi), substantia nigra pars compacta (SNc), and pars reticulata (SNr) (Nambu et al., 2002; Gerfen and Bolam, 2010; Haynes and Haber, 2013). The striatum and STN are the input structures, whereas GPi and SNr are output structures. Although most cells in the striatum, GPe, GPi, and SNr, are GABAergic with cells in SNc being largely dopaminergic, most, if not all, STN neurons use glutamate as their neurotransmitter, thus playing a unique role in the information processing and transfer in the BG (Parent and Hazrati, 1995; Hikosaka et al., 2000; Sano et al., 2013). The transmission of cortical information through the BG is mediated by three main pathways: the direct pathway, indirect pathway, and hyperdirect pathway (Nambu et al., 2002; Gerfen and Bolam, 2010; Haynes and Haber, 2013). In the direct pathway, glutamatergic corticostriatal information flows to the dopamine D1 receptor-expressing, GABAergic, often silent medium spiny neurons (D1-MSNs) in the striatum that in turn project to and inhibit the tonically firing GABAergic projection neurons in GPi and SNr, releasing the thalamocortical circuit (Figure 1A) (Hikosaka et al., 2000; Kravitz et al., 2010). In the indirect pathway, glutamatergic corticostriatal information is first transmitted to the dopamine D2 receptor-expressing, GABAergic, often silent medium spiny neurons (D2-MSNs) in the striatum that, in turn, project to
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
606 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons Anatomical studies using conventional labeling methods indicate that the glutamatergic STN neurons project not only to the neighboring SNr and SNc, GPe and GPi but also to the more distant targets, such as the striatum and even the cerebral cortex. They project in the rostral direction, and the pedunculopontine nucleus, the mesencephalic and pontine reticular formation projects in the caudal direction (Kita and Kitai, 1987; Parent and Hazrati, 1995; Sato et al., 2000; Degos et al., 2008). A recent study using a viral vector seeking enhanced visualization of the axonal arborization and the dendritic tree of single STN neurons in the rat brain has confirmed the findings from traditional tracing studies (Koshimizu et al., 2013). These projections provide the anatomical basis for glutamatergic STN neurons to affect their target neurons. Additionally, intra-STN axonal collaterals have been detected by intracellular staining, potentially enabling excitatory interaction among STN neurons, although paired recording is needed to prove this potential interaction (Kita et al., 1983; Ammari et al., 2010).
Figure 1 Location and connection of the subthalamic nucleus (STN). (A) A Nissl-stained sagittal adult mouse brain section showing the location of the STN and other major brain areas connected to the STN. Arrows indicate information flow to and from the STN. The STN receives a major glutamatergic cortical innervation (the hyperdirect pathway) and also a minor glutamatergic input from the thalamus. The STN sends a major projection to and receives a major GABAergic input from the globus pallidus external segment (GPe). The STN also intensely innervates the substantia nigra pars reticulata (SNr), forming the subthalamonigral projection. SC, superior colliculus. (B) A simplified diagram shows the main STN inputs and outputs. PPN, pedunculopontine nucleus. Glu, glutamate. GABA, γ-amino butyric acid. In both panels (A) and (B), the red lines indicate glutamatergic projections and the black lines indicate GABAergic projections. For simplicity, the substantia nigra pars compacta and its projection to the STN are neither labeled in (A) nor diagramed in (B).
and inhibit the spontaneously firing GABAergic projections in GPe. Consequently, the tonic or default GPe inhibition of STN is paused, leading to increased STN glutamatergic output to GPi and SNr and subsequent inhibition of the thalamocortical circuit (Hikosaka et al., 2000; Kravitz et al., 2010). In the hyperdirect pathway, glutamatergic cortical motor signal, and other information directly flow to and excite STN neurons, providing these glutamatergic neurons with additional excitation to drive their target neurons (Afsharpour, 1985; Nambu et al., 2002; Kita and Kita, 2012; Walker et al., 2012; Sano et al., 2013).
STN neurons are spontaneously active glutamatergic neurons Immunohistochemical and neurophysiological studies indicate that most, if not all, STN neurons use glutamate as their neurotransmitter (Smith and Parent, 1988; Lévesque and Parent, 2005; Wilson and Bevan, 2011), a unique feature in the BG that are dominated by GABAergic neurons. Studies performed in STN-containing brain slices in the presence of synaptic receptor blockers have established that STN neurons exhibit spontaneous rhythmic spiking activity around 10 Hz, generated by intrinsic ion channels independent of synaptic inputs (Nakanishi et al., 1987; Bevan and Wilson, 1999; Beurrier et al., 2000; Hallworth et al., 2003). The persistent voltage-gated sodium current is particularly important in triggering the spontaneous firing in STN neuron (Bevan and Wilson, 1999; Wilson and Bevan, 2011). In intact animals, STN neurons fire at higher rates with both regular and burst firing patterns (Bergman et al., 1994; Kreiss et al., 1997; Urbain et al., 2002). In normal monkeys, the average STN neuron firing rate is reported to be ∼25 Hz (Bergman et al., 1994; Wichmann et al., 1994; Isoda and Hikosaka, 2008). Similar data in normal humans are not available, but in Parkinson’s disease, the reported STN neuron firing rate is 40–60 Hz (Mrakic-Sposta et al., 2008; Steigerwald et al., 2008; Benedetti et al., 2009). The high-frequency spontaneous firing combined with the fact that most STN
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons 607
neurons are glutamatergic has led to the conclusion that STN neuron activity is a critical driving force in the basal ganglia (Kitai and Kita, 1987; Parent and Hazrati, 1995; Wilson and Bevan, 2011). Thus, regulation of STN neuron activity has important consequences. In this review, we focus on the 5-HT regulation of STN neuron activity.
5-HT innervation and 5-HT receptor expression in the STN Immunohistochemical studies using antibodies against 5-HT or 5-HT transporter (SERT) protein have shown that the STN receives a dense 5-HT innervation originated in the dorsal raphe nucleus (Figure 2A) (Steinbusch, 1981; Parent et al., 2010, 2011; Wallman et al., 2011). Quantitative ultrastructural studies have estimated that the SERT-positive 5-HT terminal density is 1.5 × 106/mm3 in the monkey STN and 1.2 × 106/mm3 in the rat STN (Parent et al., 2010, 2011), although the 5-HT axon terminal density in the SNr is even higher, being twice as that in the STN in both rodents and primates, including humans (Parent et al., 2010, 2011; Wallman et al., 2011). These studies further indicate that 5-HT axon terminals form both synaptic and nonsynaptic contacts with STN neurons (on dendrites and dendritic spines) in both rodents and nonhuman primates (Parent et al., 2010), similar to other brain areas (Van Bockstaele et al., 1994, 1996; Moukhles et al., 1997). Also, the majority of 5-HT axon terminal-formed synapses are asymmetric (Parent et al., 2010), although the functional implication of the morphological type of monoamine synapse is not clear, unlike the classical synapses with asymmetry being associated with excitatory synapses. These anatomical data indicate that 5-HT may influence STN neurons via the classical synaptic mode and also the volume transmission mode. 5-HT neurons often fire spontaneous action potentials at a tonic frequency of ∼1 Hz (Jacobs and Azmitia, 1992), triggering 5-HT release and providing the endogenous agonist 5-HT (Hashemi et al., 2011, 2012). The STN expresses several types of 5-HT receptors, specifically 5HT1A, 5-HT1B, 5-HT2C, and 5-HT4 receptors (Figure 2B), as detected by in situ hybridization, autoradiographic, and immunohistochemical studies (Maroteaux et al., 1992; Bruinvels et al., 1993; Boschert et al., 1994; Pompeiano et al., 1994; Wright et al., 1995; Compan et al., 1996; Eberle-Wang et al., 1997; Clemett et al., 2000; Vilaró et al., 2005). 5-HT1A and 5-HT1B receptors are coupled to Gi/o protein and are thus inhibitory, whereas 5-HT2C receptors are coupled to Gq/11 protein and 5-HT4 receptors are coupled to Gs protein; thus 5-HT2C and 5-HT4
Figure 2 5-HT innervation and 5-HT receptor expression in the STN. (A) A confocal image of a sagittal brain section showing the 5-HT innervation in the STN and neighboring brain areas. The 5-HT axons were detected by immunostaining the 5-HT transporter protein. Note that although the 5-HT innervation in STN is substantial, it is even more intense in the SNr. CP, cerebral peduncle; LHA, lateral hypothalamus (modified from Ding et al. with permission, 2013). (B) Diagram showing the established (supported by anatomical and/or physiological data), likely (some data) and potential (likely but no data yet, indicated by question marks) expression of 5-HT receptors in the STN. 5-HT1A, 5-HT1B, 5-HT2C and 5-HT4 receptors are abundant in the STN. 5-HT1A, 5-HT2C and 5-HT4 receptors are probably somatodendritic, whereas 5-HT1B receptors are expressed on the afferent axon terminals innervating the STN and also on axon terminals of the STN efferents. 5-HT1A receptor activation exerts an inhibitory effect in the STN, whereas 5-HT2C and 5-HT4 receptors activation excite STN. PPN, pedunculopontine nucleus.
receptors tend to be excitatory in neurons (Hannon and Hoyer, 2008; Millan et al., 2008). Comparison of the results from in situ hybridization studies and radioligand binding studies suggest that 5-HT2C, 5-HT4, and 5-HT1A receptors are most likely at STN neuron somatodendritic areas, whereas the 5-HT1B receptor appear be at the axon terminals of STN neurons and at the afferent axon terminals innervating STN neurons (Maroteaux et al., 1992; Bruinvels et al., 1993; Boschert et al., 1994; Pompeiano et al., 1994; Wright et al., 1995; Compan et al., 1996; EberleWang et al., 1997; Clemett et al., 2000; Vilaró et al., 2005). However, electron microscopic ultrastructural studies
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
608 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons are limited (Sari et al., 1999; Riad et al., 2000), and more studies are needed to confirm these potential subcellular localizations of 5-HT receptors in the STN and determine the subcellular expression intensity of these 5-HT receptors within STN. Despite a lack of detailed ultrastructural information, many laboratories around the world have studied how 5-HT affects STN neuronal activity and function. In the following sections, we seek to summarize and synthesize the results from these studies.
5-HT regulation of the intrinsic properties of STN neurons The regulation of 5-HT on the intrinsic properties of STN neurons has been studied by several groups using electrophysiological methods. Results from two in vitro electrophysiological studies indicate that 5-HT may excite the majority of STN neurons without affecting their firing patterns (Flores et al., 1995; Xiang et al., 2005). One group further suggested that 5-HT depolarized STN neurons through decreasing a potassium conductance, and this effect may be mediated by 5-HT2C and 5-HT4 receptors, since it was mimicked by a 5-HT2/5-HT4 receptor agonist and was partially blocked by a 5-HT4 receptor or a 5-HT2C receptor antagonist (Xiang et al., 2005). Two other groups reported a more complicated situation: 5-HT has both 5-HT2C- and 5-HT4-receptor-mediated excitatory effect and a 5-HT1A receptor-mediated inhibitory effect on STN neurons (Stanford et al., 2005; Shen et al., 2007). Shen et al. (2007) also indicated that multiple intrinsic membrane conductances were responsible for these different 5-HT effects. In a subset of STN neurons, 5-HT (30 μm) evoked a modest outward current (27 pA when clamped at -70 mV) through activation of 5-HT1A receptors, as the outward current was completely blocked by a 5-HT1A receptor antagonist. The outward current was mediated by a potassium conductance, because it was blocked by Ba2+, and its reversal potential was affected by changes in external K+ concentration. In a majority of STN neurons, 5-HT (30 μm) evoked a small inward current (17 pA at -70 mV) via activation of 5-HT2C and 5-HT4 receptors. The inward current was mediated by a reduction of an unknown potassium conductance and an increase in a mixed cation conductance. The reason is not known why only a subpopulation of STN neurons showed a 5-HT1 receptor-mediated outward current and another subpopulation of STN neurons showed a 5-HT2C/5-HT4 receptormediated inward current. Certainly, it is entirely possible that the expression of 5-HT receptors in the STN is not
homogenous. This possibility requires future combined electrophysiological and immunohistochemical studies in STN neurons. To further understand the regulatory roles in the STN, studies also have examined 5-HT effects on STN neuron activity in intact animals, and the conclusion is that the overall 5-HT effect is inhibitory (Liu et al., 2007; Aristieta et al., 2013). In one study, 5,7-dihydroxytryptamine lesion of the dorsal raphe nucleus led to an increased firing rate with more bursting firing in STN neurons (Liu et al., 2007). Based on their data that dopamine neuron lesion eliminated the effect of 5-HT neuron lesion, these authors concluded that the apparent 5-HT effects on STN neurons were mediated indirectly via nigral dopamine neurons. Although it is entirely possible for 5-HT neurons to affect dopamine neurons that, in turn, affect STN neurons, a pertinent question was, can 5-HT neurons also directly affect STN neurons in vivo giving the 5-HT innervation and 5-HT receptors in STN? This question has been answered by a recent commendable study that used several approaches to examine 5-HT effects on STN neuron activity in intact animals (Aristieta et al., 2013). This study found that 5-HT synthesis inhibition increased the STN neurons’ firing rate and bursting activity and these abnormalities were reversed by fluoxetine, a selective 5-HT reuptake inhibitor. Furthermore, they reported that in normal and 5-HT-depleted animals, direct local application of the 5-HT1A agonist 8-OH-DPAT robustly inhibited STN neuron firing in a dose-dependent manner that was prevented by pretreatment with the 5-HT1A antagonist WAY100635 (Aristieta et al., 2013). Additionally, these authors demonstrated that in normal and 5-HT-depleted animals, direct local application of the 5-HT2C agonist Ro 600175 substantially increased STN neuron firing in a dose-dependent manner that was prevented by pretreatment with the 5-HT2C antagonist SB 242084. These in vivo results from local STN drug application experiments are generally consistent with the data from in vitro studies as discussed, although the in vivo data suggest a net inhibitory 5-HT effect on STN neurons. This is understandable, because 5-HT also affects the neuronal networks that are connected to the STN such that 5-HT can influence STN neurons directly and indirectly. There is also a negative report that indicates that 5-HT depletion did not affect STN neuron firing significantly (Delaville et al., 2012a). The reasons for this discrepancy are not known. However, Delaville et al. (2012a) used a lower dose of a 5-HT synthesis inhibitor depleting brain tissue 5-HT by 73–82% while Aristieta et al. (2013) used a much higher dose of the same inhibitor depleting STN tissue 5-HT by more than 95%.
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons 609
This methodological difference may be a factor for the negative result of Delaville et al. (2012a). Taken together, available data from in vitro and in vivo studies indicate that the net serotonergic regulatory effect on the STN glutamatergic neurons may be inhibitory when the effects on intrinsic ion channels and the effects on synaptic inputs (discussed following) are combined. Further studies are needed to determine the relative influence of 5-HT1A receptor-mediated inhibition and 5-HT2C receptor/5-HT4 receptor-mediated excitation in STN neurons.
5-HT regulation of the synaptic inputs to STN neurons STN receives excitatory glutamatergic inputs from the cortex and the thalamus and inhibitory GABAergic inputs from the GPe (Afsharpour, 1985; Parent and Hazrati, 1995; Wilson and Bevan, 2011; Kita and Kita, 2012). These excitatory and inhibitory synaptic inputs have profound effects on the activities of STN neurons. In vivo recording studies in anesthetized rats indicate that the spontaneous firing of STN neurons is strongly but briefly increased by cortical glutamatergic inputs in the hyperdirect corticosubthalamic pathway evoked by electrical stimulation in sensorimotor cortical areas or the prefrontal cortex, whereas the spontaneous firing is substantially inhibited by GABAergic pallidosubthalamic inputs (Fujimoto and Kita, 1993; Maurice et al., 1998; Magill et al., 2004; Kita and Kita, 2011). These data from intact animals are further supported by in vitro studies that showed that GPe-originated GABAA inhibitory postsynaptic potentials (IPSPs) can inhibit the firing rate and change the intrinsically regular firing pattern to bursty firing patterns in STN neurons (Bevan et al., 2002a,b; Hallworth and Bevan, 2005). Studies indicate that the glutamatergic and GABAergic afferents to STN neurons may be modulated by the 5-HT system. 5-HT has been reported to inhibit the presumed corticosubthalamic glutamatergic excitatory postsynaptic currents (EPSCs) and pallidosubthalamic GABAergic inhibitory postsynaptic potentials (IPSCs) in STN neurons via activating presynaptic 5-HT1B receptors that may reduce the release of glutamate and GABA, respectively (Shen and Johnson, 2008). Moreover, in this study (Shen and Johnson, 2008), 5-HT appeared to be more potent in reducing EPSCs than IPSCs: 10 μm 5-HT reduced the EPSC amplitude by 35%, whereas 30 or 100 μm 5-HT was required to reduce the IPSC amplitude by 30% (10 μm
5-HT was ineffective). Although data on extracellular 5-HT concentrations in STN are not available, the endogenous extracellular 5-HT concentration may briefly reach 10 μm near 5-HT axon release sites but not likely to reach 30 or 100 μm. So presynaptic 5-HT1B receptors may moderately reduce the glutamatergic input to STN neurons, but the inhibitory effect on the GABAergic input is small and may not be physiologically relevant. The presynaptic 5-HT1B inhibition of synaptic inputs to STN neurons have functional importance. Because the 5-HT1B inhibition of the glutamatergic input to STN neurons appears to be larger than the 5-HT1B inhibition of the GABAergic input, when 5-HT neurons degenerate during PD, the loss of the 5-HT1B inhibitory effect on the glutamatergic input is presumably larger than that on the GABAergic input; consequently, the increase in the glutamatergic input to STN neurons also may be larger than the increase in the GABAergic input. These changes may contribute to the increased firing frequency and bursty firing pattern in STN neurons in 5-HT-depleted animals (Liu et al., 2007; Aristieta et al., 2013).
5-HT regulation of the synaptic output from STN neurons Now we turn to 5-HT regulation of STN neuron output from the efferent axon terminals in their target areas. As discussed previously, the glutamatergic STN neurons project to the GPe, GPi, and SNr (Kita and Kitai, 1987; Sato et al., 2000; Koshimizu et al., 2013). This glutamatergic STN projection is likely to be important in shaping the firing intensity and pattern of their target neurons in the GPe, GPi, and SNr (Kita and Kita, 2011). Furthermore, stimulation of the STN can evoke reverberating, long-lasting polysynaptic complex EPSCs in SNr GABA neurons (Figure 3A1, B1) and also in GPi and GPe GABA neurons (Shen and Johnson, 2006; Ammari et al., 2010; Ding et al., 2013). These intra-STN stimulation-evoked long-lasting polysynaptic complex EPSCs most likely are resulted from an intrinsic STN neuronal network for the following reasons. First, only electrical stimulation directly in the STN evokes complex EPSCs, indicating the activation of large numbers of STN neurons is required to evoke complex EPSCs. Second, upon severing the connection between the STN and pedunculopontine nucleus (PPN) and other nucleus but preserving only the connection between the STN and SNr, STN stimulation still can evoke complex EPSCs (Ammari et al., 2010; Ding et al., 2013). Third, anatomical studies have demonstrated the presence of STN neuron
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
610 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons
A
B
5-HT1B antagonist blocks 5-HT effect
A1
5-HT1B agonist mimics 5-HT effect
B1
10 µM 5-HT
10 µM CP93129
10 µM 5-HT+ 10 µM NAS-181 Control
40
Control
20
20
0
0
20
10 µM CP93129
10
0
40
0
Time course (min)
C
50 ms
B1
10 µM 5-HT Complex EPSC area (pA.s)
Complex EPSC area (pA.s)
50 pA
100 pA
10 µM NAS-181
A2
Wash
D
5-HT1A agonist has no effect
20 40 Time course (min)
5-HT1A antagonist has no effect
10 µM 5-HT 50 pA 100 pA
50 ms Control 10 µM 8-OH-DPAT Wash
Wash Control 10 µM WAY100135 +10 µM 5-HT
Figure 3 Exogenous 5-HT or 5-HT1B receptor agonist CP93129 inhibits STN-evoked complex EPSCs in SNr neurons. (A) A1: Example traces of the complex EPSCs in a SNr GABA neuron under control condition (black trace), during 10 μm 5-HT (red trace) and during 10 μm 5-HT+10 μm 5-HT1B receptor antagonist NAS-181 (blue trace). (A) A2: Scatter plot of the complex EPSC area under control condition, during 10 μm 5-HT and during 10 μm 5-HT+10 μm NAS-181 in the SNr GABA neuron shown in A1. (B) B1: Example traces of STNevoked complex EPSCs in a SNr GABA neuron before (black trace), during (red trace) and after (blue trace) bath application of 10 μm 5-HT1B receptor agonist CP93129. (B) B2: Scatter plot of the complex EPSC area before, during and after bath application of 10 μm CP93129 in the SNr GABA neuron shown in B1. (C) Example traces of STN-evoked complex EPSCs in an SNr GABA neuron before (black trace), during (red trace), and after (blue trace) bath application of 10 μm 5-HT1A receptor agonist 8-OH-DPAT. (D) Example traces of the complex EPSCs in an SNr GABA neuron under control condition (black trace), during 10 μm 5-HT (red trace), during 10 μm 5-HT+ 10 μm 5-HT1A receptor antagonist WAY100135 (green trace), and after returning to control solution (blue trace). The timescale in panel (D) is the same as in panel (C) (modified from Ding et al. with permission, 2013).
axon collaterals inside the STN that may support recurrent excitation in STN (Hammond and Yelnik, 1983; Kita et al., 1983; Ammari et al., 2010). A dendrodendritic interaction also may contribute to the complex EPSCs, although there is currently no data to support or exclude this possibility. Functionally, the STN-evoked complex EPSCs can induce bursting firing in the GPe, GPi, and SNr neurons, a form of abnormal firing pattern in these neurons often seen in Parkinson’s disease animal models and patients (Shen
and Johnson, 2006; Walters et al., 2007; Ammari et al., 2010; Ding et al., 2013). Equally important, anatomical evidence indicates that 5-HT may modulate the STN glutamatergic projections to its target nuclei and influence their firing frequency and patterns. Here we focus on 5-HT’s effects on the subthalamonigral glutamatergic projection because both anatomical and physiological data are available, whereas there is little physiological data on 5-HT regulation of other STN projections.
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons 611
Presynaptic 5-HT1B receptor activation reduces STN output to SNr GABA neurons Anatomical studies have shown that the SNr receives a very intense 5-HT innervation that is the highest among brain areas and is twice the 5-HT axon terminal density in the STN in both rodents and primates, including humans (Figure 2A) (Steinbusch, 1981; Baumgarten and Grozdanovic, 1997; Moukhles et al., 1997; Parent et al., 2010, 2011; Hashemi et al., 2011; Wallman et al., 2011). Like the intense 5-HT innervation, histochemical and ultrastructural studies also have demonstrated a strong expression of inhibitory 5-HT1B receptors in the axons but not in the neuronal somatodendritic areas in the SNr (Voigt et al., 1991; Maroteaux et al., 1992; Boschert et al., 1994; Sari et al., 1999; Riad et al., 2000; Sari, 2004). A lesion study indicated that around 50% of 5-HT1B receptors in the SNr was in striatonigral axons, whereas the sources of the remaining 50% of the receptors were not known (Sari et al., 1999). Because STN neurons have a strong expression of 5-HT1B receptor mRNA, it is likely that the STN→SNr axon terminals may express 5-HT1B receptors. The intensive 5-HT innervation in the SNr and the likely expression of the presynaptic 5-HT1B receptors on the STN efferent terminals provide the anatomical and molecular basis for 5-HT to gate or inhibit the STN→SNr glutamatergic projection and control the STN-triggered events in SNr neurons. We recently performed the following experiments to support these possibilities (Ding et al., 2013). We first examined the effect of bath-applied 5-HT on the STN-triggered complex EPSCs in SNr GABA neurons. We found that bath application of 10 μm 5-HT consistently inhibited the complex EPSCs with an IC50 of 4.24 μm (Figure 3A1). At 10 μm, 5-HT reduced the complex EPSC area by 66%. These results clearly indicated that 5-HT has a net inhibitory effect on the STN→SNr complex EPSCs. Additionally, we found that bath application of 10 μm 5-HT decreased the frequency but not the amplitude of the action potential-independent miniature EPSCs (mEPSCs) in SNr GABA neurons. These results indicate that 5-HT may inhibit complex EPSCs by activating the inhibitory 5-HT receptors on the STN→SNr axon terminals in the SNr, leading to reduced vesicular neurotransmitter release. Among the known 5-HT receptors, only 5-HT1 subfamily is inhibitory by coupling to Gi/o G-protein (Bockaert et al., 2006; Hannon and Hoyer, 2008; Millan et al., 2008). Among the brain 5-HT1 receptor subtypes, only 5-HT1A and 5-HT1B receptors are expressed at substantial levels (Bruinvels et al., 1993; Barnes and Sharp, 1999). Besides being autoreceptors in 5-HT neurons, 5-HT1B
receptors are known to selectively express in axon terminals of non-5-HT neurons, whereas 5-HT1A receptors may be expressed at the somata and also axon terminals (Sari et al., 1999; Riad et al., 2000). Thus, we reasoned that 5-HT may inhibit complex EPSCs by activating primarily 5-HT1B receptors on STN→SNr axon terminals. To test this possibility, we first examined the effect of CP93129, an established selective 5-HT1B receptor agonist (Li and Bayliss, 1998; Mizutani et al., 2006). As illustrated in Figure 3B1–2, bath application of 10 μm CP93129 reduced the complex EPSC area, mimicking the inhibitory 5-HT effect shown in Figure 3A1–2. Furthermore, bath application of 10 μm NAS181, a 5-HT1B receptor antagonist (Mizutani et al., 2006), blocked the effects of 10 μm 5-HT on complex EPSCs (Figure 3A1–2). Taken together, these results indicated that the inhibitory presynaptic 5-HT1B receptors may be mediating the 5-HT inhibition of the STN→SNr complex EPSCs. In contrast, we found that the 5-HT1A agonist 8-OH-DPAT and 5-HT1A receptor antagonist WAY100135 had no effect on the complex EPSCs or 5-HT’s effect (Figure 3C,D), indicating that 5-HT1A receptors do not make detectable contribution to the 5-HT inhibition of STN→SNr complex EPSCs, further supporting the primary role of 5-HT1B receptors. Additionally, we found that 5 μm Ro 600175, a 5-HT2C receptor agonist, and 10 μm BIMU-8, a 5-HT4 receptor agonist, failed to affect the STN→SNr complex EPSCs significantly. These results are not fully consistent with previous studies reporting that 5-HT2C and 5-HT4 receptors are expressed in the STN and increased STN neuron excitability (Pompeiano et al., 1994; Compan et al., 1996; Eberle-Wang et al., 1997; Xiang et al., 2005). One possibility is that 5-HT2C and 5-HT4 receptors may be expressed at relatively low levels in STN neurons (Clemett et al., 2000) such that they do not produce detectable effects on the recurrent excitation among STN neurons and hence the generation of the complex EPSCs. This possibility is consistent with the fact that 10 μm 5-HT reduced the interspike interval by a modest 22% (Xiang et al., 2005), whereas 10 μm 5-HT reduced the complex EPSC area by 66% (Ding et al., 2013); it is also consistent with in vivo data suggesting that 5-HT’s overall effect on STN neurons may be inhibitory (Liu et al., 2007; Aristieta et al., 2013); if the 5-HT2C/5-HT4 excitation was overwhelming, then the overall in vivo 5-HT effect would have been excitatory. The data discussed in the preceding sections were obtained with exogenously applied 5-HT ligands. So now the question is, can the 5-HT released from 5-HT axon terminals also reduce STN→SNr complex EPSCs in SNr GABA neurons? This is a particularly relevant question, because the SNr receives a dense 5-HT innervation (Figure 2A) that
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
612 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons may release sufficient amounts of 5-HT that in turn may activate presynaptic 5-HT1B receptors and inhibit glutamate release from the STN→SNr projection. We believe that endogenous 5-HT can exert effects that are qualitatively identical to these induced by exogenous 5-HT ligands. Certainly, endogenous 5-HT-induced effects may be small because the amount of endogenously released 5-HT is small compared with exogenously applied 5-HT ligands. We have recently tested this idea by examining the effect of the 5-HT1B receptor antagonist NAS-181 (Ding et al., 2013). As illustrated in Figure 4, bath application of 10 μm NAS-181 significantly increased the STN→SNr complex EPSCs area, and the effect was recovered upon washing out NAS-181. These results indicate that endogenously released 5-HT normally causes a tonic 5-HT1B activity that provides a moderate inhibitory control on glutamate release from STN→SNr axon terminals.
Presynaptic 5-HT1B receptor activation reduces STN-triggered burst firing in SNr GABA neurons The STN glutamatergic projection can trigger burst firing in SNr GABA neurons (Shen and Johnson, 2006; Ammari et al., 2010). We have discussed previously that the glutamate release from STN→SNr axon terminals is inhibited by presynaptic 5-HT1B receptors. So it can be predicted that 5-HT may inhibit STN-triggered burst firing in SNr GABA neurons. Indeed, we have shown that bath application of 10 μm 5-HT decreased the STN-triggered burst firing frequency (Ding et al., 2013). We also have observed a similar inhibitory effect of the 5-HT1B receptor agonist CP93129
B
100 pA 50 ms Control 10 µM NAS-181 Wash
Normalized complex EPSC area
A
* 100
50
0
Ctrl
NAS -181
Wash
Figure 4 Endogenous 5-HT, via 5-HT1B receptors, inhibits STNtriggered complex EPSCs in SNr GABA neurons. (A) Example traces of STN-evoked complex EPSCs in an SNr GABA neuron before (black), during (red), and after (blue) bath application of 10 μm NAS-181, a 5-HT1B antagonist. (B) Pooled data showing that bath application of 10 μm NAS-181 increased the complex EPSC area in 5 SNr GABA neurons. *p
Shengyuan Ding and Fu-Ming Zhou*
Serotonin regulation of subthalamic neurons Abstract: The subthalamic nucleus (STN) is a key component of the basal ganglia. As the only basal ganglia nucleus comprised of mostly glutamatergic neurons, STN neurons provide a key driving force to their target neurons. Thus, regulation of STN neuron activity is important. One STN regulator is the serotonin (5-HT) system. The STN receives a dense 5-HT innervation. 5-HT1A, 5-HT1B, 5-HT2C, and 5-HT4 receptors are expressed in the STN. 5-HT may regulate the STN via several mechanisms. First, 5-HT may affect STN neuron excitability directly by either inhibiting a subpopulation of STN neurons via activation of 5-HT1A receptors or exciting STN neurons through activation of 5-HT2C and 5-HT4 receptors. Second, 5-HT may affect synaptic inputs to the STN. Via activation of 5-HT1B receptors on the afferent terminals, 5-HT inhibits glutamatergic input to the STN, but the inhibitory effect on GABAergic input is smaller. Third, 5-HT may regulate the STN glutamatergic output by activating presynaptic 5-HT1B receptors, thus reducing burst firing in target neurons. Last, 5-HT may affect glutamate release at the intra-STN axon collaterals and regulate the recurrent excitation. These mechanisms may work in concert to fine-tune the intensity and pattern of STN activity and reduce STN output bursts. Keywords: 5-HT (5-hydroxytryptamine); basal ganglia; burst firing; Parkinson’s disease; presynaptic inhibition; recurrent excitation; serotonin receptor; substantia nigra; subthalamic nucleus. *Corresponding author: Fu-Ming Zhou, Department of Pharmacology, University of Tennessee College of Medicine, Memphis, TN 38163, USA, e-mail: [email protected] Shengyuan Ding: Department of Pharmacology, University of Tennessee College of Medicine, Memphis, TN 38163, USA
Introduction The subthalamic nucleus (STN) is a key component of the basal ganglia (BG). As the only BG nucleus comprised primarily of glutamatergic neurons (Smith and Parent, 1988; Lévesque and Parent, 2005; Gerfen and Bolam, 2010), the STN critically affects the activity of its target neurons, exerting its normal physiological function and also its dysfunction under pathophysiological
conditions, as demonstrated by the profound therapeutic effects of STN high-frequency stimulation in Parkinson’s disease (Perlmutter and Mink, 2006). Although intrinsically active, STN neurons are also under the regulation of several neurotransmitter systems. In this review, we discuss the serotonergic regulatory effects on STN neuron intrinsic activity and on the synaptic inputs to and synaptic output from STN neurons. Readers are also referred to several recent reviews that cover other aspects of the STN (Bevan et al., 2007; Charpier et al., 2010; Baunez et al., 2011; Wilson and Bevan, 2011).
Connectivity of the subthalamic nucleus As illustrated in Figure 1A and B, the STN is a critical component of the BG, a group of interconnected subcortical nuclei that also include the striatum (caudate-putamen), external and internal segments of the globus pallidus (GPe and GPi), substantia nigra pars compacta (SNc), and pars reticulata (SNr) (Nambu et al., 2002; Gerfen and Bolam, 2010; Haynes and Haber, 2013). The striatum and STN are the input structures, whereas GPi and SNr are output structures. Although most cells in the striatum, GPe, GPi, and SNr, are GABAergic with cells in SNc being largely dopaminergic, most, if not all, STN neurons use glutamate as their neurotransmitter, thus playing a unique role in the information processing and transfer in the BG (Parent and Hazrati, 1995; Hikosaka et al., 2000; Sano et al., 2013). The transmission of cortical information through the BG is mediated by three main pathways: the direct pathway, indirect pathway, and hyperdirect pathway (Nambu et al., 2002; Gerfen and Bolam, 2010; Haynes and Haber, 2013). In the direct pathway, glutamatergic corticostriatal information flows to the dopamine D1 receptor-expressing, GABAergic, often silent medium spiny neurons (D1-MSNs) in the striatum that in turn project to and inhibit the tonically firing GABAergic projection neurons in GPi and SNr, releasing the thalamocortical circuit (Figure 1A) (Hikosaka et al., 2000; Kravitz et al., 2010). In the indirect pathway, glutamatergic corticostriatal information is first transmitted to the dopamine D2 receptor-expressing, GABAergic, often silent medium spiny neurons (D2-MSNs) in the striatum that, in turn, project to
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
606 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons Anatomical studies using conventional labeling methods indicate that the glutamatergic STN neurons project not only to the neighboring SNr and SNc, GPe and GPi but also to the more distant targets, such as the striatum and even the cerebral cortex. They project in the rostral direction, and the pedunculopontine nucleus, the mesencephalic and pontine reticular formation projects in the caudal direction (Kita and Kitai, 1987; Parent and Hazrati, 1995; Sato et al., 2000; Degos et al., 2008). A recent study using a viral vector seeking enhanced visualization of the axonal arborization and the dendritic tree of single STN neurons in the rat brain has confirmed the findings from traditional tracing studies (Koshimizu et al., 2013). These projections provide the anatomical basis for glutamatergic STN neurons to affect their target neurons. Additionally, intra-STN axonal collaterals have been detected by intracellular staining, potentially enabling excitatory interaction among STN neurons, although paired recording is needed to prove this potential interaction (Kita et al., 1983; Ammari et al., 2010).
Figure 1 Location and connection of the subthalamic nucleus (STN). (A) A Nissl-stained sagittal adult mouse brain section showing the location of the STN and other major brain areas connected to the STN. Arrows indicate information flow to and from the STN. The STN receives a major glutamatergic cortical innervation (the hyperdirect pathway) and also a minor glutamatergic input from the thalamus. The STN sends a major projection to and receives a major GABAergic input from the globus pallidus external segment (GPe). The STN also intensely innervates the substantia nigra pars reticulata (SNr), forming the subthalamonigral projection. SC, superior colliculus. (B) A simplified diagram shows the main STN inputs and outputs. PPN, pedunculopontine nucleus. Glu, glutamate. GABA, γ-amino butyric acid. In both panels (A) and (B), the red lines indicate glutamatergic projections and the black lines indicate GABAergic projections. For simplicity, the substantia nigra pars compacta and its projection to the STN are neither labeled in (A) nor diagramed in (B).
and inhibit the spontaneously firing GABAergic projections in GPe. Consequently, the tonic or default GPe inhibition of STN is paused, leading to increased STN glutamatergic output to GPi and SNr and subsequent inhibition of the thalamocortical circuit (Hikosaka et al., 2000; Kravitz et al., 2010). In the hyperdirect pathway, glutamatergic cortical motor signal, and other information directly flow to and excite STN neurons, providing these glutamatergic neurons with additional excitation to drive their target neurons (Afsharpour, 1985; Nambu et al., 2002; Kita and Kita, 2012; Walker et al., 2012; Sano et al., 2013).
STN neurons are spontaneously active glutamatergic neurons Immunohistochemical and neurophysiological studies indicate that most, if not all, STN neurons use glutamate as their neurotransmitter (Smith and Parent, 1988; Lévesque and Parent, 2005; Wilson and Bevan, 2011), a unique feature in the BG that are dominated by GABAergic neurons. Studies performed in STN-containing brain slices in the presence of synaptic receptor blockers have established that STN neurons exhibit spontaneous rhythmic spiking activity around 10 Hz, generated by intrinsic ion channels independent of synaptic inputs (Nakanishi et al., 1987; Bevan and Wilson, 1999; Beurrier et al., 2000; Hallworth et al., 2003). The persistent voltage-gated sodium current is particularly important in triggering the spontaneous firing in STN neuron (Bevan and Wilson, 1999; Wilson and Bevan, 2011). In intact animals, STN neurons fire at higher rates with both regular and burst firing patterns (Bergman et al., 1994; Kreiss et al., 1997; Urbain et al., 2002). In normal monkeys, the average STN neuron firing rate is reported to be ∼25 Hz (Bergman et al., 1994; Wichmann et al., 1994; Isoda and Hikosaka, 2008). Similar data in normal humans are not available, but in Parkinson’s disease, the reported STN neuron firing rate is 40–60 Hz (Mrakic-Sposta et al., 2008; Steigerwald et al., 2008; Benedetti et al., 2009). The high-frequency spontaneous firing combined with the fact that most STN
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons 607
neurons are glutamatergic has led to the conclusion that STN neuron activity is a critical driving force in the basal ganglia (Kitai and Kita, 1987; Parent and Hazrati, 1995; Wilson and Bevan, 2011). Thus, regulation of STN neuron activity has important consequences. In this review, we focus on the 5-HT regulation of STN neuron activity.
5-HT innervation and 5-HT receptor expression in the STN Immunohistochemical studies using antibodies against 5-HT or 5-HT transporter (SERT) protein have shown that the STN receives a dense 5-HT innervation originated in the dorsal raphe nucleus (Figure 2A) (Steinbusch, 1981; Parent et al., 2010, 2011; Wallman et al., 2011). Quantitative ultrastructural studies have estimated that the SERT-positive 5-HT terminal density is 1.5 × 106/mm3 in the monkey STN and 1.2 × 106/mm3 in the rat STN (Parent et al., 2010, 2011), although the 5-HT axon terminal density in the SNr is even higher, being twice as that in the STN in both rodents and primates, including humans (Parent et al., 2010, 2011; Wallman et al., 2011). These studies further indicate that 5-HT axon terminals form both synaptic and nonsynaptic contacts with STN neurons (on dendrites and dendritic spines) in both rodents and nonhuman primates (Parent et al., 2010), similar to other brain areas (Van Bockstaele et al., 1994, 1996; Moukhles et al., 1997). Also, the majority of 5-HT axon terminal-formed synapses are asymmetric (Parent et al., 2010), although the functional implication of the morphological type of monoamine synapse is not clear, unlike the classical synapses with asymmetry being associated with excitatory synapses. These anatomical data indicate that 5-HT may influence STN neurons via the classical synaptic mode and also the volume transmission mode. 5-HT neurons often fire spontaneous action potentials at a tonic frequency of ∼1 Hz (Jacobs and Azmitia, 1992), triggering 5-HT release and providing the endogenous agonist 5-HT (Hashemi et al., 2011, 2012). The STN expresses several types of 5-HT receptors, specifically 5HT1A, 5-HT1B, 5-HT2C, and 5-HT4 receptors (Figure 2B), as detected by in situ hybridization, autoradiographic, and immunohistochemical studies (Maroteaux et al., 1992; Bruinvels et al., 1993; Boschert et al., 1994; Pompeiano et al., 1994; Wright et al., 1995; Compan et al., 1996; Eberle-Wang et al., 1997; Clemett et al., 2000; Vilaró et al., 2005). 5-HT1A and 5-HT1B receptors are coupled to Gi/o protein and are thus inhibitory, whereas 5-HT2C receptors are coupled to Gq/11 protein and 5-HT4 receptors are coupled to Gs protein; thus 5-HT2C and 5-HT4
Figure 2 5-HT innervation and 5-HT receptor expression in the STN. (A) A confocal image of a sagittal brain section showing the 5-HT innervation in the STN and neighboring brain areas. The 5-HT axons were detected by immunostaining the 5-HT transporter protein. Note that although the 5-HT innervation in STN is substantial, it is even more intense in the SNr. CP, cerebral peduncle; LHA, lateral hypothalamus (modified from Ding et al. with permission, 2013). (B) Diagram showing the established (supported by anatomical and/or physiological data), likely (some data) and potential (likely but no data yet, indicated by question marks) expression of 5-HT receptors in the STN. 5-HT1A, 5-HT1B, 5-HT2C and 5-HT4 receptors are abundant in the STN. 5-HT1A, 5-HT2C and 5-HT4 receptors are probably somatodendritic, whereas 5-HT1B receptors are expressed on the afferent axon terminals innervating the STN and also on axon terminals of the STN efferents. 5-HT1A receptor activation exerts an inhibitory effect in the STN, whereas 5-HT2C and 5-HT4 receptors activation excite STN. PPN, pedunculopontine nucleus.
receptors tend to be excitatory in neurons (Hannon and Hoyer, 2008; Millan et al., 2008). Comparison of the results from in situ hybridization studies and radioligand binding studies suggest that 5-HT2C, 5-HT4, and 5-HT1A receptors are most likely at STN neuron somatodendritic areas, whereas the 5-HT1B receptor appear be at the axon terminals of STN neurons and at the afferent axon terminals innervating STN neurons (Maroteaux et al., 1992; Bruinvels et al., 1993; Boschert et al., 1994; Pompeiano et al., 1994; Wright et al., 1995; Compan et al., 1996; EberleWang et al., 1997; Clemett et al., 2000; Vilaró et al., 2005). However, electron microscopic ultrastructural studies
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
608 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons are limited (Sari et al., 1999; Riad et al., 2000), and more studies are needed to confirm these potential subcellular localizations of 5-HT receptors in the STN and determine the subcellular expression intensity of these 5-HT receptors within STN. Despite a lack of detailed ultrastructural information, many laboratories around the world have studied how 5-HT affects STN neuronal activity and function. In the following sections, we seek to summarize and synthesize the results from these studies.
5-HT regulation of the intrinsic properties of STN neurons The regulation of 5-HT on the intrinsic properties of STN neurons has been studied by several groups using electrophysiological methods. Results from two in vitro electrophysiological studies indicate that 5-HT may excite the majority of STN neurons without affecting their firing patterns (Flores et al., 1995; Xiang et al., 2005). One group further suggested that 5-HT depolarized STN neurons through decreasing a potassium conductance, and this effect may be mediated by 5-HT2C and 5-HT4 receptors, since it was mimicked by a 5-HT2/5-HT4 receptor agonist and was partially blocked by a 5-HT4 receptor or a 5-HT2C receptor antagonist (Xiang et al., 2005). Two other groups reported a more complicated situation: 5-HT has both 5-HT2C- and 5-HT4-receptor-mediated excitatory effect and a 5-HT1A receptor-mediated inhibitory effect on STN neurons (Stanford et al., 2005; Shen et al., 2007). Shen et al. (2007) also indicated that multiple intrinsic membrane conductances were responsible for these different 5-HT effects. In a subset of STN neurons, 5-HT (30 μm) evoked a modest outward current (27 pA when clamped at -70 mV) through activation of 5-HT1A receptors, as the outward current was completely blocked by a 5-HT1A receptor antagonist. The outward current was mediated by a potassium conductance, because it was blocked by Ba2+, and its reversal potential was affected by changes in external K+ concentration. In a majority of STN neurons, 5-HT (30 μm) evoked a small inward current (17 pA at -70 mV) via activation of 5-HT2C and 5-HT4 receptors. The inward current was mediated by a reduction of an unknown potassium conductance and an increase in a mixed cation conductance. The reason is not known why only a subpopulation of STN neurons showed a 5-HT1 receptor-mediated outward current and another subpopulation of STN neurons showed a 5-HT2C/5-HT4 receptormediated inward current. Certainly, it is entirely possible that the expression of 5-HT receptors in the STN is not
homogenous. This possibility requires future combined electrophysiological and immunohistochemical studies in STN neurons. To further understand the regulatory roles in the STN, studies also have examined 5-HT effects on STN neuron activity in intact animals, and the conclusion is that the overall 5-HT effect is inhibitory (Liu et al., 2007; Aristieta et al., 2013). In one study, 5,7-dihydroxytryptamine lesion of the dorsal raphe nucleus led to an increased firing rate with more bursting firing in STN neurons (Liu et al., 2007). Based on their data that dopamine neuron lesion eliminated the effect of 5-HT neuron lesion, these authors concluded that the apparent 5-HT effects on STN neurons were mediated indirectly via nigral dopamine neurons. Although it is entirely possible for 5-HT neurons to affect dopamine neurons that, in turn, affect STN neurons, a pertinent question was, can 5-HT neurons also directly affect STN neurons in vivo giving the 5-HT innervation and 5-HT receptors in STN? This question has been answered by a recent commendable study that used several approaches to examine 5-HT effects on STN neuron activity in intact animals (Aristieta et al., 2013). This study found that 5-HT synthesis inhibition increased the STN neurons’ firing rate and bursting activity and these abnormalities were reversed by fluoxetine, a selective 5-HT reuptake inhibitor. Furthermore, they reported that in normal and 5-HT-depleted animals, direct local application of the 5-HT1A agonist 8-OH-DPAT robustly inhibited STN neuron firing in a dose-dependent manner that was prevented by pretreatment with the 5-HT1A antagonist WAY100635 (Aristieta et al., 2013). Additionally, these authors demonstrated that in normal and 5-HT-depleted animals, direct local application of the 5-HT2C agonist Ro 600175 substantially increased STN neuron firing in a dose-dependent manner that was prevented by pretreatment with the 5-HT2C antagonist SB 242084. These in vivo results from local STN drug application experiments are generally consistent with the data from in vitro studies as discussed, although the in vivo data suggest a net inhibitory 5-HT effect on STN neurons. This is understandable, because 5-HT also affects the neuronal networks that are connected to the STN such that 5-HT can influence STN neurons directly and indirectly. There is also a negative report that indicates that 5-HT depletion did not affect STN neuron firing significantly (Delaville et al., 2012a). The reasons for this discrepancy are not known. However, Delaville et al. (2012a) used a lower dose of a 5-HT synthesis inhibitor depleting brain tissue 5-HT by 73–82% while Aristieta et al. (2013) used a much higher dose of the same inhibitor depleting STN tissue 5-HT by more than 95%.
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons 609
This methodological difference may be a factor for the negative result of Delaville et al. (2012a). Taken together, available data from in vitro and in vivo studies indicate that the net serotonergic regulatory effect on the STN glutamatergic neurons may be inhibitory when the effects on intrinsic ion channels and the effects on synaptic inputs (discussed following) are combined. Further studies are needed to determine the relative influence of 5-HT1A receptor-mediated inhibition and 5-HT2C receptor/5-HT4 receptor-mediated excitation in STN neurons.
5-HT regulation of the synaptic inputs to STN neurons STN receives excitatory glutamatergic inputs from the cortex and the thalamus and inhibitory GABAergic inputs from the GPe (Afsharpour, 1985; Parent and Hazrati, 1995; Wilson and Bevan, 2011; Kita and Kita, 2012). These excitatory and inhibitory synaptic inputs have profound effects on the activities of STN neurons. In vivo recording studies in anesthetized rats indicate that the spontaneous firing of STN neurons is strongly but briefly increased by cortical glutamatergic inputs in the hyperdirect corticosubthalamic pathway evoked by electrical stimulation in sensorimotor cortical areas or the prefrontal cortex, whereas the spontaneous firing is substantially inhibited by GABAergic pallidosubthalamic inputs (Fujimoto and Kita, 1993; Maurice et al., 1998; Magill et al., 2004; Kita and Kita, 2011). These data from intact animals are further supported by in vitro studies that showed that GPe-originated GABAA inhibitory postsynaptic potentials (IPSPs) can inhibit the firing rate and change the intrinsically regular firing pattern to bursty firing patterns in STN neurons (Bevan et al., 2002a,b; Hallworth and Bevan, 2005). Studies indicate that the glutamatergic and GABAergic afferents to STN neurons may be modulated by the 5-HT system. 5-HT has been reported to inhibit the presumed corticosubthalamic glutamatergic excitatory postsynaptic currents (EPSCs) and pallidosubthalamic GABAergic inhibitory postsynaptic potentials (IPSCs) in STN neurons via activating presynaptic 5-HT1B receptors that may reduce the release of glutamate and GABA, respectively (Shen and Johnson, 2008). Moreover, in this study (Shen and Johnson, 2008), 5-HT appeared to be more potent in reducing EPSCs than IPSCs: 10 μm 5-HT reduced the EPSC amplitude by 35%, whereas 30 or 100 μm 5-HT was required to reduce the IPSC amplitude by 30% (10 μm
5-HT was ineffective). Although data on extracellular 5-HT concentrations in STN are not available, the endogenous extracellular 5-HT concentration may briefly reach 10 μm near 5-HT axon release sites but not likely to reach 30 or 100 μm. So presynaptic 5-HT1B receptors may moderately reduce the glutamatergic input to STN neurons, but the inhibitory effect on the GABAergic input is small and may not be physiologically relevant. The presynaptic 5-HT1B inhibition of synaptic inputs to STN neurons have functional importance. Because the 5-HT1B inhibition of the glutamatergic input to STN neurons appears to be larger than the 5-HT1B inhibition of the GABAergic input, when 5-HT neurons degenerate during PD, the loss of the 5-HT1B inhibitory effect on the glutamatergic input is presumably larger than that on the GABAergic input; consequently, the increase in the glutamatergic input to STN neurons also may be larger than the increase in the GABAergic input. These changes may contribute to the increased firing frequency and bursty firing pattern in STN neurons in 5-HT-depleted animals (Liu et al., 2007; Aristieta et al., 2013).
5-HT regulation of the synaptic output from STN neurons Now we turn to 5-HT regulation of STN neuron output from the efferent axon terminals in their target areas. As discussed previously, the glutamatergic STN neurons project to the GPe, GPi, and SNr (Kita and Kitai, 1987; Sato et al., 2000; Koshimizu et al., 2013). This glutamatergic STN projection is likely to be important in shaping the firing intensity and pattern of their target neurons in the GPe, GPi, and SNr (Kita and Kita, 2011). Furthermore, stimulation of the STN can evoke reverberating, long-lasting polysynaptic complex EPSCs in SNr GABA neurons (Figure 3A1, B1) and also in GPi and GPe GABA neurons (Shen and Johnson, 2006; Ammari et al., 2010; Ding et al., 2013). These intra-STN stimulation-evoked long-lasting polysynaptic complex EPSCs most likely are resulted from an intrinsic STN neuronal network for the following reasons. First, only electrical stimulation directly in the STN evokes complex EPSCs, indicating the activation of large numbers of STN neurons is required to evoke complex EPSCs. Second, upon severing the connection between the STN and pedunculopontine nucleus (PPN) and other nucleus but preserving only the connection between the STN and SNr, STN stimulation still can evoke complex EPSCs (Ammari et al., 2010; Ding et al., 2013). Third, anatomical studies have demonstrated the presence of STN neuron
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
610 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons
A
B
5-HT1B antagonist blocks 5-HT effect
A1
5-HT1B agonist mimics 5-HT effect
B1
10 µM 5-HT
10 µM CP93129
10 µM 5-HT+ 10 µM NAS-181 Control
40
Control
20
20
0
0
20
10 µM CP93129
10
0
40
0
Time course (min)
C
50 ms
B1
10 µM 5-HT Complex EPSC area (pA.s)
Complex EPSC area (pA.s)
50 pA
100 pA
10 µM NAS-181
A2
Wash
D
5-HT1A agonist has no effect
20 40 Time course (min)
5-HT1A antagonist has no effect
10 µM 5-HT 50 pA 100 pA
50 ms Control 10 µM 8-OH-DPAT Wash
Wash Control 10 µM WAY100135 +10 µM 5-HT
Figure 3 Exogenous 5-HT or 5-HT1B receptor agonist CP93129 inhibits STN-evoked complex EPSCs in SNr neurons. (A) A1: Example traces of the complex EPSCs in a SNr GABA neuron under control condition (black trace), during 10 μm 5-HT (red trace) and during 10 μm 5-HT+10 μm 5-HT1B receptor antagonist NAS-181 (blue trace). (A) A2: Scatter plot of the complex EPSC area under control condition, during 10 μm 5-HT and during 10 μm 5-HT+10 μm NAS-181 in the SNr GABA neuron shown in A1. (B) B1: Example traces of STNevoked complex EPSCs in a SNr GABA neuron before (black trace), during (red trace) and after (blue trace) bath application of 10 μm 5-HT1B receptor agonist CP93129. (B) B2: Scatter plot of the complex EPSC area before, during and after bath application of 10 μm CP93129 in the SNr GABA neuron shown in B1. (C) Example traces of STN-evoked complex EPSCs in an SNr GABA neuron before (black trace), during (red trace), and after (blue trace) bath application of 10 μm 5-HT1A receptor agonist 8-OH-DPAT. (D) Example traces of the complex EPSCs in an SNr GABA neuron under control condition (black trace), during 10 μm 5-HT (red trace), during 10 μm 5-HT+ 10 μm 5-HT1A receptor antagonist WAY100135 (green trace), and after returning to control solution (blue trace). The timescale in panel (D) is the same as in panel (C) (modified from Ding et al. with permission, 2013).
axon collaterals inside the STN that may support recurrent excitation in STN (Hammond and Yelnik, 1983; Kita et al., 1983; Ammari et al., 2010). A dendrodendritic interaction also may contribute to the complex EPSCs, although there is currently no data to support or exclude this possibility. Functionally, the STN-evoked complex EPSCs can induce bursting firing in the GPe, GPi, and SNr neurons, a form of abnormal firing pattern in these neurons often seen in Parkinson’s disease animal models and patients (Shen
and Johnson, 2006; Walters et al., 2007; Ammari et al., 2010; Ding et al., 2013). Equally important, anatomical evidence indicates that 5-HT may modulate the STN glutamatergic projections to its target nuclei and influence their firing frequency and patterns. Here we focus on 5-HT’s effects on the subthalamonigral glutamatergic projection because both anatomical and physiological data are available, whereas there is little physiological data on 5-HT regulation of other STN projections.
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons 611
Presynaptic 5-HT1B receptor activation reduces STN output to SNr GABA neurons Anatomical studies have shown that the SNr receives a very intense 5-HT innervation that is the highest among brain areas and is twice the 5-HT axon terminal density in the STN in both rodents and primates, including humans (Figure 2A) (Steinbusch, 1981; Baumgarten and Grozdanovic, 1997; Moukhles et al., 1997; Parent et al., 2010, 2011; Hashemi et al., 2011; Wallman et al., 2011). Like the intense 5-HT innervation, histochemical and ultrastructural studies also have demonstrated a strong expression of inhibitory 5-HT1B receptors in the axons but not in the neuronal somatodendritic areas in the SNr (Voigt et al., 1991; Maroteaux et al., 1992; Boschert et al., 1994; Sari et al., 1999; Riad et al., 2000; Sari, 2004). A lesion study indicated that around 50% of 5-HT1B receptors in the SNr was in striatonigral axons, whereas the sources of the remaining 50% of the receptors were not known (Sari et al., 1999). Because STN neurons have a strong expression of 5-HT1B receptor mRNA, it is likely that the STN→SNr axon terminals may express 5-HT1B receptors. The intensive 5-HT innervation in the SNr and the likely expression of the presynaptic 5-HT1B receptors on the STN efferent terminals provide the anatomical and molecular basis for 5-HT to gate or inhibit the STN→SNr glutamatergic projection and control the STN-triggered events in SNr neurons. We recently performed the following experiments to support these possibilities (Ding et al., 2013). We first examined the effect of bath-applied 5-HT on the STN-triggered complex EPSCs in SNr GABA neurons. We found that bath application of 10 μm 5-HT consistently inhibited the complex EPSCs with an IC50 of 4.24 μm (Figure 3A1). At 10 μm, 5-HT reduced the complex EPSC area by 66%. These results clearly indicated that 5-HT has a net inhibitory effect on the STN→SNr complex EPSCs. Additionally, we found that bath application of 10 μm 5-HT decreased the frequency but not the amplitude of the action potential-independent miniature EPSCs (mEPSCs) in SNr GABA neurons. These results indicate that 5-HT may inhibit complex EPSCs by activating the inhibitory 5-HT receptors on the STN→SNr axon terminals in the SNr, leading to reduced vesicular neurotransmitter release. Among the known 5-HT receptors, only 5-HT1 subfamily is inhibitory by coupling to Gi/o G-protein (Bockaert et al., 2006; Hannon and Hoyer, 2008; Millan et al., 2008). Among the brain 5-HT1 receptor subtypes, only 5-HT1A and 5-HT1B receptors are expressed at substantial levels (Bruinvels et al., 1993; Barnes and Sharp, 1999). Besides being autoreceptors in 5-HT neurons, 5-HT1B
receptors are known to selectively express in axon terminals of non-5-HT neurons, whereas 5-HT1A receptors may be expressed at the somata and also axon terminals (Sari et al., 1999; Riad et al., 2000). Thus, we reasoned that 5-HT may inhibit complex EPSCs by activating primarily 5-HT1B receptors on STN→SNr axon terminals. To test this possibility, we first examined the effect of CP93129, an established selective 5-HT1B receptor agonist (Li and Bayliss, 1998; Mizutani et al., 2006). As illustrated in Figure 3B1–2, bath application of 10 μm CP93129 reduced the complex EPSC area, mimicking the inhibitory 5-HT effect shown in Figure 3A1–2. Furthermore, bath application of 10 μm NAS181, a 5-HT1B receptor antagonist (Mizutani et al., 2006), blocked the effects of 10 μm 5-HT on complex EPSCs (Figure 3A1–2). Taken together, these results indicated that the inhibitory presynaptic 5-HT1B receptors may be mediating the 5-HT inhibition of the STN→SNr complex EPSCs. In contrast, we found that the 5-HT1A agonist 8-OH-DPAT and 5-HT1A receptor antagonist WAY100135 had no effect on the complex EPSCs or 5-HT’s effect (Figure 3C,D), indicating that 5-HT1A receptors do not make detectable contribution to the 5-HT inhibition of STN→SNr complex EPSCs, further supporting the primary role of 5-HT1B receptors. Additionally, we found that 5 μm Ro 600175, a 5-HT2C receptor agonist, and 10 μm BIMU-8, a 5-HT4 receptor agonist, failed to affect the STN→SNr complex EPSCs significantly. These results are not fully consistent with previous studies reporting that 5-HT2C and 5-HT4 receptors are expressed in the STN and increased STN neuron excitability (Pompeiano et al., 1994; Compan et al., 1996; Eberle-Wang et al., 1997; Xiang et al., 2005). One possibility is that 5-HT2C and 5-HT4 receptors may be expressed at relatively low levels in STN neurons (Clemett et al., 2000) such that they do not produce detectable effects on the recurrent excitation among STN neurons and hence the generation of the complex EPSCs. This possibility is consistent with the fact that 10 μm 5-HT reduced the interspike interval by a modest 22% (Xiang et al., 2005), whereas 10 μm 5-HT reduced the complex EPSC area by 66% (Ding et al., 2013); it is also consistent with in vivo data suggesting that 5-HT’s overall effect on STN neurons may be inhibitory (Liu et al., 2007; Aristieta et al., 2013); if the 5-HT2C/5-HT4 excitation was overwhelming, then the overall in vivo 5-HT effect would have been excitatory. The data discussed in the preceding sections were obtained with exogenously applied 5-HT ligands. So now the question is, can the 5-HT released from 5-HT axon terminals also reduce STN→SNr complex EPSCs in SNr GABA neurons? This is a particularly relevant question, because the SNr receives a dense 5-HT innervation (Figure 2A) that
Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 6/1/15 8:08 AM
612 S. Ding and F.-M. Zhou: Serotonin regulation of subthalamic neurons may release sufficient amounts of 5-HT that in turn may activate presynaptic 5-HT1B receptors and inhibit glutamate release from the STN→SNr projection. We believe that endogenous 5-HT can exert effects that are qualitatively identical to these induced by exogenous 5-HT ligands. Certainly, endogenous 5-HT-induced effects may be small because the amount of endogenously released 5-HT is small compared with exogenously applied 5-HT ligands. We have recently tested this idea by examining the effect of the 5-HT1B receptor antagonist NAS-181 (Ding et al., 2013). As illustrated in Figure 4, bath application of 10 μm NAS-181 significantly increased the STN→SNr complex EPSCs area, and the effect was recovered upon washing out NAS-181. These results indicate that endogenously released 5-HT normally causes a tonic 5-HT1B activity that provides a moderate inhibitory control on glutamate release from STN→SNr axon terminals.
Presynaptic 5-HT1B receptor activation reduces STN-triggered burst firing in SNr GABA neurons The STN glutamatergic projection can trigger burst firing in SNr GABA neurons (Shen and Johnson, 2006; Ammari et al., 2010). We have discussed previously that the glutamate release from STN→SNr axon terminals is inhibited by presynaptic 5-HT1B receptors. So it can be predicted that 5-HT may inhibit STN-triggered burst firing in SNr GABA neurons. Indeed, we have shown that bath application of 10 μm 5-HT decreased the STN-triggered burst firing frequency (Ding et al., 2013). We also have observed a similar inhibitory effect of the 5-HT1B receptor agonist CP93129
B
100 pA 50 ms Control 10 µM NAS-181 Wash
Normalized complex EPSC area
A
* 100
50
0
Ctrl
NAS -181
Wash
Figure 4 Endogenous 5-HT, via 5-HT1B receptors, inhibits STNtriggered complex EPSCs in SNr GABA neurons. (A) Example traces of STN-evoked complex EPSCs in an SNr GABA neuron before (black), during (red), and after (blue) bath application of 10 μm NAS-181, a 5-HT1B antagonist. (B) Pooled data showing that bath application of 10 μm NAS-181 increased the complex EPSC area in 5 SNr GABA neurons. *p
