Neurochemical characterization of NOS-only positive neurons
Our previous analysis of EPN neurons revealed the presence of SOM/PV/ChAT-triple negative neurons (Miyamoto and Fukuda
2015). The results of the present study yielded the conclusion that this population mostly corresponds to NOS-only positive neurons. Because the SOM/PV/ChAT-triple negative population consists of glutamic acid decarboxylase (GAD)-positive neurons (14.4%) and GAD-negative, presumptive glutamatergic neurons (7.7%) (Miyamoto and Fukuda
2015), it seems reasonable to deduce that NOS-only positive neurons can be divided into GAD-positive and negative populations. In fact this was confirmed in our unpublished observations in colchicine-pretreated animals, in which both GAD-positive and negative subpopulations were identified in NOS neurons of the EPN under the experimental conditions where colchicine successfully induced intrinsically weak GAD-immunoreactivity in somata of striatal medium spiny neurons (Ribak et al.
1979). Therefore, excitatory transmission in the EPN–LHb pathway (Shabel et al.
2012) will be mediated by not only SOM neurons (Wallace et al.
2017) but also the presumptive glutamatergic subpopulation of NOS neurons. It remains unknown whether GAD-positive NOS neurons release both GABA and glutamate as observed in SOM-positive EPN neurons (Shabel et al.,
2014; Wallace 2017).
The internal division of the EPN
The EPN has been divided into rostral and caudal halves depending on the type of neurons and their projection targets (van der Kooy and Carter
1981; Rajakumar et al.
1994). However, our recent study revealed that the EPN can be viewed as a double elliptical structure, which we termed core/shell organization, based on the immunohistochemical properties (Miyamoto and Fukuda
2015). The core region occupies the center of the caudal EPN and is enriched with PV neurons, whereas PV-negative neurons surround the PV-rich core in the caudal EPN and further extend rostrally to occupy the whole body of the rostral EPN. The present results show that not only SOM neurons but also NOS neurons occupy this shell-like zone at both the rostral and caudal levels. Thus, the position of the core that corresponds to the internal ellipsoid deviates caudally inside the external ellipsoid that represents the whole EPN. The tracer experiments further substantiate the core/shell organization in that the distributions of the thalamus- and lateral habenula-targeting neurons correspond well to the core and shell zones, respectively. From this perspective, NOS neurons can be characterized as shell neurons targeting mostly the lateral habenula. Similar structures are mentioned as “core” and “periphery” organization in the primate GPi (Parent and Bellefeuille
1982).
Although SOM neurons are distributed diffusely in the shell, NOS-immunoreactive neurons are mainly located in the ventromedial part of the shell. This suggests that the shell, defined by both the accumulation of lateral habenula-targeting neurons and the enrichment of PV-negative neurons, contains an NOS-rich subarea.
Projection targets of NOS-immunoreactive neurons
From past studies, the projection targets of EPN neurons can be related to the differences in the chemical characteristics of the neurons. EPN neurons projecting to the thalamus contain PV (Rajakumar et al.
1994), whereas the projection target of the SOM-containing EPN neurons is the LHb (Vincent and Brown
1986). The localization patterns of these two neuronal types coincide well with the core/shell organizations (Miyamoto and Fukuda
2015). However, this dichotomous view needs to be updated, because the present results demonstrate a new population, NOS positive neurons, in the EPN. NOS neurons were distributed in the EPN shell and projected to the LHb just as SOM-positive neurons. Moreover, a certain number of NOS-immunoreactive neurons were labeled after FG injection into the VA-VL, and these neurons were located also in the shell. This suggests two possibilities: distinct subpopulations of NOS neurons target either LHb or VA-VL but are located in the same division of the EPN, or single NOS neurons send axon collaterals to both LHb and VA-VL. The present results did not solve this issue, and previous studies regarding the connectivity between the EPN and its target regions led to controversial results. One study reported that distinct types of EPN neurons projected to the thalamus and LHb (Rajakumar et al.
1993), whereas another study showed that some EPN neurons projecting to the LHb extended the axon collateral to the thalamus (Kha et al.
2000). In relation to these two possibilities, another point to be considered is the relative abundance of the retrogradely labeled cells. Because the relative coverage of the injection site by FG was much smaller in the VA-VL as compared to the full coverage in the LHb, the number of labeled EPN cells after a single injection to the VA-VL as shown in Fig.
10 is thought to be an underestimate. Thus, apparently fewer labeling after the VA-VL injection does not necessarily exclude the possibility that many LHb-targeting neurons have collateral axons to the thalamus. Future studies are needed to resolve this important issue, because the EPN plays a pivotal role in sending information processed inside the basal ganglia toward many brain regions to execute appropriate behaviors.
Functional implications
LHb-projecting neurons in the primate GPi respond to both nonreward and reward cues, with their activities increasing and decreasing after cue presentation, respectively (Hong and Hikosaka
2008,
2013). In the rodent EPN, LHb-projecting GP (GPh) neurons are important in controlling the valence and responses to aversive stimuli (Shabel et al.
2012; Stephenson-Jones et al.
2016). A recent study using in vivo electrophysiology showed that individual GPh neurons located in the EPN are classified into two types, inhibitory (reward cue-inhibited neurons) and excitatory (reward cue-excited neurons), by presentation of reward cues, and that each type contrastingly responds toward aversive stimuli in a direction opposite to that toward reward cues (Li et al.
2019). Furthermore, LHb neurons of different subpopulations also take a response pattern similar to either of the two patterns in EPN neurons (Li et al.
2019), suggesting the possibility that EPN–LHb pathway encoding valence consists of parallel projections. However, this scheme is complicated by the co-release of GABA/glutamate from axon terminals of SOM-positive presynaptic neurons onto postsynaptic LHb neurons (Shabel et al.
2012,
2014; Wallace 2017). It has been demonstrated that the dual antagonistic responses in LHb neurons depend on the balance and synaptic strength between GABA and glutamate transmission (Li et al.
2011; Meye et al.
2016), which might be mediated through modulation of GABA/glutamate co-releasing synapses from single neurons. Alternative possibility is that the dual responses in the LHb can be mediated by afferent signals from distinct subsets of EPN neurons, one predominantly glutamatergic and the other GABAergic. Most of SOM-positive GPh neurons express both GAD and VGluT2, but they predominantly behave as excitatory neurons (Wallace et al.
2017). This brings up an issue of whether there is a subset of predominantly GABAergic EPN neurons. NOS-only positive neurons showing GAD immunoreactivity might be a candidate of GABAergic subset in the EPN-LHb circuitry.
A recent study using the single cell analysis of dissolved neurons from the EPN demonstrated the existence of purely glutamatergic GPh neurons that belong to a minor subpopulation of PV neurons (Wallace et al.
2017). These neurons were identified as PV neurons based on the results in both optogenetic manipulation of PV-Cre mice and in situ hybridization study to detect PV mRNA. However, our previous quantitative immunohistochemical analysis led to some different conclusion that all PV neurons in the EPN were GAD-positive (Miyamoto and Fukuda
2015). Moreover, in the present analysis, no PV-positive neurons were found among 823 FG-labeled EPN neurons after FG injection into the LHb (Fig.
9a). These discrepancies might be explained by the methodological difference between the mRNA-driven analysis and immunohistochemical detection of expressed protein. There are several examples of very low to almost no immunohistochemical signals in somata of neurons that should express the proteins, such as GAD in the medium-sized spiny neurons in the striatum (Ribak et al.
1979), and GAD65 in both hippocampal PV neurons (Fukuda et al.
1997) and cerebellar Purkinje cells (Esclapez et al.
1994; Obata et al.
1999). Thus, expression of PV protein in purely glutamatergic neurons of the EPN might be under the detection level in conventional immunohistochemical procedures.
Taken together, to interpret the diverse functional properties of LHb neurons in response to aversive stimuli, at least four separate groups of GPh neurons should be taken into consideration: SOM-positive neurons that co-release GABA and glutamate (Shabel et al.
2014; Wallace et al.
2017), purely glutamatergic PV neurons (Wallace et al.
2017), and NOS-positive neurons that consist of both GABAergic and glutamatergic subpopulations. The possibility of GABA/glutamate co-release from GABAergic NOS neurons needs to be determined in future studies.
NO, which is synthesized from L-arginine by the action of NOS, is used as a neurotransmitter, neuromodulator, and neural messenger in the central nervous system (Snyder and Bredt
1991). It acts diffusively on both anterograde and retrograde signals independent of synapses and receptors, because NO is a lipophilic and gaseous molecule (Brenman and Bredt
1997; Dawson and Snyder
1994). One of the important functions of NO is synaptic plasticity, such as long-term potentiation and depression (Calabresi et al.
1999; Hardingham et al.
2013; Lev-Ram et al.
1997; Schuman and Madison
1994). Since synaptic plasticity is a basic neural mechanism for learning and memory, NO might also be involved in several functions of the basal ganglia, such as the initiation of voluntary movements, control of procedural memory, and goal-directed action generation, leading to cognitive rewards. Therefore, NOS-containing neurons found in this study may be one of the important pieces to coordinate these activities.