Glutamatergic fast-spiking parvalbumin neurons in the lateral hypothalamus: Electrophysiological properties to behavior

Abstract

Throughout the central nervous system, neurons expressing the calcium-binding protein parvalbumin have been typically classified as GABAergic with fast-spiking characteristics. However, new methods that allow systematic characterization of the cytoarchitectural organization, connectivity, activity patterns, neurotransmitter nature, and function of genetically-distinct cell types have revealed populations of parvalbumin-positive neurons that are glutamatergic. Remarkably, such findings challenge longstanding concepts that fast-spiking neurons are exclusively GABAergic, suggesting conservation of the fast-spiking phenotype across at least two neurotransmitter systems. This review focuses on the recent advancements that have begun to reveal the functional roles of lateral hypothalamic parvalbumin-positive neurons in regulating behaviors essential for survival.

1. Introduction

Neuronal circuits are generally depicted as a diverse collection of excitatory and inhibitory neurons expressing the neurotransmitters glutamate and γ-aminobutyric acid (GABA), respectively. More than three decades ago, studies from cortical [1] and hippocampal [2] circuits in rodents showed that the calcium-binding protein parvalbumin (PVALB; PV) was expressed in GABAergic interneurons with fast-spiking action potential phenotypes (Fig. 1). Since then, PV has been depicted as a neuronal marker though it is also found in fast-twitch muscles [3–5]. Moreover, previous studies have implicated PV in modulating intracellular calcium dynamics in neurons [6–10]. Together, these observations led to the longstanding concept that a hallmark of fast-spiking PV neurons is their GABAergic nature. Furthermore, extensive work on the diversity of inhibitory interneurons has shown that parvalbumin-positive interneurons in both the neocortex and hippocampus are necessary and sufficient for the generation of network oscillations, a prominent feature of neuronal activity throughout the animal kingdom [11–13]. Recent studies using PV as a genetic marker to label and manipulate putative fast-spiking interneurons show that GABAergic PV neurons are also found in areas distinct from the hippocampus and cortex, such as the striatum [14]. However, there are a number of caveats concerning the identification of fast-spiking PV neurons, particularly in the hippocampus, as PV is not only expressed in fast-spiking interneurons, but also in other cell types that may be excitatory [13,15]. Thus, PV-positive GABAergic interneurons may only be a hallmark of cortical circuitry, as studies over the last several years have identified several collections of glutamatergic PV neurons throughout the brain [16–24].

Fig. 1.

Representative images depicting GABAergic and glutamatergic PV-positive neurons from different brain regions in mice. Immunohistochemical (IHC) identification of GABAergic (top) and glutamatergic (bottom) parvalbumin-expressing neurons (PVALB; green) in horizontal sections from multiple brain regions. Abbreviations: CA1, field CA1 of the hippocampus; CA3, field CA3 of the hippocampus; DG, dentate gyrus; EP, entopeduncular nucleus; LH, lateral hypothalamus; SC, superior colliculus. 10x confocal images. Scale bars = 50 μm.

Pioneering work using immunohistochemistry and gene expression analysis showed that within the lateral hypothalamus (LH), PV-expressing neurons colocalize with glutamate or mRNA for vesicular glutamate transporter 2 (Slc17a6 commonly known as Vglut2) in both rats and mice [16–18]. Another study identified a PV-positive glutamatergic neuronal population in the superior colliculus as a key mediator of fear responses in mice [21]. Separately, single-cell transcriptional and molecular analyses revealed that the entopeduncular nucleus (EP) contains a group of glutamatergic PV-positive neurons amidst a larger population of GABAergic PV neurons (Fig. 1) [20]. Moreover, ventral pallidal circuits contain a subset of PV projection neurons that express VGLUT2 [22]. These neurons have been implicated in the regulation of depression, aversion, and reward-seeking behaviors [22,24]. Additionally, the presence of overlapping markers for GABA and glutamate has not been reported in any of the PV populations mentioned above, a phenomenon recently described for neurons in the EP and ventral tegmental area (VTA) [20,25–28]. Thus, while studies have begun to identify and characterize the functional roles of PV neurons that express and release glutamate, much about their fundamental electrophysiological properties, activity dynamics, synaptic connections, and behavioral functions remains to be determined. Here, we describe studies examining a role for the small collection of glutamatergic PV-expressing neurons in the LH (LH^PV neurons), a brain region crucial for orchestrating survival behaviors, with a particular emphasis on their behavioral roles.

2. Molecular identity of LH^PV neurons

Seminal work examining the distribution of calcium-binding proteins in the rat brain revealed a “constant cluster” of PV-positive neurons in the LH among otherwise scattered PV-positive fibers of the medial forebrain bundle [16]. Two decades later, this “PV1 nucleus,” the anatomical region of the LH containing these neurons, was further characterized as a slender, longitudinal collection of cells situated parallel to the fornix, spanning over 2 mm in the rat hypothalamus and over 1 mm in the mouse hypothalamus [17]. Immunolabeling and gene expression analyses indicate that both glutamate and mRNA for Vglut2 are found in these LH^PV neurons [17,18]. However, hypocretin or other calcium-binding proteins such as calbindin or calretinin were not found in these cells [17]. These findings collectively indicate that LH^PV neurons are glutamatergic. However, it was not until recently that the fundamental electrophysiological properties of these glutamatergic LH^PV neurons were quantitatively measured and their ability to release glutamate and form excitatory synapses was demonstrated [19].

3. LH^PV neurons are fast-spiking and send long-range projections

Our work demonstrated for the first time that LH^PV neurons can sustain action potentials at high-frequency, with average maximal firing frequencies of 194 Hz [19]. Moreover, these LH^PV neurons also exhibit passive and active membrane properties such as resting membrane potential (V_rmp) and action potential half-width (AP_HW) (Table 1) that are similar to the properties of both hippocampal and neocortical PV-positive GABAergic interneurons [13,19,29–31]. Interestingly, a recent study showed that ventral pallidal glutamatergic PV-projecting neurons also display a depolarized V_rmp and high spontaneous firing activity [22]. Additionally, in LH^PV neurons, single-cell RT-qPCR analysis showed that these cells are equipped with ion channels implicated in setting and regulating fast-spiking activity, for example the delayed rectifier voltage-gated potassium channels KCNC1 (KV3.1) and KCNC2 (KV3.2) and the hyperpolarization-activated cation channels (HCNs) [11,19]. We also characterized LH^PV neuron synaptic properties using channelrhodopsin (ChR2)-assisted circuit mapping (CRACM). Using this approach, we found that LH^PV neurons provide excitatory glutamatergic synaptic inputs to neurons within the LH. This was confirmed by quantitative measurements of LH^PV neurons co-expressing Vglut2 mRNA, which revealed that approximately 95% of LH^PV cells are glutamatergic [19].

Table 1.

Electrophysiological properties of GABAergic and glutamatergic parvalbumin-positive neurons from different brain regions. Electrophysiological properties are similar between PV-positive GABAergic neurons in the hippocampus and neocortex and PV-positive glutamatergic neurons in the lateral hypothalamus. Abbreviations: V_rmp, resting membrane potential; AP_HW, action potential half-width. Values are reported as mean ± s.d. or s.e.m.

Brain region	Maximal firing frequency (Hz)	Vrmp (mV)	APHW (ms)	Reference
Hippocampus (basket cells)	230 ± 15	−62 ± 3	0.25 ± 0.04	Lübke et al., 1998 [30]
Neocortex	199 ± 64	−68 ± 6.5	0.46 ± 0.13	Schiff and Reyes, 2012 [31]
Lateral hypothalamus	194 ± 10	−66 ± 1.2	0.45 ± 0.02	Kisner et al., 2018 [19]

Several studies have demonstrated that LH^PV neurons send long-range projections to brain regions involved in reward, motivation, and nociception [19,32,33]. In particular, LH^PV neurons densely project to the ventrolateral periaqueductal gray (vlPAG), mainly to the supraoculomotor nucleus (Su3) region [19,32]. Furthermore, we demonstrated their ability to release glutamate and form excitatory synapses on neurons in the vlPAG [33]. LH^PV neurons additionally project to the lateral habenula (LHb), with sparser labeling in several other regions including the fasciculus retroflexus and surrounding parafascicular thalamic nucleus, dorsomedial tegmental area, superior colliculus, magnocellular red nucleus submedius thalamic nucleus, posterior hypothalamus, retromamillary nucleus, and reticulotegmental nucleus of the pons [19,32].

4. LH^PV neurons drive survival behaviors

For decades, lateral hypothalamic circuits have been depicted as key modulators of feeding and reward-related behaviors [34]. However, the contributions of these circuits and specific genetically-identified neuronal types to other survival behaviors require further examination, particularly those of LH^PV neurons and their long-range axonal projections. Initial studies using chemical ablation of LH^PV neurons in rats [35] or conditional inactivation of glutamatergic signaling in PV-expressing neurons in mice [36] reported a decrease in ultrasonic vocalizations and locomotion, respectively. Moreover, these conditional knockout mice (Vglut2^flox/flox;Pvalb-Cre) also displayed other behavioral changes such as an increase in social dominance and decreases in exploratory activities and thermal nociception [36]. Since deletion of Vglut2 from PV neurons was not restricted to the LH, caution should be used when interpreting these results as this deletion targeted all neurons co-expressing parvalbumin and Vglut2 throughout the brain. Thus, it is possible that the behavioral changes observed were triggered by the recently identified brain circuits that contain glutamatergic parvalbumin-positive neurons such as the superior colliculus, entopeduncular nucleus, and ventral pallidum [20–22,24]. In addition, the reported behavioral outcomes may be explained by impaired coordination and locomotor activity. Similar severe impairments of motor coordination were observed in mice with selective inactivation of the GABA_A receptor γ2 subunit from PV-positive neurons [37], indicating that impaired locomotor activity is common during whole-brain deletion of both glutamatergic and GABAergic signaling in PV neurons. Alternatively, genetic compensation, a widespread phenomenon [38], could have occurred after deletion of Vglut2 from PV neurons potentially contributing to the behavioral changes observed. Therefore, while studies such as these have begun to identify the functional roles of LH^PV neurons, the specific contributions of these neurons to behavior have yet to be determined.

Recently, we used optogenetic approaches to demonstrate that LH^PV neurons modulate nociception in mice. We found that photostimulation of LH^PV neurons reduced nociception to an acute, noxious thermal stimulus, whereas photoinhibition increased thermal nociception. We also demonstrated that activation of LH^PV axons within the vlPAG (LH^PV→ vlPAG) attenuates both thermal and visceral nociception, suggesting that this pathway not only modulates responses to acute reflex-withdrawal assays of pain but also to an ongoing noxious stimulus. In contrast to previous studies [36], we did not observe changes in locomotion or anxiety-related behaviors when activating LH^PV neurons or the LH^PV→vlPAG pathway. Thus, our results support the idea that activation of the LH^PV→vlPAG pathway may be specific to nociception insofar as has been examined [33].

Interestingly, this antinociceptive effect appears to occur independently of opioidergic mechanisms, as antagonism of opioid receptors with systemically-administered naltrexone did not abolish the antinociception encoded by the LH^PV→vlPAG circuitry [33]. Together, our findings suggest that the LH^PV→vlPAG pathway may represent a novel translational target for pain management without the use of opioids.

5. Future directions for LH^PV research

Over the past year our detailed characterization of the electrophysiological properties and neurotransmitter nature of LH^PV neurons has revealed that these neurons are a functional component within the LH glutamatergic circuitry. At the behavioral and circuit levels, we demonstrated that LH^PV neurons modulate nociception through connections in the vlPAG. However, further analyses are required to determine both the specific functional inputs that regulate LH^PV neuronal activity and the post-synaptic cell types that are crucial for orchestrating these behaviors. Moreover, future studies should seek to determine the behavioral relevance of other LH^PV axonal outputs such as those to the parafascicular nucleus and lateral habenula. Recent optogenetic studies have identified important roles for these regions in either pain or aversion [39,40], thus it is possible that LH^PV neurons integrate incoming information about noxious stimuli and regulate the activity of various downstream regions accordingly. Identifying such interactions will enable our understanding of the specific contributions of these circuitries to behaviors that are essential for survival.

Interestingly, a previous study showed that inhibitory inputs from the bed nucleus of the stria terminalis (BNST) specifically innervate and suppress glutamatergic neurons in the LH [41]. Therefore, it is possible that inputs from the BNST may modulate LH^PV neuronal activity as LH^PV neurons are a subset of glutamatergic cells within the LH. Moreover, a recent study in mice and rats using viral tracers identified other potential inputs to the lateral hypothalamic area containing PV-positive neurons [42]. These tracing experiments revealed that neurons from the lateral and ventrolateral orbitofrontal cortex (OFC) send excitatory axonal projections specifically to the LH^PV nucleus while avoiding the rest of the hypothalamus [42]. However, functional synapses between LH^PV and OFC neurons and the contributions of these interactions to behavior have yet to be examined.

The OFC is a brain region generally associated with decision-making [43], olfactory and gustatory sensory processing [44], cardiovascular and respiratory control [45,46], and emotion [47]. While initial studies on LH^PV neurons suggested their role in positive affective behaviors potentially through interactions with the OFC [35], our most recent study demonstrating their involvement in the modulation of pain through the LH^PV→vlPAG pathway [33] sets an interesting challenge for determining the specific roles of the OFC→LH pathway. Although the OFC generally takes part in higher-order decision-making processes [48], it is possible that the OFC→LH circuitry may encode decision-making for avoiding or withstanding pain depending on environmental threats. Understanding the nature of this OFC→LH→vlPAG connection should be the subject of future studies. Furthermore, the diverse collection of intermingled cell types in the LH and its multiple connections throughout the brain would suggest that LH^PV neurons may receive inputs from other areas as well. Therefore, using a combination of selective targeting, electrophysiology, CRACM, and single-cell RT-qPCR to examine the nature of both inputs to and outputs from LH^PV neurons will expand our understanding of how these circuits regulate survival behaviors.

6. Conclusions

Characterizing the heterogeneous nature of the LH has been a challenge for decades; however, recent tools have begun to make this possible. Early pioneering work demonstrated the regulation of PAG circuitry and nociception by electrical stimulation of the LH [49–52], and our optogenetic experiments have helped to elucidate at least one LH cell type that may have played a role in those findings, broadening the repertoire of behaviors governed by genetically-defined LH circuits. While our work suggests that the LH^PV→vlPAG pathway may represent an attractive translational target for pain management without the use of opioids, thus reducing the possibility of addiction, it is unclear at present how these findings may be extrapolated to clinical treatment of pain syndromes. Pivotal work examining the similarities between the human lateral tuberal nucleus (LTN; a hypothalamic structure in humans and primates) and the rodent lateral hypothalamic area containing PV-positive neurons showed that both are located within the most lateral part of the hypothalamus but differ in neuronal markers [53]. Immunohistochemical analysis showed little to no PV antibody staining in the human LTN, unlike the rodent LH^PV area, and immunoreactivity for somatostatin and neuropeptide FF receptor 1 (FF1) was detected in the human LTN but not in the rodent LH^PV area, demonstrating that these structures express different neuronal markers [53]. However, as stated by Gerig and Celio [53], phylogenetic differences in neuronal markers are not uncommon, being found in areas such as the hippocampus and basal ganglia [54,55], and as such may not preclude a conserved functional role. As hypothalamic transcriptomes for humans, non-human primates, and rodents are published with single-cell resolution, it will be important to compare cell types from this LH^PV/LTN region across organisms and determine better markers for translational studies. While the role of the human LTN is unknown, it is affected by Huntington’s disease [56,57], a neurode-generative condition which in later stages is associated with pain and negative affective disorders including anxiety and depression [58]. Therefore, it is possible that studies from LH^PV neurons in rodents will provide a better understanding of the roles of the human LTN and whether this brain region could be targeted for novel pain therapies.