Understanding the Habenula: a Major Node in Circuits Regulating Emotion and Motivation

Abstract

Over the last decade, the understanding of the habenula has rapidly advanced from being an understudied brain area with the Latin name ‘habena” meaning “little rein”, to being considered a “major rein” in the control of key monoaminergic brain centers. This ancient brain structure is a strategic node in the information flow from fronto-limbic brain areas to brainstem nuclei. As such, it plays a crucial role in regulating emotional, motivational, and cognitive behaviors and has been implicated in several neuropsychiatric disorders, including depression and addiction. This review will summarize recent findings on the medial (MHb) and lateral (LHb) habenula, their topographical projections, cell types, and functions. Additionally, we will discuss contemporary efforts that have uncovered novel molecular pathways and synaptic mechanisms with a focus on MHb-Interpeduncular nucleus (IPN) synapses. Finally, we will explore the potential interplay between the habenula’s cholinergic and non-cholinergic components in coordinating related emotional and motivational behaviors, raising the possibility that these two pathways work together to provide balanced roles in reward prediction and aversion, rather than functioning independently.

Graphical Abstract

The graphical abstract depicts the habenula, a small brain structure that is connected to several brain regions, including the Interpeduncular Nucleus (IPN), rostromedial tegmental nucleus (RMTg), ventral tegmental area (VTA), substantia nigra pars compacta (SNc), and Raphe. The habenula is divided into medial (MHb) and lateral (LHb) parts, which contain cholinergic (ChAT) and non-cholinergic components. It responds to nicotine, opioids, and stress and regulates depression and addiction. Specific synaptic mechanisms occur at MHb-IPN synapses that co-release Acetylcholine (ACh) and glutamate (Glu), are highly enriched in nicotinic Acetylcholine Receptors (nAChRs) and express Voltage gated calcium channels (VGCC) and AMPA/NMDA-type glutamate receptors. The graphical abstract highlights the importance of the habenular complex in modulating behavior, including reward processing and motivation.

1. Anatomy and connectivity of the habenula

The anatomy and topography of the habenular circuit is as striking as it is beautiful. Conserved from lampreys and zebrafish to humans, this small ancient brain structure located in the epithalamus of the diencephalon appeared in vertebrates more than 360 million years ago (Evans et al., 2018). The habenula is a bilateral structure that is asymmetric in lower vertebrates, which show lateral morphological and molecular differences and laterotopic representation of left-right information (Aizawa et al., 2005). This asymmetry is much less evident in rodents and almost nonexistent in humans, with only very subtle differences in volume, metabolism and susceptibility to damage between the left and right sides (Ahumada-Galleguillos et al., 2017). Each side of the habenula has two separate components: the medial habenula (MHb) and the lateral habenula (LHb) in mammals, equivalent to the dorsal habenula (DHb) and the ventral habenula (VHb), respectively, in lower vertebrates (Ahumada-Galleguillos, 2016; Namboodiri et al., 2016); in the human brain, though, the habenular complex arrangement is more similar to that of fish, where the MHb has been displaced to the outer portion of the complex (Hu et al., 2020). These two habenular structures have distinct connectivity, cytoarchitecture and gene expression profiles and are thought to subserve distinct but related functions (Aizawa et al., 2012; Hikosaka, 2010; Wagner et al., 2016).

1.1. Connectivity to cortical, limbic, and midbrain areas

The lateral and medial habenular systems are mono- or di-synaptically connected to cortical, limbic, and midbrain areas but largely separate from each other. Inputs to the LHb include the medial prefrontal cortex (mPFC), basal forebrain (including the bed nucleus of the stria terminalis (BNST), hypothalamus (lateral hypothalmus and suprachiasmatic nucleus (SCN)), basal ganglia (specifically the entopeduncular nucleus (EPN)), and olfactory bulb The outputs of the LHb encompass a wider number of target structures than those from the MHb, including direct outputs to the brainstem, periaqueductal grey, rostromedial tegmental nucleus (RMTg), and key monoaminergic areas. These include the dopaminergic areas: substantia nigra pars compacta (SNc) and ventral tegmental area (VTA), the serotoninergic raphe nuclei, and the locus coeruleus, which is the main site of norepinephrine production in the brain. For more detailed information on habenula connectivity see (Antolin-Fontes et al., 2015; Geisler and Trimble, 2008; Hikosaka et al., 2008; Mathis and Kenny, 2019; Namboodiri et al., 2016; Root et al., 2014; Shabel et al., 2014) Conversely, the MHb primary receives inputs from the cholinergic basal forebrain (septum and diagonal band of Broca), and its output is almost entirely to the interpeduncular nucleus (IPN). Postsynaptic IPN cells innervated by MHb neurons send direct efferents to the median, paramedian, and dorsal raphe (DR), as well as to the area surrounding the cholinergic laterodorsal tegmentum (LDTg). IPN anterograde tracing studies also show a small ascending projection back to the MHb and to the thalamus (Ables et al., 2017), while retrograde studies show that IPN populations receive major input from the MHb and raphe and smaller sources of direct input from the septum and the prefrontal cortex (Ables et al., 2017). Intra-habenular connectivity exists between the two sides of the LHb, where each LHb sends output to the contralateral LHb (Company et al., 2021; Herkenham and Nauta, 1979), while the two sides of the MHb seem to share information by converging into the IPN in the midline. There is some suggestion that connections exist between the lateral and medial habenular nuclei as well as within the MHb itself, where some MHb neurons appear to send axons within the MHb and others to the adjacent LHb, although the converse LHb to MHb connection has not been demonstrated (Cuello et al., 1978; Kim and Chang, 2005). Some of the controversy may arise because the axons from the MHb first turn laterally to join the LHb axons before forming the combined fasciculus retroflexus (fr) fiber tract to the midbrain. Future studies using 3D imaging techniques combined with genetically identified cell populations can resolve whether an intrahabenular connection exists. Within the IPN, local projections between the two intermediate sides of the IPN (IPI) areas exist, as well as local cell populations in the IPN and along the IPN-raphe axis (Ables et al., 2017). Thus, these two habenular systems seem to have evolved as two largely independent and parallel diencephalic connection pathways that share information within each system with limited exchange of information between systems.

1.2. Cytoarchitectural subdivisions of Hb and IPN with specific topographic connections

The MHb nuclei lie next to the third ventricle and contain large projection neurons that are packed together and project their axons in a bundle to form the fr that runs diagonally and ends in the IPN. The lateral habenula nuclei lie at the lateral sides of the MHb nuclei, are less neuron-dense, and have axonal projections that incorporate within the shell of the fr and continue past the IPN to other brain areas (Company et al., 2021; Herkenham and Nauta, 1979; Ichijo and Toyama, 2015). Classical MHb studies have focused on two mutually exclusive populations: ventral MHb^ChAT neurons that express choline acetyltransferase (ChAT): the enzyme that synthesizes Acetylcholine (ACh) and dorsal MHb^Tac1 neurons that express Tac1; the gene that encodes the neuropeptide substance P (Cuello, 1987). While most neurons in LHb and MHb are glutamatergic, cholinergic MHb^ChAT neurons also corelease acetylcholine (ACh) and MHb^Tac1 peptidergic neurons corelease substance P (Cuello, 1987) and glycine (Melani et al., 2019). Each habenular division can be further subdivided into distinct subnuclei based on marker expression and topographic connectivity (see Figure 1 and detailed discussion below), although there is no convention in naming the subnuclei within MHb or LHb. Generally, there is agreement in dividing the MHb into 5 subnuclei: dorsolateral (dlMHb, also known as superior), dorsomedial (dmMHb, also known as dorsal), ventrolateral (vmMHb), ventrocentral (vcMHb), and ventromedial (vmMHb), with an additional area near the stria medularis along the border of the MHb and LHb designated as habenula X (HbX) (Antolin-Fontes et al., 2015; Lee et al., 2019; Wallace et al., 2020). The LHb can be divided into at least 4 subnuclei: lateral (lLHb), central (cLHb), ventromedial (vmLHb, also called marginal), and medial (mLHb, also called oval), with some groups defining up to 6 subnuclei (Hashikawa et al., 2020; Hu et al., 2020; Lee et al., 2019; Wallace et al., 2020). The IPN can be divided into 3 midline (rostral (IPR), apical (IPA), and central (IPC)) and 4 paired subnuclei (dorsomedial (IPDM), dorsolateral (IPDL), intermediate (IPI), and lateral (IPL)), although recent advancing imaging techniques have revealed a fifth paired subnucleus (rostrolateral (IPRL)) (Lee et al., 2019; Wang et al., 2020). Topographic organization of the habenular nuclei was first demonstrated in zebrafish (Aizawa et al., 2012). There is a topographic organization of axons in the fr not only between LHb axons in the shell and MHb axons in the core but also between ventral MHb axons that terminate in the central part of the IPN and dorsal MHb axons that end in the lateral sides of the IPN (Ichijo and Toyama, 2015). Recent studies have further refined the topographic organization to demonstrate that neurons within the specific subnuclei in the MHb project to discrete and specific subnuclei within the IPN (Amo et al., 2010; Shih et al., 2014) (Figure 1). Moreover, the laterotopic organization in zebrafish is partially maintained in mammals, where cholinergic terminals crisscross the midline and end in the ipsilateral side of the IPN with respect to the left or the right ventral MHbs (vMHbs) (Figure 1) while MHb^Tac1 terminals end in both lateral sides of the IPN. Cholinergic neurons also demonstrate medial to lateral topographical organization that translates into dorsal-ventral gradients in the IPN, excluding the lateral subnuclei (Shih et al., 2014). Pictures and movies of the 3D structure of the habenula illustrate its stunning complexity (Ables et al., 2017; Amo et al., 2010; Gardon et al., 2014; Morton et al., 2018).

Figure 1. Cell types and connectivity:

(A) Subnuclei within the habenula: The medial habenula (MHb) can be subdivided into 5 subnuclei, while the lateral habenula (LHb) can be subdivided into 4 subnuclei, with an area near the stria medularis along the border between the lateral and medial habenula designated HbX.

(B) Topographic projection of neurons within the habenula: The MHb sends projections to the IPN that are arranged topographically. Dorsal MHb nuclei project bilaterally to the lateral subnuclei of the IPN, while ventral MHb subnuclei project to the rostral, intermediate, and central nuclei. Within the ventral MHb nuclei, a medial to lateral gradient of projections exists that correlates to a dorsal to ventral gradient within the IPN. Other relevant inputs into the LHb, MHb, and IPN are noted.

(C) Cell marker expression in the habenula: Each subnucleus within the habenular complex is identified by a combination of markers. Here we have displayed a select list of relevant marker expression, based on transcript expression, fluorescent-tagged protein expression and proteomics, and confirmed by in situ hybridization using the Allen Brain Atlas. The level of expression correlates with the size of the circle, with the smallest circle indicating consistent, low-level detection, the mid-size circle indicating consistent moderate expression or scattered cells in the subnuclei with intense expression, and the large circle indicating intense expression and dense coverage across the subnucleus.

2. Molecularly defined habenular cell populations harboring nicotinic acetylcholine receptors and other molecules underlying cholinergic related behaviors

Initial efforts to characterize the cell populations in habenula using markers separated the MHb from the LHb based on expression of Tac2 (Dao et al., 2014; Norris et al., 1993; Yip and Chahl, 2001) and Pchd10 (Aizawa et al., 2012; Amo et al., 2010) and further subdivided the MHb into dorsal MHb^Tac1 and ventral MHb^ChAT (Aizawa et al., 2012). The lateral habenula, however, remained an ill-defined heterogenous glutamatergic population without discrete markers. With the advent of cell-type specific transcriptomics and spatial transcriptomics, much progress has been made in the past decade to refine the cell populations within both divisions of the habenular complex as well as in the IPN, the primary target of the MHb. Further, profiling across several species has confirmed homology of the habenular complex at both the molecular and anatomic levels and demonstrated that these molecularly defined populations also demonstrate topographic connectivity (Allen_Institute, 2011; Caligiuri et al., 2022; Hashikawa et al., 2020; Wallace et al., 2020). Perhaps the most important finding from these studies is that a combination of markers is needed to define neuronal subpopulations within both divisions of the habenula, highlighting their heterogeneity.

Consistent with their prominent role in nicotine addiction, the MHb and IPN display some of the most dense expression of multiple subunits of nicotinic acetylcholine receptors (nAChRs), linked to nicotine addiction through gene-wide association studies (GWAS) (reviewed in (Antolin-Fontes et al., 2015), among others). The habenular complex and target structures also densely express components of several other neurotransmitter and neuropeptide systems that are key in regulating consummatory behavior, including opioid and endocannabinoid receptors, tachykinins and their receptors, and tyrosine hydroxylase, though the contribution of these systems within the habenular complex to nicotine addiction has only begun to be explored (see section 3). In this section, we will review in detail the advances in our understanding of the different markers that can be used to identify and target key neuronal populations in the habenula and how discrete populations connect to their target structures. Lastly, we will discuss other non-neuronal cell types that may contribute to nicotine’s action in the CNS.

2.1. Medial Habenula

As expected of a brain structure with a prominent role in nicotine addiction, the MHb expresses a wide variety of nAChRs at very high levels. Utilizing transgenic mice expressing GFP-tagged nicotinic receptor subunits, Drenan and colleagues were able to demonstrate subregion-specific expression of nAChRs in the MHb, along the fr and in the IPN ((Shih et al., 2014); see Figure 1). Expression of α3, β4, and β3 is highest in the vMHb and fr, while α4 and β2 subunits are expressed at lower levels. Interestingly, α6 expression is limited to the medial aspect of the vMHb with very little detected in the fr. Despite being highly associated with nicotine addiction, the α5 subunit was not visualized in this study, nor are transcripts detectable in WT mice (Ables et al., 2017; van de Haar et al., 2022). However, this could be a species difference, as a study utilizing rats found that reducing expression of α5 subunits in the MHb contributed to increased nicotine intake (Fowler et al., 2011), or it could be a matter of detection, as transgenic reporter lines display expression in the vlMHb and the HbX (Morton et al., 2018). Utilizing a similar fluorophore-tagged approach, detailed determination of mu opioid receptor (MOR) expression in MHb indicates significant expression in comparison to most other brain areas, with highest expression in the lateral MHb subnuclei (Gardon et al., 2014). As mentioned previously, the dorsal MHb expresses the Tac1 gene, which encodes four peptides including substance P, neurokinin A, neuropeptide K, and neuropeptide gamma. The Tac1R tachykinin receptor (also known as NK1R) is expressed in the vMHb (Dao et al., 2014; Norris et al., 1993; van de Haar et al., 2022; Yip and Chahl, 2001). Proteomic approaches have confirmed expression of these basic markers and further established that the MHb expresses a total of 262 neuropeptides in the MHb (Yang et al., 2018a). Like tachykinin, many of these neuropeptides are derived from prohormones that undergo further cleavage to generate the final biologically active peptides, including many peptides involved in metabolic process (e.g., neuropeptide Y, cholecystokinin). The presence of this wide variety of neuropeptides within the MHb suggests that some of these may be ligands for the enrichment of orphan GPCRs in the MHb, including Gpr151 and Gpr139, among others (Ehrlich et al., 2018).

More recent efforts to molecularly characterize the MHb have employed cell type-specific profiling (Ables et al., 2017; Antolin-Fontes et al., 2015; Duncan et al., 2019; Gorlich et al., 2013) or single-cell and/or spatial transcriptomics of roughly 7,000-10,000 neurons (Caligiuri et al., 2022; Hashikawa et al., 2020; Sylwestrak et al., 2022; Wallace et al., 2020). While some limitations exist in these studies – primarily the use of only male mice in some (Hashikawa et al., 2020) and sex not defined in others (Wallace et al., 2020) or both sexes pooled together (Ables et al., 2017; Antolin-Fontes et al., 2015) – overall these studies have significantly advanced our understanding of the populations of neurons within the habenula and what combinations of markers can be used to target them. There is good agreement that the MHb can be clustered into 5-6 discrete cell populations based on gene expression. Key markers of MHb populations besides ChAT, Tac1, or nAChRs include Gpr151, Gpr139, Tac2, Tac1R, Oprm1, Cartpt, Pouf4f1 (aka Brn3a) (see Figure 1) (Hashikawa et al., 2020; van de Haar et al., 2022; Wallace et al., 2020). Further refinement of cell types in the habenula can be reflected in serotype specificity for AAV infection, with the MHb being resistant to most serotypes aside from AAV1 (Sylwestrak et al., 2022). This raises the possibility of using promoters and serotype combinations to target specific populations in the habenula for functional testing.

2.2. Lateral Habenula

Though the LHb does not express nAChRs, several groups have demonstrated altered activity in the LHb in response to nicotine or withdrawal, suggesting that some cholinergic machinery must be present, specifically the muscarinic cholinergic receptor 3 (Chrm3) and a6 and b3 nAChRs (Shih et al., 2015; Wallace et al., 2020). Sparse Tac1 expressing cells are also present in the LHb and are involved in reward computations (Sylwestrak et al., 2022). Proteomic approaches revealed 177 neuropeptides isolated from the rat LHb, with 126 shared with the MHb, many with known roles in motivated behavior, addiction, and appetite (Yang et al., 2018a). Molecular profiling has also revealed that LHb can be clustered into 4-6 populations that spatially correlate to five subnuclei with distinct topographical organization of projections and can be targeted using a combination of markers, rather than single markers (Figure 1). For a more detailed review of LHb structure and function, the reader is directed to several recent publications (Dai et al., 2022; Hu et al., 2020).

2.3. Interpeduncular Nucleus

Somewhat unusually, the MHb projects to a single target, the IPN. In addition to the MHb, the IPN receives input from the raphe, the septum, and possibly the VTA (Ables et al., 2017; Molas et al., 2023). Like the MHb, the IPN is characterized by expression of unique nAChRs that contribute to the effects of nicotine use, dependence, and withdrawal. In the IPN, GFP-tagged nAChRs a3 and b4 show dense expression along the habenular axons throughout all IPN subnuclei, followed by b3 and b2 subunits. Here, b3 expression is greater in rostral nuclei and largely excluded from lateral subnuclei, with b2 following a similar pattern. a4 expression is densest along the outer fr as it enters the IPN, continuing into the dorsal IPN and some expression into more posterior ventral subnuclei. α6 seems to be expressed only at very low levels in the rostral IPN, with much more expression in the adjacent VTA (Shih et al., 2014). However, because of the density of expression of the tagged subunits in the IPN, it was unclear from this work if the expression could be attributed to terminals from habenular neurons or to the neurons of the IPN itself. Utilizing TRAP-seq to profile transcripts in genetically targeted MHb and IPN populations, work from our group was able to confirm that both the MHb and the IPN express Chrna3, Chrna4, Chrna5, Chrnb3, and Chrnb4 subunits (Ables et al., 2017; Antolin-Fontes et al., 2015). Most notably, multiple groups have demonstrated significant expression of Chrna5 in the IPN in both mice and rats (Ables et al., 2017; Correa et al., 2019; Morton et al., 2018). Unlike the MHb, the IPN also expresses Chrna7 (Correa 2019). Besides nAChRs, the IPN also expresses other receptors relevant for nicotine dependence including Tac3R (also known as NK3R; Figure 3) (Dao et al., 2014; Norris et al., 1993; Yip and Chahl, 2001), CB1 receptors (Melani et al., 2019; Soria-Gomez et al., 2015), and CrhR1 (Zhao-Shea et al., 2013), although a full discussion is beyond the scope of this review.

Figure 3. Molecular mechanisms in habenula-IPN synapses.

(A) Glutamatergic transmission: During low frequency synaptic transmission, Mg²⁺ blocks the NMDA receptor. High frequency transmission expels Mg²⁺ from the NMDA receptor, allowing Na⁺ and Ca²⁺ influx.

(B) Glutamate and ACh corelease and vesicular synergy: Acetylcholine is coreleased with glutamate by habenular cholinergic neurons. Glutamate transmission follows the “wired transmission” mode. Acetylcholine transmission follows the “volume transmission” mode (Frahm et al., 2015; Ren et al., 2011).

(C) Presynaptic facilitation by nAChRs: nAChRs located on cholinergic terminals facilitate the release of acetylcholine (ACh) constituting a fail-safe mechanism at the neuromuscular junction (Mandl and Kiss, 2007) but also in habenular cholinergic terminals (Frahm et al., 2015; Girod and Role, 2001).

(D) GABA_B coupling to R-type Ca_v2.3 channels :GABA_B mediates excitation by amplifying presynaptic Ca²⁺ entry through Ca_v2.3 channels and potentiating co-release of glutamate, acetylcholine, and neurokinin B to excite interpeduncular neurons (Zhang et al., 2016).

(E) Nitric oxide retrograde signaling: Chronic nicotine treatment in mice dramatically increases the levels of Nitric oxide synthetase (NOS1) in a specific a5-nAChR expressing IPN neuronal population. Activation of soluble guanylate cyclase (sGC) by nitric oxide (NO) in habenular terminals leads to formation of the second messenger cGMP, which inhibits synaptic transmission (Ables et al., 2017).

(F) Somatostatin feedback inhibition: Chronic nicotine treatment in mice increases somatostatin (SST) levels in in a specific a5-nAChR- expressing IPN neuronal population. SST released from the IPN activates habenular SSTR2/4, reduces cAMP, and inhibits presynaptic neurotransmission (Ables et al., 2017).

2.4. Non-neuronal Cells in Habenula

Single-cell transcriptomic studies have revealed a significant component (~50%) of non-neuronal cell types within the habenular complex, including astrocytes, oligodendrocytes and their precursors, as well as a much smaller contribution of microglia (2.6%), mural (2.1%), endothelial (1.4%) and ependymal (0.35%) cells (Hashikawa et al., 2020) with good concordance among the other data sets (Caligiuri et al., 2022; Wallace et al., 2020). Perhaps this is not surprising given the proximity of the MHb to the third ventricle and to the myelinated input and output tracts, but the response of pericytes, ependymal cells, astrocytes, and oligodendrocytes to nicotine exposure is quite striking and suggests that they contribute more than previously appreciated (Caligiuri et al., 2022). One consideration is that nicotine acts on blood vessels themselves, hence the response of these cell types to nicotine, but it should also be acknowledged that in the nicotine-exposed dataset, these non-neuronal cells are over-represented compared to control habenula due to loss of cholinergic neurons, possibly skewing the data. However, there is significant evidence to suggest that cells in the vasculature and blood brain barrier (e.g., mural, endothelial, ependymal, and choroid plexus) may contribute to behavioral responses to drugs and to mood states. In support of non-neuronal cells in habenula regulating mood states, microarray profiling of postmortem habenula from individuals who completed suicide indicate endothelial dysfunction (Kim et al., 2022).

There is an intimate interplay between endothelial cells and astrocytes in regulating access from the periphery to brain. The habenula contains two distinct subtypes of astrocytes (Hashikawa et al., 2020), marked by expression of Ntsr2; however, it remains unclear to what extent habenular astrocytes express nAChRs or respond to nicotine themselves and what functional contribution they make in the habenula. There is precedence for Chrna7 expression and function in astrocytes and microglia, where activation leads to metabotropic signaling instead of ionotropic signaling and serves to limit neuroinflammation and glutamate toxicity (reviewed in (Piovesana et al., 2021)), and precedence for a role of astrocytes in regulating mood and drug-taking (reviewed in (Lacagnina et al., 2017); see also (Cui et al., 2018; Valentinova et al., 2019)).

3. The habenula: A relay pathway with complex functions

The cholinergic composition and connectivity to serotoninergic, dopaminergic, and noradrenergic centers make the habenula a critical neuroanatomic hub regulating very diverse motivational states and functional roles. These include reward and anti-reward processing, motivational and affective behavior, behavioral adaptation, and sensory integration. It also provides a counterbalancing loop circuit to the more positively-valenced mesolimbic circuit (Hikosaka, 2010; Hikosaka et al., 2008; Li et al., 2011). Because of this, neuromodulation of the habenula using deep brain stimulation (DBS) has been proposed to treat several neuropsychiatric disorders such as depression, addiction, obsessive-compulsive disorder, and bipolar disorder (de Koning et al., 2011; Loo et al., 2011; Luigjes et al., 2012; Mayberg, 2009; Mayberg et al., 2005; Sartorius and Henn, 2007). A first successful case of habenula DBS to treat TRD (treatment-resistant major depressive disorder) was reported a decade ago (Mayberg et al., 2005) and more than five clinical trials are ongoing (Germann et al., 2021), illustrating the critical role of this brain structure in human mental health.

3.1. Parallel reward and anti-reward pathways in the habenula.

The habenula has been described as an anti-reward center because of the response of LHb neurons to aversive stimuli and negative reward prediction error (Hikosaka, 2010; Proulx et al., 2014) and, to a lesser extent, because MHb^ChAT neurons have been implicated in aversion (Boulos et al., 2020; Choi et al., 2021; Frahm et al., 2011; McLaughlin et al., 2017; Morton et al., 2018) and anxiety (Cho et al., 2020; Mathuru and Jesuthasan, 2013; Seigneur et al., 2018; Yamaguchi et al., 2013). Studies in monkeys using in vivo electrophysiology during behavior found an unexpected activation of LHb neurons to signal negative reward prediction errors (Bromberg-Martin and Hikosaka, 2011; Matsumoto and Hikosaka, 2007). This critical role in predicting negative reward renewed the interest in the habenula and has led to numerous studies in rodents and fish (Cui et al., 2018; Hu et al., 2020; Lecca et al., 2014; Li et al., 2011; Park et al., 2017). The LHb signals negative reward prediction errors, conveys this information to monoaminergic centers, and integrates multiple inputs related to these predictions (Namboodiri et al., 2016). The LHb also influences abstract inference including gathering of information, flexible reasoning, and spatial memory (Goutagny et al., 2013; Namboodiri et al., 2016; Vadovicova, 2014). Fiber photometry population dynamics in mice have confirmed that reward acquisition errors activate LHb^Tac1 neurons, while reward outcome and reward history activate dMHb^Tac1 neurons (Sylwestrak et al., 2022). MHb^ChAT neurons show pre-cue activity, suggesting a role in the attentional component of the task, and diminish their activity during reward. These studies did not investigate the response to an aversive stimulus (i.e., nicotine or a bitter solution) to evaluate whether MHb^ChAT neurons are activated by an aversive stimulus, rather than only slightly decreasing their activity during reward. Both MHb^ChAT and LHb^Tac1 neurons were activated in the open arm of the elevated plus maze (EPM), suggesting a role in anxiety. Thus, opposing valence signals for MHb^Tac1 (positive reward) and LHb^Tac1 and MHb^ChAT neurons (negative reward and anxiety) provide additional evidence of parallel reward and anti-reward pathways in the habenula. An elegant conceptualization of the role of the habenular complex in reward and anti-reward systems in the context of substance use disorders has been proposed by the laboratory of Kenny and colleagues (Mathis and Kenny, 2019). This model differentiates between the early and late stages of drug intake in non-addicted and in addicted brains. Drug intake during the early stages inhibits the habenula and disinhibits the downstream structures (HPA axis, VTA, raphe, and IPN, but not the RMTg) promoting feelings of pleasure that facilitate consumption. In the late stage of drug intake, withdrawal increases the activity of the habenular complex to signal “anti-reward” that leads to feelings of discomfort and aversion and stop drug consumption. This reward and anti-reward balance is compromised in addicted subjects who experience fewer feelings of pleasure (less reward) and become more tolerant to drug intake (less aversive/less anti-reward) (Mathis and Kenny, 2019).

3.1.1. Role of the habenula in stress responses

The LHb and, to a lesser extent, the MHb, have been shown to have a role in mediating stress responses. Early lesion studies of the habenula or transection of its efferents in the fr showed increases in endocrine and behavioral measures of stress (Lee and Huang, 1988; Murphy et al., 1996; Thornton et al., 1994). Varied stressors including noxious pain, elevated plus maze, social defeat, and physical constraint have been found to activate the habenula (Hikosaka, 2010; Namboodiri et al., 2016). Recently, it has been shown that stressful experiences reduce synaptic strength at LHb excitatory synapses and can disrupt the cognitive mechanisms underlying goal-directed behaviors like decision-making (Lewis, 2021; Nuno-Perez et al., 2021). Drug withdrawal and excessive (binge intoxication) drug intake can be considered as stressors (Chartoff and Carlezon, 2014) and will be discussed next.

3.1.2. Role of the habenula in drug withdrawal

Chronic use of addictive substances such as tobacco or opioids changes neuroplasticity and brain homeostasis and leads to physiological dependence (Chartoff and Carlezon, 2014; Koob, 2006). This dependence manifests with symptoms of withdrawal when use of the drug is abruptly stopped. Craving responses during withdrawal result in impairment of the normal stress response and other signaling mechanisms in the brain, leading to a state of anxiety and internal stress that eventually triggers relapse and use of the drug mainly to relieve this negative-affect state (Koob and Volkow, 2016; Wand, 2008). Because of the high concentration of nAChRs and opioid receptors in the habenular circuit, particularly in the MHb and IPN, nicotine and opioids have potent effects in habenula. As mentioned in sections 2.1-2.3, the MHb-IPN pathway densely expresses several nAChR subunits (Antolin-Fontes et al., 2015; Lee et al., 2019; Shih et al., 2014). Most nAChR subunits enriched in the MHb-IPN are encoded in two gene clusters: the CHRNA3-CHRNB4-CHRNA5 locus encoding a3, b4, and a5 nAChR subunits, and the CHRNA6-CHRNB3 gene locus encoding a6 and b3. GWAS have linked single nucleotide polymorphism (SNP) variants in both gene clusters to increased dependency on nicotine (Bierut, 2007; Greenbaum and Lerer, 2009; Saccone et al., 2007; Thorgeirsson et al., 2008). In humans, nicotine dependence is characterized by the difficulty to quit smoking or stop consuming nicotine products. Abrupt discontinuation of nicotine after prolonged use triggers a withdrawal syndrome characterized by somatic symptoms, including slow heart rate and gastrointestinal discomfort; affective signs of anxiety and depressed mood; and cognitive deficits such as difficulty in concentrating (Dani and De Biasi, 2013; Molas et al., 2017a). Rodents show quite similar signs of nicotine withdrawal. These are mediated through nAChR in the Hb-IPN axis – studies show that blockading with nAChRs antagonists in the MHb or IPN is sufficient to precipitate withdrawal (Salas et al., 2009; Zhao-Shea et al., 2015). nAChR knock-out (KO) mice studies have shown that specific nAChR subunits differentially regulate somatic and affective signs of withdrawal that can be measured with specific behavioral tests (Figure 2). Detailed nAChR mechanisms that mediate nicotine withdrawal symptoms are reviewed in several articles (Antolin-Fontes et al., 2015; Chellian et al., 2021; Cohen and George, 2013; Jackson et al., 2009; Jackson et al., 2015; Lee et al., 2019; Markou, 2008; McLaughlin et al., 2015; Molas et al., 2017a; Paolini and De Biasi, 2011; Salas et al., 2009; Wills and Kenny, 2021).

Figure 2. Habenula-related behaviors:

Behavioral tests employed to assess the motivational and emotional behaviors regulated by the habenula in mice and rats.

(A-C) Drug addiction is a sign of dysregulated motivation, evidenced by intense drug craving and compulsive drug-seeking behavior, and can be measured using the (A) voluntary two-bottle drinking paradigm and the (B) intravenous or intracranial self-administration. (C) The conditioned place preference test is used to evaluate whether the mouse has developed preference for the chamber where it has received nicotine after 4-5 days of conditioning.

(E-H) Anxiety-like symptoms in rodents can be evaluated using the (D) light-dark room (preference for the dark compartment), (E) elevated plus maze (preference for the enclosed arms), and (H) open field (preference for the periphery). Depressive-like behaviors can be monitored in the (F) forced swim test (FST) and the (G) sucrose preference test (SPT). (F) The FST is based on the assumption that immobility reflects a measure of behavioral despair (time not swimming). This test involves the exposure of the animals to stress, which was shown to have a role in the tendency for major depression. (G) The SPT is based on a two-bottle choice paradigm. A reduction in the sucrose preference ratio in experimental relative to control mice is indicative of anhedonia. Anhedonia is the inability to experience pleasure from rewarding or enjoyable activities and is a core symptom of depression in humans.

(I-J) Fear can be measured in the (I) contextual fear conditioning chamber where rodents learn to predict aversive events by pairing of an initially neutral stimulus (light signal and sound cue) with an aversive fear eliciting stimulus (electric foot shock). (J) The fear-potentiated startle reflex test is a paradigm in which amplitude of a simple reflex is increased when presented with a cue that has been previously paired with an aversive stimulus. A greater general startle reflex is also associated with greater anxiety levels.

(K-L) Social and novelty exploration behaviors can be measured in the three-chamber test. This test measures (K) “sociability” as the preference to spend time with another mouse, as compared to time spent alone or with an object in an identical but empty chamber, and (L) “social novelty”, or “object novelty”, as the propensity to spend time with a previously novel (never-before-met) mouse/object rather than with a familiar mouse/object.

(H-N) Locomotor activity tests include the (H) open field test which assess general locomotor activity levels, anxiety, and willingness to explore in rodents; (M) the rotarod that measures motor coordination skills; and (N) the running wheel that measures general activity, energy balance, and reward in rodents.

Nicotine withdrawal also occurs when the nAChR system interacts with other neuromodulatory systems. Among these, the tachikynins (also known as neurokinins) exert an important function in the habenular circuit. Nicotine and substance P differentially regulate neuronal excitability in subdivisions of the MHb (Dao et al., 2014; Lee and Huang, 1988). Blockade of Tac1R and Tac3R in the MHb reduces nicotine-mediated excitability of MHb neurons and precipitates nicotine withdrawal in mice (Dao et al., 2014). Another neuromodulatory system involved in nicotine withdrawal and chronic use is the corticotropin-releasing hormone (CRH) system, consisting of the neuropeptide CRH and its receptors CRHR1 and CRHR2. CRH plays a critical role in stress through activation of the hypothalamic-pituitary-adrenal axis (HPA axis) (Sukhareva, 2021). CRH is primarily synthesized in the hypothalamus but is also quite widespread in extrahypothalamic brain structures (Riad et al., 2022). Nicotine withdrawal increases CRH in the central nucleus of the amygdala (George et al., 2007) and in the VTA (Grieder et al., 2014) and also increases expression of CRHR1 in the IPN (Zhao-Shea et al., 2015). These studies have delineated a mechanism, through recruitment of cholinergic and neuromodulatory peptide systems, that underlines the anxiety-like symptoms of withdrawal.

3.1.3. Role of the habenula in drug intake

As defined by Koob and Volkow, “Drug addiction is a chronically relapsing disorder, characterized by compulsion to seek and take the drug, loss of control in limiting intake, and emergence of a negative emotional state (e.g., dysphoria, anxiety, irritability) when access to the drug is prevented” (Koob and Volkow, 2016). In rodents, drug intake can be measured with intravenous or intracranial self-administration experiments or with two-bottle choice tests (Figure 2). Excessive nicotine intake observed after abstinence (also known as binge intoxication) activates the nAChR and CRH systems. a5 nAChR KO mice (Fowler et al., 2011), b4 KO (Husson et al., 2020), and a3 nAChR hypomorphic mice (Elayouby et al., 2021) show increased nicotine intake while b4-overexpressing mice show decreased intake and increased aversion to nicotine (Frahm et al., 2011; Husson et al., 2020). These results align with findings in humans that allelic variations in the CHRNA5-CHRNA3-CHRNB4-gene cluster increase the risk of tobacco dependence. The SNP D398N (rs16969968) in CHRNA5 not only predisposes to nicotine dependency but also directly contributes to Chronic Obstructive Pulmonary Disease (COPD)-like lesions in the lungs due to oxidative stress and injury (Routhier et al., 2021). COPD has been associated with allelic variations in the Hedgehog-interacting protein (HHIP). In the brain, HHIP is very highly expressed in MHb^ChAT neurons, which receive synaptic inputs from septal neurons that synthesize and secrete Hedgehog ligands (Caligiuri et al., 2022). CRISPR/Cas9-mediated genomic cleavage of HHIP in the habenula of mice increases intravenous self-administration behavior, suggesting that the vulnerability to develop smoking-related lung diseases partially depends on the action of nicotine in the habenular circuit (Caligiuri et al., 2022). The rs16969968 variant has also been associated with addiction to opioids, and opioid users display increased habenular connectivity (Curtis et al., 2017). Several lines of evidence suggest a functional interaction of the cholinergic and opioidergic systems in the development and maintenance of drug-seeking behavior (Brynildsen and Blendy, 2021; Custodio et al., 2022). Mice with deletion of MOR in b4-nAChRs-expressing MHb neurons show attenuated physical signs of withdrawal, precipitated by either antagonists of MOR (naloxone) or of nAChR (mecamylamine) (Boulos et al., 2020). In addition, habenular neurons expressing MOR contribute to despair-like behavior and anxiety, supporting the fact that opioid-mediated inhibition of these neurons may relieve negative affect, and thus favor hedonic homeostasis and addiction (Bailly et al., 2022). Human and rodent studies indicate a bidirectional relationship between nicotine and opioid systems that is critical for the development and establishment of drug dependence (Almeida et al., 2000; Brynildsen and Blendy, 2021; Krause et al., 2018; Trigo et al., 2009; Young-Wolff et al., 2017).

Besides cholinergic, opioidergic, tachykininergic, and CRH neurotransmitter systems, other signaling molecules highly enriched in the Hb-IPN have been found to be critically important for nicotine and opioid self-administration. Activation of the receptor for glucagon-like peptide-1 (GLP-1) in the IPN abolishes nicotine reward and decreases nicotine intake (Tuesta et al., 2017), while inhibition of the signaling mediated by the diabetes-associated transcription factor Tcf7l2 increases nicotine intake in rodents and alters the circulating levels of glucagon and insulin, leading to dysregulation of blood glucose levels (Duncan et al., 2019). These studies suggest that the stimulatory actions of nicotine in the Hb-IPN circuit are linked to the diabetes-promoting actions of nicotine. This important finding by the Kenny group is reviewed separately in this issue.

Two orphan G-protein coupled receptors (GPCRs) shown to be expressed in the habenula, GPR139 and GPR151, have also been linked to substance abuse. GPR139 was identified for its anti-opioid activity in a forward genetics screen in Caenorhabditis elegans (Wang et al., 2019). GPR139 is coexpressed with MOR in opioid-sensitive brain circuits, including the habenula and striatum (Matsuo et al., 2005). GPR139 binds to MOR and opposes it action by signaling to alternate heterotrimeric guanine nucleotide–binding (Gq/G11) proteins rather than the ones employed by MORs (Stoveken et al., 2020). Deletion of Gpr139 in mice enhances opioid-induced inhibition of neuronal firing in the MHb and modulates morphine-induced analgesia, reward, and withdrawal (Wang et al., 2019). GPR151 is expressed predominantly in the habenula (vMHN and LHb) (Broms et al., 2015), signals via the inhibitory guanine nucleotide binding protein Ga_oi, and alters habenular synaptic transmission (Antolin-Fontes et al., 2020). Deletion of Gpr151 in mice decreases short-term synaptic plasticity (paired-pulse facilitation) in response to nicotine and opioids, and the changes in locomotor activity induced by nicotine and opioids are much less pronounced in Gpr151-KO (Antolin-Fontes et al., 2019). Gpr151-KO mice self-administered more intravenous nicotine infusions than wild-type mice, particularly when a higher unit dose on the descending portion of the dose-response curve was available (Antolin-Fontes et al., 2020; Wills et al., 2022). Given the converging, overwhelming evidence pointing to a critical role of the Hb-IPN in drug addiction, active compounds that could specifically target these molecules exclusively or predominantly expressed in the habenular circuit would be very beneficial to rein in addiction.

3.2. Role of the habenula in circadian rhythms and depression

Early lesion studies of the habenula or transection of the fr showed disruption of sleep (Haun et al., 1992; Valjakka et al., 1998), hyperactivity (Lee and Huang, 1988; Murphy et al., 1996; Nielson and McIver, 1966; Thornton et al., 1994), and disruption of the length of circadian rest and activity periods (Paul et al., 2011). This unrecognized function of the habenula as an alternate circadian clock besides the suprachiasmatic nucleus (SCN) has been linked to its connectivity to the SCN (Bano-Otalora and Piggins, 2017; Zhang et al., 2009) and to the circadian variation in pacemaking firing rate and resting membrane potential (Gorlich et al., 2013; Guilding et al., 2010; Sakhi et al., 2014). When isolated from the SCN, the habenula sustains rhythms in clock gene expression and neuronal activity, with the LHb expressing more robust rhythms than the adjacent MHb (Young et al., 2021). Importantly, habenular neurons respond to putative SCN output factors as well as to retinal illumination conveyed to the perihabenular area (Young et al., 2021). Rodent studies have suggested that light can influence neuronal activity in the LHb, accompanied by changes in depressive-like behaviors (Vazquez et al., 2021; Zhao and Rusak, 2005). Recently a di-synaptic visual circuit connecting the retina and LHb has been shown to underly the antidepressant effects of light therapy in the mouse (Huang et al., 2019). fMRI studies in humans have confirmed that the habenula responds to light with a decrease in activation when a change in luminance occurs. This strength is modulated in a circadian manner, being more pronounced in the morning (Kaiser et al., 2019). Notably, disruptions in circadian rhythms are well known to elevate the risk of developing mental health disorders, including depression (Young et al., 2021).

The key role of the LHb in the pathophysiology of depression has been established by several important studies (Cui et al., 2018; Fan et al., 2023; Li et al., 2013; Lin et al., 2022; Nuno-Perez et al., 2021; Yang et al., 2018b; Zheng et al., 2022) and has been the focus of many recent reviews (Hu et al., 2020; Yang et al., 2018c; Zanos and Gould, 2018; Zhang et al., 2022b). Overactivity in the LHb during depressed states drives changes in midbrain activity that suggest that the LHb could be a promising novel target for DBS procedures. As mentioned at the beginning of this section, DBS treatments have resulted in sustained full remission of depressive symptoms in cases of intractable major depressive disorder (MDD) (Mayberg et al., 2005; Sartorius and Henn, 2007). In rodents, LHb hyperactivity during depression has been linked to disfunction of the beta calcium/calmodulin-dependent protein kinase type II (βCaMKII) (Li et al., 2013). This kinase was significantly upregulated in the LHb of animal models of depression and downregulated by antidepressants. Manipulation of βCaMKII levels in the LHb affected both behavioral despair and anhedonia (Figure 2G and 2F), suggesting that LHb controls multiple aspects of depressive symptoms (Li et al., 2013). Later studies showed that burst-evoking photostimulation of LHb drives behavioral despair and anhedonia and that ketamine is able to quickly improve mood in rodents by blocking NMDAR-dependent bursting activity of LHb neurons, thus disinhibiting downstream monoaminergic reward centers (Cui et al., 2018; Sivalingam et al., 2020; Yang et al., 2018b). This research has shed light on how ketamine may work to quickly alleviate symptoms of depression, which has been demonstrated in clinical studies where a single sub-anesthetic dose of ketamine produces fast and sustained antidepressant effects (Zanos et al., 2018). Novel enantiomers that can be delivered intranasally are being tested for their efficacy in the management of MDD (Vazquez et al., 2022).

3.3. Role of the habenula in fear responses, aggression, and social behavior

There is a relatively well-developed literature for the role of the habenula in social reward and fear responses, within the context of aggression, defeat, novelty, and familiarity. Early studies in zebrafish established a critical role for the habenula in fear responses (Agetsuma et al., 2010) and identified a monosynaptic habenulo-raphe serotoninergic circuit for assessing expected danger and adaptively avoiding potential hazards (Amo et al., 2014). Manipulations of habenular specific genes have also been linked to impaired fear responses (Choi et al., 2018; Roy et al., 2021; Sivalingam et al., 2020).

Subsequent studies in mice have also revealed an important role for the MHb-IPN synapses in fear suppression. These studies have shown that MHb neurons regulate aversive memories and fear extinction by modulating cholinergic, but not glutamatergic, neurotransmission. This modulation is exerted by two types of presynaptic receptors present at axonal terminals of MHb neurons in the IPN: cannabinoid type 1 receptor (CB1) (Melani et al., 2019; Soria-Gomez et al., 2015) and GABA_B receptors (Koppensteiner et al., 2017; Zhang et al., 2016) (Figure 3D). The LHb has a more general effect on fear than the MHb (Sachella et al., 2022), as it has been shown to be required for the formation of both independent contextual and cued fear memories.

In recent years, studies in zebrafish have also implicated the habenula in social conflict resolution and aggression to establish social hierarchy (Chou et al., 2016). In mice, the LHb has been shown to modulate aggression reward (Golden et al., 2016) and aggressive behavior (Flanigan et al., 2020). LHb glutamatergic neurons projecting to the dorsal raphe nucleus promote aggressive arousal in mice. Interestingly, the major population of DRN cells that receive input from the LHb were non-serotonergic (Takahashi et al., 2022).

Novelty and familiarity are regulated by the MHb (Molas et al., 2017b). Response to novelty has been related to the vulnerability to develop addiction (Piazza et al., 1989) and is dysregulated in some neuropsychiatric disorders. KO mice of the b4 nAChR subunit displayed a significant increase in novelty-induced activity in the open field (Husson et al., 2020), and mutant mice with specific cell-ablation of cholinergic MHb neurons were maladapted when repeatedly exposed to new environments (Kobayashi et al., 2013). Exposure to an inescapable environment can elicit either an escape or an exploratory behavior; novelty-preference could be a measure of vulnerability to addiction (Belin et al., 2016), since exploration is driven by a rewarding dopamine-mediated response to novel stimuli (Molas et al., 2017b). Sociability in mice is tested by measuring the preference toward a familiar mouse versus a novel mouse or a mouse versus an inanimate object using a three-chamber box (Figure 2K and 2L). These studies showed different patterns of c-fos activation in the IPN, which became activated as novel stimuli become familiar after multiple exposures (Molas et al., 2017b). Abnormal social behaviors have also been observed in b4 nAChR KO mice (Salas et al., 2013), in Gpr151 KO mice, and in mice with partial deletion of the MOR in MHb (Allain et al., 2022). As the social brain is becoming an area of intense study, it is expected that more research on the subject will follow.

4. Electrophysiological properties of the habenular circuit: conserved and novel synaptic mechanisms

In this section, we will discuss some electrophysiological properties of the habenular circuit, including its distinctive neuronal responses depending on the receptors expressed in different subnuclei of the LHb, MHb and IPN (Figure 1), the spontaneous pacemaking and tonic firing of LHb and MHb neurons, specific mechanisms for long-term synaptic plasticity and potentiation, and presynaptic and postsynaptic signaling processes.

4.1. Electrophysiological studies in the LHb and MHb:

As previously mentioned, both the LHb and MHb express nAChRs; notably, the MHb, with 90-100% of its neurons expressing nAChRs subunits (Shih et al., 2014), has one of the highest nAChR densities in the brain. Nicotine application ex vivo in brain slices or in vivo elevates the firing rate of neurons in the LHb, as detected by whole cell brain slice recordings or extracellular single-unit recordings (Pierucci et al., 2022; Zuo et al., 2016). Nicotine also induces inward currents in LHb neurons, enhancing both inhibitory postsynaptic currents (IPSCs) and excitatory postsynaptic currents (EPSCs), suggesting that both GABAergic and glutamatergic transmissions are increased by nicotine in the LHb via nAChRs. The effects of nicotine on inhibitory and excitatory transmissions of LHb neurons are mediated by α6*-nAChRs and α4β2-nAChRs, respectively (Zuo et al., 2016). Not only ionotropic nAChRs but also metabotropic, muscarinic AChRs (mAChRs) regulate the activity of LHb neurons. Activation of M₂-mAChRs inhibits both excitatory and inhibitory transmission to the LHb, shifting the excitatory and inhibitory balance towards net inhibition. Inhibition of M₂-mAChRs in the LHb enhances impulsive cocaine-seeking behavior, indicating that M₂-mAChRs inhibit the activity of the LHb and play a role in reducing the response to cocaine (Wolfe et al., 2022). Research on cholinergic receptors in the LHb has just begun and future studies are needed to further investigate the properties and functions of LHb-mediated cholinergic pathways.

Since the MHb is one of the brain areas with the highest expression of nAChRs, it plays an important role in nicotine-related behaviors, such as aversion and withdrawal, as previously discussed in section 3 (Antolin-Fontes et al., 2015; Gorlich et al., 2013; Lee et al., 2019; Shih et al., 2014). Both chronic passive and volitional nicotine administration upregulate nAChRs in the brains of laboratory animals and human smokers (Breese et al., 1997; Marks et al., 1983; Nashmi et al., 2007; Staley et al., 2006). Electrophysiological studies have found that nicotine self-administration in rats induces nAChR functional upregulation in the MHb, increasing ACh-evoked responses both at the soma and in dendrites (Jin et al., 2020).

As we discussed in section 1 and 2, the MHb is composed of distinct subregions with different neurotransmitter release and neuronal projections (Figure 1). Neurons in the lateral part of the ventral MHb (MHbVL) fire spontaneously at ~4 Hz, while neurons in the dorsal region of the central part of the MHb (MHbD) and in the superior division of the MHb (MHbS) show irregular firing at 1 to 2 Hz (Gorlich et al., 2013; Lee et al., 2018). Nicotine application increases tonic firing in the MHbD and MHbVL but does not affect tonic firing in the MHbS (Dao et al., 2014; Gorlich et al., 2013; Lee et al., 2018). Acute nicotine application also evokes inward currents at −60 mV in the MHbD and the MHbVL, which are not observed in the MHbS. On the other hand, Substance P application elevates tonic firing and induces inward currents at −60 mV selectively in the MHbVL. The effects of Substance P in the MHbVL are facilitated after nicotine withdrawal (Lee et al., 2018). More than 80% of the cholinergic neurons in the ventral MHb show rhythmic tonic firing which is generated by hyperpolarization-activated cyclic nucleotide-gated (HCN) pacemaker channels, L-Type Ca²⁺ channels, and BK channels (Gorlich et al., 2013; Lee et al., 2018; Shih et al., 2015). Extracellular single-unit recordings in behaving rodents found that tonic firing rates in the MHb are higher during the day compared to the ones during the night, suggesting a potent role of the MHb in linking circadian and motivational pathways in the brain as previously mentioned in section 3.2 (Lee et al., 2019; Zhao and Rusak, 2005). Application of the HCN channel antagonist, ZD7288, to the MHb in acute brain slices strongly decreases tonic firing, and direct infusion of ZD7288 into the MHb of behaving mice elevates both somatic and affective signs of nicotine withdrawal in mice that have not been previously exposed to nicotine (Gorlich et al., 2013). Nicotine application increases tonic firing of neurons in the MHb through α3β4* nAChRs, and nicotine withdrawal potentiates the response to nicotine (Gorlich et al., 2013; Lee et al., 2018; Shih et al., 2015).

As previously described (Figure 1), the MHb consists of many subregions and each subregion shows distinct expression levels of nAChR subunits (Antolin-Fontes et al., 2015; Wills and Kenny, 2021). It has been also reported that nAChRs in different MHb subregions exhibit distinct responses when they are activated (Shih et al., 2014; Shih et al., 2015). Neurons in the MHbVL and in the inferior part of the ventral MHb (MHbVI) respond to ACh, showing inward currents. ACh-induced currents are partially mediated by α3β4* nAChRs both in MHbVL and MHbVI, since the α3β4* nAChRs inhibitor, SR16584 decreases ACh-evoked currents. Inhibition of β4* nAChRs or α6* nAChRs by DhβE or αCtxMⅡ reduces ACh-induced currents selectively in the MHbVI, but not in the MHbVL, suggesting that functional β4* and α6* nAChRs are expressed in the MHbVI, but not in the MHbVL. Prolonged nicotine application increases firing rate slower but does not attenuate the effect in the MHbVI compared to the MHbVL where nicotine application shows rapid but short elevation of tonic firing. Acute nicotine application decreases ACh-induced currents selectively in MHbVL, while chronic nicotine treatment increases ACh-induced currents in MHbVI, but not in MHbVL. Chronic nicotine treatment elevates tonic firing which is not affected by acute nicotine treatment (Shih et al., 2015).

4.2. Electrophysiological studies in the MHb-IPN:

The MHb-IPN circuit is a major cholinergic circuit in the brain (Antolin-Fontes et al., 2015; Koppensteiner et al., 2017; Ren et al., 2011; Zhang et al., 2016). Cholinergic terminals from MHb neurons in the IPN coexpress the synaptic vesicular transporters for glutamate (VGLUT1) and ACh (VAChT), indicating that these neurons corelease glutamate and ACh (Frahm et al., 2015; Ren et al., 2011). Channelrhodopsin-2 (ChR2) expressed in cholinergic MHb neurons of ChAT-ChR2-EYFP mice release glutamate by brief photostimulation and release ACh by tetanic photostimulation to the IPN (Ren et al., 2011). Mice with conditional deletion of ChAT in the habenula (ChAT-cKO mice) are insensitive to the rewarding effects of nicotine and show no nicotine withdrawal. In the MHb-IPN circuit, compared to wild type mice, ChAT-cKO mice exhibit smaller miniature excitatory postsynaptic currents (mEPSCs) amplitude and nicotine fails to elevate mEPSCs frequency. These electrophysiological observations suggest that ACh increases the content of glutamate in synaptic vesicles, a mechanism termed vesicular synergy (Figure 3B) and required for presynaptic nAChR-mediated synaptic facilitation (Frahm et al., 2015).

Chronic nicotine administration mediates functional upregulation of nAChR in the MHb-IPN circuit (Arvin et al., 2019; Pang et al., 2016), increases tonic firing and glutamate release, enhances nAChR function on proximal axonal membranes in the MHb and IPN (Arvin et al., 2019; Salas et al., 2009), and enhances cholinergic signaling and cell excitability in the MHb-IPN circuit (Arvin et al., 2019). In the MHb-IPN circuit, both nicotine and ACh facilitate glutamatergic neurotransmission by increasing Ca²⁺ influx through presynaptic nAChRs (Frahm et al., 2015) (Figure 3C). Among the nAChRs subunits, it is reported that presynaptic α7 subunit enhances, while presynaptic α5 subunit disrupts, the release of glutamate in the MHb-IPN circuit (Girod et al., 2000; McGehee et al., 1995). It would be interesting to evaluate the role of other presynaptic nAChR subunits on facilitation of glutamatergic neurotransmission in the MHb-IPN circuit.

GABA_B receptors are highly expressed along the axonal projections of MHb cholinergic neurons in the IPN (Bhandari et al., 2021; Margeta-Mitrovic et al., 1999; Zhang et al., 2016). Activation of GABA_B receptors has been known to be inhibitory by decreasing neurotransmitter release at presynaptic sites and hyperpolarizing postsynaptic neurons, including in the dorsal MHb-lateral IPN (LIPN) circuit (Dutar and Nicoll, 1988; Gassmann and Bettler, 2012; Melani et al., 2019; Newberry and Nicoll, 1984). However, GABA_B receptors in the ventral MHb are excitatory receptors that induce presynaptic excitation by elevating Ca₂₊ influx through the Ca_V2.3 channels (Bhandari et al., 2021; Koppensteiner et al., 2017; Zhang et al., 2016). Electrophysiological experiments conducted in the MHb-IPN circuit of ChAT-ChR2-EYFP mice found that the GABA_B receptor agonist baclofen increases glutamate-, ACh-, and neurokinin B (NKB)-mediated EPSCs amplitudes. The effect of baclofen is reversely blocked by a Ca_V2.3-containing R-type Ca²⁺ channel blocker (Zhang et al., 2016). These data suggest that GABA_B receptors in the MHb-IPN circuit elevate the corelease of multiple neurotransmitters, such as glutamate, ACh, and NKB through Ca_V2.3-containing R-type Ca²⁺ channels (Bhandari et al., 2021; Koppensteiner et al., 2017; Zhang et al., 2016), as depicted in Figure 3D.

Application of single high-frequency stimulation (HFS) to the MHb inputs in the IPN induces long-term potentiation (LTP) in the IPN in a presynaptic manner by increasing glutamate release. Experiments with antagonists revealed that induction of LTP in the MHb-IPN circuit requires Ca²⁺-permeable AMPA receptors (CPARs) (Figure 3A) and GABA_B receptors. HFS to the MHb neuronal terminals in the IPN releases glutamate and activates CPARs on GABAergic neurons in the IPN. Ensuing release of GABA from neurons in the IPN acts on presynaptic GABA_B receptors and induces a long-lasting enhancement of glutamate release (Koppensteiner et al., 2017; Zhang et al., 2016). This type of LTP in the MHb-IPN is impaired after fear conditioning and restored after fear extinction, suggesting that fear learning and inhibition regulate the mechanism underlying activity-dependent LTP in the MHb-IPN circuit (Koppensteiner et al., 2017). The MHb-IPN circuit expresses the Ca_V2.3-containing R-type Ca²⁺ channel and the auxiliary GABA_B receptor subunits and K⁺-channel tetramerization domain-containing proteins (KCTDs). It has been known that there are 4 KCTD subunits, KCTD8, 12, 12b, and 16, which bind GABA_B receptors and regulate their kinetics (Bhandari et al., 2021; Fritzius et al., 2017; Schwenk et al., 2010; Zhang et al., 2016). In the ventral MHb-rostral IPN circuit, presynaptic KCTD8 and KCTD12b bind Ca_V2.3 channels that mediate neurotransmitter release. Electrophysiological experiments on paired-pulse ratio (PPR) revealed that overexpression of KCTD8 and KCTD12b enhances and diminishes release probability of glutamate in the MHb-IPN circuit, respectively (Bhandari et al., 2021). Another study found that deletion of both KCTD8 and KCTD12 blocks HFS-induced LTP in the MHb-IPN circuit, decreasing GABA_B receptors within the MHb terminals (Ren et al., 2022). Calcium bursts from cholinergic terminals trigger excitatory currents in IPN neurons that release GABA and result in stereotypic activation of presynaptic GABA_B receptors and trans-inhibition of cholinergic and peptidergic non-cholinergic MHb axon terminals in the IPN (Zaupa et al., 2021).

MHb^Tac1 peptidergic neurons release glutamate and glycine to the LIPN. Glycine receptors interrupt HFS-induced presynaptic LTP in the MHbD-IPL circuit as blocking glycine receptors unmasks LTP in the MHbD-IPL circuit. On the other hand, Substance P application enhances LTP in the MHbD-IPL circuit in a presynaptic manner. Substance P-induced LTP requires activation of the cannabinoid receptor 1 (CB₁R) that decreases GABA release, but not glutamate release. This CB₁R-mediated inhibition of GABA_B receptor activity supports Substance P to elevate glutamate release for a long time to induce LTP in the MHbD-IPL circuit (Melani et al., 2019). Glycine receptors have been also identified in the habenula. The unconventional N-methyl-d-aspartate (NMDA) receptor subunits GluN3A and GluN3B have been shown to associate with the glycine-binding subunit GluN1 to form GluN1/GluN3A receptors which are operational in MHb neurons and generate excitatory conductances purely activated by glycine (Otsu et al., 2019).

4.3. Electrophysiological studies in the IPN

Activation of the IPN by the MHb is required for expression of nicotine aversion and withdrawal (Fowler et al., 2011; Frahm et al., 2011; Zhao-Shea et al., 2013). Neurons in the IPN are predominantly GABAergic and GABAergic neurons in the IPN show increased glutamate-mediated spontaneous responses after nicotine withdrawal (Ables et al., 2017; Zhao-Shea et al., 2013). Chronic nicotine administration upregulates β4* nAChRs in somatostatin (Sst) interneurons in the IPN and enhances nicotine-induced inward responses in IPN Sst neurons, which are attenuated by a β4* nAChR antagonist. Infusion of the β4* nAChR antagonist in the IPN induces nicotine withdrawal symptoms in naïve and nicotine-treated mice, suggesting that β4* nAChRs in the IPN blocks nicotine withdrawal symptoms (Zhao-Shea et al., 2013).

Two populations of α5* nAChR-expressing GABAergic neurons have been identified in the IPN: α5-^Amigo1 and α5-^Epyc neurons. Chronic nicotine treatment increases nitric oxide synthase (NOS1) and Sst expression in α5-^Amigo1 neurons in the IPN. Both nitric oxide (NO) and Sst decrease glutamate-mediated postsynaptic currents in the cholinergic MHb-IPN circuit (Figure 3E and 3F). Inhibition of α5-^Amigo1 neurons or NO synthesis in the IPN eliminates nicotine reward, indicating that retrograde inhibition of the MHb-IPN circuit by α5-^Amigo1 neurons regulates nicotine preference (Ables et al., 2017).

5. Concluding remarks

Just over a decade ago, groundbreaking studies conducted in monkeys revealed a significant activation of the habenula when a reward that was expected was missing (Matsumoto and Hikosaka, 2007). In addition human association studies linked nicotine and opioid dependence to variations in the CHRNA5-CHRNA3-CHRNB4 gene cluster (Bierut et al., 2008; Erlich et al., 2010). Further studies in rodents were conducted on the encoded a5 and β4 nAChRs subunits which are present in high concentrations in the habenula-IPN pathway. These mechanistic studies revealed the critical role of these subunits in drug aversion and consumption (Fowler et al., 2011; Frahm et al., 2011). Collectively, these studies shed light on the previously unknown role of the habenula in balancing responses to rewarding and aversive stimuli. It is evident from the wealth of studies that we have reviewed here that the habenula responds to both acute, novel stimuli and adaptations to chronic stimuli which may be key for neuropsychiatric diseases such as addiction, depression, and anxiety. Further, the habenular circuit modulates sleep, feeding and motivated behaviors, processes disrupted in neuropsychiatric disease.

Because of its extremely small size, efforts to isolate and manipulate the habenula are technically challenging but also offer a unique opportunity for very targeted molecular therapies. At present there are very few molecular transcriptomics studies in human habenula. Existing data consist of microarray chip datasets (Allen Brain studies described in (Le Foll and French, 2018), and suicide postmortem brain (Kim et al., 2022). Current efforts using high-throughput single-cell transcriptional profiling and more in-depth and precise methodologies (Xu et al., 2018) will provide a clearer representation of cell types and molecules in the human habenula, perhaps revealing important species differences (Wallace et al., 2020).

Given the recent understanding of the habenula in mood and addiction, novel therapies are being developed that target habenular circuitry. These include bilateral habenula deep brain stimulation (Zhang et al., 2022a) and ketamine enantiomers delivered in a nasal spray (https://www.fda.gov) to treat treatment-resistant depression. Importantly amid an unprecedented drug addiction crisis, habenular therapies based on active compounds targeting specific GPCRs enriched in the habenula, including GPR151 and GPR139, could prove critically beneficial. In a wider context, it is important to note that vaccines against nicotine, cocaine, and opioids could offer some relief. These vaccines consist of nicotine or cocaine-based haptens conjugated to carrier proteins, and they are in phase III clinical trials (Pravetoni and Comer, 2019). Vaccines to treat opioid use disorders to reduce incidence of opioid overdose are also being developed and have shown promise in reducing fentanyl versus food choice in monkeys (Townsend et al., 2021).

Because addiction therapies aiming at the mesolimbic dopaminergic reward system have proven to be extremely difficult, the delineation of the habenula as the “other half” of the reward circuitry opens new avenues to identify new strategies to treat compulsive drug-taking and mood disorders.