Concerning neuromodulation as treatment of neurological and neuropsychiatric disorder: Insights gained from selective targeting of the subthalamic nucleus, para -subthalamic nucleus and zona incerta in rodents

Neuromodulation such as deep brain stimulation (DBS) is advancing as a clinical intervention in several neurological and neuropsychiatric disorders, including Parkinson ’ s disease, dystonia, tremor


Introduction
"Life is like riding a bicycle -to keep balance, you need to keep moving" (Albert Einstein).
Many brain systems work together to enable us to move purposefully in relation to current context, and should any of these systems fail to function, clinical symptoms will guide towards diagnosis.One highly conserved brain system critical to achieve voluntary (willed, intended) movement is the one referred to as the basal ganglia, and one brain nucleus which is sometimes included in the concept of the basal ganglia, and other times referred to as an accessory to the basal ganglia, is the subthalamic nucleus (STN) (Stephenson-Jones et al., 2011) (Fig. 1).The STN and the basal ganglia together form an intracerebral loop, not directly contacting any muscles at all, but serving upon cortical command to regulate movement (Gerfen, 1992;Van Der Kooy and Hattori, 1980).
Knowledge of the importance of the STN has primarily been gained from neurological disorders in which the normal function of the STN has been disturbed, leading to failure in regulation of movement parameters.These disorders include Parkinson's disease (PD), the second most common neurodegenerative disorder (Bergman and Deuschl, 2002;McGregor and Nelson, 2019) in which STN activity is abnormal (Di Giulio et al., 2019;Hamani, 2004;Levy et al., 2000;Moro et al., 2010), and also disorders in which STN neurons degenerate, such as supranuclear palsy (PSP) (Albers et al., 1999;Hardman et al., 1997;Lubarsky and Juncos, 2008;Sakai and Yamada, 2011) and Huntington's disease (Atherton et al., 2016;Eidelberg and Surmeier, 2011;Sharma and Deogaonkar, 2015).
The role of the STN in the many different aspects of motor function is still under intense exploration, both by studies of patients and of experimental animals, primarily but not exclusively, rodents and nonhuman primates.Long believed to act as a "simple relay nucleus" to the thalamus, additional roles of the STN -beyond complex motor functions -have also been revealed in both rodents and primates.These include limbic and cognitive functions via circuits parallel to motor circuitry, including impulsivity, attention, reward-related functions, and more (Bastin et al., 2014;Baunez et al., 2002Baunez et al., , 2005;;Espinosa-Parrilla et al., 2013;Isoda and Hikosaka, 2008;Lardeux et al., 2013;Mosher et al., 2021;Pasquereau and Turner, 2017;Pelloux et al., 2018;Rouaud et al., 2010;Temel et al., 2005;Weintraub and Zaghloul, 2013).The various functions implicating the STN will be discussed in detail below, in the context of experimental approaches in rodents that have contributed to today's knowledge of neurobiological underpinnings of STN neurocircuitry in brain function and brain disorder.
The literature covering studies of the STN in human patients, including case reports, is vast and will not be addressed in any detail here.Instead, data from studies of humans will be outlined as examples to visualize the importance of increasing current knowledge of the STN and immediately surrounding structures, and how experimental studies of rodents can be used to forward such knowledge.
Beyond the multifunctionality of the STN itself, this clinically important brain nucleus is located in a functionally diverse region, neighboring zona incerta (ZI) and para-STN (pSTN) (Fig. 2).Further increasing the complexity of this brain region, these structures are surrounded by substantial white matter, passing fibers reaching vastly within the brain, including the internal capsule, and medial forebrain bundle (MFB), the latter transversing between the STN and ZI on its route from the midbrain to forebrain target areas (Franklin and Paxinos, 2008;Paxinos, 2009;Paxinos and Watson, 2014;Swanson, 2004).
In addition to the STN, ZI is also an important brain target area for DBS (ZI-DBS), primarily as intervention of essential tremor and PD (Blomstedt et al., 2011(Blomstedt et al., , 2012(Blomstedt et al., , 2018;;De Marco et al., 2020;Ossowska, 2020;Sandvik et al., 2012).However, attention directed at the pathophysiology of OCD was directed at both the STN and ZI already a couple of decades ago, described in the seminal report "Compulsions, Parkinson's disease, and stimulation" (Mallet et al., 2002).In this case study, it was described how DBS electrodes implanted in the anteromedial STN or between the anteromedial STN and ZI and/or in anterior ZI in two patients suffering from severe PD and with a long history of OCD, not only improved parkinsonian disability but also led to disappearance of compulsions and amelioration of obsessions in both patients.ZI, an area initially described as a "zone of incertainty" (hence its name), is currently under intense exploration and recent research in rodents has identified a range of important and diverse roles attributed to this rather sizeable brain area located at the dorsal aspect of the STN in rodents (Figs. 2 and 3,Fig. 4) and in the primate (Fig. 6).
Further complexity in the brain habitat surrounding the STN and ZI is mediated by the hypothalamic nucleus known as pSTN which is anatomically directly joined with the ventromedial aspect of the STN.No distinct anatomical landmark separates the STN and pSTN.The close juxtaposition of the STN and pSTN is evident in the rodent (Figs. 2,Fig. 3,Fig. 4) and in the primate (Fig. 6).As will be discussed below, the interest in using experimental rodents to decode neurocircuitry and function of the STN and pSTN has increased over the past couple of years leading to exciting new discoveries.In contrast, in the primate, and in particular in the human, the pSTNas opposed to the STN -remains an understudied brain region.In fact, human brain atlases tend to overlook the pSTN altogether.Based on mutual connections of pSTN with brain areas relaying interoceptive information and homeostatic needs, the pSTN has been considered as a nucleus involved in the regulation of autonomic functions as part of an interoceptive network (reviewed in Shah et al., 2022).As further discussed below, recent findings emerged from rodent studies have revealed additional roles for the pSTN, which has been revealed as a nucleus of strong molecular heterogeneity.
Evidently, the STN, pSTN and ZI form a clinically relevant brain area.Based on their anatomical proximity, electrical activity mediated by DBS leads implanted in one structure (STN or ZI) might hypothetically affect function also in neighboring structures, such as the pSTN and passing fibers.It may therefore be worth considering that such unintended targeting can have profound impact of the outcome of treatment.Positive effects that alleviate symptoms as well as negative effects, such as adverse side-effects, may rely on additional brain structures than those primarily intended.Complete decoding of the anatomical-functional organization of the STN, pSTN and ZI is therefore highly relevant.Clinical reports have described a broad spectrum of adverse symptoms upon DBS, including motor-related (e.g.dyskinesia, spasms), Fig. 1.Basal ganglia structures and projections in the mouse brain with emphasis on the STN.Sagittal view of the mouse brain.A. Schematic illustration of the main connections of the basal ganglia circuitry in a sagittal section.Different types of projections are presented: glutamate, GABA and dopamine shown in green, red and blue, respectively.Abbreviations: EP, entopeduncular nucleus; GP, globus pallidus; pSTN, para-STN; SNc, substantia nigra pars compacta; SNr, substantia nigra pars reticulata; STN, subthalamic nucleus; VP, ventral pallidum; ZI, zona incerta.B. Distribution of mRNA encoding expression of the Vesicular glutamate transporter 2 (Vglut2) gene, shown for visualization of distinct excitatory brain structures, e.g.thalamus and STN.Dotted black lines show areas involved in the basal ganglia circuitry.Image credit: Allen Institute.© 2004 Allen Institute for Brain Science.Allen Mouse Brain Atlas.Available from: https://mouse.brain-map.org.Experiment 71724696, image 9. C. Distribution of mRNA encoding expression of the vesicular inhibitory amino acid transporter (Viaat) gene, shown for visualization of distinct inhibitory brain structures, e.g.GP, EP, SNr, striatum.Dotted black lines show areas involved in the basal ganglia circuitry.Image credit: Allen Institute.© 2004 Allen Institute for Brain Science.Allen Mouse Brain Atlas.Available from: https://mouse.brain-map.org.Experiment 79,677,349, image 10. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)physiological (e.g.autonomic dysfunction, weight gain), affective (e.g.anxiety, depression) and cognitive (e.g.attention deficit, loss of verbal fluency) (Chou et al., 2018;Merello et al., 2009;Mosley and Akram, 2021;Romito et al., 2002;Tommasi et al., 2008;Volkmann et al., 2010;Wang et al., 2018).Several of these side-effects show different distribution among male and female DBS patients (Jost et al., 2022), an observation which warrants further investigation and for which studies in rodents can help advance knowledge.
The complexity of the STN and its heterogeneous surrounding is important to explore and uncover in detail, not least to enhance clinical possibilities towards safe and successful interventions.In addition to studies of human and non-human primates, with a focus on solving human disorders affecting the STN, the anatomy, neurocircuitry and physiology have been extensively studied in for example fish, where motor studies of the lamprey have proven significant to enhance the understanding of the role of the STN, including the evolutionary perspective (Grillner, 2021;Grillner and Robertson, 2016;Stephenson-Jones et al., 2011).
Bridging the gap towards clinical neuroscience, the laboratory rodent serves as an excellent model organism in the pre-clinical research fields for the study of anatomical-functional organization relevant to human physiology.Ample laboratory technologies in experimental, fundamental and basic neuroscience allow high spatial and temporal precision in targeted manipulations and read-out on cellular, circuitry and behavioral levels.Lesions, electrical stimulations and pharmacological interventions as well as various genetics-based methodologies, further reinforced during the past decades by optogenetics and chemogenetics, have served to experimentally manipulate neuronal components of the subthalamic area in rodents for the purpose of revealing neurobiological underpinnings of circuitry mechanisms and behavioral regulation.Lately, molecular information has informed spatial precision allowing genetically defined subpopulations of neurons within each of the three nuclei to be selectively investigated.A surprising level of heterogeneity has been revealed on molecular, anatomical, circuitry and functional levels.Basically, a multitude of experiments in rodents has revealed a plethora of functional roles for each of these highly complex brain nuclei.A range of data now exist that have provided new insights into functions engaging each of these nuclei, the STN, pSTN and ZI.Considering the translational aspect, these new discoveries should be of essence to recognize in order to fully foresee positive effects of any induced clinical manipulation, such as DBS, as well recognize risks for harmful side-effects.Thus, parallel to patient-based reports and studies in non-human primates, experiment-based studies on rodents have vastly contributed new knowledge of the STN, pSTN and ZI in both health and disease.
Here, we bring the STN, pSTN and ZI together by reviewing current literature based on experimental studies in mice and rats, summarized with specific emphasis on method of manipulation, and reported behavioral readout of each manipulation.When interpreting experimental data, the level of spatial selectivity should be taken into consideration, especially when comparing results obtained in different settings.This is particularly important when addressing heterogeneous cell populations.We have therefore paid special attention to the method of manipulation used in each study, and outlined the promoters driving spatial selectivity, such as via promoter-driven expression of Crerecombinase, when this is applicable.

Anatomy and distinguishing cellular and molecular features of the STN
The STN is a biconvex, lens-shaped diencephalic structure located between the internal capsule anterolaterally, the cerebral peduncle and substantia nigra ventrolaterally, the red nucleus posteromedially, and the thalamus and ZI dorsally (reviewed in (Weintraub and Zaghloul, 2013)).The STN is surrounded by myelinated fiber tracts that course near the its outer borders.These include pallidothalamic pathways (that connect globus pallidus interna (GPi) and thalamus), medial lemniscus, spinothalamic tract, trigeminothalamic tract, reticulothalamic tract and MFB (Hamani, 2004).The size of the STN varies between different species; the number of neurons is on average 5400 in mice, 25,000 in rats, 155,000 in macaques and 560,000 in humans (Hardman et al., 2002;Sturrock, 1991).
In primates, the STN has been shown to be composed of three anatomical domains (also referred to as subdivisions or territories) (DeLong et al., 1985;Hamani, 2004;Joel and Weiner, 1997;Krack et al., 2010;Parent and Hazrati, 1995;Temel et al., 2005).According to the anatomical-functional so-called tripartite model, each of these three domains is responsible for engaging the STN in either motor, limbic, and associative functions, respectively.The dorsolateral third is, in this model, associated with motor function, the medial-most aspect with limbic function (often referred to as the "limbic tip") and the central portion with associative function (Alkemade et al., 2015;Benarroch, 2008).A study in humans identified three functional zones in the STN using diffusion-weighted imaging (DWI) (Lambert et al., 2012).Tract-tracing studies in non-human primates (Alexander and Crutcher, 1990;Haynes and Haber, 2013;Joel and Weiner, 1997;Parent and Hazrati, 1995) and functional studies, primarily from clinical DBS studies, have lent support to the tripartite model, but also challenged it.One study showed a clear overlap between the three proposed subdivisions (Haynes and Haber, 2013).Indeed, the tripartite model of the primate STN remains a focus of debate (Keuken et al., 2012).Further work will be needed to fully outline the organization of the STN.
In addition to anatomy, connectivity, electrophysiological properties, and functional (e.g.behavioral) output as criteria for describing the complex organization of the STN, also molecular markers, i.e. gene expression patterns (mRNA and/or protein), confined either to the majority of STN neurons (such as Slc17a6/Vglut2 (VGLUT2 protein), described above), or molecular patterns and profiles that define subpopulations or clusters of neurons within the STN, are useful to unravel any internal organization (Figs. 3 and 4).Gene expression patterns and protein localization patterns within the STN of rodents and primates were recently reviewed (Prasad and Wallén-Mackenzie, 2024).For example, parvalbumin (PV, also known as Pvalb) has long been established as a marker for a subset of STN neurons in primates, primarily those neurons that can be associated with the motor domain (Kultas-Ilinsky et al., 1998).

(caption on next page)
A. Ricci et al.Homeodomain 2 (PITX2) transcription factor which is expressed within the STN from embryogenesis (Dumas and Wallén-Mackenzie, 2019;Martin et al., 2004;Skidmore et al., 2008).The Pitx2 gene is also expressed in mature STN neurons, and Pitx2 mRNAs shows 100% overlap with Vglut2 mRNA (Schweizer et al., 2014(Schweizer et al., , 2016)).Histological analysis of several differentially expressed genes identified in the snRNASeq analysis subsequently enabled the identification of three major domains in the STN of mice based on combinations of mRNA patterns.For example, PV mRNA displayed a dorsal high -ventral low gradient, while the Collagen type XXIV alpha 1 chain (Col24a1) mRNA showed an opposite gradient (dorsal low -ventral high ), and was most dense in the ventromedial STN; the combinations of these two markers allowed the subdivision into three discrete STN domains.This same study also identified distinct mRNA patterns in pSTN and ZI, and additional nearby structures such as mammillary and hypothalamic nuclei (Wallén-Mackenzie et al., 2020).For example, expression of the Tac1 (Tachykinin) gene was identified as a molecular marker for the pSTN; this will be further discussed below.Recently, two more gene expression patterns were found to be localized in the STN, the peptide neurotensin (Nts), shown to overlap with Vglut2 (Li et al., 2020) and gamma aminobutyric acid receptor subunit rho-3 (Gabrr3), shown to overlap with Pitx2 (Parolari et al., 2021).
In addition to providing insight into the molecular complexity and heterogeneity of the STN structure, promoters driving the expression of these genes can be useful as experimental tools for driving transgenic expression of Cre-recombinase, primarily in mice but also in rats.As will be reviewed below, conditional transgenics implementing Cre-Lox technology in the STN has primarily been reached by the use of Vglut2-Cre mice (Adam et al., 2022;Fife et al., 2017;Friedman and Yin, 2023;Heston et al., 2020;Jia et al., 2022;Schor et al., 2022;Serra et al., 2023) and Pitx2-Cre mice (Bhuvanasundaram et al., 2022;Guillaumin et al., 2021;Nishioka et al., 2020;Parolari et al., 2021;Schweizer et al., 2014Schweizer et al., , 2016;;Serra et al., 2023).Other transgenic mouse lines implemented to drive selectivity towards the STN include PV-Cre (Serra et al., 2023) investigating a subtype of STN neurons, Nts-Cre (Li et al., 2020) and Gabrr3-Cre (Parolari et al., 2021), further outlined below.
Gene expression patterns (mRNA) distinguishing STN, pSTN and ZI, including molecularly defined subpopulations within each area, are shown in Figs. 3 and 4.

Projections of the STN
Projections from STN reach the basal ganglia output structures, globus pallidus (GP) and substantia nigra (SN).STN primarily projects to globus pallidus externa (GPe; known as GP in mice) and GPi (known as the entopeduncular nucleus, EP or EPN, in mice) and substantia nigra, primarily substantia nigra pars reticulata (SNr) but also substantia nigra pars compacta (SNc) (Fig. 1).Additional target areas for the STN include the pedunculopontine nucleus (PPN) and ventral pallidum (VP) (Emmi et al., 2020;Hammond et al., 1983;Smith et al., 1998).Lately, a connection has been found to a subpopulation of interneurons in the striatum (Kondabolu et al., 2023;Koshimizu et al., 2013).
Projections reaching the STN originate in striatum, GP, VP, nucleus accumbens, substantia innominata, and also cerebral cortex, cerebellum, thalamus, hypothalamus, amygdala and brainstem regions (Cavdar et al., 2018).Further, a recent review forwarded the idea that the STN is part of anatomical-functional loops involved in sensory responses derived from the brainstem and engaging the STN and SNr, including the periaqueductal grey nucleus (PAG), the parabrachial nucleus (PBN), superior colliculus (SC) and the PPN (Al Tannir et al., 2023).
According to a classical basal ganglia hypothesis, activation of the socalled direct pathway leads to initiation of voluntary movement by coordinated activity from the cerebral cortex and the dopamine (DA) neurons of the nigrostriatal pathway, leading to relief of tonic inhibition of GPi, which in turn leads to disinhibition of the thalamus, allowing a "go" signal for movement.The indirect pathway, on the other hand, competes with activity on GPi neurons by activation of the STN -GPi pathway, leading to a "stop" or "pause" signal, serving to prevent contextual irrelevant movement and competing motor programs (Sano et al., 2013;Smith et al., 1998).Further, via the hyperdirect pathway, excitatory input from the cortex rapidly and directly excites STN neurons, thereby exerting immediate "stop" effect on on-going motor activity (Nambu et al., 2002).The overall circuitry organization (Cortical-Striatal-Pallidal-Thalamic-Cortical) is similar for limbic and cognitive functions, in which striatal and pallidal pathways parallel to those engaged in motor function are recruited, while the output of these non-motor loops reaches other thalamic and cortical structures than those affected in the motor programs.

Functional roles of the STN
Below, we review the literature covering experimental studies of the STN using rodents.Focus has been placed on the experimental manipulations performed in order to bring attention to the level of selectivity achieved for the intended manipulation as this is important to understand how conclusions have been reached experimentally.Table 1 contains an inclusive summary of current literature on STN studies in laboratory mice and rats, and the descriptive text below details selected studies.A summary of the findings presented in the studies cited is shown in Fig. 5.The STN is most known for regulation of motor function, and its dysregulation in PD.A vast abundance of studies is focused on rodent PD models, and DBS mechanisms.The review therefore starts with motor function, moving on to additional functions currently ascribed to the STN, including cognitive and affective functions.

The STN in regulation of motor function
In the context of movement, studies of humans and animals commonly conclude that the STN mainly acts as a "motor brake"; STN pauses, delays, slows down, inhibits, and stops motor output (Benis et al., 2016;Bonnevie and Zaghloul, 2019;Isoda and Hikosaka, 2008;Mink, 1996;Pasquereau and Turner, 2017;Schmidt et al., 2013;Schmidt and Berke, 2017;Van Den Wildenberg et al., 2006).Below follows a summary of studies addressing the role of the STN in movement parameters.For ease of reading, the vast literature on the STN in motor regulation has been subdivided into sections based on method of manipulation.
Several studies have lent support to the idea that, in the 6-OHDA PD model, a lesion of the STN is able to reverse PD motor symptoms.Usually, such STN lesion is achieved by stereotaxic injection of ibotenic acid which leads to gliosis and neuronal loss, causing a reduction in excitatory activity of the STN.This STN lesion, in turn, results in enhanced locomotion and amelioration of PD motor symptoms (Bergman et al., 1990;Centonze et al., 2005).Moreover, other pharmacological substances such as kainic acid and N-methyl-D-aspartate (NMDA) have been used to induce lesions that have turned out to be effective in rescuing motor deficits in PD animal models (Guridi et al., 1994(Guridi et al., , 1996;;Henderson et al., 1998;Rizelio et al., 2010).For example, it has been shown that STN lesions in PD rat models can improve results both in the rotarod test and the limb use asymmetry test (Centonze et al., 2005), and to reverse some motor deficits in the open field test (Rizelio et al., 2010).
Parallel to studies implementing PD models, the impact of STN lesioning in wildtype rodents under non-pathological conditions has also been reported.For example, rats bilaterally injected with ibotenic acid to achieve STN lesion in both hemispheres led to increased spontaneous locomotor activity (Eagle et al., 2008).In the stop-signal reaction-time (SSRT) test, in which a "stop" signal is presented after a "go" signal, STN-lesioned rats showed impairments in stopping behavior.STN-lesioned rats were mostly impaired in stopping their responses independently from the delay between the onset of the "go" and the "stop" trials.The authors concluded that the missed initiation of inhibition of the response was not tightly related to the SSRT in particular, but more to a general failure in activating the "stop" process, probably due to an increase of motivation to earn the reward (Eagle et al., 2008).A subsequent study showed that, if the hyperdirect projections from cortex to STN were bilaterally eliminated using a specific photodynamic technique in wild-type mice, this elimination resulted in hyperlocomotion and a decrease in cortically evoked early excitation in GPe and SNr (Koketsu et al., 2021).
In conclusion, in rodent models of PD, lesion of the STN structure helps restoring motor abilities (such as limb symmetry and balance), while in non-pathological conditions, a lesion of the STN or the hyperdirect pathway leads to hyperlocomotion and motor impairments, and reduces the ability of the STN to stop an action.These findings are in accordance with studies in non-human primate PD models and in PD patients.
1.2.1.2.STN-DBS and optogenetic DBS-like stimulations in rodent PD models.DBS implementing electrode leads positioned in, or near, the STN to deliver electrical current is employed to achieve symptom alleviation in PD patients, and, pre-clinically, for the study of mechanisms using monkey PD models (Benazzouz et al., 1993;P. Limousin et al., 1995) and rodent PD models (Chang et al., 2003;Darbaky et al., 2003;Temel et al., 2005).In the context of DBS studies in rodent PD models (reviewed extensively in (Knorr et al., 2022)), different stimulation frequencies have been employed, in which low frequency stimulation (LFS) is characterized by 20-30 Hz, and high frequency stimulation (HFS) refers to around 130 Hz.
Already in 2009, Gradinaru et al. published a seminal study which elegantly implemented optogenetics, a new technology at the time, to address the role of STN neurocircuitry in motor regulation relevant to PD (Gradinaru et al., 2009).In a hemi-parkinsonian rat model, achieved by unilateral application of 6-OHDA in the MFB, excitatory and inhibitory opsins were expressed in the STN under control of the CamKII-promoter, directing expression to excitatory neurons.No motor amelioration or recovery of PD-like motor symptoms was observed, using either excitatory (Channelrhodopsin-2, CamKIIalphaChR2) or inhibitory (Halorodopsin from Natronomonas, CamKIIalphaeNpHR) opsins.In contrast to optogenetic stimulation, it was shown that electrical HFS of the STN improved PD motor symptoms.This apparent lack of success using optogenetic HFS of the STN motivated a new round of analysis in the same study, in which afferent fibers reaching the STN were targeted, reasoning that stimulation of these fibers might be responsible for the therapeutic effect of STN-DBS.By optogenetic excitation of 6-OHDA-treated Thy1ChR2 transgenic mice, in which ChR2 was only expressed in axonal projections reaching the STN, the authors observed that optogenetic HFS led to an amelioration of PD symptoms.Indeed, excitation of motor cortex layer V was correlated with improved motor function.In conclusion, this study showed that optogenetic HFS stimulation of the cortico-STN pathway, but not of the STN itself, led to improvement of PD motor symptoms (Gradinaru et al., 2009).
This finding was confirmed by Sanders and Jager, who used a double viral strategy in mice to anterogradely express Cre-dependent ultrafast ChR2 in M1 of primary motor cortex, and to retrogradely express mCherry (encoding a fluorescent chromophore reporter), under control of Cre recombinase expressed from a genetic construct injected into the STN of wild-type mice.Using this strategy, optogenetic stimulation of cortico-STN projections resulted in an improvement of motor symptoms, such as bradykinesia (slowness of movement) in the 6-OHDA mouse PD model (Sanders and Jaeger, 2016).Another study found that not only optogenetic excitation of cortico-STN pathway is involved in rescuing PD symptoms in mice, but also another pathway from the parafascicular thalamus to the STN (Watson et al., 2021).
The ultra-fast opsin Chronos is an ultra-light-sensitive blue Channelrhodopsin with a faster kinetics than ChR2.This opsin can reliably follow a high stimulation rate, thus creating an effective way to mimic DBS with optogenetics (Hight et al., 2015;Jun and Cardin, 2020;Klapoetke et al., 2014).Using this opsin at 130 pulses per second (pps) in a rat PD model, similar effects as achieved with electrical DBS were obtained, resulting in an overall amelioration of PD symptoms.Here, the authors proposed that amelioration of PD symptoms may be due to stimulation of local STN cells (Yu et al., 2020).
A recent study has suggested that the therapeutic effect of STN-DBS is due to the interruption of movement-related STN activity (Schor et al., 2022).This idea challenges the prevailing hypothesis that DBS functions via inhibition of the STN (see below, Hypothesis on the mechanism of DBS).Schor et al. combined electrical STN-DBS with artifact-free genetically encoded calcium indicator GCaMP fiber photometry (described by (Nakai et al., 2001)) using Vglut2-Cre mice to drive expression of Cre-dependent GCaMP6s in the 6-OHDA PD mouse model (6-OHDA injected into MFB).The authors claimed that STN-DBS increases STN activity, and further, that although there is indeed a modulation of the hyperdirect M1 pathway during STN-DBS in mice, there is a little correlation between M1 activity and the observed behavioral benefits of STN-DBS; reinforced further by the finding that surgical removal of M1 does not abolish the therapeutic benefit of STN-DBS.The authors of this study propose that both optogenetic and electrical STN-DBS increase, rather than decrease, the activity in the STN and SNr.Continuous vs. 50 Hz photostimulation produced different effects on movement-related activity.From a behavioral point of view, constant stimulation did not produce a significant increase in motor parameters, whereas Vglut2-Cre mice expressing opsins and stimulated at 50 Hz showed an increase in movement speed and percentage of time moving.Based on these results, the authors were able to claim that the interruption of movement-related activity of the STN is sufficient to have therapeutic effects on PD motor symptoms, with a significant increase in velocity of movement and time spent moving (Schor et al., 2022).

Hypotheses on the mechanism of DBS.
Despite the clinical benefits of STN-DBS, the exact mechanism of action underlying its therapeutic effects still remains at the center of numerous controversies.To allow interpretation of rodent data reviewed in the present literature review, we have included a short summary of some of the current hypotheses concerning the mechanism of action of STN-DBS.For complete information, see the original articles.
In accordance with the proposed role of the STN as counteracting locomotor initiation via the indirect and hyperdirect pathways of the basal ganglia, it was initially suggested that DBS inhibits the STN (Andy et al., 1963;Bergman et al., 1990;Filali et al., 2004;Moran et al., 2011;Yoon et al., 2014).The hypothesis postulating that the clinical effects of DBS is due to an inhibition that induces a functional deafferentation from the target nuclei, acting as a "reversible lesion", is based on similarities in the therapeutic effects upon pallidal and subthalamic lesions (Benabid et al., 1994;Benazzouz et al., 1993;Patricia Limousin et al., 1995;Benazzouz and Hallett, 2000).The therapeutic similarities achieved with ablative surgery, led to the proposal that DBS works through an inhibition of cell bodies near the stimulation electrode, acting as a "reversible lesion".Indeed, stereotactic ablation of STN improves motor symptoms of PD (Alvarez, 2005), and intracranial injections of lidocaine or muscimol into the STN induce prompt reversal of parkinsonian symptoms in parallel with the focal suppression of STN neuronal activity (Levy, 2001).Furthermore, inhibition of neuronal firing has been recorded in the surroundings of stimulation sites in parkinsonian patients during STN-DBS (Filali et al., 2004;Welter et al., 2004).
In addition to this inhibitory mechanism, DBS has been shown to increase the activity of nearby axonal projections, somehow modulating the activity in nearby brain structures (which project or receive projections from the region stimulated during DBS) (Florence et al., 2016;Johnson et al., 2008;Hashimoto et al., 2003;Vitek, 2002a).Thus, although DBS may reduce cell body activation (local inhibition), some studies suggest that efferent axons may be activated (McIntyre et al., 2004;Hashimoto et al., 2003).These studies have led to the hypothesis that the effect of DBS might be due to the involvement of afferent, efferent and passing fibers (Kringelbach et al., 2007), resulting in glutamate release and activation of other basal ganglia nuclei (Windels et al., 2000).
Another hypothesis states that STN-DBS acts via antidromic activation of the motor cortex.The hyperdirect pathway consists of excitatory projections from the cerebral cortex to the STN (Nambu et al., 2000).It has been proposed that an antidromic cortical activation may be responsible for the anti-parkinsonian effects based on observations that stimulation of the cortico-subthalamic hyperdirect pathway alleviates PD motor symptoms in animal models (Chu et al., 2017;Gradinaru et al., 2009;Li et al., 2007Li et al., , 2012;;Sanders and Jaeger, 2016).In agreement with this hypothesis, cortical potentials evoked by STN-DBS have also been observed through electrophysiological recordings in humans (Chen et al., 2020;Miocinovic et al., 2018).On the other hand, one study has highlighted that antidromic activation is unstable over time and is observed during STN, but not GPi, stimulation (Johnson et al., 2020).This observation has casted doubt on whether cortical antidromic activation is the factor responsible for the therapeutic effects of DBS, given that both STN and GPi stimulation are normally effective in improving the motor symptoms of PD.
It has been established that STN neurons fire abnormally in PD patients (Vila et al., 2000) and show low-frequency synchronization with some cortical regions (Brown et al., 2001;Eusebio and Brown, 2009;Tinkhauser et al., 2018).Pathological oscillatory activity in circuits including cortex, basal ganglia, thalamus, and cerebellum have been linked to motor symptoms of PD (Herrington et al., 2016).One hypothesis suggests that STN-DBS acts therapeutically effective by interrupting these synchronous oscillations (Eusebio et al., 2011;Shimamoto et al., 2013).In particular, excessive synchronization of neuronal activity in the beta frequency band within motor circuits has been observed in PD, leading to the hypothesis that DBS acts through the modulation of these oscillatory patterns (Kuhn et al., 2008;Neumann et al., 2016).Several studies propose that disruption of beta band oscillations may be the mechanism responsible for some of the motor effects of DBS (Bronte-Stewart et al., 2009;Giannicola et al., 2010;Kuhn et al., 2008;Wingeier et al., 2006).Although beta band oscillations represent the core of oscillatory activity research in PD, other frequency bands may also play an important role.For example, increased activity of the gamma band (Florin et al., 2013;Trottenberg et al., 2006) and theta band (Alonso-Frech, 2006;Foffani, 2005) have been observed both in patients with PD and in patients with non-PD-related dyskinesia, presumably due to chronic treatment with dopaminergic drugs.Lately, it has been found that STN-DBS in PD leads to regularization, meaning an adjustment of the irregular spike train usually present in PD, of firing activity and alleviation of motor deficits of the STN by increasing endogenous histamine release (Whalen and Gittis, 2018;Zhuang et al., 2018).
Finally, as discussed above, optogenetic and electrical STN-DBS were recently found to disrupt the firing dynamics of the STN, and to increase, rather than decrease, activity in STN.Based on results from electrical STN-DBS at therapeutical settings and optogenetic STN-DBS, disrupting movement-related STN dynamics was proposed to be sufficient to result in the therapeutic effect on PD motor symptoms (Schor et al., 2022).
A general consensus about the mechanisms by which DBS promotes activity changes in the STN remains to be reached.It is possible that the efficiency of DBS is due to the synergistic effect of several mechanisms of action.Further knowledge is required to fully solve the mechanistic and neurobiological underpinnings of DBS.
1.2.1.4.Optogenetics in STN of mice and rats in, and beyond, models of human brain disorder.One important goal of this last decade of research has been to find out more about the role of the STN in motor control under non-pathological conditions, that is, rodents that have not been treated experimentally to induce human-like brain disorder, such as PD or OCD.The idea is that if we know more about the role of the STN under normal conditions, the easier it will be to better understand brain function in general, but also what goes wrong under pathological conditions, and how to correct this, for the purpose of alleviation of symptoms.Thus, optogenetics has been increasingly applied to rodents that have not undergone any pre-treatment (such as 6-OHDA) to model a human disorder, and where the purpose is not to address mechanisms of DBS, but to uncover natural roles exerted by the STN in neurocircuitry and behavioral regulation.
In this context, primarily Vglut2-Cre and Pitx2-Cre transgenic mice have been used to achieve selectivity for expression of "floxed" (flanked by lox sites) DNA constructs in the STN.The Pitx2 gene encodes the PITX2 transcription factor, expressed in the STN from early development (Dumas and Wallén-Mackenzie, 2019;Martin et al., 2004;Skidmore et al., 2008) and through to adulthood (Schweizer et al., 2014(Schweizer et al., , 2016;;Wallén-Mackenzie et al., 2020).By producing mice double-positive for the Pitx2-Cre transgene (Martin et al., 2004) and a floxed Vglut2 allele (Wallén-Mackenzie et al., 2006), such Pitx2-Cre/Vglut2 conditional knockout (cKO) mice showed reduced, but not completely ablated, Vglut2 gene expression levels (Schweizer et al., 2014).By introducing ChR2 into the STN of Pitx2-Cre/Vglut2 cKO and Pitx2-Cre-positive control mice, photostimulation evoked reduced post-synaptic glutamatergic activity in cKO compared to control mice in two STN target areas, the EP and SNr.Motor assessment revealed significant hyperlocomotion by cKO mice, while other motor parameters such as gait, coordination and balance remained intact (Schweizer et al., 2014).The concept of hyperlocomotion was subsequently discussed in an exchange that concluded a bona fide increase in locomotion in the cKO mice (Konsolaki and Skaliora, 2015;Pupe et al., 2015).Reduced Vglut2 gene expression levels in the STN thus provided an indirect way to address STN's role in movement, allowing the conclusion that reduced STN-derived glutamatergic neurotransmission increased locomotion.This finding was in accordance with the classical basal ganglia hypothesis (described above) proposing that the natural role of the STN is to, by excitation of basal ganglia target areas, inhibit movement (via the indirect and hyperdirect pathways).
To follow up on these findings, and to test the long-standing view that excitation of STN under non-pathological conditions is correlated with reduced movement, and inhibition of STN with enhanced movement, several studies have subsequently employed further optogenetic interrogations.One critical study demonstrated how transient optogenetic STN activation leads to the interruption of an on-going licking behavior.By using Vglut2-Cre mice to express opsins in the STN, the study showed that brief optogenetic STN excitation causes "stop" or "pause" of a specific behavior (e.g.milk-licking).When an optogenetic stimulation is delivered to the STN, following a self-initiated bout, the action of licking is interrupted, suggesting that the activation of the STN is correlated to stopping of an ongoing action (Fife et al., 2017; this study is also further discussed below).
Along these lines, a study from our own laboratory demonstrated that optogenetic STN activation and inhibition indeed exert opposite effects on locomotion such that optogenetic inhibition enhances locomotion, while optogenetic activation reduces locomotion.This result validated the strong impact of STN activity on movement.Further, unilateral stimulation led to rotational behavior, while when tested on the rotarod and beam walk, mice stopped upon photostimulation leading to STN excitation, and thereby fell off the apparatus.Thus, several motor parameters were affected by optogenetic activation vs. optogenetic inhibition in a manner confirming the role of the STN in the indirect and hyperdirect pathway of the basal ganglia.Curiously, however, not all types of movement were stopped upon STN excitation.Instead, self-grooming was immediately increased when mice received photostimulation leading to STN excitation (Guillaumin et al., 2021).While grooming is a natural cleansing behavior for rodents, the behavior observed did not follow the normal cephalo-caudal rule but engaged movement of front paws only and remained focused in the face; this face-grooming coincided completely with photostimulation and ceased when stimulation stopped (Guillaumin et al., 2021;Serra et al., 2023).Stereotypic grooming is commonly used as behavioral model for repetitive and stereotypic patterns of movement in OCD, with rodent grooming as a behavioral correlate of compulsive grooming in human OCD individuals.This significant finding in mice provided compelling evidence for a direct causality between STN excitation and repetitive grooming in mice (Guillaumin et al., 2021;Serra et al., 2023).
Similar results were subsequently shown by Parolari et al. implementing optogenetic stimulation of the STN in Pitx2-Cre and Gabrr3-Cre transgenic mouse lines.The Gabrr3 gene encodes Gamma-aminobutyric acid type A receptor subunit Rho 3 (GBRR3), a subunit of the GABA A receptor.In this study, it was shown that optogenetic activation of STN neurons inhibits locomotion and leads to repetitive behavior, similar as described by Guillaumin et al. Further, Parolari et al. showed that these behaviors were achieved via projections from STN to GP/EP and SNr.Moreover, selective photoinhibition of STN Gabrr3-Cre-positive neurons was shown to promote suppression of OCD features in mouse model of OCD (the Sapap3 KO model) (Parolari et al., 2021).This is an important finding which, by strengthening the observations of a role for the STN in compulsive behavior, should be further explored (see also below for additional discussion on STN in compulsive behavior).
A study by Ji et al. confirmed the data reported by Guillamin et al. and Parolari et al. and identified both parallel and collateral projections from the STN to SNr and EP.In wildtype mice, by opsin injection targeting the STN or by using a combinatorial approach (Cre-dependent and projection-specific strategy), the authors found that optogenetic activation of STN projections to EP and SNr leads to a decrease in locomotion (Ji et al., 2023).In a PD model achieved through 6-OHDA injection in the MFB, the authors observed a decrease in projections from STN to SNr, and an increase in projections from the STN to the EP.Moreover, optogenetic inhibition of these pathways was shown to reduce parkinsonian deficits, for example increasing locomotion to normal levels (Ji et al., 2023).Another study using a PD rat model created by injection of 6-OHDA in the MFB and application of an inhibitory opsin (NpHR) in the STN demonstrated that inhibition of the STN ipsilateral to the 6-OHDA lesion leads to recovery of selected motor symptoms, including contralateral forelimb akinesia (Yoon et al., 2014).
In the context of motor dysfunctions other than PD but related to STN, it has been shown that if STN is lesioned or inactivated unilaterally, uncontrolled movements can be observed; a condition known as hemiballismus (Hamada and DeLong, 1992).In mice, unilateral optogenetic excitation of the STN produces measurable ipsilateral rotational movements in agreement with the pathology of hemiballismus (Guillaumin et al., 2021).
Other studies have focused on the manipulation of the cortico-STN pathway.In PD mouse models, an increase in striato-pallidal transmission has been shown to cause a downregulation of the cortico-STN transmission and disinhibition of the STN (Mallet et al., 2006;N. Mallet et al., 2008).In a study by Chu et al. an increase in striato-pallidal activity was sufficient to downregulate cortico-STN activation in non-DA depleted mice.Further, transgenic mice homozygous for a floxed Grin 1 (encoding the NMDA receptor subunit GluN1) allele (Grin lox/lox ) were injected in the STN (to study STN disinhibition) or in the cortex (to study cortico-STN transmission) with viral vectors in which ChR2 and/or Cre recombinase constructs were under control of the human synapsin (hSyn) promoter.In addition, mice were rendered parkinsonian by 6-OHDA.In these sets of experiments, Chu et al. show that, in the mouse 6-OHDA PD model, NMDAR activation in the STN alone can lead to a downregulation of cortico-STN transmission.They also conclude that an effect of NMDAR knockdown in the STN is increased cortico-STN transmission and improve motor function in PD mice (Chu et al., 2017).
Another more recent example is provided by a report by Adam et al. which focused on the pathway from the secondary motor cortex (M2) to the STN, employing a locomotor task with a head-fixed mouse on a treadmill and LEDs as visual cues.Mice were unilaterally injected in the M2 with ChR2 under control of the CamKII promoter, and photostimulation was delivered in the STN.The authors found that this optogenetic activation stops locomotion, in particular when the stops are visually guided.Using a mixture of computational models, singleunit recording, and optogenetics in wild-type and Vglut2-Cre transgenic mice, the authors identified two pathways: STN -PPN and STN -SNr -PPN, both of which convey different signals in the stop response, with the PPN as a force that drives the integration differentiation for locomotor control (Adam et al., 2022).
A recent study gives a new perspective on motor regulation by the STN, using a 3D motion capture system (Friedman and Yin, 2023).The authors provided very brief pulses of optical stimulation at different frequencies in Vglut2-Cre mice expressing ChR2 in the STN.They show that, with very short latency from the stimulation onset, unilateral STN stimulation pulses lead mice to perform highly precise ipsilateral rotations, in accordance with previous reports (Guillaumin et al., 2021;Parolari et al., 2021).Further, unilateral stimulation led to increase in head yaw and roll, while bilateral STN stimulation instead increased a movement referred as head pitch.It was found that specific movements in head and torso have a linear relationship with the parameters of stimulation, a concept referred to as scaling of movement and which was observed during STN stimulation.In addition, optogenetic stimulation of STN terminals in the mesencephalic locomotor region of Vglut2-Cre mice induced similar motor responses as when the STN itself was stimulated (Friedman and Yin, 2023).
Lastly, a pathway from the STN to cerebellum was recently described.Viral-genetic tracing, optogenetics and electrophysiological recordings performed in Pitx2-Cre mice enabled the demonstration of a pathway from STN -via the PPN -to cerebellum.This pathway might have a role in modulating movement, possibly primarily motor coordination, commonly known as the main role of the cerebellum (Bhuvanasundaram et al., 2022).
In summary, optogenetic studies in rodents have validated the role of the STN in the modulation of motor activity, as excitation of the STN reduces both vertical and horizontal locomotion and plays a role in the interruption of ongoing actions.In further support of such a role in movement regulation, optogenetic STN inactivation increases locomotion, in accordance with cKO and lesion studies in rodents.Curiously, optogenetic STN excitation induces self-grooming activity, a putative behavioral correlate of compulsion with repetitive motor activity as behavioral output.The STN has also been proposed to scale movement, a new phenomenon that raises new questions regarding the role of STN in movement suppression, and should be further explored.Clearly, details specifying the full role of STN neurons in motor regulation will need further investigation.

The STN in regulation of non-motor function: aversion, reward and motivation
Motivation and reward research have long seen the midbrain DA system as the protagonist, while the STN was long mainly regarded as a motor structure.However, results from lesion and DBS studies over the past three decades have led to the addition of the STN to the series of brain structures that engage in motivation and reward (e.g.Baunez et al., 2005Baunez et al., , 2002;;Eagle et al., 2008;Lardeux et al., 2013;Rouaud et al., 2010).Furthermore, a role for STN in behavioral aversion and in aversive learning was recently demonstrated (Serra et al., 2023).
The section below will describe some of these studies in more detail, while all studies are summarized in Table 1.

Lesion and DBS studies: experimental basis for the idea of STN as potential brain target for treatment of addiction.
In the context of motivation for natural (e.g.sweet food) and drug (e.g.cocaine) reward, STN lesioning has reportedly given rise to two opposite responses.In one of the first studies addressing a putative reward function of the STN, rats were bilaterally lesioned with ibotenic acid and underwent a battery of behavioral paradigms, testing different reward and reinforcement paradigms.These included an operant cocaine self-administration and food self-administration paradigms, implementing both fixed and progressive ratio, and a place conditioning test (Baunez et al., 2005).Lesioning of the STN using ibotenic acid led to increased motivation for food reward and decreased motivation for cocaine, thus opposite responses.When a reversible lesion was created by injection of muscimol, a receptor agonist for GABA A R which transiently inactivated the STN (Baunez and Robbins, 1999), an interest in food reward was observed.Based on their seminal findings, the authors postulated the idea that the STN engages different microcircuits that respond differently to natural reward vs. addictive drugs (Baunez et al., 2005).
In the context of addressing STN-DBS and reward behavior in experimental mice and rats, results vary with experimental conditions, such as unilateral vs. bilateral stimulation, chronic or acute stimulation, and according to the polarity and frequency of the stimulation.By performing a systematic analysis of rodent studies assessing effects of STN-DBS on reward seeking, reward motivation and reward consumption, Vachez and Creed concluded that STN-DBS leads to a reduction in reward-seeking in the majority of experiments, and argued for the potential of optogenetic-assisted circuitry analysis to improve knowledge of the STN in motivated behavior (Vachez and Creed, 2020).
As discussed above, DBS can be delivered at high frequencies (HFS) and low frequencies (LFS).STN-HFS has been proposed to mimic the effect of STN lesion in terms of motivation by increasing the motivation for food and sucrose while decreasing motivation for cocaine (Rouaud et al., 2010;reviewed in Weintraub and Zaghloul, 2013).Furthermore, silencing of STN neurons upon STN-HFS has been reported to block the re-escalation of heroine intake in rats.In this case, the silencing was assessed through ex vivo electrophysiological recordings detecting a decrease or absence of action potentials after HFS stimulation at 130 Hz (Wade et al., 2016).
Studies identifying a role for STN in reward and motivation for reward brought forward the idea of the STN as a new treatment target for addiction (Bentzley and Aston-Jones, 2017;Creed, 2018;Degoulet et al., 2021;Uslaner et al., 2005).STN-HFS seems to prevent and reverse cocaine use (Pelloux et al., 2018;Rouaud et al., 2010) and block re-escalation of heroin intake (Wade et al., 2016) whereas STN lesion prevents alcohol intake (Pelloux and Baunez, 2017).Moreover, a recent study found that STN-LFS, but not STN-HFS, reduces pathological cocaine-seeking in a rat model of addiction (Degoulet et al., 2021).
In conclusion, experimental data in rodents support the idea that a reduction in STN activity (due to lesions or as a result of STN-DBS) reduces substance abuse and the likelihood of relapse.The range of rodent-based data presented over the past years currently forms a platform of importance for new treatment strategies for substance use disorder by exploring the STN as treatment target.

STN in reward and aversion
1.2.2.2.1.STN firing in reward and error coding.Further information concerning the role of the STN in limbic function suggests that the STN is involved in reward and error coding.Certain STN neurons in the rodent have been shown to respond to expectation of aversive and positive reward in a context-dependent manner (Breysse et al., 2015;Lardeux et al., 2009).For example, it was observed that STN neurons in the rat would fire during expectation and delivery of a predicted reward (4 or 32% sucrose solution).This led the authors to divide STN neurons into two functional subpopulations; responding to either high or low concentration of sucrose (Lardeux et al., 2009).In a subsequent study, two neuronal subpopulations of STN neurons were identified based on responses in a taste reactivity test to one specific type of substance, either rewarding (sucrose) or aversive (quinine).By studying the firing properties of rat STN neurons, STN subpopulations that respond differently to positive and negative stimuli were identified.Thus, STN was shown to encode the value of the stimuli depending on context.STN neurons that specifically encode reward prediction error when a reward is omitted were demonstrated.For example, during the taste reactivity test, a random "omission trial" (trial without reward) was presented revealing that some neurons responded only to this type of trials.The authors hypothesized that these are influenced by the dopaminergic system.Lastly, a group of neurons that only responds to errors when the preferred reward is missed; these were referred to as "oops neurons" (Breysse et al., 2015;Lardeux et al., 2009).
1.2.2.2.2.Roles of the STN in reward and aversion.In another line of research, the proposed role of STN in reward-related behavior was addressed by measuring sugar consumption achieved by Pitx2-Cre/ Vglut2 cKO mice, constructed to exert reduced glutamatergic neurotransmission from the STN (described above; Schweizer et al., 2014).These cKO mice showed increased sugar consumption compared to control mice, a finding biochemically strengthened by altered dopamine receptor binding activity in ventral vs. dorsal striatum in cKO compared to control mice (Schweizer et al., 2016).An interaction between the STN and the midbrain dopamine system in mice was thereby described, of potential interest to affective side-effects induced by STN-directed treatments such as DBS.Finally, this study also defined an alteration in the morphology of the STN structure, possibly indicating a need for normal Vglut2 gene expression levels during development (Schweizer et al., 2016).
Recently, also another study linked food reward to STN activity.By using fiber photometry, the authors observed that STN neurons are activated by feeding behavior in freely moving mice.Further, STN activity was enhanced or reduced in response to food intake such that neuronal activation was increased upon ingestion of larger food pellets and reduced upon ingestion of bitter food pellets.For optogenetic stimulation of STN, the authors injected a viral construct into wildtype mice containing excitatory or inhibitory opsins under control of the CamKII promoter.In accordance with some lesion and cKO studies, as well as DBS studies, showing that loss of STN, or its activity, increases motivation for food reward, Wu et al. found that optogenetic inhibition increases food intake while optogenetic excitation of the STN decreases food intake (Wu et al., 2020).
Based on previous findings that optogenetic photostimulation of the STN decreased several movement parameters while increasing stereotypic face-grooming and escaping behavior (Guillaumin et al., 2021), the hypothesis that STN excitation is directly correlated with aversion was recently addressed.One hypothesis holds that stereotypic grooming in rodents and primates, including humans, is caused by strongly unpleasant anxiety or aversion, as a means to control this unpleasant experience and thereby relieve the stress induced by such conditions.This is also one hypothesis, or subgroup, of OCD in human individuals.Other subgroups of OCD also exist, including loss of cognitive flexibility and/or top-down-control (Robbins et al., 2019).
Until recently, STN excitation had not been directly correlated with either anxiety or aversion.However, in a study by Serra et al. the previous findings of stereotypic behavior upon optogenetic STN excitation were followed-up with analysis of anxiety and aversion upon optogenetic excitation of the STN, and STN terminals in the VP.In this study, several Cre-recombinase mouse lines were implemented which allowed the identification of a direct correlation between optogenetic STN excitation and aversion in real time as well as aversive learning, while no correlation between STN excitation and anxiety was found.Here, Pitx2-Cre and Vglut2-Cre mice were used to target the whole STN, and PV-Cre mice to tease out a dorsal PV-positive subpopulation of STN neurons.Further, it was shown that optogenetic excitation of the Tac1-positive pSTN neurons (Tac1-Cre mice) did not cause aversive learning, verifying the specificity of this learned response to the STN over the adjacent pSTN.
STN mice (Pitx2-Cre and Vglut2-Cre) responded with strong avoidance behavior to 20 Hz optogenetic stimulation, and maintained the avoidance upon presentation to photostimulation-paired cues.While the A. Ricci et al.PV subpopulation did not show evidence of playing a major role in these responses, the involvement of the STN structure in aversive behavior was further confirmed with subsequent tests.For example, optogenetic stimulation of the STN was able to reverse the preference that mice normally have for closed over open space, and the stimulation also led to an interruption of sugar self-administration in a positive reinforcement paradigm.
As shown in a series of publications over the past years, largely based on optogenetics and viral-genetic tracing experiments, aversion neurocircuitry is centered around the lateral habenula (LHb) (Lazaridis et al., 2019;Lecca et al., 2017;Mondoloni et al., 2022;Shabel et al., 2012;Tooley et al., 2018).However, the study by Serra et al. showed that there are no direct projections from STN to LHb.To address neurocircuitry of STN-mediated aversion, a dual viral-genetic strategy allowing retrograde tracing and recording of post-synaptic responses was applied which led to the identification of an indirect connection from STN to LHb via EP and VP (STN-VP/EP-LHb).In addition, optogenetic stimulation of STN-terminals within the VP caused the same behavioral avoidance and aversive learning as when STN itself was stimulation, thus verifying the aversion circuitry engaging the STN (Serra et al., 2023).
Another recent study links a subpopulation of STN neurons with perception of threats.Tseng et al. used a combination of Cre-dependent and projection-specific chemogenetic and optogenetic strategies, together with EEG and fiber photometry.Wildtype mice were subjected to a specific predator odor (trimethylthiazoline, TMT) or a visual stimulus of a predatory rat.The results showed that corticotropin-releasing hormone (Crh) positive neurons of the medial STN are necessary to mediate adaptive defensive and threatening responses in rapid eye movement (REM) sleep and wakefulness.The authors proposed that these responses occur via the pathway connecting these neurons to the lateral GP (Tseng et al., 2022).
In conclusion, recent findings highlight a complex role for the STN in reward processing, error coding, and aversion mechanisms with the identification of different types of subpopulations responding differently to different contexts, such as positive and negative stimuli, and defensive and threatening responses.Further, while it has been reported that STN neurons are activated upon food intake, reduced activity of the STN (cKO or optogenetic inhibition) increases the motivation for food reward.Finally, optogenetic activation of the STN has been correlated with behavioral avoidance, including aversion learning.

The STN in regulation of non-motor function: cognition and impulse control
Parallel to motor and limbic function, the STN has long been shown to play important roles in attention, impulsivity and conflict (Aron, 2007;Aron and Poldrack, 2006;Cavanagh et al., 2011;Pasquereau and Turner, 2017;Zavala et al., 2013Zavala et al., , 2015)), recently validated further with the use of transgenic and viral-genetic methodology.

Cognition and attention.
Cognitive impairment studies were first presented three decades ago, in which rats lesioned in the STN with ibotenic acid underwent a 5-choice serial reaction time test (5CSRTT).The authors reported deficits beyond motor impairments, related to cognition, attention, and impulsivity, such as perseverative nose-poking and panel-pushing as well as elevated premature responding (Baunez andRobbins, 1997, 1999).In the context of cognition, STN-HFS alleviated motor, but not cognitive, phenotypes in a rat model of PD (Baunez et al., 2007;Darbaky et al., 2003).
More recently, STN-lesioned rats were observed to show a pattern of behavioral displays related to attention deficit.In the interdimensional/ extradimensional (ID/ED) test for the study of attentional set-shifting, lesion of STN and ZI was shown to cause excessive digging and abnormal learning.Moreover, these deficits were fully corrected when the lesioning using ibotenic acid in the STN/ZI lesioning was combined with 6-OHDA lesioning in the dorsomedial striatum (Tait et al., 2017), further reinforcing the relationship between the STN and the midbrain dopamine system as discussed above.
The observations made upon STN lesioning were confirmed in subsequent studies implementing other methods to target the STN.For example, one study used inhibitory chemogenetics (mutated human M4 muscarinic acetylcholine receptor (hM4Di) injection into STN followed by CNO treatment) to silence the STN in Pitx2-Cre mice and found that CNO treatment reversibly impaired attention in a 5-CSRTT test; attention impairment here is defined as a decrease of accuracy of response upon STN inhibition (Nishioka et al., 2020).
In another study, optogenetic excitation of STN at 40 Hz in Vglut2-Cre mice not only impaired motor abilities, but also cognitive performance.In a delayed match to position test, mice receiving the stimulation were slower than control mice at giving the right response and displayed an increase in omission error rate (Heston et al., 2020).According to a previous study, inhibition of the STN during surprising stimuli was shown to blunt the effect of surprise (Fife et al., 2017).In this more recent study, optogenetic inhibition of the STN was shown to lead to more errors, independently of the surprise effect.The authors conclude that STN activation or inhibition increases the error rate, with an effect on cognitive performance (Heston et al., 2020).
In conclusion, studies addressing the involvement of the STN in cognitive performance do not always seem to agree with each other.While studies that have used inhibition of the STN (obtained through a lesion, chemogenetic or optogenetic inhibition) seem to agree in suggesting that this can worsen cognitive performance, others state that an activation of the STN (through brief optogenetic stimulation) can also cause a reduction in cognitive performance.Reasoning on these results, it seems that normal endogenous activity of the STN is necessary to guarantee optimal cognitive performance.It must also be taken into consideration that the discussed results were obtained using different behavioral tests, and that the activation and inhibition of STN neurons were achieved through different experimental methods in which temporal and spatial resolution were not identical.Moreover, alterations in the functionality of the STN can induce imbalances in neurocircuitry activity which may further impact on cognitive performance.The topic of cognitive performance will be important to address further to fully recognize the neurobiological underpinnings implicating STN circuitry, including the putative contribution of distinct STN subpopulations which may help clarify the picture.

Decision-making, impulsivity and compulsivity.
In a decisionmaking context, the activation of the STN leads to a "hold your horses" type of response (Frank, 2006).This has been studied in humans (Zaghloul et al., 2012;Zavala et al., 2017) and in rodents (Schmidt et al., 2013).Schmidt et al. showed that there is a race between "stop" and "go" processes, where the striatal neurons of the basal ganglia direct pathway produce movement through an inhibition of the SNr, while the STN stops movement through the excitation of the same SNr neurons.The idea underlying this study was to test the hypothesis of a "race" between go and stop processes in the basal ganglia.Via the direct pathway, through the action of the striatum, movement is promoted ("go") through SNr inhibition, while STN inhibit movement ("stop") through excitation of SNr.The authors applied single-unit electrophysiological recordings during a stop-signal task.Here, the subject has a "go" cue, which, after a certain delay of time can be followed, in 30% of the trials, by a "stop" signal.If the rats stop in time, the response is correct and performance is rewarded by a sugar pellet.However, if they do not stop in time, the response has failed and no reward is obtained.The authors found that STN neurons responded immediately to the cue with low and fixed latency both in correct or failed stop trials, leading towards the idea that STN provides the same fixed signal to pause an action in any of the cases.On the basis of the very fast kind of responses mediated by the STN, the authors also hypothesized that the STN is one of the components of a so called "interrupt" system, in which the parafascicular thalamus and the PPN projecting to STN lead to pauses in the action (Schmidt et al., 2013).
In the context of impulsivity and compulsivity, some PD patients exhibit impaired inhibitory control and compulsive behaviors (Djamshidian et al., 2012;Gauggel et al., 2004;Kobayakawa et al., 2008;Milenkova et al., 2011;Nombela et al., 2014;Obeso et al., 2011).This could be due to neurodegeneration of dopamine neurons projecting to the ventral striatum via the mesolimbic pathway (Braak et al., 2004;Kano et al., 2011;Van Der Vegt et al., 2013).However, a greater conservation of the mesolimbic than the nigrostriatal pathway in PD (Kumakura et al., 2010) could explain the effect of supplemental dopaminergic drugs in inducing increased risk-taking behaviors (Djamshidian and Lees, 2014;Voon et al., 2010Voon et al., , 2011)).Further, the aberrant firing of STN neurons as a consequence of dopamine neuron degeneration likely contributes to increased impulsivity, compulsivity and risk-taking.Moreover, it has been shown that the STN receives cortical projections from the inferior frontal gyrus and pre-supplementary motor area via the hyperdirect pathway (Aron, 2007;Inase et al., 1999;Rae et al., 2015).From studies of PD patients, it is known that STN-DBS can lead to an increase in impulsivity (Lo Buono et al., 2021;Pham et al., 2021).These findings are aligned with studies demonstrating that STN-DBS increases impulsive behavior in naïve rats (Aleksandrova et al., 2013;Anderson et al., 2020).Confirming the involvement of the STN in impulsivity, it has been shown in a delay-discounting test, that rats lesioned in the STN with ibotenic acid are prone to be less impulsive, but to carry cognitive deficits (Winstanley et al., 2005).This finding was in accordance with other studies showing that STN-DBS decreases risk taking in risk-preferring rats (Adams et al., 2017).And high frequencies DBS stimulation leads to a decrease in premature responses in the choice reaction time task (Desbonnet et al., 2004).
In a recent study, fMRI, in vivo and ex vivo electrophysiology recordings, optogenetics, and pharmacological and genetic manipulation of metabotropic glutamate receptor 4 (mGlu4) in the STN were performed in wildtype mice tested in the go/no-go test.The authors reported that STN in general, and mGlu4 function in the STN in particular, is a central modulator for impulsive traits.Specifically, STN optogenetic inhibition was shown to results in increased pre-cue response rate, a measure of increased impulsivity; and vice versa, STN optogenetic activation caused reduced impulsivity (Piszczek et al., 2022).This finding was confirmed in another study in which chemogenetic inhibition of the STN in Pitx2-Cre mice leads to a higher number of premature responses in a 5-CSRTT test, an observation lending support to the idea of increased impulsivity upon STN inhibition (Nishioka et al., 2020).
In a separate line of research, a bi-transgenic mouse line was generated in which ChR2 gene expression was placed under control of the Tryptophan hydroxylase (Tph2) promoter (Tph2-tTA:tetO-ChR2-EYFP). In this study, the STN and pSTN are grouped together ("subthalamic/parasubthalamic nucleus") wherefore no specificity to each of these regions can be outlined.However, analysis of monoamine levels by dialysis revealed increased serotonin levels in the "subthalamic/parasubthalamic nucleus" after optogenetic excitation of serotonergic terminals in this area.In a 3CSRTT test, mice responded to photostimulation with reduced impulsive action, with no effect on anxiety or depression.The authors further found that the dorsal raphe nucleus, rather than medial raphe nucleus, provides the origin of serotonergic neurons projecting to subthalamic area (Ohmura et al., 2020).
In another recent study, Li et al. studied impulsivity upon optogenetic modulation of the STN and also of the pathway from prefrontal cortex (PFC) to STN (PFC-STN) by taking advantage of two transgenic mice lines, the Rbp4-Cre transgenic mouse line (which expresses Cre recombinase selectively in layer V pyramidal neurons of the PFC) and the Nts-Cre transgenic mouse line (which expresses Cre recombinase in glutamatergic STN neurons) (Li et al., 2020).In this study, a delayed go/no-go test was implemented, where the mice were trained to lick in response to an auditory tone.Results showed that optogenetic inhibition of the PFC-STN pathway (Rbp4-Cre mice) leads to an impaired performance, meaning a lower correct rate of response; while optogenetic activation results in a better performance.Further, selective optogenetic activation of the STN (Nts-Cre mice) at 10 Hz resulted in an improved task performance, whereas a 50 Hz photostimulation causes a severe impairment.This finding was in accordance with a study described above (Fife et al., 2017).
In the context of compulsivity, compulsive-like behavior has been observed in rats during escalation of cocaine intake, characterized by abnormal STN neural oscillations in the alpha/theta and low-beta bands (Degoulet et al., 2021).Notably, the precise origin of these oscillations is yet to identify.Oscillations of the STN have also been correlated with compulsive behaviors and compulsive/impulsive related disorders in PD patients (Rodriguez-Oroz et al., 2011) and in OCD patients (Rappel et al., 2018).
Studies have shown that in a quinpirole rat model of OCD, STN-HFS or pharmacological silencing of STN neurons can treat a compulsive phenotype (Winter et al., 2008).This was subsequently confirmed using a compulsive lever-pressing test in a rat "signal attenuation model" of OCD (Klavir et al., 2009).Chang et al. added that STN-HFS can suppress excessive self-grooming in mouse models of autism (Chang et al., 2015).Lastly, as discussed above, optogenetic excitation of the STN, using Pitx2-Cre mice to direct the targeting to the STN, inhibits locomotion and induces repetitive and stereotypic face-grooming, a phenotype associated with compulsivity and OCD (Guillaumin et al., 2021;Serra et al., 2023).A similar correlation to grooming was also observed using optogenetic activation of the STN in Gabrr3-Cre mice, a study which also found that optogenetic inhibition of STN neurons reduces pathological grooming in a Sapap3 KO mouse model of OCD (Parolari et al., 2021).These findings form a strong experimental basis for further exploration of the direct causality between STN and compulsive behavior.
To conclude, STN is strongly implicated in conflict, impulsivity and compulsivity.Regarding the involvement of the STN in impulsive responses, results of rodent studies have revealed conflicting results.In fact, both an improvement and a worsening of performance were observed with optogenetic activation but as a result of stimulation at different frequencies.An improvement in symptoms appears to be linked to the optogenetic activation of neuronal terminals in the STN.While the opposite appears to be true for both optogenetic and chemogenetic inhibition, lesions of the STN appear to reduce impulsivity.These contrasting results are not unexpected considering that even in the clinical setting, some studies have found that impulse control disorders (ICD) can be improved after DBS, while others maintain that DBS causes an increase in ICD.In the context of compulsivity, STN activation is correlated with an increase of compulsive phenotypes, and vice versa, suppression of STN activity is restorative.

The STN in regulation of non-motor function: pain processing.
A growing field of research over the past few years is the role of STN in pain processing (recently discussed in (Al Tannir et al., 2023)).Pain is a key symptom in advanced-stage PD, and pain in PD can be alleviated with STN-DBS (Cury et al., 2014;Kim et al., 2008;Schestatsky et al., 2007;Sürücü et al., 2013).This research field is rather new but there is a degree of consensus about the main findings.
Pautrat et al. made a series of experiments that demonstrated a correlation between the STN and pain (Pautrat et al., 2018).Rats analyzed by electrophysiological recordings while undergoing the hot plate test, showed a noxious response of STN neurons.Certain nociceptive neuronal responses increased in a 6-OHDA PD model.Further, an ibotenic acid-induced lesion of the STN led to an abnormal response to noxious stimuli, increasing the resistance to pain.The authors also found that PBN lesion leads to a decrease in STN nociceptive responses.Based on this result, and through in vivo electrophysiology and tracing experiments in rats, the authors propose a novel noxious pathway, in which nociceptive inputs originating in the PBN reach the STN (Pautrat et al., 2018).
In another study, Luan et al. used optogenetics in wild-type mice upon injection of a virus containing opsin constructs under control of the CamKII promoter.The authors showed that optogenetic activation of STN neurons and their projections to SNr leads to thermal, but not mechanical, pain hypersensitivity, whereas stimulation of STN-GPi and STN-VP projections reduced the mechanical, but not the thermal, pain threshold.Photoinhibition of STN-GPi, STN-VP, and STN-SNr projections increased mechanical and thermal pain thresholds differently.In fact, optogenetic inhibition of STN-SNr projections attenuated hypersensitivity to both mechanical and thermal pain while photoinhibition of STN-GPi and STN-VP projections only attenuated hypersensitivity to mechanical pain.Through in vitro electrophysiological recordings in a 6-OHDA PD mouse model, hyperactivation of STN neurons was recorded and displayed a hypersensitivity to pain that was reversible through optogenetic STN inhibition.Based on this finding, inhibition of the STN was proposed as a possible treatment for PDrelated pain hypersensitivity (Luan et al., 2020).
Lastly, Jia et al. recently used chemogenetics and optogenetics in Vglut2-Cre and Vgat-Cre mice to address the role of excitatory STN neurons (Vglut2) and inhibitory SNr (Vgat) neurons in pain processing.The authors proposed a pathway for pain modulation via GABAergic neurons in the SNr and glutamatergic neurons in the STN and lateral PBN, a SNr-STN-lateral PBN pathway in modulation of pain hypersensitivity.PBN-projecting STN glutamatergic neurons were shown to act in regulation of pain hypersensitivity and pain threshold in different pain states.For example, it was shown that the activation of this pathway induces hypersensitivity, and that its inhibition mitigates both mechanical and thermal sensitivity to pain (Jia et al., 2022).
The studies reported here jointly suggest that the STN is involved in pain perception, with STN neurons responding to nociceptive stimuli.Furthermore, an activation of STN neurons is related to a hypersensitivity to pain with different pathways involved in the perception of thermal and mechanical pain.Accordingly, the inhibition of these same pathways reduces hypersensitivity to pain.These results fit well with clinical observations showing that pain in Parkinson's disease can be alleviated with STN-DBS.Further, inhibition of STN has been forwarded as possible intervention of pain in PD, an exciting field of future research.

Conclusion: STN
Summarizing the STN, several studies over the past decades have led to a deepened understanding of the role of STN in regulation of a various physiological and behavioral functions, and also opened up for a new level of realization that this small brain area, which was long considered a "simple" rely nucleus for motor pathways, possesses great anatomical, molecular, and functional heterogeneity.Reviewing studies of the STN is rodents resembles the peeling of an onion.Layers of different types of studies, reflecting methodology of different time periods, have led to the substantial foundation of knowledge the field can witness today.Based on studies in rodents, a plethora of functional roles and associated neurocircuitry have been revealed, ranging from different aspects of movement regulation, attention, impulse control, compulsion, and motivated behavior in terms of both reward and aversion.With ever increasing complexity, the currently on-going systematic interrogations -using genetic promoters to dissociate STN neurons -will be critical to fully solve the STN.Matching molecularly defined subpopulations with neurocircuitry followed through with decoding of each such circuitry's specific contribution to the different, and sometimes seemingly opposite, functional roles of the STN will be a huge effort but important step forward for both pre-clinical and clinical sciences.Importantly, with recent revelations of molecularly defined subpopulations of neurons within the STN structure come new opportunities for promoter-selective targeting using genetics-based tools in rodents; these types of analysis are still rather few but now emerging.However, another critical aspect of securing gene expressing profiles that vary across the STN structure is the realization that different STN neurons likely have different metabolic needs for their function, a matter of importance to address in further studies, and which likely will bring an additional level of knowledge that may be useful for understanding the STN in both rodents and humans.Yet another aspect to explore further is that of the STN in motor function, while vastly studied, it still remains to fully clarify how discrete STN neurons function to regulate different movement parameters under normal, non-pathological, conditions, aspects that will be important to better grasp disabling conditions that may benefit from improved neuromodulation strategies, such as DBS to alleviate repetitive motor activity, for example in Tourette syndrome.

Beyond the STNrecent studies of the anatomically neighboring pSTN and ZI
Not only have rodent experimental studies high-lighted the STN over the past years, but pSTN and ZI, two distinct anatomical structures with own neurocircuitry and functional roles in brain-body physiology and behavioral regulation have also been carefully assessed, leading to completely new knowledge of these brain areas.Similar as for the STN, a striking heterogeneity has been identified in pSTN and ZI.
Here, we review these studies with the perspective that neurons in these STN-neighboring areas might be important in the context of clinical and experimental manipulations aimed at the STN.However, pSTN and ZI are of great interest in their own right.In addition, ZI itself is already an important target area in clinical interventions, including as DBS target in treatment of e.g.essential tremor.Below, we review the pSTN first, followed by ZI.As for the STN, we place specific emphasis on the method of manipulation implemented in order to direct awareness to spatial selectivity.

The para-subthalamic nucleus, pSTN 1.5.1. Anatomy and distinguishing cellular and molecular features of the pSTN
The pSTN is a differentiation of the lateral hypothalamus (LHA) located immediately adjacent to the ventromedial aspect of STN.This region was characterized 20 years ago in rats by Goto and Swanson who differentiated it from the more rostral regions of the LHA (Goto and Swanson, 2004).Although the first studies were performed in rats, the pSTN has later been described also in mice, non-human primates, and humans (Augood et al., 1999;Barbier et al., 2020;Franklin and Paxinos, 2008;Paxinos, 2009;Paxinos and Watson, 2014;Temiz et al., 2020).
Abundant expression of the Slc17a6 gene, encoding VGLUT2, forwards the pSTN primarily as a glutamatergic structure.The glutamatergic neurotransmitter identity is a characteristic common also to its neighboring structures STN and LHA, and different from the ZI, which is predominantly GABAergic (Chometton et al., 2016;Lein et al., 2007;Shah et al., 2022;Wallén-Mackenzie et al., 2020).Early in situ hybridization studies performed in rats reported the presence of β-preprotachykinin (β-PPT) mRNA in the pSTN.This helped to distinguish the pSTN from the surrounding regions where only scattered neurons are positive for β-PPT mRNA (Goto and Swanson, 2004).A recent study using snRNASeq to characterize the molecular profile of the pSTN in mice identified the mRNA of the Tachykinin gene, Tac1, as a useful marker to identify the pSTN (Wallén-Mackenzie et al., 2020).The Tac1 gene encodes four products of the tachykinin peptide family, including substance P and neurokinin A.
The same snRNASeq study also identified additional pSTN markers including Glra3, which is selectively found in the pSTN but has a weaker expression than Tac1, and Baiap3, which is abundant in the pSTN, absent from STN, but present also in the LHA and surrounding hypothalamic and mammillary structures.Other studies have reported that pSTN encompasses several neuropeptide-positive cells.Among these neuropeptides, pituitary adenylate cyclase-activating polypeptide (PACAP), encoded by the Adcyap1 gene, and Crh are highly abundant in the pSTN, whereas expression of pre-proenkephalin (Penk) and Nts is more scattered (Piggins et al., 1996;Shah et al., 2022;Zhu et al., 2012).
The presence of calcium-binding proteins such as calbindin and calretinin, encoded by Calb1 and Calb2 genes, respectively, has been shown to differ between the pSTN and surrounding structures.Evidence from rodent studies shows that calbindin (Calb1) is distributed over the pSTN and LHA, but absent from the STN (Chometton et al., 2016;Hontanilla et al., 1997;Kim et al., 2022).In addition, Calb1 presents some species-specific differences since its expression is abundant in the rat pSTN but few in the mouse pSTN (Barbier et al., 2020).Studies performed in mice and humans show that calretinin (Calb2) is highly enriched in the pSTN and LHA as wells as in a restricted medial domain within the STN, but scarce in the ZI (Augood et al., 1999;Guo et al., 2020;Kim et al., 2022;Wallén-Mackenzie et al., 2020).
Although Calb2 expression is denser in the pSTN than in the STN, the presence of a Calb2-positive cluster within the medial STN makes it difficult to segregate the STN and pSTN based on Calb2 labeling only.Conversely, another calcium-binding protein of relevance in this context is PV, which is strong in dorsal STN and absent in pSTN (Wallén-Mackenzie et al., 2020).In the context of markers selective for either STN or pSTN, the identification of Tac1 as distinct for pSTN and absent from STN is an important step forward to allow the separation between these anatomically joined areas (Wallén-Mackenzie et al., 2020).In addition to the mouse brain, Tac1 was recently presented to show a similar distinct pSTN expression pattern in the macaque monkey (Serra et al., 2023).In fact, this study showed that several gene expression patterns are highly conserved between mouse and macaque, in addition to Tac1, also Pitx2, Vglut2, and PV (Serra et al., 2023).
Another recent study conducted histological analyses and revealed that neurons positive for Tac1 and Crh mRNAs, respectively, represent two distinct neuronal populations in the pSTN with only sparse overlap of these two markers (Kim et al., 2022).Yet another recent study identified co-localization of PACAP, Tac1, and Crh using immunohistochemical analysis (Nagashima et al., 2022).Calb2-positive neurons were also quantified in the pSTN and account for a majority of pSTN neurons (Guo et al., 2023).Clearly, major interest in the pSTN over the past couple of years have advanced current understanding of its molecular and cellular composition greatly.
Selected gene expression patterns (mRNA) allowing the segregation between pSTN and adjacent areas as well as molecular subpopulations, distinct by mRNA labeling, of pSTN neurons, are shown in Figs. 3 and 4.
In both mice and rats, the pSTN also presents connections with the thalamus.Specifically, pSTN sends unilateral projections to the paraventricular and intermediodorsal nuclei of the dorsal thalamus (Goto and Swanson, 2004;Li and Kirouac, 2012;Zhang and Van Den Pol, 2017).Moreover, the pSTN is a downstream target of the paraventricular nucleus of the hypothalamus, lateral habenula, amygdalopiriform transition area, ventral tegmental area (VTA), and pre-locus coeruleus (Barbier et al., 2020;Chometton et al., 2016;Muzerelle et al., 2016;Santiago and Shammah-Lagnado, 2005;Shin et al., 2011).The connection with the VTA is reciprocated; evidence from tracing experiment show that dopaminergic neurons of the VTA receive its densest hypothalamic input from the pSTN (Watabe- Uchida et al., 2012).
Table 2 contains an inclusive summary of current literature on pSTN studies in laboratory rodents, and the descriptive text below details selected studies.A summary of the findings presented in the studies cited is shown in Fig. 5. 1.5.3.1. Feeding behavior.The pSTN has been found to suppress feeding in the context of satiety and in response to internal anorexigenic stimuli such as sickness-induced anorexia and fear-induced suppression of feeding.Several studies have reported a rapid expression of c-Fos, an indirect marker of neuronal activity, in the pSTN after the ingestion of food (Chometton et al., 2016;Zséli et al., 2016Zséli et al., , 2018)), suggesting that the pSTN could play a role in appetite suppression following satiety.C-Fos expression in the mouse pSTN was also detected following the administration of weight-lowering drugs (Hansen et al., 2021).
Further evidence supporting the role of pSTN in feeding comes from tracing studies.Several studies have shown the bidirectional communication between the pSTN and CeA, as well as the extensive reciprocal connectivity between the pSTN and other feeding-related centers such as hypothalamic arcuate nucleus which contains the anorexigenic, proopiomelanocortin-positive neurons (Barbier et al., 2017(Barbier et al., , 2020;;Bowen et al., 2020;Goto and Swanson, 2004;Sanchez et al., 2022;Zséli et al., 2016Zséli et al., , 2018)).Specifically, the pathway between pSTN and CeA was investigated in a study which revealed that pSTN projects to GABAergic serotonin receptor 2a (Htr2a)-expressing neurons, a neuronal subpopulation in the CeA which modulates food consumption (Douglass et al., 2017).
The paraventricular nucleus of the dorsal thalamus (PVT) is a thalamic structure involved in the control of appetitive and aversive motivation.The PVT is innervated by the pSTN, but this connection is not reciprocated.Elucidating one of the neuronal circuits which might be involved in the satiety network, photoactivation of pSTN glutamatergic neuronal terminals in the PVT inhibited food intake in ad libitum fed Vglut2-Cre mice which had been bilaterally injected with a viral vector encoding excitatory opsins in the pSTN (Zhang and Van Den Pol, 2017).
In addition to satiety, the role of the pSTN has also been investigated in conditions of sickness, induced by injection of lipopolysaccharide (LPS) or cisplatin.Both conditions caused increased c-Fos expression in the pSTN.Tac1-positive neurons were identified as the neuronal subpopulation in the pSTN responsible for feeding suppression when the animal perceives eating as a threat to physiological integrity.Chemogenetic inhibition of the pSTN in Tac1-cre mice reduced sicknessinduced anorexia and mice consumed more sucrose solution compared to controls (Barbier et al., 2020).Similar results were obtained with cell-specific caspase-3 ablation of Vglut2-positive neurons in the pSTN.Ablating this neuronal subpopulation abolished anorexia caused by glucagon-like peptide-1 (Zhang et al., 2022).Feeding is suppressed not only in conditions of sickness, but also when animals experience pain or fear.The pSTN has been found to play a role in mediating fear-induced suppression of feeding.Feeding suppression has been observed by stimulating the pSTN in PACAP-Cre mice and optogenetically activating the PBN terminals in the pSTN.These findings suggest that the PBN-pSTN pathway is involved in the inhibition of feeding behavior and PACAP neurons in the pSTN have a key role in that (Nagashima et al., 2022).

pSTN neuronal subpopulations in feeding behavior.
A recent study elucidated the role played by the Tac1-positive and Crh-positive pSTN neurons in food intake behavior, and confirmed that Tac1expressing neurons in the pSTN are involved in appetite suppression (Kim et al., 2022).Both Tac1-Cre mice and Crh-Cre mice were used in this study with the aim of testing the activity of these two neuronal subpopulations following the administration of anorexigenic hormones.Administration of amylin, cholecystokinin (CCK), and peptide YY (PYY) caused increased activation of both Tac1-positive and Crh-positive neurons as shown by increased c-Fos expression.Increased neuronal activity was also recorded using fiber photometry during administration of anorexigenic hormones in Tac1-Cre mice.To assess whether the eating-suppressing effect of amylin, CCK, and PYY is mediated by Tac1 neurons in the pSTN, chemogenetic inhibition was performed in Tac1-Cre mice.Chemogenetic inhibition of the pSTN in Tac1-Cre mice injected with anorexigenic hormones significantly attenuated feeding suppression.No difference in food intake was observed in chemogenetic inhibited Crh-Cre mice and control group, leading to the conclusion that pSTN Tac1-positive neurons, but not pSTN Crh-positive neurons, mediate the feeding-suppression caused by anorexigenic hormones.In addition, optogenetic and chemogenetic stimulation of both Tac1-Cre and Crh-Cre mice was performed to investigate the effect that a gain of activity of these two neuronal subpopulations has on feeding behavior.Stimulation of Tac1 neurons, but not Crh neurons, reduced food intake, further confirming that Crh-expressing subpopulation does not seem to have a role in appetite suppression (Kim et al., 2022).

pSTN in the satiety circuitry.
A study investigating the circuitry mediating the effect of the anorexigenic hormone CCK confirmed the results obtained by Kim et al.Part of the neural pathway involved in the eating-suppressing effect of CCK has already been established in literature: The pathway starts from the vagus nerve and involves NTS, PBN, and CeA (Becskei et al., 2007;D'Agostino et al., 2016;Rinaman, 2010).In particular, a specific neuronal subpopulation within the CeA, marked by protein kinase C-delta (PKC-δ), has been found to be preferentially activated by CCK (Cai et al., 2014).The neural circuit downstream of the CeA responsible for regulating CCK-mediated eating suppression has been identified by Sanchez et al.Their findings show that PKC-δ neurons inhibit the pSTN-projecting CeA neurons thus leading to a disinhibition of the pSTN which mediates the anorexigenic effect of CCK.Chemogenetic inhibition of the pSTN attenuated CCK-mediated eating suppression, in agreement with the results obtained by Kim et al. (Sanchez et al., 2022).
These two latter studies confirmed the role played by the pSTN in the satiety network and its function in the neuronal circuit mediating the food intake-suppressing effect of CCK.In addition, the authors also identified some of the neuronal subpopulations involved in this mechanism.Further studies are required to assess whether the pSTNprojecting CeA neurons inhibited by CeA PKC-δ neurons project to pSTN Tac1-positive neurons and whether these projections are exclusive to these Tac1 neurons or encompass also other neuronal subpopulations within the pSTN.
Although Tac1-expressing neurons in the pSTN seem to be the neuronal subpopulation mainly involved in the regulation of feeding behavior, a role of pSTN Crh neurons in eating cannot be ruled out.Increased activity, represented by higher levels of c-Fos expression, has been observed in both Tac1-positive and Crh-positive neurons of the pSTN in response to food exposure and administration of anorexigenic hormones (Kim et al., 2022).Furthermore, another study investigated the circuitry between the anterior piriform cortex and pSTN Crh-positive neurons and its involvement in mediating the anorexigenic response caused by a valine-deficient diet (VDD) (Zhu et al., 2012).A diet lacking this indispensable amino acid has been found to induce anorexia in mice and a marked induction of Crh mRNA in a pSTN neuronal subpopulation of these animals.VDD mice also presented increased expression of c-Fos in the pSTN compared to control animals.Anterior piriform cortex is the brain region responsible for detecting indispensable amino acid deficiency.Anterograde tracing from the anterior piriform cortex revealed efferent projections in close contact with Crh neurons within the pSTN (Zhu et al., 2012).1.5.3.4.Aversion and fear.The pSTN has been found to be engaged in the encoding of aversive experiences.c-Fos expression in the pSTN has been observed following conditioned taste aversion retrieval in an experiment which paired administration of sucrose with an intraperitoneal injection of lithium chloride, known to induce visceral sickness.Interestingly, pSTN was not activated when the rats were unconditioned exposed to a bitter solution of quinine.This seems to suggest that the pSTN is involved in the retrieval of an aversive visceral experience, but do not respond to gustatory aversion (Yasoshima et al., 2006).Activation of the pSTN has been reported also when mice were exposed to innate fear stimuli such as predator odor.In particular, the predator odor induced hypothermia and the pSTN has been hypothesized to be directly involved in this predator odor-evoked thermoregulatory response (Liu et al., 2021).
The connection between pSTN and PVT, suggested to have a role in feeding behavior, seems to be also involved in aversion.Optogenetic activation of axon terminals in the PVT of Vglut2-Cre mice injected bilaterally with a viral construct encoding excitatory opsins in the pSTN led to a significant aversion associated with the light-paired chamber in a two-chamber place-preference test (Zhang and Van Den Pol, 2017).The same response was observed photoactivating calcitonin gene-related peptide (CGRP) -positive neurons of the PBN projecting to the pSTN in a real-time, place-preference assay.Photostimulation of PBN terminals in the pSTN generated robust aversion in mice that thereby avoided the light-paired compartment.In addition, activation of this pathway also caused freezing behavior (Bowen et al., 2020).Another study confirmed real-time avoidance in addition to aversive learning induced by lateral PBN-pSTN optogenetic stimulation.The same tests were performed on PACAP-Cre mice to assess the role of PACAP-positive neurons, a population highly enriched in the pSTN, in avoidance behavior.The results revealed that PACAP-expressing neurons in the pSTN induce avoidance behavior and play a key role in aversive memory formation (Nagashima et al., 2022).
The role of the pSTN in mediating aversive behavior was also addressed in a recent study by Serra et al. in which the roles of the STN and pSTN in aversion and avoidance behavior were compared (Serra et al., 2023).Indeed, this was the first study to identify a role for the STN in aversive learning; described in detail above.To firmly ascertain any potential impact of the anatomically associated pSTN, the results obtained in optogenetic approaches implementing Pitx2-Cre (STN) and Vglut2-Cre (STN) mice were compared with Tac1-Cre mice, representing

Table 2
The pSTN: List of publications with summary of main findings.Calb2, calretinin; CeA, central nucleus of the amygdala; CCK, cholecystokinin; DRN, dorsal raphe nucleus; F, female; Htr2a, serotonin receptor 2a; LPS, lipopolysaccharide; M, male; N/A, not applicable; PACAP, pituitary adenylate cyclase-activating polypeptide; PBN, parabrachial nucleus; pSTN, para-subthalamic nucleus; PVT, paraventricular thalamus; Vglut2, vesicular-glutamate transporter 2; Tac1, tachykinin precursor 1. pSTN.Optogenetic stimulation of either STN or pSTN resulted in real-time avoidance.However, only STN stimulation, and stimulation of STN terminals in VP, caused aversive learning.No aversive learning was observed upon photostimulation of pSTN.Further, in a positive reinforcement paradigm, only STN photostimulation was able to interrupt sugar self-administration, while pSTN photostimulation did not negatively influence sugar consumption.In general, when Tac1-Cre-positive neurons of pSTN were optogenetically stimulated, mice showed a behavior that resembled appetitive, rather than avoidance, behavior (Serra et al., 2023).Concluding, the past few years has seen an advancement in the knowledge of the pSTN which apparently exerts strong impact on both appetitive and aversive behavior.
1.5.3.5.Novelty.The role of the pSTN has been investigated also in conditions of novelty.Novelty might encompass an aversive component.When rodents are exposed to a new food they experience noveltyinduced feeding suppression i.e. neophobia.Barbier et al. performed a neophobic experiment with sucrose solution as a new tastant.Neophobic rats consumed significantly less than rats habituated to drinking the sucrose solution, and, as measured by c-Fos expression, the pSTN was activated in neophobic rats.To better assess whether the pSTN is involved in mediating the neophobic response, experimental manipulation of the pSTN was performed in Tac1-Cre mice.Chemogenetic inhibition of Tac1-positive neurons in the pSTN suppressed the neophobic response, thereby reinstating consummatory behavior (Barbier et al., 2020).Kim et al. recorded the activity of Tac1-expressing neurons in a fiber photometry experiment where Tac1-Cre mice were exposed to a novel object.They observed no increased activity in the pSTN of these mice (Kim et al., 2022).Therefore, it is possible that the pSTN plays a main role in food novelty rather than object novelty.1.5.3.6.Wakefulness.Early evidence of an association between pSTN activity and arousal and exploration comes from a study reporting that increased c-Fos was measured in the pSTN of rats that explored a new cage for 1 h before being sacrificed (Chometton et al., 2016).Increased c-Fos expression was also observed in the pSTN of rats performing predatory hunting, an activity which is strictly related with arousal and exploration (Comoli et al., 2005).A more recent study revealed that the Calb2-expressing subpopulation in the pSTN might be involved in controlling wakefulness (Guo et al., 2023).The activity of Calb2 neurons in the pSTN recorded with fiber photometry in Calb2-Cre mice increased during wakefulness and with exploratory behaviors.Chemogenetic and optogenetic manipulations demonstrated that activation of Calb2 neurons in the pSTN induced arousal and maintained wakefulness.Specifically, optogenetic excitation of Calb2-positive neurons was sufficient to generate an immediate transition from non-REM (NREM) sleep to wakefulness.Furthermore, the activation of Calb2-positive neurons also increased locomotion and exploratory behaviors.Conversely, chemogenetic inhibition and lesioning of the Calb2-expressing neurons of the pSTN resulted in a reduction in wakefulness and increased NREM and REM sleep in both dark and light phase (Guo et al., 2023).The circuitry in which the pSTN operates to regulate arousal involves the VTA and PBN.Optogenetic stimulation of Calb2-positive pSTN neuronal terminals in the VTA or PBN produces immediate and sustained wakefulness.In addition to increased wakefulness, the activation of the pSTN -VTA pathway seems also to promote locomotion and exploratory behavior (Guo et al., 2023).

Conclusion: pSTN
Concluding, the pSTN likely deserves attention in the context of DBS aimed at STN; the fact that no specific anatomical boundary separates the medioventral STN from pSTN is intriguing when considering some of the reported side-effects of DBS.Based on recent revelations of the roles played by pSTN neurons, we tentatively propose that depression, anxiety and obesity are examples of clinical side-effects of DBS that could be related to alterations in pSTN circuitry of reward vs. aversion, feeding, novelty response, and more.Further studies will be needed to uncover neurobiological underpinnings of side-effects upon DBS, and such studies might gain from considering the pSTN which anatomically forms a continuum with the STN.As described above, rodent-based studies give a clear indication that pSTN might be a critical brain area with impact on outcome in STN-directed neuromodulations.In addition, based on the important functional outcome of pSTN stimulations as reported in the experimental studies of rodents discussed above, the pSTN could be further explored as a potential new DBS target, for example as intervention of eating and mood disorders.
As shown in rodent-based studies focused around neurocircuitry analysis with high spatial resolution over the past few years, the pSTN exerts strong impact on physiology and behavior.Despite its small size, the pSTN has recently been shown to be composed of several moleculary distinct subpopulations of neurons, giving rise to a striking heterogeneity which likely accounts for its vast influence on autonomic functions and motivated behavior, acting as a key structure of interoception.Taking advantage of selective promoters in rodents to gain access to pSTN subpopulations, further exploration of this enigmatic brain nucleus with emphasis on its anatomical, molecular and functional complexity, and by bridging findings from rodents with primate data (including clinical reports of brain stimulations in the vicinity of pSTN), the pSTN will likely soon be far more recognized than as an elongation of the hypothalamus, moving away from being an almost "blind spot" in an atlas.
1.7.Zona incerta, ZI 1.7.1.Anatomy and distinguishing cellular and molecular features of ZI ZI was first described by August Forel as "zone of uncertainty" (Forel, 1877).During the past century, some of the key features of this area have been unraveled, implicating ZI in a wide range of functions.ZI is located between the dorsal thalamus and STN, adjacent to the LHA and aligning the MFB.ZI has mainly been studied in rats, but has also been described in mouse, non-human primates, and human brain (Franklin and Paxinos, 2008;Ma et al., 1992;Mitrofanis et al., 2004;Paxinos, 2009;Paxinos and Watson, 2014;Watson et al., 2014).Considering brain size, ZI is relatively larger in the rodent brain than in the primate brain (Watson et al., 2014).In rodents, ZI occupies a larger brain area than STN and pSTN (Fig. 2).This structure is also characterized by substantial cellular and neurochemical diversity, represented by heterogeneous cytoarchitecture, expression of multiple molecular markers, and widespread connectivity (Mitrofanis, 2005).Ultimately, these features denote an area involved in a diverse set of functions that may serve as an integrative node for modulation of various behaviors.
Cells forming the ZI structure differ considerably in shape and size.Cells similar in shape and size form clusters, allowing ZI to be divided into different sectors.Early studies have reported a number of sectors varying from two to six in rats (Kawana and Watanabe, 1981;Kuzemenský, 1977;Romanowski et al., 1985;Watanabe and Kawana, 1982), and from two to four in monkeys (Ma et al., 1992).More recently, authors have argued that ZI can be divided into four sectors: rostral sector of ZI (ZIr), ventral sector of ZI (ZIv), dorsal sector of ZI (ZId), and caudal sector of ZI (ZIc) in both rodents and primates (Kim et al., 1992;Kolmac and Mitrofanis, 1999a;Mitrofanis et al., 2004;Nicolelis et al., 1992).
Further, ZI is positive for several neurochemical substances that have been used as markers to identify different neuronal subpopulations.Some of these molecular markers are PV, nitric oxide synthase (NOS), GABA, glutamate, somatostatin (SST), tyrosine hydroxylase (TH), and Calb1 (Kolmac and Mitrofanis, 1999a;Mitrofanis et al., 2004;Mitrofanis, 2005).Certain neurochemically defined cells such as those defined by PV and NOS are useful to define sector boundaries since they are principally located in ZIv and ZId, respectively (Kolmac and Mitrofanis, 1999a;Nicolelis et al., 1995).Concerning other markers, although they tend to be more expressed within a certain sector, they can also be found in other areas of ZI (Mitrofanis, 2005).This molecular heterogeneity makes it difficult to identify clear borders between sectors based on the neurochemically associated gene expression patterns.
Selected gene expression patterns (mRNA) allowing the segregation between ZI and adjacent areas, as well as molecularly defined ZI subpopulations, are shown in Figs. 3 and 4.

Functional roles of ZI
ZI is a structure involved in a heterogeneous set of functions.Early studies revealed its involvement in mainly visceral activities, arousal, attention, and locomotion (Chometton et al., 2017;Mitrofanis, 2005).Some of the results reported in these studies were obtained upon chemical or electrical lesioning of the ZI structure, an approach with limitations.Lesions could have affected the activity of the many crossing fibers described in the ZI, incorrectly attributing functions to this area.Still, these early studies provided important insights into the different functions potentially executed by the ZI (Mitrofanis, 2005).
Table 3 contains an inclusive summary of current literature on ZI studies in laboratory rodents, and the descriptive text below details selected studies.A summary of the findings presented in the studies cited is shown in Fig. 5. 1.7.3.1. Locomotion.Rodent studies have shown the role of the ZI in locomotion, and also suggested that ZI GABAergic synapses might be involved in parkinsonian symptoms (Hormigo et al., 2023;Milner and Mogenson, 1988;Mogenson et al., 1985;Mogenson and Wu, 1986;Parker and Sinnamon, 1983;Périer et al., 2002;Skinner and Garcia-Rill, 1984;Supko et al., 1991Supko et al., , 1992;;Wardas et al., 1988).Evidence of the involvement of ZI in PD came also from human studies reporting that lesioning of the ZI led to anti-parkinsonian effects (Mundinger, 1965;Patel et al., 2003).These findings were supported by results reported in DBS studies that revealed that placing the DBS electrodes dorsally to the STN, in an area that also includes the ZI, presented clinical benefits (De Chazeron et al., 2016;Herzog et al., 2004;Welter et al., 2014;Yokoyama et al., 2001).Further evidence demonstrated that a reduction of muscle rigidity and tremor in PD patients could be achieved positioning the one DBS lead inside the STN and the second lead at the border of the ZI (Henderson et al., 2002).Moreover, clinical advantages were also achieved performing DBS in regions located dorsally to the STN (Lanotte, 2002;Voges et al., 2002).
Based on these findings, ZI was implemented as a target for DBS treatment of tremor-dominant PD; in ZI-DBS, the caudal ZI region was found to be most effective (Kitagawa et al., 2005;Plaha, 2006).Although the ZI is now considered an important neuromodulation target  for treatment of motor symptoms in PD, the exact mechanism of action underlying the therapeutic effect of ZI-DBS is still debated.To date, it cannot be excluded that the beneficial clinical effects of ZI-DBS might be due to spread of current to the STN (Ossowska, 2020).Further studies in experimental rodents, in which high level precision can be achieved by the use of spatially selective promoters, might help clarify the impact of STN vs. ZI in DBS strategies.1.7.3.2.Defensive behavior.ZI has been implicated in the modulation of defensive behavior via its projections to the PAG (Chou et al., 2018).Early studies reported projections from ZI to PAG (Beitz, 1989;Grofová et al., 1978), a structure known to mediate various types of defensive behavior (Bandler and Shipley, 1994).More recently, a tracing experiment targeted ZI neurons expressing glutamic acid decarboxylase 2 (Gad2), a GABA cell marker, injecting an anterograde tracer in the ZI of Gad2-Cre mice.The results showed that primarily ZIr sends strong axonal projections to PAG (Chou et al., 2018).
Optogenetic manipulation of Gad2-Cre mice injected with a viral vector encoding excitatory opsins into ZIr showed that activation of this area reduced the flight response induced by noise, whereas inhibition of ZIr produced the opposite result leading to an increase in flight response (Chou et al., 2018).
Optogenetic manipulation had no effect on the onset latency of the flight response, suggesting that ZIr exerts only a regulatory role on this aspect of defensive behavior.Optogenetic activation of ZIr terminals in PAG, while silencing ZIr neuronal cell bodies, reduced noise-induced flight response, thus confirming that ZIr modulate flight response via its projections to PAG.The connection between ZIr and PAG has been  A. Ricci et al. further characterized revealing that GABAergic PAG-projecting ZIr neurons send inhibitory inputs only to glutamatergic PAG neurons (Chou et al., 2018).
The modulatory role of ZI in defensive behavior was confirmed also by another study which investigated the connection between ZI and the midbrain tegmentum (Hormigo et al., 2020).According to these findings, ZI can regulate avoidance, but it does not drive the behavior.This was assessed in Vgat-Cre mice injected with vectors encoding excitatory or inhibitory opsins, and tested in a signaled active avoidance task.Optogenetic manipulation of the ZI had an effect on the avoidance response only in mice that had learned to predict a harmful outcome following light stimulation and/or the presentation an auditory stimulus.Specifically, optogenetic inhibition of ZI GABAergic neurons in Vgat-Cre mice tested in a signaled active avoidance task elicited active avoidance responses.Conversely, optogenetic activation of the ZI in Vgat-Cre mice injected with a viral vector encoding ChR2 as excitatory opsin led to a suppression of signaled active avoidance responses.Mice naïve to the test who received optogenetic inhibition in the ZI did not show any avoidance response, thus demonstrating that ZI plays only a regulatory role (Hormigo et al., 2020).Recent findings further confirm that ZI is not required to drive avoidance behavior (Hormigo et al., 2023).Lesioning the ZI of Vgat-Cre mice with a viral vector expressing diphtheria toxin subunit A had little impact on their performance in signaled active avoidance tasks (Hormigo et al., 2023).
PV-positive neurons, which are particularly abundant in ZIv, have also been implicated in regulating defensive behavior.In particular, this neuronal subpopulation has been found to drive the defensive behavioral response by integrating tactile-auditory inputs.Integration of various somatosensory signals is important in the context of defensive behavior, as threats might be perceived through multiple sensory modalities.Wang et al. showed that tactile stimulation elicited by whisker deflection can enhance sound-induced flight.Chemogenetic silencing of ZIv in PV-Cre mice prevented the tactile enhancement of flight behavior, thus confirming the role of PV neurons in mediating this response.Single-unit optrode recordings further supported this finding by showing that PV neurons respond to both noise sound and whisker deflection.Taken together, these results demonstrate that ZIv PV neurons modulate defensive behavior by integrating tactile and auditory inputs (Wang et al., 2019).
The sensory integration mechanism has been observed also in a study which investigated the role of ZI in predatory hunting (Zhao et al., 2019).ZI receives inputs from structures involved in different sensory pathways and presents extensive connections throughout the brain (Roger and Cadusseau, 1985).Based on the vast connectivity, it has been reasoned that ZI might function as an integrative brain node, regulating different types of behavioral outputs (Wang et al., 2020).In addition to receiving and integrating inputs from different upstream areas, ZI also sends inhibitory top-down projections to neocortex interneurons (Schroeder et al., 2023).In particular, recent evidence has shown that projections originating from ZI GABAergic neurons preferentially target interneurons located in layer 1 of the auditory cortex transmitting integrated information that is essential for learning.The incertocortical impact on learning was assessed in a discriminative threat conditioning paradigm combined with chemogenetic inhibition of ZI synapses in the auditory cortex.Results from this experiment showed that the formation of long-term auditory threat memory is mediated by ZI-auditory cortex projections.Furthermore, learned top-down relevance is encoded in a bidirectional fashion with some boutons developing excitatory potentiation and others displaying inhibitory potentiation (Schroeder et al., 2023).
Wang et al. performed optogenetic manipulation of PV neurons in ZIv showing that activation of PV neurons enhanced the flight response induced by noise stimulation, whereas inhibition of PV neurons induced the opposite effect reducing the flight behavior (Wang et al., 2019).These results confirm the role of ZI in modulating defensive behavior, but also contrast the findings of Chou and colleagues who showed that activation of ZIr GABAergic neurons reduced noise-induced flight (Chou et al., 2018).This apparent discrepancy might be explained by the heterogeneity of ZI sectors.Different sectors or cell types might be engaged in different pathways and this can account for the different roles played in defensive behavior.ZIv differs from ZIr because it is an area rich in PV neurons and sends few axonal projections to PAG.In addition, ZIv projects to the medial posterior complex of the thalamus which has been proposed as downstream target mediating the enhancement of sound-induced flight behavior by tactile stimulation (Wang et al., 2019).
PV-expressing neurons of ZI have also been implicated in fearconditioning memory acquisition as well as remote fear retrieval.In PV-Cre mice, bilateral silencing of PV neurons in the ZI, achieved through the use of double-loxP-flanked Cre-dependent tetanus toxin light chain (TetTox), impaired both freezing and remote fear-memory retrieval in a conditioned freezing response paradigm.Tracing and electrophysiological experiments revealed that SST-positive neurons in the CeA, an essential region for fear conditioning (Fanselow and Poulos, 2005), project to PV neurons in the ZI and that this connection is inhibitory.TetTox inactivation of the CeA-ZI projection impaired both memory acquisition and remote fear memory, thus demonstrating that the inhibitory projection of CeA SST neurons to ZI PV neurons is essential for fear memory (Zhou et al., 2018).
Another report of the involvement of ZI in conditioned freezing response comes from the study of Chou et al. in which the authors paired a conditioned stimulus (CS) with foot shock and measured freezing time in Gad2-Cre mice which received either optogenetic activation or inhibition of the ZIr.Activation of GABAergic neurons in the ZIr led to a reduction in freezing time during CS presentation, while freezing time increased when ZIr neurons were inhibited.These data further demonstrate the modulatory effect exerted by ZI whose activity can dampen and enhance conditioned freezing response.PAG has been identified as the downstream target responsible for driving the freezing response.Optogenetic activation of ZIr terminals in PAG reduced the conditioned fear response.PAG can be further divided into dorsolateral PAG (dlPAG) and ventrolateral PAG (vlPAG), different compartments that drive flight and freezing behavior, respectively (Tovote et al., 2016;Xiong et al., 2015).Taken together, the results obtained by Chou et al. demonstrate that ZIr modulates both innate and learned defensive behaviors via its projections to the different PAG compartments (Chou et al., 2018).
The studies discussed above investigated learned defensive behavior in conditioned fear paradigms where fear was expressed toward threatassociated stimuli such as foot shock.However, fear response can become generalized towards non-threatening cues, a condition linked with trauma and anxiety related disorders.ZI has been found to play a role in modulating fear generalization.ZI showed decreased neuronal activity in animals that generalized fear, as demonstrated by reduced c-Fos expression.In addition, chemogenetic silencing of the ZI resulted in fear generalization, whereas chemogenetic activation evoked the opposite effect by suppressing fear generalization (Venkataraman et al., 2019).A follow-up study investigated the connectivity between ZI and the thalamic reuniens, an area playing an important role in emotional regulation, to explore the neurocircuitry underlying fear generalization (Ramanathan et al., 2018;Ramanathan and Maren, 2019;Troyner et al., 2018;Venkataraman et al., 2021).Optogenetic stimulation of GABAergic ZI projections in the thalamic reuniens led to a reduction of fear generalization in Vgat-Cre mice.The same experiment was performed on tyrosine hydroxylase (TH) -Cre mice, injected with excitatory opsin constructs in ZI to assess whether dopaminergic neurons in ZI, which are a subset of GABAergic neurons, also play a role in suppressing fear generalization.Optogenetic stimulation of dopaminergic projections from the ZI to thalamic reuniens did not inhibit fear generalization, but it had an effect in increasing extinction recall (Venkataraman et al., 2021).
Evidence also supports a role of the ZI in anxiety (Li et al., 2021).ZI neurons are activated under stressful and anxious conditions as A. Ricci et al. demonstrated by increased c-Fos expression and calcium activity.The response to anxious events was assessed in three different ZI neuronal subpopulations (SST-, Calb2-, and Vglut2-expressing neurons) which exhibited distinct behavioral phenotypes upon optogenetic manipulation.In particular, optogenetic activation of SST neurons in the ZI of SST-Cre mice evoked anxiogenesis, while the activation of the ZI in Calb2-Cre mice had opposite effects inducing anxiolysis and exploratory rearing.Finally, optogenetic activation of ZI glutamatergic neurons in Vglut2-Cre mice elicited flight-like jumps, thus demonstrating that ZI subpopulations can differentially modulate anxiety-related responses (Li et al., 2021). 1.7.3.3.Predatory hunting.ZI has been implicated in predatory hunting.A role of ZI in hunting behavior was hypothesized due to its connectivity with superior colliculus and PAG, two regions known to be involved in hunting (Furigo et al., 2010;Mota-Ortiz et al., 2012).Optogenetic inactivation of GABAergic neurons in the medial ZI (ZIm) in Gad2-Cre mice and Vgat-Cre mice has been shown to impair prey attack.Conversely, optogenetic activation of ZIm GABAergic neurons increased efficiency in predatory hunting and evoked bite attacks in live and artificial preys (Shang et al., 2019;Zhao et al., 2019).As discussed above, ZI is made up of heterogeneous groups of neurons.Predatory behavior seems to be driven specifically by GABAergic neurons of ZIm since optogenetic activation of glutamatergic or PV-positive neurons produced little effect on hunting (Zhao et al., 2019).Another important role played by GABAergic ZI neurons is the integration of vibrissal, visual, and auditory signals from cortex and superior colliculus.Sets of different neurons within ZI have been found to respond to different sensory signals, but there is a large degree of mutual overlapping between these cells indicating the integration of prey-related multisensory signals (Zhao et al., 2019). 1.7.3.4.Feeding behavior.Circuits underlying hunting behavior can be separated from those responsible for eating.Evidence shows that different sectors of ZI contribute to the circuitry responsible for hunting and eating.Tracing studies revealed that ZIm projects preferentially to PAG, a pathway that has been shown to be required for predatory hunting (Zhao et al., 2019).Zhang and Van Den Pol showed that optogenetic stimulation of ZI GABAergic neurons in Vgat-Cre mice or their axonal projections to PVT evoked binge-like eating.Conversely, ZI GABAergic neurons ablation had the opposite effect as demonstrated by a reduction in the weight of the mice (Zhang and Van Den Pol, 2017).The same results were obtained ablating dopaminergic neurons in the ZI (Ye et al., 2023).Furthermore, similar results to Zhang and Van Den Pol were obtained also with the optogenetic activation of dopaminergic projections to the PVT.In fact, excitation of dopaminergic neurons in the ZI of TH-Cre mice triggered an increase in food-seeking and consumption (Ye et al., 2023).Another pathway involved in the neurobiology of feeding behavior is the ZI projecting to the VTA (De Git et al., 2021).Results from the inactivation of this pathway in rats showed that they reduced ad libitum feeding and their motivation to work for food rewards.Conversely, chemogenetic activation of ZI neurons projecting to the VTA promoted food-motivated behavior, increasing food reward-seeking and the amount of food consumed (De Git et al., 2021).In particular, the activation of these neurons induces sleep by inhibiting wake-promoting hypocretin (Hcrt)-expressing neurons in the lateral hypothalamus (Liu et al., 2017).1.7.3.6.Motivation.Besides survival-related activities, ZI has been found to encode positive motivational valence.Optogenetic activation of ZI in Vgat-Cre mice resulted in a strong preference for the light-paired compartment and in positive conditioning in a real-time place preference test (Zhang and Van Den Pol, 2017;Zhao et al., 2019).Further, neural activation of the ZI induced a repetitive seeking for stimulation as demonstrated by persistent nose-pokes to receive laser-stimulation, indicating a strongly reinforcing motivational drive (Zhao et al., 2019). 1.7.3.7.Pain.Finally, ZI has been implicated in the modulation of pain.As part of the spinothalamic pathway, ZI has been investigated for its role in the central pain syndrome.Evidence suggests that this debilitating condition results from an increased activation of the posterior thalamus which receives abnormally reduced inhibitory inputs from the ZI (Masri et al., 2009).Suppressed GABAergic signaling in ZI has been found to cause neuropathic pain in injury models in the rat, further confirming that dysregulation of ZI activity affects how pain information is conveyed (Moon et al., 2016;Moon and Park, 2017).On the other hand, increasing the activity of neuronal cells in the ZI leads to anti-nociception.This was shown by stimulation of the ZI with local injection of glutamate which resulted in increased tail-flick latency in a tail-flick test and inhibited post-incision pain in rats (Petronilho et al., 2012).Recent findings revealed that ZI is also involved in the modulation of itch processing, an uncomfortable sensation considered strongly entangled with pain due to the similarity in brain circuitry regulating itch and pain (Dong and Dong, 2018;Koch et al., 2018).Chemogenetics was used to manipulate the activity of ZI PV neurons in PV-Cre mice.Pharmacogenetic activation of ZI PV neurons in mice injected with histamine and chloroquine to induce acute itch reduced scratching behavior.The same results were replicated in PV-Cre mice subjected to chronic itch (Li et al., 2022).

Conclusion: ZI
In conclusion, ZI is a heterogenous structure composed of several neuronal subpopulations with widespread connections.Advances in genetic tools such as selective promoters led to discovery of functional roles of ZI subpopulations, findings that support the idea that ZI may integrate diverse sensory inputs and modulate a variety of behaviors such as feeding, defensive behavior, predatory hunting, motivation, pain, and sleep.These rodent-based results might be relevant also in the context of DBS.ZI is already an important target area for DBS leads in patients suffering from essential tremor, for which ZI-DBS has a high success rate (Blomstedt et al., 2011;Holslag et al., 2018;Sandvik et al., 2012;Wong et al., 2020a).In PD, ZI-DBS alleviates motor symptoms such as bradykinesia, tremor, and muscle rigidity (Blomstedt et al., 2012(Blomstedt et al., , 2018;;De Marco et al., 2020;Ossowska, 2020).Further understanding of ZI and its neuronal subpopulations might be important to improve DBS strategies further.

Concluding remarks
Recent clinical advances position the STN at the heart of neuromodulation strategies aiming to correct faulty firing patterns and to relieve patients of disabling symptoms ranging from bradykinesia in PD, dystonia, essential tremor, and repetitive motor dysfunction (known as tics) in Tourette syndrome, and also obsessions and compulsions in OCD.For example, one recent publication pinpoints so called "sweet spots" for DBS leads within different anatomical positions of the STN, distinct for each of these severe conditions via segregated frontal lobe circuitry (Hollunder et al., 2024) (Fig. 6).In parallel, experimental neuroscience in rodents has contributed several layers of new knowledge of the multiple roles played by the STN by providing neurobiological underpinnings to behavioral regulation as well as cognitive and affective effects mediated by the STN, alongside complex neurocircuitry mechanisms and striking molecular heterogeneity of the internal organization of the STN.Similar experimental advancements have begun to uncover the closely located ZI and pSTN, two until recently rather poorly explored brain regions but which based on studies primarily on rodents have been demonstrated to possess unexpected molecular and anatomical heterogeneity, and to participate in neurocircuitry of vital physiological processes (Fig. 5).
Due to the anatomical proximity of STN, pSTN and ZI in both rodents (Figs.2-4) and primates (Fig. 6), we wished here to highlight the multitude of functions engaging each of these three brain areas as revealed by experimental approaches in laboratory mice and rats, by the implementation of methodology that allows the interrogation of discrete neuronal structures at high spatial and temporal resolution.By evaluating rodent data, new perspectives may arise that help explain neuropathology in different conditions in human individuals.Rodent studies have identified a striking diversity in STN, pSTN and ZI that today allows an enhanced understanding of the vastly complex neurocircuitry engaging these brain areas, and how they impact on physiology and behavioral regulation.Recent advances have led to the identification of unexpected molecular and functional heterogeneity within each of these areas, and different types of subpopulations have been demonstrated.Anatomical-functional decoding on subpopulation level has already been initiated, led by transgenics-based approaches that enable molecularly defined neurons to be selectively identified, targeted and functionally interrogated with high spatial and temporal precision.
With new knowledge gained in rodents, the impact of STN, pSTN and ZI in normal physiology has become clearer, bringing attention also to  1a in the recent publication "Mapping dysfunctional circuits in the frontal cortex using deep brain stimulation" (Hollunder et al., 2024).
A. Ricci et al. their importance in neurological and neuropsychiatric conditions.Further, with the diverse functional roles mediated by this narrow brain area, the idea that any of these areas might be unintentionally affected in neuromodulation strategies such as DBS is tentatively supported, such that for example the role of pSTN in appetitive behavior might contribute to weight alterations upon STN-DBS.While intending to stimulate specific domains within an area, such as aiming for the "sweet spots" within the STN structure when treating patients with PD, OCD, essential tremor, dystonia, or Tourette disorder (as illustrated in Fig. 6), positive effects of the treatment, but also adverse side-effects, might depend on induced activity of these neighboring structures.Interconnectivity between neurons and circuitries within -and betweenthe STN, the pSTN, and ZI structures should deserve further investigation.Alleviation of co-morbid symptoms of PD and OCD was initially a remarkable effect of combined stimulation of STN and ZI, a clinical observation which opened up for the intervention of OCD using DBS, as discussed above (Mallet et al., 2002), but also forwarded ZI as DBS target.Based on today's knowledge of pSTN, with its newly exposed roles in appetitive and re-feeding responses alongside additional physiological functions, it is not impossible, but still speculative, to imagine that also this area might provide new opportunities as a target in DBS strategies, perhaps directed at eating disorder.
A major leap forward would be the bridging of the gap between clinical and pre-clinical knowledge of the STN, pSTN and ZI.For example, one might anticipate that molecular knowledge gained from carefully controlled rodent studies, which now allows the decoding of subpopulations of neurons within each of these three brain areas, could find a way to enhance the neurobiological understanding of the optimal target sites for DBS, such that each of these clinically most important neuromodulation targets is fully decoded in terms of molecular needs for optimized function, as well as neurocircuitry mechanisms and their role in physiology, behavioral execution, and higher brain function, including emotion and cognition.

Fig. 2 .
Fig. 2. 3D visualization of the mouse brain highlighting the location of STN, pSTN and ZI.STN (orange), pSTN (green) and ZI (violet) from different perspectives: A, frontal; B, left side; C, top; D, rear.STN, subthalamic nucleus; pSTN, para-STN; ZI, zona incerta.Illustration created using Brainrender, a Python tool (Claudi et al., 2021).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A
.Ricci et al.

Fig. 5 .
Fig. 5. Summary of functional roles played by STN, pSTN, and ZI as revealed by studies in rodents, and cited in the present review.A. Functions mediated by the STN, pSTN, and ZI.B. Transgenic Cre recombinase mouse lines implemented in different laboratories to selectively target the STN, pSTN, or ZI.C-E.Functions revealed using Cre recombinase mouse lines targeting either the STN (C), pSTN (D), or ZI (E) as described in current literature and cited in Tables 1-3 and in main text.
1.7.3.5.Sleep.A GABAergic neuronal subpopulation of ZI characterized by the LIM homeodomain factor 6 (Lhx6) has been implicated in promoting sleep.As shown by c-Fos analysis, Lhx6-positive neurons localized in ZIv are activated by homeostatic sleep drive.Chemogenetic activation and inhibition of Lhx6-positive ZIv neurons modulate sleep.

Fig. 6 .
Fig. 6.STN is an important neuromodulation target for alleviation of symptoms in several brain disorders.A. Anatomical localization of the STN, pSTN, and ZI in the human brain.Gyral annotation, image 48.© 2010 Allen Institute for Brain Science.Allen Human Brain Atlas.Available from: human.brain-map.org.B. Optimized deep brain stimulation (DBS) electrode placement within the STN for the treatment of dystonia (DYT), Tourette's syndrome (TS), Parkinson's disease (PD), and obsessive-compulsive disorder (OCD) according to so-called DBS "Sweet Spot Mapping" in patients.Image modified from Fig.1ain the recent publication "Mapping dysfunctional circuits in the frontal cortex using deep brain stimulation"(Hollunder et al., 2024).
A.Ricci et al.