Unraveling Pallido-Retrorubral Circuits Linking the Basal Ganglia to Limbic Areas

The basal ganglia (BG) are a group of interconnected nuclei involved in learning, motivating, and performing motor and cognitive actions, including those intended to gain rewards or avoid injury. The BG integrate inputs related to the internal and external environment and through their outputs specify changes in behavior according to the present context and motivational state. How the precise organization of BG output channels enables action specification remains an intense field of study.

Multiple behavioral outputs are controlled by parallel BG loops that share a common circuit backbone but ultimately influence distinct thalamic and brainstem targets. In the classic BG circuit model, the striatum is the main integrator of cortical, thalamic, and dopaminergic inputs, the latter arising from the substantia nigra compacta (SNc; Fig. 1A). Striatal direct-pathway spiny projection neurons (dSPNs) inhibit BG output from the substantia nigra pars reticulata (SNr) and globus pallidus internal segment (GPi), while striatal indirect-pathway SPNs (iSPNs) inhibit the globus pallidus external segment (GPe), which in turn inhibits the subthalamic nucleus (STN) that ultimately excites the SNr and GPi. The classic model predicts that the direct pathway promotes actions via disinhibition of downstream brainstem and thalamic targets while the indirect pathways suppress actions via increased inhibition of these targets.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Summary of findings by Vounatsos and Gittis (2025) and comparison with established BG models. A, Classic indirect pathway model (key regions highlighted with black contour). B, Integrator–output model displaying GPe interconnection with forebrain areas and GPe output projections. Findings by Vounatsos and Gittis highlighted in orange. C, Detail of GPe-recipient RRF_GABA and RRF_DA projections (IPAC, interstitial nucleus of the anterior commissure; SNc, substantia nigra compacta; VTA, ventral tegmental area; PBN, parabrachial nucleus; PCRtA, parvicellular reticular nucleus, alpha part).

Adding complexity to this classic circuit scheme, additional intrinsic and extrinsic projections arise from the GPe. In addition to its role as an internal BG nucleus involved in the indirect pathway, the GPe is now recognized as both an integrator that receives cortical and subcortical inputs and an output structure (Courtney et al., 2023; Fig. 1B). The GPe has been linked to motor, cognitive, and limbic functions, only some of which have been attributed to specific GPe projections. GPe neurons are primarily GABAergic, and there are several subtypes identified based on their molecular and projection phenotypes. Distinct sets of parvalbumin-expressing GPe neurons target the SNr to promote locomotion and the parafascicular thalamus to modulate reversal learning (Lilascharoen et al., 2021). A GPe subpopulation expressing Npas1 directly targets the striatum and shows elevated activity during trained stop tasks; its activation is sufficient to interrupt locomotion (reviewed in Courtney et al., 2023). These studies demonstrate that the GPe can directly modulate downstream targets to promote certain behavioral outputs. In contrast, how GPe outputs modulate limbic functions such as avoidance behaviors (Isett et al., 2023) is not understood.

One way that GPe may influence limbic functions is through regulation of dopaminergic neurons. GPe is known to project to midbrain dopamine neurons in the VTA and SNc, two nuclei with well-known roles in motivated behaviors. Whether the GPe also projects to a third, less studied group of midbrain dopamine neurons located in the retrorubral field (RRF) has been unclear. The RRF is considered a hub for threat evaluation, reward processing, and motivation (Moaddab and McDannald, 2021), and it is interconnected with limbic structures including the amygdala. In addition to dopamine neurons (RRF_DA), the RRF contains GABAergic (RRF_GABA) neurons. Whether RRF_DA and RRF_GABA neurons have similar projection patterns has been unknown.

Because GPe projections to midbrain dopamine neurons might hold the key to understanding its roles in limbic functions, Vounatsos and Gittis (2025) recently asked whether GPe projects to the RRF. Tracing experiments in which GABAergic GPe axons and synaptic puncta were selectively labeled with a Cre-dependent reporter revealed robust labeling of GPe axons and puncta within the RRF. Analysis of synaptic densities showed a medial-lateral gradient, indicating that the GPe preferentially targets the lateral RRF.

Next, Vounatsos and Gittis (2025) asked which RRF neuronal populations receive GPe inputs. To address this, they expressed channelrhodopsin in the GPe, then photostimulated GPe axons in the RRF, while they made patch-clamp recordings from fluorescently labeled RRF_GABA or RRF_DA neurons. Synaptic responses to GPe axon activation in RRF_GABA and RRF_DA did not differ in conductance or short-term plasticity. These experiments show unambiguously that GPe targets both RRF_GABA and RRF_DA neurons.

To identify the targets of GPe-recipient RRF_GABA and GPe-recipient RRF_DA neurons, Vounatsos and Gittis (2025) used a viral vector capable of transsynaptic anterograde transport to express a Cre-dependent Flp in GPe neurons of Vgat-Cre or DAT-Cre mice. With this approach, Flp was expressed selectively in GPe-recipient GABAergic neurons (in Vgat-Cre mice) or GPe-recipient dopaminergic neurons (in DAT-Cre mice). A second virus was then injected into the RRF to express the fluorescent protein MCherry in a Flp-dependent manner. Thus, MCherry was expressed selectively in GPe-recipient RRF_GABA or GPe-recipient RRF_DA neurons. Subsequent axonal tracing revealed that GPe-recipient RRF_DA neurons project to the ventral, dorsal and tail of the striatum, as well as associated areas, including the interstitial nucleus of the posterior limb of the anterior commissure and the olfactory tubercle. In contrast, GPe-recipient RRF_GABA neurons had widespread projections to brain regions associated with limbic function, including several nuclei in the hypothalamus (dorsomedial nucleus, parafascicular nucleus, and submedial nucleus), parasubthalamic nucleus (pSTN), central amygdala, bed nucleus of the stria terminalis, parabrachial nucleus (PBN), and parvicellular reticular nucleus, alpha part. In addition, GPe-recipient RRF_GABA neurons projected to SNc and VTA. Thus, these experiments uncovered remarkably distinct axonal projection patterns of GPe-recipient RRF_DA and GPe-recipient RRF_GABA populations (Fig. 1C, summary).

The study by Vounatsos and Gittis (2025) expands our understanding of the organization of the BG by showing that the RRF is a target of the GPe. Specifically, the GPe monosynaptically targets both RRF_GABA and RRF_DA neurons, which in turn innervate distinct postsynaptic areas, many of which participate in limbic functions. Limbic subterritories within the striatum have previously been identified based on the distribution of inputs from limbic cortices. In turn, limbic-receiving striatal SPNs can influence SNc-mediated DA release directly, via monosynaptic projections, or indirectly, via disynaptic connections through the GPe (Xiao et al., 2020; Lazaridis et al., 2024). The findings by Vounatsos and Gittis (2025) add to this scheme by showing that that GPe-RRF subcircuits are also well positioned to mediate interactions between BG and limbic structures.

Although Vounatsos and Gittis (2025) did not investigate the functions of GPe-recipient RRF neurons, we may speculate about their function based on their projections patterns. The projection pattern of GPe-recipient RRF_DA neurons partly overlaps with that of mesolimbic and nigrostriatal pathways innervating the ventral and dorsal striatum. Striatal RRF_DA projections have a ventrolateral bias, suggesting a small contribution of RRF_DA projections to reward processing by the nucleus accumbens. RRF_DA projections to the tail of the striatum suggest a role in threat avoidance behaviors (Green et al., 2024). RRF_GABA projections may influence the BG broadly by targeting the VTA and SNc. RRF_GABA neurons also project to hypothalamic, thalamic, and amygdala targets, suggesting roles in fear memory, nociceptive modulation, hyperalgesia, and stress. The largest portion of RRF_GABA axons were found in the PBN and a fraction in the PCRtA, two brainstem nuclei known to participate in sensory processing relevant for modulating pain and autonomic functions (Palmiter, 2018). In addition, RRF_GABA projections to PBN and those ascending to the pSTN could play a role in appetite regulation (Palmiter, 2018; Kim et al., 2022).

Previous studies have found that RRF responses to threat or safety stimuli are heterogeneous across neurons (Moaddab and McDannald, 2021). Future work may use the viral strategies adopted by Vounatsos and Gittis (2025) to test the function of distinct RRF subpopulations. For example, it should be possible to virally express calcium indicators in RRF populations defined by their neurotransmitter phenotype, inputs (e.g., GPe), and outputs (e.g., tail of the striatum). In vivo imaging experiments may then show how processing of reward, safety, and threat stimuli is reflected by the calcium dynamics of specific RRF subpopulations. Vounatsos and Gittis (2025) show that the GPe links basal ganglia circuits to limbic structures via RRF projections. Future investigations may use retrograde tracing or monosynaptic rabies tracing from RRF_GABA and RRF_DA neurons to clarify which among GPe neuronal subtypes target the RRF, revealing if RRF-projecting GPe neurons belong to previously identified subclasses or perhaps form a new subclass of GPe neurons. In addition, understanding the upstream inputs to RRF-projecting GPe neurons will be important to further clarify their position within the broader BG network.

In conclusion, the study by Vounatsos and Gittis (2025) shows that GPe is more than a simple node in the classic indirect pathway. It has specific output projections to RRF. Through GPe-RRF pathways, the GPe is well-positioned to control downstream nuclei important for the control of bodily functions, motivation, and threat assessment, processes compromised in neuropsychiatric conditions affecting BG function.