Parvalbumin-expressing basal forebrain neurons mediate learning from negative experience

The basal forebrain (BF) is crucial for cognitive functions, containing both cholinergic and GABAergic neurons. While cholinergic neurons have been extensively studied, the role of the larger GABAergic population, particularly parvalbumin-expressing (PV) neurons (BFPVNs), remains unclear. BFPVNs are known to regulate cortical gamma oscillations and promote wakefulness and arousal. However, their involvement in awake behaviors and associative learning is less understood. Evidence suggests a potential role: first, BFPVN degeneration is linked to cognitive decline in Alzheimer's disease and aging; second, non-selective BF lesions cause more significant learning deficits than selective cholinergic lesions; and third, non-cholinergic BF neurons respond to behaviorally salient stimuli. These observations motivate the investigation into the specific role of BFPVNs in associative learning, particularly focusing on their response to different reinforcement types (reward and punishment) and their downstream effects on brain areas critical for learning and memory.

Previous research has highlighted the importance of the basal forebrain in cognitive function, with a focus primarily on cholinergic neurons. However, studies showing that non-selective lesions of the basal forebrain result in greater attentional and learning deficits than selective lesions of cholinergic neurons suggest a crucial role for the larger GABAergic population. Further, work demonstrating that non-cholinergic basal forebrain neurons respond phasically to behaviorally salient stimuli regardless of valence, and that response strength correlates with decision-making speed, hints at a more nuanced role in cognitive processes beyond simple arousal. This study builds upon these previous findings to directly investigate the role of parvalbumin-expressing GABAergic neurons in associative learning, particularly focusing on their response to aversive stimuli.

The study used male PV-Cre mice (expressing Cre recombinase under the parvalbumin promoter) to selectively target BFPVNs in the horizontal limb of the diagonal band of Broca (HDB). Mice underwent a probabilistic Pavlovian conditioning task, associating auditory cues with water rewards (positive reinforcement) or air puffs (punishment). In vivo electrophysiological recordings using tetrodes combined with optogenetic tagging (ChR2) were used to identify and monitor the activity of individual BFPVNs in response to cues and outcomes. Bulk calcium measurements (fiber photometry with GCaMP6s) were also employed to monitor BFPVN activity both somatically and in projection areas. Optogenetic inhibition (ArchT) was utilized to causally assess the contribution of BFPVNs to learning during the punishment phase of the task. Anterograde and retrograde tracing techniques (using AAV viral vectors) were applied to map the long-range afferent and efferent connections of HDB BFPVNs. Furthermore, to evaluate the role of BFPVNs in mediating aversion, a conditioned place aversion test was also carried out. The study also explored responses to other aversive stimuli (foot shocks, fox odor) and compared BFPVN responses to those of somatostatin-expressing (SOM) neurons within the HDB. Finally, channelrhodopsin-assisted circuit mapping was used in acute brain slices to determine the downstream effects of BFPVN activity on target neurons in specific brain regions.

BFPVNs in the HDB exhibited distinct response profiles to different stimuli in the Pavlovian conditioning task. They showed rapid, phasic activation to punishment (air puffs), but smaller and delayed responses to rewards and reward-predicting cues. Optogenetic inhibition of BFPVNs specifically during punishment impaired the animals' ability to form cue-outcome associations, demonstrating a causal role in learning from negative reinforcement. Connectivity studies revealed that HDB BFPVNs receive strong inputs from aversion-related brain regions, including the lateral hypothalamus, septal complex, and median raphe nucleus. These neurons project to multiple limbic structures (medial septum, hippocampus, retrosplenial cortex), broadcasting information about aversive stimuli. Further analysis showed that the projections exhibited phasic punishment responses, suggesting a widespread broadcast function. Electrophysiological studies in acute brain slices revealed that BFPVN inputs onto target neurons showed varying synaptic plasticity characteristics dependent on the target region (short-term synaptic depression in hippocampus, facilitation in retrosplenial cortex). These findings confirm that BFPVNs play a significant role in associative learning, specifically processing and relaying information about negative reinforcement, and influencing downstream neuronal activity and information processing in learning and memory related brain structures.

The findings demonstrate that HDB BFPVNs are not simply involved in general arousal but play a specific role in processing and using aversive information for associative learning. The robust and homogeneous response to punishment, combined with the impaired learning observed during optogenetic inhibition, strongly supports this. The widespread projections to limbic areas suggest a broadcast mechanism for disseminating information about negative experiences. The varying synaptic plasticity observed in different target regions indicates a dynamic and context-dependent modulation of downstream circuits. This contrasts with the role of cholinergic BF neurons that have been shown to encode prediction errors related to both rewards and punishments. This suggests a functional division within the basal forebrain, with cholinergic neurons focused on reward prediction, and BFPVNs specializing in the processing and learning from negative reinforcement.

This study provides compelling evidence for a crucial role of parvalbumin-expressing GABAergic neurons in the basal forebrain in learning from negative experiences. Their phasic responses to aversive stimuli and causal involvement in associative learning highlight their significance beyond general arousal functions. Future research should focus on the precise mechanisms of BFPVN-mediated disinhibition in target areas and their interaction with other basal forebrain cell types. Investigating the potential for therapeutic interventions targeting BFPVNs in neurodegenerative diseases characterized by cognitive decline warrants further consideration.

The study primarily focused on male mice, limiting the generalizability to females. While multiple aversive stimuli were used, further investigation into the specific sensory pathways involved in BFPVN activation would be beneficial. The optogenetic inhibition approach, although effective, might have non-specific effects that warrant further examination. Finally, the sample sizes for some of the experiments could be considered relatively small and may influence the statistical power of some analyses.