Excitatory and inhibitory neurons in the dorsal periaqueductal gray encode decisions to assess and escape natural threats

1. Problem Statement

The dorsal periaqueductal gray (dPAG) is a critical brainstem structure for organizing defensive behaviors, specifically the transition from threat assessment (approach) to escape (flight). Previous electrophysiological studies identified two distinct neuronal populations in the dPAG:

• Assessment+ neurons: Increase firing during threat approach and cease firing at escape onset.
• Escape+ (Flight+) neurons: Are suppressed during approach and fire abruptly at escape onset.

However, the genetic identity and circuit architecture of these populations remained unclear. A prevailing hypothesis suggested that Assessment+ neurons were inhibitory (GABAergic) and Escape+ neurons were excitatory (glutamatergic), forming a "push-pull" mechanism. Conversely, an alternative model proposed that local feedback inhibition among excitatory neurons drives this switch. Furthermore, it was unknown whether the dPAG processes different types of threats (predator, social, prey) via distinct or overlapping neuronal ensembles.

2. Methodology

The authors employed a multimodal approach combining in vivo calcium imaging, optogenetics, and advanced behavioral analysis in mice.

• Animal Models: Transgenic mice expressing Cre-recombinase in specific cell types: Vglut2::Cre (glutamatergic/excitatory) and Vgat::Cre (GABAergic/inhibitory).
• Imaging: Chronic implantation of GRIN lenses and miniaturized microscopes (miniscopes) to record GCaMP6f fluorescence in the dPAG.
• Behavioral Paradigms:
  • Predator Threat: Exposure to a laboratory rat (separated by a barrier allowing sniffing but preventing attack).
  • Social Threat: Exposure to an aggressive conspecific (CD1 male mouse) following social defeat.
  • Prey Threat: Exposure to a live cockroach.
  • Protocol: Mice were exposed to all three threats in a single session to allow longitudinal comparison of neuronal responses.
• Optogenetics: In Vgat::Cre mice, ChR2 was expressed in GABAergic neurons. Blue light (465 nm) was delivered to the dPAG during predator exposure to test causal roles.
• Data Analysis:
  • Single-unit extraction: Identification of putative somata from miniscope videos.
  • Behavioral Annotation: Manual annotation and unsupervised machine learning (Keypoint-MoSeq) to classify behaviors (risk assessment, escape, rearing) into "syllables."
  • Statistical Modeling:
    • Linear Time Warping: To normalize trial durations for comparing neural dynamics across different behavioral epochs.
    • CEBRA (Contrastive Embedding for Behavioral Representation Analysis): A nonlinear dimensionality reduction technique used to decode behavioral states (approach vs. escape) from population activity.

3. Key Contributions

1. Genetic Identity of Functional Populations: The study definitively maps the functional "Assessment+" and "Escape+" phenotypes to both glutamatergic and GABAergic neurons, overturning the hypothesis that these roles are segregated by cell type.
2. Circuit Mechanism: The findings support a model where local feedback inhibition among excitatory neurons, rather than a simple excitatory/inhibitory segregation, drives the transition between assessment and escape.
3. Convergent Threat Processing: The study demonstrates that the dPAG utilizes largely overlapping neuronal ensembles to process diverse natural threats (predator, social, and prey), suggesting a generalized "trigger" mechanism for defensive behavior.
4. Causal Role of Inhibition: Optogenetic activation of GABAergic dPAG neurons was shown to suppress defensive behaviors and promote exploration, clarifying the inhibitory control of the escape circuit.

4. Results

A. Genetic Identity of Assessment+ and Escape+ Neurons

• Glutamatergic (Vglut2+) Neurons: Confirmed the presence of both Assessment+ (16–21% of population) and Escape+ (33–40%) neurons.
• GABAergic (Vgat+) Neurons: Unexpectedly, the study found that GABAergic neurons also contained distinct populations of Assessment+ (10–13%) and Escape+ (25–44%) neurons.
• Dynamics: While both cell types showed similar encoding profiles, GABAergic neurons exhibited slower temporal dynamics (ramping up/down more gradually) compared to the sharper transitions in glutamatergic neurons.

B. Causal Role of GABAergic Neurons

• Optogenetic Activation: Stimulating Vgat+ neurons in the dPAG did not induce flight (consistent with prior literature).
• Behavioral Modulation: However, stimulation significantly reduced risk assessment behaviors and escape velocity.
• Unsupervised Analysis: Keypoint-MoSeq revealed that Vgat+ stimulation increased time spent in rearing (exploratory behavior) and decreased time in risk assessment. This suggests GABAergic neurons act as a tonic brake on defensive circuits, and their activation suppresses the drive to flee.

C. Convergent Encoding of Threats

• Overlap: When mice were exposed to rats, aggressive conspecifics, and cockroaches, the same neuronal ensembles were recruited.
  • ~56% of responsive Vglut2+ neurons and ~46% of Vgat+ neurons responded to multiple threat types.
• Decoding: CEBRA models trained on calcium traces could accurately distinguish between "risk assessment" and "escape" states across all three threat types. Furthermore, models could distinguish the threat context (predator vs. social vs. prey) based on population activity, indicating that while the core circuit is shared, the specific activity patterns contain context-specific information.

5. Significance and Conclusion

This study fundamentally revises the understanding of dPAG circuitry:

• Paradox Resolved: The co-existence of Assessment+ and Escape+ neurons in both excitatory and inhibitory populations suggests a complex local microcircuit. The authors propose a model where GABAergic interneurons provide local feedback inhibition to distinct sets of projection neurons, facilitating the rapid switch from assessment to escape.
• Generalized Defense: The dPAG acts as a convergence point for diverse threats, utilizing a shared neural substrate to initiate stereotyped defensive behaviors, rather than maintaining parallel, threat-specific pathways.
• Therapeutic Implications: Understanding how excitatory and inhibitory balance controls the "decision" to flee could inform treatments for anxiety disorders and PTSD, where the transition from assessment to panic/flight is dysregulated.

The paper concludes that the dPAG functions as a tightly coordinated hub where local feedback inhibition shapes the decision-making process between assessing a threat and executing an escape, regardless of the threat's biological nature.