Prefrontal correlates of fear generalization during endocannabinoid depletion

Abstract

Maladaptive fear generalization is one of the hallmarks of trauma-related disorders. The endocannabinoid 2-arachidonoylglycerol (2-AG) is crucial for modulating anxiety, fear, and stress adaptation, but its role in balancing fear discrimination versus generalization is not known. To address this, we used a combination of plasma endocannabinoid measurement and neuroimaging in a childhood maltreatment-exposed and -nonexposed mixed population, combined with human and rodent fear-conditioning models. Here we show that 2-AG levels were inversely associated with fear generalization at the behavioral level in both mice and humans. In mice, 2-AG depletion increased the proportion of neurons that respond to, and the similarity of neuronal representations for, both threat-predictive and neutral stimuli within prelimbic prefrontal cortex neuronal ensembles. In humans, increased dorsolateral prefrontal cortical-amygdala resting-state connectivity was inversely correlated with fear generalization. These data provide convergent cross-species evidence that 2-AG is a key regulator of fear generalization and further support the notion that 2-AG deficiency could represent a trauma-related disorder-susceptibility endophenotype.

Keywords: Behavior; Clinical Research; Neuroimaging; Neuroscience; Psychiatric diseases.

Figure 1. Decreased levels of 2-AG are associated with increased generalization to neutral stimuli and self-reported emotional regulation.

(A) Experiment schematic. (B) Generalization between CS⁺ and CS^– and (C) difficulties in emotional regulation as a function of baseline peripheral 2-AG levels. Scatter plots: yellow, participants with CM and SUD history; red, participants with CM history; green, participants with SUD history; and gray, participants with no CM or SUD history. (D) Experiment schematic depicting stimuli for differential fear-conditioning and recall test sessions. (E) Percentage of time freezing in response to CS⁺ and CS^–. Arrow, i.p. injection of vehicle or DO34 before recall test. (F) Generalization index for CS^– t₁ and t₂ for vehicle and DO34 groups. (G) Experiment schematic depicting stimuli for partial differential fear-conditioning and recall test sessions. (H) Percentage of time freezing to CS⁺ and CS^–. (I) Generalization index for CS^– t₁ and t₂ for vehicle and DO34 groups. (J) Experiment schematic depicting stimuli for classical fear-conditioning and recall test sessions. (K) Percentage of time freezing to CS⁺ and NT. (L) Generalization index for NT t₁ and t₂ for vehicle and DO34 groups. (M) Experiment schematic depicting stimuli for contextual fear-conditioning and recall test sessions. (N) Percentage of time freezing to conditioning context (CtxA) and novel context (NCtx). (O) Generalization index for NCtx relative to freezing to CtxA on conditioning day 2 (CtxC). (P) Experiment schematic depicting stimuli for classical fear-conditioning and recall test sessions. (Q) Sagittal depiction (left) and coronal section (right) showing optic fiber and GRABeCB2.0 expression in PL. The dashed lines correspond to the subdivisions of the medial prefrontal cortex. (R) Average NT response for trials with freezing to tones above 75% (high generalization) and below 25% (low generalization). (S) Peak population NT response for high- and low-generalization trials. Repeated-measures or 2-way ANOVA followed by Šídák’s multiple-comparison test or Student’s t test where appropriate. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 2. Neuronal tone responses in response to NTs and CS⁺.

(A) Schematic of experimental timeline and setup. (B) Sagittal depiction of GRIN lens and microscope implanted above the region where GCaMP7f was expressed in the PL. (C) Left: Example coronal section of the lens track above GCaMP expression in PL. Right: Maximum projection image as seen through the miniaturized microscope, with identified neurons extracted using the CNMFe algorithm surrounded by white lines. (D) Left: Average PL neuronal response to NT for all recorded neurons. Middle: NT response for neurons in PL that exceeded +3 Z-scores during tone presentation. Right: CS⁺ response for neurons in PL below –3 Z-scores during tone presentation. (E) Left: Average PL neuronal response to conditioned tone (CS⁺) for all recorded neurons. Middle: CS⁺ response for neurons in PL that exceed +3 Z-scores during tone presentation. Right. CS⁺ response for neurons in PL below –3 Z-scores during tone presentation. (F) Top: Pie charts for vehicle- and (G) DO34-exposed mice showing proportion of neurons that significantly responded to tone presentations (±3 Z-scores). Bottom: Proportion of neurons that increased activity to NT [(+) NT] or CS⁺ [(+) CS⁺], decreased activity to NT [(–) NT] or CS⁺ [(–) CS⁺], or changed activity to both NT and CS⁺ (dual). (H) Histogram of neuronal stimulus preference strength for NT versus CS⁺. Hierarchical tree clustering of neuronal activity to NT and CS⁺ from vehicle- (I) and DO34-exposed (J) mice. Data are shown as mean ± SEM. Repeated-measures ANOVA followed by Šídák’s multiple-comparison test, Fisher’s exact test, and Kolmogorov-Smirnov test where appropriate. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 802 and 885 neurons from n = 7 and 8 mice treated with vehicle and DO34, respectively.

Figure 3. Reduced 2-AG is associated with increasing similarity of population metrics.

(A) Trajectory of neuronal population activity in 3D PC space during presentation of the CS⁺ (open dots) and NT (filled dots) for both DO34- (red) and vehicle-treated (blue) mice. (B) Fifteen PCs explain 91.5% of the variance in the neuronal data set. (C) Mahalanobis distance between NT and CS⁺ trajectories in PC space for vehicle- and DO34-exposed mice. (D) Instantaneous Euclidean distance between NT and CS⁺ trajectories across tone duration. Student’s t test. ****P < 0.0001.

Figure 4. dlPFC-amygdala resting-state connectivity is inversely correlated to fear generalization.

(A) Depiction of increased resting state between the right amygdala and dlPFC. (B) Generalization between CS^– and CS⁺ as a function of β coefficient for dlPFC. (C) Baseline 2-AG levels as a function of β coefficient for dlPFC.