The appropriate selection of behavioral responses in dangerous situations is vital for survival and is influenced by the perceived proximity and intensity of threats. Various studies have investigated the biological underpinnings surrounding the four most common responses to a perceived threat: fight, flight, freeze, and fawn (Zingela et al. 2022). The mechanisms underlying shifts between different response states, however, remain incompletely understood. The eCB has emerged as a key modulator of threat appraisal and defensive behavior selection (Maldonado et al. 2020). The present study extends these findings by delineating the role of the eCB system in shaping passive-to-active behavioral transitions in response to both learned and innate aversive stimuli.

To discern the role of 2-AG in regulating passive (freezing) and active (jumping, fleeing) defensive behaviors, we first employed the SCS paradigm. On day 1 of conditioning, pharmacological reduction of 2-AG via DO34 treatment led to an increased number of jumps during the WN component compared to vehicle-treated mice. Interestingly, on Day 2 we observed that DO34-treatment led to heightened passive responses to the pure tone and increased active responses to white noise. It is important to contextualize these findings within the overall freezing patterns across conditioning. On Day 1, average pre-SCS freezing values were elevated (~ 40%) for both treatment groups and freezing during the pure tone and white noise periods did not significantly exceed this baseline. However, freezing levels during trial 1 were comparable to those on Day 0 (Supplemental Fig. 2), suggesting that initial elevated freezing reflects a generalized fear response to the first tone-shock pairing rather than a drug-induced locomotor impairment. This interpretation is further supported by the similar pre-SCS freezing levels between DO34 and vehicle groups, and prior evidence that DO34 does not impair general locomotion (Ramos-Medina et al. 2024). While cue-specific freezing was limited on Day 1, clear cue discrimination emerged by Day 2 and during extinction, indicating intact associative learning. These findings suggest that 2-AG depletion may facilitate fear memory consolidation, thereby amplifying defensive responses during subsequent SCS exposure.

Extinction of conditioned fear responses is also essential for adaptive behavior, enabling reductions in defensive responses when threat is no longer present. Impaired extinction is a hallmark of stress-related disorders, including PTSD (Norrholm et al. 2011). Prior work has demonstrated that diminished eCB signaling can impair fear extinction (Marsicano et al. 2002; Hill et al. 2018; Cavener et al. 2018; Ramos-Medina et al. 2024), whereas enhancement of eCB signaling is linked to facilitation of fear extinction (Chhatwal et al. 2005, 2009; Bitencourt et al. 2008; Gunduz-Cinar et al. 2013, 2023). Interestingly, we found that during early extinction (trials 1–4), previously DO34-treated mice showed significantly attenuated active fear responses (Fig. 3G, H). However, their calculated escape scores were higher than controls, which appears to be a limitation of the metric itself: if a mouse exhibited complete freezing during the pre-SCS period, the resulting score was artificially inflated due to a near-zero denominator (Supplemental Fig. 6). These results suggest that 2-AG may promote fear extinction by preferentially dampening active defensive responses. The neural underpinnings of these effects may involve circuit-specific modulation, as two different cell types in the central amygdala – somatostatin and corticotropin-releasing factor neurons– initiate passive freezing behavior and mediate conditioned flight responses, respectively (Fadok et al. 2017). Thus, we conclude that 2-AG depletion during conditioning promotes the subsequent extinction of active defensive responses to SCS presentation, however the effects of 2-AG depletion during extinction training itself were not evaluated.

In contrast, global enhancement of 2-AG levels via MAGL inhibition (JZL-184) did not have significant effects on fear acquisition or extinction. These findings partially align with a previous work from our lab in which JZL-184 pre-treatment had no impact on cued-conditioning fear learning (Hartley et al. 2016). However, while the former study saw impairment in fear extinction, we did not replicate that effect with our SCS protocol. This discrepancy may stem from the timings of the JZL-184 administration. In the present study, our findings inform us of the effects of varying eCB tone during fear acquisition on extinction learning, as drug treatments only occurred on conditioning days. Meanwhile, our prior work examined the effects of enhanced 2-AG levels during extinction learning and, thus, drug treatments were delivered one hour prior to extinction sessions (Hartley et al. 2016).

Similarly, augmenting AEA levels via FAAH inhibition (PF-3845) did not alter fear learning or extinction. One possible explanation is that naturally occurring AEA signaling may be maximally engaged during conditioning, and further elevations in AEA content cannot affect defensive response generation. This is supported by studies that show increased AEA levels in the medial prefrontal cortex, amygdala, hippocampus, and periaqueductal gray following footshock stress (Hohmann et al. 2005; Morena et al. 2014). However, some studies have demonstrated contrasting results with reductions in AEA content globally (Bluett et al. 2014) and in the amygdala (Vecchiarelli et al. 2022) after footshock stress. Thus, these findings highlight the complexity of AEA signaling dynamics in response to stress, suggesting that its effects on defensive behavior may be region-specific, temporally regulated, and highly sensitive to experimental context.

Taken together, these data indicate that 2-AG deficiency is associated with progressively enhanced magnitude of passive and active responses to SCS across conditioning days, thereby amplifying both defensive response strategies to SCS presentation. However, analysis of escape scores suggests DO34 treatment may bias behavior toward active coping as threat imminence increases, driving a transition from passive to active responses. These data point to an important role for endogenous 2-AG in constraining both active and passive defensive responses during escalating threat imminence.

Adaptive responses to predator-like threats generally fall into two primary categories: (1) avoiding detection through behaviors like freezing and (2) avoiding capture through actions such as fleeing or fighting (Yilmaz and Meister 2013; Shang et al. 2018; Salay et al. 2018). In this study, we used a looming shadow paradigm to elicit these behaviors and examine how eCBs regulate the repertoire and shift of behaviors to a perceived imminent threat. Consistent with previous reports, we observed that the predominant defense responses to looming shadow stimuli were flight/escape behaviors, including darting and transitions from freezing to darting (Fig. 4D) (Yilmaz and Meister 2013; Daviu et al. 2020b). Our results showed that pharmacological inhibition of 2-AG synthesis led to animals spending significantly more time in the tent, with no changes in latency to respond to the shadows. Additionally, DO34-treated mice timed-out significantly more than vehicle-treated mice, suggesting exaggerated passive defensive responses at the expense of exploratory behavior during the inter-trial interval when no threat was present. While we did not analyze behavior as a function of time across the looming stimulus itself, doing so in future work could reveal whether 2-AG dynamically modulates the shift from passive to active responses as the perceived threat escalates. This would offer further insight into how 2-AG regulates adaptive defense strategies in a threat-imminence-dependent manner. Taken together, endogenous 2-AG may be important for terminating passive defensive responses in the absence of threat and promoting optimal exploratory behavior in the absence of threat.

Importantly, these effects appear to be specific to reductions in 2-AG, as enhancing 2-AG levels via the MAGL inhibitor JZL-184 had no significant effect on shadow-evoked behaviors, latency to respond, or time spent in the tent. In contrast, augmenting AEA levels by blocking its degradation promoted more passive behavioral responses, although latency and time in tent remained unaffected. Our data suggest that 2-AG signaling may be necessary for adaptive behavioral transitions in the absence of threat, while AEA may serve to modulate coping style toward more passive responding without altering threat detection or appraisal. These results align with previous work in which mice showed reduced flight behaviors to an approaching robo-beetle after enhancement of AEA levels (Heinz et al. 2017). Similarly, we observed fewer darts and decreased reactivity to shadow stimuli, indicative of a shift toward more passive coping strategies. However, while they also reported that enhanced 2-AG signaling increased flight responses (Heinz et al. 2017), which we did not observe. Notably, their study lacked a designated ‘safe zone’ for the mice to retreat to, which may have influenced the expression of escape responses. This difference in task design may explain the divergent behavioral outcomes and suggests that environmental context can shape how 2-AG modulates threat responses.

One alternative interpretation is that eCBs induce alterations in visual perception that affect defense responses. While it is well-established that eCBs are known to play important roles in visual system development (Bouchard et al. 2016), less is known about the acute effects of 2-AG depletion on visual processing. However, a recent study examined how different cannabinoids affect mouse visual acuity. They report that pharmacological augmentation and attenuation of 2-AG decreased and increased visual acuity, respectively, while augmenting AEA did not affect acuity (Cécyre et al. 2020). This aligns with our finding that DO34 does not impair threat detection, as treated mice displayed normal latencies to respond to the looming stimulus, suggesting intact visual processing. Future studies directly assessing visual function during acute 2-AG depletion will be necessary to fully rule out this possibility.

Beyond behavioral outcomes, future studies should also dissect the neural circuitry by which 2-AG signaling shapes defensive state transitions. Selective targeting of eCB machinery using promising tools such as the DAGLα^f/f (Winters et al. 2021) or CB1^f/f (Marcus et al. 2020; Kondev et al. 2023b) mouse lines may provide insight into how local eCB tone modulates defensive behavior. Notably, recent work has demonstrated that the visual cortex can instruct the suppression of innate defensive responses through a top-down pathway to the ventrolateral geniculate nucleus (vLGN), with this learning-induced plasticity requiring eCB-mediated long-term suppression of inhibitory synapses onto vLGN neurons (Mederos et al. 2025). Such findings reinforce the idea that eCB signaling is critical for adaptive behavioral plasticity in response to complex and evolving threat environments.

In summary, our study demonstrates that 2-AG signaling plays a critical role in regulating behavioral state transitions across both learned and innate fear responses. During learned fear paradigms, such as the SCS, endogenous 2-AG appears to constrain the magnitude of both passive and active defense strategies, potentially promoting adaptive responses to escalating threat imminence. Similarly, in innate fear paradigms like the looming shadow test, endogenous 2-AG appears to promote exploratory behavior between threat presentations and thus may serve an adaptive role to maximize exploratory/foraging in the absence of threat. These findings highlight the context-dependent role of the endocannabinoid system in facilitating appropriate behavioral selection during perceived threats. By revealing how 2-AG shapes the repertoire and transition of defensive states, this study underscores the broader impact of the eCB system on adaptive responses to fear and stress. Elucidating these mechanisms may ultimately inform novel therapeutic strategies for fear dysregulation in stress-related psychiatric disorders.