Introduction

Exposure to severe stress can lead to the development of psychiatric conditions such as post-traumatic stress disorder (PTSD). While PTSD encompasses a range of symptoms, alterations in fear learning are widely regarded as a core feature underlying the disorder (Lissek and Meurs, 2015). Accordingly, Pavlovian fear conditioning, a form of associative learning, has been the dominant framework for studying how fear is triggered by cues or contexts associated with a prior stressful experience. However, stress also engages a second mechanism often overlooked in Pavlovian procedures: fear sensitization, a nonassociative form of learning. Fear sensitization is characterized by a widespread potentiation of fear and defensive responses that extends beyond the initial stressful experience. Therefore, after stress exposure, innocuous stimuli may elicit fear even in contexts quite different from the original stressor. Nonassociative learning processes, such as sensitization, are highly conserved across species, from organisms with primitive nervous systems to humans (Harris, 1943). Sensitization is also likely to underlie a subset of PTSD symptoms (Stam, 2007b; Charney et al., 1993; McFarlane, 2010). Core symptoms—including heightened arousal and reactivity, hypervigilance, and exaggerated startle reactions—closely resemble fear sensitization. These symptoms can manifest independently of cues or contexts related to the traumatic event and may therefore be resistant to conventional therapeutics, such as exposure therapy, which primarily target associative fear processes. Despite its clinical relevance, the neural basis of fear sensitization in the mammalian nervous system remain largely unknown and underexplored.

Most research on fear sensitization has focused on a phenomenon called stress-enhanced fear learning (SEFL). Developed by Fanselow and colleagues (e.g., Rau et al., 2005), this model demonstrates that a prior history of inescapable footshock stress potentiates subsequent fear conditioning to a novel context. This phenomenon is thought to reflect stress enhancement of fear acquisition rather than fear expression. For instance, Rau et al., 2005 showed that footshock stress failed to enhance the expression of conditioned fear that was acquired before and tested after exposure to footshock stress. Brain regions implicated in SEFL include the amygdala (Perusini et al., 2016; Pennington et al., 2024), hippocampus (Hersman et al., 2019), and prefrontal cortex (Pennington et al., 2017), with the basolateral amygdala (BLA) playing a critical role. For instance, disrupting neuronal activity or protein synthesis in the BLA blocks stress-induced enhancement of fear learning (Perusini et al., 2016; Pennington et al., 2024).

In addition to enhancing associative fear learning, stress can also sensitize unlearned or unconditioned fear responses, a phenomenon we term stress-enhanced fear responding (SEFR). For example, prior footshock stress potentiates freezing evoked by a novel, innocuous tone stimulus (Hassien et al., 2020; Kamprath and Wotjak, 2004; Siegmund and Wotjak, 2007a, b). This stressinduced sensitization of unlearned fear persists even after animals undergo fear extinction in the stress context, indicating a nonassociative process that is independent of the fear memory of the initial stressful event (Hassien et al., 2020; Rau et al., 2005; Siegmund and Wotjak, 2007a). Clinically, SEFR may parallel features of exaggerated threat sensitivity, which is strongly associated with PTSD (Stam, 2007a). However, the neural mechanisms of SEFR are poorly understood, and it is unclear whether SEFL and SEFR reflect distinct processes or share overlapping neural mechanisms.

Here, we investigated the neural basis of SEFR and asked whether these mechanisms also mediate SEFL. Using c-Fos mapping and fiber photometry, we identified the posterior paraventricular thalamus (pPVT) as a region that is hyperactive in stressed animals during SEFR expression. Using chemogenetic manipulations, we show that pPVT hyperactivity during and after stress is essential for SEFR but not SEFL. We conclude that sensitization of fear learning and sensitization of unlearned fear are dissociable at the behavioral and neural levels. Additionally, we identify pPVT hyperactivity as a mechanism for stress-induced sensitization of unlearned fear and as a potential contributor to arousal and reactivity symptoms in trauma- and stressor-related disorders.

Results

Exposure to footshock stress induces stress-enhanced fear responding (SEFR) and stress-enhanced fear learning (SEFL)

Our laboratory previously established a mouse behavioral protocol for within-subjects assessment of SEFR and SEFL (Fig. 1A) (Hassien et al., 2020), based on the work of Rau et al., 2005 and Siegmund and Wotjak, 2007a. Mice were assigned to either a Stress group, which received a single session of four 1 mA footshocks, or a NoStress group, which underwent context exposure for an equivalent duration without footshock (Fig. 1B). Subsequently, all mice were subjected to a series of behavioral tests designed to evaluate associative fear memory of the initial stress context, SEFR, and SEFL. The tests were conducted in order of increasing severity to minimize hysteresis.

Figure 1.

To test associative fear memory, mice were re-exposed to the chamber in which the footshock stress occurred. Mice in the Stress group exhibited significantly elevated freezing levels compared to those in the NoStress group (𝑡₂₁ = 5.078, 𝑝 < .0001), indicating robust associative fear memory of the stress context (Fig. 1D). To assess sensitization of unlearned fear, we employed a behavioral assay referred to as the SEFR tone test. Mice were placed in a novel chamber and exposed to three 30 sec auditory tones following a 3 min baseline period (Fig. 1C). Unlike Pavlovian cued fear conditioning, tone-elicited fear in this task reflects an "unlearned" or "unconditioned" fear response, as mice had no prior experience with the tone or the context. During the pre-tone baseline period, we observed low levels of freezing in all groups. However, during the tone and inter-tone periods ("tone" in figures), the Stress group exhibited significantly higher freezing relative to the NoStress group (𝑝 < .01). A two-way repeated measures ANOVA revealed a significant Group x Tone Period interaction (𝐹_1,21 = 9.870, 𝑝 = .0049). Finally, we assessed sensitization of fear learning using the two-day SEFL procedure. On day 1 (SEFL conditioning), mice were placed in another novel context and received a single 0.75 mA footshock. All groups displayed low levels of freezing before shock delivery (Fig. 1E). On day 2 (SEFL test), mice were returned to the same context of the previous day for assessment of context-elicited fear. The Stress group exhibited significantly greater freezing than the NoStress group (𝑡₁₉ = 5.050, 𝑝 < .0001), indicating enhanced fear learning (Fig. 1F).

It is important to highlight that mice displayed little to no freezing prior to the onset of the tone in the SEFR tone test (Fig. 1C "Pre-Tone") or the onset of the footshock in the SEFL conditioning session (Fig. 1E "Pre-Shock"). These observations demonstrate that mice distinguished between contexts and that subsequent freezing is these tests is not likely attributed to associative fear processes, such as generalization of fear of the stress context. Collectively, these results demonstrate that footshock stress induced both associative fear of the stress context and fear sensitization in the forms of SEFR and SEFL. We propose that SEFR and SEFL are distinct types of fear sensitization: the sensitization of unlearned fear responses (SEFR), and the sensitization of fear learning (SEFL).

Associative fear, SEFR, and SEFL are dissociable at the behavioral level

To determine whether associative fear, SEFR and SEFL are dissociable at the behavioral level, we examined how the three assays respond to variations in footshock stress intensity and the extent to which freezing responses in the three assays intercorrelate. Mice were assigned to one of four treatment groups corresponding to the amplitude of footshock delivered during the stress session: NoStress, 0.25mA-Stress, 0.5mA-Stress, and 1mA-Stress. During the footshock stress session (Fig. 2A), we detected a significant effect of Group (𝐹_3,23 = 13.04, 𝑝 < .0001) and Group x Shock Pe-riod interaction (𝐹_12,92 = 6.384, 𝑝 < .0001). Mice exposed to footshock showed increased levels of freezing relative to the NoStress group. In the SEFR tone test (Fig. 2B), a RM-AVOVA detected a significant Group x Tone Period interaction (𝐹_3,23 = 3.607, 𝑝 = .0286). Pre-tone levels of freezing were low and similar between groups. During the tone period, freezing levels in the 0.5mA-Stress and 1mA-Stress groups were significantly higher than the NoStress group (all 𝑝^′𝑠 < .05). Moreover, no significant differences in freezing were observed between the 0.25mA-Stress and NoStress groups.

Figure 2.

In the stress context test (Fig. 2C), results from a one-way ANOVA revealed a significant effect of Group on freezing levels (𝐹_3,23 = 30.46, 𝑝 < .0001). Specifically, all groups that received footshock during the stress session, regardless of shock amplitude, exhibited significantly higher levels of freezing relative to the NoStress group (all 𝑝^′𝑠 < .05). Lastly, in the SEFL procedure, mice were given a 1-shock conditioning session (Fig. 2D), in which significant effects were observed for Group (𝐹_3,23 = 3.413, 𝑝 = .0344) and Group x Shock Period interaction (𝐹_3,23 = 3.439, 𝑝 = .0336). The following day, in the SEFL test (Fig. 2E), we observed a significant main effect of Group (𝐹_3,23 = 3.391, 𝑝 = .0351). Post hoc comparisons indicated that both the 0.5mA-Stress and 1mA-Stress groups exhibited significantly higher levels of freezing compared to the NoStress group (all 𝑝^′𝑠 < .05). The pattern of results indicates that associative contextual fear had a lower threshold for induction than either form of sensitization. Lastly, we examined whether freezing behavior in stressed mice correlated across the different behavioral assays (Fig. 2F). No significant correlations were observed between associative fear of the stress context, SEFR, and SEFL, suggesting that these fear responses reflect dissociable outcomes of footshock stress.

Stressed mice exhibit potentiated c-Fos expression in the paraventricular thalamus during SEFR

To identify brain regions that might mediate sensitization of unlearned fear, we examined the effects of prior footshock stress on c-Fos expression evoked during the SEFR tone test. On the first day of testing, mice underwent footshock stress (Stress groups) or an equal duration of context exposure (NoStress groups). Following 24 hrs, mice were placed in a novel context and underwent the SEFR tone test (ToneTest groups) or were left undisturbed in their homecage (HomeCage groups). After 90 min, brain tissue was collected and processed for immunohistochemistry against the c-Fos protein. This experimental design established the following groups: NoStress-HomeCage, Stress-HomeCage, NoStress-ToneTest, Stress-ToneTest.

Next, we quantified c-Fos expression across several limbic brain structures regions previously implicated in the regulation of fear and other defensive behaviors (Fig. 3A). A two-way ANOVA revealed a significant Group x Region interaction (𝐹_15,173 = 5.036, 𝑝 < .0001), indicating that stress effects on neural activity varied across brain areas. Post hoc comparisons showed that relative to the NoStress-ToneTest group, the Stress-ToneTest group exhibited significantly elevated levels of c-Fos expression in the paraventricular thalamus (𝑝 < .0001). The c-fos levels in these groups did not differ in the basolateral amygdala (𝑝 = .8889), central amygdala (𝑝 = .9993), lateral parabrachial nucleus (𝑝 = .9841), dorsolateral periaqueductal gray (𝑝 = .3871), or lateral periaqueductal gray (𝑝 = .3361). c-fos levels in the NoStress-HomeCage and Stress-HomeCage groups did not differ significantly in any region (𝑝 > .05), indicating that stress did not alter baseline c-Fos expression.

Figure 3.

These results demonstrate that footshock stress potentiates PVT activity in response to the SEFR tone test. Based on prior literature indicating functional distinctions between the anterior (aPVT) and posterior (pPVT) subdivisions of the PVT, we conducted subregional analyses and found that c-Fos expression in both the aPVT and pPVT was significantly elevated in the Stress-ToneTest group relative to the NoStress-ToneTest group. Lastly, we examined the relationship between c-Fos expression and freezing behavior during the SEFR tone test (Fig. 3C). c-Fos expression in the PVT was significantly correlated with freezing to the tone (𝑟 = .67, 𝑝 = .003), whereas c-Fos expression in other regions did not correlate significantly. These findings highlight PVT activity as a potential regulator of sensitized unlearned fear responses.

Stressed mice display elevated pPVT neural activity to a tone stimulus, but not to footshock

Our c-Fos analysis suggests that stressed mice display neuronal hyperactivity in the pPVT during the SEFR tone test. Based on prior evidence that the pPVT is a stress-responsive region and can modulate fear and other defensive behaviors (Kirouac, 2015), we sought to further investigate its putative role in fear sensitization. We expressed the genetically encoded calcium indicator GCaMP8s in the pPVT and implanted an optical fiber above the region (Fig. 4A.) After the mice recovered, they were subjected to either footshock stress (Stress group) or context exposure with no footshocks (NoS-tress group). We then monitored bulk Ca²+ signals using fiber photometry during the SEFR tone test and SEFL conditioning session. The aim of this experiment was to confirm whether stress-induced alterations in pPVT activity are present during exposure to the novel tone stimulus used to evoke SEFR and/or the single mild footshock used to elicit SEFL.

Figure 4.

In the SEFR tone test, all mice exhibited Ca²+ signals in pPVT neurons that peaked shortly after tone onset and gradually declined over the 30-s stimulus (Fig 4B, left panel). Notably, the Stress group displayed higher levels of pPVT activity during the tone compared to the NoStress group (Fig. 4B, right panel; 𝑡₁₄ = 2.358, 𝑝 = .0335). We next examined pPVT activity responses to the footshock used in the SEFL conditioning session. Here, the presentation of the footshock evoked a rapid increase of Ca²+ signals that were comparable between the NoStress and Stress groups (Fig. 4C, left panel). AUC analysis during the 2-s presentation of the footshock revealed no significant group differences (Fig. 4C, right; (𝑡₉ = .00014, 𝑝 = .9989). In summary, stress exposure enhanced pPVT Ca²+ responses in the SEFR tone test but did not alter pPVT responses during fear conditioning in a novel context. These results suggest that pPVT hyperactivity in stressed mice is a distinct feature of sensitization of unlearned fear, but not sensitization of fear learning.

Inhibition of pPVT during footshock stress disrupts the induction of SEFR but not SEFL

We next tested the hypothesis that pPVT neuronal activity during footshock stress is necessary for the induction of fear sensitization. An adeno-associated virus (AAV) encoding the inhibitory hM4Di receptor (AAV8-hSyn-hM4Di-mCherry) or the fluorescent mCherry reporter (AAV8-hSyn-mCherry) was infused in the pPVT of mice (Fig. 5A). Following recovery from surgery, we tested whether CNO-mediated inhibition of pPVT neurons during footshock stress would prevent subsequent stress-induced alterations in fear and anxiety-like behavior. All mice underwent the same behavioral testing procedure depicted in Fig. 5B.

Figure 5.

CNO was administered 30 min prior to footshock stress (Fig. 5C). During this session, freezing levels increased progressively as a function of shock number, but no significant differences between the groups were detected (𝐹_1,27 = .01788, 𝑝 = .895). Subsequent behavioral assays were conducted without CNO administration. In the SEFR tone test (Fig. 5D), a RM-ANOVA revealed a significant Group x Tone Period interaction (𝐹_1,27 = 6.097, 𝑝 = .0202). Freezing behavior was notably lower in the hM4Di group compared to the mCherry group. Post hoc tests confirmed that these differences in freezing were restricted to the tone period (𝑝 = .0013) and not the pre-tone baseline period (𝑝 = .9759).

Next, we conducted subsequent behavioral assays in the same cohort of mice to further investigate the role of pPVT in regulating additional stress-altered fear and anxiety-like behaviors. Our previous report showed that total distance traveled in the open field is reliably suppressed after footshock stress (Hassien et al., 2020). In the open field (Fig. 5E), the distance traveled was not significantly different between groups (𝑡(25) = .9920, 𝑝 = .3307). Time and distance in the center arena also did not differ between groups (data not shown). In the test for associative fear of the stress context (Fig. 5F), we also observed no significant group differences in freezing behavior (𝑡(27) = .1735, 𝑝 = .8636). Lastly, in the SEFL test (Fig. 5G), expression of contextual fear did differ between the mCherry and hM4Di groups (𝑡(24) = .1506, 𝑝 = .8815). Collectively, the results from this experiment demonstrate that inhibiting the pPVT during footshock stress attenuated SEFR but did not affect conditioning of the stress context, open field exploration, or SEFL.

Inhibition of pPVT attenuates the expression of SEFR but not SEFL

The findings thus far suggest that pPVT activity is necessary for the induction of SEFR but not SEFL. Next, we asked whether pPVT activity is required for the expression of either of these forms of sensitization. As in the previous experiment, we used an AAV approach to express the inhibitory hM4Di receptor or the control mCherry fluorescent reporter in pPVT neurons (Fig. 6A). After recovery from surgery, all groups were exposed to footshock stress without CNO treatment. Subsequently, mice underwent behavioral testing under CNO treatment (CNO administered 30 min prior to each test) to assess the impact of pPVT inhibition on fear and anxiety-like behavior (Fig. 6B).

Figure 6.

During the footshock stress session conducted in the absence of CNO (Fig. 6C), mice displayed progressively increasing levels of freezing in response to the footshocks. No significant differences were observed between groups that received the hM4Di or mCherry viral infusions (𝐹_1,28 = .8117, 𝑝 = .3753). 24 hours later, mice underwent the SEFR tone test with CNO on board (Fig. 6D). A RM-ANOVA revealed a significant Group x Tone period interaction (𝐹_1,28 = 4.422, 𝑝 = .0446). As confirmed by post hoc analysis, pre-tone baseline levels of freezing were similar between groups (𝑝 = .9995), but during the tone period, the hM4Di group displayed significantly less freezing than the mCherry group (𝑝 = .0178). These data suggest that CNO-mediated inhibition of pPVT neurons attenuated the behavioral expression of SEFR.

Next, we conducted the open field test with CNO on board (Fig. 6E). No significant differences in distance traveled or time spent in the center arena (data not shown) were observed between the hM4Di and mCherry groups (𝑡₂₆ = .0568, 𝑝 = .9551), indicating no effect of pPVT inhibition on locomotor activity or anxiety-like behavior. We then examined the effect of pPVT inhibition on associative fear memory retrieval by measuring context-evoked fear of the footshock stress context with CNO on board (Fig. 6F). Here, no significant difference in freezing was found between groups (𝑡₂₇ = 1.234, 𝑝 = .2278). Lastly, we performed the two-day SEFL procedure with CNO administered during both the SEFL conditioning and SEFL test sessions. During SEFL conditioning (Fig. 6G), we observed no significant main effect of Group (𝐹_1,28 = .6567, 𝑝 = .4246) or Group x Shock Period in-teraction (𝐹_1,28 = .0077, 𝑝 = .9308). In the SEFL test session (Fig. 6H), freezing behavior did not differ significantly between groups (𝑡₂₈ = 1.245, 𝑝 = .2234). Overall, CNO-mediated inhibition of pPVT did not affect anxiety-like behavior, expression of associative fear of the stress context, or SEFL. Notably, the effect of this manipulation on the SEFR tone test, but lack of an effect on SEFL, indicates that the pPVT selectively supports the expression of sensitization of unlearned fear but not the sensitization of fear learning.

Artificial activation of pPVT in stress-naïve mice potentiates unlearned fear but not fear learning

Next, we asked whether chemogenetic activation of pPVT neurons in stress-naïve mice was sufficient to recapitulate the sensitized fear observed in footshock-stressed mice. An AAV encoding either the excitatory hM3Dq receptor (AAV8-hSyn-hM3Dq-mCherry) or the fluorescent mCherry reporter (AAV8-hSyn-mCherry) was infused into the pPVT (Fig. 7A). After recovery, mice were tested in the SEFR tone test, open field, and SEFL procedure. CNO was administered 30 min prior to each of the following behavioral sessions: the SEFR tone test, open field, and SEFL conditioning (Fig. 7B). In the SEFR tone test (Fig. 7C), a RM-ANOVA revealed a significant Group x Tone Period inter-action (𝐹_1,31 = 6.133, 𝑝 = .0189). Post hoc analysis showed no group differences during the pretone baseline period (𝑝 = .9992). However, during the tone period, the hM3Dq group displayed significantly higher levels of freezing relative to the mCherry group (𝑝 = .0056). In the open field test (Fig. 7D), chemogenetic activation of the pPVT did not significantly affect distance traveled (𝑡₂₇ = 1.787, 𝑝 = .0852). Finally, to examine whether activation of pPVT enhances fear learning to a novel context, we conduced the SEFL procedure with CNO administered prior to the SEFL conditioning session. During SEFL conditioning (Fig. 7E), a RM-ANOVA revealed a significant main effect of Shock Period (𝐹_1,26 = 31.12, 𝑝 < .0001) and a significant Group x Shock Period interaction (𝐹_1,26 = 5.939, 𝑝 = .0220). Pre-shock freezing levels did not differ between groups, indicating that CNO administration did not alter baseline fear. However, following footshock, the hM3Dq group displayed significantly elevated freezing compared to the mCherry group (𝑝 = .0099). Despite this, freezing levels in the SEFL test session did not differ between groups (Fig. 7E; 𝑡₂₆ = 0.3591, 𝑝 = .7224). In summary, chemogenetic activation of the pPVT in stress-naïve mice was sufficient to recapitulate the sensitization of unlearned fear to a novel tone (SEFR), but did not enhance fear learning (SEFL).

Figure 7.

Inhibition of pPVT fails to attenuate spontaneous tone-elicited fear in stress-naïve mice

Our results indicate that pPVT neuronal activity is required for the sensitizing effects of stress in the SEFR tone test. However, it is unclear whether PVT activity modulates fear expression generally or only after exposure to footshock stress. To test whether chemogenetic inhibition of pPVT could diminish freezing behavior in the SEFR tone test with stress-naïve mice, we conducted a tone test in which the tone amplitude was increased to a level (115 dB) at which it evokes freezing even in stress-naive mice. Mice received viral infusion of AAV encoding the inhibitory hM4Di receptor or the fluorescent mCherry reporter (Fig. 8A). Following recovery, all mice were tested in the modified SEFR tone test with CNO on board (Fig. 8B). A RM-ANOVA on freezing behavior revealed a main effect of Tone Period (𝐹_1,17 = 37.50, 𝑝 < .0001), demonstrating that the tone was of sufficient intensity to elicit elevated levels of freezing relative to the pre-tone period. However, we did not detect a significant main effect of Group (𝐹_1,17 = .07030, 𝑝 = .7941) or a significant Group x Tone Period interaction (𝐹_1,17 = .06350, 𝑝 = .8041). This finding suggests that pPVT neuronal activity does not modulate expression of fear in mice that have not been exposed to stress.

Figure 8.

Discussion

In this study, we identified a novel role for the pPVT in stress-induced fear sensitization. We found that stress produces long-lasting hyperexcitability of the pPVT, selectively in response to a novel tone used to demonstrate SEFR, but not to the footshock used to induce SEFL. Using chemogenetic inhibition, we demonstrated that pPVT activity is necessary for both the induction and expression of SEFR, but not SEFL. Conversely, artificial activation of the pPVT mimicked the effects of footshock in the SEFR tone test but failed to enhance fear conditioning in a novel context. Together, these findings suggest that pPVT hyperactivity mediates SEFR but not SEFL.

We first established that associative fear of the stress context, SEFR, and SEFL are dissociable outcomes of footshock stress at the behavioral level. We showed that different intensities of foot-shock stress yielded distinct behavioral patterns across these assays. While exposure to 0.25 mA stress was sufficient to induce associative fear memory of the stress context, only higher intensities (0.5 mA and 1 mA) significantly enhanced freezing in the SEFR tone test and enhanced fear conditioning in the SEFL procedure. Importantly, we found no significant correlations between freezing behavior in the stress context test, SEFR tone test, and SEFL test in stressed mice, providing direct evidence from our study for the behavioral dissociation of these stress-induced fear responses.

To investigate the neural basis of SEFR, we employed c-Fos mapping and identified the pPVT as a region exhibiting persistently elevated activity after footshock stress in response to a novel tone. Furthermore, c-Fos expression in the PVT was significantly correlated with the level of freezing to the tone in the SEFR test, suggesting a link between PVT activity and the behavioral expression of sensitized unlearned fear. We corroborated this finding using fiber photometry to monitor realtime Ca²+ signals in pPVT neurons during testing. We observed that footshock stress potentiated pPVT Ca²+ responses specifically during the presentation of the novel tone in the SEFR test. In contrast, pPVT responses to the single footshock used during the SEFL conditioning session were comparable between stressed and non-stressed mice.

To test the causal role of pPVT activity in SEFR, we employed chemogenetic manipulations. We found that pPVT activity during footshock stress was necessary for the induction of SEFR, as mice receiving pPVT inhibition during stress subsequently showed attenuated tone-elicited freezing in the SEFR tone test. Notably, pPVT inhibition during stress did affect acquisition of associative fear of the stress context or contextual fear in the SEFL test. pPVT activity during the SEFR tone test itself was necessary for the expression of SEFR, but inhibition of pPVT during testing did not affect anxiety-like behavior in the open field, associative fear memory retrieval, or contextual fear in the SEFL test. Conversely, chemogenetic activation of pPVT neurons in stress-naive mice was sufficient to recapitulate the sensitized freezing response to the novel tone in the SEFR test, effectively phenocopying the effect of footshock stress. This activation did not affect exploratory behavior in the open field or contextual fear in the SEFL test. Importantly, chemogenetic inhibition of pPVT in stress-naive mice did not attenuate freezing responses to a loud tone (115dB), suggesting that pPVT activity does not modulate expression of unlearned fear generally, but rather selectively supports the sensitized fear response induced by stress. Collectively, these chemogenetic results establish pPVT as critical for both the induction and expression of SEFR but dispensable for SEFL and associative fear.

Although silencing the pPVT in stressed mice produced robust effects on SEFR, the same manipulation did not affect associative fear of the stress context or SEFL. These findings were somewhat unexpected, given prior studies showing that the PVT modulates both the acquisition and expression of Pavlovian fear conditioning (Penzo et al., 2015). However, the literature on this topic is mixed. While some studies demonstrate a role for the PVT in associative fear learning (Penzo et al., 2015; Do-Monte et al., 2015), others have reported selective or no effects following PVT lesions or inactivation (Li et al., 2014; Choi et al., 2019). These inconsistencies could be explained by the functional heterogeneity of the PVT (McGinty and Otis, 2020). Distinct PVT cell populations can exert opposing effects on behavior (Ma et al., 2021), highlighting a potential limitation of global silencing, which may obscure effects of behaviorally relevant subpopulations. This possibility warrants further investigation.

Our observation that the pPVT contributes to SEFR is consistent with recent evidence that stress inclines animals towards heightened levels of fear when faced with uncertain or ambiguous threats and that the PVT mediates this stress-induced shift in threat evaluation (Wang et al., 2023, 2024). In our study, the SEFR procedure involves presenting a novel tone in a novel chamber to mice with no prior experience with the stimulus or context. This makes the tone stimulus, although loud, inherently ambiguous in terms of its predictive value for danger. We suspect that the ambiguous nature of the tone engages the pPVT, and stress-induced hyperactivation of this region biases mice towards increased fear responses to such stimuli. This interpretation may also explain the lack of pPVT involvement in SEFL, where the explicit pairing of a context with footshock makes the threat value of the context unambiguous.

Our findings identify pPVT hyperactivity as a mechanism by which stress sensitizes unlearned fear. This interpretation aligns with prior reports showing that acute stress primes PVT responses to subsequent threats (Pliota et al., 2018; Beas et al., 2018; Jász et al., 2025). One potential driver of this hyperactivity is stress-induced plasticity within the pPVT. Notably, the PVT is composed exclusively of glutamatergic projection neurons whose activity is tightly regulated by stress hormones, neurotransmitters, and neuropeptides (Li and Kirouac, 2012; Curtis et al., 2021). These modulatory signals may potentiate PVT activity through alterations in intrinsic neuronal properties. For example, stress-related neuromodulators such as dynorphin (Chen et al., 2015) and orexins (Kolaj et al., 2007) are known to influence cellular excitability in the PVT. Alternatively, stress may alter presynaptic inputs to the PVT to facilitate increased activity. For instance, dopamine released from locus coeruleus terminals during footshock stress suppresses inhibitory pre-synaptic input onto PVT neurons, thereby increasing their excitability (Beas et al., 2018).

Our study shows that pPVT is necessary for the induction of SEFR during stress, suggesting that it receives and processes information about the stressor. Identifying the specific projection pathways that transmit stress information to the pPVT is crucial for understanding how stress initiates the sensitization process. Anatomical studies point to brainstem inputs like the lateral parabrachial nucleus (lPBN) and dorsolateral periaqueductal gray (dlPAG) as potential sources of nociceptive information projecting to pPVT (Li and Kirouac, 2012). Both lPBN and dlPAG are involved in transmitting threat signals and modulating defensive behavior (Li and Kirouac, 2012; Sato et al., 2015; Kim et al., 2013). Future directions should involve leveraging techniques like viral-mediated retrograde tracing to identify specific neuronal populations projecting to pPVT that are activated by aversive stimuli and contribute to SEFR induction.

Furthermore, it remains unclear whether specific populations of pPVT neurons are responsible for both the induction and expression of fear sensitization. The possibility that stressful experiences are "encoded" by a subset of pPVT neurons and subsequently reactivated by ambiguous threats to sensitize fear responses is intriguing. A recent study using activity-tagging demonstrated that PVT ensembles active during early life stress regulate adult behavior (Kooiker et al., 2024), supporting the idea of stress-specific memory traces. The PVT is known to exhibit significant heterogeneity, with different cell types potentially having opposing influences on behavior and distinct "stress-responsive" populations identified (Gao et al., 2020; Beas et al., 2018). Future characterization of pPVT heterogeneity using modern techniques like single-nucleus RNA-sequencing is a critical next step to clarify whether distinct neuronal subpopulations underlie the induction and expression of SEFR.

Stress is known to sensitize a range of fear and defensive behaviors, but few studies have directly tested whether different forms of fear sensitization arise from a shared mechanism or are mediated by behavior-specific circuits. Recent work suggests the latter. Manipulations of the dorsal hippocampus or ventromedial prefrontal cortex affect cued and contextual SEFL differently (Hersman et al., 2019; Pennington et al., 2017). Moreover, Pennington et al., 2024 showed that footshock stress-induced alterations in SEFL and avoidance behavior are selectively mediated by the basolateral amygdala and ventral hippocampus, respectively. Our results support this emerging view, revealing that SEFR and SEFL are dissociable forms of fear sensitization governed by distinct neural substrates.

The PVT has been proposed to integrate prior aversive experiences and regulate affective behaviors through its widespread limbic connections. Supporting this role, previous studies have shown that PVT activity contributes to stress-induced alterations in anxiety-like behavior, passive coping, and avoidance (Zhu et al., 2022; Pliota et al., 2018; Ma et al., 2021; Dong et al., 2020). However, these studies often do not clearly distinguish whether such changes arise from associative or nonassociative processes. For instance, some stress-induced behavioral alterations can be reversed by extinction or through pharmacological suppression of the stress memory, suggesting that associative mechanisms may be involved (Hassien et al., 2020; Blundell et al., 2005). To clarify the specific contribution of the pPVT to nonassociative fear sensitization, we demonstrated that it is essential for SEFR, which is a nonassociative behavioral phenotype. These findings expand the known behavioral functions of the PVT and position the pPVT as a key node in the circuitry linking stress to heightened unlearned fear responses.

In conclusion, our study characterized a set of behavioral and neural mechanisms by which exposure to stress leads to the induction and expression of sensitization of unlearned fear (SEFR). We established that SEFR and SEFL are distinct behavioral manifestations of fear sensitization supported by discrete neural mechanisms. Specifically, we found that exposure to footshock stress leads to long-lasting alterations in pPVT neural activity patterns that selectively underlie SEFR. These results have important implications for PTSD, particularly for symptoms such as hyperarousal and reactivity, which our SEFR tone test is intended to model. This work emphasizes the need for greater research into the nonassociative consequences of stress to achieve a comprehensive understanding of the full range of PTSD symptoms. Our findings highlight the pPVT as a potential therapeutic target for stress-induced sensitization of unlearned fear, a class of symptoms that may be resistant to therapies focused solely on associative processes.

Methods and Materials

Subjects

Adult male and female mice (12-18 weeks) from 129S6/SvEvTac background (Taconic, Germantown, NY) were bred in house, weaned at postnatal day 21 and group housed (3-5 per cage). Mice were maintained on a 12h light/dark cycle with food and water available ad libitum. Experiments were conducted during the light phase. Mice were randomly assigned to groups prior to each experiment. Since no significant sex-by-group interactions were observed, data from males and females were combined for analysis. All animal procedures were conducted according to the National Institutes of Health guidelines on the Care and Use of Laboratory Animals and were approved by the University of Texas at Austin Institutional Animal Care and Use Committee.

Apparatus

Conditioning chambers (30.5 × 24 × 21) consisted of three aluminum walls and ceiling, a clear Plexiglas door, and stainless-steel grid flooring (Med Associates VFC-005A). Chambers were individually housed in sound-attenuating cabinets and illuminated with overhead white light throughout the entire session. The conditioning chambers were equipped with near-infrared cameras that recorded at 30 frames/s and a speaker for tone delivery. Videos were automatically scored (Video Freeze software; Med Associates) using a linear pixel change algorithm that defined freezing as the absence of all movement, except for those related to breathing. In order to design distinct contextual environments across behavioral procedures (Footshock stress, SEFR tone test, and SEFL procedure), modifications were made to the chamber’s scent, floor, walls, ceiling, and transportation.

Open field apparatus consisted of a 40 x 40 cm² arena with opaque plastic walls 35 cm high. The arena was illuminated with white incandescent bulbs producing 85 lux at the center of the arena. Behavior was recorded using digital video cameras mounted above the arenas and analyzed using AnyMaze software (Stoelting Inc.).

Behavioral procedures

Footshock stress: Mice received four footshocks (2 sec duration, 1.0 mA intensity), delivered through the grid floor of the chamber at 160, 240, 320, and 400 sec after placement. Forty seconds following the final footshock, mice were returned to their homecage. NoStress control mice were exposed to the chamber for the same duration (7 min total) but did not receive any shocks. To promote contextual discrimination between the stress context and future testing environments, mice underwent extinction sessions consisting of 5-minute re-exposures to the footshock context once daily for five consecutive days following the stress session. The first extinction session also served as the test for stress context recall.

SEFR tone test: Following a 3 min acclimation period, mice were presented with three 30 sec tones (90 dB, 9 kHz, 50 ms rise/fall time), each separated by a 30 sec interstimulus interval. For analysis, freezing behavior was segmented into two epochs: the pre-tone period (defined as the duration before the onset of the first tone) and the tone period (defined as the duration following the onset of the first tone). The tone period includes freezing during all three tone presentations and the corresponding interstimulus intervals, as freezing behavior was consistent across these events. An earlier study from our group reported that female, but not male, mice exhibited elevated freezing to the tone regardless of prior stress exposure (Hassien et al., 2020). Since that publication, the protocol has been refined to eliminate this effect. Specifically, we increased the discriminability between the footshock stress and SEFR tone test contexts by modifying visual, olfactory, and transportation cues. As a result, we no longer observe enhanced freezing in stress-naïve female mice under the updated protocol.

Open Field: Testing was conducted in a separate behavioral room from the other behavioral experiments. Mice were allowed to explore the arena for 30 min.

Stress-enhanced fear learning: On Day 1 of the SEFL procedure (SEFL conditioning session), mice received a single footshock (2 sec, 0.75 mA) delivered through the grid flooring 3 min after the start of the session. Mice were removed from the chambers 30 sec after footshock delivery and returned to homecage. On day 2 (SEFL test session), mice were placed into the same context from the previous day for 5 min to assess context-elicited freezing.

Surgery

Mice were anesthetized using isoflurane (3% in induction chamber, maintained at 1.5% for remainder of surgery) and were placed in a stereotaxic apparatus. Prior to skin incision to reveal the skull, mice received a subcutaneous injection of buprenorphine (10 mg/kg) and carprofen (0.1mg/kg). Body temperature was maintained at 37 °C with a heating pad and ophthalmic ointment was applied to prevent the eyes from drying. The skull was exposed and a 1 mm diameter craniotomy was made at the site of injection. A pulled glass containing virus was slowly lowered to the target site. Viral coordinates were selected according to the atlas of Paxinos and Franklin. After viral infusions, the injector remained in place for 10 min to allow for viral diffusion.

For chemogenetic manipulation experiments, mice received unilateral viral infusion of AAV8-hSyn-hM3Dq-mCherry (Addgene: 50474), AAV8-hSyn-hM4Di-mCherry (Addgene: 50475), or AAV8-hSyn-mCherry (Addgene: 114472) in the pPVT (AP -1.6, ML 0.0, DV -3.3). Virus was delivered at a rate of 23.5nL/min for a total of amount of 138 nL. Following 3-4 weeks of recovery, clozapine-N-oxide (5mg/kg) was administered via I.P. injection 30 min prior to the start of behavior testing.

For fiber photometry experiments, mice received unilateral viral infusion of AAV1-syn-jGCaMP8s-WPRE (Addgene: 162374) in pPVT (AP -1.6, ML 0.0, DV -3.3). Following viral infusion, a second craniotomy was made and acidic gel etchent was applied to the surface of the skull followed by a layer of OptiBond epoxy (Kerr Corporation). Optical fiber cannulas with 400 um core diameter (Doric Lenses) were unilaterally implanted at a 7 degree angle targeting the pPVT (AP -1.6, ML 0.0, DV -3.3). Dental cement (Bosworth) was applied to secure the cannula to the skull. Mice were given 2 weeks to recover prior to photometry recordings.

Tissue preparation and histology

Mice were deeply anesthetized with ketamine/xylazine (150/15mg/kg) and transcardially perfused with 0.01M phosphate buffered saline (1x PBS), followed by 4% paraformaldehyde (PFA) in 1x PBS. Brains were extracted and post-fixed at 4C in 4% PFA overnight then transferred to 30% sucrose solution in 1x PBS at 4C for 24 hr prior to cryopreservation. Coronal tissue sections (35um) were prepared on a cryostat.

For c-Fos immunohistochemistry, sectioned tissue was incubated in 1x PBS/methanol solution (1:1) with 1% hydrogen peroxide for 15 min. Sections were then washed in 1x PBS with 0.5% Triton-X (1x PBST) three times and blocked at room temperature in 1x PBST with 5% normal donkey serum (NDS) for 1.5 hr. Sectioned tissue was then incubated in for 24 hr at 4C degrees in a primary antibody solution (1:2000 rabbit anti-c-Fos polyclonal; Synaptic Systems 226-003) diluted in 1x PBST with 5% NDS. Sections were then washed in 1x PBST and then treated with secondary antibody solution (biotinylated goat anti-rabbit 1:500; Abcam ab64256) for 2 hrs at room temperature. Tissue sections were washed in 1x PBS then incubated in Vectastain Elite ABC reagent for 30 min followed by 1x PBS wash. Next, tissue sections were incubated in DAB peroxidase substrate solution (Vector laboratories) for 8 min then immediately washed in 1x PBS and mounted onto slides. Lastly, tissue sections were dehydrated through a graded series of alcohol concentrations, cleared with Citrasolv (Fisher Scientific), and coverslipped with Fluoromount-G (SouthernBiotech).

For fiber photometry and chemogenetic manipulation experiments, brain tissue was collected at the conclusion of behavioral testing. Tissue sections were stained with 1:1000 DAPI solution and mounted onto slides before being coverslipped with Fluoromount-G (SouthernBiotech). Fluorescent images were obtained using a Zeiss Axio Zoom.V16. Viral expression and optical fiber placement were confirmed according to the mouse brain atlas of Franklin and Paxinos.

Cell counting and quantification

Images of c-Fos positive cells were obtained using a Zeis Axio Lab.A1 microscope and Zen imaging software. Images were collected at the same exposure settings and dimensions. Regions of interest were chosen a priori and a sample size of approximately 3-5 images per region were collected for each subject. Images were quantified in ImageJ using an automated counting algorithm based on experimenter-defined cut off metrics including the size, circularity, and signal intensity relative to background. Automated cell quantification was validated by comparing results to manual counts in a randomly selected subset of images. The c-Fos quantification was calculated by averaging the number of c-Fos positive cells per mm2, per animals, then per group.

Fiber photometry

in vivo recordings of fluorescent signal were collected using a commercial fiber photometry system (Tucker-Davis Technologies; RZ10x) and data acquisition was performed on TDT synapse software. Excitation of GCaMP8s and isosbestic signals was achieved using blue 465 nm (Lx465; TDT) and ultraviolet 405 nm (Lx405) LEDs sinusoidally modulated at 331 Hz and 211 Hz, respectively. The mean fiber power was 40-60𝜇W. Excitation light was passed through a dichroic mirror (Doric Minicube) coupled to a fiber optic patch cord (400 um fiber core diameter, 0.66 NA; Doric Lenses) that was connected to the cranial implanted fiber (200 um fiber core diameter, 0.66 NA; Doric Lenses). Emission signals were filtered at 6 Hz and acquired at a sampling frequency of 1017 Hz. Fiber photometry data were analyzed using custom-written MATLAB scripts adapted from Barker et al., 2017. To account for time-dependent slow drift artifacts caused by photobleaching, a least-squares linear fit was applied to the isosbestic signal to obtain a fitted control signal. 𝑑𝐹 / 𝐹 was computed by subtracting the fitted control signal (𝐹₀) from the 465 nm raw signal (𝐹), and dividing by the fitted control signal (𝐹 − 𝐹₀/𝐹₀). For epoch-based analysis, Z-scores were calculated as (𝑑𝐹 / 𝐹 − 𝑑𝐹 / 𝐹_{𝑚𝑒𝑎𝑛})/𝑑𝐹 / 𝐹_𝑆𝐷) where the mean and standard deviation correspond to the 5 seconds prior to each event of interest.

Statistical analysis

For freezing behavior analysis, we calculated the percentage of time mice spent freezing during predefined behavioral event windows. For the footshock stress session, freezing was quantified during the pre-shock baseline period and each of the four post-shock periods. For the SEFR tone test, if no significant differences were observed between the three tone presentations and the subsequent interstimulus intervals, freezing was calculated across two time windows: the pre-tone period (time prior to the first tone) and the tone period (time following the onset of the first tone, including all tones and intervals). For the SEFL conditioning session, freezing was measured during the pre-shock baseline and the 30 sec post-shock period prior to removal from the chamber. Lastly, for the stress context test and SEFL test, the percentage of time spent freezing was calculated for the entire 5 min test sessions. Data were analyzed using ANOVA or repeated-measures ANOVA when appropriate. Significant main effects and interactions were probed using Sidak’s post hoc tests for between-subject pairwise comparisons and Bonferroni-adjusted paired t-tests for within-subjects comparisons, except for Fig. 2 in which Dunnett’s test was used to compare multiple stress groups against a single NoStress group. Statistical analyses were performed in Prism 6 (GraphPad Software).

Acknowledgements

Supported by NIH grants R01MH117426 and R21MH128610 to MRD, T32MH106454 to KJN.