Exercise Leverages a Post-Stress Therapeutic Window in Hypothalamic CRH Neurons to Enable Recovery from Threat Sensitization

16 17 Exercise provides a range of benefits for physical and mental health. It relieves stress, 18 yet paradoxically, it recruits the body’s stress response system. Here we show that 19 activation of corticotropin- releasing hormone cells of the paraventricular nucleus of the 20 hypothalamus (CRHPVN) and the endocrine arm of the stress axis is necessary, but not 21 sufficient for the beneficial effects of exercise. Specifically, glucocorticoids act 22 synergistically with brain -derived neurotrophic factor (BDNF) on CRH PVN cells during 23 exercise to reverse behavioral and synaptic sensitization after stress. In the absence of 24 exercise, optogenetic activation of the tropomyosin-related kinase B (TrkB) receptor after 25 stress is sufficient to reverse behavioral sensitization and synaptic metaplasticity. Our 26 findings reveal a novel, time-sensitive mechanism by which exercise alleviates the impact 27 of acute stress. 28 29 30 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 2 Exercise is widely recognized as a preventive strategy against various metabolic and 31 neuropsychiatric conditions (1-3). Growing evidence supports the role of regular exercise 32 in fostering stress resilience (4 -6). Exercise also promotes adaptive responses to acute 33 stress (4, 7, 8) and single bouts of exercise have immediate mood-enhancing effects (9). 34 Paradoxically, exercise itself activates the hypothalamic pituitary adrenal (HPA) axis and 35 increases circulating stress hormones (10-13). How exercise can elicit both a short-term 36 stress response and engage compensatory neurobiological processes promoting long-37 term stress regulation is not known. 38 39 Here, we leveraged synaptic and behavioral readouts of stress sensitization to 40 understand how exercise modulates the lasting consequences of acute stress. Using a 41 combination of ex vivo electrophysiology, genetic, optogenetic, and behavioral 42 approaches, we identify a critical period immediately after stress that may serve as a 43 therapeutic window for the effects of exercise. Specifically, we show that exercise after 44 stress increases both circulating corticosterone (CORT) and hypothalamic brain-derived 45 neurotrophic factor (BDNF) concentrations. The synergistic action of these signaling 46 elements in CRH PVN neurons reverses behavioral threat sensitization and synaptic 47 metaplasticity induced by stress. These findings provide the first mechanistic insights into 48 how exercise alleviates the lasting effects of stress and reconcile previous observations 49 regarding the timing and efficacy of exercise as a stress intervention. 50 51

52 53 Exercise prolongs CRH PVN activation and CORT secretion after stress but reverses its 54 synaptic consequences when delivered within a critical period 55 56 Exercise relieves stress but also promotes endocrine stress signaling, highlighting a 57 physiological paradox (12, 14). To understand this, we first assessed the effects of acute 58 stress and exercise on two components of the HPA axis: the activity of CRHPVN neurons 59 and circulating concentrations of CORT. Mice expressing GCaMP6f in CRHPVN neurons 60 were unilaterally implanted with a ferrule for single fiber photometry (Fig. 1A-C). Baseline 61 measurements were obtained in the home cage (HC). One group of mice received 10 foot 62 shocks (FS) in a 5- min period; another group was placed on a treadmill and allowed to 63 run for 1 hour (Fig. 1E). FS evoked a robust increase in CRH PVN activity that returned to 64 baseline levels ~10 minutes after returning mice to HC (Fig. 1e, left panel). Exercise also 65 evoked a robust increase in CRH PVN activity that remained elevated through the entirety 66 of running session (Fig. 1E, right panel). Tail blood was collected before experimental 67 procedures started for quantifying baseline (BL) CORT levels. Then, a second sample 68 was collected after treatments for post-stimuli measures (Post). This is consistent with the 69 classical view that both acute stress and exercise increase CORT (Fig. 1F) and provides 70 new information that exercise directly increases the activity of CRHPVN neurons. 71 72 Exercise is often used as a tool for stress relief. To better understand the connection 73 between stress and exercise, we examined the effects of acute stress and exercise on 74 CRHPVN activity and CORT. Mice were subjected to FS and then placed on the treadmill 75 immediately (Fig. 1G). As expected, FS strongly increased CRHPVN activity. Exercise after 76 stress sustained this increase in activity (Fig. 1 G). When averaging CRH PVN activity 77 across the first hour after FS, mice that exercised after stress showed significantly higher 78 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 3 activity levels compared to stressed mice that did not exercise (Fig. 1H ). In contrast, FS 79 did not alter overall CRH PVN activity during running itself, as both exercise groups 80 displayed comparable levels (Fig. 1H ). To measure the effects of these treatments on 81 CORT levels, we collected blood samples before stress , after FS, and after exercise. In 82 FS mice, in the absence of exercise, CORT levels were increased 15 minutes after FS, 83 then declined one hour later (Fig. 1I). In contrast, post- stress exercise further elevated 84 CORT concentrations beyond those induced by stress alone (Fig. 1I ). Both stress and 85 exercise strongly activate the HPA axis, with post -stress exercise maintaining the 86 elevation in CRHPVN activity and slowing the recovery of the CORT response. 87 88 Next, we probed the effects of exercise after stress on a synaptic biomarker that has been 89 identified after acute stress and may contribute to sensitization of this system : 90 metaplasticity at glutamate synapses on CRH PVN neurons (15, 16, 17). Following acute 91 stress, glutamatergic synapses exhibit activity -dependent short-term potentiation (STP) 92 that is not evident in naïve (unstressed) mice or rats. STP requires local CRH release (16) 93 and can be elicited by optogenetic activation of CRHPVN neurons in the absence of actual 94 stress (17). To test the effects of exercise on STP after FS, we prepared brain slices from 95 naïve and stressed mice, and from mice that exercised after stress, and whole-cell patch 96 clamp recordings were obtained from CRH PVN neurons (Fig. 1J). In naïve animals, high 97 frequency stimulation (HFS) of glutamate inputs had no effect on excitatory postsynaptic 98 current (EPSC) amplitude (Fig. 1 K), consistent with previous reports that STP is not 99 evident if animals are not stressed (15, 16, 17) . As expected, STP was evident at 100 glutamate synapses from stressed mice. We have previously shown that STP can be 101 reliably induced up to three days after a single acute stress (15). If mice were subjected 102 to exercise after FS, however, we observed no STP (Fig. 1L). This suggests that exercise, 103 despite increasing CRHPVN activity, attenuates STP after stress ( Supplemental 104 information Fig. S1). These observations show that although exercise causes a sustained 105 increase in the activity of CRH PVN neurons and amplifies HPA axis activation beyond 106 stress levels, it counteracts the effects of stress on excitatory inputs onto CRH PVN 107 neurons. 108 109 Delivering interventions within a post-stress sensitive window may be critical to modulate 110 synaptic plasticity such as STP (17, 18). Similar plasticity windows have been described 111 (19), and evidence suggests that timely interventions can buffer the consequences of 112 stress (20). To test whether delaying exercise after stress affected its capacity to buffer 113 STP, mice subjected to FS were returned to HC (Fig. 1m) for fourteen hours before having 114 access to exercise (S14h delayed EX ) or remain ing in their HC (S 14h delay). Robust STP was 115 detected in neurons from the control group S14h delay mice (Fig. 1N), confirming the lasting 116 impact of stress in CRH PVN neurons (17). The S14h delayed EX mice also show ed a robust 117 increase in EPSC amplitude that was indistinguishable from the S 14h delay mice (Fig. 1O). 118 This evidence demonstrates that delayed exercise is ineffective in reversing the synaptic 119 changes induced by acute stress, hinting at a potential sensitive window immediately after 120 stress. 121 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 4 122 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 5 Fig. 1. Exercise prolongs CRH activation and CORT after stress but reverses the synaptic 123 consequences of stress when delivered within a sensitive window. A. Ferrule implantation 124 for fiber photometry recordings. B. Representative confocal image of CRH PVN neurons 125 expressing GCaMP6f. C. Schematic of single-fiber photometry method. D. Experimental 126 design. E. In vivo recordings of CRHPVN activity in stress (S: N = 10), exercise (E: N = 11), 127 and stress followed by exercise ( S + E: N = 12) mice. F. CORT concentration from S (N 128 = 11) and E mice (N = 10) before (baseline: BL) and after treatments (Post). M ixed 129 ANOVA: Time (F(1, 19) = 330.1, P < 0.001), Group (F(1, 19) = 12.88, P = 0.002), and 130 Time × Group (F(1, 19) = 22.63, P < 0.001). G. CRHPVN activity in S+E mice. H. Average 131 CRHPVN activity after FS and during exercise. One- way ANOVA: F(2, 31) = 10.44, P < 132 0.001. I. CORT concentrations for S (N = 22) and S + E (N = 23) mice . Mixed ANOVA: 133 Time (F(1.86, 80.00) = 136.2, P < 0.001), Group (F(1, 43) = 16.21, P < 0.001), and Time 134 × Group (F(2, 86) = 68.43, P < 0.001). J. Whole-cell patch clamp recordings from CRHPVN 135 neurons in naïve (N), S, and S+E mice. K. EPSC amplitudes following HFS (grey bar) 136 relative to BL (doted line). Mixe d ANOVA: Time (F(4.434, 199.5) = 7.454, P < 0.001), 137 Group (F(2, 45) = 7.682, P = 0.001), and Time × Group (F(8.867, 1995) = 3.232, P = 138 0.001). EPSC amplitude increased after HFS (min 5 vs min 7: P = 0.003) in S mice only. 139 L. Average EPSC amplitude one minute after HFS. O ne-way ANOVA: F(2, 45) = 11.62, 140 P < 0.001. M. Experimental design of whole- cell patch clamp N. EPSC amplitudes 141 following HFS. Mixed ANOVA: Time (F(4.219, 105.5) = 16.44, P < 0.001). Both S14h delay 142 (P = 0.013) and S 14h delayed E (P = 0.005) mice showed increased EPSC amplitude after 143 HFS. O. Average EPSC amplitude one minute after HFS was equally elevated in both 144 groups. Data are mean ± SEM. ***: Within-group, P < 0.001. 145 146 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 6 Exercise reverses stress-induced behavioral threat sensitization 147 148 A single acute stress also induces behavioral sensitization (21–23). To probe the links 149 between acute stress, exercise and behavioral threat response, we used the Dark -Light 150 test (DL) which takes advantage of the fact that mice naturally avoid open, brightly 151 illuminated spaces (24). We first investigated the effects of FS on multiple parameters in 152 this test. Mice were habituated to the dark compartment and then placed in one of two 153 groups: unstressed (FS context, no FS) and acute stress (FS) (Fig. 2A). The following 154 day, mice were placed into the dark compartment of the DL chamber, and the connecting 155 door was opened, providing free access to the light compartment for 5 minutes. 156 157 In the control group, each mice entered the light compartment within the first minute, so 158 we focused our analysis on this initial period (Supplemental information F ig. S2 for the 159 complete 5-minute analysis). Analysis of movement in the light compartment revealed 160 that naïve mice entered the compartment within the first minute (time to 1st entry: 14.68 ± 161 3.761 s) and explored the entire area, with bouts of exploration interspersed with periods 162 in the dark compartment (Fig. 2B, C, left panels). In contrast, stress exposure prolonged 163 time to first entry into the light compartment, and suppressed exploration of the light 164 compartment (Fig. 2 B-F). Specifically, stressed mice showed a delay in the first 165 exploration bout compared to naive mice (time to 1st entry: 85.46 ± 17.46 s, P < 0.001 vs 166 unstressed), with only 46% of the animals entering the light compartment within the first 167 minute of testing (Fig. 2E ; Supplemental information Fig. S2 G). Additionally, stressed 168 mice covered significantly less distance while exploring the light compartment (Fig. 2F ). 169 The third group was subjected to FS, followed immediately by exercise. A single session 170 of exercise fully reversed this behavioral phenotype (Fig. 2B , C). In mice subjected to 171 stress followed by exercise, the time spent in the light compartment was higher than in 172 stressed mice and comparable to that of naïve mice (Fig. 2 D). Moreover, exercise 173 restored the early exploration pattern observed in non- stressed mice (time to 1 st entry: 174 21.12 ± 5.65 s, P < 0.001 vs stressed), with 93% of these mice entering the light 175 compartment within the first testing minute (Fig. 2 E). Mice subjected to exercise after 176 stress also traveled a greater distance than stressed mice, with cumulative levels over 177 time being indistinguishable from those of naive mice (Fig. 2F ). These findings 178 demonstrate that a single session of exercise reverses stress -induced alterations in 179 approach-avoidance dynamics, consistent with a decrease in behavioral sensitization to 180 potential threat. 181 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 7 182 183 Fig. 2. Exercise reverses stress-induced behavioral threat sensitization. A. Experimental 184 design for the dark/light test (DL). Each group consisted of 13 male mice. Animals were 185 tested in the DL 24 hours after treatment. B. Heat maps representing the spatial 186 distribution of the average exploration time in the light compartment (top section of each 187 panel). Tracking plot of the exploration trajectory of a representative animal per group. 188 Behavioral time budget plot showing individual visits to the light (white bars) and dark 189 compartment (black bars) of the DL test during the first minute of testing (bottom section 190 of each panel). C. Frequency analysis of the exploration time in the light compartment 191 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 8 binned every five seconds. D. Summary of time spent in the light compartment. One-way 192 ANOVA: F(2, 36) = 6.132, P =.005). E. Frequency analysis of light compartment 193 exploration latencies during the first minute of testing. One-way ANOVA: F(2, 18)= 7.569, 194 P= 0.004. F. Distance travelled in the light compartment . One-way ANOVA: F(2, 36) = 195 8.900, P = 0.007). S showed a slow increase of cumulative distance over the first minute 196 of test, with S + E and N showing a similar pattern of fast and steady increase (right panel). 197 Tukey's multiple comparisons test was used. Data are mean ± SEM. 198 199 Exercise has no effect on contextual fear memory 200 201 Aversive stimuli such as FS generate lasting conditioned fear memory, expressed as 202 freezing when animals are re- exposed to the threat -associated context (25, 26). Since 203 exercise following stress negates stress-induced metaplasticity and threat behavioral 204 sensitization, we investigated whether it may impact the memory of the threat . Mice 205 received FS and were either returned to HC or allowed to run on the treadmill (Fig. 3A). 206 An additional group of naïve controls was exposed to the FS chamber for 5 minutes 207 without receiving shocks. Twenty-four hours later, mice were re-exposed to the FS context 208 for 5 minutes. Compared to naïve controls, FS context exposure elicited robust freezing 209 in both stressed mice and those that exercised after stress, with no detectable differences 210 between the stress and stress + exercise groups (Fig. 3B). Exploratory behaviors, 211 including walking and rearing, were markedly reduced in these groups relative to naïve 212 mice (Fig. 3C–D). Additionally, FS context elevated circulating CORT levels in both stress 213 mice and mice exercising after stress, beyond those observed in naïve subjects (Fig. 3E). 214 These observations indicate that exercise had no effect on the contextual fear memory or 215 the CORT response. 216 217 218 219 Fig. 3 . Exercise has no effect on contextual fear memory. A. Experimental design for 220 contextual fear conditioning. Each group consisted of 8 male mice. Mice in all groups were 221 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 9 exposed to the FS chamber for 5 minutes. Stress (S) and Stress + Exercise (S+E) mice 222 were exposed to FS and then either returned to their HC (S) or placed on a treadmill for 1 223 hour of running (S+E). Naïve (N) mice were not exposed to FS and were returned directly 224 to HC. Twenty-four hours later, mice were re -exposed to the FS chamber for 5 minutes 225 and the following behaviors were analyzed automatically: freezing (percentage of total 226 time), walking (meters), and rearing (seconds). Mice returned to HC after that, and tail 227 blood was collected 15 minutes later for CORT quantification using ELISAS. B. Individual 228 freezing percentage per group. One- way ANOVA: F(2, 21) = 63.13, P < 0.001. C. 229 Individual distance travelled per group. One-way ANOVA: F(2, 21) = 33.32, P < 0.001. D. 230 Individual rearing duration. One-way ANOVA: F(2, 21) = 34.37, P < 0.001. E. Individual 231 CORT concentration per group. One- way ANOVA: F(2,21) = 4.90, P = 0.018. Tukey's 232 multiple comparisons test was used. Data are mean ± SEM. 233 234 Elevated CORT synergizes with a TrkB ligand to prevent stress-induced metaplasticity 235 236 Post-stress exercise elevates HPA axis activity beyond stress levels, yet it buffers the 237 synaptic consequences of stress. One possible explanation is that the prolonged 238 elevation in CORT is sufficient to buffer the effects of stress. To test whether the increased 239 and prolonged CORT output is sufficient to deter STP expression, we mimicked exercise-240 induced modulation of the HPA axis by using a 1- hour immobilization (IMMO) protocol. 241 Mice were unilaterally implanted for fiber photometry recordings of CRHPVN activity during 242 IMMO (Fig. 4A). IMMO increased CRHPVN activity for the duration of the session (Fig. 4A). 243 As expected, CORT was markedly elevated 1 hour after IMMO onset (Fig. 4B). Next, we 244 evaluated the effects of IMMO on STP (Fig. 4 C, top section). W hole-cell patch clamp 245 recordings reveal ed robust STP (Fig.4C , bottom section). Thus, prolonged CRH PVN 246 activity and elevated CORT do not deter the expression of STP. It is unclear, however, 247 whether CORT itself after stress would affect STP. To test this idea, brain slices from 248 stressed mice were incubated in a bath containing artificial cerebrospinal fluid (aCSF; 249 SaCSF) or CORT (100 nM) diluted in aCSF (S CORT) for 1 hour (Fig. 4D ). Whole-cell patch 250 clamp recordings of both S aCSF and SCORT mice showed STP (Fig. 4e ). Together, these 251 observations indicate that the elevated circulating CORT induced by post-stress exercise 252 is insufficient to prevent STP, suggesting that a distinct molecular mediator may be 253 responsible for the buffering effects of exercise. 254 255 Regular exercise promotes BDNF release, a neurotrophin critical for neuronal growth, 256 survival, and diverse forms of synaptic plasticity, which acts primarily through its high -257 affinity receptor, tropomyosin receptor kinase B (TrkB) (27–32). T he capacity of a single 258 exercise bout to elevate BDNF varies between brain regions (33, 34) , and t o our 259 knowledge, it is unknown whether acute exercise elevates BDNF in the PVN, especially 260 in stressed mice. Therefore, we assessed BDNF concentrations in bilateral PVN samples 261 from stressed mice and mice that exercised after stress (Fig. 4F). Protein quantification 262 revealed a significant increase in BDNF concentrations in mice that exercised after stress 263 compared to stressed controls (Fig. 4 G). Since post -stress exercise locally increases 264 BDNF in the PVN, we next tested whether activating TrkB receptors in the PVN of 265 stressed mice could buffer STP. Brain slices were incubated for 1 hour in aCSF containing 266 the TrkB ligand 7,8‑Dihydroxyflavone (DHF; 20 µM; S DHF condition; Fig. 4H) (35–37) . 267 Whole-cell patch clamp recordings showed STP after HFS in the S DHF group (Fig. 4I ), 268 indicating that TrkB activation after stress is insufficient to prevent STP. 269 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 10 270 Glucocorticoid receptors (GR) and TrkB share key intracellular signaling pathways ( 38), 271 and some forms of synaptic plasticity require a synergistic interaction between CORT and 272 BDNF ( 39). We hypothesized that post -stress exercise may induce adaptive GR/TrkB 273 activation in a time -dependent manner, contributing to STP buffering. To test this, brain 274 slices from stressed mice were incubated for 1 hour in aCSF containing both CORT and 275 DHF (SCORT + DHF condition; Fig. 3J). STP was fully prevented in this group (Fig. 4K). EPSC 276 amplitudes during the first minute post-HFS were significantly reduced in SCORT + DHF cells 277 compared to SaCSF, SCORT, and SDHF conditions (Fig. 4L). These findings indicate that the 278 co-activation of GR and TrkB is sufficient to fully block STP in vitro. 279 280 281 282 Fig. 4. Elevated CORT synergizes with a TrkB ligand to prevent stress -induced 283 metaplasticity. A. CRHPVN calcium response (Z-score) during one hour of immobilization 284 (IMMO; N = 9). IMMO elevated average CRHPVN calcium response compared to BL (two-285 tailed, one sample t-test: t(7)=4.478, P = 0.003). B. Quantification of plasma 286 concentrations of CORT (N = 13) before and after IMMO . Two-tailed paired t-test: 287 t(12)=15.92, P < 0.001. C. Schematic of whole-cell patch clamp recordings from CRHPVN 288 neurons after 1- hour IMMO (top section). Repeated-measures ANOVA: F (13, 104) = 289 6.762, P < 0.001; min 5 vs min 7: P < 0.001. D. Experimental design for CORT incubation 290 of brain slices. E. A mixed ANOVA revealed a main effect of Time ( F (5.485, 234.6) = 291 16.85, P < 0.001). F. Schematic of PVN tissue collection for protein quantification of brain-292 derived neurotrophic factor (BDNF). G. Individual concentration of BDNF in the PVN. Two 293 tail t-test: t(44)=2.901, P = 0.006. H. Experimental design for 7,8-Dihydroxyflavone (DHF) 294 incubation of brain slices. I. CRH PVN neurons from S DHF mice showed a significant 295 increase in EPSC amplitude after HFS (repeated measures ANOVA: F (13, 221) = 5.791, 296 P < 0.001; min 5 vs min 7: P < 0.001). J. Experimental design for CORT + DHF incubation 297 of brain slices. K. Following HFS, no changes were observed in the EPSC amplitudes of 298 CRHPVN neurons in SCORT + DHF mice. L. Summary of the average EPSC amplitude per cell 299 in all drug -treated slices. One -way ANOVA: F (3, 69) = 3.490, P = 0.020. Tukey’s and 300 Fisher's LSD multiple comparisons tests were used. Geisser-Greenhouse correction was 301 used. Data are mean ± SEM. 302 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 11 Post-stress activation of TrkB in CRHPVN cells prevents stress-induced STP and anxiety-303 like behaviors 304 305 Next, we asked whether the effects of Tr kB activation and CORT on CRH PVN neurons 306 were necessary and sufficient to reverse the effects of stress. To address this question, 307 we selectively disrupted TrkB signaling by overexpressing a truncated version of this 308 receptor (TrkB.T1) in CRH PVN neurons, then test ed whether post -stress exercise (S + 309 ETrkB.T1) was still capable of buffering STP (Fig. 5A-C). CRHPVN neurons from S + ETrkB.T1 310 mice showed robust STP (Fig. 4D), demonstrating that TrkB activation in CRHPVN neurons 311 during exercise is necessary to buffer the synaptic effects of stress. Importantly, 312 overexpression of TrkB.T1 in CRHPVN neurons did not induce STP or block stress-induced 313 STP ( Supplemental information F ig. S3 A-B ). DHF buffered STP in the presence of 314 CORT, but this does not indicate whether direct TrkB activation in CRH PVN neurons after 315 stress –but in the absence of exercise– is sufficient to buffer STP. We tested this idea by 316 expressing a light-activated TrkB (Opto-cytTrkB) (40) in CRHPVN neurons and delivering 317 light stimulation through an optic fiber for 15 minutes immediately after FS (SOpto-TrkB; Fig. 318 4A-C). Since circulating CORT is rapidly elevated after FS (Fig. 1), no exogenous CORT 319 was supplemented. We observed no STP in CRH PVN neurons of SOpto-TrkB mice (Fig. 5E; 320 Supplemental information Fig. S3 D). During the first minute post-HFS, EPSC amplitudes 321 were smaller in SOpto-TrkB compared to S + ETrkB.T1 mice (Fig. 5F). This demonstrates that, 322 in the presence of elevated CORT, TrkB activation in CRH PVN cells is necessary and 323 sufficient to buffer stress-induced STP. 324 325 Next, we tested whether these molecular and electrophysiological changes translate into 326 behavioral outcomes. Specifically, we asked if TrkB activation in CRH PVN neurons is 327 necessary and sufficient to reverse threat behavioral sensitization following stress. Thus, 328 S + E TrkB.T1 and SOpto-TrkB mice were tested in the DL box 24 hours after treatment (Fig. 329 5G). An additional group of mice expressing mCherry in CRHPVN neurons was implanted 330 with an optic fiber and light was delivered after FS (Fig. 4A, G), serving as a stress control 331 group (SmCherry). The analysis revealed sparse and delayed light-compartment exploration 332 bouts in S mCherry mice, increased but delayed visits in S + E TrkB.T1 mice, while prompt, 333 frequent exploration in S Opto-TrkB mice (Supplemental information F ig. S4 A–B; Five -334 minutes analysis: Supplemental information F ig. S4 D-I). The analysis of visit durations 335 revealed that S mCherry mice spent most of their time in the dark compartment (Fig. 5 H). 336 Similarly, S + E TrkB.T1 mice spent limited time in the light, although their stay was 337 descriptively longer compared to SmCherry mice. In contrast, post-FS optical stimulation of 338 TrkB fully reversed the effects of stress, with SOpto-TrkB mice showing the longest duration 339 in the light compartment compared to all other groups (Fig. 5H). The analysis of latencies 340 confirmed delayed first exploration bouts in SmCherry mice, with only 50% of them entering 341 the light compartment within the first testing minute (Fig. 5 I). In S + ETrkB.T1 mice, shorter 342 latencies were detected, with only 70% of the animals vising the light compartment during 343 the same period. In contrast, 100% of S Opto-TrkB animals explored the light compartment 344 within the first minute of testing, exhibiting the shortest exploration latencies among any 345 of the groups (Fig. 5I). No significant difference was detected between S mCherry and S + 346 ETrkB.T1 animals. The analysis of distance traveled revealed a similar pattern. SmCherry mice 347 exhibited the shortest displacement within the light compartment, followed by S + ETrkB.T1 348 animals, whose increased exploration did not reach statistical significance compared to 349 SmCherry mice (Fig. 5J, left panel). In contrast, S Opto-TrkB mice showed the most extensive 350 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 12 exploration across all groups. Exploration kinetics further revealed that early light -351 compartment exploration in S Opto-TrkB mice was accompanied by greater cumulative 352 distances traveled, a pattern that was less pronounced in the other groups. (Fig. 5J, right 353 panel). These findings demonstrate that TrkB signaling plays a fundamental role in 354 mediating the behavioral effects of post -stress exercise. While S + E TrkB.T1 animals 355 exhibited a partial recovery from the stress -induced threat behavioral sensitization, the 356 expression of CRH TrkB.T1 attenuated the otherwise robust and consistent behavioral 357 benefits of post -stress exercise. Moreover, the ability of optogenetic TrkB activation to 358 fully reverse stress -induced behavioral sensitization highlights a critical role for TrkB 359 signaling in CRHPVN neurons in modulating the brain’s response to stress. 360 361 362 363 Fig. 5. Post -stress activation of TrkB in CRH PVN cells reverses stress -induced STP and 364 threat behavioral sensitization. A. Constructs of the following Cre -dependent viruses: 365 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 13 AAV1-hSyn-DIO-mCherry, AAV1 -EF1a-DIO-trkB.DN-mCherry, and AAV1 -hSyn1-DIO-366 Opto-cytTrkB(E281A)-HA (top section). Schematic showing bilateral virus injection in the 367 PVN (bottom left). Schematic of ferrule implantation for optogenetic stimulation of the PVN 368 (bottom right). B. Confirmatory confocal image of virus expression in CRHPVN cells. Traces 369 of the ferrule implantation in the tissue are delineated. C. Experimental design for genetic 370 and optogenetic manipulation of TrkB after stress. D. After HFS (grey bar), CRH PVN 371 neurons from S + E TrkB.T1 showed a significant increase in EPSC amplitudes (repeated 372 measures ANOVA: F (13, 143) = 10.53, P < 0.001, min 5 vs min 7: P < 0.001). E. CRHPVN 373 neurons from S OptoTrkB showed no changes in EPSC amplitudes after HFS. F. Average 374 EPSC amplitude per cell one minute after HFS. Unpaired two-tailed t-test: t(22)=2.777, P 375 = 0.011. F. Experimental design for genetic and optogenetic manipulation of TrkB after 376 stress and its effects on the DL test. H. Time in the light compartment during the first 377 minute of testing. One-way ANOVA: F (2, 33) = 19.13, P < 0.001). I. Frequency analysis 378 of light compartment exploration latencies . The percentage of both S mCherry (P < 0.001) 379 and S + E TrkB.T1 mice ( P < 0.004) initiating their first exploration bout in the light 380 compartment was lower than in S OptoTrkB mice (One-way ANOVA: F (2, 18) = 15.03, P < 381 0.001). J. Analysis of the distance travelled in the light compartment . Both SmCherry (P < 382 0.001) and S + ETrkB.T1 mice (P < 0.001) travelled less distance than in SOptoTrkB mice (One-383 way ANOVA: F (2, 33) = 21.57, P < 0.001) . Cumulative distance travelled in the light 384 compartment (right panel). Tukey's multiple comparisons test and Geisser -Greenhouse 385 correction were used. Data are mean ± SEM. 386 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 14

387 388 The ability of exercise to activate the HPA axis presents a physiological paradox, 389 complicating efforts to delineate the neurobiological mechanisms by which it exerts 390 stress-relieving effects. Here, we demonstrate that a single bout of aerobic exercise , 391 immediately after acute stress, selectively reverses stress- induced threat sensitization 392 and synaptic metaplasticity in CRHPVN neurons, while sparing adaptive fear memory. This 393 reversal requires a localized, exercise- induced increase in BDNF within the PVN, which 394 when coinciding with elevated CORT, activates TrkB receptors in CRH PVN neurons to 395 promote adaptive plasticity. Notably, this effect occurs only when exercise is performed 396 after stress exposure, revealing a sensitive window during which stress-induced synaptic 397 changes remain liable and subject to intervention. 398 399 CORT and BDNF exhibit complex, temporally dynamic interactions in the PVN. Acute 400 stress increases circulating CORT within 15 minutes while reducing BDNF protein content 401 and simultaneously elevating BDNF mRNA expression in parvocellular PVN neurons (41–402 43). PVN BDNF content increases 1 hour after stress ( 41), and central BDNF 403 administration increases circulating CORT within 30 minutes (42). Conversely, prolonged 404 acute stress reduces BDNF mRNA in the PVN without affecting BDNF content ( 42). GR 405 activity suppresses TrkB signaling but BDNF/TrkB activation does not alter GR function 406 (43). In our study, both short and prolonged acute stress induced STP. Such plasticity 407 was unaffected by activating GR or TrkB alone, but simultaneous activation of both 408 pathways fully reversed STP, highlighting the necessity of temporal coordination and 409 combinatorial signaling. Although stress can elevate both BDNF and CORT in the PVN, 410 their endogenous release is either temporally misaligned or subthreshold, rendering them 411 ineffective to prevent or reverse synaptic metaplasticity. This points to the necessity of 412 synchronizing these signals post -stress to effectively modulate plasticity and mitigate 413 stress-related circuit changes. 414 415 The capacity of acute exercise to increase BDNF in subcortical regions like the PVN is 416 not well described. Treadmill running elevates BDNF content in sensorimotor cortex but 417 not in the hippocampus ( 34). However, it increases hippocampal BDNF mRNA shortly 418 after exercise and for several minutes ( 45, 46), with BDNF content increasing 6 hours 419 later (47 ). Running also increases BDNF in the prefrontal cortex ( 48). Importantly, 420 treadmill exercise sensitizes TrkB receptors in the hippocampus, promoting stronger and 421 prolonged activation (50). In contrast, acute treadmill running reliably engages the HPA 422 axis, inducing a dose -dependent activation of CRH PVN cells ( 14, 50) and increasing 423 circulating CORT (34, 47, 51, 52). Here, we demonstrated that acute post-stress treadmill 424 exercise increases BDNF concentrations in the PVN and provided in vivo evidence of the 425 temporal dynamics of CRHPVN activity following exercise. We also demonstrated that the 426 temporally coordinated rise in both BDNF and CORT levels within the PVN following 427 exercise is both necessary and sufficient to reverse stress -induced STP in CRH PVN 428 neurons. Our findings suggest a hormone- dependent gating mechanism whereby 429 glucocorticoids prime CRH PVN neurons to respond adaptively to environmental 430 interventions. The intracellular mechanisms by which CORT enables TrkB activation to 431 buffers stress -induced plasticity within the post -stress critical window remain to be 432 elucidated. STP expression in CRH PVN neurons requires a downregulation of NMDA 433 receptors (16). It is possible that enhanced BDNF/TrkB signaling post -exercise may 434 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 15 promote increased intracellular Ca 2+ release through canonical ( 29, 53, 54) and non -435 canonical pathways (55), preventing STP. Further investigation is required to determine 436 the precise intracellular mechanism involved. 437 438 This work shows that, although different stimuli including exercise activate CRH PVN 439 neurons, only stressors prime them for STP. This indicates that BDNF may act as the 440 molecular mediator shifting CRHPVN activation from driving synaptic metaplasticity to 441 enabling stress buffering. Notably, CORT negative feedback was insufficient to prevent 442 STP. While CORT rapidly suppresses CRHPVN neuronal activity (59), it did not block the 443 synaptic metaplasticity, underscoring a dissociation between hormonal feedback and 444 local synaptic plasticity mechanisms. Given the local effects of BDNF/TrkB signaling in 445 CRHPVN neurons on threat behavioral sensitization, our findings support that stress -446 induced synaptic metaplasticity serves as a cellular correlate of stress memory. The 447 persistence of this metaplasticity days after stress (17) and its association with behavioral 448 sensitization suggests that CRH PVN neurons maintain a latent “readiness” that encodes 449 prior aversive experience. This synaptic trace may act as a predictive internal model of 450 future threat. Our observations also underscore the critical role of CRH PVN neurons in 451 integrating aversive and rewarding signals to guide behavior (17, 56 – 58 ). Interestingly, 452 this phenomenon diverges from canonical contextual fear memory. Contextual fear 453 serves an adaptive role by minimizing predator detection and promoting energy 454 conservation (25, 26). Exercise selectively buffers CRH PVN plasticity and its associated 455 behaviors while leaving the adaptive processes of contextual fear learning that promote 456 threat avoidance and survival intact. 457 458 Clinical evidence suggests that a single bout of exercise rapidly enhances mood ( 9, 60), 459 reduces anxiety and depressive symptoms ( 62, 63), and attenuates HPA axis response 460 to acute stress (4, 63). However, the mechanisms by which acute exercise modulates 461 stress responsiveness remain understudied. Building upon these clinical insights, our 462 preclinical work uncovered a possible “therapeutic” window during which lasting stress 463 effects can be reversed. Given the role of CRHPVN neurons in anxiety and stress (64–68), 464 our findings provide a mechanistic basis for this intervention and highlight a novel, non-465 invasive strategy for remodeling stress-sensitive circuits. Importantly, the existence of this 466 therapeutic window opens new avenues for pharmacological and non- pharmacological 467 interventions. Although efforts to develop TrkB agonists (e.g., positive allosteric 468 modulators) have faced challenges and remain in early stages, several existing 469 compounds such as ketamine and certain psychedelics, engage BDNF/TrkB signaling 470 pathways ( 69–71). This raises the exciting possibility that post -stress interventions – 471 whether behavioral or pharmacological – could be optimized to target a temporally defined 472 window of plasticity within stress-responsive brain regions. Together, these insights pave 473 the way for the development of time-sensitive, mechanism-based strategies to prevent or 474 reverse the maladaptive effects of acute stress. 475 476 477 478 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 16

479 Subjects and general procedures: Mice strains with a C57BL/6 background were used 480 (Jackson Laboratory). Males and female mice were used in all experiments unless 481 specified. For whole-cell patch-clamp recordings, a mouse line expressing Cre in a CRH-482 dependent manner (CRH -IRES-Cre) was crossed with a Cre- dependent tdTomato 483 reporter line (Ai14) to generate mice with tdTomato- labeled CRH⁺ cells (CRH -IRES-Cre 484 × Ai14) ( 76). For in vivo fiber photometry experiments, the Ai148 mouse line – which 485 expresses the calcium sensor GCaMP6s in a Cre-dependent manner – was crossed with 486 the CRH-IRES-Cre line to generate mice expressing GCaMP6s specifically in CRH⁺ cells 487 (Ai148 × CRH- IRES-Cre) (66) . Cre-dependent adeno- associated viruses (AAVs) were 488 injected bilaterally into CRH-IRES-Cre or Ai148 × CRH -IRES-Cre mice, to obtain viral 489 expression in CRH+ cells only. For behavioral studies, all mouse lines were used. For all 490 in vivo experiments, animals were habituated to experimental handling using the following 491 procedure: once daily for three consecutive days, mice were scooped and held in the 492 experimenter’s hands for 1 minute, followed by 1- minute rest periods, for a total of 10 493 minutes per session under dimmed light. Mice used in fiber photometry experiments were 494 additionally habituated to fiber optic attachment and detachment. This was accomplished 495 by gradually increasing the level of restraint during the handling sessions. Over the final 496 two days of habituation, mice were fitted with fiber optic cables for 20 minutes per day 497 while freely exploring their home cages. 498 Mice were 6–8 weeks old at the time of surgical procedures and viral injections, and 8–10 499 weeks old at the time the experiments. All male mice were single housed for at least one 500 week prior to the start of experimental procedures. To avoid sex-specific effects of acute 501 social isolation or social buffering on PVN electrophysiology (Sterley et al., 2018), female 502 mice were group-housed until the start of experimental procedures. Female mice used in 503 in vivo experiments were single- housed one week prior to the beginning of testing. All 504 animals were housed in transparent polycarbonate cages (35.5 × 17.5 × 13 cm) under a 505 12:12-hour light–dark cycle (lights on at 7:00 a.m.), with ad libitum access to food and 506 water. 507 Stereotaxic procedures: Mice were anesthetized with isoflurane during all surgical 508 procedures. For in vivo fiber photometry recordings, a 400 µm -diameter mono fiber optic 509 ferrule (Doric Lenses; MFC_400/430-0.48_5.5mm_MF2.5_FLT) was implanted targeting 510 the PVN using the following stereotaxic coordinates: anterior –posterior (AP): −0.7 mm; 511 medial–lateral (ML): ±0.5 mm; dorsal–ventral (DV): −4.8 mm from the dura. To minimize 512 tissue damage, a stainless-steel tracking ferrule (diameter: 0.33 mm) was first lowered at 513 a rate of 1.5 mm/min to a depth of DV: −4.5 mm, then immediately withdrawn. The optical 514 ferrule was subsequently lowered to the final target depth at 0.2 mm/min. The ferrule was 515 secured to the skull using Metabond and dental cement. For viral injections, a pulled glass 516 micropipette containing the viral construct was lowered to the PVN (AP: −0.7 mm, ML: 517 ±0.5 mm, DV: −4.7 mm from the dura), and a total volume of 210 nL was delivered via 518 pressure injection (Nanoject II, Drummond Scientific). After injection, the micropipette was 519 left in place for 5 minutes to allow for viral diffusion before being slowly retracted. 520 Postoperatively, mice received analgesia (Meloxicam, 5 mg/kg, subcutaneous) 24 hours 521 after surgery and were monitored daily for 7 days. All mice were given a 2-week recovery 522 period before experimental procedures started. 523 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 17 Viruses: The following Cre- dependent AAV plasmids (pAAV) were obtained from 524 Addgene and packaged into AAV1 vectors: pAAV -hSyn1-DIO-Opto-cytTrkB(E281A)-HA 525 (5.85 × 10¹³ GC/mL; Addgene #180590) and pAAV -EF1a-DIO-trkB.DN-mCherry (1.95 × 526 10¹³ GC/mL; Addgene #121502). The AAV1- hSyn1-DIO-Opto-cytTrkB(E281A)-HA was 527 used to enable optogenetic manipulation of TrkB signaling, while AAV1- EF1a-DIO-528 trkB.DN-mCherry was used to reduce endogenous TrkB activity via expression of a 529 dominant-negative mutant. Both viral vectors were stereotaxically injected into the PVN 530 of CRH- IRES-Cre or Ai148 × CRH -IRES-Cre mice for cell- type-specific expression in 531 CRH⁺ neurons. 532 Histology: To verify viral expression and ferrule placement, mice were anesthetized with 533 isoflurane and their brains rapidly extracted and submerged in ice- cold aCSF for 534 approximately 5 minutes. Coronal brain slices (200 µm) containing the PVN were 535 prepared using a vibratome (VT1200 S, Leica) in ice-cold aCSF. Slices were fixed in 4% 536 paraformaldehyde (PFA) in phosphate buffer (PB) at 4 °C for 5 minutes, followed by 537 immersion in 30% sucrose in phosphate-buffered saline (PBS) for an additional 5 minutes. 538 For animals injected with HA -tagged viral constructs, immunohistochemistry was 539 performed. Slices were incubated for 24 h at 4 °C with a primary antibody against the HA 540 epitope (1:1000 dilution; rabbit monoclonal antibody, C29F4; Cell Signaling Technology). 541 After three washes in PBS containing 0.3% Triton X -100, slices were incubated with a 542 secondary Alexa Fluor 488- conjugated anti -rabbit antibody (1:1000 dilution; Cell 543 Signaling Technology) for 2 h at room temperature. Following final washes in PBS, slices 544 were mounted and coverslipped using Vectashield mounting medium. Images were 545 acquired using a confocal microscope (Olympus BX50 Fluoview) and processed using 546 ImageJ (NIH). 547 Slice preparation and ex vivo electrophysiology: All preparations and recordings were 548 done as previously reported (15; 27). Mice were anesthetized (Isoflurane) and 549 decapitated 5-10 minutes after experimental manipulations end. The brain was removed 550 and submerged into ice- cold slicing solution containing (in nM): 87 NaCl, 2.5 KCl, 0.5 551 CaCl2, 7 MgCl2, 25 NaHCO3, 25 d-glucose, 1.25 NaH2PO4 and 75 sucrose, saturated with 552 95% O2/5% CO2. Coronal slides (250 µm) obtained using a vibratome (VT 1200 S, Leica) 553 were kept for 15 minutes in a 30°C N -methyl-d-glucamine (NMDG) -recovery solution 554 containing (in nM): 2.5 KCI, 25 NaHCO3, 0.5 CaCl2, 10 MgCl2, 1.2 NaH2PO4, 25 glucose, 555 110 NMDG and 110 HCI, saturated with 95% O 2/5% CO2. Then, slides were incubated 556 for 1 hour in 30 °C artificial cerebrospinal fluid (aCSF) containing (in mM): 126 NaCl, 2.5 557 KCl, 26 NaHCO 3, 2.5 CaCl 2, 1.5 MgCl 2, 1.25 NaH 2PO4 and 10 glucose, saturated with 558 95% O2/5% CO2. 559 Electrophysiological recording protocol: All recordings were obtained in aCSF containing 560 picrotoxin (100 μM) at 30-32 °C, perfused at 1 mL/min. Neurons were visualized using an 561 upright microscope fitted with differential interference contrast and epifluorescence optics 562 and a camera. Borosilicate pipettes (2.5 - 4.5 mΩ) were filled with internal solution 563 containing (in mM): 108 K-gluconate, 2 MgCl2, 8 sodium-gluconate, 8 KCl, 1 K-EGTA, 4 564 K-ATP, 0.3 Na-GTP and 10 HEPES buffer. Current-clamp recordings were acquired at a 565 −70 mV membrane potential. To assess synaptic currents, cells were voltage-clamped at 566 −70 mV. A monopolar aCSF-filled electrode placed in the vicinity of the cell (~20 μM) were 567 used to evoke excitatory postsynaptic currents (EPSCs) 50 ms apart at 0.2 Hz intervals. 568 After a 5 -minute baseline, high- frequency stimulation (four 1 -s stimulations at 100 -Hz 569 applied every 10 s) was delivered, followed by a 10 min recording. Access resistance (<20 570 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 18 MΩ) was assessed every 3 minutes, and recordings were accepted for analysis if changes 571 were <15%. 572 Pharmacological manipulations: Stock solutions of CORT (5 mM; Tocris, UK) and 7,9 -573 Dihydroxiyflavone (DHF) (100 mM; Tocris, UK) dissolved into DMSO were stored at -20°C 574 until used. Stock solutions were diluted in aCSF for a final concentration of 100 nM of 575 CORT and 20 µM of DHF. After incubating for 1 hour, slices were transferred from the 576 drug-containing bath into fresh aCSF and patched within 4 hours. 577 Corticosterone immunoassays: Blood (~15 µL) was collected from the tail vein into ice-578 cold Microvette® capillary tubes (Sarstedt, Germany) and centrifuged at 8100 RPM for 25 579 minutes at –4 °C using an Eppendorf 5430 R centrifuge. Serum samples were processed 580 and analyzed using the DetectX® Corticosterone Immunoassay Kit (Arbor Assays, USA). 581 For processing, 5 µL of serum were incubated with 5 µL of dissociation reagent (Arbor 582 Assays) for approximately 8 minutes, followed by dilution with 490 µL of 1X assay buffer 583 (Arbor Assays), resulting in a final 1:100 dilution. Processed samples were stored at –584 20 °C until analysis. All samples were analyzed in triplicate on the same day. When 585 possible, repeated samples from the same animal were analyzed on the same assay 586 plate. Plates with standard curves exhibiting an R² < 0.99 were reanalyzed to ensure 587 accuracy. Corticosterone concentrations were reported in ng/mL. 588 Brain-derived neurotrophic factor immunoassays: Mice (CRH -IRES-Cre × Ai14 589 [tdTomato]) were anesthetized with isoflurane, and their brains were rapidly extracted and 590 submerged in ice-cold aCSF for ~5 minutes. Coronal brain slices (100 µm) containing the 591 PVN were prepared using a vibratome (VT1200 S, Leica) in ice- cold aCSF. Slices were 592 inspected under a fluorescence microscope to identify CRH⁺ neurons expressing 593 tdTomato. PVN regions containing fluorescent signal were microdissected using a scalpel 594 under ice-cold conditions, collected into microcentrifuge tubes, flash- frozen on dry ice, 595 and stored at –80 °C until processing. 596 For BDNF quantification, samples were homogenized in 500 µL of a custom-made protein 597 extraction buffer (prepared according to kit specifications) using a TissueLyser LT 598 (Qiagen). Homogenates were centrifuged, and the supernatant was aliquoted and stored 599 at –80 °C. BDNF levels were measured using a Mouse BDNF ELISA Kit (EZ0309, Bolster, 600 USA) and quantified using a SpectraMax 190 plate reader (Molecular Devices). All 601 samples were run in triplicate on the same day. Total protein content was assessed using 602 the Pierce BCA Protein Assay Kit (ThermoFisher), and BDNF concentrations were 603 normalized to total protein. Results are expressed as pg of BDNF per mg of protein. 604 Optogenetics: Continuous optical stimulation (1 s every 5 s) with blue light (465 nm) 605 reliably activates intracellular TrkB signaling in cells expressing Opto-cytTrkB(E281A)-HA 606 (50). For stimulation, a 400 µm core -diameter fiber optic cable (Doric Lenses; 607 MFP_400/430/1100-0.48_2m_FC-MF2.5) was connected to a 465 nm laser, which was 608 controlled by a pulse generator to deliver 1-second light pulses (20 mW) every 5 seconds 609 over a 15-minute period. 610 Fiber photometry recordings: Calcium transients in freely moving mice were recorded 611 using a Doric fiber photometry system. The setup consisted of two excitation LEDs 612 (470 nm and 405 nm), controlled by an LED driver and a Doric Studio-compatible console 613 (Doric Lenses). The 470 nm LED (Ca²⁺-dependent signal) was modulated at 208.616 Hz, 614 and the 405 nm LED (isosbestic control) at 572.205 Hz. Signals were demodulated in real 615 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 19 time using lock-in amplification. Excitation light was delivered via a Doric Mini Cube filter 616 set (FMC5_E1(405)_E2(460–490)_F1(500–550)_S), coupled to a 400 µm core- diameter 617 fiber optic patch cable (Doric Lenses; MFP_400/430/1100 -0.48_2m_FC-MF2.5) 618 connected to the implanted ferrule. LED power at the fiber tip was adjusted to ~30 μW. 619 Emitted fluorescence was collected through the same fiber and detected by a 620 photoreceiver (Newport Model 2151). 621 Fiber photometry data analysis: Fluorescent signal data were acquired at a sampling rate 622 of 100 Hz using the Doric system and exported to MATLAB (MathWorks) for offline 623 analysis with custom -written scripts (https://github.com/leomol/FPA). First, the 470 nm 624 and 405 nm signals were each individually fitted with a second-order polynomial to correct 625 for photobleaching artifacts; the fitted curves were then subtracted from the raw data. 626 Next, a least-squares linear regression was applied to the 405 nm signal to align it with 627 the 470 nm channel. The change in fluorescence (ΔF) was calculated by subtracting the 628 405 nm Ca²⁺-independent baseline from the 470 nm Ca²⁺-dependent signal at each time 629 point. To minimize the influence of handling- and novelty-induced activity on bleaching 630 correction, a 20-minute baseline recording was obtained in the animals' home cage (HC). 631 The initial 10 minutes of this baseline were discarded; when the signal reached an 632 asymptotic phase, a 5-minute epoch was selected for curve fitting. Following treatments, 633 subjects were returned to their HC, and Ca²⁺ signals were recorded for an additional 20 634 minutes. A 5 -minute epoch was then selected within the last 10 minutes of this post -635 treatment recording. Both 5- minute baseline and post -treatment epochs were used to 636 calculate a modified Z-Score (https://github.com/leomol/FPA) . Epochs were defined using 637 synchronized video recordings. Compound traces from the selected epochs were 638 presented as continuous time series. 639 Stress induction: Electric foot shocks (FS; 0.5 mA) were delivered using either a manually 640 operated scrambler (42.6 × 21 × 29 cm; SMSCK, Kinder Scientific) or an automatic shuttle 641 box (20 × 19.5 × 25.5 cm; Imetronic, France). Acute stress was induced by a single 642 session consisting of ten 3-second foot shocks applied every 27 seconds over a 5-minute 643 period. Stress by immobilization involved restraining animals by tying their limbs and torso 644 onto a flat, highly illuminated (~630 lux) table using Transpore surgical tape (3M, 645 Germany). Animals remained immobilized for 60 minutes. 646 Treadmill training and exercise protocol: The treadmill (TM; LE7808; Panlab, Spain) 647 features a 30 × 10 cm running area with a 10 × 10 cm metal grid at one end. The running 648 area is enclosed by transparent Plexiglas walls and a custom -made lid with an opening 649 to accommodate a photometry optic fiber during experiments. Mice were trained to run on 650 the treadmill more than 24 hours before experiments using the following procedure: 651 subjects were placed on the moving treadmill set at 15 cm/s with a 5% inclination, gently 652 guided forward for 1 minute, and then allowed to run freely for 10 minutes. If a mouse 653 stopped running and was fully dragged onto the metal grid at the end of the treadmill, a 654 mild foot shock (0.1 mA) was delivered. Two training sessions were performed in a single 655 day, with at least 30 minutes of rest between sessions. During the exercise session, mice 656 were placed on the treadmill and allowed to run for 1 hour. Animals were continuously 657 monitored and removed from the session if necessary. Subjects receiving more than 10 658 seconds of cumulative foot shock during the running session were excluded from the 659 study. 660 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 20 Behavioral tests: Assessments were conducted between 8:00 and 14:00 hours. Animals’ 661 schedule assignments were balanced across experimental conditions within each daily 662 session and throughout the testing period. Behavioral assessments involved multiple 663 cohorts evaluated at different time points. 664 Dark/Light box : The Dark/Light (D/L) box consists of two rectangular Plexiglas 665 compartments (43 × 21.5 × 30.5 cm; Med Associates) connected by a door. The dark 666 compartment is fully enclosed, painted black, and unilluminated, while the light 667 compartment is open, white, and illuminated (~240 lux). Animals were habituated to the 668 testing environment by being placed daily in the dark compartment with the door closed 669 for 5 minutes on two consecutive days. Experimental manipulations were conducted more 670 than 2 hours after the final habituation session. On testing day, animals were placed in 671 the dark compartment, and the door was immediately opened to allow access to the light 672 compartment for 5 minutes. All sessions were video recorded (DMK 22AUC03, The 673 Imaging Source), and the following behaviors were analyzed using AnyMaze tracking 674 software (v12, Stoelting): latency to enter the light compartment (seconds), frequency and 675 duration of visits (seconds), and distance traveled (meters) within the light compartment. 676 Contextual Fear Conditioning: Mice were placed on an automatic shuttle box (20 × 19.5 677 × 25.5 cm; Imetronic, France), where they received ten 3- second foot shocks (0.5 mA) 678 delivered every 27 seconds over a 5- minute period. For contextual fear memory recall, 679 mice were reintroduced to the foot shock chamber for 5 minutes while the duration of 680 freezing, walking, and rearing behaviors were automatically scored (Imetronic, France). 681 Statistical analysis: Data was analyzed using GraphPad Prism (v10.2.0) and IBM SPSS 682 Statistics (v29.0.1.1). Group differences were assessed using one-tailed t-tests, one-way, 683 or two -way ANOVAs as appropriate. Bonferroni correction was applied for multiple 684 pairwise comparisons when necessary. Repeated- measures ANOVAs were used to 685 analyze group differences over time. If Mauchly’s test of sphericity indicated violation (P 686 < 0.05), Greenhouse-Geisser correction was applied. 687 AI use: Language refinement in the Introduction and Discussion sections were assisted 688 by the large language model ChatGPT (OpenAI, GPT- 4, May–September 2025). Some 689 schematics (i.e., mouse icons) were created by the authors using Adobe Illustrator with 690 AI-assisted design tools. All scientific content was conceived, written, and verified by the 691 authors. 692 Code availability : Scripts used to analyze fiber photometry are deposited here: 693 https://github.com/leomol/FPA; https://doi.org/10.5281/zenodo.5708470. 694 695 696 697 698 699 700 701 702 703 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 21

704 705 1. Schmitt, A., et al. Effects of aerobic exercise on metabolic syndrome, cardiorespiratory 706 fitness, and symptoms in schizophrenia include decreased mortality. Front. psychiatry 707 9, 690 (2018). 708 2. Dauwan, M., Begemann, M. J., Slot, M. I., Lee, E. H., Scheltens, P., & Sommer, I. E. 709 Physical exercise improves quality of life, depressive symptoms, and cognition across 710 chronic brain disorders: a transdiagnostic systematic review and meta- analysis of 711 randomized controlled trials. J. Neurology 268(4), 1222-1246 (2021). 712 3. Baskerville, R., McGrath, T., & Castell, L. (2023). The effects of physical activity on 713 glutamate neurotransmission in neuropsychiatric disorders. Front. Sports Act. Living 714 5, 1147384 (2023). 715 4. Childs, E., & De Wit, H. Regular exercise is associated with emotional resilience to 716 acute stress in healthy adults. Front. Physiology 5, 161(2014). 717 5. Borrega-Mouquinho, Y., Sánchez -Gómez, J., Fuentes -García, J. P., Collado -Mateo, 718 D., & Villafaina, S. Effects of high -intensity interval training and moderate- intensity 719 training on stress, depression, anxiety, and resilience in healthy adults during 720 coronavirus disease 2019 confinement: a randomized controlled trial. Front. Psych 12, 721 643069 (2021). 722 6. Popov, S., Sokić, J., & Stupar, D. Activity matters: Physical exercise and stress coping 723 during the 2020 COVID-19 state of emergency. Psihologija 54(3), 307-322 (2021). 724 7. Cairney, J., Kwan, M. Y., Veldhuizen, S., & Faulkner, G. E. Who uses exercise as a 725 coping strategy for stress? Results from a national survey of Canadians. J. Phys. Act. 726 Health 11(5), 908-916 (2014). 727 8. Faulkner, G., et al. Physical activity as a coping strategy for mental health due to the 728 COVID-19 virus: a potential disconnect among Canadian adults?. Front. Comm 5, 729 571833 (2020). 730 9. Ensari, I., Greenlee, T. A., Motl, R. W., & Petruzzello, S. J. (2015). Meta‐analysis of 731 acute exercise effects on state anxiety: An update of randomized controlled trials over 732 the past 25 years. Depression and Anxiety 32(8), 624-634 (2015). 733 10. Brownlee, K. K., Moore, A. W., & Hackney, A. C. (2005). Relationship between 734 circulating cortisol and testosterone: influence of physical exercise. J. Sports Sci. Med 735 4, 76 (2005). 736 11. Hill, E. E., Zack, E., Battaglini, C., Viru, M., Viru, A., & Hackney, A. C. Exercise and 737 circulating cortisol levels: the intensity threshold effect. J. Endocrinol. Investig 31, 587–738 591 (2008). 739 12. Chen, C., Nakagawa, S., An, Y., Ito, K., Kitaichi, Y., & Kusumi, I. The exercise-740 glucocorticoid paradox: How exercise is beneficial to cognition, mood, and the brain 741 while increasing glucocorticoid levels. Front. Neuroendocrinol 44, 83–102 (2017). 742 13. Caplin, A., Chen, F. S., Beauchamp, M. R., & Puterman, E. The effects of exercise 743 intensity on the cortisol response to a subsequent acute psychosocial stressor. 744 Psychoneuroendocrinology 131, 105336 (2021). 745 14. Morikawa, R., Kubota, N., Amemiya, S., Nishijima, T., & Kita, I. Interaction between 746 intensity and duration of acute exercise on neuronal activity associated with 747 depression-related behavior in rats. J. Physiol. Sci 71, 1–11 (2021). 748 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 22 15. Daviu, N. et al. Paraventricular nucleus CRH neurons encode stress controllability and 749 regulate defensive behavior selection. Nat. Neurosci 23, 398–410 (2020). 750 16. Kuzmiski, J. B., Marty, V., Baimoukhametova, D. V., & Bains, J. S. Stress -induced 751 priming of glutamate synapses unmasks associative short -term plasticity. Nat. 752 Neurosci 13, 1257–1264 (2010). 753 17. Sterley, T. L., et al. Social transmission and buffering of synaptic changes after stress. 754 Nat. Neurosci 21, 393–403 (2018). 755 18. Rasiah, N. P., Loewen, S. P., & Bains, J. S. Windows into stress: a glimpse at 756 emerging roles for CRHPVN neurons. Physiol Rev 103, 1667–1691 (2023). 757 19. Goto, A. et al. Stepwise synaptic plasticity events drive the early phase of memory 758 consolidation. Science 374, 857–863 (2021). 759 20. Dong, Y. et al. Stress relief as a natural resilience mechanism against depression-like 760 behaviors. Neuron 111, 3789–3801 (2023). 761 21. Belda, X., Fuentes, S., Daviu, N., Nadal, R., & Armario, A. Stress -induced 762 sensitization: the hypothalamic–pituitary–adrenal axis and beyond. Stress 18, 269–279 763 (2015). 764 22. Belda, X., Fuentes, S., Nadal, R., & Armario, A. A single exposure to immobilization 765 causes long-lasting pituitary -adrenal and behavioral sensitization to mild stressors. 766 Horm. Behav 54, 654–661 (2008). 767 23. Belda, X., Rotllant, D., Fuentes, S., Delgado, R., Nadal, R., & Armario, A. Exposure to 768 severe stressors causes long‐lasting dysregulation of resting and stress‐induced 769 activation of the hypothalamic‐pituitary‐adrenal axis. Ann. N. Y. Acad. Sci 1148, 165–770 173 (2008). 771 24. Gencturk, S., & Unal, G. Rodent tests of depression and anxiety: Construct validity 772 and translational relevance. Cogn. Affect. Behav. Neurosci 24, 191–224 (2024). 773 25. Blanchard, D. C., Griebel, G., Pobbe, R., & Blanchard, R. J. Risk assessment as an 774 evolved threat detection and analysis process. Neurosci. Biobehav. Rev 35, 991–998 775 (2011). 776 26. Tseng, Y. T., Schaefke, B., Wei, P., & Wang, L. Defensive responses: behaviour, the 777 brain and the body. Nat. Rev. Neurosci 24, 655–671 (2023). 778 27. Notaras, M., & van den Buuse, M. Neurobiology of BDNF in fear memory, sensitivity 779 to stress, and stress-related disorders. Mol. Psychiatry 25, 2251–2274 (2020). 780 28. Brigadski, T., & Leßmann, V. The physiology of regulated BDNF release. Cell Tissue 781 Res 382, 15–45 (2020). 782 29. Sequeira-Cordero, A., Brenes, J. C., & Vindas-Smith, R. The role of BDNF on food 783 intake and overweight: Linking neuroplasticity and obesity. Prog. 784 Neuropsychopharmacol. Biol. Psychiatry 111457 (2025). 785 30. Leger, C. et al. Impact of Exercise Intensity on Cerebral BDNF Levels: Role of 786 FNDC5/Irisin. Int. J. Mol. Sci 25, 1213 (2024). 787 31. Shahroodi, A., Mohammadi, F., Vafaei, A. A., Miladi -Gorji, H., Bandegi, A. R., & 788 Rashidy-Pour, A. Impact of different intensities of forced exercise on deficits of spatial 789 and aversive memory, anxiety-like behavior, and hippocampal BDNF during morphine 790 abstinence period in male rats. Metab. Brain Dis 35, 135–147 (2020). 791 32. Yan, L. et al. Physical exercise mediates a cortical FMRP–mTOR pathway to improve 792 resilience against chronic stress in adolescent mice. Transl. Psychiatry 13, 16 (2023). 793 33. Rasmussen, P. et al. Evidence for a release of brain‐derived neurotrophic factor from 794 the brain during exercise. Exp. Physiol 94, 1062–1069 (2009). 795 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 23 34. Thacker, J. S., Xu, Y., Tang, C., Tupling, A. R., Staines, W. R., & Mielke, J. G. A single 796 session of aerobic exercise mediates plasticity-related phosphorylation in both the rat 797 motor cortex and hippocampus. Neuroscience 412, 160–174 (2019). 798 35. Devi, L., & Ohno, M. 7, 8-dihydroxyflavone, a small-molecule TrkB agonist, reverses 799 memory deficits and BACE1 elevation in a mouse model of Alzheimer's disease. 800 Neuropsychopharmacology 37, 434–444 (2012). 801 36. Chen, Y. S. et al. Early 7, 8-dihydroxyflavone administration ameliorates synaptic and 802 behavioral deficits in the young FXS animal model by acting on BDNF-TrkB pathway. 803 Mol. Neurobiol 60, 2539–2552 (2023). 804 37. Sun, S. R., et al. Pharmacologically activating BDNF/TrkB signaling exerted rapid-805 acting antidepressant -like effects through improving synaptic plasticity and 806 neuroinflammation. Metab. Brain Dis 40, 158 (2025). 807 38. Chen, D. Y., & Bambah-Mukku, D., Pollonini, G., Alberini, C. M. Glucocorticoid 808 receptors recruit the CaMKIIα -BDNF-CREB pathways to mediate memory 809 consolidation. Nat. Neurosci 15, 1707–1714 (2012). 810 39. Thacker, J. S., & Mielke, J. G. The combined effects of corticosterone and brain-811 derived neurotrophic factor on plasticity -related receptor phosphorylation and 812 expression at the synaptic surface in male Sprague- Dawley rats. Horm. Behav 145, 813 105233 (2022). 814 40. Hong, J., & Do Heo, W. Optogenetic modulation of TrkB signaling in the mouse brain. 815 J. Mol. Biol 432, 815–827 (2020). 816 41. Rage, F., Givalois, L., Marmigere , F., Tapia- Arancibia, & L., Arancibia, S. 817 Immobilization stress rapidly modulates BDNF mRNA expression in the hypothalamus 818 of adult male rats. Neuroscience 112, 309–318 (2002). 819 42. Givalois, L., Naert, G., Rage, F., Ixart, G., Arancibia, S., & Tapia-Arancibia, L. A single 820 brain-derived neurotrophic factor injection modifies hypothalamo– pituitary–821 adrenocortical axis activity in adult male rats. Mol. Cell. Neurosci 27, 280–295 (2004). 822 43. Schaich, C. L., Wellman, T. L., Einwag, Z., Dutko, R. A., & Erdos, B. Inhibition of BDNF 823 signaling in the paraventricular nucleus of the hypothalamus lowers acute stress -824 induced pressor responses. J. Neurophysiol 120, 633–643 (2018). 825 44. Jeanneteau, F. D. et al. BDNF and glucocorticoids regulate corticotrophin- releasing 826 hormone (CRH) homeostasis in the hypothalamus. Proc. Natl. Acad. Sci. U.S.A 109, 827 1305–1310 (2012). 828 45. Venezia, A. C., Quinlan, E., & Roth, S. M. A single bout of exercise increases 829 hippocampal Bdnf: influence of chronic exercise and noradrenaline. Genes Brain 830 Behav 16, 800–811 (2017). 831 46. Venezia, A. C., Hyer, M. M., Glasper, E. R., Roth, S. M., & Quinlan, E. M. Acute forced 832 exercise increases Bdnf IV mRNA and reduces exploratory behavior in C57BL/6J 833 mice. Genes Brain Behav 19, e12617 (2020). 834 47. Soya, H., et al. BDNF induction with mild exercise in the rat hippocampus. Biochem. 835 Biophys. Res. Commun 358, 961–967 (2007). 836 48. Baranowski, B. J., Hayward, G. C., Marko, D. M., & MacPherson, R. E. Examination 837 of BDNF treatment on BACE1 activity and acute exercise on brain BDNF signaling. 838 Front. Cell. Neurosci 15, 665867 (2021). 839 49. Guo, W., Ji, Y., Wang, S., Sun, Y., & Lu, B. Neuronal activity alters BDNF– TrkB 840 signaling kinetics and downstream functions. J. Cell Sci 127, 2249–2260 (2014). 841 50. Nishii, A., Amemiya, S., Kubota, N., Nishijima, T., & Kita, I. Adaptive changes in the 842 sensitivity of the dorsal raphe and hypothalamic paraventricular nuclei to acute 843 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 24 exercise, and hippocampal neurogenesis may contribute to the antidepressant effect 844 of regular treadmill running in rats. Front. Behav. Neurosci 11, 235 (2017). 845 51. Goekint, M., Bos, I., Heyman, E., Meeusen, R., Michotte, Y., & Sarre, S. Acute running 846 stimulates hippocampal dopaminergic neurotransmission in rats, but has no influence 847 on brain-derived neurotrophic factor. J. Appl. Physiol 112(4), 535–541 (2012). 848 52. Chen, X. et al. HPA- axis and inflammatory reactivity to acute stress is related with 849 basal HPA-axis activity. Psychoneuroendocrinology 78, 168–176 (2017). 850 53. Kim, J., He, M. J., Widmann, A. K., & Lee, F. S. The role of neurotrophic factors in 851 novel, rapid psychiatric treatments. Neuropsychopharmacology 49, 227–245 (2024). 852 54. Wang, Y., Liang, J., Xu, B., Yang, J., Wu, Z., & Cheng, L. TrkB/BDNF signaling 853 pathway and its small molecular agonists in CNS injury. Life Sci 336, 122282 (2024). 854 55. Lao-Peregrin, C. et al. Synaptic plasticity via receptor tyrosine kinase/G -protein-855 coupled receptor crosstalk. Cell Rep 43, 113595 (2024). 856 56. Füzesi, T. et al. Hypothalamic CRH neurons represent physiological memory of 857 positive and negative experience. Nat. Commun 14, 8522 (2023). 858 57. Yuan, Y. et al. Reward inhibits paraventricular CRH neurons to relieve stress. Curr. 859 Biol 29, 1243–1251 (2019). 860 58. Xu, X. et al. Hypothalamic CRF neurons facilitate brain reward function. Curr. Biol 34, 861 389–402 (2024). 862 59. Jiang, Z. et al. Stress-induced glucocorticoid desensitizes adrenoreceptors to gate the 863 neuroendocrine response to somatic stress in male mice. Cell Rep 41, 111509 (2022). 864 60. Crush, E. A., Frith, E., & Loprinzi, P. D. Experimental effects of acute exercise duration 865 and exercise recovery on mood state. J. Affect. Disord 229, 282–287 (2018). 866 61. Connor, M., Hargreaves, E. A., Scanlon, O. K., & Harrison, O. K. The effect of acute 867 exercise on state anxiety: a systematic review. Sports 11, 145 (2023). 868 62. Weinstein, A. A., van Aert, R. C., Donovan, K., Muskens, L., & Kop, W. J. Affective 869 responses to acute exercise: A meta- analysis of the potential beneficial effects of a 870 single bout of exercise on general mood, anxiety, and depressive symptoms. 871 Psychosom. Med 1097 (2024). 872 63. Zschucke, E., Renneberg, B., Dimeo, F., Wüstenberg, T., & Ströhle, A. The stress -873 buffering effect of acute exercise: Evidence for HPA axis negative feedback. 874 Psychoneuroendocrinology 51, 414–425 (2015). 875 64. Kim, J. et al. Rapid, biphasic CRF neuronal responses encode positive and negative 876 valence. Nat. Neurosci 22, 576–585 (2019). 877 65. Füzesi, T., Daviu, N., Wamsteeker Cusulin, J. I., Bonin, R. P., & Bains, J. S. 878 Hypothalamic CRH neurons orchestrate complex behaviours after stress. Nat. 879 Commun 7, 11937 (2016) 880 66. McGill, B. E., Bundle, S. F., Yaylaoglu, M. B., Carson, J. P., Thaller, C., & Zoghbi, H. 881 Y. Enhanced anxiety and stress -induced corticosterone release are associated with 882 increased Crh expression in a mouse model of Rett syndrome. Proc. Natl. Acad. Sci. 883 U.S.A 103, 18267–18272 (2006). 884 67. Yu, L. et al . Propofol Alleviates Anxiety‐Like Behaviors Associated with Pain by 885 Inhibiting the Hyperactivity of PVNCRH Neurons via GABAA Receptor β3 Subunits. 886 Adv. Sci 11, 2309059 (2024). 887 68. Liu, Y. et al. Distinct hypothalamic paraventricular nucleus inputs to the cingulate 888 cortex and paraventricular thalamic nucleus modulate anxiety and arousal. Front. 889 Pharmacol 13, 814623 (2022). 890 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 25 69. Brunello, C. A., Cannarozzo, C., & Castrén, E. Rethinking the role of TRKB in the 891 action of antidepressants and psychedelics. Trends Neurosci 47, 865–874 (2024). 892 70. Moliner, R., et al. Psychedelics promote plasticity by directly binding to BDNF receptor 893 TrkB. Nat. Neurosci 26, 1032–1041 (2023). 894 71. Kim, J., He, M. J., Widmann, A. K., & Lee, F. S. The role of neurotrophic factors in 895 novel, rapid psychiatric treatments. Neuropsychopharmacology 49, 227–245 (2024). 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 26

939 940 We thank Ms. Alexis Passmore, Dr. Leonardo Molina, and Dr. Jianjun Sun for their 941 technical assistance. We are also grateful to Dr. Jonathan Thacker, Dr. Grant Gordon, 942 Sierra Stokes-Heck, Cheryl Breiteneder, Govind Peringod, and Patrick Grouve for their 943 valuable advice and support. We acknowledge the Cumming School of Medicine 944 Optogenetics Core Facility for their continued assistance. Language refinement in the 945

and discussion sections was assisted by the large language model ChatGPT 946 (OpenAI, GPT -4, May –September 2025). Some schematics (i.e., mouse icons) were 947 created by the authors using Adobe Illustrator with AI -assisted design tools. M.R-C. was 948 supported by graduate scholarships from the Cumming School of Medicine and the Killam 949 Trust. 950 951 952 Author contributions: M.R.C. conceptualized, designed, and conducted the experiments; 953 analyzed the data; prepared the figures; wrote the manuscript; and administered the 954 project. D.B. contributed to experimental design, performed experiments, and analyzed 955 data. S.G.C. assisted in conducting experiments. T.F. and N.D. contributed to 956 conceptualization, experimental design, and supervision of the project. T.F. also 957 contributed to figure preparation and to reviewing and editing the manuscript. M.N.H. 958 contributed to reviewing the manuscript and provided financial support for the project. 959 J.S.B. contributed to conceptualization, experimental design, supervision, and 960 administration of the project. J.S.B. also contributed to reviewing and editing the 961 manuscript and secured funding for the project. 962 963 Competing interests: We declare no competing interests. 964 965 Data and materials availability : All data are available in the manuscript or the 966 Supplemental information figures. 967 968 969 Funding: 970 971 Canadian Institutes of Health Research (CIHR) Foundation Grant: FDN-148440 (J.S.B.). 972 Canadian Institutes of Health Research (CIHR) Foundation Grant: FDN-426504 (M.N.H.). 973 974 975 976 977 978 979 980 981 982 983 984 985 986 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 27 Supplemental Information 987 988 989 990 991 992 993 994 995 996 997 Figure S1. Exercise does not induce STP. A. Experimental design for whole- cell patch 998 clamp recordings from CRH PVN neurons of mice exercising on a treadmill for 1 hour. B. 999 Summary of excitatory post-synaptic current (EPSC) amplitudes following HFS (grey bar) 1000 relative to baseline (BL; doted line). A repeated- measures ANOVA revealed an effect of 1001 Time (F (13, 143) = 2.175, P = 0.013). However, no significant increase in the EPSC 1002 amplitudes was observed after HFS (min 5 vs min 7: P = 1.00). Bonferroni's multiple 1003 comparisons adjustment was used. Data shown represent the mean ± standard error of 1004 the mean (SEM). 1005 1006 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 28 1007 1008 1009 Figure S2. Exercise reverses stress-induced anxiety-like behaviors (five-minute analysis). 1010 A. Heat maps representing the spatial distribution of the average exploration time in the 1011 light compartment (top section of each panel). Tracking plot of the exploration trajectory 1012 of a representative animal per group. Behavioral time budget plot showing individual visits 1013 to the light (white bars) and dark compartment (black bars) of the DL test during the test 1014 (bottom section of each panel). B. Frequency analysis of the exploration time in the light 1015 compartment binned every five seconds. C. Cumulative time in the light compartment. 1016 The analysis of the average duration per group revealed S + E mice spent significantly 1017 more time in the light compartment compared to the stressed group (P= 0.0145; One-way 1018 ANOVA: F (2, 36) = 4.596, P = 0.0167). D. Stress significantly reduced the total number 1019 of visits to the light compartment compared to N mice (P < 0.001), with S + E mice showing 1020 increased number of visits compared to S mice (P= 0.002; One-way ANOVA: F (2, 36) = 1021 9.539, P = 0.0005). E. No differences in the distribution of bout duration were found 1022 between groups. F. Cumulative distance traveled in the light compartment over testing 1023 minutes. The analysis of the average distance travelled per group revealed that stress 1024 significantly reduced animals’ displacement compared with N mice (P = 0.025), with S + 1025 E mice showing increased levels compared to S subjects ( P = 0.013; One-way ANOVA: 1026 F (2, 36) = 4.128, P = 0.0243). G. Stress also significantly increases the latency to explore 1027 the light compartment compared to N mice (P < 0.001), with S + E mice showing a reduced 1028 latency compared to S subjects (P < 0.001; One- way ANOVA: F (2, 36) = 13.09, P < 1029 0.001). Tukey's multiple comparisons test was used. Data shown represents the mean ± 1030 standard error of the mean (SEM). 1031 1032 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 29 1033 1034 1035 Figure S3. Overexpression of TrkB.T1 in CRHPVN neurons does not induce STP or blocks 1036 stress-induced STP. A. Experimental design for whole-cell patch clamp recordings from 1037 CRHPVN neurons of naïve and stress mice expressing TrkB.T1 receptor in CRH PVN 1038 neurons. B. Summary of excitatory post- synaptic current (EPSCs) amplitudes following 1039 HFS (grey bar) relative to baseline (BL; dotted line). A mixed ANOVA revealed Time ( F 1040 (13, 377) = 8.769, P < 0.001) and Group (F (1, 30) = 8.630, P = 0.006) main effects, with 1041 a Time x Group interaction (F (13, 377) = 4.637, P < 0.001). EPSC amplitude significantly 1042 increased after HFS (min 5 vs min 7: P < 0.001) in STrkB.T1 mice only. C. Average EPSC 1043 amplitude per cell one minute after HFS. STrkB.T1 mice showed higher amplitude compared 1044 to NTrkB.T1 animals (P < 0.001). D. Viral construct of the Opto -cytTrkB(E281A)-HA AAV 1045 (top). Confirmatory confocal image of CRHPVN cells expressing Opto-cytTrkB(E281A)-HA 1046 labeled with Alexa-488 (bottom). Tukey's multiple comparisons test was used. Geisser -1047 Greenhouse correction was used. Data shown represents the mean ± SEM. 1048 1049 1050 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 30 1051 1052 1053 Figure S4. Post-stress activation of TrkB in CRH PVN cells reverses stress -induced STP 1054 and anxiety -like behaviors (five -minute analysis). A. Heat maps represent the spatial 1055 distribution of the average exploration time in the light compartment (top section of each 1056 panel) during the first testing minute. Tracking plot of the exploration trajectory of a 1057 representative animal per group. Behavioral time budget plot showing individual visits to 1058 the light (white bars) and dark compartment (black bars) of the DL test during the first 1059 minute of testing (bottom section of each panel). B. Frequency analysis of the exploration 1060 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint 31 time in the light compartment binned every five seconds during the first testing minute. C. 1061 Heat maps represent the spatial distribution of the average exploration time in the light 1062 compartment (top section of each panel) during the five minutes of the test. Tracking plot 1063 of the exploration trajectory of a representative animal per group. Behavioral time budget 1064 plot showing individual visits to the light (white bars) and dark compartment (black bars) 1065 of the DL test during the five minutes of the test (bottom section of each panel). D. 1066 Frequency analysis of the exploration time in the light compartment binned every five 1067 seconds during the five minutes of testing. E. Cumulative time in the light compartment. 1068 The analysis of the average duration per group revealed that both S + E TrkB.T1 (P = 0.04) 1069 and SOptoTrkB mice (P < 0.001; showed increased time in the light compartment compared 1070 to S mChery, One -way ANOVA: F (2, 33) = 6.584, P = 0.004). F. No between- group 1071 differences were found when analyzing the total number of visits to the light compartment. 1072 G. No differences in the distribution of bout duration were found between groups. H. 1073 Cumulative distance travelled in the light compartment. The analysis of the average 1074 duration per group revealed that S OptoTrkB mice travelled longer distances in the light 1075 compartment compared to S mChery animals ( P = 0.002; One- way ANOVA: F (2, 33) = 1076 7.245, P = 0.002). I. The analysis of the latency to explore the light compartment revealed 1077 shorter latencies in SOptoTrkB mice compared to SmChery (P = 0.013) and S + ETrkB.T1 subjects 1078 (P = 0.032; One-way ANOVA: F (2, 33) = 4.006, P = 0.028). Tukey's multiple comparisons 1079 test was used. Data shown represents the mean ± SEM. 1080 1081 1082 1083 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 8, 2026. ; https://doi.org/10.64898/2026.01.07.698178doi: bioRxiv preprint