Environmental enrichment reverses ELS-induced visceral pain and depression through a prefrontal-limbic circuit involving CB1Rs

Abstract

Neonatal colorectal distension (CRD), as a mechanical stress stimulus in early life to simulate irritable bowel syndrome (IBS), increases susceptibility to visceral pain and depression-like behaviors in adulthood. Enriched environment (EE) treatment effectively counteracts these effects, but the underlying neural circuit and molecular mechanisms remain poorly defined. Here, we investigate whether EE reverses visceral pain and depression by modulating cannabinoid type-1 receptors (CB1Rs) in the PrL^Glu→avBNST pathway. A multidisciplinary combination of behaviors, chemogenetics, optogenetics, pharmacology, molecular and electrophysiological approaches is applied. In CRD rats, CB1Rs-expressing glutamatergic neurons in prelimbic cortex (PrL) projecting to the anteroventral bed nucleus of the stria terminalis (avBNST) promote visceral pain and depression. EE exerts analgesic and antidepressant effects by rescuing PrL^Glu→avBNST^GABA pathway activity and reducing CB1Rs expression. EE also reverses the CRD-induced paraventricular nucleus (PVN) hyperactivity by reversing the dysfunction of this pathway. Together, EE alleviates chronic visceral pain and comorbid depression by restoring homeostasis in the CB1Rs-modulated PrL^Glu→avBNST^GABA→PVN pathway. Combining EE and CB1Rs-targeted pharmacological modulation offers a promising therapeutic strategy for early-life developmental disorder-related pain-depression comorbidities.

Environmental enrichment sheds light on the reversal of ELS-induced visceral pain and depression through a prefrontal-limbic circuit involving CB1Rs.

Introduction

Early life stress (ELS), encompassing chronic adverse experiences (e.g., poverty, neglect, or trauma), induces persistent maladaptations in stress response systems, leading to long-term consequences. Compelling epidemiological evidence identifies ELS as a significant environmental determinant in the pathogenesis of disorders of gut-brain interaction (DGBIs), particularly IBS. IBS, a gut-brain axis disorder characterized by recurrent abdominal pain, altered bowel habits, and abdominal bloating¹, affects 5–10% of the global population with 2:1 female predominance^2,3. Notably, 29% of IBS patients exhibit depression, 39% anxiety, and 23% exhibit both^4–6. Currently, IBS management includes peripheral (e.g., antispasmodics) and central (e.g., antidepressants) pharmacotherapies^7,8. While antidepressants improve symptoms with demonstrated cost-effectiveness^9,10, their adverse effects necessitate alternatives such as cognitive behavioral therapy (CBT), which alleviates overall symptomatology^11,12. Emerging integrated CBT-drug approaches highlight the need to elucidate precise neural circuits and molecular pathophysiology.

As a well-established risk factor for both IBS and subsequent emotional comorbidities in adulthood¹³, ELS leads to persistent alterations in hypothalamic-pituitary-adrenal (HPA) axis function¹⁴. Neuroanatomical studies by Sawchenko and Swanson first characterized the predominant origin of PVN-projecting BNST neurons in the anteroventral bed nucleus of the stria terminalis (avBNST)¹⁵. The avBNST consists of predominantly GABAergic neurons (~80-90%) and a smaller population of glutamatergic neurons¹⁶. Our previous findings have identified that the excitability of GABAergic and glutamatergic neurons in the avBNST that project to PVN decreased and elevated respectively, contributing to HPA axis activation and chronic visceral pain^17–20. The GABAergic neurons in the avBNST can be subdivided into distinct subtypes based on molecular markers (e.g., somatostatin (SST), calretinin (CR), and corticotropin-releasing hormone (CRH)) and projection targets, which differentially regulate stress responses, anxiety, and reward processing, highlighting their functional heterogeneity in emotional and motivational circuits²¹. What neural circuits and molecular mechanisms drive the reduced excitability of avBNST GABAergic neurons?

The corticolimbic system, particularly its descending regulatory pathways, plays a pivotal role in modulating chronic visceral pain and its affective comorbidities. As an important part of the limbic system, avBNST receives substantial inputs from the medial prefrontal cortex (mPFC) along with other limbic structures, including the hippocampus and amygdala. Previous studies suggest that ELS alters structural and functional plasticity in the medial prefrontal cortex (mPFC). As a central hub linking chronic pain to its mental comorbidities, the mPFC’s plasticity changes are thought to contribute to the altered resting-state connectivity observed in these conditions²². Anatomically, the mPFC is subdivided into the prelimbic cortex (PrL), the infralimbic cortex (IL), the dorsal peduncular cortex (DP), and the anterior cingulate cortex (ACC). The mPFC neural architecture is dominated by glutamatergic pyramidal neurons (80-90%) and inhibitory GABAergic interneurons (~10–20%). While deep layer (V-VI) mPFC pyramidal neurons predominantly mediate long-distance subcortical projections, superficial layers (II-III) participate in both local microcircuits and corticocortical connections²³. In particular, prefrontal circuits involving descending projections to limbic areas, which regulate stress, emotion, motivation, and pain, could be a crucial output pathway. A recent meta-analysis of imaging data demonstrated that, different from the greater engagement of areas correlated with emotional arousal and homeostatic afferent processing in IBS patients, controls show more pronounced activation in cortical regions involved in the regulation of pain and emotion (especially lateral and medial PFC), which is largely absent in IBS patients²⁴. Congruent with these human brain imaging findings, chronic stress rats presented a marked reduction in CRD-evoked functional brain activation in PrL region²⁵. Our previous study also showed that glutamatergic neuronal excitability of PrL is decreased significantly in depression induced by restraint-stress paradigm²⁶. What are the precise molecular mechanisms driving PrL circuit dysfunction in visceral pain and stress-related comorbidities?

The endocannabinoid (eCB) system, particularly the cannabinoid receptors in peripheral gastrointestinal tract, has been delineated as a promising therapeutic target for IBS²⁷. However, the clinical application of cannabinoid agonists for visceral pain is constrained by their psychoactive side effects mediated through central mechanisms. The endocannabinoid system signals through two principal cannabinoid receptor subtypes with distinct distributions and functions. Cannabinoid type 1 receptors (CB1Rs) are abundantly expressed at presynaptic terminals throughout the central nervous system (CNS), where their modulation of neurotransmitter release and synaptic plasticity explains both the therapeutic benefits and psychoactive effects of cannabinoid compounds^28,29. In contrast, cannabinoid type 2 receptors (CB2Rs) exhibit predominant expression in peripheral immune cells, with more restricted CNS localization primarily observed in microglia and select neuronal subpopulations³⁰. In patients with IBS, visceral pain was exaggerated during episodes of stress³¹, but the endocannabinoid-like dietary supplement could effectively alleviate abdominal pain and discomfort³². The animal experiments also supported CB1Rs- and CB2Rs-mediated amelioration of stress-induced visceral hypersensitivity^33–36. CB1R signaling plays a critical role in regulating both emotional and pain-related processes through modulation of corticolimbic circuits. Evidence from psychiatric disorders and CB1R knockout models indicates that CB1R functions as a context-dependent regulator of neural excitability, linking affective dysfunction with altered pain sensitivity^37,38. CB1Rs-mediated retrograde signaling regulates presynaptic neurotransmitter release and synaptic plasticity, thereby shaping glutamatergic neuronal excitability in the PrL, which may contribute to chronic visceral pain and associated affective dysfunction.

Empirical evidence demonstrates that positive caregiving experiences protect against the stress-sensitive effects of childhood neglect on externalizing problems during adolescence. Environmental enrichment (EE) in rodents is a well-known environmental intervention analogous to CBT, which consists of more social interactions, novel object stimuli, and physical exercise among others³⁹. EE can reverse many harmful effects of ELS by promoting neurogenesis, dendrite development, and neuroplasticity, and modulating HPA activity to stress^40,41. Many studies have shown that EE increased the activity and functionality of the PFC^41–43. However, there is a lack of preclinical data on whether EE can reverse ELS-induced visceral pain and pain-related disorders. Here, we define EE as an environmental paradigm involving enhanced sensory, cognitive, social, and physical stimulation, which has been historically associated with alleviation of visceral pain and mood disorders. It is critically important to further investigate the role of EE in the prefrontal-limbic pathway.

In this study, we investigated whether CB1Rs-regulated visceral pain and depression-like behaviors are associated with suppressed PrL^Glu→avBNST pathway activity, and whether EE reverses this effect. Our findings demonstrate that CB1Rs play a critical role in modulating a corticolimbic circuit that forms a complex neural network governing both visceral pain and associated emotional comorbidities. These results highlight the potential of targeting CB1Rs-mediated pathways for therapeutic intervention in pain and mood-related disorders.

Results

Reduced excitation of avBNST GABAergic neurons contributes to early life stress-induced visceral hypersensitivity and depression-like behaviors in adulthood

ELS is a major risk factor for visceral pain and depression-related disorders. Our previous studies demonstrated that neonatal CRD induces chronic visceral pain in SD rats, with PVN-projecting GABAergic neurons in avBNST playing a key role in visceral hyperalgesia^17,44. In this study, we employed a battery of behavioral tests to assess visceral hypersensitivity and depression-like behaviors. A schematic diagram of the neonatal CRD protocol and related behavioral tests in adulthood is shown in Fig. 1A. CRD-exposed rats exhibited reduced visceral pain threshold (Fig. 1B), elevated AWR scores (Fig. 1C), enhanced EMG responses to graded CRD (Fig. 1D, E), increased immobility in the FST (Fig. 1F), and decreased sucrose preference (Fig. 1G), confirming concurrent visceral hypersensitivity and depression-like behaviors.

Fig. 1. Functional characterization of avBNST GABAergic neurons in visceral pain and depression-like behaviors.

A Experimental timeline of neonatal CRD model. B Adult rats with neonatal CRD showed reduced visceral pain threshold (n = 50 rats). C Neonatal CRD increased the AWR scores at 20–50 mmHg (n = 10 rats). D Schematic diagram and heatmap of the EMG recordings. E CRD group showed increased EMG activity in response to CRD at 30–50 mmHg (n = 10 rats). F–G CRD rats exhibited prolonged immobility time in the FST and reduced sucrose preference (n = 50 rats). H Patch-clamp recording configuration (left) and representative traces (right) of avBNST GABAergic neurons. I Decreased firing frequency in CRD group (n = 6 neurons from 3 rats). J–M Exogenous NMDA increased pain threshold, decreased AWR scores, reduced immobility time in the FST, and increased sucrose preference in CRD rats (n = 6–8 rats). N–Q Exogenous AMPA increased pain threshold, decreased AWR scores, reduced immobility time in the FST, and increased sucrose preference (n = 6 rats). Data are expressed as the mean ± SEM. *P < 0.05, **P < 0.01, as determined by two-tailed Student’s t-test (B, F, G, I–N, P, Q), and two-way repeated-measures ANOVA (C, E, K, O).

Based on our finding that disinhibition of PVN-projecting avBNST GABAergic neurons participates in the visceral hypersensitivity in rats induced by neonatal CRD¹⁷, we investigated neuronal activity and glutamatergic synaptic input onto avBNST GABAergic neurons. Whole-cell recordings revealed decreased firing frequencies in avBNST GABAergic neurons from CRD rats versus controls (Fig. 1H, I). Our whole-brain mapping of PrL glutamatergic projections clearly demonstrates that axonal terminals from the PrL have the densest projection density within the avBNST, with significantly less innervation in the AL or AM subnuclei (Fig. S1A). This identifies the avBNST as the primary structural target for PrL modulation. Histological verification of our micro-infusion sites confirms that the 0.2 μL injection volume was localized within the avBNST (Fig. S1B).

Given that PrL projections to avBNST are primarily glutamatergic, we next examined whether enhancing glutamatergic signaling could rescue CRD-induced behavioral abnormalities. Exogenous administration of N-methyl-D-aspartic acid (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA) was performed, and behavioral responses were assessed 30 min post-administration. Consequently, both NMDA and AMPA administration significantly increased visceral pain threshold (Fig. 1J, N), reduced AWR scores (Fig. 1K, O), decreased immobility time in the FST (Fig. 1L, P), and increased sucrose preference (Fig. 1M, Q) in CRD rats. Taken together, these results demonstrated that avBNST neurons played an important role in mediating both visceral hypersensitivity and depression-like behaviors induced by neonatal CRD exposure.

Hypoactivity of PrL glutamatergic neurons contributes to visceral hypersensitivity and depression-like behaviors in CRD rats

We first assessed neuronal activity in forebrain nuclei via c-Fos expression. CRD rats showed altered c-Fos levels in multiple regions, including the PrL, anterolateral BNST (alBNST), avBNST, and PVN (Fig. S2). To map connectivity between the avBNST and forebrain, we employed retrograde and anterograde viral tracing. The first strategy used a retrograde transmonosynaptic virus system (Fig. 2A, left). Representative images of starter cells from the avBNST are shown (Fig. 2A, right). As demonstrated by our retrograde tracing in Fig. 2B, these projection neurons are predominantly located in the deep layers (Layers 5 and 6) of the PrL (Fig. 2B, left). We further found the majority of PrL neurons projecting to the avBNST provide dense glutamatergic afferents to innervate GABAergic neurons within the avBNST (Fig. 2B, right). The second strategy employed an anterograde transsynaptic tracing system (Fig. 2C, left). Confocal microscopy revealed that the majority of green fluorescent-labeled neurons projecting from PrL glutamatergic neurons to the avBNST region were co-localized with GAD67 immunoreactivity (Fig. 2C, right). These results identified the existence of an anatomical PrL^Glu→avBNST^GABA neural circuit.

Fig. 2. Glutamatergic PrL neurons were hypoactive in visceral pain and depression-like behaviors.

A The timeline for retrograde rabies virus injection (left) and the representative images of DsRed+ neurons in avBNST (right). B Schema of mPFC and immunofluorescent staining of CaMKIIα and DsRed in PrL. C Anterograde transsynaptic viral vector strategy to target PrL to avBNST projections and immunofluorescent staining of GAD67 and WGA in avBNST. D Representative images of CaMKIIα and c-Fos expression in PrL. Scale bar = 100 μm. E The percentage of c-Fos-positive CaMKIIα neurons in PrL (n = 5 rats). F Schematic of whole-cell patch-clamp recording of PrL^Glu neurons. G, H Representative traces and summarized data for the firing rate in response to the current intensity at 0–140 pA in control and CRD rats (n = 9 or 13 neurons from 6 rats). I Time flow for the chemogenetic modulation of PrL glutamatergic neurons (left) and representative images of hM3Dq-mCherry expression in PrL (right). J–M Chemogenetic activation of PrL glutamatergic neurons elevated pain threshold, decreased AWR scores and immobility time in the FST, and enhanced sucrose preference in CRD rats (n = 6 rats). N–Q Chemogenetic inhibition of PrL glutamatergic neurons decreased pain threshold, elevated AWR scores and immobility time in the FST, and reduced sucrose preference in control rats (n = 6 rats). R Timeline for the optogenetic modulation of PrL glutamatergic neurons (left) and representative images of ChR2-mCherry and eNpHR-EYFP expression in the PrL (right). S–U Optogenetic activation of PrL glutamatergic neurons enhanced pain threshold, decreased AWR scores, and decreased immobility time in the FST in CRD rats (n = 3 rats). V–X Optogenetic inhibition of PrL glutamatergic neurons decreased pain threshold and elevated the AWR scores and immobility time in the FST in control rats (n = 3 rats). Data are expressed as the mean ± SEM. *P < 0.05, **P < 0.01, as determined by two-tailed Student’s t test (E, J, L, M, N, P, Q, S, U, V, X), and repeated-measures ANOVA (H, K, O, T, W).

Next, we investigated whether chemogenetic manipulation of PrL glutamatergic neuron excitability could reverse pain and depression-related behaviors. CRD rats exhibited reduced excitability of PrL glutamatergic neurons, evidenced by decreased c-Fos/CaMKIIα co-labeling (Figs. 2D, E) and lower firing frequency in slice recordings (Fig. 2F–H). Additionally, we injected AAV-CaMKIIα-cre and AAV-DIO-hM3Dq-mCherry or AAV-DIO-mCherry in the PrL region of CRD rats, followed by CNO administration on day 21, and conducted relevant behavioral tests 30 min after CNO injection (Fig. 2I). Immunofluorescence revealed hM3Dq-mCherry was expressed in the PrL (Fig. 2I, right). Chemogenetic activation of PrL glutamatergic neurons in CRD rats resulted in increased visceral pain threshold (Fig. 2J), decreased AWR scores (Fig. 2K) and FST immobility time (Fig. 2L), and elevated sucrose preference (Fig. 2M). Furthermore, chemogenetic inhibition of PrL glutamatergic neurons in control rats induced visceral hypersensitivity and depression-like behaviors (Fig. 2N–Q).

Next, we assessed whether photomanipulation of PrL glutamatergic neurons in vivo altered the visceral pain and depression-like behaviors. A schematic of virus injection and timeline is shown (Fig. 2R). Optogenetic activation of PrL glutamatergic neurons in CRD rats elevated visceral pain threshold (Fig. 2S), reduced AWR scores (Fig. 2T), and shortened immobility time in the FST (Fig. 2U). Conversely, optogenetic inhibition of PrL glutamatergic neurons in control rats induced visceral hypersensitivity (Fig. 2V), elevated AWR scores (Fig. 2W), and prolonged immobility time in the FST (Fig. 2X). These results indicate that reduced activity of PrL glutamatergic neurons contributes to the pathogenesis of both visceral hypersensitivity and depression-like behaviors in CRD rats.

The PrL^Glu→avBNST^GABA circuit modulates visceral hypersensitivity and depression-like behaviors

To elucidate the functional role of PrL^Glu→avBNST^GABA circuit in visceral hypersensitivity and depression-like behaviors, we employed chemogenetic and optogenetic approaches to modulate this pathway. A schematic diagram of the circuit manipulation strategy is shown in Fig. 3A. The effects of hM3Dq-mCherry virus were verified in our previous study¹⁷. Chemogenetic activation of the PrL^Glu→avBNST circuit in CRD rats significantly increased visceral pain threshold and sucrose preference while reducing AWR scores, EMG activity, and immobility time in the FST (Fig. 3B–G). Conversely, circuit inhibition in control rats significantly decreased visceral pain threshold and sucrose preference while elevating AWR scores, EMG activity, and immobility time in the FST (Fig. 3H–M).

Fig. 3. Modulation of visceral pain and depression-like behaviors by PrL^Glu$\to$ avBNST projections.

A Experimental diagram for genetic strategy. B, C Chemogenetic activation of PrL^Glu→avBNST neurons increased pain threshold and decreased the AWR scores in CRD rats (n = 6 rats). D Schematic diagram and heatmap of the external oblique muscle. E Decreased EMG responses in CRD rats after chemogenetic activation (n = 6 rats). F, G Chemogenetic activation of PrL^Glu→avBNST neurons reduced immobility time in the FST and elevated the sucrose preference in CRD rats (n = 6 rats). H–I Chemogenetic inhibition of PrL^Glu→avBNST neurons decreased pain threshold and increased the AWR scores in control rats (n = 6 rats). J Schematic diagram and heatmap of the external oblique muscle. K Control rats with hM4Di showed increased myoelectric activity in response to CRD at 40 and 50 mmHg (n = 6 rats) compared with the mCherry group. L, M Chemogenetic inhibition of PrL^Glu→avBNST neurons increased immobility time in the FST and decreased the sucrose preference in control rats (n = 6 rats). N Patch-clamp recordings of PrL glutamatergic neurons. O Blue light-evoked APs and EPSCs. P Representative traces of avBNST neuron activity evoked by blue light stimulation of PrL→avBNST terminals. Q Representative images showing immunostaining for the optogenetic manipulation of PrL glutamatergic projection neurons that innervate the avBNST. R–T The influences of optogenetic activation of the PrL^Glu→avBNST circuit on pain sensation and FST behavior in CRD rats. U Patch-clamp recording of avBNST neurons receiving PrL glutamatergic input (above) and representative traces of yellow light evoked suppression of APs (below). V–X The influences of optogenetic inhibition of PrL^Glu→avBNST circuit on the pain sensitivity and FST in control rats. Data are expressed as the mean ± SEM. *P < 0.05, **P < 0.01, as determined by two-tailed Student’s t-test (B, F, G, H, L, M, R, T, V, X), and two-way repeated-measures ANOVA (C, E, I, K, S, W).

To validate functional synaptic connectivity, we injected AAV-CaMKIIα-ChR2-mCherry into the PrL and performed ex vivo electrophysiological recordings (Fig. 3N). Blue light stimulation (470 nm) evoked frequency-dependent (2–20 Hz) excitatory postsynaptic currents (eEPSCs) and action potentials (APs) in ChR2-expressing PrL neurons (Fig. 3O). Whole-cell recordings in avBNST slices revealed that photostimulation of PrL terminals elicited EPSCs, which were completely blocked by tetrodotoxin (TTX). Moreover, EPSCs were abolished by TTX and subsequently restored by 4-aminopyridine (4-AP), indicating monosynaptic connectivity. The synaptic responses were further confirmed to be glutamatergic, as they were almost completely blocked by the AMPA receptor antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX). We found that CNQX almost completely blocked the eEPSCs (Fig. 3P). These data demonstrated that glutamatergic PrL neurons provided functional glutamatergic input to avBNST neurons.

In the optogenetic regulation of the specific PrL-avBNST circuit, glutamatergic projection neurons are primarily located in layers V and VI of the PrL (Fig. 3Q). Given the anatomical distribution of these cells in L5/6, light delivered via the fiber effectively modulates this specific output population regardless of the broader light spread, as non-projecting neurons in the superficial layers do not express the viral transgene. Consistent with our chemogenetic findings, optogenetic activation of the PrL^Glu→avBNST circuit in CRD rats significantly increased visceral pain threshold (Fig. 3R) while reducing both AWR scores and FST immobility time (Fig. 3S–T). To characterize circuit inhibition, we expressed eNpHR-EYFP in PrL neurons and recorded from PrL-innervated avBNST cells; the APs firing was abolished after yellow light stimulation (Fig. 3U). In control rats, optogenetic inhibition of PrL^Glu→avBNST circuit decreased the visceral pain threshold (Fig. 3V) while increasing both AWR scores and FST immobility time (Fig. 3W–X). Thus, our data suggest that the PrL does not act through isolated neurons but rather as a coordinated functional unit, with the deep-layer output to subcortical targets like the avBNST being a key component of this system.

Previous studies have demonstrated distinct subregional distributions of BNST GABAergic neurons expressing different isoforms of glutamic acid decarboxylase (GAD)²¹. Specifically, we confirmed that both SOM+ and CR+ neurons in avBNST are essential for mediating visceral hypersensitivity and depression-like behaviors (Fig. S3). Given their overlapping roles in these pathological processes, subsequent experiments did not further differentiate between GABAergic neuronal subtypes in the avBNST. This reduced activity of PrL^Glu→avBNST circuit likely contributes to the observed behavioral phenotypes, suggesting a key role for this pathway in the comorbidity of visceral pain and affective disorders.

Upregulated CB1 receptors on avBNST-projecting PrL glutamatergic neurons contribute to visceral hypersensitivity and depression-like behaviors in adult rats exposed to early life stress

CB1Rs regulate neuronal excitability and synaptic plasticity in the central nervous system. To investigate the mechanism of reduced excitability in PrL glutamatergic neurons of CRD rats, we first confirmed CB1Rs expression in these neurons using FISH with CaMKIIα and CB1R probes (Fig. 4B). Notably, 88.8% of CaMKIIα neurons expressed CB1Rs mRNA, and 95.5% of CB1Rs mRNA localized to CaMKIIα neurons (Fig. 4C). Total CB1Rs mRNA levels in the PrL were increased (Fig. 4D). Western blot showed increased CB1Rs protein in the avBNST of CRD rats (Fig. 4E). Since CB1Rs are primarily presynaptic, we verified their presence in PrL→avBNST glutamatergic terminals by injecting AAV-CaMKIIα-EGFP into PrL (Fig. 4F). As expected, CB1Rs antibody co-labeled CaMKIIα-EGFP axons in avBNST (Fig. 4G and Supplementary Video 1).

Fig. 4. CB1R localization at PrL terminals in avBNST modulates visceral pain and depression-like behaviors.

A Schematic diagram depicting the experiment. B Representative images of in situ hybridization using CaMKIIα and CB1R probes in PrL. C Co-labeled ratio of CaMKIIα and CB1R (n = 3 rats). D The CB1R mRNA levels in PrL (n = 4 rats). E CB1R immunoreactivity was significantly enhanced in CRD rats (n = 6 rats). F Virus injection strategy (left) and CaMKIIα-EGFP expression in PrL (right). G CaMKIIα-EGFP-labeled nerve terminals were co-labeled with anti-CB1R antibody and counterstained with DAPI. H The timeline for virus injection and behavioral detection. I CB1R mRNA level in PrL was significantly decreased after sh-CB₁R-1 or sh-CB₁R-2 treatment (n = 5 rats). J CB1R immunoreactivity in the avBNST was significantly reduced after sh-CB₁R treatment (n = 5 rats). K The visceral pain threshold was significantly increased after sh-CB₁R-1 or sh-CB₁R-2 injection into the PrL (n = 6 rats). L The immobility time of FST was significantly decreased after sh-CB₁R-1 or sh-CB₁R-2 injection (n = 6 rats). M Visceral pain threshold increased 30 min after AM251 injection in CRD rats (n = 6 rats). N AWR scores decreased 30 min after AM251 injection in CRD rats (n = 6 rats). O Immobility time of FST declined 30 min after AM251 injection in CRD rats (n = 6 rats). P CB1R, Q GAD67, and R GAD65 immunoreactivity levels after AM251 treatment (n = 3–5 rats). In control rats (n = 6 rats), 30 min after WIN55,212-2 injection, visceral pain threshold decreased (S), AWR scores elevated (T), and immobility time in the FST increased significantly (U). V CB1R, W GAD67, and (X) GAD65 immunoreactivity signals after WIN55,212-2 treatment (n = 5–6 rats). Data are expressed as the mean ± SEM. *P < 0.05, **P < 0.01, as determined by two-tailed Student’s t-test (C–E, P–R, V–X), and two-way ANOVA followed by Tukey’s post hoc test (I–L, M, O, S, U), and repeated-measures ANOVA (N, T).

To assess functional roles, we injected AAV-CaMKIIα-CB1Rs-shRNA (or Scramble) into PrL of CRD rats (Fig. 4H). CB1Rs knockdown reduced CB1Rs mRNA expression in PrL (Fig. 4I) and attenuated CB1Rs immunoreactivity in the avBNST (Fig. 4J), concomitant with an elevated visceral pain threshold and reduced immobility time in the FST (Fig. 4K, L). Pharmacological manipulation further supported these findings. Microinjecting the CB1Rs antagonist AM251 into avBNST increased pain threshold, lowered AWR scores, and decreased FST immobility in CRD rats (Fig. 4M–O). Notably, AM251 treatment reduced detectable CB1Rs immunoreactivity (Fig. 4P) but altered GAD67 and GAD65 immunoreactivity patterns (Fig. 4Q, R). These acute shifts in protein immunoreactivity suggest that CB1Rs antagonism might facilitate the activation of GABAergic neurons. Furthermore, we performed additional control experiments by administering AM251 to control rats and assessing its impact on visceral sensitivity and mood-related behaviors (Fig. S4). Our data indicate that AM251 reduced the visceral pain threshold in naive rats, but did not affect their immobility time in the forced swimming test. WIN55,212-2 was initially chosen due to its high efficacy and extensive characterization in central pain-processing models, where CB1Rs are the predominant isoform^45,46. Conversely, the CB1Rs agonist WIN55,212-2 induced a lower visceral pain threshold and increased depressive-like behaviors in control rats (Fig. 4S–U), while augmenting CB1Rs immunoreactivity (Fig. 4V), and suppressing GAD67 while increasing GAD65 signals (Fig. 4W–X). This suggests that acute CB1R agonism inhibits GABAergic transmission in the avBNST.

Furthermore, we characterized the effects of WIN 55,212-2 in CRD rats (Fig. S5). Our results confirm that pharmacological activation of the cannabinoid system in this circuit effectively induces visceral hypersensitivity and depression-like behaviors similar to those observed after neonatal stress. Given that WIN 55,212-2 is a non-selective agonist, we conducted additional experiments to verify whether these effects are exclusively mediated via CB1Rs rather than CB2Rs. Microinfusion of the selective CB1Rs agonist ACEA⁴⁷ (2.5 μM, 0.2 μL) into the avBNST of CRD rats yielded pronociceptive and depressogenic effects consistent with those of WIN 55,212-2 (Fig. S6). Collectively, these findings suggest that CB1Rs signaling on PrL→avBNST glutamatergic terminals is critical for driving visceral hypersensitivity and depressive-like behaviors, likely via the acute presynaptic modulation of GABAergic transmission.

The PVN is a key downstream output of the PrL^Glu→avBNST^GABA circuit mediating chronic visceral hypersensitivity

Given that our previous work identified the PVN as a key node in ELS-induced visceral pain, we next asked whether it acts as a downstream target of the PrL^Glu→avBNST^GABA circuit. We used retrograde transsynaptic tracing to map this pathway. We injected AAV-Retro-Vgat-Cre into the PVN and AAV-DIO-TVA-EGFP + AAV-DIO-RVG into the avBNST, followed by RV-EnvA-ΔG-DsRed (Fig. 5A). EGFP+ neurons in avBNST confirmed GABAergic projections to the PVN (Fig. 5B), while RV-DsRed-labeled neurons in the PrL further identified a PrL→avBNST^GABA→PVN polysynaptic pathway (Fig. 5C).

Fig. 5. Dissection of the PrL^Glu$\to$ avBNST^GABA$\to$ PVN circuit.

A Schematic of viral injection for triple retrograde tracing. B Representative images of viral expression in avBNST. C DsRed signals from the avBNST were detected in PrL. D Stereotaxic delivery of viruses encoding a Cre-dependent hM3Dq into the PVN-projecting GABAergic avBNST neurons receiving PrL projection. E The expression of hM3Dq-positive neurons in avBNST was stained with anti-GABA antibody. Scale bar = 100 μm. F Representative images of c-Fos expression of PVN-projecting GABAergic avBNST neurons receiving PrL projection after hM3-CNO administration. Scale bar = 50 μm. G c-Fos expression in GABAergic avBNST neurons of the PrL→avBNST^GABA→PVN circuit in hM3Dq group compared with the control group. H Intraperitoneal injection of CNO significantly elevated visceral pain threshold following activation of GABAergic avBNST neurons in the PrL^Glu→avBNST^GABA→PVN pathway (n = 8 rats). However, immobility time in the FST I and sucrose preference J showed no significant changes (n = 8 rats). K Stereotaxic delivery of viruses encoding a Cre-inducible hM4Di into the PVN-projecting GABAergic avBNST neurons receiving PrL projection in control rats. L The expression of hM4Di-positive neurons in avBNST was stained with anti-GABA antibody. Scale bar = 100 μm. M Patch-clamp recordings of GABAergic neurons in avBNST belonging to the PrL^Glu→avBNST^GABA→PVN pathway. N Sample traces for firing rate recorded from the avBNST GABAergic neurons. O Representative images of c-Fos expression of PVN-projecting avBNST GABAergic neurons receiving PrL projection after hM4-CNO administration. Scale bar = 100 μm. P c-Fos expression in GABA avBNST neurons of PrL→avBNST^GABA→PVN circuit in hM4Di group compared with control group. Q Intraperitoneal injection of CNO significantly decreased the visceral pain threshold after inhibition of GABAergic avBNST neurons in PrL^Glu→avBNST^GABA→PVN pathway (n = 6 rats). Whereas, R immobility time of FST and S sucrose preference showed no significant changes (n = 6 rats). Data are expressed as the mean ± SEM. **P < 0.01, as determined by two-tailed Student’s t-test (G–J, P–S).

To test whether this pathway modulates visceral pain and depressive-like behaviors, we activated PrL→avBNST→PVN GABAergic neurons using a FLP/Cre-dependent DIO-hM3Dq system (Fig. 5D). hM3Dq-mCherry+ neurons colocalized with GABA (Fig. 5E). CNO administration increased c-Fos in hM3Dq-mCherry+ GABAergic neurons (Fig. 5F, G) and elevated visceral pain threshold (Fig. 5H), but did not alter depression-like behaviors (FST, SPT; Fig. 5I, J). Next, we inhibited GABAergic neurons in this pathway in controls using hM4Di (Fig. 5K). hM4Di-mCherry+ GABAergic neurons colocalized with GABA (Fig. 5L), and CNO reduced both firing frequency in patched avBNST GABAergic neurons (Fig. 5M, N) and c-Fos expression in GABAergic neurons in the avBNST (Fig. 5O-P). PrL→avBNST→PVN GABAergic inhibition decreased visceral pain threshold (Fig. 5Q) but did not affect immobility time in FST or sucrose preference (Fig. 5R-S). Taken together, the PrL→avBNST^GABA→PVN pathway selectively regulates visceral hypersensitivity but did not significantly alter baseline depression-like behaviors in rats.

A key phenotype of enriched environments (EE) is the reversal of early life stress-induced visceral hypersensitivity and depression-like behaviors

To determine whether EE reverses ELS-induced behavioral abnormalities, rats were assigned to control, CRD, and CRD + EE groups (Fig. 6A). EE treatment significantly reversed the decreased visceral pain threshold observed in CRD rats (Fig. 6B) and reduced their AWR scores (Fig. 6C). Analysis of external oblique muscle EMG activity (Fig. 6D, E) revealed significantly decreased responses to graded CRD (30, 40 and 50 mmHg) in CRD + EE rats compared to CRD controls (Fig. 6F). Furthermore, EE-treated rats showed reduced immobility time in FST (Fig. 6G) and increased sucrose preference (Fig. 6H) compared with CRD rats. The EE apparatus and stimulation components are further illustrated in Supplementary Video 2.

Fig. 6. The effects of EE treatment on neonatal CRD-induced visceral pain and depression-like behaviors.

A Representative photograph of the EE cages used in the current experiment. B Visceral pain threshold was significantly elevated in the EE group vs the CRD group (n = 10 rats). C EE reversed the elevated AWR scores induced by neonatal CRD (n = 10 rats). D Sample traces and E heatmap of the EMG activity of the external oblique muscle. F EE reversed neonatal CRD-induced elevated EMG activity in response to CRD at 30, 40, and 50 mmHg (n = 10 rats). G Immobility time in the FST decreased significantly in the EE group vs the CRD group (n = 10 rats). H Sucrose preference increased significantly in the EE group vs the CRD group (n = 10 rats). I Stereotaxic viral injection. J, K Representative traces and summary data of the firing rate of PrL^Glu neurons from three groups in response to the current intensities at 0–140 pA (n = 7–13 neurons from 3 to 6 rats). L, M Representative traces and summary data of the spontaneous firing frequencies of avBNST^GABA neurons in three groups (n = 4 rats). N, O Representative traces and summary data of the spontaneous firing frequencies of PVN^CRH neurons in three groups (n = 10 neurons from 4 rats). P Schematic diagram of CB1Rs protein expression. Q Expression of CB1Rs protein in the avBNST region in control, CRD, CRD + EE groups. Data are expressed as the mean ± SEM. **P < 0.01, as determined by one-way ANOVA (B, G, H, M, O, Q), and repeated-measures ANOVA (C, F, K).

To explore the circuit basis underlying the beneficial effects of EE, we next examined the excitability of PrL glutamatergic neurons, avBNST GABAergic neurons, and PVN CRH neurons following viral labelling (Fig. 6I). PrL glutamatergic neurons of EE rats exhibited normalized excitability in response to a stepwise increase in stimulus intensity (40–140 pA) as compared with CRD rats (Fig. 6J, K). Conversely, avBNST GABAergic neurons showed significantly increased firing rates following EE intervention (Fig. 6L, M). Furthermore, the spontaneous firing rates of CRH neurons in the PVN were decreased in EE-treated rats compared to the CRD group (Fig. 6N, O). At the molecular level, EE treatment reversed the elevated CB1Rs expression observed in CRD rats, suggesting a key mechanism underlying its beneficial effects (Fig. 6P, Q). Together, these findings indicate that EE attenuates ELS-induced visceral hypersensitivity and depressive-like behaviors, accompanied by normalization of neuronal excitability along the PrL^Glu→avBNST^GABA→PVN^CRH circuit and reversal of CB1Rs upregulation.

Enriched environment alleviates visceral hypersensitivity and depression-like behaviors via CB1-dependent modulation of the PrL ^Glu→avBNST^GABA circuit

As mentioned above, experimental manipulation like EE may reverse the effects of ELS induced by neonatal CRD in rats. However, the upstream circuit mechanisms underlying EE-mediated rescue remain unclear. We next examined whether EE modulates the PrL→avBNST pathway. Workflow diagram depicting the experiment is shown (Fig. 7A). Since EE restores PrL glutamatergic neuronal excitability, we hypothesized that EE alleviates chronic visceral pain in CRD rats by normalizing PrL^Glu→avBNST circuit function. To test this hypothesis, we selectively ablated avBNST-projecting PrL glutamatergic neurons using a pro-apoptotic viral approach (AAV-Ef1α-flex-taCasp3 in PrL combined with retro-CaMKIIα-cre in avBNST; Fig. 7B). Caspase-mediated ablation reversed EE’s beneficial effects, resulting in renewed visceral hyperalgesia (Fig. 7C–G), worsened depression-like behavior demonstrated by prolonged immobility time in the FST (Fig. 7H), but notably maintained baseline levels of sucrose preference (Fig. 7I). These results implied that the avBNST-projecting PrL glutamatergic neurons mediate EE’s analgesic effects, but their role in depression requires further investigation. In addition, we performed pathway-specific caspase-3-mediated ablation of the PrL-avBNST circuit in control rats and assessed their behavioral profiles (Fig. S7). Interestingly, our results revealed a partially dissociable regulation of sensory and affective components. Ablation of the PrL-avBNST pathway in control rats significantly decreased the visceral pain threshold, mirroring the hypersensitivity seen in CRD rats. While the ablation induced robust pain, the induction of depression-like behaviors (measured via FST and SPT) was less pronounced in control rats compared to the CRD group. This suggests that this circuit exerts a constitutive “top-down” inhibitory tone on nociceptive processing. Similar to our pharmacological findings with the CB1R antagonist AM251, removing this “tonic brake” leads to immediate hypersensitivity.

Fig. 7. EE reversed the visceral hypersensitivity and depression-like behaviors via the PrL$\to$ avBNST pathway.

A Representative photograph of the experimental design. B Schematic diagram of pathway-specific caspase-mediated ablation strategy. C Visceral pain threshold was decreased after Casp3 treatment in EE rats (n = 6 rats). D AWR scores were increased after Casp3 treatment in EE rats (n = 6 rats). E, F The schematic diagram and heatmap of the EMG activity of the external oblique muscle. G EE group showed increased myoelectric activity in response to CRD at 40 and 50 mmHg after Casp3 treatment (n = 6 rats). H Immobility time in the FST increased after Casp3 treatment in EE rats (n = 6 rats). I However, there was no significant change in sucrose preference (n = 6 rats). EE rats showed decreased pain threshold J and sucrose preference (M), but increased AWR scores K and immobility time in the FST L after avBNST injection of WIN55,212-2 (n = 6 rats). Data are expressed as the mean ± SEM. *P < 0.05, **P < 0.01, as determined by two-tailed Student’s t-test (C, H, I, L, M), two-way ANOVA followed by Tukey’s post hoc test (J), and repeated-measures ANOVA (D, G, K).

We next investigated the involvement of CB1Rs in EE-mediated circuit modulation. Pharmacological activation of CB1Rs using WIN55,212-2 in EE-treated rats reinstated visceral hyperalgesia (Fig. 7J, K) and increased depressive-like behaviors (Fig. 7L, M). These findings indicated that EE restores normal neural circuit function by reversing CB1Rs overexpression in PrL→avBNST glutamatergic neurons, thereby alleviating both chronic visceral hypersensitivity and depression-like phenotypes.

Discussion

This study demonstrates that EE alleviates ELS-induced chronic visceral pain and depression-like behaviors through modulation of the PrL^Glu→avBNST^GABA→PVN circuit involving CB1Rs signaling. Using a well-established rat model of functional chronic visceral pain induced by neonatal CRD, we identified reduced excitability of PrL glutamatergic neurons and impaired PrL^Glu→avBNST circuit function as key contributors to ELS-induced visceral hypersensitivity and depression-like phenotypes. EE intervention effectively reversed these pathological alterations, likely through restoration of PrL glutamatergic neuronal activity. At the molecular level, EE normalized the aberrant upregulation of CB1Rs in avBNST-projecting PrL glutamatergic neurons of CRD rats. Collectively, these findings indicate that EE restores functional integrity of the PrL^Glu→avBNST^GABA→PVN circuit through normalization of CB1Rs signaling, thereby alleviating ELS-induced chronic visceral pain and depression-like behaviors.

Cognitive and emotional processes critically modulate pain perception. Patients with chronic visceral pain disorders, such as IBS, frequently exhibit comorbid psychopathologies including depression, anxiety, and somatization^48–50. Rodent models, particularly rats, provide well-validated translational systems for investigating IBS-related pathophysiology. Early-life noxious CRD induces long-lasting alterations in visceral sensitivity, nociceptive processing, and affective behaviors. The neonatal CRD model used in our laboratory demonstrates high stability and reproducibility, with a success rate of approximately 95% in producing persistent visceral hypersensitivity¹⁷. Our preliminary observations indicate that this potent stressor induces robust and qualitatively similar behavioral phenotypes and circuit-level alterations in both sexes. This is consistent with previous literature suggesting that physical stressors during the neonatal period—a critical window for programming stress-responsive circuits—can produce significant long-term outcomes regardless of sex^51,52. Current pharmacological interventions often demonstrate limited efficacy in symptom management, likely due to shared neural circuitry between visceral pain and emotional disturbances⁵³. This therapeutic challenge has spurred interest in non-pharmacological approaches. Notably, centrally targeted interventions like CBT and physical therapy have shown both acute and long-term benefits for visceral pain and its neuropsychological comorbidities⁵³. EE serves as a preclinical analog to CBT, providing a controlled paradigm to investigate how environmental factors modulate disease progression, particularly during critical developmental windows. Our findings demonstrated that EE effectively alleviated neonatal CRD-induced visceral hypersensitivity and depression-like behaviors in adulthood (Fig. 6B–H), which is consistent with EE’s beneficial effects in maternal separation (MS) model⁵⁴. Importantly, we identified specific prefrontal-limbic circuit mechanisms (the PrL→avBNST→PVN pathway) and molecular targets (CB1Rs) underlying these therapeutic effects. These discoveries provide a foundation for developing combined therapeutic strategies integrating EE principles with circuit-specific interventions.

The BNST represents a critical limbic hub involved in stress integration and affective regulation. Our previous studies demonstrated that the avBNST^GABA→PVN^CRH circuit mediates visceral hypersensitivity in maternal separation models^17–19. Although the avBNST contains a relatively small population of glutamatergic neurons that may contribute to aversive processing, our present findings, together with previous circuit-level evidence¹⁷, indicate that GABAergic neurons represent the dominant regulatory component of the PrL-avBNST-PVN axis. Consequently, our pharmacological results primarily capture the predominant influence of these inhibitory neurons on visceral pain and mood, even if a minor supporting role for local glutamatergic signaling cannot be entirely ruled out. This interpretation is further supported by the fact that GABAergic neurons constitute approximately 80–90% of neuronal populations within the avBNST, making them the most likely substrate mediating behavioral recovery observed following EE intervention.

In the current study, we extended these findings by investigating how BNST GABAergic neuron excitability contributes to the comorbidity of visceral pain and depressive-like behaviors in neonatal CRD-exposed rats. Pharmacological evidence suggested that the spontaneous glutamate release onto BNST GABAergic neurons was reduced, resulting in decreased synaptic drive and overall hypoexcitability of this population in CRD rats (Fig. 1J–Q). These observations align with the established framework of stress-induced visceral hyperalgesia, which involves dysregulation of prefrontal cortico-limbic circuits. In addition, although the PrL and IL cortices exhibit distinct projection patterns and functional roles^55,56, our anterograde and retrograde tracing data demonstrated that avBNST receives predominant direct innervation from PrL glutamatergic neurons (Fig. 2A–C). Specifically, we found that PrL glutamatergic neurons exhibited marked hypoactivity in CRD rats, as evidenced by reduced c-Fos expression and decreased firing frequency. These results were in line with previous studies showing that the glutamatergic neuronal activity of PrL was reduced in several animal models of chronic pain and pain-related behavior^25,57–59. The molecular mechanisms underlying reduced PrL excitability, therefore, warrant further investigation.

The endocannabinoid system serves as a promising therapeutic target for IBS, given its crucial role in regulating both visceral pain and stress response, mediated principally by widely distributed CB1Rs throughout the central nervous system²⁷. As key regulators of neuronal excitability and synaptic plasticity^60–62, CB1Rs participate in stress-related behaviors by providing autoregulatory feedback that prevents glutamate-mediated postsynaptic overactivation³⁰. While previous work localized CB1Rs on excitatory afferents from the IL to the anteromedial BNST⁶³, our study revealed an additional population of CB1Rs expressed on PrL-originating glutamatergic terminals in the avBNST. Notably, these receptors were upregulated in rats with neonatal CRD-induced chronic visceral pain and depression-like behaviors. This finding aligns with Grueter et al.’s demonstration that activation of group 1 metabotropic glutamate receptors (mGluRs) in BNST inhibits excitatory transmission via a CB1Rs-dependent mechanism, as evidenced by the complete abolition of this effect in CB1Rs knockout mice^30,64. It is important to note that the alterations in CB1R and GAD signal intensity observed in our Western blot assays likely reflect acute, functional modulations of protein states rather than shifts in total cellular abundance. Given the 90 min interval between pharmacological treatment and tissue collection—a timeframe insufficient for significant de novo protein synthesis or degradation—we interpret these findings as a biochemical “snapshot” of the circuit’s rapid response to ligands. Methodological evidence suggests that Western blot band density represents immunoreactivity, which can be influenced by protein conformation, phosphorylation-induced epitope masking, and subcellular redistribution. Specifically, the apparent shifts in CB1R signal following agonist (WIN 55,212-2) or antagonist (AM251) treatment are consistent with the rapid kinetics of G-protein coupled receptor (GPCR) trafficking, where agonist binding triggers phosphorylation and β-arrestin recruitment, leading to receptor sequestration into endosomes within minutes. Such internalization can alter the mask antibody epitopes, resulting in measurable changes in detectable protein signals. Similarly, GAD enzymes are subject to rapid regulation via phosphorylation, which modifies their conformation and membrane association, thereby interfering with antibody recognition. Here, when shRNA was used to knock down CB1Rs in PrL glutamatergic neurons of CRD rats, chronic visceral pain and depressive-like behaviors were significantly relieved (Fig. 4I–L). In line with this, pharmacological interventions further identified that reversing the dysfunction of CB1Rs in avBNST mitigated visceral hypersensitivity and depressive-like behaviors (Fig. 4M–X). These results suggest that enhanced CB1Rs signaling reduces glutamatergic drive within the PrL^Glu→avBNST pathway, thereby contributing to circuit hypoactivity and behavioral dysfunction.

To further ensure anatomical specificity, we considered the functional roles of adjacent structures. The functional effects of CB1R manipulation in the avBNST were distinct and often opposite to those reported in adjacent regions, providing strong evidence for site-specificity. The nucleus accumbens (NAc) and ventral pallidum (VP) are primary hubs for hedonic “liking” and motivational “wanting”. For example, the endocannabinoid 2-arachidonoyl glycerol (2-AG) in the limbic forebrain—including the NAc shell—is dynamically regulated by metabolic state, with increased levels promoting feeding behavior, supporting the role of CB1 receptor signaling in enhancing the incentive salience of natural rewards⁶⁵. In contrast, our micro-infusion of ACEA/WIN55,212-2 into the avBNST induced visceral hypersensitivity and depression-like behaviors. If our drug had diffused significantly into the NAc, we would have expected to see a masking of the depressive phenotype or an increase in pain thresholds, which is the opposite of our observed results. While the VP is involved in “liking” and hedonic impact⁶⁶, the avBNST is uniquely positioned as a “functional relay node” for the PrL-PVN axis. Our specific finding—that AM251 in the avBNST modulates visceral pain but not basal mood in naive rats—aligns with the specialized role of the BNST in sensory-autonomic gating, a feature not typically attributed to the VP or NAc in the same manner. Most importantly, our optogenetic and chemogenetic manipulations (which are genetically restricted to the PrL→avBNST pathway) yielded behavioral phenotypes identical to our pharmacological results in neonatal CRD model. This cross-validation confirms that the observed effects are driven by the specific neural architecture of the avBNST and not by non-specific drug diffusion into neighboring structures.

Early and continuous EE paradigm was implemented to mitigate cellular dysfunction, neurocircuitry impairments, and behavioral deficits in the rat model of neonatal CRD-induced chronic visceral pain and depression-like behaviors. Electrophysiological analyses revealed that EE significantly enhanced the excitability of glutamatergic neurons in PrL of CRD-exposed rats, suggesting a potential mechanism for therapeutic effects. Notably, previous clinical studies have demonstrated that schizophrenia patients exhibit a marked reduction in CB1Rs expression in the dorsolateral PFC (Brodmann area 9) compared to both healthy controls and individuals with major depressive disorder (MDD)⁶⁷. In alignment with these findings, our study identified that the hypoexcitability of PrL^Glu→avBNST^GABA neural circuit—a key pathway implicated in pain and affective regulation—was mediated by an upregulation of CB1Rs on glutamatergic neurons. Importantly, CB1Rs expression was already significantly elevated at P21 and remained significantly higher at P49 (Fig. S8). This temporal pattern demonstrates that the molecular “seeds” of circuit dysfunction are firmly sown immediately following the neonatal insult, with the cannabinoid system exhibiting significant dysregulation as early as weaning. This early dysregulation consolidates further during adolescence, a period of intensive circuit refinement. Our finding that CB1Rs levels are already significantly elevated at P21 provides a clear and compelling rationale for the efficacy of EE. By introducing EE at this early stage, we intercept this pathological trajectory before it undergoes further reinforcement during adolescence. Strikingly, EE intervention normalized CB1Rs levels, thereby restoring functional activity within the PrL^Glu→avBNST^GABA circuit (Fig. 8). This restoration was associated with a significant alleviation of chronic visceral pain and depression-like behaviors in CRD-exposed rats, highlighting EE as a developmentally sensitive intervention capable of reshaping maladaptive circuit maturation.

Fig. 8. Schematic model of the proposed mechanism.

In neonatal CRD-induced visceral pain and depression, PrL^Glu neurons and PrL^Glu→avBNST pathway are hypoactive, attributed to upregulation of CB1 receptors, leading to reduced excitatory drive onto avBNST GABAergic and subsequent disinhibition of PVN CRH neurons. Encouragingly, enriched environment effectively reversed the increase in CB1 receptors and restored activity of the PrL^Glu→avBNST pathway. This provides a circuit-level mechanistic framework for the comorbid visceral pain and depression-like behaviors. Comprehensive strategies combining behavioral/cognitive interventions with precise molecular targeting provide a robust theoretical foundation and broad therapeutic prospect.

Conclusions

In conclusion, this study not only elucidates the critical role of prefrontal-limbic circuitry in ELS-induced comorbid visceral pain and depression but importantly proposes a novel multimodal therapeutic strategy combining EE with CB1Rs-targeted pharmacological interventions. This integrated approach, by simultaneously modulating synaptic plasticity in the PrL^Glu→avBNST^GABA→PVN pathway and CB1Rs-mediated neurotransmission, offers a more comprehensive and sustainable solution for functional visceral pain and its psychiatric comorbidities. Future studies should further explore the synergistic effects between EE and other neuromodulatory approaches to develop personalized combination therapies.

Material and methods

Animals

The care and use of animals and the experimental protocols (202201A026) used in this study were approved by the Institutional Animal Care and Use Committee and the Office of Laboratory Animal Resources of Xuzhou Medical University. We have complied with all relevant ethical regulations for animal use. All experimental procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 2011). Pregnant Sprague-Dawley (SD) rats (8–12 weeks, 270–310 g) were provided by the Laboratory Animal Center of Xuzhou Medical University (SYXK (苏) 2026-0017). Rats were randomly assigned to plastic cages with nesting material and maintained on a 12/12 h light-dark cycle at 23 ± 2 °C with ad libitum access to food and water.

We employed neonatal CRD to establish an ELS-induced visceral hypersensitivity model in rats^18,44. In line with current guidelines advocating for sex-balanced research, both male and female rats were included to ensure translational relevance. In brief, rats underwent CRD on postnatal days 8, 10, and 12 by inserting an angioplasty balloon (length, 20.0 mm; diameter, 3.0 mm) through the rectum into the descending colon. The balloon was distended with 0.3 mL water at a pressure of 60 mmHg (~8 kPa) for 1 min prior to deflation and withdrawal. The distention was repeated twice a day at an interval of at least 30 min. Male and female pups were weaned at postnatal day 21 (P21), group-housed in standard or enriched cages, and left undisturbed. Follow-up experiments were conducted at P60.

To investigate whether EE could ameliorate or even reverse detrimental effects of neonatal CRD, the experiment consisted of three groups: naïve controls, rats exposed to neonatal CRD only (CRD group), and rats exposed to neonatal CRD followed by EE housing (CRD + EE group). Neonatal SD rats were weaned on P21 and subsequently housed in standard or enriched environments as assigned. CRD group was housed in a standard cage, and EE group was housed in an enriched environment. The standard environment consisted of a 44 × 27 × 19 cm transparent polypropylene cage containing only a nestlet. Enrichment consisted of a larger cage (54 × 50 × 40 cm) containing a running wheel, trapeze, plastic tubing, multiple houses, and nestlets. EE rats were group-housed (3–5/cage) with rotating enrichment items to maintain novelty. The bedding was changed every week. All rats were kept in the same room under standard conditions.

Visceral pain threshold tests and Abdominal withdrawal reflex (AWR) scores

In adulthood, visceral hypersensitivity was quantitatively evaluated through CRD-induced pain responses using an established protocol^17,18,44. Graded CRD stimuli (10, 20, 30, 40, 50 mmHg) were applied sequentially, with each pressure maintained for 10 s followed by a 4 min recovery interval to prevent sensitization. AWR responses were scored according to the following standardized criteria: 0, absence of significant behavioral response; 1, immobility or only brief head movement; 2, contraction of the abdominal muscles without lifting off the table; 3, obvious contraction of the abdominal muscles with lifting off the table; and 4, arcing of the abdominal wall and trunk. The visceral pain threshold was defined as the first pressure eliciting an AWR score of 3 (obvious contraction of the abdominal wall), representing a clinically relevant nociceptive response. Each pressure level was tested in triplicate with randomized presentation order, and the average response across trials was calculated for analysis.

Electromyography (EMG) recording

EMG responses to graded CRD (10, 20, 30, 40, and 50 mmHg) were quantitatively assessed under light anesthesia (isoflurane, 0.5–1.5%). As described by our previous study⁴⁴, the EMG signals were recorded with a BL-420N system from Chengdu Taimeng Co., Ltd. (Chengdu, China). The detailed recording parameters were as follows: high frequency filter, 2 kHz; time constant, 0.01 s; sampling frequency, 2000 Hz; sensitivity, 500 μV; and scanning speed, 250 ms/div. EMG activity was continuously recorded during both the 10 s distension period and subsequent 4-min recovery phase. The rectified EMG signal was quantified as area under the curve (AUC) during the 10 s distension period.

Forced swimming test (FST)

FST, a well-validated paradigm for assessing behavioral despair in rodents, was conducted using a standardized two-session protocol. Rats were placed individually in a transparent Plexiglas inescapable Plexiglas cylinder (height: 45 cm; diameter: 20 cm) containing 30 cm of water maintained at 25 ± 1 °C. The procedure included an initial 15-min pre-test session on day 1 to establish baseline behavior, followed by a 6-min test session conducted 24 h later⁶⁸. During the test session, immobility time was quantified as the primary behavioral endpoint, operationally defined as passive floating with only minimal limb movements necessary to maintain the head above water. This immobility response has been extensively characterized as reflecting behavioral despair, passive stress coping strategies, or learned helplessness in the literature. Water was changed between subjects to prevent olfactory cue interference.

Sucrose preference test (SPT)

SPT was performed to evaluate hedonic behavior in rats⁶⁹. Prior to testing, rats were individually housed and acclimated to a two-bottle choice paradigm. Following 24 h water deprivation, rats were presented with simultaneous access to both 1% sucrose and water in identical bottles for a 2 h test period. To control for side preference, bottle positions were systematically alternated after the first hour. Fluid consumption was measured gravimetrically, with sucrose preference percentage calculated as: [sucrose solution intake/(sucrose solution intake + water intake)] × 100%.

Stereotaxic surgeries and injections

Rats were anesthetized with isoflurane (induced concentration 4–5%, maintained at 1.5–2%) before undergoing stereotaxic surgery for viral injection and fiberoptic implantation. A dental drill was used to carefully remove the skull above the target region. Viral vector or drugs (0.5 μL per PrL/PVN; 0.2 μL per avBNST) were infused at 100 nL/min using a 10 μL Hamilton syringe connected to a glass microelectrode and controlled by a syringe pump. The CB1R antagonist AM251 (100 pmol in 0.2 μL, 0.5 μM), agonist WIN55,212-2 (100 pmol in 0.2 μL, 0.5 μM), and selective agonist ACEA (0.5 pmol in 0.2 μL, 2.5 μM) were bilaterally microinfused into the avBNST in a volume of 0.2 μL. After injection, the needle was left in place for 10 min to allow viral diffusion, then slowly withdrawn over 2 min to minimize backflow. For optogenetic experiments, bilateral fiberoptic cannulae were implanted above the target region and secured with dental cement. Stereotaxic coordinates (in mm) were defined as anterior-posterior (AP) from bregma, mediolateral (ML) from the midline, and dorsoventral (DV). We utilized the following stereotactic coordinates: PrL (AP: +3.2 mm; ML: ± 0.75 mm; DV: −4.0 mm), avBNST (AP: −0.3 mm; ML: ±1.25 mm; DV: −7.2 mm), and PVN (AP: -1.8 mm; ML: ±0.25 mm; DV: −8.0 mm). The viral expression was verified by immunostaining. Any animals showing significant viral spread into the IL (defined as expression extending beyond the ventral border of the PrL) or into the adjacent ACC were excluded from all behavioral and molecular analyses. The details of drugs and fluorescent tracing viral tools employed in this study are as follows (Table 1).

Table 1.

The details of drugs and fluorescent tracing viral tools

Reagent or resource	Source	Identifier
Antibodies
Rabbit anti c-Fos	CST	2250S
Guinea pig anti c-Fos	Synaptic systems	226 308
Rabbit anti-CaMKIIα	HuaBio	JM11-07
Mouse anti-CaMKIIα	CST	50049S
Rabbit anti-CB1R	Proteintech	17978-1-AP
Rabbit anti-GABA	Millipore	A2052
Rabbit anti-GAD67	Proteintech	10408-1-AP
Rabbit anti-GAD65	Proteintech	20746-1-AP
Donkey anti-rabbit (Alexa Fluor 594)	Invitrogen	A21207
Goat anti-guinea pig (Alexa Fluor 594)	Invitrogen	A11076
Donkey anti-rabbit (Alexa Fluor 405)	Invitrogen	WJ338368
Donkey anti-mouse (Alexa Fluor 488)	Invitrogen	A21202
Drugs
NMDA	MedChemExpress	HY-17551
AMPA	MedChemExpress	HY-17551
CNQX	MedChemExpress	HY-15066
4-AP	MedChemExpress	HY-B0604
AM251	Tocris	1117/1
WIN55.212-2	Tocris	1038
ACEA	MCE	HY-110004
CNO	MedChemExpress	HY-17366
Bacterial and virus strains
rAAV-Vgat1-EGFP-WPRE-hGH polyA	BrainVTA	PT-3176
AAV2/9-CaMKIIα-mCherry-WPRE-Pa	BrainVTA	PT-0108
AAV2/9-CaMKIIα-EGFP-WPRE-Pa	BrainVTA	PT-0290
AAV2/9-Vgat1-Cre-WPRE-Pa	BrainVTA	PT-0346
rAAV-CR-Cre-WPREs	BrainVTA	PT-1771
rAAV-fSST-CRE-WPRE-hGH-Pa	BrainVTA	PT-0279
rAAV-EF1α-DIO-EGFP-T2A-TVA-WPRE-hGH pA	BrainVTA	PT-0062
rAAV-EF1α-DIO-RVG-WPRE-pA	BrainVTA	PT-0223
RV-EnvA-ΔG-DsRed	BrainVTA	R01002
AAV2/9-CaMKIIα-Cre-WPRE-Pa	BrainVTA	PT-0220
rAAV-CAG-DIO-WGA-FLP-WPRE-hGH-pA	BrainVTA	PT-0557
AAV2/9-Ef1α-DIO-hM3D(Gq)-mCherry-WPRE-Pa	BrainVTA	PT-0042
AAV2/9-Ef1α-DIO-hM4D(Gi)-mCherry-WPRE-Pa	BrainVTA	PT-0017
rAAV-Ef1α-DIO-hChR2(H134R)-mCherry-WPRE-Pa	BrainVTA	PT-0002
rAAV-Ef1α-DIO-eNpHR3.0-EYFP-WPRE-Pa	BrainVTA	PT-0006
rAAV-EF1α-DIO-mcherry-WPRE-hGH-pA	BrainVTA	PT-0013
rAAV-EF1α-DIO-EGFP-WPRE-hGH pA	BrainVTA	PT-0795
AAV1-hSyn-CRE-WPRE-hGH pA	BrainVTA	PT-0136
AAV2-Retro-EF1α-DIO-FLP-WPRE-hGH pA	BrainVTA	PT-0075
rAAV-nEF1a-fDIO-mCherry-WPRE-PA	BrainVTA	PT-0339
rAAV-EF1α-fDIO-hM4D(Gi)-mCherry-WPREs	BrainVTA	PT-0170
AAV2 Retro/ScAAV-hSyn-EGFP-WPREs	BrainVTA	PT-2315
rAAV-EF1α-fDIO-hM3D(Gq)-mCherry-WPREs	BrainVTA	PT-4809
AAV2-Retro-CaMKIIα-EGFP-WPRE-hGH pA	BrainVTA	PT-0290
rAAV-SST-mCherry-WPRE-bGH pA	BrainVTA	PT-1215
rAAV-flex-taCasp3-TEVp-WPRE-Pa	BrainVTA	PT-0206
rAAV-CR-BFP-WPRE-Pa	BrainVTA	PT-2270
rAAV-nEF1α-fDIO-EGFP-WPRE-hGH pA	BrainVTA	PT-0435
rAAV-CaMKIIa-eNpHR3.0-mCherry-WPRE-hGH pA	BrainVTA	PT-0009
rAAV-EF1α-DIO-FLP-WPRE-hGH Pa	BrainVTA	PT-0075
pAAV-CaMKIIα-mCherry-3xFLAG-miR30s-hRNA(Cnr1)-WPRE	OBIO	RC1203
pAAV-CaMKIIα-mCherry-3xFLAG-miR30s-hRNA(NC2)-WPRE	OBIO	OP0913

In vitro electrophysiology

Rats were anesthetized with isoflurane and transcardially perfused with 20 mL ice-cold cutting solution (containing, in mM): 80 NaCl, 3.5 KCl, 0.5 CaCl₂, 4.5 MgSO₄, 1.25 NaH₂PO₄, 90 sucrose, 25 NaHCO₃, and 10 glucose. After decapitation, the brain was transferred into ice-cold oxygenated cutting solution. Coronal slices (300 μm) containing PrL, avBNST and PVN were cut using a Leica VT1200S vibrating microtome in oxygenated artificial cerebrospinal fluid (ACSF) (in mM) at 32 °C: 126 NaCl, 2.5 KCl, 1.2 NaH₂PO4, 1.2 MgSO₄, 26 NaHCO3, 10 glucose, and 2.4 CaCl₂. ACSF was continuously saturated with 95% O₂ and 5% CO₂. Slices were incubated for at least 1 h before transfer to a recording chamber perfused with ACSF at room temperature (RT).

Whole-cell patch clamp recordings were performed in the PrL glutamatergic or avBNST GABAergic neurons with standard internal solution containing (in mM) 10 phosphocreatine-Tris, 2 ATP-Mg, 10 HEPES, 10 EGTA, 0.5 GTP-Na, 1.5 MgCl₂, 20 KCl, and 115 potassium gluconate (pH = 7.25). The patch pipettes had a resistance of 5–10 MΩ. PVN CRH neurons, PrL glutamatergic, or avBNST GABAergic neurons were targeted for recording. Spontaneous discharge (firing frequency) was recorded in current-clamp mode (I = 0) for a minimum of 5 min to ensure baseline stability. Additionally, evoked action potentials (APs) were recorded in response to depolarizing step current injections (0–140 pA, 20 pA increments). The somata and terminals of PrL ChR2-expressing neurons were photostimulated using 473 nm blue laser pulses. For somatic stimuli, slices were photostimulated at 2, 5, 10, and 20 Hz. For the terminal stimuli, tetrodotoxin (TTX, 1 μM), 4-AP (50 μM), and CNQX (20 μM)⁷⁰ were utilized for pharmacological characterization of the synaptic driven avBNST bursting evoked by 10 Hz optostimulation. For eNpHR experiments, 589 nm laser (a continuous 800-ms light pulse) was applied.

Western blotting analysis

Protein expression was evaluated via Western blotting. Briefly, total protein was extracted using ice-cold RIPA buffer supplemented with protease inhibitors and quantified via a BCA assay kit. Equal protein amounts were resolved by SDS-PAGE and subsequently immobilized onto PVDF membranes. After blocking with 5% non-fat milk for 2 h at RT, membranes were probed overnight at 4 °C with primary antibodies: rabbit anti-CB1Rs (1:1000, Proteintech, #17978-1-AP), rabbit anti-GAD67 (1:1000, Proteintech, #10408-1-AP), and rabbit anti-GAD65 (1:1000, Proteintech, #20746-1-AP). Following incubation with appropriate HRP-conjugated secondary antibodies (1:2000), immunoreactive bands were visualized using an enhanced chemiluminescence (ECL) kit (Millipore, USA) and imaged with the Image-Pro Plus analysis system (Media Cybernetics, USA). Densitometric analysis was performed using ImageJ software (NIH, USA), and target protein expression was normalized to GAPDH as the internal loading control. The full, uncropped Western blot images are provided in Supplementary Fig. 9.

Immunofluorescence

Rats were deeply anesthetized with isoflurane and transcardially perfused sequentially with 0.9% saline followed by ice-cold 4% (w/v) paraformaldehyde (PFA; 100 ml/100 g). The brains were immediately dissected and post-fixed in 4% PFA overnight at 4 °C, followed by immersion in 30% sucrose at 4 °C until saturation. The brain tissues were coronally serially sectioned (30 μm) in a cryostat (CM1800, Leica, Germany) and used for immunofluorescence. The sections were blocked with 10% donkey serum for 2 h at RT, and incubated with the primary antibodies at 4 °C for 24 h, followed by the corresponding secondary antibodies (Invitrogen, USA) for 2 h at RT. DAPI (KGI Biological, China) was utilized for nuclear staining. Antibody details are provided in Table 1. Imaging was performed using a confocal microscope (FV1000, Olympus, Tokyo, Japan; LSM880, Zeiss, Germany).

Fluorescent in situ hybridization (FISH)

Fluorescently labeled RNA probes targeting CaMKIIα (Cy3-conjugated) and CB1Rs (FITC-conjugated) were custom-designed and synthesized by GenePharma (Shanghai, China). Hybridization was performed overnight with Cy3-labeled CaMKIIα or FITC-labeled CB1R probes. After PBS rinses, slides were counterstained with DAPI and mounted using antifade mounting medium. The fluorescence was visualized and captured by a Leica confocal laser scanning microscope system.

Optogenetic manipulations in vivo and in vitro

For in vivo optical stimulation, the optic fibers were bilaterally implanted 200 μm above the PrL (for somatic stimulation) or avBNST (for terminal stimulation). The delivery of a 5–10 min pulse of blue light (473 nm laser, 10 Hz, 10 ms pulse width, 7–10 mW for somata, 15–20 mW for terminals; BL473T3-100FC, Lasercentury, China) or yellow (589 nm, 5–8 mW, 8 s on/2 s off; YL589T6-100FC, Lasercentury, China) light pulses were delivered via a stimulator (QAXK-TTL-MC, Thinkertech, China). Cables were shielded with aluminum foil to prevent light leakage.

Chemogenetic manipulation

The DREADDs (designer receptors exclusively activated by designer drugs) approach was implemented in rats aged 8–10 weeks. Neural circuits were selectively modulated using AAV-mediated expression of hM3Dq (Gq-coupled) for neural activation and hM4Di (Gi-coupled) for inhibition. During behavioral testing, rats received intraperitoneal injections of clozapine-N-oxide (CNO, 3 mg/kg) 30 min prior to habituation, allowing optimal time for DREADD receptor activation. Behavioral measurements, including visceral pain threshold, AWR scores, EMG, SPT, and FST, were systematically assessed 30 min after injection. Viral expression was confirmed histologically via mCherry reporter fluorescence following behavioral testing.

Statistics and reproducibility

Data are presented as the mean ± SEM. Statistical analyses were performed with Graphpad Prism 8.3.0 software package. Electrophysiological data were analyzed using Clampfit 10.7. Statistical details can be found in the figure legends. Differences between two groups were evaluated using two-tailed unpaired Student’s t-test. Datasets among multiple groups were analyzed by one-way analysis of variance (ANOVA). Datasets with two independent variables were analyzed by two-way ANOVA. Two-way repeated measures ANOVA was used to detect significant interactions between group and treatment. If a significant difference was found, Tukey’s multiple comparisons post hoc test was performed. Normality was assessed using the Shapiro–Wilk test. Homogeneity of variances was confirmed by Brown–Forsythe test. P < 0.05 was considered statistically significant (*P < 0.05 and **P < 0.01). All experiments were conducted in at least three independent cohorts, and the specific number of replicates is indicated in the figure legends. Figures were generated using Adobe Illustrator 2019.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization: Yong-Mei Zhang. Data curation: Si-Ting Huang, Ning-Ning Ji. Formal analysis: Yong-Mei Zhang, Si-Ting Huang, Ning-Ning Ji. Funding acquisition: Yong-Mei Zhang, Ning-Ning Ji, Si-Ting Huang. Methodology and Investigation: Ying Wang, Ning-Ning Ji, Si-Ting Huang, Yi-Wen Niu, Jian-Hua Xi, Long Huang, Zi-Ming Zhou. Supervision: Yong-Mei Zhang. Writing-original draft: Si-Ting Huang. Writing-review & editing: Yong-Mei Zhang, Si-Ting Huang

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Alban Latremoliere and Joao Manuel de Sousa Valente.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information file. Source data for the main figures are provided as Supplementary Data. Additional raw data are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10105-2.