CRF1-R activation of the dynorphin/kappa opioid system in the mouse basolateral amygdala mediates anxiety-like behavior

Abstract

Stress is a complex human experience and having both rewarding and aversive motivational properties. The adverse effects of stress are well documented, yet many of underlying mechanisms remain unclear and controversial. Here we report that the anxiogenic properties of stress are encoded by the endogenous opioid peptide dynorphin acting in the basolateral amygdala. Using pharmacological and genetic approaches, we found that the anxiogenic-like effects of Corticotropin Releasing Factor (CRF) were triggered by CRF(1)-R activation of the dynorphin/kappa opioid receptor (KOR) system. Central CRF administration significantly reduced the percent open-arm time in the elevated plus maze (EPM). The reduction in open-arm time was blocked by pretreatment with the KOR antagonist norbinaltorphimine (norBNI), and was not evident in mice lacking the endogenous KOR ligand dynorphin. The CRF(1)-R agonist stressin 1 also significantly reduced open-arm time in the EPM, and this decrease was blocked by norBNI. In contrast, the selective CRF(2)-R agonist urocortin III did not affect open arm time, and mice lacking CRF(2)-R still showed an increase in anxiety-like behavior in response to CRF injection. However, CRF(2)-R knockout animals did not develop CRF conditioned place aversion, suggesting that CRF(1)-R activation may mediate anxiety and CRF(2)-R may encode aversion. Using a phosphoselective antibody (KORp) to identify sites of dynorphin action, we found that CRF increased KORp-immunoreactivity in the basolateral amygdala (BLA) of wildtype, but not in mice pretreated with the selective CRF(1)-R antagonist, antalarmin. Consistent with the concept that acute stress or CRF injection-induced anxiety was mediated by dynorphin release in the BLA, local injection of norBNI blocked the stress or CRF-induced increase in anxiety-like behavior; whereas norBNI injection in a nearby thalamic nucleus did not. The intersection of stress-induced CRF and the dynorphin/KOR system in the BLA was surprising, and these results suggest that CRF and dynorphin/KOR systems may coordinate stress-induced anxiety behaviors and aversive behaviors via different mechanisms.

Figure 1. CRF-induced anxiety-like behavior is mediated by the dynorphin/KOR system.

A, CRF injection (1 µg, i.c.v., 30 min prior to assay) produced a significant anxiety-like effect (decrease in % open arm time) in the elevated plus maze (EPM) compared with saline injected controls. The effect of CRF was blocked by pretreatment with the KOR antagonist norBNI (10 g/kg, i.p., 2.5 hrs prior to test) (n = 7–8; two-way ANOVA; main effect of CRF, F_(1,24) = 6.323, p<0.05; interaction of CRF and pretreatment, F_(1,24) = 7.897 p<0.05; ** p<0.01 Bonferroni post hoc)B, CRF (1 µg, i.c.v., 30 min prior to assay) produced a significant anxiety-like effect in wildtype littermate control Dyn(+/+) mice, but had no effect in prodynorphin knockout animals (Dyn −/−) (n = 4–8 per group, p<0.01 CRF in Dyn +/+ vs Dyn −/−). CRF also caused a significant increase in open arm time in Dyn (−/−) animals (p<0.01, t-test, open arm vs. closed arm time, data not shown) C, Activation of KOR by the selective KOR agonist U50,488 (U50, 5 mg/kg, i.p.) was also sufficient to cause significant anxiety-like behavioral effects, whereas treatment with norBNI alone had no effect on the percent open arm or closed arm time in the EPM under these conditions. (n = 8, one-way ANOVA, **p<0.01, U50 vs. saline).

Figure 2. CRF-induced anxiety-like behavior is mediated by CRF₁-R but not CRF₂-R activation.

A, Administration of stressin 1 (0.5 µg, i.c.v.), a selective CRF₁-R agonist, produced significant anxiety-like behavioral effects. Pretreatment with the KOR antagonist norBNI (10 mg/kg, i.p., 2.5 hrs prior to test) blocked the stressin1-induced (Stsn) decrease in percent open arm time (n = 6–8, one-way ANOVA, F_{(3, 20)} = 4.658; **p<0.05, Saline vs. Stressin 1). In contrast, the EPM scores for wildtype mice injected with the CRF₂-R selective agonist urocortin III (Uro3) (0.5 µg, i.c.v.), were not significantly different from the saline-treated group (n = 6–8). B, Consistent with the above results, CRF administration to CRF₂-R (−/−) mice still produced significant anxiety-like behavior (n =  4–8, two-way ANOVA; main effect of CRF, F_{(1, 21)} = 19.02, p<0.001; no main effect of genotype, F_(1,22) = 0.26, p>0.05; Bonferroni post-hoc, ** p<0.01 saline (−/−) vs. CRF (−/−)).

Figure 3. CRF-induced conditioned place aversion is CRF₂-R dependent.

A, Schematic of the conditioning procedure used to assay aversion. B, Conditioned place aversion expressed as the individual animal pre-test (squares) connected by line to the same animal's post-test (circles) times spent in the CRF-paired compartment. Dashed lines represent the mean times during either the pre-test or post-test for each group. There was no significant difference between CRF₂-R (+/+) and CRF₂-R(−/−) mice in pretest scores. CRF (1 µg, i.c.v.) induced a significant place aversion in CRF₂-R +/+ mice (post-test minus pre-test in the drug paired side, seconds) of -129±43 sec. In contrast, CRF injected in CRF₂-R(−/−) mice induced a significant place preference (p<0.05) of 235±68 sec (n = 5–7, p<0.05 pre-test minus post-test for both groups).

Figure 4. CRF-induces CRF₁-R-dependent dynorphin/KOR activation in the BLA.

A, CRF (1 µg, i.c.v., 30 min) administration in wildtype mice increased in phospho-KOR-ir in the BLA (upper row, second panel). Pretreatment of wild type mice with antalarmin (10 mg/kg, i.p., injected 1hr before CRF) blocked the increased CRF-induced KORp-ir in the BLA (upper row, right panel). Similarly, KORp-ir did not qualitatively increase in CRF₁-R (−/−) mice following CRF challenge (panel 3, top row, n = 2). As expected, mice injected with saline (vehicle, i.c.v.) showed low levels of KORp-ir in the BLA (top left). B , Local infusion of norBNI into the BLA of wildtype animals reduced the CRF-induced KORp-ir in the BLA (far right panel) but had no effect on CRF-induced KORp-ir the dorsal medial hypothalamus (DMH), a nearby brain structure with comparable levels KORp-ir in the untreated animal. This control region allowed for a regional confirmation of the sphere of norBNI blockade of KOR in injected animals. As predicted norBNI into the BLA had no effect on KORp-ir in the DMH. C, Schematic of the area imaged and injected with local norBNI; the right box outlines the region imaged including the BLA and the left box outlines the dorsal medial hypothalamus (DMH), the closest adjacent brain structure in the same slice where there is high expression of KOR and KORp-ir was visualized following CRF-injection. D, Quantification (per 600 µm²) of BLA KORp-ir following saline or CRF-injection. CRF induced a significant increase in KORp-ir as compared to saline, or antalarmin-treated groups. In animals locally injected with norBNI into the BLA, CRF-induced KORp-ir cell staining was not significantly different than saline treated mice. (*p<0.05, CRF vs. saline, †† p<0.01 CRF vs. CRF + antalarmin). Data are from 2–4 independent experiments.

Figure 5. Stress and CRF-induced anxiety is mediated by dynorphin/KOR in the BLA.

A, Mice injected with saline (1 µl) into the BLA showed a decrease in percent open arm time behavior in the EPM following subsequent CRF (i.c.v) injection, whereas bilateral infusion of norBNI (2.5 µg/side, see methods) in the BLA significantly blocked the CRF-induced (1 µg, i.c.v.) decrease in percent open arm response. Control injection of norBNI into a nearby brain structure, the ventral posteromedial thalamic nucleus (VPN), had no effect on the CRF-induced anxiety-like behavior evident in the EPM. (*p<0.05, CRF-injected norBNI/BLA vs. CRF-injected saline/BLA; † p<0.05, CRF-injected norBNI/BLA vs. CRF-injected norBNI/VPN; n = 6–8 per group). B, In further support of a CRF-dependent dynorphin/KOR interaction in the BLA, mice were exposed to a single acute swim stress (5 min), placed back in their home cage for 30 min and tested in the EPM. As expected, swim stress induced a significant decrease in the percent time in the open arm as compared to locally injected non-stressed control animals. The anxiogenic-like response to acute stress was significantly blocked in mice injected with norBNI in the BLA (Fig. 5B ) (n = 5–6, F_{(1, 16)} = two-way ANOVA; main effect of stress, F _(1,16) = 23.54; main effect of treatment, F _(1,16) = 5.387; p<0.05, Bonferroni post-hoc).