Insular cortex corticotropin-releasing factor integrates stress signaling with social affective behavior

Abstract

Impairments in identifying and responding to the emotions of others manifest in a variety of psychopathologies. Therefore, elaborating the neurobiological mechanisms that underpin social responses to social emotions, or social affective behavior, is a translationally important goal. The insular cortex is consistently implicated in stress-related social and anxiety disorders, which are associated with diminished ability to make and use inferences about the emotions of others to guide behavior. We investigated how corticotropin-releasing factor (CRF), a neuromodulator evoked upon exposure to stressed conspecifics, influenced the insula. We hypothesized that social affective behavior requires CRF signaling in the insular cortex in order to detect stress in social interactions. In acute slices from male and female rats, CRF depolarized insular pyramidal neurons. In males, but not females, CRF suppressed presynaptic GABAergic inhibition leading to greater excitatory synaptic efficacy in a CRF receptor 1 (CRF₁)- and cannabinoid receptor 1 (CB₁)-dependent fashion. In males only, insular CRF increased social investigation, and CRF₁ and CB₁ antagonists interfered with social interactions with stressed conspecifics. To investigate the molecular and cellular basis for the effect of CRF we examined insular CRF₁ and CB₁ mRNAs and found greater total insula CRF₁ mRNA in females but greater CRF₁ and CB₁ mRNA colocalization in male insular cortex glutamatergic neurons that suggest complex, sex-specific organization of CRF and endocannabinoid systems. Together these results reveal a new mechanism by which stress and affect contribute to social affective behavior.

Fig. 1. CRF alters intrinsic properties of male and female insular cortex pyramidal neurons in whole-cell recordings.

A Representative single action potential (AP) recordings of deep layer insular cortex pyramidal neurons at baseline (aCSF-gray) and after application of 50 nM Corticotropin-releasing factor (CRF-blue). B CRF decreased the resting potential of male (n = 14) and female (n = 12) pyramidal neurons, F_CRF(1, 24) = 72.93, P < 0.0001, with this effect being stronger in males than females as indicated by a CRF × sex interaction, F_{CRF × SEX}(1, 24) = 5.124, P = 0.033. C Action potential rise rate was reduced by CRF in both males and females, F_CRF(1, 24) = 18.93, P = 0.0002. D Action potential half-width increased following CRF application in male and female recordings, F_CRF(1, 24) = 16.69, P = 0.0004. E CRF reduced the amplitude of the after depolarization (ADP) in both male and female recordings, F_CRF(1, 24) = 26.83, P < 0.0001. F CRF increased the current required to trigger burst firing in male and female neurons, F_CRF(1, 24) = 38.82, P < 0.0001. G Representative family of 1 s hyperpolarizing and depolarizing current injections used characterize passive membrane properties and spike rate in aCSF (gray) and after 50 nM CRF (blue). H Example steady-state current–voltage dependence plot. Input resistance was determined by linear fit and slope at 0pA and deviation from fit indicates rectification. I CRF reduced membrane input resistance in male and female neurons, F_CRF(1, 24) = 5.985, P = 0.022; this effect appeared most robustly in males. J CRF increased rectification of membrane potential in males and females, F_CRF(1, 24) = 35.28, P < 0.0001. K CRF did not alter firing rates in response to 1 s depolarizing current injections in either males or females. Bar graphs indicate mean with individual replicates, line graphs mean (±SEM). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (Sidak’s tests).

Fig. 2. CRF has dose- and sex-dependent synaptic effects in insular cortex slices.

A Representative traces of male (above) and female (below) field excitatory postsynaptic potentials (fEPSP) at 1, 3 and 5 V under aCSF (gray) and after 50 nM (left, blue) or 300 nM CRF (right, blue) conditions. For analysis, traces were normalized to the peak amplitude of the fEPSP evoked at 5 V in aCSF. B Bath application of 50 nM CRF significantly increased fEPSPs in male insular cortex slices in biphasic 0–5 V I/O curves F_{Voltage × CRF} (20, 80) = 5.791, P < 0.0001 with post hoc tests showing CRF being significantly increased over baseline at 3 V (P = 0.0425), 3.5 V (P < 0.0001), 4 V (P = 0.0001), 4.5 V (P = 0.0091) and 5 V (P = 0.0009). However, there was no significant effect of CRF on female slices F_{Voltage × CRF} (20, 80) = 0.5351, P < 0.5667. A three-way ANOVA revealed significant interactions between voltage and CRF and sex: F_{Voltage × CRF}(10, 80) = 3.654, P = 0.0005, F_{Voltage × Sex}(10, 80) = 2.910, P = 0.0037 as well as a main effect of sex, F_Sex(1, 8) = 10.53, P = 0.0118. C Bath application of 300 nM CRF led to a sex difference in fEPSPs such that males showed increased synaptic efficacy but not females resulting in a significant three-way interaction, F_{Voltage × Sex × CRF} (10, 100) = 5.306, P < 0.0001. Males showed significant increases in fEPSP under CRF conditions via Tukey’s multiple comparison tests at 2.5 V (P = 0.0370), 3 V (P = 0.0212), 3.5 V (P = 0.0124), 4 V (P = 0.0063), 4.5 V (0.0040) and 5 V (P = 0.0058). D Comparing 5 V responses (normalized to female aCSF 5 V) under 50 nM versus 300 nM CRF by sex revealed main effects of sex, F_Sex(1, 18) = 5.737, P = 0.0277, and CRF, F_CRF(1, 18) = 7.855, P = 0.0118. Sidak’s post hoc tests showed that there was a significant dose effect in males t(18) = 2.981, P = 0.0471, but not in females t(18) = 0.9826, P = 0.9165. E CRF₁ antagonist CP154526 (10 μm) coapplied with 300 nM CRF prevented CRF from increasing fEPSPs in slices from male rats. The dashed gray line depicts the effect of 300 nM CRF alone for comparison. While there was a significant interaction F_{Voltage × CRF}(20, 89) = 3.276, P < 0.0001. There was no main effect of CRF, F_CRF(2, 10) = 0.0362, P = 0.9646. No post hoc comparisons were significant across treatments at different voltages. F Coapplication of the GABA_A antagonist SR95531 prevented the enhancing effect of 300 nM CRF and led to a significant decrease in fEPSPs F_{Voltage × CRF}(10, 40) = 3.464, P = 0.0024 in slices from male rats. Significant Tukey’s post hoc comparisons were found at 2.5 V (P = 0.0422), 3 V (P = 0.0144), 3.5 V (P = 0.0038), 4 V (P = 0.0026), 4.5 V (P = 0.0014), and 5 V (P = 0.0116). G 5 V fEPSPs after CRF, CRF + CP154526 and CRF + SR95531 (300 nM) were normalized to the relative 5 V fEPSP in aCSF in (D) to summarize the effect of CRF₁ and GABA_A receptor antagonist on fEPSP. Both CRF₁ antagonist and GABA_A antagonist prevented the increase in fEPSP caused by CRF, F(2, 14) = 19.42, P < 0.0001. Tukey’s post hoc tests show a significant difference between CP154526 + CRF and 300 nM CRF (P = 0.0031) and SR95531 + CRF and 300 nM CRF (P < 0.0001). H Voltage-clamp recordings of evoked inhibitory postsynaptic currents (eIPSC) from deep layer insular cortex pyramidal neurons under baseline (gray-aCSF with glutamatergic synaptic antagonists) and after 50 nM CRF (blue) in slices from male or female rats. Twenty eIPSCs were evoked by extracellular bipolar electrodes at 5 Hz (the first 10 are shown). For analysis, eIPSC amplitudes were normalized using z-scores computed from the mean and standard deviation of the aCSF recordings (panels I, J). Basal eIPSC amplitudes did not differ between male and female recordings. I CRF significantly reduced the amplitude of eIPSCs in males, F_CRF (1, 8) = 7.006, P = 0.0294. J CRF did not alter eIPSC amplitudes in females F_CRF(1, 8) = 0.0531, P = 0.8235. K Pretreatment of the slice with CRF₁ antagonist eliminated the effect of CRF on eIPSCs in slices from male rats, F_CRF(1, 7) = 0.0547, P = 0.8218.

Fig. 3. CRF augments social behavior and is necessary for social affective behavior.

A A diagram laying out the experimental procedure for social exploration tests. Cannula was placed in the insular cortex. On the day of testing, rats were given 1 h to acclimate to the testing cage. CRF or saline vehicle infusions were made 40 min prior to social interaction with a juvenile (P30) or adult (P50) conspecific for 5 min. B In male rats, CRF increased social exploration of juvenile conspecifics, F_CRF(1, 13) = 48.5, P < 0.0001. Sidak’s post hoc tests revealed significantly increased social exploration at both 50 nM (P = 0.0044) and 300 nM (P < 0.0001). C In male rats, CRF also increased social exploration of P50 conspecifics, F_CRF(1, 15) = 24.99, P = 0.0002. Sidak’s post hoc tests showed that social exploration was increased at both 50 nM (P = 0.0450) and 300 nM (P = 0.0007) doses. D In female rats, 300 nM CRF did not alter social interaction with juvenile conspecifics. Female data were compared males at 300 nM (data replotted from panel B to facilitate comparison) revealing a sex-specific effect of CRF F_{Sex × CRF}(1, 13) = 6.517, P = 0.0241 such that males showed increased social exploration following CRF treatment (P = 0.0033) but females did not (P = 0.8615). E Social exploration by males (n = 11) of juvenile (PN30) conspecifics was altered by CRF and the CRF₁ antagonist CP154526. A two-way repeated-measures ANOVA revealed a significant interaction F_{CRF × CRF1antagonist}(1, 10) = 12.82, P = 0.005. 300 nM CRF increased social exploration that was significantly greater than both the vehicle condition (Tukey’s post hoc test, P = 0.0359) and the combined CRF and CRF₁ antagonist condition (P = 0.0152). Independently, the CRF₁ antagonist had no effect on social exploration (P = 0.6093). F In tests of male rats (n = 8) with adult conspecifics, CRF₁ antagonist blocked the increase in social interaction caused by CRF, F_{CRF × CRF1antagonist}(1, 6) = 12.67, P = 0.0119. Mean social interaction time was greatest in the group that received CRF alone that differed from the vehicle (P = 0.0142) and combined CRF and CRF₁ antagonist conditions (P = 0.0457). G Diagram of the social affective behavior test (SAP) paradigm. Rats received insular cannula implants. On the test day, drug infusions were made 40 min before SAP tests consisting of a 5-min interaction with a naive and stressed same-sex conspecific. The amount of time spent socially investigating each conspecific is recorded. H When tested under vehicle conditions with PN30 conspecifics, male rats (n = 19) exhibit greater exploration of the stressed rat (P = 0.0027); this pattern was blocked by the CRF₁ antagonist (P = 0.8293) supported by a significant interaction, F_{CRF1antagonist × Stress}(1, 18) = 5.225, P = 0.0346. I Experimental male rats (n = 11) spent less time interacting with stressed PN50 adult conspecifics in the vehicle condition but this pattern was blocked by the CRF₁ antagonist, F_{CRF1antagonist × Stress} (1,10) = 6.133, P = 0.0327. Post hoc comparisons revealed a preference for more interaction with naive adults in vehicles (P = 0.0020) but no difference with the CRF₁ antagonist (P = 0.5₁50). J For comparison, time spent interacting with naive and stressed conspecifics from panels H and I was converted to a preference score (% preference = time investigating stressed conspecific/total investigation time × 100). In vehicle conditions, experimental rats preferred interaction with stressed juveniles, but avoided stressed adults and CRF₁ antagonist treatment appeared to reduce these preferences, F_{Age × Drug} (1, 28) = 11.30, P = 0.0023. When comparing juveniles, the percent preference for the stressed conspecific was significantly reduced by CRF₁ antagonist (P = 0.0227). When comparing adults, although the CRF₁ antagonist appeared to eliminate the preference for naive conspecifics, the Sidak-corrected post hoc test did not reach significance (P = 0.0768). K Cannula maps showing the placement of in-dwelling cannula across all experiments related to Fig. 3. Diagrams in panels A and G were created with BioRender.com.

Fig. 4. Cannabinoid 1 receptor is necessary for the behavioral and synaptic effects of insular CRF.

A fEPSPs recorded from insular cortex slices (n = 8) were insensitive to 300 nM CRF when applied with CB₁ receptor antagonist AM251 (2 μM), F_{Voltage × Drug} (1,11) = 2.992, P = 0.0959. B In voltage-clamp recordings of insular cortex pyramidal neurons (n = 9), AM251 prevented the inhibition of eIPSCs previously caused by CRF (50 nM) application, F_Drug(1, 8) = 0.0157, P = 0.9032. C In a 5-min social interaction test with male rats (n = 15) CRF injected into the insular cortex increased exploration of juvenile conspecifics (P = 0.0270). CRF given in combination with AM251 did not increase social exploration, F_{CRF × AM251} (1, 14) = 9.102, P = 0.0092. D For comparison, raw social interaction times from panel C are shown as percent of time relative to the no drug condition. CRF increased interaction (one-sample t-test compared to 100%, t(14) = 3.422, P = 0.0041, AM251 alone increased interaction, t(14) = 2.620, P = 0.0202, but CRF given with AM251 did not differ from vehicle levels, t(14) = 0.5935, P = 0.5623. Importantly, social interaction was greater with CRF alone than in combination with AM521, (P = 0.0110, Sidak test after significant one-way ANOVA, F(2, 28) = 3.705, P = 0.0374). E In SAP tests with juveniles (n = 11), AM251 prevented the formation of a preference for stressed juvenile conspecifics, F_{Drug × Stress}(1, 9) = 22.53, P = 0.0010, such that the preference for stressed juveniles present during vehicle testing (P < 0.0001) was eliminated during AM251 testing (P = 0.0940). F In SAP tests with adult conspecifics (n = 16), AM251 eliminated the preference of test rats for naive adult F_{Drug × Stress}(1, 13) = 19.93, P = 0.0006. A significant preference for naive adults was present in vehicle test rats (P = 0.0013) that was not present in AM251 rats (P = 0.1691). G Percent preference for stressed conspecifics was significantly altered by a combination of AM251 treatment and age of conspecific F_{Age × Drug}(1, 20) = 43.12, P < 0.0001. Specifically, the percent preference for stressed juveniles was significantly reduced (P < 0.0001) while the percent preference for stressed adults was significantly increased (P = 0.0054). H Cannula placements of all animals in the experiments contained in Fig. 4. I Summary of behavioral pharmacology. (1) Exposure to the stressed conspecific causes emotion transfer and release of CRF (purple) from the paraventricular nucleus of the hypothalamus. (2) CRF binds to principal neurons in the insular cortex leading to depolarization and (3) synthesis of endocannabinoids, such as 2-AG. (4) Endocannabinoids (green) bind to CB₁ receptors on presynaptic GABAergic interneurons (orange) leading to hyperpolarization and suppression of local inhibition. (5) The loss of GABAergic inhibition permits greater excitatory synaptic transmission among principal neurons (6) whose output shapes social approach or avoidance by projections to nodes in the social decision-making network. Diagram created in BioRender.com. Bar graphs indicate mean with individual replicates, line graphs mean (±SEM). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (Sidak’s tests).

Fig. 5. Cellular distribution of CRF₁ and CB₁ receptor mRNA in the insular cortex.

A qPCR analysis of relative mRNA expression revealed greater CRF₁ mRNA in females compared to males t(12) = 2.728, P = 0.183, and equal CB₁ mRNA across sexes t(14) = 0.090, P = 0.929. B RNAScope was performed for CRF₁, CB₁, and vglut1 mRNAs. Fluorescent grains were counted in the left and right hemispheres of the posterior insular cortex. The total number of cells was determined by counting DAPI nuclei in each hemisphere. Nuclei containing 3 or more fluorescent grains were considered mRNA expressing cells and shown as the % of the total cells. The number of vglut1 cells was equal between sexes, but the portion of cells expressing CRF₁ F_Sex(1, 14) = 11.19, P = 0.005, or CB₁, F_Sex(1, 14) = 13.07, P = 0.003, mRNA was approximately double in males compared to females. C Looking at expression of CRF₁ and CB₁ mRNA in vglut1 cells, shown as a percent of the total vglut1 cells per hemisphere, male rat sections contained more CB₁ mRNA expressing vglut cells, and more vglut cells expressing both CB₁ and CRF₁ mRNA, F_Sex(1, 14) = 4.489, P = 0.044). The number of vglut cells expressing CRF1 mRNA was greater on average in males than females, but did not reach significance F_Sex(1, 14) = 3.944, P = 0.067. In males, there were more vglut cells colocalized with both CRF₁ and CB₁, F_Sex(1, 14) = 6.576, P = 0.023. D, E Representative digital photomicrographs of RNAScope in situ hybridization and fluorescent visualization of DAPI and vesicular glutamate transporter 1 (vglut1), CRF₁ (crhr1), and CB₁ (cn1r) mRNA from male (A) and female (rats) insular cortex coronal sections (20 um, n = 8/sex). Sections were selected from subjects nearest to the mean values for CRF₁ + DAPI colabeling. Colored arrowheads indicate cells with coexpression of all three mRNAs with DAPI. Bar graphs indicate mean (±SEM) with individual replicates. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (Sidak’s tests).