CRH Engagement of the Locus Coeruleus Noradrenergic System Mediates Stress-Induced Anxiety

Abstract

The locus coeruleus noradrenergic (LC-NE) system is one of the first systems engaged following a stressful event. While numerous groups have demonstrated that LC-NE neurons are activated by many different stressors, the underlying neural circuitry and the role of this activity in generating stress-induced anxiety has not been elucidated. Using a combination of in vivo chemogenetics, optogenetics, and retrograde tracing, we determine that increased tonic activity of the LC-NE system is necessary and sufficient for stress-induced anxiety and aversion. Selective inhibition of LC-NE neurons during stress prevents subsequent anxiety-like behavior. Exogenously increasing tonic, but not phasic, activity of LC-NE neurons is alone sufficient for anxiety-like and aversive behavior. Furthermore, endogenous corticotropin-releasing hormone(+) (CRH(+)) LC inputs from the amygdala increase tonic LC activity, inducing anxiety-like behaviors. These studies position the LC-NE system as a critical mediator of acute stress-induced anxiety and offer a potential intervention for preventing stress-related affective disorders.

Figure 1. Chemogenetic inhibition of LC-NE neurons prevents stress-induced anxiety

(A) Cartoon of restraint stress-induced anxiety paradigm. (B and C) Stressed animals spend significantly less time exploring the center of the OFT than non-stressed controls with no change in locomotor activity (Data represented as mean ± SEM, n=7–8/group: Student’s t-test, p<0.001). (D and E) Representative immunohistochemistry (IHC) and quantification show restraint stress increases c-fos immunoreactivity (IR) in LC neurons (Red=c-fos, green=tyrosine hydroxylase, arrows indicate example co-localization, scale bars=100 μm; data represented as mean ± SEM, n=3 slices from 3 animals/group: Student’s t-test, p<0.0001). See locomotor activity data in Fig. S1. (E) Cartoon of viral strategy with high power confocal image of post-CNO internalized mCherry expression in LC-NE neurons. Scale bar=10 μm. (G and H) Representative IHC and quantification show hM4Di inhibition of LC neurons decreases c-fos IR in LC neurons (Red=c-fos, green=tyrosine hydroxylase, Scale bars=100 μm; data represented as mean ± SEM, n=3 slices from 3 animals/group: One-Way ANOVA, Bonferroni post-hoc, No stress/Cre⁻ vs. No stress/Cre⁺ **p<0.01, No stress/Cre⁻ vs. Stress/Cre⁻ ****p<0.0001, No stress/Cre⁻ vs. Stress/Cre⁺ ****p<0.0001, No stress/Cre⁺ vs. Stress/Cre⁻ ****p<0.0001, No stress/Cre⁺ vs. Stress/Cre⁺, not significant). See also Fig. S1. (I) Cartoon of LC-NE inhibition paradigm. (J) Inhibition of LC-NE neurons during stress blocks stress-induced anxiety. Data represented as mean ± SEM, n=8–13/group: One-Way ANOVA, Newman-Keuls post-hoc, No stress/Cre⁻ vs. No stress/Cre⁺, not significant, No stress/Cre⁻ vs. Stress/Cre⁻ **p<0.01, No stress/Cre⁻ vs. Stress/Cre⁺, not significant, No stress/Cre⁺ vs. Stress/Cre⁻ **p<0.01, No stress/Cre⁺ vs. Stress/Cre⁺, not significant). (K) Inhibition of LC-NE neurons has no effect on stress-induced bowel motility. Data represented as mean ± SEM, n= n=8–13/group: One-Way ANOVA, Newman-Keuls post-hoc, No stress/Cre⁻ vs. No stress/Cre⁺, not significant, No stress/Cre⁻ vs. Stress/Cre⁻ *p<0.05, No stress/Cre⁻ vs. Stress/Cre⁺ *p<0.05, No stress/Cre⁺ vs. Stress/Cre⁻, not significant, No stress/Cre⁺ vs. Stress/Cre⁺, not significant). (L) Inhibition of LC-NE neurons has no effect on locomotor activity. Data represented as mean ± SEM. (M) Representative heat maps of activity during OFT.

Figure 2. Optogenetic targeting to selectively drive tonic LC-NE activity

(A) Cartoon of viral strategy for slice experiments. (B and C) Representative IHC shows selective targeting of ChR2-eYFP to TH⁺ LC neurons (Red= tyrosine hydroxylase, green=ChR2-eYFP, 4V = 4^th ventricle, Scale bars= 100 and 50 μm, respectively). (D and E) Whole cell current- and voltage-clamp recordings of an LC-NE neuron expressing ChR2. D, scale bar is 200 pA x 100 ms and E, scale is 20 ms. (F) Slice recording of a single LC-NE neuron demonstrating action potentials over 30 min in response to 5 Hz 470 nm light. (G) Cartoon of viral and multielectrode delivery for anesthetized, in vivo recordings. (H) Peristimulus time histogram (PSTH) showing increased LC neuron firing during a 20 s optical stimulation at 5 Hz. (I) Firing rate of n=16 cells before and during 5 Hz photostimulation (Paired Student’s t-test, p<0.01). (J) Correlation of baseline activity to photostimulated activity (r=0.8938, p<0.0001).

Figure 3. High tonic LC-NE neuronal activity is sufficient to induce anxiety-like behavior

(A) Cartoon of viral and fiber optic delivery. (B) Calendar of pre-stimulation OFT studies. (C) 5 Hz photostimulation prior to OFT causes an anxiety-like phenotype of ChR2-expressing Th-Cre⁺ animals compared to Th-Cre⁻ controls (Data represented as mean ± SEM, n=14–15/group: Mann-Whitney t-test, p<0.01) with (D) a significant decrease in locomotor activity (Data represented as mean ± SEM, n=14–15/group: Student’s t-test, p<0.05). (E) Calendar of concurrent stimulation OFT studies. (F) 5 Hz photostimulation drives anxiety-like behavior in OFT of Th-Cre^LC:ChR2 animals compared to Th-Cre^LC:eYFP controls (Data represented as mean ± SEM, n=10/group: Student’s t-test, p<0.01) with (E) no change in locomotor activity (Data represented as mean ± SEM, n=10/group). (H) Representative heat maps of activity in the OFT. See also Fig. S2. (I) Calendar of systemic antagonism in EZM studies. (J) Representative heat maps of activity in the EZM. (K) 5 Hz photostimulation drives anxiety-like behavior in EZM of Th-Cre^LC:ChR2 animals compared to Th-Cre^LC:eYFP controls, which is blocked by β-adrenergic antagonism (Prop), but not α1 antagonism (Praz) (Data represented as mean ± SEM, n=6–9/group; Kruskal-Wallis One-way ANOVA, *p<0.05, **p<0.01).

Figure 4. LC-NE photostimulation drives both real-time and learned aversions

(A) Current-clamp whole cell recording at 2, 3, 5 and 8 Hz. (B) Cartoon of viral and fiber optic delivery and calendar of behavioral studies. (C) Frequency response of RTPT and (D) locomotor activity at 0, 1, 2, 5, and 10 Hz. Data represented as mean ± SEM, n=6–7/group: Two-Way ANOVA, Bonferroni post-hoc, 5 Hz ChR2 vs. 5 Hz eYFP *p<0.05, 10 Hz ChR2 vs. 10 Hz eYFP ***p<0.001. (E) Representative traces of behavior at different frequencies. (F) Phasic stimulation does not drive aversion, n=6/group. (G) 5 Hz photostimulation causes a real-time place aversion in RTPT of Th-Cre⁺ animals expressing ChR2 in the LC compared to Th-Cre⁻ control animals that do not express ChR2. This effect is not reversed by Propranolol (Prop) pre-treatment, but is by Prazosin (Praz) (Data represented as mean ± SEM, n=6–10/group; One-Way ANOVA, Bonferonni post-hoc, ****p<0.0001, **p<0.01, ns=no significance). (H) Timeline and cartoon of 5 Hz CPA experiment. (H) Mean preference (s) ± SEM, post-test minus pre-test (n=7–9; Student’s t-test, p<0.05). (I) No locomotor effect in either pre- or post-test. (J) Representative traces of behavior in the pre- and post-test. See also Fig. S3.

Figure 5. Galanin containing LC-NE neurons are sufficient to drive place aversion

(A and B) Galanin labeling in Gal-Cre x Ai9-tdTomato compared to in situ images from the Allen Institute for Brain Science in a sagittal section highlighting presence of galanin in the LC. All images show tdTomato (red) and Nissl (blue) staining. See also Fig. S4. (C) Cartoon of viral and fiber optic delivery. (D) IHC of ChR2-eYFP targeting to DBH⁺ LC neurons. Scale bar=100 μm. (E) Quantification of Cre-dependent eYFP viral expression in the LC of Th-Cre and Gal-Cre mouse lines. Data represented as mean ± SEM, n=3 slices from 3 animals/group: Student’s t-test, p<0.0001. See also Fig. S5. (F) 5 Hz stimulation of LC-Gal neurons drives aversion (n=6, Paired Student’s t-test, p<0.01). (G) Representative traces of behavior at different frequencies.

Figure 6. Identifying a CRH⁺ CeA input to the LC

(A) Calendar of pharmacological experiment. (B) CRF-R1 antagonism blocks stress-induced anxiety-like behavior (C) with no significant effect on locomotor activity (n=6–8/group, Student’s t-test, *p<0.05). (D) Representative heat maps show behavior in the OFT. (E) Cartoon depicting fluorgold tracing. (F) Representative image shows robust retrograde labeling of the CeA (Fluorgold pseudocolored red, Nissl=grey). See also Fig. S6. (G) Cartoon depicting dual injection tracing for CTB-594 and DIO-ChR2-eYFP. (H) Representative IHC shows retrograde labeling in CeA of CTB-594 (red) and anterograde labeling of CRH⁺ cells (green). Arrow indicates example co-localization. (I) ~25% of each label co-labels with the other. (J) Cartoon depicting anterograde tracing. (K) 71° off of sagittal slice of Crh-Cre mouse expressing DIO-eYFP in the CeA. Image shows intact projections from CeA to LC. Arrow indicates fiber optic placement. (L and M) Coronal image depict robust eYFP labeling in the CeA and LC of the same mouse. All scale bars=100 μm. + is 4^th ventricle.

Figure 7. CRH⁺ CeA-LC terminals modulate LC activity and drive anxiety through CRFR1 activation

(A) Cartoon of viral and multi-electrode array delivery for anesthetized, in vivo recordings. (B and C) Representative PSTHs of putative LC neurons responding to 20s of 10 Hz, 10 ms pulse width photostimulation (473 nm, ~10 mW). (D) Representative principal component analysis plot showing the first two principal components with clear clustering of units. (E) Total recorded sample shows significant increase in firing rate to 10 Hz photostimulation (n=35, Wilcoxon matched- pairs signed rank test, p<0.05). (F) n=15 units increase firing rate by >10% during 10 Hz photostimulation (Wilcoxon matched- pairs signed rank test, p<0.0001). (G) Response latency following onset of photostimulation for cells that increase firing. (H) n=15 units decrease firing rate by >10% during 10 Hz photostimulation (Wilcoxon matched- pairs signed rank test, p<0.0001). See also Fig. S7A. (I) Response latency following onset of photostimulation for cells that increase firing. (J) Correlation of baseline activity to activity during photostimulation (r=0.7029, p<0.0001). Representative voltage-clamp traces following (K) 10 Hz, 3 ms photostimulation and (L) 50 Hz, 3 ms photostimulation of Crh-Cre^CeA-LC:ChR2 terminals. Scale bars 30 pA; 50 ms. (M) Magnitude of the event after the first photostimulation. Data represented as mean ± SEM, n=35 cells from 4 brains.

Figure 8. CRH⁺ CeA-LC terminals drive aversion and anxiety-like behavior through CRFR1 activity

(A) Cartoon of viral and fiber optic delivery and (B) calendar of CPA behavior. (C and D) Crh-Cre^CeA-LC:ChR2 show a significant CPA compared to Crh-Cre^CeA-LC:eYFP controls. Representative traces of behavior in the pre- and post-test and mean preference (s) ± SEM, post-test minus pre-test (n=10–12/group; Student’s t-test, p<0.05). (E) Cartoon of viral, cannula, and fiber optic delivery and calendar of EZM behavior. (F) 10 Hz photostimulation drives anxiety-like behavior in EZM of Crh-Cre^CeA-LC:ChR2 animals compared to Crh-Cre^CeA-LC:eYFP controls, which is reversed by intra-LC α-helical-CRF (αHCRF) pretreatment (Data represented as mean ± SEM, n=7/group: One-Way ANOVA, Newman-Keuls, Crh-Cre^CeA-LC:eYFP vs. Crh-Cre^CeA-LC:ChR2 *p<0.05, Cre^CeA-LC:eYFP vs. Crh-Cre^{CeA-LC:ChR2+AHCRF} **p<0.01, Cre^CeA-LC:ChR2 vs. Crh-Cre^{CeA-LC:ChR2+AHCRF} ***p<0.001). See also Fig. S7B&C. (G) Systemic CRHR1 antagonism reverses photostimulation-induced anxiety-like behavior (n=6–8/group, Student’s t-test, p<0.001). (H) Model and summary of results. See also Fig. S7D.