Networks of habenula-projecting cortical neurons regulate cocaine seeking

Abstract

How neurons in the medial prefrontal cortex broadcast stress-relevant information to subcortical brain sites to regulate cocaine relapse remains unclear. The lateral habenula (LHb) serves as a “hub” to filter and propagate stress- and aversion-relevant information in the brain. Here, we show that chemogenetic inhibition of cortical inputs to LHb attenuates relapse-like reinstatement of extinguished cocaine seeking in mice. Using an RNA sequencing–based brain mapping procedure with single-cell resolution, we identify networks of cortical neurons that project to LHb and then preferentially innervate different downstream brain sites, including the ventral tegmental area, median raphe nucleus, and locus coeruleus (LC). By using an intersectional chemogenetics approach, we show that inhibition of cortico-habenular neurons that project to LC, but not to other sites, blocks reinstatement of cocaine seeking. These findings highlight the remarkable complexity of descending cortical inputs to the habenula and identify a cortico-habenulo-hindbrain circuit that regulates cocaine seeking.

Fig. 1.. mPFC_➔LHb neurons signal the occurrence of stressful events.

(A) Graphical representation of intracranial injection procedure to label cortico-habenular neurons (top left). Representative fluorescence from tdTom⁺ neurons in the LHb after local injection of rg-tdTom (bottom left). Representative tdTom⁺ cells in the mPFC of mice that received intra-LHb injection of rg-tdTom (right). A total of n = 7 mice were imaged. (B) Numbers of tdTom⁺ cells detected throughout the mPFC. MO, medio-orbital cortex; PrL, prelimbic cortex; IL, infralimbic cortex; Cg1, anterior cingulate cortex region 1; Cg2, anterior cingulate cortex region 2. (C) Graphical representation of intracranial injection procedure to label cortico-habenular neurons in Vgat-Cre::tdTom reporter mice (top left). Representative fluorescence from GFP⁺ neurons in the LHb after local injection of rg-GFP (top right). Representative fluorescence image showing that GFP was detected exclusively in non-tdTom⁺ cells in the mPFC of Vgat-Cre::tdTom mice injected with rg-GFP in the LHb (bottom). A total of n = 4 Vgat-Cre::tdTom mice were imaged. (D) Schematic of photometry-based in vivo calcium imaging procedure in mice (left). Graphical representation of intracranial injection procedure to express GCaMP6m in cortico-habenular neurons (top right). Representative GCaMP6m expression in the mPFC of mice used in the photometry experiment (bottom right). (E) Representative trace showing increase in GCaMP6m-derived fluorescence after delivery of a noxious foot-shock stressor. A total of n = 3 mice were recorded. (F) z score representation of calcium response to foot shock [one-way ANOVA: F_(3,80) = 44.66, P < 0.0001; ***P < 0.001 compared with −2 s time point].

Fig. 2.. mPFC_➔LHb neurons regulate cocaine seeking.

(A) Summary of virus injection procedure and representative fluorescence image of GFP⁺ neurons in the LHb. (B) Representative image from the mPFC of virus-injected mice. (C) Time course of CNO-induced hyperpolarization of membrane potential in hM4Di-expressing mPFC neurons. (D) Example of CNO-induced reduction of whole-cell membrane resistance. (E) Representative traces of voltage changes in whole-cell mode in hM4Di-expressing mPFC neurons depolarized by current injections (500 ms; 10 to 80 pA in 5-pA steps) before [artificial cerebrospinal fluid (aCSF)] and after CNO (20 μM). (F) Mean (±SEM) plasma corticosterone levels in yohimbine-treated (Yoh; n = 5) and saline-treated (n = 7) mice. **P = 0.005, unpaired two-tailed t test. (G) Summary of intravenous cocaine self-administration (IVSA) procedure. After extinction training, mice were injected with CNO (3 mg kg⁻¹) and then yohimbine (2 mg kg⁻¹), and responding was recorded. (H) Yohimbine reinstated cocaine seeking in mCherry but not hM4Di mice [main effect of Virus during yohimbine and post-Yoh sessions; F_(1,16) = 5.042, *P = 0.0392]. Data are presented as mean (±SEM) responses on active lever. (I) Yohimbine did not alter inactive lever responding [F_(1,16) = 1.573, P = 0.2278]. Data are presented as means (±SEM). (J) Summary of conditioned place preference (CPP) procedure and experimental design. (K) Mean (±SEM) percentage of time spent in cocaine-paired side during habituation (Hab), preference test for cocaine-paired side (Test), and after extinction sessions (Ext) in mice expressing hM4Di (n = 11) or mCherry (n = 7) in cortico-habenular neurons. ***P < 0.001, unpaired two-tailed t test. (L) Mean (±SEM) % time spent in cocaine-paired side during reinstatement session in hM4Di (n = 11) and mCherry (n = 7) mice injected with CNO and then yohimbine. ***P < 0.001, one-sample t test. (M) Mean (±SEM) % time spent in cocaine-paired side on the day after reinstatement session. ***P < 0.001, one-sample t test. (N) Graphical representation of cortico-habenular neurons highlighting their responsiveness to stressful stimuli and their involvement in stress-induced reinstatement of cocaine seeking (cocaine relapse).

Fig. 3.. Connectomes of mPFC_➔LHb neurons.

(A) Summary of MAPseq procedure. (B) Percentage of barcodes detected in the mPFC also detected in brain sites listed on y axis. Data were collected from n = 2 mice. (C) Heatmap representation of mPFC barcodes (% total number) detected in brain regions shown on y axis and in regions shown on x axis. (D) Histogram of connectome of mPFC_➔PVT, mPFC_➔VTA, and mPFC_➔LHb neurons. Shown is the distribution of barcodes (%) for mPFC cells that project to the PVT, VTA, and LHb. (E) Summary of injection procedure to label cortical-habenular neurons that project to VTA and fluorescence image of mPFC from mice injected with CTb retrograde tracers in the LHb and VTA. Shown are CTb-594 (red; mPFC_➔LHb neurons)–labeled and CTb-488 (green; mPFC_➔VTA neurons)–labeled cells. White arrows identify dual-labeled cells (yellow; mPFC_➔LHb^VTA neurons) in the mPFC. A total of n = 5 mice were imaged. (F) Summary of injection procedure to label cortical-habenular neurons that project to the LC and associated fluorescence image of mPFC from injected mice (red; mPFC_➔LHb neurons) and GFP (green; mPFC_➔VTA neurons). White arrows identify dual-labeled cells (yellow; mPFC_➔LHb^LC neurons) in the mPFC. A total of n = 5 mice were imaged. (G) Injection procedure (the tracing the relationship between input and output method, also known as the TRIO method) to label cortico-habenular neurons that project to VTA. Shown are mCherry⁺ cells in the mPFC, which are cortico-habenular neurons that synapse into LHb neurons that project to VTA. A total of n = 4 mice were imaged. (H) Injection procedure to label cortico-habenular neurons that project to the LC. mCherry⁺ cells were detected in the mPFC (right), which are cortico-habenular neurons that synapse into LHb neurons that project to the LC. A total of n = 3 mice were imaged. (I) Graphical representation of the cortico-habenular connectome. A population of mPFC neurons (shown in blue) projects concurrently to the LHb and then to the same monoaminergic brain centers to which LHb neurons also project (shown in red).

Fig. 4.. mPFC_➔LHb^LC neurons regulate cocaine seeking.

(A) Summary of intersectional genetic approach to express hM4Di in cortico-habenular neurons that project to the VTA, MRn, or LC. (B) Only cortico-habenular neurons that project to targeted downstream sites can contain FLPo and fDIO-Cre necessary to express DIO-hM4Di. (C) CNO hyperpolarized mPFC neurons that express hM4Di. *P < 0.05, paired t test. RMP, resting membrane potential. (D) Summary of procedure to express hM4Di-mCherry (or only Cherry) in mPFC_➔LHb^VTA neurons. Representative image of hM4Di-mCherry⁺ cells in mPFC_➔LHb^VTA neurons. (E) Mean (±SEM) time (%) spent in the cocaine-paired side of CPP apparatus during habituation, testing, and extinction sessions in mice expressing hM4Di (n = 10) or mCherry (n = 7) in mPFC_➔LHb^VTA neurons. ***P < 0.001 and *P < 0.05, unpaired two-tailed t tests. (F) Mean (±SEM) time (%) spent in cocaine-paired side during reinstatement in hM4Di (n = 11) and mCherry (n = 7) mice injected with CNO and then yohimbine. (G) Summary of strategy to target mPFC_➔LHb^MRn neurons and fluorescence image of hM4Di-mCherry in mPFC_➔LHb^MRn neurons. (H) Mean (±SEM) time (%) spent in cocaine-paired side during habituation, testing, and extinction sessions in hM4Di (n = 13) and mCherry (n = 12) mice. ***P < 0.001, unpaired two-tailed t tests. (I) Mean (±SEM) time (%) spent in cocaine-paired side during reinstatement session in hM4Di (n = 13) and mCherry (n = 12) mice injected with CNO and then yohimbine. (J) Summary of strategy to target mPFC_➔LHb^LC neurons and representative image of hM4Di-mCherry in mPFC_➔LHb^LC neurons. (K) Mean (±SEM) time (%) spent in cocaine-paired side during habituation, testing, and extinction sessions in hM4Di (n = 17) and mCherry (n = 13) mice. ***P < 0.001, unpaired two-tailed t tests. **P < 0.01.(L) Mean (±SEM) time (%) spent in cocaine-paired side during reinstatement session in hM4Di (n = 17) and mCherry (n = 13) mice injected with CNO and then yohimbine. *P < 0.05, unpaired two-tailed t tests. (M) Graphical representation of the cortico-habenular neurons that project to the LC and regulate cocaine seeking.