Lateral Preoptic Control of the Lateral Habenula through Convergent Glutamate and GABA Transmission

Abstract

The lateral habenula (LHb) is a brain structure that participates in cognitive and emotional processing and has been implicated in several mental disorders. Although one of the largest inputs to the LHb originates in the lateral preoptic area (LPO), little is known about how the LPO participates in the regulation of LHb function. Here, we provide evidence that the LPO exerts bivalent control over the LHb through the convergent transmission of LPO glutamate and γ-aminobutyric acid (GABA) onto single LHb neurons. In vivo, both LPO-glutamatergic and LPO-GABAergic inputs to the LHb are activated by aversive stimuli, and their predictive cues yet produce opposing behaviors when stimulated independently. These results support a model wherein the balanced response of converging LPO-glutamate and LPO-GABA are necessary for a normal response to noxious stimuli, and an imbalance in LPO→LHb glutamate or GABA results in the type of aberrant processing that may underlie mental disorders.

Keywords: GABA; aversion; calcium imaging; electron microscopy; glutamate; habenula; optogenetics; preoptic; reward; stress; synapse.

Figure 1. The major input from the lateral preoptic area (LPO) to the lateral habenula (LHb) is from glutamatergic neurons

(a) Iontophoretic delivery of the retrograde tract tracer Fluorogold (FG) into the LHb. (b) FG injection site in the LHb and (c) retrogradely labeled LPO➔LHb neurons. (d–e) Phenotypic characterization of LPO➔LHb neurons. FG is seen as white–or brown after its immunodetection (FG-IR). (d) LPO➔LHb neuron expressing GADs mRNA (detection with non-radioactive probe; purple cell). (e) LPO➔LHb neuron expressing VGluT2 mRNA (detection with radioactive probe; cell with green or silver grains). (f) Frequency of different phenotypes of LPO➔LHb neurons (mean ± SEM; 10–15 sections from each of three rats). Most LPO➔LHb neurons expressed VGluT2 mRNA (FG-VGluT2; 74.7% ± 3.2%), some expressed GADs (FG-GADs; 16.0% ± 3.2%, fewer lacked detectable levels of VGluT2 or GADs mRNA (FG; 5.9% ± 1.8%) and infrequently expressed both transcripts (FG-VGluT2-GADs; 3.5% ± 2.8%). (g) The frequency of FG-GADs neurons was confirmed with a radioactive probe for GAD 65/67 detection. Scale bar in c, 40 μm; d, and e, 10 μm.

Figure 2. LPO neurons establish functional glutamate and GABA synapses on LHb neurons

(a) Delivery of AAV-DIO-ChR2-mCherry (for electron microscopy) or AAV-DIO-ChR2-eYFP (for in vitro electrophysiology) into the LPO of VGluT2∷Cre or VGAT∷Cre mice. LHb electron micrographs (b, c, d) and corresponding diagrams (e, f, g) showing LPO axon terminals (AT) expressing mCherry (dark diffuse material, yellow arrow). (b, e) LPO-mCherry terminal in the LHb of a VGluT2-LPO^ChR2-mCherry mouse expressing VGluT2 (gold particles seen; black dots; green arrow-head) establishing an asymmetric synapse (green arrow) on a dendrite (De) of a LHb neuron. (c, d, f, g) LPO mCherry terminal in the LHb of a VGAT-LPO^ChR2-mCherry mouse expressing VGAT (gold particles seen as black; dots; red arrow-heads) establishing a symmetric synapse (red arrow) on a dendrite (De) (c, f) or soma (d, g) of LHb neurons. (h) EPSCs recorded in voltage clamp mode in a LHb neuron after LHb photostimulation (blue line) of LPO-VGluT2 fibers were not affected by Bicuculline (10 μM), but were abolished by the subsequent addition of CNQX (10 μM), (Baseline = −138.18 ± 27.98 pA; Bicuculline = −148.83 ± 29.67 pA; Bicuculline + CNQX = −2.84 ± 0.40 pA; F2, 29 = 23.56, *P ˂ 0.0001, repeated measures ANOVA, post hoc Dunnett’s test; n = 10 cells from 6 mice) error bars correspond to SEM. (i) IPSCs recorded in voltage clamp mode in a LHb neuron after LHb photostimulation (blue line) of LPO-VGAT fibers were not affected by CNQX (10 μM), but were abolished by the subsequent addition of Bicuculline (10 μM), (Baseline = 20.23 ± 3.60 pA; CNQX = 17.58 ± 3.58 pA ; CNQX + Bicuculline = 1.38 ± 0.26 pA; F2, 20 = 21.56, *P = 0.0001, repeated measures ANOVA, post hoc Dunnett’s test; n = 7 cells from 5 mice) error bars correspond to SEM. Scale bars: b, c, d 200 nm.

Figure 3. Axon terminals from LPO-glutamate and LPO-GABA neurons innervating the LHb have distinct topographical distributions within the LHb

(a) Delivery of AAV-CaMKII-ChR2-mCherry into the LPO of wild-type mice (WT-LPO^ChR2-mCherry). (b) LHb at low magnification showing LPO-mCherry fibers in the LHb and axon terminals with either VGluT2 or VGAT labeling. (c–d) High magnification of the medial (c) and lateral (d) aspects of the LHb showing LPO terminals containing VgluT2 (blue arrows) or VGAT (yellow arrows). (e) LPO-mCherry fibers were more prevalent in the medial than in the lateral aspects of the LHb, and more frequently contained VgluT2 than VGAT. Within the lateral aspects of the LHb, the LPO-mCherry-VGAT were more abundant than LPO-mCherry-VGluT2 (n=24236 terminals n=3 mice with three replications each; F1, 8=49.02, P=0.0001, Two-way Repeated Measures ANOVA, Sidak adj. post-hoc, *P < 0.05). (f–g) Diagrams showing a high concentration of LPO-mCherry-VGluT2 axon terminals in the medial aspect of the LHb (f) and a broad distribution of LPO-mCherry-VGAT axon terminals across the LHb (g). Scale Bars: b, 100 μm, c–d, 2 μm.

Figure 4. Converging LPO-glutamatergic and LPO-GABAergic inputs onto single LHb neurons

(a) Delivery of AAV-CaMKII-ChR2-eYFP into the LPO of wild-type mice (WT-LPO^ChR2-eYFP). (b, d, f) LHb neurons were recorded under voltage clamp at three holding potentials: 0 mV, to record inhibitory currents; −60 mV to record excitatory currents, and −45 mV to simultaneously record both inhibitory and excitatory currents. (b–c) LPO-glutamatergic input on a LHb neuron. EPSCs recorded in voltage clamp mode in a LHb neuron at three holding potentials after LHb photostimulation (blue line) of LPO-fibers were all blocked by CNQX (10μM) (b), indicating LPO-glutamatergic neurotransmission on LHb neurons (~29%; 6 of 21 recorded LHb neurons) (c). (d–e) LPO-GABAergic input on a LHb neuron. IPSCs recorded in voltage clamp mode at three holding potentials in a LHb neuron after LHb photostimulation (blue line) of LPO-fibers were not affected by CNQX (10μM), but were all blocked by the subsequent addition of Bicuculline (10μM) (d), indicating LPO-GABAergic neurotransmission on LHb neurons (~9%; 2 of 21 recorded LHb neurons) (e). (f–g) Convergent neurotransmission of LPO-glutamate and LPO-GABA inputs on a single LHb neuron. Both EPSCs and IPSCs were recorded in voltage clamp mode on a single LHb neuron after LHb photostimulation. EPSCs were blocked by CNQX (10μM) and the remaining IPSCs were blocked by the subsequent addition of Bicuculline (10μM) (f), indicating that LPO-glutamate and LPO-GABA neurons provide convergent neurotransmission onto single LHb neurons (~62%; 13 of 21 recorded LHb neurons) (g).

Figure 5. Simultaneous activation of LPO-VGluT2 and LPO-VGAT fibers in the LHb by an aversive stimulus and its predictive cue

(a) Delivery of AAV-DIO-GCAMP6s into the LPO in VGAT-LPO^GCaMP6s mice or VGluT2-LPO^GCaMP6s mice, and optic fiber placement over the LHb for recording Ca2⁺ activity in LPO-VGluT2 fibers and LPO-VGAT fibers. (b) Ca2⁺ imaging traces during an entire aversive conditioning task. (c–d) Heat maps of Ca2⁺ activity over successive aversive conditioning trials (top) and peri-event histograms showing the average trace during aversive conditioning (bottom) in a VGAT-LPO^GCaMP6s mouse (c) and VGluT2-LPO^GCaMP6s mouse (d). (e) Average Ca2⁺ activity (± SEM) for VGluT2-LPO^GCaMP6s mice or VGAT-LPO^GCaMP6s mice in a 25 s window encompassing the tone and footshock period. (f) Area under the curve (ΔF/F) for Ca2⁺ activity in LPO-VGluT2 terminals or LPO-VGAT terminals during baseline (VGluT2-LPO^GCaMP6s mice: −293.52 ± 155.47; VGAT-LPO^GCaMP6s mice: −103.12 ± 91.13), tone (VGluT2-LPO^GCaMP6s mice: 1183.86 ± 562.63; VGAT-LPO^GCaMP6s mice: 590.88 ± 149.85), shock (VGluT2-LPO^GCaMP6s mice: 940.03 ± 434.93; VGAT-LPO^GCaMP6s mice: 365.13 ± 116.72), and post-shock epochs (VGluT2-LPO^GCaMP6s mice: −121.05 ± 85.55; VGAT-LPO^GCaMP6s mice: −147.96 ± 58.50) (n=13 VGluT2-LPO^GCaMP6 mice and n=14 VGAT-LPO^GCaMP6 mice; Mixed ANOVA, main effect of epoch F3, 75=24.61, P<0.0001, GLM, Sidak post-hoc test, *P < 0.001 vs. baseline; Epoch x Group and Group effects, N.S.)

Figure 6. LHb photostimulation of LPO-glutamate is aversive, while LHb photostimulation of LPO-GABA fibers is rewarding

(a) Diagram of virus injection in LPO and photostimulation of LPO-VGluT2 or LPO-VGAT fibers in the LHb. (b) Timeline for the place conditioning procedure. (c) VGluT2-LPO^ChR2-eYFP mice (n=9) spent less time in the chamber paired with photostimulation on conditioning days (blue line) and exhibited place aversion during stimulation-free test sessions. VGluT2-LPO^eYFP mice (n=7) spent equal time in the photostimulation paired and unpaired chambers during conditioning (blue line) and test sessions (F20, 280)=3.78, P<0.0001, Mixed ANOVA, Sidak adjusted post-hoc, *p<0.05 Paired vs. Unpaired chamber). (d) VGAT-LPO^ChR2-eYFP mice (n=7) spent more time in the photostimulation paired chamber during conditioning (blue line) sessions. Stimulation in VGAT-LPO^ChR2-eYFP mice did not produce a conditioned place preference for the photostimulation-paired chamber on test sessions, although there was a trend for animals to spend more time in the photostimulation paired chamber. VGAT-LPO^eYFP mice (n=7) did not show a preference for either the paired or unpaired chamber during conditioning sessions (blue line) or test sessions (F20, 260= 1.76, p=0.26, Mixed ANOVA, Sidak adjusted post-hoc, *p<0.05 Paired vs. Unpaired chamber).

Figure 7. Functional neuroanatomy of the LPO➔LHb circuit

a) Electrophysiological and anatomical evidence indicate that LPO-glutamate and LPO-GABA neurons both synapse on single LHb neurons to provide convergent neurotransmission of glutamate and GABA or independently synapse on LHb neurons to release only glutamate or only GABA. b) Recording LPO-VGluT2 or LPO-VGAT terminals by fiber photometry Ca²⁺ imaging demonstrated that noxious footshock induces the simultaneous and balanced activation of LPO-glutamatergic and LPO-GABAergic inputs to the LHb. c) Independent optogenetic activation of LPO➔LHb GABA produces reward while (d) independent optogenetic activation of LPO➔LHb glutamate produces aversions. These results suggest that the balance of LPO-glutamate and LPO-GABA is critical for an organisms normal response to aversive stimuli and that shifts in this balance may produce a psychopathological state.