Ventral tegmental area astrocytes orchestrate avoidance and approach behavior

Abstract

The ventral tegmental area (VTA) is a heterogeneous midbrain structure, containing neurons and astrocytes, that coordinates behaviors by integrating activity from numerous afferents. Within neuron-astrocyte networks, astrocytes control signals from distinct afferents in a circuit-specific manner, but whether this capacity scales up to drive motivated behavior has been undetermined. Using genetic and optical dissection strategies we report that VTA astrocytes tune glutamatergic signaling selectively on local inhibitory neurons to drive a functional circuit for learned avoidance. In this circuit, astrocytes facilitate excitation of VTA GABA neurons to increase inhibition of dopamine neurons, eliciting real-time and learned avoidance behavior that is sufficient to impede expression of preference for reward. Loss of one glutamate transporter (GLT-1) from VTA astrocytes selectively blocks these avoidance behaviors and spares preference for reward. Thus, VTA astrocytes selectively regulate excitation of local GABA neurons to drive a distinct avoidance circuit that opposes approach behavior.

Fig. 1

Neuronal afferent activity changes glutamate transport in VTA astrocytes. a Schematic showing the microcircuitry of the VTA and recordings performed. b Recording from a VTA astrocyte showing the current generated by electrical stimulation (black; 50 Hz, 20 pulses), the current remaining after blocking action potentials with 250 µM lidocaine (red), and after washing out lidocaine (green). c Summary of the currents before (black) and after lidocaine application (red; paired t-test t₍₅₎ = 2.679, P = 0.04, n = 6 cells, 4 mice). d Current generated in a VTA astrocyte by electrical stimulation of afferents before (black) and after adding 15 µm TFB-TBOA (red). e Summary of the currents generated by electrical stimulation in VTA astrocytes before (black) and after (red) addition of TFB-TBOA (paired t-test t₍₄₎ = 2.159, P = 0.04, n = 5 cells, 5 mice). *: P < 0.05

Fig. 2

Astrocytic GLT-1 increases the excitation of VTA GABA neurons. a Schematic showing the microcircuitry of the VTA and manipulations performed. b Immunocytochemistry images (×20 objective) showing lack of co-localization between ChR2 (green, note characteristic astrocyte morphology) and NeuN (red) within VTA. c Example voltage-clamp recordings (left V_h = −55 mV, right V_h = +10 mV). The inward current (left), but not the outward current (right), increases when astrocytes are photoactivated (ChR2, blue hash marked horizontal line above traces) concurrently with electrical stimulation (Elec.) of neuronal afferents (red traces) in a GLT-1^+/+ mouse. ChR2 photoactivation alone produces no response (blue trace; see Supplementary Fig. 8). d Summarized data from VTA GABA neurons where ChR2 stimulation on VTA astrocytes increases peak EPSC amplitude (GLT-1^+/+, paired t-test t₍₉₎ = 4.457, P = 0.016, n = 10 cells) without altering IPSCs (paired t-test t₍₇₎ = 1.318, P = 0.22, n = 8 cells, 4 mice). e Five overlaid traces of a cell-attached recording from a VTA GABA neuron. Electrical stimulation of neuronal afferents in the VTA (black traces) elicits a pause in firing activity (gray area). However, when ChR2 on VTA astrocytes (gfaABC1D::ChR2^VTA) is photoactivated concurrently with electrical stimulation (red traces) the pause duration is decreased (see spikes within the gray area, which is pause duration of electrical stimulation alone). f Summarized data from GABA neurons demonstrating a decrease in pause duration (paired t-test t₍₄₎ = 3.345, P = 0.028, n = 5 cells, 2 mice) elicited by ChR2 stimulation on VTA astrocytes. g The increase in EPSC amplitude is absent in a GLT-1 cKO^{VTA Astrocyte} mouse (GLT-1^f/f, gfaABC1D::Cre^VTA + gfaABC1D::ChR2^VTA). h Summarized data comparing the change in the EPSC amplitude between GLT-1^+/+ (GLT-1^+/+, gfaABC1D::ChR2^VTA), GLT-1^f/f (GLT-1^f/f, gfaABC1D::ChR2^VTA), and GLT-1 cKO^{VTA Astrocyte} mice (One-way ANOVA, F_(2,22) = 7.352, P = 0.0036; post-hoc Tukey’s test GLT-1^+/+ vs GLT-1^f/f P > 0.05, GLT-1^+/+ vs GLT-1 cKO^{VTA Astrocyte} P < 0.05, GLT-1^f/f vs GLT-1 cKO^{VTA Astrocyte} P < 0.05, n = 11, n = 6, n = 8 cells, respectively). **P < 0.01, *P < 0.05; Error bars indicate ± SEM

Fig. 3

Astrocytic GLT-1 increases inhibition of VTA dopamine neurons. a Schematic showing the microcircuitry of the VTA and manipulations performed. b Example voltage-clamp recordings (left V_h = −55 mV, right V_h = +10 mV). For dopamine neurons, the outward current (right), not the inward current (left), increases when ChR2 is photoactivated concurrently with electrical stimulation (Elec.) of neuronal afferents (red traces) in a GLT-1^+/+ mouse. Photoactivation of ChR2 alone elicits no response (blue trace, and see Supplementary Fig. 8). c Summarized data from VTA dopamine neurons where ChR2 stimulation on VTA astrocytes increases peak IPSC amplitude (GLT-1^+/+, paired t-test t₍₁₂₎ = 5.894, P = 0.0001, n = 13 cells, 8 mice) without altering EPSCs (paired t-test t₍₉₎ = 1.129, P = 0.28, n = 10 cells, 8 mice). d Nine overlaid traces of a cell-attached recording and e summarized data from VTA dopamine neurons demonstrating an increase in pause duration (paired t-test t₍₄₎ = 2.815, P = 0.04, n = 5 cells, 2 mice) when ChR2 stimulation on VTA astrocytes is delivered concurrently with electrical stimulation of neuronal afferents (red traces; gray areas are the pause duration following electrical stimulation alone). f, g The increase in IPSC amplitude (red trace, top) is abolished by pharmacologically blocking GLT-1 with 300 µM DHK (bottom traces). h When GLT-1 is conditionally knocked out from VTA astrocytes there is no change in IPSC caused by activating ChR2 on astrocytes. i Summarized data comparing the change in IPSC amplitude between GLT-1^+/+ (GLT-1^+/+, gfaABC1D::ChR2^VTA), GLT-1^f/f (GLT-1^f/f, gfaABC1D::ChR2^VTA), GLT-1^f/f, and GLT-1 cKO^{VTA Astrocyte} mice (One-way ANOVA, F_(2,27) = 7.732, P = 0.0022; post-hoc Tukey’s test GLT-1^+/+ vs GLT-1^f/f P > 0.05, GLT-1^+/+ vs GLT-1 cKO^{VTA Astrocyte} P < 0.05, GLT-1^f/f vs GLT-1 cKO^{VTA Astrocyte} P < 0.05, n = 15, n = 7, n = 8 cells, respectively). **P < 0.01, *P < 0.05; Error bars indicate ± SEM

Fig. 4

Increase in dopamine neuron inhibition is dependent on glutamate receptor activation. a Schematic showing the microcircuitry of the VTA and manipulations performed. b On the left is a voltage-clamp (V_h = +10 mV) recording from a VTA dopamine neuron where pairing electrical simulation (Elec., black traces) with ChR2 stimulation on VTA astrocytes (red traces) causes an increase in IPSC amplitude. On the right, when the GABA_A antagonist, bicuculline (30 µM), is added during the same recording the IPSC is abolished. c Data summarizing the effect of bicuculline. d The increase in IPSC amplitude was eliminated by glutamate receptor antagonists (10 µM NBQX + 100 µM AP-5) with no effect on the amplitude of the IPSC elicited by electrical stimulation alone (black traces) (paired t-test, P = 0.99, n = 8 cells, 4 mice). e Data summarizing the effect of glutamate receptor antagonists. *P < 0.05

Fig. 5

Astrocyte-dependent increase in dopamine neuron inhibition is mediated by VTA GABA neurons. a Experimental manipulation. b Immunocytochemistry demonstrating TH positive cells (red) and eNpHR expression (yellow). On the right are the two images merged together. c Voltage-clamp traces (V_h = +10 mV) from a VTA dopamine neuron. ChR2 stimulation during electrical stimulation (Elec + ChR2, red trace; ChR2 stimulated at 405 nm to minimize activation of eNpHR) increases the IPSC amplitude relative to electrical stimulation alone (Elec, black trace). When ChR2 on astrocytes is activated concurrently with eNpHR activation on GABA neurons during electrical stimulation (Elec + ChR2 + eNpHR, green trace) the increase in IPSC amplitude was eliminated. Electrical stimulation with GABA neuron hyperpolarization alone (Elec + eNpHR, yellow trace) has no difference relative to electrical stimulation along with ChR2 on astrocytes and eNpHR activation on GABA neurons (Elec + ChR2 + eNpHR, green trace). d The increase in IPSC amplitude when ChR2 was photoactivated with electrical stimulation (Elec + ChR2, red) was blocked when GABA neurons were simultaneously hyperpolarized (Elec + eNpHR + ChR2, green). e IPSC amplitudes during electrical stimulation with GABA neuron hyperpolarization (Elec + eNpHR, yellow) were not changed when ChR2 stimulation was added (Elec + eNpHR + ChR2, purple-yellow; paired t-test t₍₆₎ = 0.1247,P = 0.90, n = 7 cells, 4 mice). ***P < 0.001; Error bars indicate ± SEM

Fig. 6

Real-time avoidance is dependent on VTA astrocytes and mediated by local GABA neurons. a Experimental manipulations for panels b–e. b Three track plots from one ChR2^{VTA Astrocyte} mouse (GLT-1^+/+, gfaABC1D::ChR2^VTA). (left) The mouse avoided the side of the chamber where ChR2 was stimulated in the VTA (473 nm Laser on). When the laser was left off both sides (Laser off) the mouse showed no avoidance. (right) With laser stimulation on the opposite side, the mouse then avoided the opposite side. c Summarized data showing avoidance of the laser-paired side (Repeated Measures ANOVA, F_{(1.413, 11.3)} = 18.43, P = 0.0006; post-hoc Tukey’s test, stimulation side 1 vs no stimulation P < 0.05, stimulation side 1 vs stimulation side 2 P > 0.05, Stimulation side 2 vs no stimulation P < 0.05, n = 9 mice). d (left) Track plot from a mouse expressing GFP on VTA astrocytes (GFP^{VTA Astrocyte}; GLT-1^f/f, gfaABC1D::GFP^VTA) shows no avoidance. (Right) track plot from a mouse expressing ChR2 on, but with GLT-1 conditionally knocked out from, VTA astrocytes (GLT-1 cKO^{VTA Astrocyte}; GLT-1^f/f, gfaABC1D::ChR2^VTA + gfaABC1D::Cre^VTA) showing no avoidance. e Summarized data showing avoidance elicited by ChR2 activation in GLT-1^+/+ and GLT-1^f/f mice. However, loss of GLT-1 from VTA astrocytes eliminates avoidance (One-way ANOVA, F_(3,49) = 16.11, P = 0.0001; post-hoc Tukey’s test, GLT-1^f/f (n = 10) relative to: GLT-1 cKO^{VTA Astrocyte} P < 0.05, GLT-1^+/+ P > 0.05 (n = 18), GFP P < 0.05 (n = 14); cKO^{VTA Astrocyte} (n = 11) relative to: GLT-1^+/+ P > 0.05, GFP P < 0.05; GLT-1^+/+ P < 0.05). f Experimental manipulations for panels g–h. g Track plots from a VGAT^CRE::eNpHR^VTA mouse with ChR2 stimulation on VTA astrocytes (eNpHR^{VTA GABA} + ChR2^{VTA Astrocyte}). Avoidance observed during ChR2 stimulation (left; ChR2 stimulated at 405 nm to minimize activation of eNpHR) was blocked when astrocytes were photoactivated concurrently with GABA neuron hyperpolarization (right). h Summarized data demonstrating GABA neuron hyperpolarization blocks avoidance elicited by ChR2 on VTA astrocytes.**P < 0.01, *P < 0.05; Error bars indicate ± SEM

Fig. 7

VTA Astrocytes mediate learned avoidance and prevent learned preference for reward. a Experimental design and manipulations for panels b–d. b Track plots showing that conditioning with ChR2 photoactivation on VTA astrocytes elicits learned avoidance in a GLT-1^+/+ mouse (gfaABC1D::ChR2^VTA). c The learned avoidance is absent in a mouse where GLT-1 was removed from VTA astrocytes (gfaABC1D::ChR2^VTA, GLT-1 cKO^{VTA Astrocyte}). d Comparing the difference scores in mice expressing either ChR2 (GLT-1^+/+ and GLT-1^f/f, GLT-1 cKO^{VTA Astrocyte}) or GFP (gfaABC1D::ChR2^eYFP) in VTA astrocytes. e Experimental design and manipulations for panels f–h. f Track plots showing astrocyte activation paired with cocaine administration does not elicit either learned avoidance or preference in a GLT-1^f/f mouse (gfaABC1D::ChR2^VTA). g However, when GLT-1 is removed from VTA astrocytes (gfaABC1D::ChR2^VTA, GLT-1 cKO^{VTA Astrocyte}) a mouse displays robust learned preference for VTA astrocyte ChR2 photoactivation paired with cocaine administration. h Comparing the difference scores in mice expressing either ChR2 (GLT-1^+/+ and GLT-1^f/f, GLT-1 cKO^{VTA Astrocyte}) or GFP (gfaABC1D::ChR2^eYFP) in VTA astrocytes. **P < 0.01, *P < 0.05; Error bars indicate ± SEM

Fig. 8

GLT-1 removal from VTA astrocytes disrupts endogenous avoidance behavior. a Schematic depicting experimental design. b Track plots from a GLT-1^f/f mouse with only a fluorescent reporter (eYFP) on astrocytes displaying normal thigmotaxis in an open field (left), and a GLT-1 cKO^{VTA Astrocyte} (GLT-1^f/f, gfaABC1D::Cre^VTA) mouse with increased tendency to explore the center of the open field (right). c, d GLT-1 cKO^{VTA Astrocyte} mice spent more time exploring the center of the open field without affecting locomotor activity (d; unpaired t-test t₍₁₆₎ = 1.981, P = 0.065, n = 9 mice). **p < 0.01; Error bars indicate ± SEM

Fig. 9

A neuron-astrocyte circuit for avoidance. In a sequence of events, (1) reduction of glutamate uptake by the astrocytic glutamate transporter, GLT-1, leads to (2) increased glutamate concentration at synapses on VTA GABA neurons. (3) The resultant increase in GABA release at synapses on DA neurons inhibits DA neuron activity, eliciting (4) real-time and learned avoidance behavior