Dopamine Regulates Aversive Contextual Learning and Associated In Vivo Synaptic Plasticity in the Hippocampus

Abstract

Dopamine release during reward-driven behaviors influences synaptic plasticity. However, dopamine innervation and release in the hippocampus and its role during aversive behaviors are controversial. Here, we show that in vivo hippocampal synaptic plasticity in the CA3-CA1 circuit underlies contextual learning during inhibitory avoidance (IA) training. Immunohistochemistry and molecular techniques verified sparse dopaminergic innervation of the hippocampus from the midbrain. The long-term synaptic potentiation (LTP) underlying the learning of IA was assessed with a D1-like dopamine receptor agonist or antagonist in ex vivo hippocampal slices and in vivo in freely moving mice. Inhibition of D1-like dopamine receptors impaired memory of the IA task and prevented the training-induced enhancement of both ex vivo and in vivo LTP induction. The results indicate that dopamine-receptor signaling during an aversive contextual task regulates aversive memory retention and regulates associated synaptic mechanisms in the hippocampus that likely underlie learning.

Keywords: LTP; fear conditioning; inhibitory avoidance; in vivo recording; memory; passive avoidance; ventral tegmental area.

Fig. 1. Midbrain DAT-positive neurons project to the CA1

(A) Didactic vector map of the AAV-EF1a-DIO-synaptophysin:GFP virus constructed for use in the viral tracing experiments. This DJ8 vector specifically expressed synaptophysin, which was primarily targeted in the fiber terminals of infected neurons. (B) AAV-EF1a-DIO-synaptophysin:GFP was injected into the midbrain dopamine area (VTA/SNc) of Slc6a3^ires-cre/+ mice and WT mice served as controls. (C) Confocal image (10x) taken from the VTA of a Slc6a3^ires-cre/+ mouse injected with Synaptophysin-GFP virus. The light blue is DAPI (Vector laboratories, Burlingame CA), and green fluorescence indicates the reporter from Synaptophysin-GFP. Cell bodies and processes were labeled in the VTA. (D) As a positive control, a confocal image is shown of the dense innervation of the ventral striatum, which receives innervation from VTA DA neurons. (E) A second positive control showing green DA-terminal puncta in the medial prefrontal cortex (PFC). (F) Image from Slc6a3^ires-cre/+ mouse indicating direct dopaminergic projections from the midbrain DA area revealed as green puncta in the CA1. (G) The CA1 of a Slc6a3^+/+ (WT littermate) shows practically no green fluorescence. (H) Confocal image (40x) of DAT terminals projecting directly to cell bodies in the PCL of the hippocampal CA1. (I) Quantification of the number of green pixels from Slc6a3^ires-cre/+ mice compared to WT littermates: 0.83% ± 0.03, n = 5 for Slc6a3^ires-cre/+; 0.09% ± 0.004, n = 3 for WT. (J) Locus coeruleus image illustrating Synaptophysin-GFP (DAT) labeled terminals, but no cell somas, indicating that these are not DAT positive cell bodies. SO: stratum oriens, PCL: pyramidal cell layer, SR: stratum radiatum. Scale bars: 100 μm (C, D, E); 50 μm (F, G, J); 25 μm (H).

Figure 2. Dopaminergic terminals and axons in the CA1 show high co-localization with tyrosine hydroxylase (TH)

(A) Confocal images of the CA1 field of a Slc6a3^ires-cre/+ mouse injected with Synaptophysin-Ruby-Red virus in the VTA/SNc area. Images show punctate Synaptophysin-Ruby-Red signal (in red) mainly along the pyramidal cell layer of CA1 and TH immunoreactivity (in green). A merged overlay of the two signals, shows a high degree of co-localization (yellow-orange) of Synaptophysin-Ruby-Red with TH (Ruby-Red/TH double-positive: 93 ± 2%, n = 6) providing further evidence of the existence of midbrain dopaminergic innervation of the hippocampus. SO: stratum oriens, PCL: pyramidal cell layer, SR: stratum radiatum. Scale bars, 50 μm. (B) High magnification images of the white boxed area in (A), showing individual and composite images of the different labels including the co-localization (Merged). Scale bars, 5 μm.

Fig. 3. Dopamine regulated acquisition of a IA long-term memory

(A) Didactic illustration of the mouse in the IA training and testing box. The mouse was placed on the light side, and after a short time a door opened enabling the mouse to move to the dark side where it could be foot shocked. (B) IA training typically elevated the approach latency when tested 24 h later (saline injection, black bars). A low dose of SCH 23390 (0.05 mg/kg, i.p., red bars) just before training blocked this effect at different footshock intensities: n = 9 - 11/group. The results are the following: 0.3 mA, 0.4 mA, and 0.8 mA: Sal 0.3 mA = 52.1 ± 13.80 s, SCH 0.3 mA = 16.44 ± 3.19 s, n = 10, 9; Sal 0.4 mA = 78.5 ± 16.47 s, SCH 0.4 mA = 7.4 ± 2.73 sec, n =10,10; Sal 0.8 mA = 106.17 ± 23.54 s, SCH 0.8mA = 28.78 ± 7.04 sec, n = 12, 9. (C) The same SCH dose (red bars) did not block short-term retention in the IA task: 1 h retention, SCH, n = 32 and Sal, n = 15, p > 0.05; 3 h retention, n = 5, 5, p > 0.05. (D) Local bilateral infusion of 1-μl SCH (1 mg/ml concentration, red bars) into the CA1 prior to training significantly reduced memory retention in the IA paradigm 24 h after training: n = 7, 9 p < 0.05. The insert indicates post hoc staining, indicating the location of the infusion of SCH into the dorsal CA1 region. (E) Systemic injections of a high dose of SCH (0.2 mg/kg) immediately after IA training did not impair retention of the footshock at the 24 h interval: Training = 11.78 ± 1.36 s, Testing = 91.20 ± 21.77 s; p < 0.01. (F) Systemic injection of DA D1-like receptor agonist, SKF 81297 (0.9 mg/kg), enhanced retention of a footshock (0.4 mA) when tested at the 72 h retention interval: Sal = 45.90 ± 13.80 s, n = 10; SKF = 92.5 ± 12.64 s, n = 32; p < 0.05. (G) Two doses of β2-adrenergic antagonist, Tim (i.p.), prior to IA training did not block the retention of a footshock: approach latency was 83.3 ± 16.7 s, p < 0.01, after 10 mg/kg Tim; and was 103.1 ± 21.0 s, p < 0.01 after 20 mg/kg Tim , n = 10, 10.

Fig. 4. IA training increased the AMPA/NMDA current ratio in CA1 pyramidal neurons, but not in dentate gyrus granule cells

(A) Diagram illustrating the whole-cell recording from CA1 pyramidal neurons in hippocampal slices. The stimulating electrode (S) was placed on the Schaffer collateral path and the recording electrode (R) onto a CA1 pyramidal neuron. (B) and (D) Representative traces of AMPA (grey traces) and NMDA (black traces) receptor mediated whole-cell currents recorded from CA1 pyramidal neurons from control (unshocked, left) and IA (shocked, right) mice, which were decapitated either 1.5 h after training (B), or after testing (D). (C) and (E) The average of AMPA/NMDA ratios from CA1 pyramidal neurons are plotted. IA training significantly increased the AMPA/NMDA ratio in slices prepared from animals decapitated 1.5 h after training: control vs. IA, 0.59 ± 0.06 vs. 0.93 ± 0.06, n = 12, 6, p < 0.01), but not in slices prepared from animals after testing: control vs. IA, 0.66 ± 0.10 vs. 0.59 ± 0.10, n = 6, 6, p > 0.05. (F) Diagram illustrating the whole-cell recording from dentate gyrus granule cells. The stimulating electrode (S) was placed on the medial perforant path (MPP) and the recording electrode (R) onto a DG granule cell. (G) and (I) Representative traces of AMPA (grey) and NMDA (black) receptor mediated whole-cell currents recorded in granule cells from control (unshocked, left) and IA (shocked, right) mice, which were sacrificed either 1.5 h after training (G) or after testing (I). (H) and (J) IA training had no effect on the AMPA/NMDA ratio measured from dentate gyrus granule cells at these time points: 1.5 h, control vs. IA, 0.77 ± 0.06 vs. 0.63 ± 0.07, n = 10, 11, p > 0.05; post-test, control vs. IA, 0.70 ± 0.10 vs. 0.57 ± 0.06, n = 5, 7, p > 0.05.

Fig. 5. IA training enhanced the slope of the in vivo fEPSP of the CA3-CA1 circuit

(A) Illustration of the post-experimental positioning of stimulating (left) and recording (right) electrodes with white circles for walk-through controls and black circles for IA recordings. All of the illustrated sites indicate successful recordings that produced stable input-output curves on all the recording days and remained within the CA1 (Recording) and Schaffer collateral (Stimulate) pathway. (B) Representative traces from the CA1 taken from mice recorded before (1) and after (2) IA training (black lines) or walk-through controls (unshocked, gray lines). (C) Walk-through control mice that were exposed to the experimental IA chamber but did not receive footshock did not have a significant change in the fEPSP slope (white circles, p > 0.05). Mice trained in the IA paradigm showed a significant increase in the fEPSP slope (p < 0.01). Following testing, the fEPSP slope returned to baseline levels: controls, IA both p > 0.05.

Fig. 6. D1/D5 receptor antagonist, but not β-adrenergic receptor antagonist, blocked IA training-induced increases in the in vivo fEPSP slope

(A) Illustration of the post-experimental positioning of stimulating (left) and recording (right) electrodes with grey circles for Tim and red circles for SCH 23390 treated mice. (B) Representative traces from the CA1 taken from mice recorded before (1) and after (2) IA training after injections of either SCH (red traces) or Tim (grey traces). (C) Mice treated with Tim (10 mg/kg, i.p., grey circles) showed a significant increase in the CA1 fEPSP slope: p < 0.01. IA + vehicle treated mice are re-represented from Fig. 5 for comparison (black circles). Mice treated with SCH (0.05 mg/kg, i.p., red circles) showed no change in the fEPSP slope relative to baseline recordings: n = 7, p > 0.05.

Fig. 7. D1/D5 receptor antagonist, but not β-adrenergic receptor antagonist, blocked the increased AMPA/NMDA current ratio in CA1 pyramidal neurons from mice 1.5 h after training

Representative traces of AMDA (grey) or NMDA (black) mediated currents after treatment with SCH 23390 + IA (A), SCH only (B), Tim + IA (D), or Tim only (E). (C) Summary of the average data for IA only (black bar), SCH only (open red bar), and SCH + IA (red bar): F _{(2, 16)} = 15.15, p < 0.01, ANOVA; Tukey post hoc multiple comparison test, IA vs. SCH only: 0.93 ± 0.06 vs. 0.55 ± 0.06, p < 0.01; IA vs. SCH + IA: 0.93 ± 0.06 vs. 0.52 ± 0.04, p < 0.01. (F) Summary of the average data for IA only (black bar), Tim only (Tim, open black bar), and Tim + IA (grey bar). TIM treatment did not prevent IA from increasing the AMPA/NMDA ratio: F _{(2, 18)} = 10.62, p < 0.01, ANOVA; Tukey post hoc multiple comparison test, IA vs. TIM only, 0.93 ± 0.06 vs. 0.51 ± 0.04, p < 0.01; TIM + IA vs. TIM only, 0.82 ± 0.07 vs. 0.51 ± 0.04, p < 0.01. The IA data (black bars) were duplicated from Fig. 4C.