Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Nov 21.
Published in final edited form as: Neuron. 2018 Oct 11;100(4):916–925.e3. doi: 10.1016/j.neuron.2018.09.028

Dopamine neurons reflect the uncertainty in fear generalization

Yong Sang Jo 1, Gabriel Heymann 1, Larry S Zweifel 1,2,*
PMCID: PMC6226002  NIHMSID: NIHMS1509348  PMID: 30318411

Summary:

Generalized fear is a maladaptive behavior in which non-threatening stimuli elicit a fearful response. Here we demonstrate that discrimination between predictive and non-predictive threat stimuli is highly sensitive to probabilistic discounting and increasing threat intensity in mice. We find that dopamine neurons of the ventral tegmental area (VTA) encode both the negative valence of threat-predictive cues and the certainty of threat prediction. As fear generalization emerges the dopamine neurons that are activated by a threat predictive cue (CS+) decrease the amplitude of activation and an equivalent signal emerges to a non-predictive cue (CS−). Temporally precise enhancement of dopamine neurons during threat conditioning to high-threat levels or uncertain threats can prevent generalization. Moreover, phasic enhancement of genetically captured dopamine neurons activated by threat cues can reverse fear generalization. These findings demonstrate the dopamine neurons reflect the certainty of threat prediction that can be used to inform and update the fear engram.

eTOC

Jo et al., demonstrate that dopamine neurons of the ventral tegmental area encode the prediction of future threat. They demonstrate that elevated threat levels and increased uncertainty of threats induce fear generalization that is reflected in the dopamine code.

Introduction

In Pavlovian conditioning a neutral sensory cue paired with an unconditioned stimulus (US) will become a conditioned stimulus (CS) that elicits behavioral responses appropriate to the US (Pavlov, 1927). As neutral stimuli become perceptually similar to a CS they can also elicit conditioned responses (Dunsmoor and Paz, 2015; Pavlov, 1927). In fear generalization when the intensity of a US increases, the discrimination between a US-predictive signal (CS+) and a non-predictive safety signal (CS−) decreases, even if the CS− maintains dissimilarity to the CS+ (Ghosh and Chattarji, 2015; Sanford et al., 2017).

Dopamine neurons of the ventral midbrain play an essential role in Pavlovian associative processes by providing a teaching signal relating to the value and saliency of stimuli (Bromberg-Martin et al., 2010; Clark et al., 2012; Dayan and Berridge, 2014; Montague, 2007; Schultz et al., 2017). In addition to their role in reinforcement learning, dopamine plays a critical role in fear conditioning (Borowski and Kokkinidis, 1996; Fadok et al., 2009; Guarraci and Kapp, 1999; Pezze and Feldon, 2004; Zweifel et al., 2011). Several studies have demonstrated that subsets of dopamine neurons are activated by threatening or noxious stimuli (Brischoux et al., 2009; Chiodo et al., 1979; Mantz et al., 1989; Schultz and Romo, 1987; Wang and Tsien, 2011), as well as threat-predictive stimuli following fear conditioning (Gore et al., 2014; Guarraci and Kapp, 1999). Genetic suppression of the capacity of dopamine neurons to engage phasic increases in firing promotes sustained generalized fear and anxiety following exposure to moderate intensity threat (Zweifel et al., 2011). Additionally, enhancement or suppression of dopamine receptor signaling in the central nucleus of the amygdala (CeA) can either promote or diminish fear discrimination, respectively (De Bundel et al., 2016). Although considerable evidence points to a role for dopamine neurons in discriminatory fear learning, the manner in which dopamine signals contribute to this process remains unresolved.

Uncertainty is a potent stressor (Peters et al., 2017) and dopamine neurons are sensitive to both stress and stress-associated signaling pathways (Margolis et al., 2003; Saal et al., 2003; Ungless et al., 2003; Wanat et al., 2008). Dopamine neurons are also sensitive to uncertainty in reinforcement learning (Fiorillo et al., 2003). Threatening stimuli are inherently stressful and stress has long been known to negatively affect behavioral performance and learning (Yerkes and Dodson, 1908). Thus, as threat intensity increases elevated stress levels would be predicted to inhibit discriminatory learning processes resulting in fear generalization. However, whether there is a link between uncertainty and fear generalization is not known. Given the sensitivity of dopamine neurons to uncertainty and the observation of subsets of dopamine neurons being activated by threat-predictive cues, we hypothesized that these neurons may provide an important link between these two phenomena and yield new insight into the basis of fear generalization.

Results

Dopamine neuron activity tracks behavioral threat discrimination

To test whether dopamine neurons track uncertainty in fear conditioning and whether fear generalization associated with elevated threat levels parallel these responses, we tracked dopamine neuron responses using probabilistic and non-probabilistic fear conditioning paradigms. Mice expressing Cre-recombinase under the control of the endogenous dopamine transporter locus (Slc6a3IRES-Cre, or DAT-Cre; (Zhuang et al., 2005) were injected with conditional adeno-associated virus for expression of channelrhodopsin (Boyden et al., 2005)(AAV1-FLEX-ChR2-mCherry) and implanted with an optetrode array for recording (Figure S1A). Dopamine neurons were identified using an optical-tagging method (Lima et al., 2009) (Figure 1A–D). To test the sensitivity of threat discrimination to probabilistic discounting, mice were conditioned to auditory cues paired with a mild US that promotes discriminatory fear (0.3 mA), but with diminishing discrimination indices (DI = 1.0 to 0.4) in separate cohorts of implanted mice, or with a strong US that promotes generalization (0.5 mA) but with a high discrimination index (DI = 1.0) (Figure 1B). Analysis of discriminatory fear (CS+ freezing minus CS− freezing) in recorded mice demonstrated that animals discriminated the predictive and non-predictive cues at a DI = 1.0, 0.8, and 0.6, but not 0.4 following the first day of conditioning and discriminated at DI = 1.0 and 0.8, but not 0.6 or 0.4 following the second day of conditioning (Figure 1C and Figure S1B–E). Mice conditioned with a strong US intensity of 0.5 mA at DI = 1.0 did not discriminate following either day of conditioning (Figure 1C and Figure S1F).

Figure 1. Genetic isolation of dopamine neurons during fear conditioning.

Figure 1.

Open in a new tab

(A) Neural activity and freezing behavior were measured in distinct contexts in response to two auditory stimuli before (Pre) and after (Post) fear conditioning (FC). (B) Schematic of tone-US pairing to manipulate US probabilities associated with two tones. Among ten sets of tone A and B trials, high- and low-probability sets were pseudo-randomly ordered. Tone A was associated with US on high-probability sets, whereas tone B was paired with US on low-probability set(s). Tone A at P=1 and B at P=0 were interchangeably used as CS+ and CS−, respectively. (C) Discriminative freezing of recorded mice (N = 4–5 mice per condition per condition) calculated as freezing to tone A (CS+) minus freezing to tone B (CS−). Mice were conditioned with 0.3 mA foot shock with different DIs (1 – 0.4) or 0.5 mA foot shock (DI = 1). (D) Characteristics of light-evoked responses. A cluster analysis was performed based on spike probability and spike latency in response to 10 pulses of blue light (5-ms width, 20 Hz; 473 nm). The cluster in black was categorized as dopaminergic. Inset histograms show firing patterns of 5 representative VTA neurons (in red) in response to the light. Light-evoked waveforms in dopamine cells were correlated to spontaneous waveforms (Figure S1F).

Neurons were classified as being dopaminergic if they had a short latency light-evoked response (≤ 8 ms) and a high probability of generating a light-evoked action potential (P ≥ 0.6) (Figure 1D, Figure S1G). Genetically defined dopamine neurons were mostly non-responsive before fear conditioning, but exhibited phasic excitation or inhibition at the onset of tones after conditioning (Figure 2A–F and Figure S2A-E). These dopamine responses to tones were strongly modulated by uncertainty associated with probabilistic discounting. When behavioral discrimination and certainty were high (0.3 mA US, DI = 1, or 0.8) the excitation of a subset of dopamine neurons was significantly stronger in response to tone A than tone B. As the DI decreased to 0.6 and 0.4 to lower predictive certainty, excited responses to tone A were diminished and responses by the same neurons to tone B were reciprocally increased (Figure 2A–C). For a distinct subset of neurons displaying phasic inhibition, differential firing rates between the two tones were also observed at DI = 1 and 0.8 (Figure 2D–F). As the DI diminished, inhibition in response to tone B was significantly enhanced in the same cells that exhibited inhibition in response to tone A. The inhibited responses to tone A showed no change across DIs.

Figure 2. Dopamine neuron activity tracks fear discrimination.

Figure 2.

Open in a new tab

(A) Average excited dopamine neuron responses during diminishing certainty or increased US intensity (±S.E.M.). Shaded areas represent cue presentations. (B and C), Normalized excited responses to tone A (relative to pre-tone baseline) were negatively correlated with DI and intensity during post 1 (B; Pearson’s correlation, r = −0.53, ***p < 0.001) and post 2 tests (C; r = −0.63, ***p < 0.001). Normalized responses to tone B we positively correlated in post 2 (***r = 0.65, p values < 0.001). (D) Average inhibited dopamine neuron responses during diminishing certainty or increased US intensity (±S.E.M.). (E and F) Normalized firing rates of inhibited dopamine neurons in response to tone B are negatively correlated with DI during both post 1 (E; r = −0.77, ***p < 0.001) and 2 tests (F; r = −0.65, ***p < 0.001). (G-I) Relationships between dopamine neuron activity and freezing behavior. Magnitudes of tone A (G) and B (H) responses relative to basal firing before tone onset and preferential firing to tone A (CS+) over B (CS−) (I) and were computed from pooled electrophysiology data as (auROC – 0.5) × 2, where auROC is the area under the ROC curve. Freezing behavior was also expressed as (CS+ minus CS−) / (CS+ plus CS−). Spearman’s rank correlation tests revealed that discriminative freezing behavior had significant positive relationships with dopaminergic excitation to the CS+ (G; r = 0.73, ***p < 0.001), dopaminergic inhibition to the CS− (H; r = 0.57, ***p < 0.001), and differential excitation between the CSs (I; r = 0.72, ***p < 0.001). By contrast, discriminative performance was negatively correlated with dopaminergic excitation to CS− (H; r = −0.46, ***p < 0.001) and differential inhibition (I; r = −0.73, ***p < 0.001). No correlation was found between freezing behavior and dopaminergic inhibition to CS+.

To further establish a relationship between dopamine neuron responses to predictive and non-predictive cues and behavior discrimination, we performed receiver-operator characteristic (ROC) analysis. This analysis confirmed that freezing behavior was significantly correlated with the following: 1) the peak of the excited response to the CS+ (Tone A) was related to behavioral discrimination (Figure 2G), but the trough of the inhibited response was not (Figure 2G). 2); the peak of the excited response to the CS− (Tone B) was related to generalization (Figure 2H), as was the trough of the inhibited response (Figure 2H); and 3) the differential responses to the CS+ and CS− of both excited and inhibited cells was related to discrimination (Figure 2I).

Dopamine neurons have been shown to display prediction errors following reinforcement learning when CS+ stimuli are presented and rewards are omitted (Schultz et al., 1997). Because dopamine neuron responses were recorded during fear recall in which CS stimuli were presented alone, we assessed whether a similar prediction error occurs. Neurons across conditioning groups that were either activated by the CS+ or CS− or inhibited by the CS+ or CS− were pooled and activity at cue offset was analyzed. In activated neurons US omission did not result in a significant change in activity (Figure S2F); however, inhibited neurons showed an increased firing rate above baseline at cue offset (Figure S2F).

Dopamine activity during CS+ facilitates fear learning

To begin to establish the role of dopamine neuron activation for discriminatory fear learning, we selectively inhibited all dopamine neurons of the VTA at the onset of the CS+ during fear conditioning using the inhibitory opsin Jaws (Chuong et al., 2014). To determine the optimal inhibition protocol, DAT-Cre mice were injected with AAV1-FLEX-Jaws-GFP and an optetrode array was implanted above the VTA (Figure 3A). Inhibition of dopamine neurons using a square pulse of red light (1 s) resulted in a large rebound excitation (Figure 3B and 3C). However, inhibiting these neurons using a ramp-down illumination (1-s on with 1-s ramp-down) greatly attenuated this effect (Figure 3B and 3C). Based on these observations, we selectively inhibited all dopamine neurons bilaterally at the onset of the CS+ during fear conditioning using the 1-s on with 1-s ramp-down method (Figure 3D). Using this protocol Inhibition of dopamine neurons with Jaws approximates the responses observed in a subset of dopamine neurons that are inhibited by the CS+ following conditioning and should prevent the activation of the subset that show increased activity to the CS+.

Figure 3. Dopamine activation is necessary for acquisition of fear memory.

Figure 3.

Open in a new tab

(A) Schematic of optetrode recording and viral-mediated expression of Jaws for analysis of inhibition. (B) Two representative dopamine neurons in response to a square pulse of red light (1 s, left, top and bottom) and two representative dopamine neurons in response to a ramp-down illumination of red light (1-s on followed by 1-s ramp-down, right, top and bottom). A square pulse suppressed neuronal activity, but rebound excitation occurred after the pulse. A gradual ramp-down illumination decreased this rebound excitation. (C) Population activity of 34 dopamine neurons in response to the two types of red light. (D) Schematic of bilateral cannulation for Jaws-mediated inhibition of dopamine cells. Ramp-down red light (1-s on and 1-s ramp-down) was paired with the onset of CS+, but not CS−. (E) Differential freezing responses between CSs during retention tests showed that bilateral inhibition of dopamine cells significantly impaired discriminative fear learning (significant group × testing interaction in a two-way ANOVA, F(2,36) = 4.32, p = 0.02; Subsequent Bonferroni pairwise comparisons, *p values < 0.05; N = 10 mice per group). (F and G) Comparisons of CS+ (F) and CS− (G) responses between Jaws- and mCherry-expressing groups. A significant group difference was found for CS+ responses (significant group effect in a two-way ANOVA, F(1,18) = 4.46, *p = 0.049).

Mice were fear conditioned at 0.3 mA, DI = 1 to promote threat discrimination. Bilateral inhibition of dopamine neurons at the onset of the CS+ during conditioning significantly attenuated discriminatory threat learning (Figure 3E, Figure S3). This reduction in threat discrimination was associated with a reduction in freezing to the CS+, without altering freezing to the CS− (Figure 3F and 3G). These results indicate that activation of a subset of neurons is important for discriminatory fear learning.

Enhancement of CS+ dopamine signals attenuates fear generalization

Threat discrimination is correlated with the peak amplitude of the dopamine response to the CS+, the differential activation of dopamine neurons to the CS+ and CS−, and the differential inhibition to the CS+ and CS− (Figure 4A). To further probe the functional relationship between dopamine neurons and fear discrimination, we optogenetically manipulated these cells unilaterally (Figure 4B) by conditionally expressing either stimulatory or inhibitory opsins (AAV1-FLEX-ChR2-mCherry and AAV1-FLEX-Jaws-GFP, respectively) in the VTA of DAT-Cre mice (Figure S4A and S4B). The rationale for unilateral manipulation is as follows: 1) If the amplitude of dopamine excitation to the CS+ is a critical signal that facilitates behavioral discrimination, then a unilateral increase in this signal at a strong US intensity that promotes generalization should prevent generalized fear. 2) If equivalent activation in response to the CS+ and CS−promotes fear generalization, then increasing dopamine neuron activity at CS− onset at a mild US intensity that promotes discrimination should cause generalization. Alternatively, increasing activation of dopamine neurons to both the CS+ and CS− during high intensity foot shock conditioning should not prevent generalization. 3) If the equivalent inhibition to CS+ and CS− is a dominant signal that promotes fear generalization, then unilateral inhibition of dopamine cells at CS− onset at a mild US intensity that promotes discrimination should cause generalization.

Figure 4. Amplitude of dopamine neuron excitation promotes acquisition of discriminatory fear.

Figure 4.

Open in a new tab

(A) Relationships between dopamine neuron responses to CSs and fear discrimination: 1) Amplitude of excited responses to CS+, 2) differential excited response to CS+ and CS−, and 3) differential inhibited response to CS+ and CS−. (B) Optogenetical manipulations of dopamine signaling during fear conditioning. Cre-dependent Jaws and ChR2 were unilaterally injected into the VTA of Dat-cre mice to inhibit and stimulate dopamine cells, respectively. (C) Protocol for ChR2 stimulation: 5-ms blue light at 20 Hz for 1 s at CS onset, or Jaws inhibition: red light, 1-s on followed by 1-s ramp-down at CS onset. (D-F) ChR2 enhancement of activation to either CS+ (D) or both CSs (F), but not CS− alone (E) during fear conditioning with 0.5 mA stimulus improved discrimination (significant interactions in two-way ANOVAs, (D) F(4,54) = 3.25, p = 0.02, (F) F(4,54) = 3.63, p = 0.01; follow-up Bonferroni pairwise comparisons, *p values < 0.05; N = 10 mice per group per condition). Unilateral manipulation of inhibition with Jaws had no effect under any conditions. (G and H) Neither manipulation of stimulation or inhibition produced differences in fear discrimination in mice conditioned at 0.3 mA (N = 10 mice per group per condition). (I) Dopamine stimulation improved discriminative freezing under increased uncertainty (DI=0.4, 0.3 mA foot shock; significant group effect in a two-way ANOVA, F(1,18) = 4.74, *p = 0.04; N = 10 mice per group).

To test these different possibilities, dopamine neurons were manipulated at the onset of the CS+ and/or CS− during fear conditioning using the following parameters, blue light, 1 s, 20Hz for ChR2 and red light, 1-s on with 1-s ramp-down for Jaws (Figure 4C). To determine whether manipulation of dopamine signals can alter fear generalization, mice were fear conditioned with 0.5 mA foot shock, DI = 1. Unilateral stimulation of dopamine neurons at the onset of the CS+ promoted fear discrimination (Figure 4D, Figure S4C and S4D). Unilateral inhibition with Jaws at CS+ onset in a separate cohort of mice under these conditions had no effect (Figure 4D, Figure S4E). Similarly unilateral stimulation or inhibition at the onset of the CS− in different groups of mice had no effect on fear generalization (Figure 4E, Figure S4F–H). Unilateral stimulation at the onset of the CS+ and CS− promoted discrimination, but unilateral inhibition to both in a separate cohort of mice had no effect (Figure 4F, Figure S4I–K). These results suggest that the amplitude of the activation to the CS+ is a critical signal that promotes discrimination.

To establish whether manipulation of dopamine signals can alter fear discrimination, mice were fear conditioned with 0.3 mA foot shock, DI = 1 and unilaterally stimulated or inhibited at the onset of the CS+ or CS−. Neither unilateral stimulation nor inhibition of dopamine neurons at the onset of the CS+ or CS− in district cohorts of mice altered fear discrimination (Figure 4G and 4H, Figure S5A–F). To confirm that unilateral inhibition of dopamine neurons is sufficient to induce negative affect, we monitored real-time conditioned place aversion. Unilateral inhibition resulted in a significant place avoidance of the red light-paired side of the chamber (Figure S6A and S6B). We also found that unilateral inhibition facilitated extinction of instrumental lever pressing for food reward (Figure S6C and S6D). Thus, in specific behavioral contexts unilateral inhibition can serve as a dominant signal to modulate behavior. Overall, we find that only the experiments in which dopamine neurons were unilaterally stimulated at the onset of the CS+ during fear conditioning with US intensity of 0.5mA resulted in a behavioral change (Figure S5G).

Our electrophysiology data shows that dopamine neuron responses following 0.5 mA foot shock, DI = 1 are similar to those in mice conditioned with 0.3 mA foot shock, DI = 0.4 (Figure 2A), indicating that fear conditioning with elevated US intensity or increased uncertainty results in similar behavioral outcomes that are also similar at the level of dopamine neuron responses. To determine whether increasing dopamine activity unilaterally in mice conditioned under increased uncertainty can also prevent fear generalization, we fear conditioned mice with 0.3 mA foot shock, DI = 0.4 and stimulated dopamine cells at the onset of CS+ (1 s, 20Hz). Like increasing CS+ signals under elevated US intensity, increasing this signal under increased uncertainty also enhanced discrimination (Figure 4I, Figure S5H and S5I).

Enhancement of CS activated dopamine neurons alone reverses generalization

A caveat to unilateral stimulation is that it will increase activity at the onset of the CS+, but also attenuate or block the inhibited responses on that side of the VTA. To address this, we genetically captured CS responsive VTA dopamine neurons using a Tet-tag strategy (Liu et al., 2012; Zhang et al., 2015) (Figure 5A–B). Following injection of two viral vectors (AAV-fos-tTA and AAV-TRE-DIO-ChR2-mCherry), mice were maintained on doxycycline (Dox) diet to prevent ChR2 expression. After two weeks, mice were conditioned with a 0.5 mA US then removed from Dox for 3 days. On the third day off-Dox mice were placed in the testing environment and presented with no stimuli (control) or exposed to 3CS+ and 3 CS− presentations. Both groups were returned to the Dox diet following the exposure. On the subsequent 2 days, mice were reconditioned with blue light stimulation at CS+ onset as above (Figure 5B). Mice exposed to CS presentations on the third day off Dox showed a significantly higher number of dopamine neurons expressing ChR2 than mice only exposed to the testing environment on the third day off Dox, or mice that were continuously maintained on Dox (Figure 5C–D). The control group showed generalized freezing behavior throughout all post-conditioning tests. The experimental group also displayed non-discriminative freezing during the first post-conditioning test. However, subsequent reconditioning with dopamine stimulation significantly improved discriminative freezing (Figure 5E and Figure S7A and S7B).

Figure 5. Manipulation of CS excited dopamine neurons reverses generalization upon reconditioning.

Figure 5.

Open in a new tab

(A) Cartoon depicting AAV viruses used to conditionally express ChR2 in an activity-dependent manner. The presence of Dox prevents c-fos-promoter-driven tTA from binding TRE site, and thus ChR2 expression. (B) Schematic of fear conditioning procedure, regiment of Dox treatment, and timing of ChR2 stimulation (20 Hz, 5ms, 1s, at CS+ onset). (C) ChR2-mCherry expression (magenta) in tyrosine hydroxylase (TH)-positive neurons (green) in the VTA of experimental and control groups, as well as mice with Dox continuously present. Scale bar: 250μm. (D) Percentages of dopamine cells expressing ChR2 (significant group effect in one-way ANOVA, F(2,17) = 67.83, p < 0.001; subsequent post-hoc comparisons, *p < 0.05, ***p values < 0.001; N = 8 for exp and con, N = 4 for on dox). (E) Differential freezing behavior during retention tests (significant group × testing interaction in a two-way ANOVA, F(3,42) = 3.46, p = 0.03; follow-up Bonferroni pairwise comparisons, *p values < 0.05).

Dopamine signals in the CeA modulate fear discrimination

The amygdala is a key site of fear learning (LeDoux, 2003). Recent evidence suggests that the CeA plays an important role in fear generalization (Botta et al., 2015) and anxiety (Ahrens et al., 2018). Dopamine signaling in the CeA has also been shown to be potent modulator of discriminatory fear (De Bundel et al., 2016). To determine whether phasic activation of dopamine projections to the CeA is sufficient to prevent generalized fear responses, we unilaterally implanted an optic fiber in the CeA of mice expressing ChR2 in dopamine cells and optically stimulated dopamine terminals (1 s, 20 Hz) at the onset of CS+ during fear conditioning with a 0.5 mA foot shock. Similar to unilateral cell body stimulation, unilateral stimulation of dopamine terminals in the CeA prevented generalized fear responses (Figure 6A–C and Figure S8A–C). Unlike unilateral VTA cell body stimulation that is well known to promote reward seeking behavior (Tsai et al., 2009; Witten et al., 2011), unilateral stimulation of dopamine terminals in the CeA did not promote real-time conditioned place preference (Figure S8D). Thus, the enhancement of threat discrimination observed with CeA terminal stimulation appears unrelated to reinforcing effects of dopamine. Unlike CeA terminal stimulation, stimulation of dopamine projections to the nucleus accumbens (NAc) is reinforcing (Steinberg et al., 2014). We find that stimulation of these terminals does not attenuate fear generalization (Figure 6D–F, Figure S8E–G).

Figure 6. Dopamienergic projections to the CeA are critical for fear discrimination.

Figure 6.

Open in a new tab

(A) Unilateral AAV1-FLEX-ChR2-mCherry injection into the VTA and an optic fiber implanted above the CeA. (B) Representative image of ChR2-mCherry fibers in the CeA and placement of optic fiber (OF). Scale bar: 250μm. (C) ChR2 stimulation of terminals during CS+ presentation improved discrimination in mice conditioned with 0.5 mA foot shock (significant group × testing interaction in a two-way ANOVA, F(2,36) = 4.97, p = 0.01; follow-up Bonferroni pairwise comparisons, *p < 0.05, **p < 0.05; N = 10 mice per group). (D) Unilateral AAV1-FLEX-ChR2-mCherry injection into the VTA and an optic fiber implanted above the NAc. (E) Representative image showing the placement of an OF above ChR2-mCherry fibers in the NAc. Scale bar: 500μm. (F) Terminal stimulation in the NAc failed to improve fear discrimination (N = 10 mice per group). (G) Bilateral AAV1-FLEX-Jaws-GFP injection into the VTA and an optic fiber implanted above the CeA. (H) Representative image of Jaws-GFP fibers in the CeA. Scale bar: 250μm. (I) Bilateral inhibition of dopaminergic fibers in the CeA significantly disrupted normal discriminative fear responses seen by the control group (significant interaction in a two-way ANOVA, F(2,40) = 5.91, p = 0.006; subsequent Bonferroni pairwise comparisons, *p < 0.05, ***p < 0.001; N = 11 per group).

We next sought to determine whether dopamine projections to the CeA are critical for fear discrimination. To test this, mice were bilaterally injected with AAV1-FLEX-Jaws-GFP and bilaterally implanted with optic fibers over the CeA (Figure 6E and 6F, Figure S9A). Mice were conditioned with 0.3 mA foot shock, DI = 1 and terminals were inhibited at the onset of the CS+ (1-s on followed by 1-s ramp-down). Bilateral inhibition of dopamine terminals in the CeA significantly reduced discriminatory fear compared to AAV1-FLEX-GFP injected controls (Figure 6G, Figure S9B and S9C). Unlike bilateral cell body inhibition, terminal inhibition in the CeA did not significantly reduce freezing to the CS+ following conditioning (Figure S9D), but resulted in a significant increase in freezing to the CS− (Figure S9E).

Discussion

Accurate threat discrimination is essential for an animal’s ability to avoid deleterious outcomes while maintaining neutrality towards non-threatening stimuli and incentive motivation for appetitive stimuli. Unpredictable receipt of an aversive stimulus or deficits in discriminatory fear learning can increase generalized anxiety and fear responses (Grillon, 2002; Grillon and Davis, 1997). Consistent with this impaired contingency awareness model of generalized fear and anxiety, we previously reported that mice with reduced phasic dopamine activation have deficits in cued-fear discrimination (Jones et al., 2015; Zweifel et al., 2009) and develop persistent anxiety phenotypes following an aversive experience (Zweifel et al., 2011). Here, we demonstrate that non-discriminatory fear associated with increased uncertainty or increased US intensity has similar effects on dopamine neuron responses to CS+ and CS− cues. Artificially increasing dopamine neuron activity during CS+ presentation during conditioning can attenuate deficits in discriminatory fear associated with either increased uncertainty or increased US intensity. Collectively, these data provide a link between uncertainty and fear generalization.

Dopamine neurons reflect accurate threat discrimination

We observed two distinct populations of dopamine responses to CS stimuli following fear conditioning, consistent with previous findings (Guarraci and Kapp, 1999). One population was transiently inhibited by the CS+ stimulus, but not the CS− when mice displayed discriminatory fear responses. As uncertainty or US intensity increased, inhibitory responses to the CS+ were equivalent in discriminating and non-discriminating mice. However, as threat discrimination decreased equivalent inhibitory responses to the CS− emerged in cells that showed inhibited responses to the CS+. These results are consistent with the assignment of negative valence to the CS− cue (Bromberg-Martin et al., 2010). Activation of inhibited cells at the omission of the US is also consistent with these neurons being related to dopamine neurons that display prediction errors following reinforcement learning (Schultz et al., 1997)

Like dopamine cells inhibited by the CS+, cells activated by the CS+ did not respond to the CS− in mice displaying threat discrimination. Unlike inhibited cells, the magnitude of the activated neuron responses to the CS+ was not equivalent between discriminating and non-discriminating mice, but rather decreased in a manner proportional to the degree of generalization. As threat discrimination decreased in our probabilistic fear conditioning paradigm, or as US intensity increased, we observed equivalently small activated responses to the CS− in cells activated by the CS+. These findings are remarkably similar to those reported for dopamine neuron responses to CS+ cues in a probabilistic reinforcement task (Fiorillo et al., 2003). These results indicate that subsets of dopamine neurons are sensitive to uncertainty associated with appetitive or threat-predictive cues. Thus, subsets of dopamine neurons report either a negative valence and negative outcome prediction error (Schultz et al., 1997) or predictive certainty (Fiorillo et al., 2003).

The exact nature of how fear generalization associated with elevated US intensity and increased CS/US uncertainty are linked remains elusive. However, based on our observations, we propose that fear generalization is the result of impairment in associative learning that results in the aberrant assignment of salience and negative valence to both CS+ and CS− cues. Under high threat conditions in which the CS+ is always paired with the US, repeated US presentations elevate stress levels that impairs learning. This impairment results in aberrant salience and negative valence assigned to the CS−. Because the CS− signal is not temporally linked to the US under these conditions a type probabilistic discounting occurs. Thus, the animal incorrectly assumes that both the CS+ and the CS− are predictive of the US outcome, but they are correct only half of the time which results in further uncertainty. Future experiments designed to elucidate the neural circuits and signaling pathways that connect US intensity, stress, and uncertainty to fear generalization will prove highly valuable.

Phasic dopamine signals regulate threat discrimination

Bilateral inhibition of dopamine neurons at the onset of the CS+ during conditioning significantly impaired freezing to the CS+ when mice were probed for fear recall without manipulating dopamine signals during test sessions. These results are consistent with previous data demonstrating mice lacking the ability to synthesize dopamine have reduced fear following conditioning (Fadok et al., 2009). Because Jaws-mediated inhibition was similar to the inhibitions observed in a subset of dopamine neurons, these results further suggest that inhibited dopamine neurons alone are not sufficient to promote fear. Thus, some level of dopamine is required for both fear learning and fear discrimination. This is further supported by our previous findings that reducing, but not eliminating phasic dopamine using a genetic strategy increased susceptibility to the development of generalized fear following an aversive experience (Zweifel et al., 2011).

Artificial enhancement of dopamine signals at the onset of the CS+ promoted fear discrimination under conditions of increased uncertainty and increased US intensity. Selectively enhancing these signals in dopamine neurons previously activated by CS presentations in generalizing mice reversed generalization upon reconditioning. These findings indicate that phasic dopamine signals facilitate discriminatory fear learning and can update previously established fear engrams. Artificially introducing a negative valence signal unilaterally to the CS− using Jaws did not promote fear generalization, consistent with the notion that these negative valence signals alone are not sufficient to evoke fear-related responses. Unilateral Jaws-mediated inhibition to the CS+ also did not alter fear discrimination, indicating that an intact phasic dopamine signal on the contralateral side is sufficient to promote discriminatory fear learning. The function of the subset of inhibited dopamine neurons in fear-related learning remains unclear. Our observation that inhibited dopamine neurons show a type of prediction error upon US omission suggests that these signals may play an important role in extinction learning. Consistent with this notion, a recent study found that manipulation of inhibited dopamine signals during fear extinction significantly impacts this learning process (Luo et al., 2018).

The role of dopamine projections to the CeA for discriminatory fear

We find that stimulation of dopamine terminals in the CeA at the onset of the CS+ is sufficient to enhance threat discrimination under strong US intensity conditioning. In agreement with this observation, bilateral inhibition of dopamine terminals as the onset of the CS+ under mild US intensity conditioning is sufficient to promote generalization. Importantly, unlike bilateral inhibition of dopamine neuron cell bodies, inhibition of terminals in the CeA did not significantly reduce conditioned responses to the CS+, but rather promoted a generalized increase in conditioned responses to the CS−.

Based on previous observations that the CeA plays a key role in generalized fear responses (Botta et al., 2015; Sanford et al., 2017), and that dopamine signaling in the CeA regulates threat discrimination (De Bundel et al., 2016), we reasoned that the CeA is a critical site for dopamine predictive signals to promote discrimination. Our results support this hypothesis; however, it should be noted that dopamine projections to the CeA represent a relatively small proportion of the overall dopamine neuron population. Thus, it is highly likely that activated neurons project to numerous other brain regions. This is supported by previous observations that dopamine synthesis is required in both projections to the amygdala and the ventral striatum for long-term fear memory formation (Fadok et al., 2010). Furthermore, previous studies have demonstrated that dopamine neurons projecting to the nucleus accumbens lateral shell and medial prefrontal cortex undergo plasticity following an aversive experience (Lammel et al., 2011), and selective targeting of inputs to dopamine neurons projecting the medial prefrontal cortex can elicit aversive behavior (Lammel et al., 2012). Future experiments designed to address how the different targets of dopamine neurons coordinate conditioned threat discrimination and the mechanisms that increase uncertainty reflected by dopamine neurons under high threat conditions will provide an important next step in the resolution of threat generalization.

STAR Methods

Contact for reagent and resource sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Larry Zweifel (larryz@uw.edu).

Experimental Model and Subject Details

Male and female Slc6a3Cre/+ (DAT-Cre) mice (3–7 month old) were housed on a 12 h light/dark cycle (lights on at 7 am) with ad libitum access to food and water. All experiments were conducted during the light phase of the cycle under the guidelines of University of Washington’s Institutional Animal Care and Use Committee.

Method details

Viral production.

AAV viral vectors (serotype 1) were generated as previously described (Gore et al., 2013). Briefly, pAAV shuttle plasmids for ChR2 (pAAV-Ef1α-DIOChR2-mCherry or pAAV-TRE(Tight)-FLEX-ChR2-mCherry), Jaws (pAAV-CAG-FLEX-Jaws-GFP), and tTA (pAAV-fos-tTA) were co-transfected with the packing plasmid pDG1 (Nature Biotechnologies) into HEK293T/17J cells (ATCC). 72 hours post-transfection cells were harvested and viral vector was liberated by repeated freeze/thaw. Vectors were purified by multiple rounds of cesium chloride gradient centrifugation, sucrose gradient centrifugation, and dialysis. Viral vectors were stereotaxically injected at a final titer of 1–3 × 1012 particles/ml.

Surgery.

Mice were anesthetized under isoflurane (1.5 – 4%) and placed in a stereotaxic frame (Model 1900, David Kopf Instruments). The skull was exposed and adjusted to place bregma and lambda on the same horizontal plane. The anterior-posterior coordinates of target brain structures were normalized to the standard mouse bregma-lambda distance of 4.21 mm using a correction factor (F = Bregma-Lambda distance/4.21). After small burr holes were drilled, virus (0.5 μl) was unilaterally injected into the VTA (3.25 mm posterior, 0.5 mm lateral, and 4.5 mm ventral to bregma) at a rate of 0.2 μl/min. Subsequently a microdrive or an optic fiber (200 μm inner core, 0.22 numerical aperture) was implanted dorsal to the VTA (3.7 mm to the brain surface and 4.0 mm to bregma, respectively). The microdrive contained four tetrodes (25 μm diameter tungsten wire; California Fine Wire) and an optic fiber. Tetrode tips were cut and gold-plated to reach impedances of 200–400 kΩ tested at 1 kHz. The distance between the fiber and tetrode tips was shorter than 500 μm. For stimulation of dopaminergic terminals, an optic fiber was implanted into the CeA (1.2 mm posterior, 2.9 mm lateral, and 4.5 mm ventral to bregma). Only optic fibers that had light loss not exceeding 30% were used thought the study. The implants were secured in place with anchoring screws and dental cement. Mice were allowed to recover for 2–4 weeks following surgery before experimentation.

Fear conditioning.

Age-matched littermate mice were randomly assigned to ChR2, mCherry, or Jaws groups. For controls, half of the mice received red light stimulation and half received blue light stimulation. A t-test was used to confirm that the two control subgroups did not differ significantly from each other. Acquisition of fear memory was conducted in four identical chambers (context B; 21.6 × 17.8 × 12.7 cm; Med Associates Inc.) placed inside sound-attenuating boxes. Each chamber was made of two aluminum and two Plexiglas side walls. A speaker was mounted on one of the aluminum walls for delivering auditory tones. The floor consisted of 24 stainless steel rods which were wired to a scrambled shock generator. To add a context-specific odor, the chamber was cleaned with a 1% acetic acid solution between mice and a stainless-steel pan containing the same solution was placed under the grid floor. Retention of Fear memory was tested in a different environment (context A) where white plastic panels were inserted into the chamber to cover the walls and floor. The chamber was wiped with 70% ethanol between animals.

Conditioning procedures started with measuring baseline freezing levels in response to two tones (4 and 12 kHz; 10 s) which were randomly assigned as CS+ and CS−. The assignment of CS type was counterbalanced within each group. Mice were habituated to context A for 2 min and three CS+ and three CS− trials were presented in a pseudo-randomized order (CS+ was always presented on the first trial; intertrial interval = 60 s). The mice were returned to their home cages. About two hours later, the mice were placed and habituated to context B for 2 min. Then the animals received ten CS+ trials which co-terminated with a 0.5-s foot shock (0.3 or 0.5 mA). CS+ presentations were interleaved with ten CS− trials which were never paired with the US. On the next day, mice underwent the identical conditioning procedures which included a post-conditioning test followed by a conditioning session. Only a post-conditioning test was given on the third day. When CS/US probabilities were manipulated, all conditioning procedures were identical except for tone-US pairing during the acquisition of fear memory. Specifically, ten sets of tone A and B trials were serially ordered (A,B,A,B, A,B…), then probabilistic shock pairings for the different discrimination indices for high-probability and low-probability sets were pseudo-randomly selected with the first A,B set always assigned as A-paired, B-unpaired.

During the conditioning procedures, each mouse’s behavior was recorded using a video camera mounted above the chamber. Movement velocities were calculated offline by a video tracking software (Ethovision XT 8.5, Noldus Technology). Freezing behavior during the presentations of CS+ and CS− was scored if velocities were less than 0.75 cm/s for at least 1 s. The cutoff velocity for immobility was chosen based on the comparison between automatic and manual scoring of freezing behavior using a sample data set.

Single-unit recording.

Within the home cage, the microdrive was connected to a preamplifier linked to a Cheetah data acquisition system (Digital Lynx 4SX, Neruralynx). Continuously sampled data were amplified, filtered between 100 and 9,000 Hz, and digitized at 32 kHz. Discretely sampled data were filtered between 100 and 6,000 Hz and recorded when spikes exceeded a predetermined threshold. To identify dopamine neurons that expressed ChR2, 10 blue light pulses (473 nm; 5-ms long at 20 Hz; Laserglow technologies) were presented via the optic fiber of the microdrive. If at least two light-responsive units were observed out of four tetrodes, the light intensity (5–15 mW/mm2) was adjusted, so that light-evoked spike waveforms were similar to spontaneous ones. Then the mouse was placed in a brown recording box (27 × 18 × 8 cm). After a 10-min habituation period, ten CS+ and ten CS− presentations were randomly ordered. The duration of CS presentations was shortened to 3 s to minimize possible extinction effects on the subsequent recall test. At the end of the recording session, 10 blue light pulses were delivered 20 times in order to confirm the stability of the recorded units and quantify their firing characteristics in response to the light. If only one or no light-sensitive unit were found, all tetrodes were lowered in 80 μm increments until more than two light-sensitive units were encountered. These neuronal responses to CSs were recorded on the following day.

Immunohistochemistry.

After completion of all experiments, mice were deeply anesthetized with 50 mg/kg of Beuthanasia-D and transcardially perfused with phosphate-buffered saline (PBS) and 4% paraformaldehyde. Brains were extracted and fixed overnight in 4% paraformaldehyde, and then transferred to a 30% sucrose solution at 4°C for 72 hours. The brains were frozen and cut in coronal sections (30 μm) on a cryostat (Leica CM 1850). The sections were treated with a blocking solution (PBS containing 3% normal donkey serum and 0.3% Triton) for 1 hour and incubated overnight at 4°C with the following primary antibodies: 1) anti-tyrosine hydroxylase (monoclonal, 1:1000; Millipore, MAB318), and anti-dsRed (polyclonal, 1:1000; Clontech, 632496) for ChR2-injected mice, 2) anti-tyrosine hydroxylase (polyclonal, 1:1000; Millipore, AB152), and anti-GFP (monoclonal, 1:1000; Millipore, MAB3580) for Jaws-injected mice. Sections were washed three times, then incubated with secondary antibodies conjugated to AlexaFluor 488 or CY3 (donkey anti-rabbit or mouse, 1:200; Jackson Immunoresearch) for 1 hour at room temperature. The sections were washed three times and coverslipped with Fluoromount-G (Southern Biotech). Using a Nikon upright microscope, fluorescent images were taken to examine recording sites, fiber placements, and protein expression levels.

Quantification and statistical analysis

Single unit isolation was conducted using an Offline Sorter (Plexon). VTA single units were identified based on various waveform features, such as peak, valley, width, energy, and principal component (Jo et al., 2013; Jo and Mizumori, 2016). Only units showing good recording stability throughout the whole recording session were further analyzed using Matlab software (Mathworks). To classify dopamine neurons, a cluster analysis was performed based on spike probability and spike latency in response to blue light pulses. The cluster that had the highest correlation between light-evoked and spontaneous waveforms was considered as dopaminergic. The two types of waveforms were measured using continuously sampled data.

To examine dopamine responses to CSs, peri-event time histograms (PETHs; 50-ms bins) were constructed around the time of CS presentations. A dopamine cell was categorized as excited by a CS if its firing rates within the 0.5-s window after the onset of the CS were significantly higher than the basal average firing (2-s epoch before the CS) using a Wilcoxon signed-rank test. A dopamine neuron was considered as inhibited if its firing rates during a 0.8-s window (0.1 to 0.9 s) after the onset of a CS were significantly lower the basal firing activity. In order to compare the magnitude of CS responses across different fear conditioning procedures, neuronal firing rates were transformed to z-scores relative to the basal firing activity before CS onset. Average z-scores of excited and inhibited responses were measured within the same windows used for the categorization.

ROC curves for differential responses to CSs were calculated by comparing two distributions of firing rates between CS+ and CS− trials during the excited (0 to 0.5 s) and inhibited time windows (0.1 to 0.9 s). ROC curves for the magnitude of CS responses were computed by comparing the distribution of CS−evoked firing rates across trials to the distribution of basal firing rates (2 s prior to CS onset). The auROC values were then transformed to a dopamine firing index ranging from −1 to 1: (auROC – 0.5) × 2.

Statistical tests for electrophysiological and behavioral results were performed with mixed-design ANOVA that contained within-subjects variables (e.g., CS and test day) as well as between-subjects factors (e.g., group). Once significant interaction effects were found, Bonferroni corrected t-tests were used for post-hoc pairwise comparisons. Pearson’s Spearman’s correlation tests were used to establish a relationship between two variables. Two-tailed P values < 0.05 were considered statistically significant. Data were expressed as mean ± SEM unless otherwise indicated.

Data and software availability

All data and custom software that support the current findings are available upon request from the Lead Contact, Larry Zweifel (larryz@uw.edu).

Key resources table

REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
anti-tyrosine hydroxylase Millipore MAB318
anti-tyrosine hydroxylase Millipore AB152
anti-dsRed Clontech 632496
anti-GFP Millipore MAB3580
Secondary antibodies conjugated with AlexaFluor 488 or Cy3 Jackson Immunoresearch
Bacterial and Virus Strains
AAV-Ef1α-DIO-ChR2-mCherry University of Washington N/A
AAV-CAG-FLEX-Jaws-GFP University of Washington N/A
AAV-TRE-FLEX-ChR2-mCherry University of Washington N/A
AAV-fos-tTA University of Washington N/A
Experimental Models: Organisms/Strains
Slc6a3Cre/+ (DAT-Cre) mice Jackson laboratory 20080
Software and Algorithms
Offline sorter for cluster cutting Plexon v3.3.5
Ethovision for tracking mice movement Noldus XT8.5
Matlab script for data analysis Yong Sang Jo N/A
Med Associates script for fear conditioning Yong Sang Jo N/A
Open in a new tab

Supplementary Material

1

Highlights.

  • Dopamine responses to CS reflect threat discrimination

  • Generalization is associated with reduced predictive coding by dopamine neurons

  • Selective enhancement of dopamine predictive coding prevents fear generalization

Acknowledgments:

This work was supported by National Institutes of Health grants P50-MH106428, R01-MH094536, and R01-MH104450 (LSZ). We thank Dr. James Allen for assistance in the production of viral vectors and members of the Zweifel lab for thoughtful discussion.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of interests: The authors have no competing interests.

References:

  1. Ahrens S, Wu MV, Furlan A, Hwang GR, Paik R, Li H, Penzo MA, Tollkuhn J, and Li B (2018). A Central Extended Amygdala Circuit That Modulates Anxiety. I. Neurosci 38, 5567–5583. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Borowski TB, and Kokkinidis L (1996). Contribution of ventral tegmental area dopamine neurons to expression of conditional fear: effects of electrical stimulation, excitotoxin lesions, and quinpirole infusion on potentiated startle in rats. Behav. Neurosci 110, 1349–1364. [DOI] [PubMed] [Google Scholar]
  3. Botta P, Demmou L, Kasugai Y, Markovic M, Xu C, Fadok JP, Lu T, Poe MM, Xu L, Cook JM, et al. (2015). Regulating anxiety with extrasynaptic inhibition. Nat. Neurosci 18, 1493–1500. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Boyden ES, Zhang F, Bamberg E, Nagel G, and Deisseroth K (2005). Millisecond-timescale, genetically targeted optical control of neural activity. Nat. Neurosci 8, 1263–1268. [DOI] [PubMed] [Google Scholar]
  5. Brischoux F, Chakraborty S, Brierley DI, and Ungless MA (2009). Phasic excitation of dopamine neurons in ventral VTA by noxious stimuli. Proc. Natl. Acad. Sci. U.S.A 106, 4894–4899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bromberg-Martin ES, Matsumoto M, and Hikosaka O (2010). Dopamine in motivational control: rewarding, aversive, and alerting. Neuron 68, 815–834. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chiodo LA, Caggiula AR, Antelman SM, and Lineberry CG (1979). Reciprocal influences of activating and immobilizing stimuli on the activity of nigrostriatal dopamine neurons. Brain Res. 176, 385–390. [DOI] [PubMed] [Google Scholar]
  8. Chuong AS, Miri ML, Busskamp V, Matthews GA, Acker LC, Sorensen AT, Young A, Klapoetke NC, Henninger MA, Kodandaramaiah SB, et al. (2014). Noninvasive optical inhibition with a red-shifted microbial rhodopsin. Nat. Neurosci 17, 1123–1129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Clark JJ, Hollon NG, and Phillips PE (2012). Pavlovian valuation systems in learning and decision making. Curr. Opin. Neurobiol 22, 1054–1061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dayan P, and Berridge KC (2014). Model-based and model-free Pavlovian reward learning: revaluation, revision, and revelation. Cogn. Affect. Behav. Neurosci 14, 473–492. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. De Bundel D, Zussy C, Espallergues J, Gerfen CR, Girault JA, and Valjent E (2016). Dopamine D2 receptors gate generalization of conditioned threat responses through mTORC1 signaling in the extended amygdala. Mol. Psych 21, 1545–1553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dunsmoor JE, and Paz R (2015). Fear Generalization and Anxiety: Behavioral and Neural Mechanisms. Biol. Psych 78, 336–343. [DOI] [PubMed] [Google Scholar]
  13. Fadok JP, Darvas M, Dickerson TM, and Palmiter RD (2010). Long-term memory for pavlovian fear conditioning requires dopamine in the nucleus accumbens and basolateral amygdala. PloS one 5, e12751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Fadok JP, Dickerson TM, and Palmiter RD (2009). Dopamine is necessary for cue-dependent fear conditioning. J. Neurosci 29, 11089–11097. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Fiorillo CD, Tobler PN, and Schultz W (2003). Discrete coding of reward probability and uncertainty by dopamine neurons. Science 299, 1898–1902. [DOI] [PubMed] [Google Scholar]
  16. Ghosh S, and Chattarji S (2015). Neuronal encoding of the switch from specific to generalized fear. Nat. Neurosci 18, 112–120. [DOI] [PubMed] [Google Scholar]
  17. Gore BB, Soden ME, and Zweifel LS (2013). Manipulating gene expression in projection-specific neuronal populations using combinatorial viral approaches. Curr. Prot. Neurosci 65, 4.35 31–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gore BB, Soden ME, and Zweifel LS (2014). Visualization of plasticity in fear-evoked calcium signals in midbrain dopamine neurons. Learn. Mem 21, 575–579. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Grillon C (2002). Startle reactivity and anxiety disorders: aversive conditioning, context, and neurobiology. Biol. Psych 52, 958–975. [DOI] [PubMed] [Google Scholar]
  20. Grillon C, and Davis M (1997). Fear-potentiated startle conditioning in humans: explicit and contextual cue conditioning following paired versus unpaired training. Psychophysiology 34, 451–458. [DOI] [PubMed] [Google Scholar]
  21. Guarraci FA, and Kapp BS (1999). An electrophysiological characterization of ventral tegmental area dopaminergic neurons during differential pavlovian fear conditioning in the awake rabbit. Behav. Brain. Res 99, 169–179. [DOI] [PubMed] [Google Scholar]
  22. Jo YS, Lee J, and Mizumori SJ (2013). Effects of prefrontal cortical inactivation on neural activity in the ventral tegmental area. J. Neurosci 33, 8159–8171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Jo YS, and Mizumori SJ (2016). Prefrontal Regulation of Neuronal Activity in the Ventral Tegmental Area. Cereb. Cortex 26, 4057–4068. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Jones GL, Soden ME, Knakal CR, Lee H, Chung AS, Merriam EB, and Zweifel LS (2015). A genetic link between discriminative fear coding by the lateral amygdala, dopamine, and fear generalization. eLife 4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lammel S, Ion DI, Roeper J, and Malenka RC (2011). Projection-specific modulation of dopamine neuron synapses by aversive and rewarding stimuli. Neuron 70, 855–862. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lammel S, Lim BK, Ran C, Huang KW, Betley MJ, Tye KM, Deisseroth K, and Malenka RC (2012). Input-specific control of reward and aversion in the ventral tegmental area. Nature 491, 212–217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. LeDoux J (2003). The emotional brain, fear, and the amygdala. Cell. Mol. Neurobiol 23, 727–738. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lima SQ, Hromadka T, Znamenskiy P, and Zador AM (2009). PINP: a new method of tagging neuronal populations for identification during in vivo electrophysiological recording. PloS one 4, e6099. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Liu X, Ramirez S, Pang PT, Puryear CB, Govindarajan A, Deisseroth K, and Tonegawa S (2012). Optogenetic stimulation of a hippocampal engram activates fear memory recall. Nature 484, 381–385. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Luo R, Uematsu A, Weitemier A, Aquili L, Koivumaa J, McHugh TJ, and Johansen JP (2018). A dopaminergic switch for fear to safety transitions. Nat. Comm 9, 2483. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mantz J, Thierry AM, and Glowinski J (1989). Effect of noxious tail pinch on the discharge rate of mesocortical and mesolimbic dopamine neurons: selective activation of the mesocortical system. Brain Res. 476, 377–381. [DOI] [PubMed] [Google Scholar]
  32. Margolis EB, Hjelmstad GO, Bonci A, and Fields HL (2003). Kappa-opioid agonists directly inhibit midbrain dopaminergic neurons. J. Neurosci 23, 9981–9986. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Montague PR (2007). Neuroeconomics: a view from neuroscience. Funct. Neurol 22, 219–234. [PubMed] [Google Scholar]
  34. Pavlov IP (1927). Conditional reflexes: an investigation of the physiological activity of the cerebral cortex (Oxford, England: Oxford Univ. Press; ). [Google Scholar]
  35. Peters A, McEwen BS, and Friston K (2017). Uncertainty and stress: Why it causes diseases and how it is mastered by the brain. Prog. Neurobiol 156, 164–188. [DOI] [PubMed] [Google Scholar]
  36. Pezze MA, and Feldon J (2004). Mesolimbic dopaminergic pathways in fear conditioning. Prog. Neurobiol 74, 301–320. [DOI] [PubMed] [Google Scholar]
  37. Saal D, Dong Y, Bonci A, and Malenka RC (2003). Drugs of abuse and stress trigger a common synaptic adaptation in dopamine neurons. Neuron 37, 577–582. [DOI] [PubMed] [Google Scholar]
  38. Sanford CA, Soden ME, Baird MA, Miller SM, Schulkin J, Palmiter RD, Clark M, and Zweifel LS (2017). A Central Amygdala CRF Circuit Facilitates Learning about Weak Threats. Neuron 93, 164–178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Schultz W, Dayan P, and Montague PR (1997). A neural substrate of prediction and reward. Science 275, 1593–1599. [DOI] [PubMed] [Google Scholar]
  40. Schultz W, and Romo R (1987). Responses of nigrostriatal dopamine neurons to high-intensity somatosensory stimulation in the anesthetized monkey. J. Neurophys 57, 201–217. [DOI] [PubMed] [Google Scholar]
  41. Schultz W, Stauffer WR, and Lak A (2017). The phasic dopamine signal maturing: from reward via behavioural activation to formal economic utility. Curr. Opin. Neurobiol 43, 139–148. [DOI] [PubMed] [Google Scholar]
  42. Steinberg EE, Boivin JR, Saunders BT, Witten IB, Deisseroth K, and Janak PH (2014). Positive reinforcement mediated by midbrain dopamine neurons requires D1 and D2 receptor activation in the nucleus accumbens. PloS one 9, e94771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tsai HC, Zhang F, Adamantidis A, Stuber GD, Bonci A, de Lecea L, and Deisseroth K (2009). Phasic firing in dopaminergic neurons is sufficient for behavioral conditioning. Science 324, 1080–1084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Ungless MA, Singh V, Crowder TL, Yaka R, Ron D, and Bonci A (2003). Corticotropin-releasing factor requires CRF binding protein to potentiate NMDA receptors via CRF receptor 2 in dopamine neurons. Neuron 39, 401–407. [DOI] [PubMed] [Google Scholar]
  45. Wanat MJ, Hopf FW, Stuber GD, Phillips PE, and Bonci A (2008). Corticotropin-releasing factor increases mouse ventral tegmental area dopamine neuron firing through a protein kinase C-dependent enhancement of Ih. J. Physiol 586, 2157–2170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Wang DV, and Tsien JZ (2011). Convergent processing of both positive and negative motivational signals by the VTA dopamine neuronal populations. PloS one 6, e17047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Witten IB, Steinberg EE, Lee SY, Davidson TJ, Zalocusky KA, Brodsky M, Yizhar O, Cho SL, Gong S, Ramakrishnan C, et al. (2011). Recombinase-driver rat lines: tools, techniques, and optogenetic application to dopamine-mediated reinforcement. Neuron 72, 721–733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Yerkes RM, and Dodson JD (1908). The relation of strength of stimulus to rapidity of habit-formation. J. Comp. Neurol. Psycho 18, 459–482. [Google Scholar]
  49. Zhang Z, Ferretti V, Guntan I, Moro A, Steinberg EA, Ye Z, Zecharia AY, Yu X, Vyssotski AL, Brickley SG, et al. (2015). Neuronal ensembles sufficient for recovery sleep and the sedative actions of alpha2 adrenergic agonists. Nat. Neurosci 18, 553–561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Zhuang X, Masson J, Gingrich JA, Rayport S, and Hen R (2005). Targeted gene expression in dopamine and serotonin neurons of the mouse brain. J. Neurosci. Meth 143, 27–32. [DOI] [PubMed] [Google Scholar]
  51. Zweifel LS, Fadok JP, Argilli E, Garelick MG, Jones GL, Dickerson TM, Allen JM, Mizumori SJ, Bonci A, and Palmiter RD (2011). Activation of dopamine neurons is critical for aversive conditioning and prevention of generalized anxiety. Nat. Neurosci, 14, 620–626. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Zweifel LS, Parker JG, Lobb CJ, Rainwater A, Wall VZ, Fadok JP, Darvas M, Kim MJ, Mizumori SJ, Paladini CA, et al. (2009). Disruption of NMDAR-dependent burst firing by dopamine neurons provides selective assessment of phasic dopamine-dependent behavior. Proc. Natl. Acad. Sci. U. S. A 106, 7281–7288. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1509348-supplement-1.pdf (2.1MB, pdf)

RESOURCES