Aversive stimuli bias corticothalamic responses to motivationally significant cues

  1. Federica Lucantonio
  2. Eunyoung Kim
  3. Zhixiao Su
  4. Anna J Chang
  5. Bilal A Bari
  6. Jeremiah Y Cohen  Is a corresponding author
  1. The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, United States

Abstract

Making predictions about future rewards or punishments is fundamental to adaptive behavior. These processes are influenced by prior experience. For example, prior exposure to aversive stimuli or stressors changes behavioral responses to negative- and positive-value predictive cues. Here, we demonstrate a role for medial prefrontal cortex (mPFC) neurons projecting to the paraventricular nucleus of the thalamus (PVT; mPFC→PVT) in this process. We found that a history of aversive stimuli negatively biased behavioral responses to motivationally relevant cues in mice and that this negative bias was associated with hyperactivity in mPFC→PVT neurons during exposure to those cues. Furthermore, artificially mimicking this hyperactive response with selective optogenetic excitation of the same pathway recapitulated the negative behavioral bias induced by aversive stimuli, whereas optogenetic inactivation of mPFC→PVT neurons prevented the development of the negative bias. Together, our results highlight how information flow within the mPFC→PVT circuit is critical for making predictions about motivationally-relevant outcomes as a function of prior experience.

Introduction

Effective decision making requires anticipating the outcomes associated with environmental stimuli. It also requires balancing the goals of a decision—for instance, acquiring a reward or avoiding a punishment—with uncertainty about the outcome of the decision. For example, when the outcome of a decision to approach or avoid a stimulus is ambiguous, the nervous system must weigh the cost of receiving a punishment, or missing out on a reward, with the benefit of obtaining the reward, or avoiding the punishment.

Many decisions are influenced by background emotional state. For example, both positive and negative mood affect decision making (Deldin and Levin, 1986; Wright and Bower, 1992; Bechara et al., 2000; Hockey et al., 2000; Dolan, 2002; Harding et al., 2004). This background state can be driven by prior experience. For example, prior exposure to aversive stimuli or stressors changes behavioral responses to ambiguous stimuli (Harding et al., 2004; Boleij et al., 2012; Rygula et al., 2014). How the balance between competing behaviors is weighed in the brain or how prior experience with an environment shifts this balance is still poorly understood.

The medial prefrontal cortex (mPFC) is critical for regulating cue-mediated behaviors in both appetitive and aversive domains (Ishikawa et al., 2008; Burgos-Robles et al., 2009; Sotres-Bayon and Quirk, 2010; Amemori and Graybiel, 2012; Burgos-Robles et al., 2013; Courtin et al., 2014; Sangha et al., 2014; Sparta et al., 2014; Burgos-Robles et al., 2017; Otis et al., 2017). Behavioral manifestations of appetitive and aversive conditioning correlate with changes in neural activity within the mPFC (Burgos-Robles et al., 2009; Peters et al., 2009; Amemori and Graybiel, 2012; Burgos-Robles et al., 2013; Moorman and Aston-Jones, 2015) and pharmacological or optogenetic manipulations of mPFC alter both reward-seeking and fear-related behaviors (Morgan and LeDoux, 1995; Blum et al., 2006; Corcoran and Quirk, 2007; Sierra-Mercado et al., 2011; Sangha et al., 2014; Sparta et al., 2014; Bari et al., 2019).

The mPFC has dense projections to subcortical structures involved in motivated behavior, including the paraventricular thalamus (PVT) (Vertes, 2004; Li and Kirouac, 2012). Like the mPFC, PVT is recruited by cues or contexts previously associated with rewarding or aversive outcomes (Beck and Fibiger, 1995; Schiltz et al., 2007; Yasoshima et al., 2007; Choi et al., 2010; Igelstrom et al., 2010; Haight and Flagel, 2014; Hsu et al., 2014; Do-Monte et al., 2015; Kirouac, 2015; Matzeu et al., 2015; Penzo et al., 2015; Li et al., 2016; Zhu et al., 2016; Do-Monte et al., 2017; McNally, 2021). PVT neurons are activated by multiple forms of stressors (Chastrette et al., 1991; Sharp et al., 1991; Cullinan et al., 1995; Bubser and Deutch, 1999; Spencer et al., 2004) and coordinate behavioral responses to stress (Hsu et al., 2014; Do-Monte et al., 2015; Penzo et al., 2015; Zhu et al., 2016; Do-Monte et al., 2017; Beas et al., 2018). On the other hand, under conditions of opposing emotional valence, PVT plays a role in multiple forms of stimulus-reward learning and PVT neurons have been reported to show reward-modulated responses (Schiltz et al., 2005; Igelstrom et al., 2010; Martin-Fardon and Boutrel, 2012; James and Dayas, 2013; Browning et al., 2014; Haight and Flagel, 2014; Li et al., 2016; Choi et al., 2019). Recent work suggests that PVT is especially critical during motivational conflict (Li et al., 2014; Choi and McNally, 2017; Zhu et al., 2018; Engelke et al., 2021). Inhibition across the anterior–posterior axis of the PVT disrupts arbitration between appetitive and aversive behaviors when they are in conflict but has no effect when these behaviors are assessed in isolation (Choi et al., 2019). Moreover, activity in mPFC neurons projecting to the PVT also suppresses both the acquisition and expression of conditioned reward seeking (Otis et al., 2017).

Taken together, these studies place the mPFC-to-PVT projection in a position to integrate information about positive and negative motivationally relevant cues and translate it into adaptive behavioral responses (McNally, 2021). How these projection-specific mPFC neurons regulate behavioral responses in ambiguous settings and how their neural activity may be altered upon presentation of an aversive stimulus is unknown.

To address these questions, we trained mice on a go/no-go discrimination task with sweet- and bitter-predicting odor cues. Subsequently, mixtures of varying proportions of those cues were presented to probe behavioral responses to ambiguous stimuli. We recorded extracellularly from single neurons in mPFC and from identified mPFC→PVT neurons. We then tested whether stimulating or inhibiting mPFC→PVT projections modulated behavioral responses to the learned stimuli and their ambiguous mixtures.

Results

Aversive stimuli negatively bias behavioral responses to motivationally-significant cues

To assess the effect of aversive stimulus history on decisions about motivationally-significant outcomes, we developed a go/no-go discrimination task in head-fixed mice consisting of four phases: conditioning, probe test, reversal, and a second probe test (Figure 1A). In the conditioning phase, four odor cues (A, B, C, and D, counterbalanced) were presented. Odor A predicted an appetitive sweet solution (3 μl of 5% sucrose water). Odor B predicted an aversive bitter solution (3 μl of 10 mM denatonium water). Odor C was associated with no reinforcement. Odor D predicted a punishment (an unavoidable air puff delivered to the mouse’s right eye) in mice assigned to the air-puff group and was associated with no reinforcement in mice assigned to the no-air-puff group. Each behavioral trial began with an odor (1 s; conditioned stimulus, CS), followed by a 1 s delay and an outcome (unconditioned stimulus, US). Mice showed essentially binary responses to cues that predicted sucrose or denatonium: they licked in anticipation of sucrose and did not in anticipation of denatonium (Figure 1B,C). We designed a probe test, in which, in addition to the four conditioning odors, we measured behavioral responses to ambiguous stimuli. We designed parametrically-varying mixtures of stimuli between appetitive and aversive solutions. We exposed mice to unreinforced mixtures of varying proportions of odors A and B: 85%A/15%B, 50%A/50%B, and 15%A/85%B. To ensure that behavioral responses to ambiguous stimuli were not driven by a mouse’s preference for a particular odor, after completion of the probe test, cue-outcome associations for odors A and B were reversed, and each mouse was re-tested in a second set of probe stimuli.

Figure 1 with 1 supplement see all
Behavioral responses to motivationally-significant predictive cues are modulated by history of aversive stimuli.

(A) Task design and experimental timeline. During ten conditioning sessions, odors (A, B, C, and D) predicted an appetitive sucrose solution, an aversive denatonium solution, no outcome, and either no outcome or an unavoidable air puff, respectively. During the first probe test, A and B were mixed in three different ratios: 85%B/15%A (light red), 50%B/50%A (gray), 15%B/85%A (light blue). After completion of 10 reversal training sessions, in which A and B contingencies were reversed, mice were re-trained in a second probe test. (B) Licking rates in no air puff (top, 7 mice) or air puff (bottom, 5 mice) groups across days, during odor and delay period, for sucrose (blue), denatonium (red), and no-outcome (black) trials. Dashed lines indicate reversals on day 11. (C) Licking behavior from a representative test session from a mouse without (left) and one with (right) exposure to air puffs. Color gradations between blue and red indicate odor mixtures as in (A). Gray bars indicate a period of odor presentation. Dashed lines indicate outcome delivery. (D) Licking rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for no air puff (black) and air puff (orange) groups, during odor and delay period. (E) Mean trial-by-trial licking rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for each ambiguous cue (light red, gray, light blue squares) for no air puff (black) and air puff (orange) groups, during odor and delay period. Line and error bars represent mean ± SEM.

As predicted, mice in both air-puff and no-air-puff groups quickly learned the CS-US associations: they showed increases in anticipatory licking responses to the positive, sucrose-predicting cue and in the delay before sucrose arrived across conditioning sessions, while withholding licking after sampling the negative, denatonium-predicting cue (Figure 1B). A three-factor ANOVA (session × cue × group) comparing licking during sucrose and denatonium cue presentation and delay period demonstrated a significant main effect of session (F1,9=29.74, p<0.01) and cue (F1,1=64.77, p<0.01). Moreover, mice exposed to air puffs responded to the sucrose-predicting odor with fewer licks (cue × group interaction, F1,1=7.84, p<0.05). During reversal learning, in which A and B contingencies were reversed, mice in both no-air-puff and air-puff groups quickly re-adapted to the new associations (Figure 1B). A three-factor ANOVA (session × cue × group) comparing licking rates during sucrose and denatonium cue presentation and delay period demonstrated a significant interaction between cue and session (F1,9=8.4, p<0.01).

At the end of conditioning and reversal training, mice received a single probe test session. Behavioral responses from the initial probe test were not statistically different from data gathered in the second test and thus these sessions were analyzed together in the text (Figure 1—figure supplement 1). Licking rates for odor mixtures scaled with the proportion of the mixture that was the sucrose-predicting odor. This indicates that mice responded to parametrically varying ambiguous stimuli with smoothly varying behavioral responses (Figure 1C). Interestingly, mice exposed to air puffs responded to ambiguous odor mixtures with fewer licks during the anticipatory odor and delay period indicating that exposure to aversive stimuli biased decisions about ambiguous outcomes toward avoidance and away from approach (Figure 1D). A two-factor ANOVA (cue × group) comparing licking rates during cue presentation and delay period in test days demonstrated a significant interaction between cue and group (F1,4=5.32, p<0.01; Tukey HSD, 0.71 ± 0.41 licks s−1, p<0.001). Lick rates were significantly larger for mixtures of odors in mice unexposed to air puffs (t69=2.70, p<0.01). Notably, the reduced licking to ambiguous cues in the air-puff group was evident on the first trial of the probe test and persisted throughout all probe test trials (Figure 1E). Thus, the decline in responding was not due to effects of extinction in the probe test. Indeed, both groups showed similar extinction of responding to ambiguous cues across trials resulting from outcome omission. A three-factor ANOVA (cue × trial × group) revealed a significant interaction between cue and trial (F1,14=5.33, p<0.01). Importantly, the interaction between cue, group and trial was not significant (F1,14=1.21, p=0.27).

Aversive stimuli modulate mPFC neuronal responses to motivationally relevant cues

Medial PFC (mPFC) is known to be involved in learning (Holland and Gallagher, 2004; Luk and Wallis, 2009; Alexander and Brown, 2011; Del Arco et al., 2017; Otis et al., 2017; Orsini et al., 2018), stress (Wellman, 2001; Cook and Wellman, 2004; Radley et al., 2004; Radley et al., 2005; Liston et al., 2006; Radley et al., 2006; Cerqueira et al., 2007; Wei et al., 2007; Liu and Aghajanian, 2008; Radley et al., 2008; Goldwater et al., 2009; Yuen et al., 2012; Adhikari et al., 2015), and uncertainty (Ernst and Paulus, 2005; Opris and Bruce, 2005; Sugrue et al., 2005; Bach et al., 2009; Levy et al., 2010; Orsini et al., 2018). These functions are critical for making predictions about previously unobserved stimuli. Such predictions derive from prior knowledge, as well as experience with the context of an environment. We thus asked whether mPFC neurons responded differently to motivationally relevant stimuli in the presence or absence of a history of aversive cues.

We recorded action potentials extracellularly from 2208 mPFC neurons in 12 mice, five exposed to air puffs (929 neurons), seven unexposed (1279 neurons), while mice performed the go/no-go discrimination task (Figure 2—figure supplement 1). Most neurons showed firing rate increases or decreases following odor cues, largely with persistent activity that lasted beyond cue offset (Figure 2A). There was a positive correlation between firing rates in response to sucrose-predicting and denatonium-predicting CS in mice exposed (r=0.59±0.04, 95% CI, p<0.0001) and unexposed (r=0.58±0.04, 95% CI, p<0.0001) to air puffs. To quantify the responses of the population, we measured the temporal response profile of each neuron during sucrose trials by quantifying firing rate changes relative to denatonium trials in 100 ms bins using a receiver operating characteristic (ROC) analysis (Figure 2B). We calculated the area under the ROC curve (auROC), comparing sucrose trials to denatonium trials. This analysis determines to what extent an ideal observer could discriminate between activities on the two trial types. Values of 0.5 indicate no discriminability, whereas values of 0 or 1 indicate perfect discriminability. A large fraction of neurons (270 of 929 in mice exposed to air puffs and 325 of 1279 in mice unexposed) showed auROC values greater than 0.7, indicating substantially greater activity for sucrose compared to denatonium trials, or values less than 0.3, indicating substantially greater activity for denatonium compared to sucrose trials (Figure 2B). These patterns were observed in mice exposed and unexposed to air puffs (Figure 2C). Neurons from mice exposed to air puffs responded robustly to both air puff-predicting cues and the air puffs themselves (Figure 2D; Wilcoxon rank sum test, p<0.0001).

Figure 2 with 1 supplement see all
Neuronal responses across learning.

(A) Firing rates of neurons in mice unexposed (top) and exposed (bottom) to air puffs relative to pre-CS firing rates. Histogram scale bars: 700 neurons (top), 500 neurons (bottom). (B) Discriminability (auROC) between sucrose and denatonium trials of neurons in mice unexposed (left) and exposed (right) to air puffs with significant firing rate changes during the cue. Increases (yellow) and decreases (cyan) in firing rate in sucrose trials relative to denatonium trials. Each row represents one neuron. (C) Average firing rates of all neurons with auROC values greater than 0.7 or less than 0.3 in at least one bin in no air puff (left) and air puff (right) mice during sucrose (blue), denatonium (red), no-outcome (gray) and air puff (orange) trials. Gray bars indicate a period of odor presentation. Dashed lines indicate outcome delivery. (D) auROC values for responses to air puff-predicting CSs (left) and air puff (right).

We next asked whether a history of air puffs changed the responses of mPFC neurons to aversion- and reward-predictive cues together with ambiguous stimuli. We recorded from 136 neurons from mPFC in the no-air-puff group and 106 neurons from mPFC in air-puff exposed mice during the probe sessions (Figure 3). We calculated firing rates in response to odors predicting sucrose or denatonium, as well as the three proportions of mixtures of the two odors. These populations included 37 in the no-air-puff group and 25 in the air-puff group that exhibited auROC discrimination between sucrose and denatonium trials greater than 0.7 or less than 0.3 during the probe sessions (see Figure 3A,B for two example neurons in mice unexposed or exposed to air puffs). To compare firing rates of those neurons between mice exposed or unexposed to air puffs, we calculated a generalized linear model (Poisson regression) to predict spike counts during the cues as a function of air puff exposure and odor type. This type of regression tests for effects of categorical variables (i.e. air puff exposure) on variables generated from a discrete nonnegative probability distribution (i.e. spike counts). Cue-evoked firing rates were significantly higher in mice exposed to air puffs (Figure 3C–E; odor mixture z=0.18±0.015, air-puff group z=0.043±0.014, stimulus-group interaction z=0.033±0.019, p<0.001). Firing rates were higher at each odor mixture in mice exposed to air puffs (Wilcoxon rank sum tests, all p<0.05). When we use the term odor mixture in the context, we refer to all mixtures of sucrose-predicting and denatonium-predicting odors: 100%/0%, 85%/15%, 50%/50%, 15%/85%, and 0%/100%.

Aversive stimuli increase mPFC firing rates to motivationally relevant cues.

(A–B) Average firing rates (top) and histograms of firing rates during odor and delay period (bottom) from example neurons in a mouse unexposed (A) and exposed (B) to air puffs. Red: denatonium trials. Blue: sucrose trials. Graded colors indicate mixtures as in Figure 1A. Gray bars indicate a period of odor presentation. Dashed lines indicate outcome delivery. (C) Mean trial-by-trial firing rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for each ambiguous cue (light red, gray, light blue squares) for no air puff (black) and air puff (orange) groups, during odor and delay period. Firing rates were calculated during the CS. (D) Mean ± SEM firing rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for no air puff (black) and air puff (orange) groups, during odor and delay period. (E) Mean firing rates of mPFC neurons in mice unexposed (left) or exposed to air puffs (right).

Aversive stimuli modulate corticothalamic neuronal responses to motivationally-significant cues

The neural data described above suggest that mPFC neurons that discriminate between sucrose and denatonium trials are modulated by a history of aversive stimuli. One of the major projection targets of the mPFC is the paraventricular nucleus of the thalamus (mPFC→PVT). Studies examining the role of PVT in regulating behavioral responses have found that PVT neurons are activated by cues or contexts previously associated with positive or negative emotional outcomes (Beck and Fibiger, 1995; Schiltz et al., 2007; Yasoshima et al., 2007; Choi et al., 2010; Igelstrom et al., 2010; Haight and Flagel, 2014; Hsu et al., 2014; Do-Monte et al., 2015; Kirouac, 2015; Matzeu et al., 2015; Penzo et al., 2015; Li et al., 2016; Zhu et al., 2016; Do-Monte et al., 2017). Additionally, activity in these corticothalamic neurons suppresses both the acquisition and expression of conditioned reward seeking (Otis et al., 2017). Thus, we hypothesized that mPFC→PVT neurons encode information about appetitive and aversive stimuli, and that this information is critical for weighing prior experience in those predictions.

To test this hypothesis, we performed projection-specific electrophysiological recordings from mPFC→PVT neurons while mice performed the go/no-go task. We expressed the light-gated ion channel channelrhodopsin-2 (ChR2, using adeno-associated viruses, AAV1-CaMKIIa-ChR2-eYFP) in pyramidal neurons of the mPFC and we implanted an optic fiber above the posterior portion of the PVT, to activate mPFC→PVT cells antidromically (Figure 4A,B). Virus expression and optic fiber implantation were verified histologically (Figure 4—figure supplement 1). At the end of each recording session, we used ChR2 excitation to observe antidromically-evoked spikes. For each neuron, we measured the response to light stimulation and the shape of spontaneous spikes (Figure 4C). To unequivocally identify mPFC→PVT neurons, ChR2-expressing cells in the mPFC were identified with axonal photostimulation and extracellular recordings in mPFC using a collision test (Figure 4D; Paintal, 1959; Bishop et al., 1962; Darian-Smith et al., 1963; Economo et al., 2018; Bari et al., 2019). Based on the parameters of cells that passed the collision test, we only included units that responded to light with a latency less than 15 ms and spiked in response to at least 70% of all pulses (in response to 10 Hz pulses; Figure 4D). These criteria are comparable to the fastest responses seen using antidromic stimulation with collision tests in corticothalamic neurons in sensory regions of neocortex (Swadlow and Weyand, 1981; Swadlow, 1998; Stoelzel et al., 2017). Not every neuron was subjected to a collision test, therefore it is plausible that some of these neurons may have exhibited orthodromic activation via local synapses.

Figure 4 with 1 supplement see all
Air-puff-predicting stimuli modulate mPFC→PVT neuron firing rates.

(A) Schematic drawings of viral stereotaxic injection of AAV1-CaMKII-ChR2-eYFP and tetrode bundle into mPFC and optic fiber over PVT. (B) eYFP (green), and DAPI (blue) in mPFC (top) and PVT (bottom) coronal sections from BL6 mice that received AAV1-CaMKII-ChR2-eYFP and tetrode bundle into mPFC and an optic fiber over PVT (scale bar, 100 μm). (C) Example of an identified corticothalamic neuron responding to a sequence of light stimuli (cyan) with action potentials (top) but not when the light stimuli followed spontaneous action potentials (bottom). (D) Antidromically-tagged corticothalamic neurons (blue) and antidromically-tagged corticothalamic neurons that passed collision tests (magenta). White points are neurons that were not identified. Crosses are neurons that passed collision tests, but were not recorded during behavior. (E–F) Average firing rates from example mPFC→PVT neurons showing firing rate increase (E) or decrease (F) to the air puff-predicting cue. Orange: air puff trials. Gray: CS - trials. Gray bars indicate odor presentation. Dashed lines indicate outcome delivery. (G) Scatter plot showing relationship between the change in firing rate to the air puff-predicting cue compared to baseline firing activity. Orange: neurons in which the firing rate during the air puff-predicting cue was significantly different from baseline firing activity (t-test, p<0.05).

We identified 39 and 45 neurons as projecting to PVT in mice unexposed and exposed to air puffs, respectively (Figure 4D; no air puff group: 18 and 21 neurons during conditioning and probe sessions, respectively; air puff group: 35 and 10 neurons during conditioning and probe sessions, respectively). We first asked whether these neurons responded to aversive stimuli. Previous studies have emphasized the role of the PVT in shaping behavioral responses to stimuli that predict aversive outcomes (Beck and Fibiger, 1995; Yasoshima et al., 2007; Hsu et al., 2014; Do-Monte et al., 2015; Penzo et al., 2015), but it is unknown which of its inputs may drive those responses. We found that 43% of mPFC→PVT neurons recorded from mice in the air puff group showed firing rate changes (5 of 35 neurons with increases and 10 of 35 neurons with decreases) in response to air-puff-predicting stimuli during conditioning (Figure 4E–G). This demonstrates that mPFC→PVT neurons, thought to be involved in behavioral responses to aversive-predicting stimuli, are indeed modulated by those stimuli.

We next compared the responses of mPFC→PVT neurons to sucrose- and denatonium-predictive cues and ambiguous stimuli (see Figure 5A,B for two example neurons in mice unexposed or exposed to air puffs, respectively). Interestingly, corticothalamic neurons from mice exposed to air puffs responded to the sucrose-predictive cue and ambiguous odor mixtures with higher phasic activity during the anticipatory odor and delay period, indicating that exposure to aversive stimuli biases neural responses of corticothalamic neurons to reward-predictive and ambiguous stimuli (Figure 5C–F; Poisson regression, odor mixture z=0.26±0.10, p<0.01, air-puff group z=0.73±0.15, p<0.01, stimulus-group interaction z=0.27±0.22, p>0.2). Firing rates were higher at each odor mixture in mice exposed to air puffs (Wilcoxon rank sum tests, all p<0.05).

Aversive stimuli increase corticothalamic firing rates to motivationally-relevant cues.

(A–B) Average firing rates (left) and histograms of firing rates during odor and delay period (right) from example neurons in a mouse unexposed (A) and exposed (B) to air puffs. Red: denatonium trials. Blue: sucrose trials. Graded colors indicate mixtures as in Figure 1A. Gray bars indicate a period of odor presentation. Dashed lines indicate outcome delivery. (C) Mean trial-by-trial firing rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for each ambiguous cue (light red, gray, light blue squares) for no air puff (black) and air puff (orange) groups, during odor and delay period. (D) Mean ± SEM firing rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials for no air puff (black) and air puff (orange) groups, during odor and delay period. (E) Mean firing rates of corticothalamic neurons in mice unexposed to air puffs. (F) Mean firing rates of corticothalamic neurons in mice exposed to air puffs.

Excitation of corticothalamic neurons modulates behavioral responses to motivationally-relevant cues

The neural data described above suggest that elevated activity in corticothalamic neurons to the aversion- and reward-predictive cues together with the ambiguous cues is critical for modulating behavioral response to those stimuli. To test this hypothesis, we next used optogenetic methods to activate corticothalamic neurons selectively at the time of presentation of those cues in mice trained in the same go/no-go task described above but with no exposure to air puffs. Mice received bilateral infusions of either AAV1/5-CaMKIIa-ChR2-EYFP or AAV1/5-CaMKIIa-EYFP (control) into mPFC; expression was verified histologically (Figure 6—figure supplement 1). Mice also received fiber optic implants over posterior PVT (Figure 6A). Three weeks after surgery, these mice began training in the go/no-go discrimination task and, during the probe session, light was delivered into the PVT in five random trials of all sucrose and denatonium trials and in half of all the ambiguous trials, in the cue and delay period (Figure 6B). We used a 10 Hz frequency stimulation because it approximated the firing rate of corticothalamic neurons in response to ambiguous stimuli observed in neural recordings; we also used a 20 Hz frequency in order to obtain a frequency response curve. All mice underwent conditioning sessions and there were neither main effects nor any interactions of group on conditioned responding across conditioning (F<1.14; p>0.33; Figure 6C). During the subsequent probe test, ChR2 mice showed a reduction in response to denatonium- and sucrose-predictive cues and ambiguous cues in the trials in which light was delivered in a frequency-dependent manner, whereas eYFP mice that received the same treatment responded equally in trials with or without light delivery (Figure 6D,E). A three-factor ANOVA (cue × stimulation × group) comparing licking during cue presentation and delay period in stimulated versus unstimulated trials in EYFP, 10 Hz and 20 Hz groups revealed a significant main effect of cue (F1,4=13.8, p<0.01) and stimulation (F1,1=30.11, p<0.01). Moreover, there was a significant interaction between cue and group (F1,8=4.8, p<0.01) and stimulation and group (F1,2=4.68, p<0.05).

Figure 6 with 2 supplements see all
Optogenetic excitation of corticothalamic neurons negatively biases responses to motivationally-relevant stimuli.

(A) Schematic of viral stereotaxic injection of AAV1/5-CaMKIIa-ChR2-eYFP or AAV1/5-CaMKIIa-eYFP into mPFC and optic fiber over PVT. (B) Optical stimulation was delivered during presentation of the cue and during the 1 s delay before outcome delivery. (C) Licking rates in eYFP (left, n=3), 10 Hz ChR2-eYFP (center, n=5), and 20 Hz ChR2-eYFP (right, n=7) groups across conditioning, during odor and delay period, for sucrose (blue), denatonium (red), and no-outcome (black) trials. (D) Licking rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials of each mixed stimulus for eYFP (left), 10 Hz ChR2-eYFP (center) and 20 Hz ChR2-eYFP (right) groups, during odor and delay period, with (light blue) or without (white) laser stimulation. (E) Trial-by-trial licking rates during 85%A/15%B (light red), 50%A/50%B (gray), 15%A/85%B (light blue) trials for eYFP (left), 10 Hz ChR2-eYFP (center) and 20 Hz ChR2-eYFP (right) groups, during odor and delay period, with (light blue shadows) or without laser stimulation. Line and error bars represent the mean ± SEM.

Stimulation of the same neurons did not, however, disrupt licking for an unpredictable reward. Outside of the task, we delivered randomly-timed sucrose rewards (3 μl). Mice licked at high rates to consume unexpected rewards (Figure 6—figure supplement 2). Importantly, stimulation of mPFC→PVT axons during reward delivery at the same frequencies used in the behavioral task in half of the total number of trials did not change the number of licks in response to these rewards. A two-factor ANOVA (frequency of stimulation × group) comparing licking during stimulated and unstimulated trials revealed no significant main effect or interaction with group (all F<0.23, p>0.65). Thus, uncued licking is not altered by optogenetic excitation of mPFC→PVT cells and the optogenetic effects are not due to a light-induced impairment in licking in general.

Inhibition of corticothalamic neurons prevents negative biases to motivationally-relevant cues

The optogenetic data described above suggest that artificially increasing activity in corticothalamic neurons to the aversion- and reward-predictive cues together with the ambiguous cues is sufficient for modulating behavioral response to those stimuli. To test the necessity of this pathway in generating negative biases to motivationally relevant stimuli, we next used optogenetic inhibition to inhibit corticothalamic neurons selectively during air-puff trials during conditioning in mice trained in the same go/no-go task described above. Mice received bilateral infusions of either AAV1-CaMKIIa-eNpHR3.0-EYFP or AAV1-CaMKIIa-EYFP (control) into mPFC; expression was verified histologically (Figure 6—figure supplement 1). Mice also received fiber optic implants over the posterior portion of PVT (Figure 7A). Three weeks after surgery, these mice began training in the go/no-go discrimination task and, during the air-puff-predictive cue, light was delivered into the PVT (Figure 7B). All mice underwent conditioning sessions and there were main effects of odor (F2,228=124.7, p<0.0001) and day (F1,228=205.5, p<0.0001). During the late phase of training, licking rates were not significantly different between the two groups of mice (Wilcoxon rank sum test, p>0.05; Figure 7C). During the probe test, mice expressing EYFP showed a reduction in response to denatonium- and sucrose-predictive cues and ambiguous cues in the trials compared to mice expressing eNpHR3.0-EYFP (Figure 7D). A two-factor ANOVA (cue × group) comparing licking during cue presentation and delay period in eYFP versus virus groups revealed a significant main effect of cue (F1,4=39.8, p<0.05) and group (F1,1=16.6, p<0.05).

Optogenetic inhibition of corticothalamic neurons prevents negatively bias responses to motivationally-relevant stimuli.

(A) Schematic of viral stereotaxic injection of AAV1-CaMKIIa-eNpHR3.0-EYFP or AAV1-CaMKIIa-EYFP into mPFC and optic fiber over PVT. (B) Optical stimulation was delivered during presentation of the air-puff-predicting cue, during the 1 s delay before air-puff delivery and during air-puff delivery, only during learning. (C) Licking rates in EYFP (left, n=3), eNpHR3.0-eYFP (right, n=5) groups across conditioning, during odor and delay period (sucrose: blue, denatonium: red, no-outcome: black). (D) Licking rates during sucrose (blue square) and denatonium (red square) trials and during the eight probe trials of each mixed stimulus for EYFP (white circles), eNpHR3.0-EYFP (green circles) groups, during odor and delay period. Line and error bars represent the mean ± SEM.

Discussion

In this study, we examined the projection from mPFC to PVT in mice approaching or avoiding negative and positive valence-predictive stimuli. We found that a history of aversive stimuli negatively biased behavioral responses to those motivationally relevant cues while increasing excitatory responses of PVT-projecting mPFC neurons. Indeed, mice exposed to aversive stimuli showed reduced approach behavior to reward-predictive and ambiguous stimuli compared to mice unexposed to aversive stimuli. Moreover, artificially increasing activity from mPFC to PVT quantitatively mimicked the negative behavioral bias induced by presentation of aversive stimuli while selectively inhibiting this pathway during learning prevented the formation of negative biases.

Cognitive processes—appraisals of stimuli, events and situations—play an important role in generating affective states, and these affective states influence cognitive functioning by inducing attentional, memory, and judgment biases (Lerner and Keltner, 2000; Haselton et al., 2009; Harding et al., 2004; Enkel et al., 2010; Rygula et al., 2012; Papciak et al., 2013; Rygula et al., 2013; Parker et al., 2014; Rygula et al., 2014). Among brain regions that are engaged in affective processing, the mPFC has long been implicated in adaptive responding by signaling information about expected outcome and by regulating sensitivity to reward and punishment (Holland and Gallagher, 2004; Luk and Wallis, 2009; Alexander and Brown, 2011; Del Arco et al., 2017; Orsini et al., 2018). Here, we found that firing rates in mPFC neurons reflect cue-evoked expectations for aversion- and reward-predictive cues and were enhanced in mice exposed to aversive stimuli, which correlates with a reduction in anticipatory responses to the same stimuli. These observations are consistent with a negative bias and suggest that negative events can bias decisions by altering the activity of mPFC neurons. Several studies have implicated the prefrontal network in the pathophysiology of affective disorders (Phillips et al., 2003; Drevets et al., 2008) and chronic stress—a crucial factor in increasing the risk of developing affective disorders—has profound detrimental effects on the anatomy and physiology of mPFC neurons (Wellman, 2001; Cook and Wellman, 2004; Radley et al., 2004; Liston et al., 2006; Radley et al., 2006; Cerqueira et al., 2007; Wei et al., 2007; Liu and Aghajanian, 2008; Radley et al., 2008; Goldwater et al., 2009; Yuen et al., 2012; Adhikari et al., 2015). For example, chronic stress induces significant regression of the apical dendrites of pyramidal neurons in mPFC (Cook and Wellman, 2004; Radley et al., 2004; Liston et al., 2006; Goldwater et al., 2009), which may in turn impact mPFC function.

The mPFC densely projects to subcortical structures relevant for our behavioral task, including PVT, amygdala, hippocampus, and nucleus accumbens (Vertes, 2004; Li and Kirouac, 2012). Among all brain regions that receive strong projections from the mPFC, PVT has long been considered a stress detector and implicated in the emergence of adaptive responding to stress (Chastrette et al., 1991; Sharp et al., 1991; Cullinan et al., 1995; Bubser and Deutch, 1999; Spencer et al., 2004; Hsu et al., 2014; Do-Monte et al., 2015; Penzo et al., 2015; Zhu et al., 2016; Do-Monte et al., 2017; Beas et al., 2018; Choi et al., 2019). At the same time, PVT has also been considered a potential mediator of motivated behavior responding to both food- and drug-associated cues (Schiltz et al., 2005; Igelstrom et al., 2010; Martin-Fardon and Boutrel, 2012; James and Dayas, 2013; Browning et al., 2014; Haight and Flagel, 2014; Li et al., 2016). Consistent with these findings that place PVT in a position to integrate information about positive and negative motivationally relevant cues and translate it into adaptive behavioral responses (McNally, 2021), we found that mPFC neurons projecting to the PVT maintain cue-evoked expectations for motivationally-relevant outcomes and their neural activity was enhanced in mice exposed to aversive stimuli. A feature of our experimental design is worth comment. In our experiments, mice were exposed to aversive stimuli in the same context in which they received motivationally-relevant cues. This design cannot rule out the influence of context-derived motivational conflict on the observed behavioral outcome. Indeed, recent work suggests that PVT is important during motivational conflict (Li et al., 2014; Choi and McNally, 2017; Zhu et al., 2018; Choi et al., 2019; Engelke et al., 2021). PVT seems important for behavioral control when appetitive and aversive behaviors are elicited at the same time and the animal must select between them (Li et al., 2014; Choi and McNally, 2017; Zhu et al., 2018; Choi et al., 2019). Future studies in which aversive stimuli and rewarding stimuli are presented in two different contexts will help clarify the behavioral relevance of such context-derived variables.

By mimicking the increased response of mPFC→PVT neurons induced by exposure to aversive stimuli, we observed a reduction in anticipatory responses to the predictive cues. By contrast, selective inhibition of this pathway prevented the formation of negative biases. These findings suggest that information about cue interpretation is transferred from mPFC to PVT and this pathway is crucial for adaptive responding toward those stimuli as a function of previous experience. These results are also consistent with a recent study showing that activity in mPFC neurons projecting to the PVT suppresses both the acquisition and expression of conditioned reward seeking (Otis et al., 2017). Based on recent evidence showing two genetically, anatomically, and functionally distinct cell types across the anteroposterior axis of the PVT (Gao et al., 2020), it will be interesting to investigate in future studies whether the information from mPFC is transferred to anatomically or molecularly segregated cell types or projection-specific neurons within the PVT. Indeed, signals from mPFC to PVT inputs have been recently shown to propagate onto PVT projections to nucleus accumbens neurons (Otis et al., 2019) and are responsible for influencing reward seeking by adjusting a multiplexed reward signal in downstream PVT→nucleus accumbens neurons. Future studies may help clarify whether this pathway is also involved in negative bias or dynamic salience processing (Zhu et al., 2018), which is a behavioral domain that is well positioned to shape the negative bias itself.

Studying the mechanisms by which neuronal responses to cues are updated as a function of prior experience in the mPFC→PVT circuit is crucial for understanding learning and decision making more generally. Adaptive tuning of this network is modulated by ascending monoaminergic systems (Arnsten et al., 2012) and maladaptive changes of these systems has been implicated in cognitive deficits associated with several affective disorders (Enkel et al., 2010; Kukolja et al., 2008). For example, acute pharmacological stimulation of the serotonergic and dopaminergic systems has been shown to influence cognitive bias in rodents (Rygula et al., 2014). In particular, citalopram, a selective serotonin reuptake inhibitor, and amphetamine, a powerful psychostimulant, both induced a positive cognitive processing bias (Rygula et al., 2014). Those results are important from a clinical point of view, knowing that negative cognitive bias lies at the core of the pathophysiology of several affective disorders and it has been extensively studied in humans (Wright and Bower, 1992; MacLeod and Byrne, 1996; Beck, 2008). For example, it has been shown that patients with anxiety and depression interpret ambiguous information with a negative bias (Schwarz and Clore, 1983; Eysenck et al., 1987; Wright and Bower, 1992; MacLeod and Byrne, 1996; Lawson et al., 2002; Beck, 2008; Chan et al., 2008; Pizzagalli et al., 2008; Dearing and Gotlib, 2009). Thus, understanding how cognitive biases develop and act in several chronic and debilitating neuropsychiatric disorders may offer an opportunity for designing novel treatments, aimed at ameliorating the proper functional tuning and connectivity of prefrontal-orchestrated neuronal circuits.

In summary, by using exposure to aversive stimuli, we induced a negative bias in mice in response to ambiguous stimuli, which was linked to an increase in firing rates of PVT-projecting mPFC neurons. Artificial activation of the same pathway recapitulated the behavioral outcome while inhibition prevented the formation of negative biases. Thus, our results highlight a fundamental role for the mPFC→PVT circuit in shaping adaptive responses by modulating predictions about imminent motivationally-relevant outcomes as a function of prior experience.

Materials and methods

Subjects

Wild-type C57BL/6J male mice (The Jackson Laboratory, 000664), 8–10 weeks old at the time of surgery, were housed in a reverse 12 hr light-dark cycle room (lights on at 20:00). All mice were given ad libitum water except during testing periods. During behavioral testing, mice were water deprived by giving 1 ml of water per day. Food was freely available throughout the experiments. All testing was conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Johns Hopkins University Animal Care and Use Committee (protocol MO19M424).

Stereotaxic surgeries

Request a detailed protocol

All mice were surgically implanted with custom-made titanium head plates using dental adhesive (C and B-Metabond, Parkell) under isoflurane anesthesia (1.0–1.5% in O2). Surgeries were conducted under aseptic conditions and analgesia (ketoprofen, 5 mg/kg and buprenorphine, 0.05–0.1 mg/kg) was administered postoperatively. Mice recovered for 7–10 days before starting behavioral testing.

For electrophysiological experiments, we implanted unilaterally a custom microdrive containing eight drivable tetrodes made from nichrome wire (PX000004, Sandvik) and positioned inside 39 ga polyimide guide tubes. We targeted mPFC under stereotaxic guidance at 2.2 mm anterior and 0.4 mm lateral to bregma and 1.6 mm ventral to the skull. Tetrodes were advanced subsequently into final positions in mPFC during recording. For identifying corticothalamic neurons, an optic fiber was implanted over PVT under stereotaxic guidance at 1.4 mm anterior and 1.3 mm lateral to bregma and 3.6 mm ventral to the skull with a 225° angle.

For optogenetic excitation experiments, AAV1/5-CamKIIa-hChR2(H134R)-eYFP or AAV1/5-CamKIIa-eYFP (AAV5: from UNC GTC Vectore Core; AAV1: from Addgene) was injected bilaterally in mPFC under stereotaxic guidance at 2.2 mm anterior and 0.3 mm lateral to bregma and 1.6 mm ventral to the skull. pAAV-CaMKIIa-hChR2(H134R)-EYFP was a gift from Karl Deisseroth (Addgene viral prep 26969-AAV1; http://n2t.net/addgene:26969; RRID:Addgene_26969) (Lee et al., 2010). A total of 300 nl of virus (titer ∼ 1013 GC/mL) per hemisphere was delivered at the rate of 1 nl/s (MMO-220A, Narishige). The injection pipette was left in place for 5 min after each injection. Optic fibers (200 μm diameter, 0.39 NA, Thorlabs) were implanted bilaterally over mPFC (at 2.2 mm anterior and 0.6 mm lateral to bregma and 1.3 mm ventral to the skull with a 10° angle) or unilaterally over PVT (at 1.4 mm anterior and 1.3 mm lateral to bregma and 3.6 mm ventral to the skull with a 22.5° angle). For inhibition experiments, AAV1-CaMKIIa-eNpHR 3.0-EYFP or AAV1-CaMKIIa-EYFP was injected as described above. pAAV-CaMKIIa-eNpHR 3.0-EYFP was a gift from Karl Deisseroth (Addgene viral prep 26971-AAV1; http://n2t.net/addgene:26971; RRID:Addgene_26971).

Behavioral task

Request a detailed protocol

Following recovery from surgery, water-restricted mice were habituated for 3 days while head-fixed before training on the go/no-go task. Each mouse performed behavioral tasks at the same time of day (between 08:00 a.m. and 2:00 p.m). All behavioral tasks were performed in dark, sound-attenuated chamber, with white noise delivered between 2 and 60 kHz (L60 Ultrasound Speaker, Pettersson). Odors were delivered with a custom-made olfactometer (Cohen et al., 2012). Each odor was dissolved in mineral oil at 1:10 dilution. Diluted odors (30 μl) were placed on filter-paper housing (Whatman, 2.7 μm pore size). Odorized air was further diluted with filtered air by 1:10 to produce a 1.0 l/min flow rate. Licks were detected by charging a capacitor (MPR121QR2, Freescale). Task events were controlled with a microcontroller (ATmega16U2 or ATmega328). Reinforcements were 3 μl of sucrose (an appetitive sweet solution), denatonium (an aversive bitter solution) or air puff (40 psi), delivered using solenoids (LHDA1233115H, The Lee Co). Intertrial intervals (ITIs) were drawn from an exponential distribution with a rate parameter of 0.3, with a maximum of 30 s. This resulted in a flat ITI hazard function, ensuring that expectation about the start of the next trial did not increase over time (Luce, 1986). The mean ITI was 7.2 s (range 2.4–30.0 s).

Mice underwent 10 conditioning sessions

Request a detailed protocol

In each session, mice received 50 1 s presentations of four different olfactory stimuli (A, B, C, and D). The order of odor presentations was randomized among mice and among sessions. For all conditioning, A, B, C, and D consisted of (+)-limonene, p-cymene, penthylacetate, and acetophenone, respectively (counterbalanced). 1 s after termination of A, sucrose was delivered and 1 s after termination of B, denatonium was delivered. C was paired with no reinforcement. In the air puff group, 1 s after termination of D, an unavoidable air puff was delivered to their right eye, while in the control group, D was paired with no reinforcement. Four s after the presentation of each odor, a vacuum was activated to remove any residual of sucrose or denatonium. After the activation of the vacuum, there was a fixed 3 s delay and then the variable ITI will follow. After completion of conditioning training, mice received a single extinction probe session. During the probe session, the four conditioning odors were continued to be presented (53 trials for each odor), but mice also received eight non-reinforced presentations of three mixtures of varying proportions of A and B odors: 85%A/15%B, 50%A/50%B, 15%A/85%B. These odor mixture trials were interleaved with the four conditioning odor trials in a randomized order.

In mice designated for the mPFC electrophysiological experiments, following the probe test, mice underwent reversal learning, in which A and B were reversed. 1 s after termination of A, denatonium was delivered and 1 s after termination of B, sucrose was delivered. C and D were continued to be presented as in conditioning. Then, mice received another single extinction probe session, identical to the one received after conditioning. Neural data from the initial extinction days were not statistically different from data gathered in later rounds of training and thus these neurons were analyzed together in the text.

In mice designated for the PVT-projecting mPFC electrophysiological experiments, following the probe test, mice repeated 3 days of conditioning and then underwent additional rounds of probe test days in order to acquire additional data. This was done up to three times for a given mouse. Neural data from the initial extinction days were not statistically different from data gathered in later rounds of training and thus these neurons were analyzed together in the text.

In mice designated for the optogenetic experiments of stimulation, training began approximately 3 weeks after viral injection and fiber implantation, and light (473 nm, 10–12 mW) was delivered into the PVT during the probe session. During the behavioral task, light was delivered in half of all the ambiguous trials, during the cue and delay epoch. Moreover, light was also delivered in five random trials of all sucrose and denatonium trials, during the cue and delay epoch. The primary measure of conditioning to cues was the number of licks during odor presentation and the second preceding reinforcement delivery. During the uncued stimulation trials, light was delivered in half of all trials, during the presentation of reward delivery and lasted for 1500 ms.

In mice designated for the optogenetic experiments of inhibition, training began approximately 3 weeks after viral injection and fiber implantation, and light (532 nm, 10–12 mW) was delivered into the PVT during air puff trials during conditioning days. During the behavioral task, light was delivered in all air puff trials, during the cue and delay epoch. The primary measure of conditioning to cues was the number of licks during odor presentation and the second preceding reinforcement delivery.

Electrophysiology

Request a detailed protocol

Throughout the discrimination task, mice were attached to the recording cable and before each session, tetrodes were screened for activity. Active tetrodes were selected for recording, and the session was begun. On the rare occasion that fewer than four of eight tetrodes had single units, the tetrode assembly was advanced 40 or 80 μm at the end of the session. Otherwise, the tetrode assembly was kept in the same position between sessions until the probe test day. A neuron may be represented more than one time in the dataset. After the extinction probe test, the tetrode assembly was advanced 80 μm regardless of the number of active tetrodes in order to acquire activity from a new group of neurons in any subsequent training.

We recorded extracellularly (Digital Lynx 4SX, Neuralynx Inc) from multiple neurons simultaneously at 32 kHz using custom-built screw-driven microdrives with eight tetrodes coupled to a 200 μm fiber optic (32 channels total). All tetrodes were gold-plated to an impedance of 200–300 kΩ prior to implantation. Spikes were bandpass filtered between 0.3–6 kHz and sorted online and offline using Spikesort 3D (Neuralynx Inc) and custom software written in MATLAB. To measure isolation quality of individual units, we calculated the L-ratio (Schmitzer-Torbert et al., 2005) and fraction of interspike interval (ISI) violations within a 2 ms refractory period. All single units included in the dataset had an L-ratio less than 0.05 and less than 0.1% ISI violations. We only included units that had a firing rate of greater than 0.5 spikes s−1 over the course of the recording session.

Optogenetic identification

Request a detailed protocol

To verify that our recordings targeted corticothalamic neurons, at the end of daily recording sessions, we used channelrhodpsin excitation to observe stimulation-locked spikes, by delivering 3–5 ms pulses of 473 nm light at 15 mW using a diode-pumped solid-state laser (Laserglow), together with a shutter (Uniblitz). Spike shape was measured using a broadband signal (0.1 Hz-9 kHz) sampled at 32 kHz. This ensured that particular features of the spike waveform were not missed. We delivered 10 trains of light (10 pulses per train, 10 s between trains) at 10 Hz, resulting in 100 total pulses. For collision tests, we delivered light over PVT (2 ms pulse, 473 nm, 15 mW) triggered by a spontaneous action potential. Identified neurons did not show stimulus-locked spikes following spontaneous spikes (‘collisions’).

Histology

Request a detailed protocol

At the end of behavioral testing, all mice were deeply anesthetized and then transcardially perfused with 4% paraformaldehyde (wt/vol). The brains were removed and processed for histology using standard techniques. For the electrophysiological experiments, we verified recording sites histologically with electrolytic lesions (15 s of 10 μA direct current across two wires of the same tetrode). For optogenetic experiments, virus expression was examined using a confocal microscope (Zeiss LSM 800). After histological verification, mice with incorrect virus injection or tetrode implantation were excluded from data analysis.

Data analysis

Request a detailed protocol

All analyses were performed with MATLAB (Mathworks) and R (http://www.r-project.org/). All data are presented as mean ± SEM unless reported otherwise. All statistical tests were two-sided. For nonparametric tests, the Wilcoxon rank-sum test was used, unless data were paired, in which case the Wilcoxon signed-rank was used. To estimate neuronal learning rates, we used logistic functions of the form f(x)=L1+e-k(x-x0). Learning rates are estimates of the k parameter.

Data availability

Data have been deposited on Dryad.

The following data sets were generated
    1. Cohen JY
    (2020) Dryad Digital Repository
    Punishment history biases corticothalamic responses to motivationally-significant stimuli.
    https://doi.org/10.5061/dryad.hdr7sqvf5

References

Decision letter

  1. Joshua Johansen
    Reviewing Editor; RIKEN Center for Brain Science, Japan
  2. Michael J Frank
    Senior Editor; Brown University, United States
  3. Mario Penzo
    Reviewer; National Institute of Mental Health, United States
  4. James M Otis
    Reviewer; Medical University of South Carolina, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

A history of aversive experiences biases decision making, but the neural mechanisms through which this occurs is unclear. Using an odor guided go, no-go task, Lucantonio et al. show that when rodents have previously experienced aversive stimuli, they are more cautious and exhibit reduced reward seeking behavior. Importantly, this is accompanied by increased activity in neurons in the medial prefrontal cortex which project to the paraventricular nucleus of the thalamus and optogenetically increasing the activity of these cells mimics the effect of prior stress.

Decision letter after peer review:

Thank you for submitting your article "Punishment history biases corticothalamic responses to motivationally-significant stimuli" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Michael Frank as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Mario Penzo (Reviewer #2); James Otis (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before a decision is reached, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Summary:

The authors use electrophysiological and optogenetic techniques to study how prior aversive experiences alter olfactory cue based decision making. They report that when an aversive stimulus (airpuff) is added to the training regimen the resulting decision making behavior to other cues is more cautious and animals don't seek rewards as readily. This is accompanied by an increase in cue-evoked responding in medial prefrontal cortex (mPFC), and paraventricular thalamus (PVT) projecting mPFC, neurons. Finally, they show that optogenetic stimulation mPFC projections to PVT reduced cue-evoked reward seeking.

Essential revisions:

While all of the Reviewers thought that there were some potentially interesting results they felt that the analyses were seriously under-described and, in some cases, did not support the claims of the paper. Furthermore, there was a consensus opinion that the addition of a loss of function study was necessary to (a) go beyond what is already known from the literature, (b) test whether this pathway is important for prior aversive history to influence/inhibit reward seeking behavior and (c) to demonstrate that the projections to PVT, and not collateral projections to other brain regions, are important. Details of these and other concerns are outlined below.

1) The paper requires a loss of function study to adequately test a central claim of the paper that this pathway is essential for prior aversive history to influence/inhibit reward seeking behavior. The stimulation effects they see are a replication of previous results (Otis et al., 2017; Otis et al., 2019) and these effects are difficult to ascribe to prior history effects because optogenetic excitation may be impacting behavioral performance via mechanisms not related to punishment history. Ideally this would be a terminal inhibition experiment as the optogenetic terminal stimulation experiment leaves open the possibility of antidromic activitation of cell bodies and axon collaterals, resulting in effects which are not specific to projections to PVT.

2) A major issue with this paper is the way in which the results of statistical tests are presented. For example, in most cases the authors refer to overall ANOVA effects, but no group comparison tests are included. Specifically, on lines 258-262 the authors say that cue evoked firing rates were higher in cells from animals exposed to airpuff, statistics are provided and Figure 3C-D is referred to. They mention that they used a poisson regression to analyze this data, but very little information is given about this analysis, how it was used and the justification for its use. Furthermore, there is no statistical analysis which corresponds to the data presented in Figure 3C-D (this is also true for 1D and 5D). Figures reflecting the statistics with some kind of posthoc tests should be included so that the reader can get an idea of what odor mixtures this occurs in.

3) The authors want to emphasize the specific function of mPFC-PVT projection by measuring the different firing patterns of mPFC-PVT neurons with and w/o air-puff stressor. However, how specific this change is to this projection is unclear and this could simply reflect a general change occurring across most mPFC cells. They could do this by examining population tendencies across all mPFC cells and compare this to mPFC-PVT projection neurons.

4) Based on the figure legends and information in the figures, it is very difficult to understand what is represented in Figure 3B and the right panels in Figure 5A and B. These figures should be explained in the legend and explained/referred to in the text.

5) Also related to data presentation. While auROC is certainly a valuable way to identify neurons that can discriminate between reward and aversive cues, my understanding is that this is essentially a normalization and at such it limits appreciation of when a cue leads to an excitatory or an inhibitory response. It would be useful to also see the raw firing rate data for the neurons recorded. Also, for visualization purposes the authors should include pie charts summarizing electrophysiological data. This will help assess the data better and the actual proportion of cells with excitatory and inhibitory responses as well as cue discrimination.

6) In Figure 2 the auROC compares mPFC neuronal activity during sucrose vs denatonium trials rather than responses to each odor per se. Considering this analysis, technically a neuron could show greater inhibition to denatonium and no response to sucrose, and this would be completely hidden by the current analyses and figures. Data should be shown in the rawest form possible. One idea is that the authors could show spike frequency heatplots in addition to the existing auROC heatplots.

7) Description of data for Figure 4E-G indicate that 43% of neurons increase or decrease their activity in response to the sucrose-predictive odor, however the number that increase and the number that decrease in response to each odor is not indicated. This should be rather obvious to the reader and within both the figures and analysis.

8) To give the reader a better understanding of the dynamics of the cue response during the probe trials population/trial averaged PSTHs (similar to Figure 2B) should be presented for each stimulus condition in Figure 3C and 5C. Similarly, in Figure 3C and 5C, cue evoked firing rates are presented, but it is not clear how these were binned. Do they include only cue period, trace only, cue and trace, pre/post-stimulus? Finally, for Figure 1E, 3C and 5C it is unclear what is represented on the y-axis. Are these average values? This should be clarified.

9) Related to all electrophysiological figures, simply showing examples and then an average of all neurons does not take advantage of the powerful data collection that the authors have completed, could be quite misleading to readers, and is particularly important to consider given the heterogeneity of cue responses in the PFC-PVT pathway (shown here, also shown in Otis et al., 2017; Otis et al., 2019). I highly recommend inserting "example" right on the figure panel when it is an example, and "population average" when appropriate. Finally, to show the full and powerful datasets collected, authors should consider plotting and comparing cue responses for every recorded neuron. For example, for Figure 4 authors could plot sucrose odor vs baseline auROC for each cell on one axis and denatonium odor vs baseline auROC for each cell on the opposing axis for both air puff and control mice.

10) The authors have framed their study as addressing the effect of "punishment history" on motivationally-relevant stimuli. However, at present, their experimental design does not seem to rule out the possibility that what the authors are actually evaluating here is not the effect of punishment history, but rather the response to motivational conflict. Have the authors taken any steps to ensure that context was not the main driver of the behavioral and electrophysiological changes detected? If mice are being tested in the same context in which air puffs were presented, then their decreases in reward seeking may result from this context-derived motivational conflict.

11) Also related to experimental design is the notion that ambiguity seems to serve no purpose here. With the exception of the near zero response to Odor B (bitter solution), mice response to parametrically-varying odor mixtures and to Odor A itself were similarly decreased for the air puff group. There is a sense that the authors initially predicted an effect of punishment history and ambiguity on response ratio. This may have been the case had mice been tested in a different context, or trained using a different behavioral task in which punishment was not directly associated with the reward (as seems to be the case based on the current experimental design).

12) The authors claim that behavior for the first and second probe tests were not different and as such the data was pooled together. They should include the comparisons for the first and second probe test in the supplementary material, as it would be useful to see this quantification.

13) For optogenetic antidromic identification of mPFC-PVT projection neurons, the authors find cells that respond with a range of latencies and with a range of fidelity. One issue with the antidromic approach, however, pertains to their mention that only neurons with latencies under 15 ms and at least 70% spike probability were labeled as corticothalamic. But looking at the data shown in Figure 4d it seems that two of the neurons that passed the collision test had spike probability below 70%. Furthermore, I don't think this approach rules out orthodromic activation as many cells were not tested for collision and it is not possible to test the ability of cells to follow very high frequency stimulation or to get as accurate a readout of latency as would be possible with electrical antidromic approaches for which these criteria were developed. These issues should be clarified and discussed.

14) For the PVT projection neurons (Figure 5D) the authors note that "…in both groups, firing rates of mPFC-PVT neurons for odor mixtures scaled with the proportion of the mixture that was the sucrose predicting odor…". They make similar claims about the mPFC population as a whole (Figure 3D) However, based on the data in these figures it appears that, except for the non-ambiguous cue predicting bitter outcome, the cue response in both airpuff and non-airpuff groups was fairly flat, showing that there was no proportional scaling of the response. The authors should show some statistical support for this claim or change the interpretation.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Aversive stimuli bias corticothalamic responses to motivationally-significant cues" for further consideration by eLife. Your revised article has been reviewed by 3 reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Michael Frank as the Senior Editor.

The manuscript has been greatly improved but there are some remaining issues that need to be addressed/clarified, as outlined in the reviewer comments below:

Reviewer #1:

The authors have addressed my previous concerns.

Reviewer #2:

It is this reviewer's opinion that the authors have appropriately addressed the concerns raised during the initial submission. In particular, the loss of function experiment has significantly improved the quality of the study by demonstrating that inhibition of the mPFC-PVT pathway prevents the formation of negative bias. Similarly, the addition of missing statistical information has facilitated the assessment of the authors' findings.

Reviewer #3:

The revisions overall were good, although I have some residual concerns listed below.

On one hand, addition of the optogenetic inhibition experiment adds causal information beyond what is already known regarding PFC◊PVT involvement in reward seeking. Additional data analysis and figure presentation has improved the manuscript significantly. New Figure 2A is fantastic – although I would have liked to see this in the phototagging experiment as well.

On the other hand, below are some of my concerns that either remain unaddressed or I simply can't find the in-text revisions. It would be helpful if authors indicate where exactly they have addressed each concern (e.g., line number).

Concerns Not Addressed:

1. The number of mice used for Figure 3, Figure 4, and Figure 5 remains unreported. Of concern is Figure 5F, where only 10 neurons are represented. Please report how many mice these 10 neurons come from.

2. Authors suggest that nomenclature of s-1 (which is used for neuronal and behavioral data analysis) is now described in the text, but I cannot find it in the methods, results, or legends. Am I missing it? Why not include it in the figure legend, like mean +/- SEM?

Additional Concern

1. The authors mention within their resubmission that recordings from Figure 2 were across 10 conditioning sessions, so is it possible that the same neurons were represented 10 times within a single dataset? If so, how do the authors account for sampling the same neurons repeatedly? Additionally, do the data from earlier conditioning sessions look the same as in later conditioning sessions? Considering that learning occurred across conditioning, it would be surprising if they were the same based on previous findings showing learning dependent cue-reward response development in dmPFC neurons that are stably maintained after learning (see PMID 34184635; Figure 4B). Please clarify this for me or simply add text within the results or discussion to indicate this limitation of the current analysis.

https://doi.org/10.7554/eLife.57634.sa1

Author response

Essential revisions:

While all of the Reviewers thought that there were some potentially interesting results they felt that the analyses were seriously under-described and, in some cases, did not support the claims of the paper. Furthermore, there was a consensus opinion that the addition of a loss of function study was necessary to (a) go beyond what is already known from the literature, (b) test whether this pathway is important for prior aversive history to influence/inhibit reward seeking behavior and (c) to demonstrate that the projections to PVT, and not collateral projections to other brain regions, are important. Details of these and other concerns are outlined below.

1) The paper requires a loss of function study to adequately test a central claim of the paper that this pathway is essential for prior aversive history to influence/inhibit reward seeking behavior. The stimulation effects they see are a replication of previous results (Otis et al., 2017; Otis et al., 2019) and these effects are difficult to ascribe to prior history effects because optogenetic excitation may be impacting behavioral performance via mechanisms not related to punishment history. Ideally this would be a terminal inhibition experiment as the optogenetic terminal stimulation experiment leaves open the possibility of antidromic activitation of cell bodies and axon collaterals, resulting in effects which are not specific to projections to PVT.

We agree with the reviewers that a loss of function experiment is crucial to support our conclusions. We added a new experiment in which we prevent the formation of negative bias by inhibiting mPFC→PVT axons during air-puff trials only during conditioning days (Figure 7 and Supplementary Figure 4). Mice receiving inhibition of mPFC axons over PVT during aversive cues had increased anticipatory licking to motivationally-relevant stimuli on probe trials on later days. There was no inactivation during probe trials. Thus, axon terminal inhibition during learning affected future behavior. This demonstrates that mPFC→PVT activity during learning affects future responses to motivationally-relevant stimuli.

2) A major issue with this paper is the way in which the results of statistical tests are presented. For example, in most cases the authors refer to overall ANOVA effects, but no group comparison tests are included. Specifically, on lines 258-262 the authors say that cue evoked firing rates were higher in cells from animals exposed to airpuff, statistics are provided and Figure 3C-D is referred to. They mention that they used a poisson regression to analyze this data, but very little information is given about this analysis, how it was used and the justification for its use.

We apologize for the lack of detail in motivating and describing statistical tests. We used a Poisson regression which tests for effects of categorical variables (in our case, air puff exposure) on variables generated from a discrete nonnegative probability distribution (in our case, spike counts).

Using this regression, we found that cue-evoked firing rates were significantly higher in mice exposed to air puffs (Figures 3C and 3D; odor mixture z = 0.18 ± 0.015, air-puff group z = 0.043 ± 0.014, stimulus-group interaction z = 0.033 ± 0.019, p < 0.001). As we also explain in the text, when we use the term “odor mixture,” we refer to all combinations of sucrose and denatonium proportions including pure denatonium (100/0 mixture) and pure sucrose (0/100 mixture). As suggested by the reviewers, we also added group comparison tests where appropriate (Tukey HSD or Wilcoxon rank sum tests). We conclude that firing rates were higher at each odor mixture in mice exposed to air puffs (Figure 3D, all p < 0.05). We clarified these analyses in the text.

Furthermore, there is no statistical analysis which corresponds to the data presented in Figure 3C-D (this is also true for 1D and 5D). Figures reflecting the statistics with some kind of posthoc tests should be included so that the reader can get an idea of what odor mixtures this occurs in.

As described above, we expanded our description of the Poisson regression test that was used for analyzing data in Figures 3C, 3D, and 5D, and we also ran post hoc tests to make specific comparisons.

3) The authors want to emphasize the specific function of mPFC-PVT projection by measuring the different firing patterns of mPFC-PVT neurons with and w/o air-puff stressor. However, how specific this change is to this projection is unclear and this could simply reflect a general change occurring across most mPFC cells. They could do this by examining population tendencies across all mPFC cells and compare this to mPFC-PVT projection neurons.

We thank the reviewers for raising this concern. We did not intend to draw conclusions about the specificity of mPFC→PVT neurons; indeed, we do not have experiments comparing these identified cells to cells with other projection targets. We have revised the text to make it clear.

4) Based on the figure legends and information in the figures, it is very difficult to understand what is represented in Figure 3B and the right panels in Figure 5A and B. These figures should be explained in the legend and explained/referred to in the text.

We apologize for the lack of clarity of those figure panels. We added more details to the legend.

5) Also related to data presentation. While auROC is certainly a valuable way to identify neurons that can discriminate between reward and aversive cues, my understanding is that this is essentially a normalization and at such it limits appreciation of when a cue leads to an excitatory or an inhibitory response. It would be useful to also see the raw firing rate data for the neurons recorded. Also, for visualization purposes the authors should include pie charts summarizing electrophysiological data. This will help assess the data better and the actual proportion of cells with excitatory and inhibitory responses as well as cue discrimination.

We thank the reviewers for this suggestion. We analyzed firing rate for each neuron in reward and aversive trials without any normalization. Those results are summarized in Figures 2A. Specifically, we compared firing rates of neurons for sucrose and denatonium trials in mice exposed and unexposed to air puffs relative to their pre-CS firing rates. Based on those results, there is a positive correlation between firing rates in response to sucrose-predicting and denatonium-predicting CS in mice exposed (r = 0.59 ± 0.04, 95% CI, p < 0.0001) and unexposed (r = 0.58 ± 0.04, 95% CI, p < 0.0001) to air puffs.

6) In Figure 2 the auROC compares mPFC neuronal activity during sucrose vs denatonium trials rather than responses to each odor per se. Considering this analysis, technically a neuron could show greater inhibition to denatonium and no response to sucrose, and this would be completely hidden by the current analyses and figures. Data should be shown in the rawest form possible. One idea is that the authors could show spike frequency heatplots in addition to the existing auROC heatplots.

As described in the response above, we analyzed raw firing rates for each neuron in reward and aversive trials without normalization. Those results are summarized in Figure 2A.

7) Description of data for Figure 4E-G indicate that 43% of neurons increase or decrease their activity in response to the sucrose-predictive odor, however the number that increase and the number that decrease in response to each odor is not indicated. This should be rather obvious to the reader and within both the figures and analysis.

Sorry for not having made this explicit. We added this information in the current version of the manuscript. To be clear, data presented in this figure are relative to air-puff-predicting cues.

8) To give the reader a better understanding of the dynamics of the cue response during the probe trials population/trial averaged PSTHs (similar to Figure 2B) should be presented for each stimulus condition in Figure 3C and 5C.

We agree with the reviewers that showing PSTHs will give the reader a better understanding of the dynamics of the cue response so we have now added these data in Figure 3E and 5E.

Similarly, in Figure 3C and 5C, cue evoked firing rates are presented, but it is not clear how these were binned. Do they include only cue period, trace only, cue and trace, pre/post-stimulus?

Cue-evoked firing rates were calculated during the cue and excluded the trace period. We clarified this in the text.

Finally, for Figure 1E, 3C and 5C it is unclear what is represented on the y-axis. Are these average values? This should be clarified.

Yes, those are average values. We clarified this in the text.

9) Related to all electrophysiological figures, simply showing examples and then an average of all neurons does not take advantage of the powerful data collection that the authors have completed, could be quite misleading to readers, and is particularly important to consider given the heterogeneity of cue responses in the PFC-PVT pathway (shown here, also shown in Otis et al., 2017; Otis et al., 2019). I highly recommend inserting "example" right on the figure panel when it is an example, and "population average" when appropriate.

Thank you for this suggestion. We have now made very clear in the text and in the figures where examples vs population averages are shown.

Finally, to show the full and powerful datasets collected, authors should consider plotting and comparing cue responses for every recorded neuron. For example, for Figure 4 authors could plot sucrose odor vs baseline auROC for each cell on one axis and denatonium odor vs baseline auROC for each cell on the opposing axis for both air puff and control mice.

Thank you for suggesting these analyses. As previously mentioned, we have analyzed and plotted firing rate for each neuron in reward and aversive trials and those results are summarized in Figure 2A. We compared firing rates of neurons for sucrose and denatonium trials in mice exposed and unexposed to air puffs relative to their pre-CS firing rates.

10) The authors have framed their study as addressing the effect of "punishment history" on motivationally-relevant stimuli. However, at present, their experimental design does not seem to rule out the possibility that what the authors are actually evaluating here is not the effect of punishment history, but rather the response to motivational conflict. Have the authors taken any steps to ensure that context was not the main driver of the behavioral and electrophysiological changes detected? If mice are being tested in the same context in which air puffs were presented, then their decreases in reward seeking may result from this context-derived motivational conflict.

We would like to thank the reviewers for this important point. We agree with the reviewers that mice being tested in the same context in which air puffs are presented may result in a conflict and the current experimental design does not allow to rule out an effect of context-derived motivational conflict on the obtained results. We have now added text to the Discussion about this issue.

11) Also related to experimental design is the notion that ambiguity seems to serve no purpose here. With the exception of the near zero response to Odor B (bitter solution), mice response to parametrically-varying odor mixtures and to Odor A itself were similarly decreased for the air puff group. There is a sense that the authors initially predicted an effect of punishment history and ambiguity on response ratio. This may have been the case had mice been tested in a different context, or trained using a different behavioral task in which punishment was not directly associated with the reward (as seems to be the case based on the current experimental design).

As for the previous point, we have discussed the methodological issue of having tested the mice in the same context in which the air puff has been presented in the current version of the manuscript. However, in our current experimental design, the punishment was never directly associated with the reward.

12) The authors claim that behavior for the first and second probe tests were not different and as such the data was pooled together. They should include the comparisons for the first and second probe test in the supplementary material, as it would be useful to see this quantification.

We agree with the reviewers that presenting the data for the first and second probe tests separately is important, so we have now added those results in Supplementary Figure 1.

13) For optogenetic antidromic identification of mPFC-PVT projection neurons, the authors find cells that respond with a range of latencies and with a range of fidelity. One issue with the antidromic approach, however, pertains to their mention that only neurons with latencies under 15 ms and at least 70% spike probability were labeled as corticothalamic. But looking at the data shown in Figure 4d it seems that two of the neurons that passed the collision test had spike probability below 70%. Furthermore, I don't think this approach rules out orthodromic activation as many cells were not tested for collision and it is not possible to test the ability of cells to follow very high frequency stimulation or to get as accurate a readout of latency as would be possible with electrical antidromic approaches for which these criteria were developed. These issues should be clarified and discussed.

We agree that neurons that were not subjected to collision tests may have exhibited orthodromic activation via local synapses. We revised the text to clarify this caveat. Regarding the criteria for identifying cells, we used relatively conservative criteria for identifying corticothalamic neurons. Although its use is still relatively new, we used stimulation frequencies in the range of other studies that used collision tests with excitatory opsins to identify corticothalamic and corticostriatal neurons (Economo et al., 2018; Bari et al., 2019).

14) For the PVT projection neurons (Figure 5D) the authors note that "…in both groups, firing rates of mPFC-PVT neurons for odor mixtures scaled with the proportion of the mixture that was the sucrose predicting odor…". They make similar claims about the mPFC population as a whole (Figure 3D) However, based on the data in these figures it appears that, except for the non-ambiguous cue predicting bitter outcome, the cue response in both airpuff and non-airpuff groups was fairly flat, showing that there was no proportional scaling of the response. The authors should show some statistical support for this claim or change the interpretation.

Thank you for the comment. We agree with the reviewer and we have changed the interpretation of the results.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #3:

The revisions overall were good, although I have some residual concerns listed below.

On one hand, addition of the optogenetic inhibition experiment adds causal information beyond what is already known regarding PFC◊PVT involvement in reward seeking. Additional data analysis and figure presentation has improved the manuscript significantly. New Figure 2A is fantastic – although I would have liked to see this in the phototagging experiment as well.

On the other hand, below are some of my concerns that either remain unaddressed or I simply can't find the in-text revisions. It would be helpful if authors indicate where exactly they have addressed each concern (e.g., line number).

Concerns Not Addressed:

1. The number of mice used for Figure 3, Figure 4, and Figure 5 remains unreported. Of concern is Figure 5F, where only 10 neurons are represented. Please report how many mice these 10 neurons come from.

Thank you for this comment and sorry for the confusion. The number of mice for Figure 3 are reported in line 143 (12 mice, 5 exposed to air puffs, 7 unexposed). The number of mice for Figures 4 and 5 are reported in the caption of Figure 4—figure supplement 1 (9 mice, 3 unexposed and 6 exposed to air puff). The 10 neurons in Figure 5F are from the 6 mice exposed to air puffs.

2. Authors suggest that nomenclature of s-1 (which is used for neuronal and behavioral data analysis) is now described in the text, but I cannot find it in the methods, results, or legends. Am I missing it? Why not include it in the figure legend, like mean +/- SEM?

If we understand correctly, the reviewer is asking about the “s−1” notation, which we used to abbreviate “per second.”

Additional Concern

1. The authors mention within their resubmission that recordings from Figure 2 were across 10 conditioning sessions, so is it possible that the same neurons were represented 10 times within a single dataset? If so, how do the authors account for sampling the same neurons repeatedly? Additionally, do the data from earlier conditioning sessions look the same as in later conditioning sessions? Considering that learning occurred across conditioning, it would be surprising if they were the same based on previous findings showing learning dependent cue-reward response development in dmPFC neurons that are stably maintained after learning (see PMID 34184635; Figure 4B). Please clarify this for me or simply add text within the results or discussion to indicate this limitation of the current analysis.

Thank you for this comment. During conditioning sessions, tetrodes were advanced 40–80 µm at the end of the session when fewer than 4 tetrodes had single units. It is highly unlikely that the same neuron is included multiple times in the data set. To clarify this point in the text, we added a sentence in the methods section in which we specify that we cannot completely rule out that a specific neuron may be represented more than one time in the dataset (lines 472–473).

https://doi.org/10.7554/eLife.57634.sa2

Article and author information

Author details

  1. Federica Lucantonio

    The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, Baltimore, United States
    Present address
    Department of Psychiatry and Taylor Family Institute for Innovative Psychiatric Research, Washington University School of Medicine, St Louis, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  2. Eunyoung Kim

    The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, Baltimore, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  3. Zhixiao Su

    The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, Baltimore, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  4. Anna J Chang

    The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, Baltimore, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Bilal A Bari

    The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, Baltimore, United States
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  6. Jeremiah Y Cohen

    The Solomon H Snyder Department of Neuroscience, Brain Science Institute, Kavli Neuroscience Discovery Institute, The Johns Hopkins University School of Medicine, Baltimore, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing
    For correspondence
    jeremiah.cohen@jhmi.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4768-7124

Funding

National Institute on Drug Abuse (R01DA042038)

  • Jeremiah Y Cohen

National Institute of Neurological Disorders and Stroke (R01NS104834)

  • Jeremiah Y Cohen

Esther A. and Joseph Klingenstein Fund

  • Jeremiah Y Cohen

MQ

  • Jeremiah Y Cohen

Brain and Behavior Research Foundation

  • Jeremiah Y Cohen

Whitehall Foundation

  • Jeremiah Y Cohen

National Institute of Mental Health (F30MH110084)

  • Bilal A Bari

Life Sciences Research Foundation

  • Federica Lucantonio

National Institute of Neurological Disorders and Stroke (P30NS050274)

  • Jeremiah Y Cohen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank T Shelley for machining, and G Schoenbaum, M Pignatelli, and the Cohen Lab for comments. This work was supported by Life Sciences Research Foundation (FL), F30MH110084 (BAB), Klingenstein-Simons, MQ, NARSAD, Whitehall, R01DA042038, and R01NS104834 (JYC), and P30NS050274.

Ethics

Animal experimentation: All testing was conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Johns Hopkins University Animal Care and Use Committee (protocol MO19M424).

Senior Editor

  1. Michael J Frank, Brown University, United States

Reviewing Editor

  1. Joshua Johansen, RIKEN Center for Brain Science, Japan

Reviewers

  1. Mario Penzo, National Institute of Mental Health, United States
  2. James M Otis, Medical University of South Carolina, United States

Version history

  1. Received: April 6, 2020
  2. Accepted: October 11, 2021
  3. Version of Record published: November 5, 2021 (version 1)

Copyright

© 2021, Lucantonio et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 871
    Page views
  • 117
    Downloads
  • 2
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Federica Lucantonio
  2. Eunyoung Kim
  3. Zhixiao Su
  4. Anna J Chang
  5. Bilal A Bari
  6. Jeremiah Y Cohen
(2021)
Aversive stimuli bias corticothalamic responses to motivationally significant cues
eLife 10:e57634.
https://doi.org/10.7554/eLife.57634

Further reading

    1. Neuroscience
    Federico G Segala, Aurelio Bruno ... Daniel H Baker
    Research Article

    How does the human brain combine information across the eyes? It has been known for many years that cortical normalization mechanisms implement ‘ocularity invariance’: equalizing neural responses to spatial patterns presented either monocularly or binocularly. Here, we used a novel combination of electrophysiology, psychophysics, pupillometry, and computational modeling to ask whether this invariance also holds for flickering luminance stimuli with no spatial contrast. We find dramatic violations of ocularity invariance for these stimuli, both in the cortex and also in the subcortical pathways that govern pupil diameter. Specifically, we find substantial binocular facilitation in both pathways with the effect being strongest in the cortex. Near-linear binocular additivity (instead of ocularity invariance) was also found using a perceptual luminance matching task. Ocularity invariance is, therefore, not a ubiquitous feature of visual processing, and the brain appears to repurpose a generic normalization algorithm for different visual functions by adjusting the amount of interocular suppression.

    1. Neuroscience
    Meghan Jelen, Pierre-Yves Musso ... Michael D Gordon
    Research Article

    Tastes typically evoke innate behavioral responses that can be broadly categorized as acceptance or rejection. However, research in Drosophila melanogaster indicates that taste responses also exhibit plasticity through experience-dependent changes in mushroom body circuits. In this study, we develop a novel taste learning paradigm using closed-loop optogenetics. We find that appetitive and aversive taste memories can be formed by pairing gustatory stimuli with optogenetic activation of sensory neurons or dopaminergic neurons encoding reward or punishment. As with olfactory memories, distinct dopaminergic subpopulations drive the parallel formation of short- and long-term appetitive memories. Long-term memories are protein synthesis-dependent and have energetic requirements that are satisfied by a variety of caloric food sources or by direct stimulation of MB-MP1 dopaminergic neurons. Our paradigm affords new opportunities to probe plasticity mechanisms within the taste system and understand the extent to which taste responses depend on experience.