Abstract
Ventral tegmental area (VTA) glutamatergic neurons participate in reward, aversion, drug-seeking, and stress. Subsets of these neurons cotransmit glutamate and GABA (VGluT2+VGaT+ neurons), transmit glutamate without GABA (VGluT2+VGaT− neurons), or cotransmit glutamate and dopamine (VGluT2+TH+ neurons), but whether these molecularly distinct subpopulations show behavior-related differences is not wholly understood. We identified in male and female mice that VGluT2+ subpopulations are sensitive to the reward value in unique ways. VGluT2+VGaT+ neurons increased maximum activity with increased sucrose concentration, whereas VGluT2+VGaT− neurons increased maximum and sustained activity with increased sucrose concentration, and VGluT2+TH+ neurons increased sustained but not maximum activity with increased sucrose concentration. VGluT2+ subpopulations also uniquely signaled consumption of sweet/noncaloric (saccharine) and nonsweet/high-calorie rewards (fat). VGluT2+VGaT+ neurons uniquely signaled lower-calorie sucrose over fat, whereas both VGluT2+VGaT− neurons and VGluT2+TH+ neurons showed a signaling preference for higher-calorie fat over sucrose but in temporally distinct ways. Further experiments suggested that VGluT2+VGaT+ consummatory reward-related activity was related to sweetness, partially modulated by prefeeding, and not dependent on caloric content. Additionally, aversive stimuli increased activity for each VGluT2+ subpopulation, but VGluT2+VGaT+ neurons uniquely scaled their magnitude and sustained activity with footshock intensity. Optogenetic activation of VGluT2+VGaT+ neurons during low-intensity footshock enhanced fear-related behavior without inducing place preference or aversion. About half of VGluT2+VGaT+ sucrose-sensitive neurons were transcriptionally activated by footshock. We interpret these data such that VTA glutamatergic subpopulations signal different elements of rewarding and aversive experiences and highlight the unique role of VTA VGluT2+VGaT+ neurons in salience signaling.
Significance Statement
Ventral tegmental area (VTA) glutamate neurons play a role in reward- and aversion-based–motivated behaviors. We identify that genetically distinct VTA glutamatergic subpopulations show differences in their signaling of consummatory rewards and aversive experiences. While all glutamatergic subpopulations signaled rewarding and aversive experiences, glutamatergic subtypes differed in their phasic magnitude and sustained activity profiles in response to the value of consummatory rewards, comparisons between multiple present rewards, and the value of aversive stimuli. Vesicular glutamate transporter (VGluT2)+ vesicular GABA transporter (VGaT)+ neurons showed unique profiles related to both rewarding and aversive events. Based on these results, we hypothesize that VTA VGluT2+ VGaT+ neurons have a role in signaling the general salience of positive and negatively valenced behavioral experiences.
Introduction
The ventral tegmental area (VTA) is a cellularly heterogeneous midbrain region that is conserved across rodents, nonhuman primates, and humans (Root et al., 2016; Morales and Margolis, 2017). VTA dopamine neurons, defined by tyrosine hydroxylase (TH), are well-established regulators of a wide range of motivated behaviors involving reward- and aversion-related stimuli (Bromberg-Martin et al., 2010; Watabe-Uchida et al., 2017; Keiflin et al., 2019; de Jong et al., 2022; Lowes and Harris, 2022). VTA GABA neurons, defined by the vesicular GABA transporter (VGaT), regulate dopamine reward- and aversion-related processes (Tan et al., 2012; van Zessen et al., 2012; Galaj et al., 2020). VTA glutamatergic neurons, defined by vesicular glutamate transporter (VGluT2; Yamaguchi et al., 2007, 2011, 2015), also regulate reward- and aversion-related processing.
VTA VGluT2 neurons are activated by rewarding and aversive stimuli (Qi et al., 2016; Root et al., 2018; McGovern et al., 2021), drug-seeking (McGovern et al., 2023), and threatening stimuli (Barbano et al., 2020; McGovern et al., 2024). Optogenetic stimulation of VTA VGluT2 neurons during real-time place conditioning has been reported to result in place preference or place aversion (Wang et al., 2015; Yoo et al., 2016; Bimpisidis et al., 2020; Zell et al., 2020; Warlow et al., 2024). Circuit-specific activation of the VTA VGluT2 pathway to the lateral habenula or nucleus accumbens medial shell results in conditioned place aversion (Root et al., 2014a; Lammel et al., 2015; Qi et al., 2016; Yoo et al., 2016) but can also support self-stimulation behavior (Yoo et al., 2016; Warlow et al., 2024).
From their initial discovery, it was recognized that VTA VGluT2 neurons are molecularly diverse, suggesting the existence of multiple VTA VGluT2 neuron subtypes (Yamaguchi et al., 2007, 2011, 2015; Root et al., 2014b, 2020; Zhang et al., 2015; Miranda-Barrientos et al., 2021). We hypothesize that the heterogeneous behavioral roles of VTA VGluT2 neurons are influenced by the diverse molecular subtypes of VTA VGluT2 neurons. VTA VGluT2 subtypes include those that release glutamate and GABA (VGluT2+ VGaT+ neurons), glutamate without GABA (VGluT2+ VGaT−), or glutamate and dopamine (VGluT2+ TH+). Using intersectional and subtractive genetic techniques, it was recently shown that VGluT2+ VGaT+ and VGluT2+ VGaT− neurons differentially signal pavlovian reward- and aversion-related stimuli (Root et al., 2020). While both VGluT2+ VGaT+ and VGluT2+ VGaT− neurons are activated by sweet rewards, as well as aversive footshocks, VGluT2+ VGaT− neurons signaled learned predictors (cues) of each while VGluT2+ VGaT+ neurons did not. Furthermore, VGluT2+ VGaT+ neurons detected errors in the expected receipt of reward, while VGluT2+ VGaT− neurons did not. The reward- and aversion-related signaling patterns of VTA VGluT2+ TH+ neurons are unknown. Here, we sought to identify whether VTA VGluT2+ subtypes show differences in their sensitivity toward different aspects of rewarding or aversive experiences. We found that VTA VGluT2+ subpopulations share features on general activation by rewarding and aversive stimuli. However, each subpopulation signaled reward and aversive value and showed differential sensitivity to calorically distinct consummatory rewards in temporally distinct ways. Furthermore, VGluT2+ VGaT+ neurons uniquely scaled neuronal activity with the sweet-reward value and aversive value. Half of VGluT2+ VGaT+ neurons transcriptionally activated by sucrose consumption were also transcriptionally activated by footshock. This subset of salience-detecting VGluT2+ VGaT+ neurons appears localized to the interfascicular nucleus. While activation of VGluT2+ VGaT+ neurons was not inherently rewarding or aversive, activation of VGluT2+ VGaT+ neurons timelocked to a low-value aversive stimulus resulted in enhanced fear-related behavior. Together, we interpret these results such that VTA glutamatergic subpopulations signal different elements of rewarding and aversive experiences. Furthermore, while VGluT2+ VGaT+ neuronal activity is biased toward sweet rewards, they are simultaneously capable of detecting aversive events and inflating the salience of aversive outcomes.
Materials and Methods
Animals
Male and female VGluT2-IRES::Cre mice (Slc17a6tm2(cre)Lowl/J; Jax Stock #016963) were crossed with either VGaT-2A::FlpO mice (Slc32a1tm2.1(flpo)Hze/J; Jax Stock #031331) or TH-2A::FlpO mice (C57BL/6N-Thtm1Awar/Mmmh; MMRRC 050618-MU) at the University of Colorado to produce VGluT2::Cre/VGaT::FlpO offspring or VGluT2::Cre/TH::FlpO offspring, respectively. Mice were maintained in a reverse light/dark cycle 12:12 h (lights off at 10 A.M.) and were group housed by sex and experimental condition with a maximum of five mice per cage. For all consummatory reward and behavioral economics experiments, mice were restricted to 85% of their ad libitum feeding bodyweight and were provided with access to water ad libitum. Mice were weighed daily and fed following the behavioral tasks in all experimental conditions, except for postprandial conditions. For postprandial (prefed) conditions, mice were weighed and separated into individual cages and given their daily food ration 1 h prior to two-bottle choice sessions. The remaining food was returned to the home cage after the completion of the experiment. For all footshock and place conditioning experiments mice were provided food and water ad libitum. All experiments were performed during the dark cycle. All experiments were conducted in accordance with the regulations by the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee at the University of Colorado Boulder.
Stereotactic surgery
VGluT2::Cre/VGaT::FlpO or VGluT2::Cre/TH::FlpO mice were anesthetized with 1–3% isoflurane and secured in a stereotactic frame (David Kopf Instruments). AAV8-EF1a-Con/Fon-GCaMP6M (Addgene 137119), AAV8-EF1a-Con/Foff-GCaMP6m (Addgene 137120), AAV8-nEF-Con/Fon-ChR2-mCherry (Addgene 137142), or AAV8-EF1a-Con/Fon-mCherry (Addgene 137132) was injected into the VTA (relative to the bregma AP, −3.2 mm; ML, 0.0 mm relative to the midline; DV, −4.3 mm from the skull surface). For the detection of single VGluT2+ VGaT+ neurons sensitive to reward and shock, VGluT2::Cre/VGaT::Flp mice were injected in VTA with a mixture of AAV1-TRE-tight-mKate2-WPRE-RAM-TetR (Addgene 84474, Virovek) and AAV1-pTRE-FSF-FLEX-EGFP-WPRE-bGHpA (Addgene 65453). The injection volume (400 nl) and flow rate (100 nl/min) were controlled with a microinjection pump (Micro4; World Precision Instruments). Following injection, the needle was left in place for an additional 10 min to allow for virus diffusion, after which the needle was slowly withdrawn. For fiber photometry experiments, mice were implanted with an optic fiber cannula (400 µm core diameter, 0.66 NA; Doric Lenses) dorsal to the VTA (AP, −3.2 mm relative to the bregma; ML, −1.0 mm at 9.5° relative to the midline; DV, −4.2 mm from the skull surface) that was secured with screws and dental cement to the skull. For optogenetic experiment mice, mice were implanted with optic fiber cannula (200 µm core diameter, 0.37 NA; Doric Lenses) at the same coordinate as recording fibers. All mice were allowed to recover for at least 3–4 weeks before experimentation.
Calcium recordings
GCaMP6m was excited at two wavelengths (465 and 405 nm isosbestic control) with amplitude-modulated signals from two light-emitting diodes reflected off dichroic mirrors and then coupled into an optic fiber (McGovern et al., 2021). Signals from GCaMP and the isosbestic control channel were returned through the same optic fiber and acquired with a femtowatt photoreceiver (Newport), digitized at 1 kHz, and then recorded by a real-time signal processor (Tucker-Davis Technologies, TDT). Analysis of the recorded calcium signal was performed using custom-written MATLAB scripts available at https://www.root-lab.org/code. For analysis, signals (465 and 405 nm) were downsampled (10×), and perievent time histograms were created trial-by-trial between −10 and 10 s surrounding lick bout onset or footshock. For each bout trial, data were detrended by regressing the isosbestic control signal (405 nm) on the GCaMP signal (465 nm) and then generating a predicted 405 nm signal using the linear model generated during the regression. The predicted 405 nm channel was subtracted from the 465 nm signal to remove movement, photobleaching, and fiber-bending artifacts (ΔF). Each trial's ΔF was then z-scored. Baseline maximum z-scores were taken from −6 to −3 s prior to lick bout onset or footshock. Reward or footshock maximum z-scores (normalized dF) were taken from 0 to 2 s following lick bout onset or footshock onset, respectively. Due to different calcium dynamics between cell types, the timepoint in which 50% of the maximum normalized dF occurred after event onset was determined as the half maximum. The area under the curve (AUC) was calculated between bout onset and the half maximum time to account for variability of calcium signals that returned to the baseline following event onset between cell types. The half maximum time was also used as its own metric of sustained activity following behavioral events.
Two-bottle choice recordings
Mice were brought daily to Med Associates chambers outfitted with two spouts connected to bottles filled between 30 and 50 ml of a solution. Behavioral timestamps of bottle licks were digitized in MED-PC or Synapse software (TDT) by TTL input from MED-PC or a custom lickometer. Mice were given 1 h daily access to reward solutions (8, 16, 32% sucrose, or 8% intralipid fat, 0.3% saccharine, or water) respective to experimental condition and testing. Eight percent sucrose was chosen for its comparison to prior investigations of VTA VGluT2+ neurons (Root et al., 2020; McGovern et al., 2021). A 0.3% saccharine was chosen for its high consumption and sweetness profile (Collier and Novell, 1967; Moskowitz, 1970; Sclafani et al., 2010). Two solutions were always presented, and the solution sides as well as solution combinations were counterbalanced and randomized. Sucrose scaling experiments consisted of daily training with either 8, 16, or 32% and water. Each sucrose concentration was paired with water to ensure that the value of the sucrose concentration was reliably represented in the neuronal response and not modified by the presence of a higher or lower sucrose concentration on testing day. Behavioral preference experiments consisted of daily training with 8% sucrose and 8% intralipid fat. Following sucrose versus fat recordings in the VTA VGluT2+ VGaT+ condition, a subset of mice advanced to a satiety and noncaloric sweetener condition. In the satiety conditions, mice were given access to their daily food ration for the hour immediately preceding their fiber photometry recording. For the noncaloric sweetener condition, mice were trained to consume 0.3% saccharine, and this solution was presented with 8% sucrose for at least 3 d prior to recording.
To behaviorally assess licking, the total number of licks were counted from each reward spout. The interlick interval was calculated as the median time between all licks from each reward spout. Neuronal analyses of licks were centered around the onset of a bout of licks for each spout. The bout onset was defined by at least two consecutive licks with less than a 400 ms interlick interval and was preceded by a minimum of 3 s of no licks. The bout offset was defined by the last lick before a subsequent lick with greater than 400 ms interlick interval or the last lick before a subsequent lick on the other spout. To calculate estimated kcal consumed, we assumed that each lick resulted in 0.55 µl consumed (Dotson and Spector, 2005). Then, we converted kilocalories per milliliter of each solution (kilocalories per milliliter, 8% sucrose 0.32; 16% sucrose 0.64; 32% sucrose 1.28; 0.3% saccharin 0.012; 8% intralipid 0.8) to kcal/0.55 µl and multiplied by the number of licks.
Unsignaled shock recordings
GCaMP-expressing mice were brought to behavior chambers outfitted with rod flooring which was electrically connected to a shock generator (Med Associates). After 2 min, mice received a randomly selected footshock of varying intensity (0.25, 0.50, 0.75, 1.00 mA) once per minute. Three footshocks per intensity were delivered.
Optogenetic aversion scaling
ChR2- and mCherry-expressing mice were brought to behavior chambers outfitted with rod flooring which was electrically connected to a shock generator (Med Associates). After 2 min, a single 0.25 mA (0.5 s) footshock was delivered each minute for a total of 10 footshocks. Optical stimulation was 2 s trains of 473 nm pulses of 5 ms duration at 20 Hz starting at the footshock onset (10–15 mW, CNI Laser). Mice were video recorded by AnyMaze (Stoelting, 30 Hz), which quantified freezing behavior for each minute following shock presentation. AnyMaze freezing detection was set to Level 1 and required at least 1 s of freezing to identify a freezing bout.
Real-time place preference (RTPP)
ChR2- and mCherry-expressing mice were habituated to a three-chamber apparatus (ANY-Box, Stoelting), consisting of a black chamber with white vertical stripes (striped), a black chamber with no stripes (solid), and a smaller connecting chamber (connecting). Mice were given 15 min to habituate in this apparatus 24 h prior to testing; the percentage time spent in each chamber was calculated per animal. During testing, 473 nm light was delivered (10–15 mW, 5 ms duration at 20 Hz) when mice were within the designated light-paired chamber. Light terminated when mice left the light-paired chamber. For 10 min, the light-paired chamber was randomly assigned to either the striped or the solid chamber of the apparatus, and the percentage time in each chamber was calculated. After the initial chamber paired stimulation, the laser paired side was reversed to the opposite chamber for 10 min, and the percentage time was calculated for the reversed position. The initial side pairing was randomized to prevent potential order effects.
Detection of reward- and shock-sensitive single neurons
One day before surgery, VGluT2::Cre/VGaT::Flp mice were individually housed and fed ad libitum with doxycycline-treated rodent diet (Inotiv, 40 mg/kg). Mice continued feeding ad libitum on doxycycline diet for 3 weeks. Mice were then food-restricted to 85% ad libitum feeding body weight with the same doxycycline-treated diet. For 3 individual days, mice were brought to Med Associates chambers equipped with bottles containing 16% sucrose (the middle value of previously tested concentrations) for 1 h and immediately brought back to the colony. One day following the third sucrose consumption training, mice were provided normal rodent chow without doxycycline to maintain 85% ad libitum feeding body weight. After 3 d, mice were again brought to Med Associates chambers and consumed 16% sucrose for 1 h. After consumption, mice were returned to the colony room and over 3 d were provided daily rations of normal rodent chow to maintain 85% ad libitum feeding body weight. Finally, mice were brought to a contextually different Med Associates chamber and delivered ten 0.50 mA footshocks (the middle value of previously tested currents) separated by 1 min. Ninety minutes later, mice were euthanized as described below in Histology. mRuby and GFP coexpressing VTA neurons that lacked or coexpressed c-Fos were counted in Fiji.
Slice preparation
Mice were anesthetized with isoflurane and transcardially perfused with ice-cold cutting solution containing the following (in mM): 75 NaCl, 2.5 KCl, 6 MgCl2, 0.1 CaCl2, 1.2 NaH2PO4, 25 NaHCO3, 2.5 d-glucose, and 50 sucrose. Coronal slices (240 μm) with the VTA were cut in the same cutting solution that was used for transcardial perfusion. Slices were maintained at 32°C in aCSF containing the following (in mM): 126 NaCl, 2.5 KCl, 1.2 MgCl2, 1.2 NaH2PO4, 2.5 CaCl2, 21.4 NAHCO3, 11.1 d-glucose, and 10 μm MK-801. After at least 30 min of incubation, slices were transferred to a recording chamber and continually perfused with 34 ± 2°C aCSF at a rate of 2 ml/min. All solutions were always bubbled with 95% O2, 5% CO2.
Electrophysiology
All whole-cell recordings were performed using an Axopatch 200B amplifier (Molecular Devices). Data were acquired using an ITC-18 interface (InstruTech) and Axograph X software (Axograph Scientific) at 10 KHz and filtered to 2 KHz. Neurons were visualized on a BX51WI microscope (Olympus) with an infrared LED and filter cube (Thorlabs). Cells that showed mCherry fluorescence were selectively used for whole-cell recordings. For whole-cell voltage–clamp recordings, cells were held at a voltage of −60 mV. Widefield activation of ChR2 was activated with collimated light from a LED (470 nm) through the 40× water immersion objective (10 pulses of 5 ms at 20 Hz). Patch pipettes (2.5–3 MΩ) were pulled from borosilicate glass (World Precision Instruments). The internal pipette solution contained (in mM) 135 d-gluconic acid (K), 10 HEPES (K), 0.1 CaCl2, 2 MgCl2, and 10 BAPTA and 0.1 mg/ml GTP, 1 mg/ml ATP, and 1.5 mg/ml phosphocreatine, pH 7.3 (280 mOsm).
Histology
Mice were anesthetized with isoflurane and perfused transcardially with phosphate buffer (PB) followed by 4% (w/v) paraformaldehyde in 0.1 M PB, pH 7.3. Brains were extracted, postfixed overnight in the same fixative, and cryoprotected in 18% sucrose in PB at 4°C. Coronal sections containing the VTA (30 μm) were taken on a cryostat, mounted to gelatin-coated slides, and imaged for GCaMP6m, mCherry, and optical fiber cannula placement on a Zeiss Axioscope. For VTA sections in above, Detection of reward- and shock-sensitive single neurons, experiment, sections were washed in PB, blocked in PB supplemented with 4% bovine serum albumin and 0.3% Triton X-100 (blocking buffer) for 1 h, and incubated over 48 h in mouse anti-GFP (JL-8, Takara Bio, 1:500) and guinea pig anti-c-Fos (Synaptic Systems 226004, 1:500) at 4°C. Sections were then washed and incubated with donkey anti-mouse Alexa Fluor 488 (Jackson Secondaries, 1:200) and donkey anti-guinea pig Alexa Fluor 647 for 2 h at room temperature before a final wash and mounted to gelatin-coated slides and coverslipped with DAPI-infused Fluoromount-G. Sections were imaged on a Nikon AXR confocal at 20× in resonant mode at 2,048 × 2,048 resolution. Mice with no virus expression or optic fibers not localized to VTA expressing neurons were removed from the study.
Statistics
Tests were conducted in SPSS (IBM) or Prism (GraphPad Software). Sucrose concentration licks, interlick intervals, and estimated kilocalorie were compared with the Friedman test followed by Dunn’s multiple-comparison tests. Sucrose/fat and sucrose/saccharine licks and interlick intervals were compared with Wilcoxon tests. GCaMP data were analyzed by comparing the maximum GCaMP z-score during the baseline with the maximum GCaMP z-score following a bout of licking on a specific spout or footshock. GCaMP analyses were conducted in stages. To test if an event (e.g., 8% sucrose consumption or footshock) differed from the baseline, Friedman tests were conducted and if significant were followed by Dunn’s multiple-comparison tests of the baselines against their associated events. To test if events with three or more conditions differed from each other (e.g., 8% sucrose, 16% sucrose, 32% sucrose), we conducted Friedman tests on the events alone, and if significantly different, we followed by Dunn’s multiple-comparison tests. For event comparisons with two events (e.g., 8% sucrose vs 8% fat), Wilcoxon tests were conducted. Because cell types visually differed in their sustained activity profiles, AUC duration was defined between event onset (e.g., lick bout onset or footshock onset) until the signal returned to half of the maximum value observed (half maximum). Sucrose concentration value AUC, footshock value AUC, and time to half maximums were compared with the Friedman test followed by Dunn’s multiple-comparison tests. Sucrose/fat AUC, sucrose/saccharine AUC, and time to half maximums were compared with Wilcoxon tests. Data did not differ between male or female mice; therefore, data were collapsed across sex.
For real-time place conditioning, repeated-measure ANOVAs compared time in each side (Side 1, connecting, Side 2) with day (pre, stimulation Side 1, stimulation Side 2) for mCherry and ChR2 groups. To assess potential changes in locomotion as a consequence of stimulation, a 2 × 2 repeated-measure ANOVA compared average speed in the outer sides (stimulation side vs nonstimulation side) for initial stimulation and reversal stimulation across groups. In the unsignaled footshock experiment, a group × time mixed ANOVA was conducted. Because differences in freezing behavior have been observed between sexes (Gruene et al., 2015), sex was used as a covariate. For all ANOVAs, if the assumption of sphericity was not met (Mauchley's test), the Greenhouse–Geisser correction was used. Sidak-adjusted pairwise comparisons followed up significant main effects or interactions.
Results
VTA VGluT2+ subpopulations differentially scale neuronal activity with the sucrose reward value
We previously found that VTA VGluT2+ VGaT+ neurons and VGluT2+ VGaT− neurons were activated by sucrose reward in the context of pavlovian conditioning (Root et al., 2020). Here, we assessed whether VTA VGluT2+ subpopulations differentially signaled consummatory reward value alone by recording VGluT2+ subpopulations in response to consumption of different concentrations of sucrose. To target VGluT2+ VGaT+ neurons (glutamate GABA neurons), VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding GCaMP6m dependent on the expression of Cre and Flp (AAV8-EF1a-Con/Fon-GCaMP6m; Root et al., 2020). To target VGluT2+ VGaT− neurons (non-GABA glutamate neurons), VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding GCaMP6m dependent on the expression of Cre and the absence of Flp (AAV8-EF1a-Con/Foff-GCaMP6m; Root et al., 2020). To target VGluT2+/TH+ neurons (glutamate dopamine neurons), VGluT2::Cre/TH::Flp mice were injected in VTA with AAVs encoding GCaMP6m dependent on the expression of Cre and Flp (AAV8-EF1a-Con/Fon-GCaMP6m; Chuhma et al., 2018; Poulin et al., 2018; Mingote et al., 2019; Buck et al., 2023; Prevost et al., 2025). All mice were implanted with optic fibers dorsal to VTA and food-restricted to promote consumption of reward (Fig. 1). Mice were presented with two bottles, one containing a sucrose solution of differing concentrations (8, 16, or 32%) and the other containing water. While we did not observe changes in neuronal activity to water consumption across cell types, the number of water licks was too low to allow for a systematic investigation of water-reward dynamics. Major results are summarized in Table 1.
Localization of recording fibers across recording VTA VGluT2+ subpopulation recordings. A–C, VGluT2::Cre/VGaT::Flp mice and VGluT2::Cre/TH::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent GCaMP6m to target VGluT2+ VGaT+ neurons and VGluT2+ TH+ neurons, respectively. VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre-dependent and Flp-lacking GCaMP6m to target VGluT2+ VGaT− neurons. For all mice, fiber optics were implanted dorsal to VTA. Representative GCaMP-expressing VTA neurons and fiber-optic localization for VGluT2+ VGaT+ mice (A), VGluT2+ VGaT− mice (B), and VGluT2+ TH+ mice (C). D, Recording fiber localizations. Yellow are VGluT2+ VGaT+ mice, green are VGluT2+ VGaT− mice, and blue are VGluT2+ TH+ mice. Number refers to distance from bregma. Scale bars in A, B, C, 500 µm.
Summary of major results
VGluT2+ VGaT+ neuron-targeted mice showed different numbers of licks and estimated kilocalorie consumed in response to varying sucrose concentrations (Fig. 2A–D; licks Friedman test = 10.89; p = 0.0029; kilocalorie Friedman test = 16.22; p < 0.001; n = 9 mice) but did not change interlick intervals (Fig. 2D; Friedman test = 2.2; p > 0.05; n = 9 mice). Mice significantly decreased their licks at 32% compared with 8% sucrose (Dunn’s z = 3.3; p = 0.0029; Fig. 2B) likely reflecting the higher estimated kcal of 32% compared with 8% sucrose (Dunn’s z = 4.007; p = 0.0002; Fig. 2C). The lack of change in the interlick interval indicates that mice show a consistent licking rhythm across reward conditions (Wiesenfeld et al., 1977). We examined neuronal activity in two ways. To assess phasic signaling, we captured the maximum change in neuronal activity within 2 s of initiating a bout of licks. To assess sustained signaling, we first calculated the duration of neuronal changes by calculating the timepoint where the maximum change in activity following a lick bout decayed to 50% of its maximum value (termed time to half maximum). We also assessed AUC between the lick bout onset and the time to half maximum. VGluT2+ VGaT+ neurons significantly increased maximum neuronal activity from the baseline in response to each sucrose concentration (Fig. 2E,F; Friedman test = 37.13; p < 0.001; n = 9 mice; Dunn-corrected tests BL × 8% z = 2.646; p = 0.0245; BL × 16% z = 3.654; p = 0.0008; BL × 32% z = 3.906; p = 0.0003). Thirty-two percent maximum sucrose-induced neuronal activity was significantly greater than 8% sucrose-induced neuronal activity (Friedman test = 6.889; p = 0.0307; n = 9 mice; Dunn’s test z = 2.593; p = 0.0286; Fig. 2F). Sustained activity of VGluT2+ VGaT+ neurons was significantly elevated in the 32% concentration compared with 8% measured by AUC (Friedman test = 10.89; p = 0.0029; n = 9 mice; Dunn’s z = 3.3; p = 0.0029) but not in time to half maximum (Friedman test = 5.556; p = 0.0689; n = 9 mice; Fig. 2G,H), suggesting AUC may have been influenced by changes in signal amplitude rather than a change in response duration.
VTA VGluT2+ subpopulations differentially signal changes in sucrose value. A, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). VGluT2+ VGaT+ VTA neurons were recorded during consumption of 8, 16, or 32% sucrose in independent sessions with water in a two-bottle choice. B–D, The number of licks (B), estimated kilocalorie consumed (C), and interlick intervals across sucrose recordings (D). E, VGluT2+ VGaT+ normalized dF (z-score) during sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. Eight percent sucrose is blue; 16% sucrose is purple; 32% sucrose is red. F, Maximum neuronal activity of individual mice compared with the baseline for all sucrose value conditions. G, Sustained activity analysis, AUC of individual mice for all sucrose value conditions. H, Sustained activity analysis, time to half maximum (sec) of individual mice for all sucrose value conditions. I, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre-dependent and Flp-lacking GCaMP6m (Con/Foff where C, Cre; F, Flp; on, must express; off, must not express). VGluT2+ VGaT− VTA neurons were recorded during consumption of 8, 16, or 32% sucrose in independent sessions with water in a two-bottle choice. J–L, The number of licks (J), estimated kilocalorie consumed (K), and interlick intervals across sucrose recordings (L). M, VGluT2+ VGaT− normalized dF (z-score) during sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. N, Maximum neuronal activity of individual mice compared with that of the baseline for all sucrose value conditions. O, Sustained activity analysis, AUC of individual mice for all sucrose value conditions. P, Sustained activity analysis, time to half maximum (sec) of individual mice for all sucrose value conditions. Q, Experimental setup: VGluT2::Cre/TH::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). VGluT2+ TH+ VTA neurons were recorded during consumption of 8, 16, or 32% sucrose in independent sessions with water in a two-bottle choice. R–T, The number of licks (R), estimated kilocalorie consumed (S), and interlick intervals across sucrose recordings (T). U, VGluT2+ TH+ normalized dF (z-score) during sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. V, Maximum neuronal activity of individual mice compared with that of the baseline for all sucrose value conditions. W, Sustained activity analysis, AUC of individual mice for all sucrose value conditions. X, Sustained activity analysis, time to half maximum (sec) of individual mice for all sucrose value conditions. *p < 0.05; **p < 0.005; ***p < 0.001, ****p < 0.0001.
VGluT2+ VGaT− neuron-targeted mice also showed differential licking responses and estimated kilocalorie consumed to varying sucrose concentrations (Fig. 2I–L; licks Friedman test = 19.86; p < 0.001; kilocalorie Friedman test = 24.57; p < 0.0001; n = 14 mice) but did not change interlick intervals (Fig. 2L; Friedman test = 1.857; p > 0.05; n = 14 mice). Mice significantly decreased their licks at 32% compared with 8% sucrose (Dunn’s z = 3.3; p = 0.0029) and 16% sucrose (Dunn’s z = 3.024; p = 0.0075; Fig. 2J) likely reflecting the higher estimated kilocalorie of 32% compared with both 8 and 16% sucrose (32 × 8% Dunn’s z = 4.914; p < 0.0001; 32 × 16% Dunn’s z = 3.024; p = 0.0075; Fig. 2K). VGluT2+ VGaT− neurons significantly increased maximum neuronal activity from the baseline in response to each sucrose concentration (Fig. 2M,N; Friedman test = 53.18; p < 0.001; n = 14 mice; Dunn-corrected tests BL × 8% z = 3.232; p = 0.0037; BL × 16% z = 4.445; p < 0.001; BL × 32% z = 4.243; p < 0.001). Thirty-two percent sucrose-induced maximum neuronal activity was significantly greater than 8% sucrose-induced neuronal activity (Friedman test = 11.57; p = 0.0031; n = 14 mice; Dunn’s test z = 3.402; p = 0.002; Fig. 2N). Sustained activity of VGluT2+ VGaT− neurons was also significantly elevated in the 32% concentration compared with 8% measured by AUC (Friedman test = 19.00; p < 0.001; n = 14 mice; Dunn’s z = 4.347; p < 0.001), as well as time to half maximum (Friedman test = 20.57; p < 0.001; n = 14 mice; Dunn’s 8 × 32% z = 4.536; p < 0.001; Fig. 2O,P), suggesting VGluT2+ VGaT− neurons signal in both phasic and sustained changes in activity.
VGluT2+ TH+ neuron-targeted mice showed different numbers of licks and estimated kilocalorie consumed in response to varying sucrose concentrations (Fig. 2Q–T; licks Friedman test = 6.2; p = 0.0456; kilocalorie Friedman test = 18.2; p < 0.0001; n = 10 mice) but did not change interlick intervals (Fig. 2T; Friedman test = 0.3684; p > 0.05; n = 10 mice). Mice significantly decreased their licks at 32% compared with 8% sucrose (Dunn’s z = 2.46; p = 0.0417; Fig. 2R) and had higher estimated kilocalorie consumed comparing both 32 and 8% sucrose as well as 32 and 16% sucrose (32 × 8% Dunn’s z = 4.249; p < 0.0001; 32 × 16% Dunn’s z = 2.46; p = 0.0417; Fig. 2S). VGluT2+ TH+ neurons significantly increased maximum neuronal activity from the baseline in response to each sucrose concentration (Fig. 2U,V; Friedman test = 40.97; p < 0.001; n = 10 mice; Dunn-corrected tests BL × 8% z = 3.586; p = 0.001; BL × 16% z = 3.227; p < 0.0038; BL × 32% z = 3.944; p < 0.0002). In contrast to VGluT2+ VGaT+ and VGluT2+ VGaT− neurons, VGluT2+ TH+ neurons showed no significant differences in maximum activity levels between sucrose concentrations (Fig. 2V; Friedman test = 4.2; p = 0.1352; n = 10 mice). However, the sustained activity of VGluT2+ TH+ neurons was significantly elevated in the 32% concentration compared with 8%, as well as in the 16% concentration compared with 8%, measured by both AUC (Friedman test = 15.8; p < 0.001; n = 10 mice; Dunn’s 32% z = 3.801; p < 0.0004; 16% z = 2.907; p = 0.011) and time to half maximum (Friedman test = 15; p < 0.001; n = 10 mice; Dunn’s 32% z = 3.354; p = 0.0024; 16% z = 3.354; p = 0.0024; Fig. 2W,X), suggesting VGluT2+ TH+ neurons use sustained changes in activity to discriminate the sucrose reward value. Together, sucrose rewards increase the activity of all VTA VGluT2+ subpopulations, but their signaling dynamics are cell-type specific. VGluT2+ VGaT+ neurons scaled sucrose-induced activity in signal amplitude, VGluT2+ VGaT− neurons scaled sucrose-induced activity with signal amplitude and duration, and VGluT2+ TH+ neurons scaled sucrose-induced activity in duration of activity but not in signal amplitude.
VTA VGluT2+ subpopulations differentially signal the consumption of sucrose and fat rewards
We next assessed whether sucrose reward signaling patterns were influenced by the presence of a second reward by recording VGluT2+ subpopulations in response to consumption of sucrose and a behaviorally preferred, higher-calorie intralipid fat solution (Sakamoto et al., 2015). VGluT2+ VGaT+ recorded mice had significantly more licks for fat than sucrose (Fig. 3A,B; Wilcoxon z = −2.819; p = 0.005; n = 20 mice) as well as higher estimated kilocalorie from fat consumption (Fig. 3C; Wilcoxon z = −3.92; p < 0.0001) but did not differ in the interlick interval (Fig. 3D; Wilcoxon z = −0.966; p > 0.05). VGluT2+ VGaT+ neurons significantly increased maximum neuronal activity from the baseline in response to each reward (Fig. 3E,F; Friedman test = 53.13; p < 0.001; n = 20 mice; Dunn-corrected tests BL × sucrose z = 5.756; p < 0.001; BL × fat z = 4.042; p = 0.0001). Despite the animal's behavioral preference for fat, VGluT2+ VGaT+ maximum neuronal activity was significantly greater following sucrose consumption compared with fat consumption (Wilcoxon z = −3.808; p < 0.001; Fig. 3F), as was AUC (z = −2.613; p = 0.009), but not time to half maximum (z = −1.157; p > 0.05; Fig. 3G,H).
VTA VGluT2+ subpopulations differentially signal preferred consummatory rewards. A, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). VGluT2+ VGaT+ VTA neurons were recorded during consumption of 8% intralipid fat and 8% sucrose in a two-bottle choice. B–D, The number of licks (B), estimated kilocalorie consumed (C), and interlick intervals (D). E, VGluT2+ VGaT+ normalized dF (z-score) during sucrose and fat consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. Eight percent sucrose is purple; fat is orange. F, Maximum neuronal activity of individual mice compared with that of the baseline. G, Sustained activity analysis, AUC of individual mice. H, Sustained activity analysis, time to half maximum (sec) of individual mice. I, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre-dependent and Flp-lacking GCaMP6m (Con/Foff where C, Cre; F, Flp; on, must express; off, must not express). VGluT2+ VGaT− VTA neurons were recorded during consumption of 8% intralipid fat and 8% sucrose in a two-bottle choice. J–L, The number of licks (J), estimated kilocalorie consumed (K), and interlick intervals (L). M, VGluT2+ VGaT− normalized dF (z-score) during fat and sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. N. Maximum neuronal activity of individual mice compared with that of the baseline. O, Sustained activity analysis, AUC of individual mice. P, Sustained activity analysis, time to half maximum (sec) of individual mice. Q, Experimental setup: VGluT2::Cre/TH::Flp mice were injected in VTA with AAVs encoding Cre and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). VGluT2+ TH+ VTA neurons were recorded during consumption of 8% intralipid fat or 8% sucrose in a two-bottle choice. R–T, The number of licks (R), estimated kilocalorie consumed (S), and interlick intervals (T). U, VGluT2+ TH+ normalized dF (z-score) during fat and sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. V, Maximum neuronal activity of individual mice compared with that of the baseline. W, Sustained activity analysis, AUC of individual mice. X, Sustained activity analysis, time to half maximum (sec) of individual mice. *p < 0.05; **p < 0.005; ***p < 0.001, ****p < 0.0001.
VGluT2+ VGaT− recorded mice had significantly more licks for fat than sucrose (Fig. 3I,J; Wilcoxon z = −2.48; p = 0.013; n = 14 mice) as well as higher estimated kilocalorie from fat consumption (Fig. 3K; Wilcoxon z = −3.233; p = 0.0012) but did not differ in the interlick interval (Fig. 3L; Wilcoxon z = −0.734; p > 0.05). VGluT2+ VGaT− neurons significantly increased maximum neuronal activity from the baseline in response to each reward (Fig. 3M,N; Friedman test = 33.43; p < 0.001; n = 14 mice; Dunn-corrected tests BL × sucrose z = 4.099; p < 0.001; BL × fat z = 3.513; p = 0.0009). However, VGluT2+ VGaT− neurons showed an opposite reward signaling preference compared with VGluT2+ VGaT+ neurons. VGluT2+ VGaT− maximum neuronal activity was significantly greater following fat consumption compared with sucrose (Wilcoxon z = −2.04; p = 0.041; n = 14 mice; Fig. 3N), as were the sustained activity measures AUC (z = −2.48; p = 0.013) and time to half maximum (z = −2.652; p = 0.009), consistent with their behavioral preference for higher-calorie fat (Fig. 3O,P).
VGluT2+ TH+ recorded mice had significantly more licks for fat than sucrose (Fig. 3Q,R; Wilcoxon z = −2.803, p = 0.005, n = 10 mice) as well as higher estimated kilocalorie consumed from fat (Fig. 3S; Wilcoxon z = −2.803, p = 0.005) but did not differ in the interlick interval (Fig. 3T; Wilcoxon z = −0.51; p > 0.05). VGluT2+ TH+ neurons significantly increased maximum neuronal activity from baseline in response to each reward (Fig. 3U,V; Friedman test = 23.52; p < 0.001; n = 10 mice; Dunn-corrected tests BL × sucrose z = 3.464; p = 0.0011; BL × fat z = 2.771; p = 0.0112). VGluT2+ TH+ neurons differed from both VGluT2+ VGaT+ and VGluT2+ VGaT− neurons in that they did not show a signaling preference for either reward in maximum activity levels (Wilcoxon z = −1.376; p > 0.05; n = 10 mice; Fig. 3V). However, the sustained activity measures AUC (z = −2.09; p = 0.037) and time to half maximum (z = −2.652; p = 0.008) were significantly greater following fat consumption compared with sucrose (Fig. 3W,X). Together, VGluT2+ subpopulation signaling patterns involving two rewards was cell-type specific. VGluT2+ VGaT+ neurons showed a signaling bias toward the less behaviorally preferred lower-calorie sucrose solution in signal amplitude, VGluT2+ VGaT− neurons showed a signaling bias toward the behaviorally preferred higher-calorie fat solution in signal amplitude and duration, and VGluT2+ TH+ neurons showed a signaling bias toward the behaviorally preferred fat solution in duration of activity but not signal amplitude.
VTA VGluT2+ VGaT+ neuronal activity is partially regulated by satiety and related to sweetness
Given that VTA VGluT2+ VGaT+ neurons showed higher phasic signaling for sucrose reward over fat, we next evaluated whether this signaling pattern was sensitive to prefeeding or sweetness of the reward. To test this, we prefed a subset of mice recording VGluT2+ VGaT+ neurons their daily chow ration for 1 h and then consumed sucrose and fat postprandially (Fig. 4A). Mice had significantly more licks for fat over sucrose (Wilcoxon z = −2.275; p = 0.023; n = 12 mice; Fig. 4B), likely reflecting the higher estimated kilocalorie consumed from fat (Wilcoxon z = −3.059; p = 0.0022; Fig. 4C), and had no difference in interlick intervals between rewards (Wilcoxon z = −1.177; p = 0.239; Fig. 4D). VGluT2+ VGaT+ neurons significantly increased maximum neuronal activity from the baseline in response to each reward (Fig. 4E,F; Friedman test = 29.3; p < 0.001; n = 12 mice; Dunn-corrected tests BL × sucrose z = 4.269; p < 0.001; BL × fat z = 3.32; p = 0.0018). However, prefeeding abolished the previously observed sucrose signaling bias by VGluT2+ VGaT+ neurons (Wilcoxon z = −1.49; p > 0.05; Fig. 4F). There were no differences between fat or sucrose-induced sustained signaling measures of AUC (z = −1.334; p = 0.182) or time to half maximum (z = −0.863; p = 0.388; Fig. 4G,H).
VTA VGluT2+ VGaT+ neuronal signaling is partially modified by prefeeding and sweetness. A, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). Mice were prefed their daily food ration for 1 h, and then VGluT2+ VGaT+ VTA neurons were recorded during consumption of 8% intralipid fat and 8% sucrose in a two-bottle choice. B–D, The number of licks (B), estimated kilocalorie consumed (C), and interlick intervals (D). E, VGluT2+ VGaT+ normalized dF (z-score) during sucrose and fat consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. Eight percent sucrose is purple; fat is orange. F, Maximum neuronal activity of individual mice compared with that of the baseline. G, Sustained activity analysis, AUC of individual mice. H, Sustained activity analysis, time to half maximum (sec) of individual mice. I, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). VGluT2+ VGaT− VTA neurons were recorded during consumption of 0.3% saccharine and 8% sucrose in a two-bottle choice. I, J–L, The number of licks (J); estimated kilocalorie consumed (K); and interlick intervals (L). M, VGluT2+ VGaT+ normalized dF (z-score) during saccharine and sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. The 0.3% saccharine is green; 8% sucrose is purple. N, Maximum neuronal activity of individual mice compared with that of the baseline. O, Sustained activity analysis, AUC of individual mice. P, Sustained activity analysis, time to half maximum (sec) of individual mice. Q, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). Mice were prefed their daily food ration for 1 h, and then VGluT2+ VGaT+ VTA neurons were recorded during consumption of 0.3% saccharine or 8% sucrose in a two-bottle choice. R–T, The number of licks (R), estimated kilocalorie consumed (S), and interlick intervals (T). U, VGluT2+ VGaT+ normalized dF (z-score) during saccharine and sucrose consumption (means are the solid line; SEMs are shading). X-axis timepoint zero is aligned to the lick bout onset. Eight percent sucrose is purple; 0.3% saccharine is green. V, Maximum neuronal activity of individual mice compared with that of the baseline. W, Sustained activity analysis, AUC of individual mice. X, Sustained activity analysis, time to half maximum (sec) of individual mice. *p < 0.05; **p < 0.005; ***p < 0.001, ****p < 0.0001; ns not significant.
We next assessed whether the VGluT2+ VGaT+ sucrose signaling bias was influenced by calories or sweetness. To test this, a subset of mice recording VGluT2+ VGaT+ neurons consumed 8% sucrose and a sweeter, noncaloric 0.3% saccharine solution (Fig. 4I). Mice showed no difference in licks (Wilcoxon z = −1.014; p = 0.31; n = 7 mice; Fig. 4J) despite the higher estimated kilocalorie consumed from sucrose (Wilcoxon z = −2.366; p = 0.0179; Fig. 4K). There was no interlick interval difference between rewards (Wilcoxon z = −0.845; p = 0.398; Fig. 4L). VGluT2+ VGaT+ neurons significantly increased maximum neuronal activity from the baseline in response to each reward (Fig. 4M,N; Friedman test = 19.29; p = 0.0002; n = 7 mice; Dunn-corrected tests BL × sucrose z = 2.484; p = 0.026; BL × saccharine z = 3.12; p = 0.0019). Maximum neuronal activity was significantly greater following saccharine consumption compared with sucrose (Wilcoxon z = −2.366; p = 0.018; Fig. 4N), as was AUC (z = −2.028; p = 0.043), but not time to half maximum (z = 0; p > 0.05; Fig. 4O,P).
To identify whether the saccharine-signaling preference was sensitive to prefeeding, we prefed a subset of mice recording VGluT2+ VGaT+ neurons for 1 h and then they consumed 8% sucrose and 0.3% saccharine postprandially (Fig. 4Q). Mice showed no difference in licks (Wilcoxon z = −1.363; p = 0.173; n = 6 mice; Fig. 4R) despite the higher estimated kilocalorie consumed from sucrose (Wilcoxon z = −2.201; p = 0.0277; Fig. 4S). There was no interlick interval difference between rewards (Wilcoxon z = −1.153; p = 0.249; Fig. 4T). VGluT2+ VGaT+ neurons significantly increased maximum neuronal activity from the baseline in response to saccharine but not sucrose reward (Fig. 4U,V; Friedman test = 15.2; p < 0.001; n = 6 mice; Dunn-corrected tests BL × sucrose z = 2.236; p = 0.0506; BL × saccharine z = 3.13; p = 0.0035). Maximum neuronal activity was again significantly greater following saccharine consumption compared with sucrose (Wilcoxon z = −1.992; p = 0.046; Fig. 4V) but was not different in AUC (z = −1.363; p = 0.173) or time to half maximum (z = −0.734; p = 0.463; Fig. 4W,X). Together, VTA VGluT2+ VGaT+ neurons showed a signaling bias toward sweet rewards in signal amplitude, and this bias was influenced by prefeeding when mice compared sucrose and fat rewards but not when mice compared sucrose and saccharine rewards.
VTA VGluT2+ subpopulations differentially scale neuronal activity with footshock intensity
In addition to their reward signaling, VGluT2+ subpopulations are highly sensitive to aversive stimuli (Root et al., 2018; McGovern et al., 2021, 2024). Therefore, we next assessed whether VTA VGluT2+ subpopulations differentially signaled aversive value by recording VGluT2+ subpopulations in response to different intensities of unsignaled footshocks (0.25, 0.50, 0.75, or 1.00 mA). VGluT2+ VGaT+ neurons significantly increased maximum neuronal activity from the baseline in response to each footshock intensity and peaked at 0.75 mA (Fig. 5A–F; Friedman test = 51.25; p < 0.001; n = 9 mice; Dunn-corrected tests BL × 0.25 mA z = 2.791; p = 0.021; BL × 0.50 mA z = 3.849; p = 0.0005; BL × 0.75 mA z = 4.234; p < 0.0001; BL × 1.00 mA z = 2.983; p = 0.0114). Examining maximum neuronal activity at each footshock intensity showed that 0.75 mA footshock-induced neuronal activity was significantly greater than 0.25 mA footshock-induced neuronal activity (Friedman test = 10.47; p = 0.015; n = 9 mice; Dunn’s test z = 3.104; p = 0.0115; Fig. 5F). VGluT2+ VGaT+ sustained neuronal activity was also significantly elevated in the 0.75 mA condition compared with 0.25 mA measured by AUC (Friedman test = 10.33; p = 0.0159; Dunn’s z = 2.921; p = 0.0209) and time to half maximum (243 ± 74 ms longer at 0.75 mA; Friedman test = 9.133; p = 0.0276; Dunn’s z = 2.921; p = 0.0209; Fig. 5G,H).
VTA VGluT2+ subpopulations differentially signal footshock intensity. A, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). VGluT2+ VGaT+ VTA neurons were recorded during unsignaled footshock deliveries of differing intensity (0.25–1.00 mA). B–E, VGluT2+ VGaT+ normalized dF (z-score) during unsignaled footshock delivery of 0.25 mA (B, blue), 0.50 mA (C, purple), 0.75 mA (D, orange), and 1.00 mA (E, red). F, Maximum neuronal activity of individual mice compared with that of the baseline. G, Sustained activity analysis, AUC of individual mice. H, Sustained activity analysis, time to half maximum (sec) of individual mice. I, Experimental setup: VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre-dependent and Flp-lacking GCaMP6m (Con/Foff where C, Cre; F, Flp; on, must express; off, must not express). J–M, VGluT2+ VGaT− normalized dF (z-score) during unsignaled footshock delivery of 0.25 mA (J, blue), 0.50 mA (K, purple), 0.75 mA (L, orange), and 1.00 mA (M, red). N, Maximum neuronal activity of individual mice compared with that of the baseline. O. Sustained activity analysis, AUC of individual mice. P, Sustained activity analysis, time to half maximum (sec) of individual mice. Q, Experimental setup: VGluT2::Cre/TH::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent GCaMP6m (Con/Fon where C, Cre; F, Flp; on, must express). R–U, VGluT2+ TH+ normalized dF (z-score) during unsignaled footshock delivery of 0.25 mA (R, blue), 0.50 mA (S, purple), 0.75 mA (T, orange), and 1.00 mA (U, red). V, Maximum neuronal activity of individual mice compared with that of the baseline. W, Sustained activity analysis, AUC of individual mice. X, Sustained activity analysis, time to half maximum (sec) of individual mice. *p < 0.05; **p < 0.005; ***p < 0.001, ****p < 0.0001.
VGluT2+ VGaT− neurons significantly increased neuronal activity from the baseline in response to each footshock current and peaked at 1.00 mA (Fig. 5I–N; Friedman test = 79.05; p < 0.001; n = 14 mice; Dunn’s multiple-comparison tests BL × 0.25 mA z = 2.932; p = 0.0135; BL × 0.50 z = 4.0890; p = 0.0002; BL × 0.75 mA z = 4.861; p < 0.001; BL × 1.00 mA z = 5.401; p < 0.001). Examining neuronal activity at each footshock intensity showed that both 1.00 and 0.75 mA footshock-induced maximum neuronal activities were significantly greater than 0.25 mA footshock-induced neuronal activity (Fig. 5N; Friedman test 12.60; p = 0.0056; n = 14 mice; Dunn’s z = 3.074; p = 0.0127; z = 3.074; p = 0.0127), respectively. However, there was no significant difference in sustained activity measured by AUC (Friedman test = 2.675; p = 0.4476) or time to half maximum (Friedman test = 6.857; p = 0.0766; Fig. 5O,P).
VGluT2+ TH+ neurons significantly increased neuronal activity from the baseline in response to each footshock current except 0.25 mA and peaked at 1.00 mA (Fig. 5Q–V; Friedman test = 56.81; p < 0.001; n = 10 mice; Dunn’s multiple-comparison tests BL × 0.25 mA z = 2.419; p = 0.0622; BL × 0.50 z = 3.88; p = 0.0004; BL × 0.75 mA z = 3.651; p = 0.001; BL × 1.00 mA z = 4.656; p < 0.001). While there was no significant difference in maximum neuronal activity between footshock currents (Friedman test = 5.88; p = 0.1176) or change in time to half maximum (Friedman test = 6.96; p = 0.0732), AUC was significantly greater at 1.00 mA compared with 0.25 mA (Friedman test = 9.996; p = 0.0189; Dunn’s z = 2.944; p = 0.0194; Fig. 5V–X). Together, aversive footshocks increase the activity of all VTA VGluT2+ subpopulations. However, VGluT2+ VGaT+ neurons scaled footshock-induced activity in signal amplitude and duration, VGluT2+ VGaT− neurons scaled footshock-induced activity in signal amplitude, and VGluT2+ TH+ neurons less reliably scaled footshock-induced activity in duration of activity but not signal amplitude.
VTA VGluT2+ VGaT+ optogenetic stimulation increases fear in response to low-intensity footshock but does not support place preference or aversion
Based on the ability of VTA VGluT2+ VGaT+ neurons to scale both rewarding and aversive stimulus values in signal amplitude, we reasoned that stimulation of VTA VGluT2+ VGaT+ neurons may influence general salience of motivationally relevant outcomes. Specifically, we hypothesized that optogenetically stimulating VTA VGluT2+ VGaT+ neurons during low-intensity aversive stimuli would result in enhanced aversion-related behavior. To test this, we injected VGluT2::Cre/VGaT::Flp mice in VTA with AAVs encoding Cre- and Flp-dependent channelrhodopsin tethered to mCherry (ChR2; n = 9 mice) or mCherry (n = 9 mice; Fig. 6A–I). By whole-cell recordings, we found that 470 nm light activation of ChR2 elicited robust inward currents and reliable action potentials in response to 5 ms photostimulus pulses delivered at 20 Hz (Fig. 6A–F). Given that we previously found VTA VGluT2+ VGaT+ neurons did not signal footshock-associated pavlovian cues (Root et al., 2020), we delivered low-intensity footshocks (0.25 mA) once per minute over 10 min while optogenetically stimulating VTA VGluT2+ VGaT+ neurons at each aversive stimulus and assessing freezing levels after each footshock (Fig. 6J). A mixed ANOVA yielded a significant interaction between shock number and stimulation group (F(9,108) = 3.54; p = 0.019; n = 18 mice). Sidak-adjusted pairwise comparisons showed no significant differences in freezing between groups following the first four footshocks (p > 0.05) after which ChR2 mice showed significantly more freezing than mCherry mice following the fifth, sixth, eighth, and ninth footshocks (all p < 0.05; Fig. 6K). At the final footshock, no group differences in freezing were again observed. Finally, we tested whether optogenetic stimulation of VTA VGluT2+ VGaT+ neurons alone results in general reward or aversion using real-time place conditioning (Fig. 7A). Mice initially explored a three-chamber apparatus without stimulation (pretest); then 473 nm light was delivered to VTA when mice were within Side 1 (initial stimulation) or Side 2 (reversal stimulation). For ChR2 mice, a 3 (side) × 3 (stage) within-subject ANOVA yielded a significant side × stage interaction (F(4,28) = 4.035; p = 0.010). Sidak-adjusted pairwise comparisons found no significant differences in time spent between stages within each side, as well as between sides within each stage, except for time in the middle chamber compared with the nonstimulated chamber during reversal (p = 0.021; Fig. 7B). For mCherry mice, the 3 × 3 within-subject ANOVA yielded no significant side × stage interaction (F(4,28) = 0.867; p > 0.05; Fig. 7C). Finally, comparing stimulated and nonstimulated sides across stimulation and reversal conditions, we found no statistically significant changes in locomotor speed (no group or group interactions with stimulation conditions or side; all F < 2.4; p > 0.05). Together, stimulation of VTA VGluT2+ VGaT+ neurons at the time of a low-value aversive stimulus increased freezing behavior, and this stimulation was not inherently rewarding or aversive, nor induced changes in locomotor speed.
VTA VGluT2+ VGaT+ neuron activation during footshock increases fear-related behavior. A, VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent ChR2–mCherry. mCherry-expressing VTA neurons were whole-cell recorded in response to 470 nm light pulses (5 ms). B, Representative ChR2-induced current. C, Photoevoked ChR2 currents were significantly larger than spontaneously observed currents; t(8) = 6.66; p < 0.001. Nine neurons from three mice. D, Representative ChR2–mCherry neuron showing high-fidelity action potentials in response to 20 Hz ChR2 photoactivation. E, Action potentials evoked by 20 Hz stimulation were reliable across stimulation trains. Seven of nine ChR2–mCherry neurons had 100% action potential fidelity in response to photoactivation. F–H, VGluT2::Cre/VGaT::Flp mice were injected in VTA with AAVs encoding Cre- and Flp-dependent ChR2–mCherry or mCherry, and a fiber optic was implanted dorsal to VTA. Representative viral expression and fiber of Con/Fon-mCherry mice (F) and Con/Fon-ChR2-mCherry mice (G). Fiber localizations shown in H. I, Experimental setup: mice were delivered 10 0.25 mA footshocks, once per minute. During footshocks, 473 nm light (5 ms pulse duration, 20 Hz) was delivered to VTA. J, Time spent freezing each minute following single 0.25 mA footshocks. *p < 0.05. Scale bar in F, 500 µm, and applies to G.
VTA VGluT2+ VGaT+ neuron ChR2 activation does not support real-time place conditioning. A, Experimental setup: ChR2 or mCherry mice explored a three-chamber apparatus with two outer chambers and a middle connecting chamber over three stages (pretest, initial stimulation, reversal stimulation). After pretest, ChR2 was activated when mice entered Side 1 for initial stimulation. After initial stimulation, ChR2 was activated when mice entered Side 2 for reversal stimulation. B, C, Time in each chamber during each stage for ChR2 mice (B) and mCherry mice (C).
A subset of single VTA VGluT2+ VGaT+ neurons are transcriptionally activated by rewarding and aversive stimuli
We next examined whether single VTA VGluT2+ VGaT+ neurons are sensitive to both rewarding and aversive stimuli. Mice were placed on an ad libitum doxycycline-infused diet (40 mg/kg) and injected in VTA with a mix of the pan-neuronal robust activity marking (RAM)-mKate vector AAV1-RAM-TetR-mKate2 and the TRE, Cre, and Flp-dependent vector encoding EGFP (AAV1-TRE-FSF-FLEX-EGFP; Fig. 8A). Three weeks later, mice were food-restricted to 85% ad libitum feeding body weight with doxycycline diet and trained to consume 16% sucrose. Mice were then provided daily diet without doxycycline to open the window of RAM-labeling while maintaining bodyweights over 3 d. After the third day, mice consumed 16% sucrose for 1 h and were returned to the colony with normal diet. After 3 d mice were administered ten 0.5 mA footshocks and perfused 90 min later to examine footshock-related c-Fos labeling in mKate2 and GFP coexpressing neurons (sucrose consumption RAM-activated VGluT2+ VGaT+ neurons). VGluT2::Cre single transgenic control showed very low GFP coexpression with mKate2 (mean, 1.25 neurons/section), whereas double transgenic VGluT2::Cre/VGaT::Flp mice observed 27.65 ± 1.44 RAM-activated GFP coexpressing neurons/section (n = 5 mice). Of mKate2 and GFP coexpressing neurons in VGluT2::Cre/VGaT::Flp mice, c-Fos coexpressing neurons induced following footshock were 49.72 ± 3.77% and were largely confined to the interfascicular nucleus in the ventral midline VTA (Fig. 8B–D). mKate2 and GFP coexpressing neurons that lacked c-Fos were largely found outside the interfascicular nucleus and represented 50.28 ± 3.77% of mKate+ GFP+ neurons. Thus, at the single neuron level, about half of transcriptionally reward-activated VGluT2+ VGaT+ neurons, especially within the interfascicular nucleus, are also transcriptionally activated by footshock.
Single VGluT2+ VGaT+ neurons are transcriptionally activated by sucrose consumption and footshock. A, Experimental setup. VGluT2::Cre/VGaT::Flp mice were injected with a mix of AAVs encoding RAM-gated mKate and tet trans activator (tTA) as well as a tet-response element (TRE)-, Cre-, and Flp-gated GFP. Mice were maintained on doxycycline diet (40 mg/kg). Mice were removed from doxycycline diet and consumed sucrose. Days later, mice received footshocks and were euthanized to examine c-Fos protein 90 min later. B, Example VTA section. Red, mKate; green, GFP; blue, c-Fos. Scale bar, 500 µm. C, High magnification of areas outlined in B. Top are example mKate+ GFP+ c-Fos neurons (white in merge), and bottom are example mKate+ GFP+ c-Fos− neurons (yellow in merge). Scale bar+ 50 µm. D, Mean ± SEM of mKate+ GFP + c-Fos+ neurons and mKate+ GFP+ c-Fos− neurons.
Discussion
VTA VGluT2-expressing neurons participate in reward and aversion-related behaviors (Root et al., 2014a, 2018, 2020; Wang et al., 2015; Qi et al., 2016; Yoo et al., 2016; Morales and Margolis, 2017; Barbano et al., 2020; Zell et al., 2020; McGovern et al., 2021, 2024). However, subtypes of VTA VGluT2+ neurons exist, most defined by the release of one or more neurotransmitters, and the behavioral roles of each cell type are not understood. Using double transgenic mouse lines and INTRSECT viral vectors (Fenno et al., 2020), we and others have shown that GCaMP transfected Cre and Flp coexpressing VTA neurons in VGluT2::Cre/VGaT::Flp mice coexpress VGluT2 and VGaT mRNA, that GCaMP transfected Cre and not Flp VTA neurons in VGluT2::Cre/VGaT::Flp mice express VGluT2 without VGaT mRNA, and that GCaMP transfected Cre and Flp coexpressing VTA neurons in VGluT2::Cre/TH::Flp mice coexpress VGluT2 mRNA and TH mRNA (Root et al., 2020; Miranda-Barrientos et al., 2021; Prevost et al., 2025). We previously found that VTA VGluT2+ VGaT+, VGluT2+ VGaT−, and VGluT2− VGaT+ neurons show pavlovian task-related changes in neuronal activity (Root et al., 2020; McGovern et al., 2021). Here, we sought to identify whether VTA VGluT2+ subtypes show differences in their sensitivity toward different aspects of rewarding or aversive experiences.
We first examined cell-type–specific signaling of reward value via different sucrose reward concentrations versus water. In both VGluT2+ VGaT− and VGluT2+ VGaT+ subpopulations, maximum activity levels were greater for the highest sucrose concentration compared with those for the lowest, suggesting that these subpopulations integrate an element of the reward value. Water consumption did not result in neuronal activation in any cell type, although the low number of water licks prevented a systematic investigation of water-reward dynamics.
In contrast to other VGluT2+ cell types, VGluT2+ TH+ neurons scaled sustained, but not maximum activity, with higher sucrose concentrations. Sucrose concentration-dependent changes in sustained neuronal activity were also identified in VGluT2+ VGaT− neurons. VGluT2+ VGaT+ neurons showed increased AUC but no change in time to half maximum across sucrose concentrations, suggesting AUC was influenced by maximum activity. We infer that VGluT2+ subpopulations signal sucrose reward value in different ways: VGluT2+ VGaT+ neurons by changes in magnitude and not sustained activity, VGluT2+ VGaT− neurons by changes in both magnitude and sustained activity, and VGluT2+ TH+ neurons by sustained and not maximum activity levels. One possibility for VGluT2+ TH+ sustained activity patterns might be to regulate dopamine release that operates on longer timescales than glutamate. A second possibility is that VGluT2+ subtypes buffer calcium differentially.
We next investigated how VGluT2+ subpopulations signaled multiple presented consummatory rewards. VTA neurons are sensitive to consumption of fat, caloric sweet, and noncaloric sweet rewards (Beeler et al., 2012; McCutcheon et al., 2012; Rada et al., 2012), but it is unclear how VTA VGluT2+ subpopulations signal these rewards when presented together. Behaviorally, mice preferred fat over sucrose reward, likely due to the high calorie content of fat compared with sucrose. Fat preference persisted prefeeding before the task, indicating that fat preference is not completely dependent on hunger drive. In VTA VGluT2+ VGaT− neurons, maximum and sustained activity were higher following fat consumption compared with sucrose, consistent with the animal's behavioral preference for fat. VGluT2+ TH+ neurons showed higher sustained activity, not maximum activity, for fat over sucrose. Surprisingly, VGluT2+ VGaT+ neurons showed higher maximum activity for sucrose over fat. Because our sucrose and fat solutions differed in macronutrients and calories, it is unclear how nutritional factors interacted with behavioral preference and neuronal signaling dynamics. Mice sampled solutions for multiple days before recordings, and thus we cannot disentangle preference from potential learned associations between taste and postingestive effects. Nevertheless, both VGluT2+ VGaT− and VGluT2+ TH+ neurons showed elevated activity patterns that varied with kilocalorie consumption, VGluT2+ TH+ neurons in sustained patterns and VGluT2+ VGaT− neurons in both maximum and sustained patterns. VGluT2+ VGaT+ neuron phasic activity varied with kilocalorie consumption when sucrose concentrations were presented alone, but not when fat and sucrose were presented together nor when sucrose and saccharine were presented together, which in these cases preferentially signaled the sweeter reward option. We infer that each VGluT2+ subpopulation is sensitive to multiple rewards, but their consumption-related signaling likely depends on different factors.
When examining the neuronal responses to fat, but not sucrose, it was observed that there were sometimes two peaks. While the second peak may reflect a resumption of licking after a brief pause, we do not observe this second component on the sucrose spout, which we assume would have similar lick architecture based on interlick intervals. The second peak was largest in VGluT2+ TH+ neurons that showed sustained reward-related responses. We hypothesize the second peak to reflect an enhanced long-duration signal rather than a second phasic signal. However, further research will be necessary to identify the potential signaling properties of this second peak.
Due to their unique signaling of sucrose over fat, we evaluated if VGluT2+ VGaT+ neuronal signaling was influenced by prefeeding and whether a noncaloric sweet reward also increases their activity. While VGluT2+ VGaT+ neurons increased activity in response to both fat and sucrose consumption, the neuronal preference for sucrose over fat was abolished by prefeeding. This signaling modulation may be regulated by neurons within the lateral hypothalamus (Morgane, 1961; Castro et al., 2015), a primary input to VTA VGluT2+ VGaT+ neurons (Prévost et al., 2024), but requires testing. Additionally, the macronutrient composition of the animals’ standard chow diet may have biased a hunger state toward fat rather than sucrose following prefeeding. Prefeeding kept bodyweight at ∼85% ad libitum feeding levels and thus was not designed to fully satiate that might reduce consumption too low for analysis. This resulted in no licking differences between prefed and nonprefed conditions. We interpret our prefed condition testing as a small reduction in hunger but not full satiety. Future studies will be necessary to delineate the interactions between macronutrient composition, satiety state, and neuronal activity.
To test the influence of calories or sweetness, mice consumed both saccharine and sucrose in a two-bottle choice. Mice showed no preference for consuming one reward over the other despite more calories in the sucrose solution. Saccharine consumption activated VTA VGluT2+ VGaT+ neurons to a greater degree than sucrose. This saccharine-signaling preference remained following prefeeding, suggesting caloric content is not a critical factor for VTA VGluT2+ VGaT+ signaling. Given that saccharine is sweeter than sucrose (Moskowitz, 1970), we interpret the larger VGluT2+ VGaT+ signaling to saccharine over sucrose, as well as sucrose over fat, as a signal related to sweetness. However, because saccharine also has bitter qualities (Kuhn et al., 2004) and VGluT2+ VGaT+ neurons are activated by aversive events (Root et al., 2020; McGovern et al., 2024), it is possible that bitter taste influenced VGluT2+ VGaT+ signaling. Together, VTA VGluT2+ VGaT+ neurons appear to signal sweet rewards that do not depend on caloric content or behavioral preference, and this signaling is partially modulated by prefeeding.
In addition to reward signaling, we found that all VGluT2+ subtypes were activated by footshock. Activation by aversive stimuli appears to be an essential feature of all VTA VGluT2+ subtypes examined thus far (Root et al., 2018, 2020; Barbano et al., 2020; McGovern et al., 2021, 2024; Prevost et al., 2025). However, the aversion-related dynamics differed between VGluT2+ subtypes. In response to increasing footshock intensity, VGluT2+ VGaT− neurons scaled in maximum but not sustained activity, and VGluT2+ TH+ neuronal activity did not scale. VTA VGluT2+ VGaT+ neurons scaled in both magnitude and sustained activity with increasing footshock intensities. The phasic and tonic aversion-related VGluT2+ VGaT+ activity supports prior research showing their higher c-Fos activity following inescapable stress compared with VGluT2+ VGaT− and VGluT2+ TH+ neurons (McGovern et al., 2024). A limitation of our experiments was the lack of a consummatory comparison of rewarding solutions with an aversive solution, such as using bitter quinine.
The unique sensitivity of VTA VGluT2+ VGaT+ neurons to aversive value, together with their reward sensitivity, is consistent with these neurons playing a role in general salience. We confirmed that at the single neuron level, about half of sucrose-transcriptionally activated VGluT2+ VGaT+ neurons are also transcriptionally activated by footshock. These neurons were almost entirely localized to the interfascicular nucleus, and neurons activated by only sucrose were found outside the interfascicular nucleus. A limitation of this experiment is that whether footshock-activated neurons are transcriptionally activated by reward was not examined. To further assess potential salience, we optogenetically activated VTA VGluT2+ VGaT+ neurons while mice received unsignaled low-intensity footshocks and measured the development of fear-related freezing. Activation of VGluT2+ VGaT+ neurons during low-intensity footshock led to higher rates freezing than fluorophore controls. However, optogenetic stimulation during real-time place conditioning failed to result in preference or aversion. We interpret these results such that while optogenetic activation of VGluT2+ VGaT+ neurons does not induce valence-related behaviors, they may amplify unconditioned stimuli relevant to the animal. Stimulation also failed to change locomotor speed, indicating that the observed increased freezing levels were not a consequence of changes in locomotion. Given that optogenetic stimulation of VTA VGluT2+ neurons can result in either preference or aversion (Root et al., 2014a; Wang et al., 2015; Yoo et al., 2016; Bimpisidis et al., 2020; Zell et al., 2020), as well as promote or disrupt reward-seeking behavior (Yau et al., 2016; Yoo et al., 2016; Tsou et al., 2023), our results suggest that different subpopulations of VGluT2+ neurons are capable of differentially affecting motivated behavior.
Interpretations of our results are limited by additional factors. First, the anatomical locations and clustering of targeted neuronal populations and photometry fibers likely differ between cell types. While VTA expression and fiber placement were verified in all mice, it is unclear if group differences in fiber distribution impacted the neuronal recordings. Second, because Con/Foff vectors in VGluT2::Cre/VGaT::Flp mice selects non-GABA glutamate neurons, this targeting strategy does not exclude glutamate–dopamine neurons. In other words, the VGluT2+ VGaT− neurons recorded here reflect the summation of glutamate-only and glutamate–dopamine neurons. Indeed, glutamate–dopamine and glutamate-only neurons exhibit different activation profiles to motivationally salient stimuli (Prevost et al., 2025) which are not captured in the present results.
In conclusion, we identify that genetically distinct VTA glutamatergic subpopulations show differences in their signaling of consummatory rewards and aversive experiences. While all VGluT2+ subpopulations signal rewarding and aversive experiences, VGluT2+ subtypes have different phasic magnitude and sustained activity profiles in response to the value of consummatory rewards, multiple presented rewards, and the value of aversive stimuli. In particular, VTA VGluT2+ VGaT+ neurons showed a preference for sweet reward that is partially modulated by prefeeding but does not depend on caloric content. While activation of VTA VGluT2+ VGaT+ neurons was not inherently rewarding or aversive, activation of VTA VGluT2+ VGaT+ neurons amplified aversive value by increasing freezing behavior in response to low-intensity footshocks. Based on these, results we hypothesize that VTA VGluT2+ VGaT+ neurons have a role in signaling the general salience of a variety of behavioral experiences.
Footnotes
This research was supported by the Webb-Waring Biomedical Research Award from the Boettcher Foundation (D.H.R.), The Shurl and Kay Curci Foundation (E.D.P.), the Institute for Cannabis Research (D.H.R.), the National Institute on Drug Abuse DA047443 (D.H.R.) and DA035821 (C.P.F.), and National Institute on Mental Health Grants MH125569 (D.J.M.) and MH132322 (A.L.). We thank Dr. Yingxi Lin for the advice on the use of RAM. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Prism, Photoshop, and Biorender were used to make figures and schematics.
The authors declare no competing financial interests.
- Correspondence should be addressed to David H. Root at david.root{at}colorado.edu.
