Cell Type–Specific Encoding of Cocaine‐Conditioned Responses in the Lateral Preoptic Area

TL;DR

This study shows how different types of neurons in the lateral preoptic area respond to cocaine-related cues in mice, revealing distinct patterns linked to drug-associated behaviors.

Contribution

The study identifies cell type–specific neural activity patterns in the LPO during cocaine conditioning, linking them to conditioned behavioral outcomes.

Findings

Both glutamatergic and GABAergic neurons in the LPO showed reduced activity in the cocaine-paired chamber after conditioning.

Only glutamatergic neurons increased activity in the saline-paired chamber, correlating with preference for that context.

Reduced GABAergic activity was associated with stronger preference for the cocaine-paired chamber.

Abstract

The lateral preoptic area (LPO) is a functionally heterogeneous hypothalamic structure that is increasingly recognised for its role in motivated and drug‐seeking behaviours. Although prior studies have shown that LPO neurons exhibit diverse activity patterns during cocaine self‐administration, the specific contributions of glutamatergic and GABAergic populations to conditioned responses to cocaine remain unclear. In this study, we recorded the activity of LPO glutamatergic (vglut2‐expressing) and GABAergic (vgat‐expressing) neuronal subpopulations during cocaine conditioning to identify how these cell types respond to drug‐associated cues in awake, behaving mice.

Our results revealed that, after cocaine conditioning, both glutamatergic and GABAergic populations showed reduced activity upon entry into the cocaine‐paired chamber. However, only glutamatergic neurons exhibited increased activity when mice entered the saline‐paired chamber. The magnitude of these activity changes was also correlated with behavioural outcomes: reduced GABAergic activity was associated with greater preference for the cocaine‐paired chamber, whereas stronger glutamatergic responses were associated with greater preference for the saline‐paired chamber.

These findings indicate that LPO GABAergic and glutamatergic neurons exhibit distinct, cell type–specific patterns during cocaine‐conditioned place preference. Rather than indicating opposing motivational states, the results highlight that LPO activity relates to conditioned behaviours in a cell type– and context‐dependent manner. By linking activity in glutamatergic and GABAergic neurons with conditioned responses, this work provides initial evidence for how LPO circuits may contribute to drug‐associated learning.

Cocaine conditioning induces context‐specific and cell type–specific dependent changes in activity in glutamatergic and GABAergic neuronal populations of the lateral preoptic area.

Figures

All vgat‐cre and vglut2‐cre mice exhibited a conditioned response to the rewarding effects of cocaine. (A) Vgat‐cre (N = 7 males; N = 5 females) and vglut2‐cre (N = 6 males; N = 6 females) mice on a CB57BL/6J background were injected with a calcium sensor (AAV‐DIO‐GCaMP‐7) into the LPO and implanted with a fibre optic cannula above the injection site, to record LPO neuronal activity before and after saline and cocaine (15 mg/kg) conditioning. (B) Histology confirmed GCaMP‐7 expression in the LPO and the location of the fibre implant over the LPO (images taken at 20× magnification). (C) After 2 weeks of recovery and viral expression, mice were run in a cocaine‐CPP paradigm that included habituation, a pretest, four conditioning sessions and a posttest. Tissue was collected at the end of the experimental timeline. It was observed that (D) vgat‐cre and (E) vglut2‐cre mice spent more time in the cocaine‐paired chamber than the saline‐paired chamber after conditioning (p < 0.001*, p < 0.0001).

LPO GABA neurons are inhibited during entry into the cocaine‐paired chamber. (A, B) In vgat‐cre (N = 7 males; N = 5 females) mice, cocaine conditioning, but not saline, led to changes in calcium activity. Time 0 marks entry from the neutral chamber into the conditioned chambers. Traces show mean ΔF/F ± SEM across all subjects. (C) No significant change was observed in the AUC for the saline‐paired chamber, although a trend toward increased activity was present (F(1,11) = 4.63, p = 0.055, η 2 = 0.30). (D) In contrast, the AUC significantly decreased after entry into the cocaine‐paired chamber after conditioning relative to both the pretest and the pre‐entry baseline (p < 0.05*, p < 0.01). (E) A moderate positive correlation was observed between the average number of calcium transients per entry in the saline‐paired chamber and time spent in the saline chamber (R 2 = 0.33, p = 0.05). (F) Conversely, fewer calcium transients per entry into the cocaine‐paired chamber were associated with greater time spent in the cocaine chamber (R 2 = 0.34, p = 0.04).

LPO glutamatergic neurons are inhibited during entry into the cocaine‐paired chamber. (A, B) In vglut2‐cre (N = 6 males; N = 6 females) mice, calcium activity changed across test sessions in response to both cocaine and saline conditioning. Time 0 marks entry from the neutral chamber into the conditioned chambers. Traces show mean ΔF/F ± SEM across all subjects. (C) The AUC significantly increased during entries into the saline‐paired chamber after conditioning, relative to both the pre‐entry baseline and the pretest (*p < 0.05). (D) In contrast, the AUC decreased significantly during entries into the cocaine‐paired chamber after conditioning, with reductions observed relative to both the pretest and pre‐entry baseline (p < 0.01). (E) A strong positive correlation was found between the number of calcium transients per entry in the saline‐paired chamber and time spent in the saline chamber (R 2 = 0.48, p = 0.01). (F) However, there was no relationship between glutamatergic activity in the cocaine‐paired chamber and time spent in the cocaine chamber (R 2 = 0.05, p = 0.48).

Tables

TABLE 1. Sex‐specific analyses of cocaine reward.

	Averaged change scores		Average AUC for GABAergic neurons				Average AUC for glutamatergic neurons
	Saline	Cocaine	Pretest saline	Posttest saline	Pretest cocaine	Posttest cocaine	Pretest saline	Posttest saline	Pretest cocaine	Posttest cocaine
Male	−154.6 ± 38.78	280.45 ± 31.8	−3469.12 ± 1780.74	532.01 ± 262.89	2397.98 ± 927.37	929.62 ± 550.88	1892.99 ± 1255.98	−674.1 ± 347.94	530.93 ± 390.12	506.55 ± 818.77
Female	−238.49 ± 83.35	370.45 ± 33.83	9841.9 ± 5838.45	1083.46 ± 555.66	1155.05 ± 534.3	2758.94 ± 965.79	2121.86 ± 593.3	804.03 ± 371.21	1011.27 ± 292.1	2853.54 ± 876.43

TABLE 2. Control analyses ruling out alternative explanations for calcium activity patterns.

Potential confound	Analysis	Result	Conclusion
Locomotor activity	Ca2+ peaks vs. locomotion for vglut2‐cre and vgat‐cre mice (distance travelled) in each chamber, before and after conditioning, for both cell types*	All p > 0.05, all R 2 < 0.25	Ca2+ signals not driven by movement
Pre‐existing aversion	Pretest vglut2‐cre Ca2+ peaks (cocaine chamber) vs. pretest time spent in cocaine chamber	R 2 = 0.10, p = 0.38	No innate bias reflected in baseline activity
Baseline activity predicting preference	Pretest vglut2‐cre Ca2+ peaks (cocaine chamber) vs. posttest time spent in cocaine chamber	R 2 = 0.33, p = 0.08	Baseline activity does not predict postconditioning preference

Full text

Introduction

1

The lateral preoptic area (LPO) is an anatomically and functionally diverse hypothalamic region implicated in motivated behaviour, reward processing and drug seeking (Barker et al. 2017; Coffey et al. 2021). LPO neurons project broadly to components of the mesocorticolimbic system, including the ventral tegmental area, lateral habenula and other limbic sites, positioning the LPO to broadly regulate arousal, reinforcement and affective states (Barker et al. 2017; Coffey et al. 2021; Gordon‐Fennell et al. 2020). Although historically understudied in addiction research, growing evidence highlights the LPO as a key hub for integrating motivational signals (Barker et al. 2017, 2023).

Electrophysiological work shows that LPO neurons display heterogeneous firing patterns during cocaine self‐administration, with some increasing and others decreasing activity during drug seeking (Barker et al. 2015; Coffey et al. 2021). They also exhibit slower tonic changes in firing that track cocaine pharmacokinetics, supporting functional diversity among LPO cell types (Barker et al. 2015). Anatomical and molecular studies indicate that the LPO is composed primarily of glutamatergic (vglut2‐expressing) and GABAergic (vgat‐expressing) neurons, which exert distinct influences on downstream circuits and behaviour (Barker et al. 2017, 2023). Prior optogenetic studies indicate that activating LPO glutamatergic neurons produces aversive‐like responses, whereas activating GABAergic neurons produces reward‐related behaviours. Cocaine itself can support both conditioned place preference (CPP) and avoidance, reflecting the engagement of multiple motivational and contextual processes (Nicot et al. 2025). Together, these findings suggest that LPO cell types contribute in distinct, context‐dependent ways to cocaine‐associated motivation, rather than encoding simple or opposing motivational valence. However, no prior studies have directly assessed whether LPO glutamatergic and GABAergic populations exhibit distinct activity patterns during cocaine‐associated learning or how these patterns relate to conditioned behaviours.

To address this gap, we employed in vivo calcium imaging in genetically defined vglut2‐cre and vgat‐cre mice during a cocaine‐CPP task. This enabled examination of LPO glutamatergic and GABAergic neuronal activity before and after conditioning, and assessment of how activity patterns changed in relation to conditioned responses for a drug‐paired context. These findings provide the first cell type–specific evidence of how LPO neurons contribute to cocaine conditioning, laying the groundwork for future studies on projection‐defined circuits or targeted therapeutic strategies.

Methods

2

General Procedures

2.1

Animal Care and Housing

2.1.1

Experiments were conducted on 3‐ to 6‐month‐old male and female vgat‐cre and vglut2‐cre mice (Jackson Labs). Mice were group housed (two to five per cage) on ventilated racks in a temperature‐controlled vivarium on a 12‐h light–12‐h dark cycle (dawn on at 7 am), with standard chow (Mouse Diet 5001; LabDiet) and water available ad libitum throughout testing. All protocols complied with the Guide for the Care and Use of Laboratory Animals (NIH, Publication 865‐23) and were approved by the Institutional Animal Care and Use Committee, Rutgers University.

Surgeries

2.1.2

Mice were anaesthetised with isoflurane (1%–5% induction; 1%–3% maintenance) and received carprofen (20 mg/kg), baytril (5 mg/kg) subcutaneously and a local injection of bupivacaine (2.5 mg/kg). Vgat‐cre (N = 7 males; N = 5 females) and vglut2‐cre (N = 6 males; N = 6 females) mice received a cre‐dependent AAV‐Syn‐FLEX‐GCaMP7f‐WPRE injection (150 nL; Adgene #104492) into the LPO (Figure 1A) using a Micro4 controlled and UltraMicroPump (coordinates: AP = +0.15 to +0.3; ML = +0.85 to +0.9; DV = −5.05 to −5.10). A 400‐μm core optic fibre (0.48 NA; Doric Lenses) was implanted at the same coordinates. Three screws (AMS‐1 20‐3) were used to secure the fibre optic implant to the skull, along with dental acrylic.

All vgat‐cre and vglut2‐cre mice exhibited a conditioned response to the rewarding effects of cocaine. (A) Vgat‐cre (N = 7 males; N = 5 females) and vglut2‐cre (N = 6 males; N = 6 females) mice on a CB57BL/6J background were injected with a calcium sensor (AAV‐DIO‐GCaMP‐7) into the LPO and implanted with a fibre optic cannula above the injection site, to record LPO neuronal activity before and after saline and cocaine (15 mg/kg) conditioning. (B) Histology confirmed GCaMP‐7 expression in the LPO and the location of the fibre implant over the LPO (images taken at 20× magnification). (C) After 2 weeks of recovery and viral expression, mice were run in a cocaine‐CPP paradigm that included habituation, a pretest, four conditioning sessions and a posttest. Tissue was collected at the end of the experimental timeline. It was observed that (D) vgat‐cre and (E) vglut2‐cre mice spent more time in the cocaine‐paired chamber than the saline‐paired chamber after conditioning (p < 0.001**, p < 0.0001****).*

Postoperatively, mice recovered on heating pads, were monitored for 3 days and received daily carprofen (20 mg/kg), with Baytril (5 mg/kg) administered as needed. All animals recovered for at least 2 weeks to ensure viral expression and surgical healing before behavioural testing.

Sample Size Determination and Experimental Design

2.1.3

Sample size was determined based on established effect sizes from prior fibre photometry studies in our laboratory and published work using similar methodologies (Mejaes et al. 2024; Morishita et al. 2025). Based on medium to large effect sizes (f = 0.25–0.40) typically observed in these types of recordings, and using standard parameters (α = 0.05, power = 0.80), we aimed for n = 6–7 per group to ensure adequate statistical power while accounting for anticipated exclusions. Final groups after exclusions: vglut2‐cre (six males, six females), vgat‐cre (seven males, five females; N = 24 total).

At the start of the experiment, mice were randomly selected from the breeding colony by selecting cages without consideration of individual animal characteristics, ensuring unbiased sampling. After surgery and behavioural testing, all exclusion decisions were made by an experimenter without knowledge of behavioural outcomes. Mice were excluded for incorrect viral expression or fibre placement (Section 2.5.1) as determined by histological verification, or for insufficient signal quality (signal‐to‐noise ratio <2:1; Section 2.6.2) as assessed from photometry recordings (Mejaes et al. 2022). Linear mixed models accommodated unbalanced groups resulting from these exclusions (Section 2.7).

Conditioned Place Preference

2.2

Apparatus

2.2.1

As illustrated in Figure 1, CPP was conducted in a three‐compartment black acrylic apparatus (Stoelting Co.), consisting of two pairing chambers and a smaller neutral middle chamber. One pairing chamber had smooth floors with vertical white‐striped walls, whereas the other had textured metal floors with black walls. A camera positioned ~1 m above each apparatus recorded behaviour using AnyMaze software. During conditioning, the two larger compartments were paired with either cocaine or saline injections, and the centre chamber allowed free movement between them during pretest and posttest sessions. This design isolated the distinct sensory environments associated with each drug condition while facilitating transitions between them during preference tests.

Training Procedure

2.2.2

Mice underwent a biased CPP paradigm, where cocaine was paired with the side least preferred during the pretest, for an average duration of 1 week. After 30–45 min of habituation to the testing room, a pretest was conducted the following day. Over 4 days, eight conditioning sessions took place, with saline pairings each morning and cocaine pairings each afternoon (Figure 1C). Mice were habituated to the CPP room in cages for 15 min on pairing days. Mice were injected with 0.3 mL of 0.9% saline (i.e., physiological saline; i.p.) and isolated in the saline‐paired chamber in the morning. In the afternoon, mice were injected with 15 mg/kg of cocaine dissolved in an equivalent volume of saline (i.p.) and isolated in the cocaine‐paired chamber. Each pairing session lasted 30 min, and after conditioning, mice were given one 15‐min posttest.

Fibre Photometry Data Collection

2.3

During the CPP pretest and posttest (Figure 1C), fibre photometry data were collected using Tucker‐Davis Technologies hardware and software that was interfaced with AnyMaze to track chamber position. Implanted optical fibres were attached to the recording system via a fibre optic patch cable. GCaMP‐7 was excited at 490 nm (calcium dependent) and 405 nm (isobestic control) using amplitude‐modulated light‐emitting diodes routed through a Doric fluorescence mini cube (Lerner et al. 2015). The 405‐nm wavelength served as an isosbestic reference to correct for noncalcium‐related fluctuations such as motion and photobleaching (Bruno et al. 2021; Mejaes et al. 2022). Emitted fluorescence from both channels travelled back through the same optic fibre and passed through a 500‐ to 550‐nm emission filter before being acquired via a Doric Fluorescence Detector (Doric).

Signals were digitised at ~1 kHz using a real‐time signal processor (RZ5D; Tucker‐Davis Technologies, TDT, Alachua, FL) running Synapse software (TDT). Position data tied to zones for each CPP chamber were transmitted to the RZ5D from an AnyMaze AMi‐2 I/O device and integrated in Synapse to align behavioural and neural timestamps. To eliminate LED onset/offset artefacts, the first and last 2 s of each recording were excluded from analyses.

Histology

2.4

Mice were anaesthetised with isoflurane and transcardially perfused with a 0.1‐M phosphate buffer, followed by 4% paraformaldehyde. Brains were postfixed overnight, transferred to 18% sucrose until they sank, frozen on powdered dry ice and stored at −80°C. Coronal sections (30 μm) were sliced on a cryostat and mounted onto charged slides (Superfrost Plus).

Microscopy

2.5

Keyence BZ‐X800

2.5.1

Brains were imaged at 2× and 20× magnification on a Keyence BZ‐X800 microscope to verify viral expression and optic fibre placement (Figure 1B). Only mice with confirmed LPO‐targeted virus injection and correct fibre localisation were included in analyses; animals lacking expression or accurate placement were excluded.

Data Analysis

2.6

Across all experiments, initial analyses included sex as a between‐subjects factor. No main effects or interactions involving sex were observed (all p > 0.05; Table 1). Although all data are presented stratified by sex in figures, sex was not included as a factor in the final statistical models, and the analyses reported below combine data from both sexes.

CPP Data Analysis

2.6.1

For CPP, time spent in each chamber during the pretest and posttest was recorded with AnyMaze, which tracked the subjects' movements within the apparatus. A change score (ΔS) was calculated as the difference in time spent in the drug‐paired chamber from pretest to posttest (ΔS = posttest − pretest). AnyMaze also synchronised signals to the TDT fibre photometry system, allowing alignment of behavioural events with photometry data collected in Synapse.

Fibre Photometry Data Analysis

2.6.2

Fibre photometry data collected by the TDT hardware and software were exported to and analysed by custom MATLAB scripts. The initial and final 2 s of each recording were removed to eliminate potential rise/fall artefacts from the LEDs. Signals were initially plotted over the whole session control channel to evaluate the quality of each recording. To maintain experimental rigour, mice that did not exhibit at least a 2:1 signal‐to‐noise ratio between GCaMP‐7 transients and background noise were removed from the dataset.

For each included subject, we generated peri‐event time histograms (PETHs) and heatmaps using previously published MATLAB scripts (Barker et al. 2017). These visualisations assessed response stability across repeated entries into the cocaine‐ or saline‐paired chambers throughout the session. PETH event windows were set from 5 s before to 10 s after chamber entry (event). Pre‐entry baseline activity was defined as the 5 s immediately before chamber entry (Figure 2A,B, Figure 3A,B), during which subjects were typically located in the neutral chamber, with access to both conditioned chambers, as confirmed by position tracking.

LPO GABA neurons are inhibited during entry into the cocaine‐paired chamber. (A, B) In vgat‐cre (N = 7 males; N = 5 females) mice, cocaine conditioning, but not saline, led to changes in calcium activity. Time 0 marks entry from the neutral chamber into the conditioned chambers. Traces show mean ΔF/F ± SEM across all subjects. (C) No significant change was observed in the AUC for the saline‐paired chamber, although a trend toward increased activity was present (F(1,11) = 4.63, p = 0.055, η 2 = 0.30). (D) In contrast, the AUC significantly decreased after entry into the cocaine‐paired chamber after conditioning relative to both the pretest and the pre‐entry baseline (p < 0.05, p < 0.01**). (E) A moderate positive correlation was observed between the average number of calcium transients per entry in the saline‐paired chamber and time spent in the saline chamber (R 2 = 0.33, p = 0.05). (F) Conversely, fewer calcium transients per entry into the cocaine‐paired chamber were associated with greater time spent in the cocaine chamber (R 2 = 0.34, p = 0.04).*

*LPO glutamatergic neurons are inhibited during entry into the cocaine‐paired chamber. (A, B) In vglut2‐cre (N = 6 males; N = 6 females) mice, calcium activity changed across test sessions in response to both cocaine and saline conditioning. Time 0 marks entry from the neutral chamber into the conditioned chambers. Traces show mean ΔF/F ± SEM across all subjects. (C) The AUC significantly increased during entries into the saline‐paired chamber after conditioning, relative to both the pre‐entry baseline and the pretest (*p < 0.05). (D) In contrast, the AUC decreased significantly during entries into the cocaine‐paired chamber after conditioning, with reductions observed relative to both the pretest and pre‐entry baseline (*p < 0.01). (E) A strong positive correlation was found between the number of calcium transients per entry in the saline‐paired chamber and time spent in the saline chamber (R 2 = 0.48, p = 0.01). (F) However, there was no relationship between glutamatergic activity in the cocaine‐paired chamber and time spent in the cocaine chamber (R 2 = 0.05, p = 0.48).

Neural activity was quantified in two ways. First, the area under the curve (AUC) was calculated to capture the magnitude of calcium signal changes surrounding chamber entries. Second, the number of calcium transients (peaks) was measured across entries into each chamber to provide an index of overall neural activity during time spent in a chamber. Calcium activity in the neutral chamber was not analysed because animals spent very little time there and entries were primarily brief, transitional crossings. For correlational analyses, peaks per entry assessed relationships between neural activity and behavioural preference scores, as this reflects the dynamic frequency of calcium transients throughout the animal's time in the chamber, whereas AUC represents the total magnitude of activity over a period and may obscure more nuanced patterns of neural dynamics.

Statistical Analysis

2.7

All statistical analyses were conducted in GraphPad Prism. To examine conditioning‐induced changes within each cell type, separate one‐ and two‐way repeated measures analyses of variance (ANOVAs) were conducted for vgat‐cre mice (Section 3.3) and vglut2‐cre mice (Section 3.4). Linear mixed models were used in place of ANOVAs to handle unbalanced group sizes. Greenhouse–Geisser and Sidak post hoc corrections were applied to address violations of the sphericity assumption and control the familywise error rate, respectively. Finally, correlation analyses were performed to examine relationships between neural activity and behavioural preference scores. For these analyses, we quantified the number of calcium transients (peaks) per chamber entry, as this measure reflects the overall level of neural activity during time spent in a chamber. Pearson correlations were used unless otherwise specified.

Results

3

Cocaine Conditioning Increased Preference for the Cocaine‐Paired Chamber in Both vglut2‐cre and vgat‐cre Mice

3.1

During the CPP assay, all mice, regardless of genotype, increased time spent in the cocaine‐paired chamber after conditioning (F(1.58, 36.43) = 43.79, p < 0.0001, η ^2^ ₚ = 0.655; Figure 1D, E). Change scores (ΔS) were significantly higher for the cocaine‐paired chamber compared to both saline‐paired (t(23) = 10.03, p < 0.0001, d ₚ = 2.05) and neutral chambers (t(23) = 8.69, p < 0.0001, d ₚ = 1.77). On average, time spent in the cocaine‐paired chamber increased from 188.88 ± 22.06 to 427.35 ± 38.38 s, consistent with a conditioned preference.

Neural Correlates for Place Conditioning in LPO GABA and Glutamate Neurons

3.2

We next assessed calcium activity before and after conditioning in vgat‐ and vglut2‐cre mice to determine whether neural activity in LPO GABA neurons (Figure 2) and glutamate neurons (Figure 3) is responsive to cocaine‐associated cues. We analysed both chamber‐specific activity across the session and time‐locked responses to entries into the cocaine‐ and saline‐paired chambers to determine how neural signals relate to the expression of conditioned behaviour. Pre‐entry baseline activity was defined as the 5 s immediately before chamber entry during both the pretest and posttest. Additional control analyses ruled out potential confounds including locomotor differences, pre‐existing chamber preferences and baseline neural activity as explanations for observed patterns (Table 2). Below, we present conditioning‐induced changes in calcium activity for each cell type.

Activity of LPO GABAergic Neurons Decreased During Entries Into the Cocaine‐Paired Chamber

3.3

We first quantified calcium responses in vgat‐cre mice by calculating the AUC for activity surrounding chamber entries. After conditioning, AUC values showed a nonsignificant trend toward increasing values in the saline‐paired chamber (F(1,11) = 4.63, p = 0.055, η ^2^ = 0.30; Figure 2C), whereas AUC responses in the cocaine‐paired chamber significantly decreased (F(1,11) = 8.33, p < 0.05, η ^2^ = 0.43; Figure 2D). Posttest AUC values were reduced relative to both the pre‐entry baseline (t(11) = 3.05, p < 0.05, d ₚ = 0.88) and the pretest chamber entry (t(11) = 3.55, p < 0.05, d ₚ = 1.02).

We next examined whether these activity patterns related to behavioural preference. We observed a positive correlation between the number of calcium transients in the saline‐paired chamber and the time spent in the chamber (R ^2^ = 0.33, p = 0.05; Figure 2E). Conversely, reduced activity in the cocaine‐paired chamber was associated with greater preference for that chamber (R ^2^ = 0.34, p = 0.04; Figure 2F). Overall, GABAergic activity decreased during entries into the cocaine‐paired chamber, showed little change in the saline chamber and inversely tracked preference for the cocaine‐paired environment.

LPO Glutamatergic Neurons Are Inhibited During Entry Into the Cocaine‐Paired Chamber

3.4

In vglut2‐cre mice, conditioning altered activity in both chambers. AUC values in the saline‐paired chamber significantly increased after conditioning (F(1,10) = 5.33, p < 0.05, η ^2^ = 0.35; Figure 3C), and posttest AUCs were elevated relative to both the pre‐entry baseline (t(22) = 2.28, p < 0.05, d ₚ = 0.49) and the pretest chamber entry (t(22) = 2.17, p < 0.05, d ₚ = 0.46). In contrast, AUC responses in the cocaine‐paired chamber significantly decreased after conditioning (F(1,8) = 11.47, p < 0.01, η ^2^ = 0.59; Figure 3D), and posttest values were significantly lower than both the pretest (t(20) = 3.48, p < 0.01, d ₚ = 0.76) and the pre‐entry baseline (t(20) = 3.48, p < 0.01, d ₚ = 0.76).

To examine relationships with behaviour, we assessed calcium transients per chamber entry. A positive correlation was observed between glutamatergic activity in the saline‐paired chamber and time spent in that chamber (R ^2^ = 0.48, p = 0.01; Figure 3E). No significant relationship was found between glutamatergic activity in the cocaine‐paired chamber and time spent in the chamber (R ^2^ = 0.05, p = 0.48; Figure 3F). Overall, glutamatergic activity increased during entries into the saline‐paired chamber, decreased during entries into the cocaine‐paired chamber and only the saline‐related activity tracked behavioural preference.

Discussion

4

Our findings reveal that LPO GABAergic and glutamatergic neurons exhibit distinct activity patterns after cocaine conditioning, suggesting cell type–specific contributions to conditioned responses. After conditioning, both populations showed reduced activity when mice entered the cocaine‐paired chamber, whereas only glutamatergic neurons increased activity in the saline‐paired chamber, highlighting new evidence of cell type–specific differences in how LPO neurons respond to drug‐ and nondrug‐associated environments.

Interpreted alongside prior work, the patterns observed in LPO GABAergic neurons provide insight into how these cells respond to cocaine after conditioning. We observed decreased activity during cocaine‐paired entries and a modest increase in the saline chamber. Although these changes must be interpreted cautiously, they reveal a relationship far more intricate than simple reward encoding. Artificial stimulation of LPO GABAergic neurons has been reported as rewarding (Barker et al. 2017; Gordon‐Fennell et al. 2020), but this apparent discrepancy with our findings may highlight the diversity of LPO signalling to different downstream targets. Artificial stimulation may preferentially suppress specific circuits to produce reward, whereas the naturally occurring decrease in activity during cocaine chamber entries could reflect the disinhibition of specific LPO connections. Nonetheless, we also cannot rule out the possibility that the reduction in activity observed in LPO GABA neurons may reflect recognition of the absence of cocaine during the drug‐free test session.

LPO glutamatergic neurons showed decreased activity in the cocaine‐paired chamber and increased activity in the saline‐paired chamber, patterns that may reflect multiple nonexclusive processes. Reduced activity in the cocaine‐paired context may attenuate aversion‐related signalling and facilitate approach to the drug‐associated environment. In contrast, increased activity in the saline‐paired chamber may reflect the absence of expected reward, the relative devaluation of the nondrug context or contextual features associated with nonreinforced experience. Our prior work shows that strong experimental activation of LPO glutamatergic neurons can drive aversion‐related behaviours (Barker et al. 2017, 2023), suggesting that this population contributes to motivational signalling across a reward–aversion spectrum. However, because the present data are correlational, we cannot conclude that elevated activity in the saline‐paired chamber reflects aversion, per se. Rather, this activity may index subtler motivational states, including the omission of expected reward, anticipatory negative contrast or reduced motivational value of a nonreinforced context.

Correlational analyses further support distinct cell type–specific roles. For both cell types, increased activity correlated with more time spent in the saline chamber. However, only GABAergic neurons showed a negative correlation between activity and time in the cocaine chamber, such that decreased GABAergic activity predicted greater preference for the drug‐paired side. Overall, this pattern may best align with a disinhibition model, wherein reduced GABAergic activity releases downstream reward circuits from inhibition to enhance cocaine‐associated motivation. Although both cell types exhibited a suppression in AUC at entry into the cocaine‐paired chamber, only GABAergic neurons showed a sustained reduction in calcium activity that tracked behavioural preference. By contrast, glutamatergic neurons showed reduced overall activity without accompanying changes in peaks per entry. Moreover, only GABAergic neurons showed a contrast between saline‐ and cocaine‐related peak activity, suggesting condition‐dependent modulation of GABAergic signalling. Overall, these findings suggest that suppression manifests differently across cell types, with GABAergic neurons showing reductions in both peak‐related and overall activity, and glutamatergic neurons showing reductions primarily in overall activity.

Together, these findings provide the first evidence that LPO glutamatergic and GABAergic neurons exhibit distinct activity patterns during conditioned responses to cocaine. These findings extend prior work by identifying cell type–specific dynamics during re‐exposure to drug‐associated contexts, supporting a framework in which LPO circuits differentially process drug and nondrug environments. Future studies manipulating projection‐defined subpopulations will be needed to determine how these signals influence downstream targets such as the ventral tegmental area and lateral habenula, and whether similar patterns occur during active drug seeking. More broadly, this data highlight the LPO as an integrative area where excitatory and inhibitory signals shape motivated behaviour, offering a foundation for understanding its contribution to addiction circuitry.

Author Contributions

D.J.B. and J.I.M. designed the study. J.I.M., P.Y., R.R. and K.S. collected the data. J.I.M., R.R., K.S., P.Y. and A.B. curated and analysed the data. J.I.M., P.Y., R.R., K.S. and A.B. created visualisations. K.S. and A.B. contributed to formal analysis and investigation. P.Y. led methodology with support from K.S. J.I.M. led project administration and validation with support from D.J.B. and P.Y. D.J.B. secured funding, provided resources and software and supervised the project. J.I.M. wrote the original draft. J.I.M., R.R., K.S., P.Y., A.B. and D.J.B. reviewed and edited the manuscript. All authors approved the final version of the manuscript.

Funding

This work was supported by the National Institute on Drug Abuse, grant no. DA043572.

Conflicts of Interest

The authors declare no conflicts of interest.