Adolescent THC exposure history reduces morphine-induced locomotion, with no impact on morphine reward

We first exposed adolescent mice (PD30) to a daily 5 mg/kg/day dose of THC (or its vehicle) for 14 days (Fig. 1A). This is considered a moderate dose in mice that is comparable to doses self-administered by humans [31, 32], and 14 days of administration leads to long-lasting physiological and behavioral changes in mice and rats [4, 33,34,35,36,37,38,39]. Following vehicle or THC administration, mice were left in their home cages undisturbed besides weekly cage changes until postnatal day 70, at which time morphine administration commenced (Fig. 1B). To test how adolescent THC exposure history impacts morphine reward, we conducted a conditioned place preference (CPP) test using 10 mg/kg morphine. We induced a significant preference to the morphine-paired side in both vehicle and THC exposed animals, to an approximately equivalent extent (paired t-tests vehicle, saline t10 = 0.3108, p = 0.7623; vehicle, morphine t11 = 2.352, p = 0.0384; THC, morphine t31 = 3.136, p = 0.0037; Fig. 1C).

Fig. 1: Effects of adolescent THC exposure on behavioral responses to morphine administration later in life.

A Timeline of THC or vehicle injections during adolescence. Injections were repeated once per day for 14 total days. B Timeline of morphine injections and behavioral testing during adulthood. C Time spent in the morphine-paired chamber in the pre-test and post-test (in the vehicle/saline-paired group, both chambers were paired with saline). Vehicle/saline 762 s pre, 778 s post, p = 0.22; Vehicle/morphine 783 s pre vs. 887 s post, p = 0.038; THC/morphine 785 s pre vs. 887 s post, p = 0.0017. D Mice treated with THC during adolescence showed an overall reduction in morphine-induced locomotion relative to mice treated with saline during adolescence. Repeat measures two-way ANOVA between conditions p = 0.045; Tukey’s multiple comparisons tests morphine day 1 p = 0.68, morphine day 2 p = 0.24, morphine day 3 p = 0.31, morphine day 4 p = 0.50, morphine day 5 p = 0.75. Vehicle/saline-treated mice are shown for comparison but were not included in statistical tests. Two-way ANOVA only included morphine injection days, not habituation days. For this and all figures, error bars = SEM, ns p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. E Repeated morphine administration led to a reduction in the time spent in the center of an open field relative to mice treated with saline during adolescence, and saline during adulthood. Adolescent THC exposure blunted this reduction. One-way ANOVA p = 0.027; pairwise t-tests, vehicle/saline vs. vehicle/morphine 22.66 s vs. 12.55 s, p = 0.030; vehicle/saline vs. THC/morphine 22.66 s vs. 15.85 s, p = 0.077; vehicle/morphine vs. THC/morphine 12.55 s vs. 15.85 s, p = 0.59. F There was no difference in locomotion during the five-minute open field test between groups. Two-way ANOVA p = 0.63. G Repeated morphine administration led to a reduction in the time spent in the open arms of an elevated plus maze, indicative of anxiety-like behavior relative to mice treated with saline during adolescence, and saline during adulthood. Adolescent THC exposure prevented this morphine-induced anxiety. One-way ANOVA p = 0.0004; pairwise t-tests, vehicle/saline vs. vehicle/morphine 34.97 s vs. 17.84 s, p = 0.035; vehicle/saline vs. THC/morphine 34.97 s vs. 42.39 s, p = 0.35; vehicle/morphine vs. THC/morphine 17.84 s vs. 42.39 s, p = 0.0002. H Repeated morphine administration led to an elevation in the number of marbles buried, indicative of anxiety-like behavior relative to mice treated with saline during adolescence, and saline during adulthood. Adolescent THC exposure prevented this morphine-induced anxiety. One-way ANOVA p = 0.0006; pairwise t-tests, vehicle/saline vs. vehicle/morphine 0.93 vs. 3.67, p = 0.0048; vehicle/saline vs. THC/morphine 0.93 vs. 0.50, p = 0.83; vehicle/morphine vs. THC/morphine 3.67 vs. 0.50, p = 0.0006. I Mice given saline vs. THC during adolescence showed no difference in baseline mechanical sensitivity. Vehicle/morphine vs. THC/morphine 0.85 g vs 0.86 g, p = 0.93. J Repeated morphine administration led to a reduction in the 50% mechanical threshold, indicative of pain hypersensitivity relative to mice treated with saline during adolescence, and saline during adulthood. Adolescent THC exposure blunted this morphine-induced pain hypersensitivity. One-way ANOVA p = 0.0032; pairwise t-tests, vehicle/saline vs. vehicle/morphine 1.09 vs. 0.42, p = 0.0027; vehicle/saline vs. THC/morphine 1.09 vs. 0.69, p = 0.052; vehicle/morphine vs. THC/morphine 0.42 vs. 0.69, p = 0.30.

To determine whether adolescent THC treatment impacted morphine-induced locomotion, we measured locomotor behavior while mice were in an open field. Following a two-day habituation protocol where mice were given saline injections, 10 mg/kg morphine was given once per day for five days to test the sensitization or tolerance to morphine’s effects on locomotion. We observed that morphine increased locomotion the most following the first dose, and this elevation was reduced with each subsequent dose, for both groups (two-way ANOVA conditions term F1,62 = 6.776, p = 0.0115; Fig. 1D). We also found that adolescent THC exposure overall blunted morphine-induced locomotion during the five-day morphine administration window, though no individual day was significantly different between the groups following multiple comparisons corrections. These data indicate that adolescent THC exposure reduced morphine-induced locomotion following morphine exposure later in life.

Adolescent THC exposure prevents the development of anxiety-related behavior and pain hypersensitivity during morphine withdrawal

We next assessed whether mice treated with THC during adolescence were differentially impacted by withdrawal from repeated morphine administration. Following the five days of morphine given for locomotor testing and two days of no morphine for CPP extinction, mice were left in their home cages for an additional 8 days, for a total abstinence period of ten days (Fig. 1B). Following this ten-day forced abstinence, mice were tested for anxiety-like behavior using three tests: the open field test, elevated plus maze, and marble burying tests. In the open field test, mice are both compelled to explore new environments, and avoid open areas; less time spent exploring the center of the arena is interpreted as anxiety-like behavior. We found that repeated morphine injections reduced time spent in the center of the open field, and THC history reduced this anxiety-like behavior (One-way ANOVA F2,59 = 3.838, p = 0.0271; multiple comparisons-corrected unpaired t-test vehicle/morphine vs. vehicle/saline 95% CI −19.40 to −0.8231, p = 0.0298; Fig. 1E). Notably, no differences were observed in overall locomotion during this five-minute test (One-way ANOVA F2,59 = 0.4621, p = 0.6322; Fig. 1F). The elevated plus maze tests how mice balance their tendency to explore against their preference for enclosed spaces, and reduction in the time spent exploring the open arms is interpreted as anxiety-like behavior. Mice treated with repeated morphine injections spent less time in the open arms of the elevated plus maze relative to those treated with repeated saline injections (One-way ANOVA F2,59 = 9.026, p = 0.0004; multiple comparisons-corrected unpaired t-test vehicle/morphine vs. vehicle/saline 95% CI −33.23 to −1.021, p = 0.0347; Fig. 1G). However, adolescent THC exposure blocked the reduction in time spent in the open arms induced by repeat morphine injection, suggesting that adolescent THC exposure suppressed withdrawal-induced anxiety behaviors in response to adult opioid exposures (multiple comparisons-corrected unpaired t-test THC/morphine vs. vehicle/morphine 95% CI 10.64 to 38.46, p = 0.0002; Fig. 1G). Lastly, in the marble burying test mice tend to bury marbles that are exposed in the bedding of a new cage, and more marbles buried is interpreted to reflect increased anxiety-like behavior. We found that mice undergoing forced abstinence following repeat morphine injections buried more marbles after the 30-min test, whereas adolescent THC administration reduced this number, consistent with a reduced anxiety-like behavior (one-way ANOVA F2,42 = 8.931, p = 0.0006; multiple comparisons-corrected unpaired t-tests vehicle/morphine vs. vehicle/saline 95% CI 0.7527 to 4.714, p = 0.0048; THC/morphine vs. vehicle/morphine 95% CI −5.073 to −1.261, p = 0.0006; Fig. 1H). Therefore, the results from each of these tests support the conclusion that adolescent THC exposure decreases the expression of anxiety-like behaviors in morphine-treated mice.

Next, we used a von Frey assay to examine the effects of adolescent THC exposure on mechanical hypersensitivity, both before and after morphine injection. In addition to an increase in anxiety-like behavior, mice undergoing physiological withdrawal following repeated opioid administration show an increased pain sensitivity [40, 41]. Mice treated with repeated THC doses during adolescence showed no difference in mechanical thresholds when tested one day prior to morphine CPP testing (unpaired t-test t44 = 0.08606, p = 0.9318; Fig. 1I). However, when tested again following repeat morphine exposure and forced abstinence, mice given repeated morphine showed an increased mechanical hypersensitivity relative to saline-treated controls, and adolescent THC treated reduced the extent of this hypersensitivity (one-way ANOVA F2,44 = 6.550, p = 0.0032; multiple comparisons-corrected unpaired t-tests vehicle/morphine vs. vehicle/saline 95% CI −1.143 to −0.2146, p = 0.0027; Fig. 1J; THC/morphine vs. vehicle/morphine 95% CI −0.1656 to 0.7097, p = 0.2973). Together, these results indicate that adolescent THC exposure reduces the expression of morphine withdrawal-induced behaviors following repeated morphine administration during adulthood.

Adolescent THC exposure enhances drug-induced reinstatement of morphine CPP

Following forced abstinence, we next assessed whether mice differentially reinstated CPP following a 5 mg/kg exposure of morphine in the CPP box. Drug exposure following abstinence can reinstate previously extinguished drug-associated behaviors such as CPP, and therefore can serve as a model of relapse in mice. Both groups of morphine-treated mice showed normal CPP (e.g., Fig. 1C). Following CPP, mice then underwent extinction (Fig. 2A). While mice treated with vehicle during adolescence did not significantly reinstate CPP following the priming 5 mg/kg morphine injection using this protocol, mice treated with THC during adolescence did significantly reinstate their CPP (paired t-tests vehicle/morphine t9 = 1.039, p = 0.3260; THC/morphine t7 = 3.212, p = 0.0148; Fig. 2B). This means that adolescent THC exposure facilitates reinstatement of CPP in response to a moderate priming morphine dose during adulthood.

Fig. 2: Mice treated with THC during adolescence show an elevation in drug-induced reinstatement of morphine-seeking behaviors later in life.

A Time spent in the morphine-paired chamber during CPP pre-test, post-test, extinction, and reinstatement. Mice treated with THC or vehicle during adolescence showed a similar trajectory of CPP and extinction. B Quantification of the time spent in the morphine-paired chamber in the reinstatement task relative to the final extinction day. Vehicle/morphine-treated mice 660 s vs 732 s, p = 0.33, n = 10; THC/morphine-treated mice 781 s vs. 1015 s, p = 0.015, n = 8.

Whole-brain analysis shows differences in brain activity following drug-primed reinstatement in adolescent vehicle- vs. THC-treated mice

Given that adolescent THC-treated mice showed morphine-induced reinstatement whereas vehicle-treated controls did not, we next assessed what brain regions may be responsible for this effect. To do this, we perfused the mice 60 min following the reinstatement test, cleared the brains using iDISCO + , and immunostained for the immediate early gene cFos to obtain a brain-wide map of recent neuronal activity (Fig. 3A–C). We observed an overall elevation of recent neuronal activity in response to 5 mg/kg morphine in mice treated with THC during adolescence than vehicle, as evidenced by a higher number of cFos+ cells upon morphine exposure (unpaired t-test vehicle/morphine vs. THC/morphine t24 = 3.079, p = 0.0051; Fig. 3D, E). This increase was exhibited in most brain regions, while only a small number of brain regions showed a reduction in cFos labeling in adolescent THC-treated mice, which included both the medial and lateral habenula (Supplemental Figs. 1–3, Supplemental Table 2). We counted cells manually to test the accuracy of cells detected by ClearMap, and found significant correlation between manual counts and ClearMap counts across four representative regions: the globus pallidus externus (GPe), prelimbic area, hippocampus, and midline group of the dorsal thalamus (MTN) (Supplemental Table 3–4, Supplemental Fig. 5). These regions tended to show differences in cFos staining and were expected to show different trends of over or undercounting of cFos labeling by ClearMap. The averaged ratio of the number of ClearMap counts to the number of manual counts was 0.843 for the GPe, 1.12 for the prelimbic area, 1.14 for the hippocampus, and 1.18 for the MTN. Linear regression models were fit on ClearMap counts as an independent variable and manual counts as a dependent variable, with R^2 values of 0.86 for GPe, 0.78 for prelimbic area, 0.98 for hippocampus, and 0.91 for MTN. Slope coefficients for the models all were relatively close to each other at 0.976 ± 0.0371 for hippocampus, 0.766 ± 0.0623 for MTN, 1.054 ± 0.109 for GPe, and 0.920 ± 0.125 for the prelimbic area. These regression results imply that trends in cFos labeling detected by ClearMap reflect trends in the actual cFos labeling.

Fig. 3: Brain-wide activity patterns in adolescent vehicle- and THC-treated mice following a reinstatement dose of morphine.

A Representative sagittal section of a cFos-stained brain of an animal treated with vehicle during adolescence and following a reinstatement dose of morphine. Scale, 2 mm. B Representative section of a cFos-stained brain of an animal treated with THC during adolescence and following a reinstatement dose of morphine. C Representative sagittal images of the GPe in adolescent vehicle-treated vs. THC-treated mice. Scale, 500 μm. D More cFos-expressing cells were detected in adolescent THC-treated mice than vehicle-treated mice following a dose of 5 mg/kg morphine (averages of 548,481 vs. 325,789 cells, p = 0.0051). E Pie chart showing the percentage of brain regions with elevated or depressed cFos-expressing cells in adolescent THC-treated mice relative to adolescent vehicle-treated mice. F Anatomical connectivity matrix of the mouse brain, broken into four communities based on the Louvain method for community detection. G Log fold-change in number of cFos labeled cells between THC and Vehicle for all regions grouped by network community (One-way ANOVA p = 0.0002; purple (1) vs. blue (2) 0.68 vs. 0.75, p = 0.76; purple (1) vs. red (3) 0.68 vs. 0.61, p = 0.82; purple (1) vs. yellow (4) 0.68 vs. 0.38, p = 0.0024; blue (2) vs. red (3) 0.75 vs. 0.61, p = 0.31; blue (2) vs. yellow (4) 0.68 vs. 0.38, p = 0.0002; red (3) vs. yellow (4) 0.61 vs. 0.38, p = 0.05). H The number of hierarchical clusters in adolescent vehicle-treated and THC-treated mice across a range of heights for cluster cut-offs in hierarchical trees. The THC-treated mice show reduced modularity across a range of cutoff values. The dotted red line in this and (I, J) equals the cutoff used for cluster definition in (I, J) (0.75). I Hierarchical clustering of regions based on the cFos correlation matrix in adolescent vehicle-treated mice. The VTA is within a separate activity-defined cluster from the frontal cortex regions of interest (prelimbic cortex, PL; infralimbic cortex, ILA; anterior insula cortex, AI; anterior insula cortex, ventral part, AIv; ACAd, anterior cingulate cortex, dorsal part; ACAv, anterior cingulate cortex, ventral part). J Hierarchical clustering of regions based on the cFos correlation matrix in adolescent THC-treated mice. The VTA co-clusters with regions of the anterior cortex in these mice.

As the total cFos labeling was increased in the brains of mice treated with THC, we wanted to assess whether this elevation was larger for some pathways or whether the activation was uniform across the brain. To test this, we built a quantitative connectivity matrix of the mouse brain, using the Allen Mouse Brain Connectivity Atlas as a template [42]. Doing so revealed the existence of four major network communities within the brain, largely consisting of subcortical regions including the VTA (community 1, purple), cortical regions (community 2, blue), the hippocampus and related connections (community 3, red), and mid/hindbrain regions (community 4, yellow) (Fig. 3F). We then assessed the relative increase in cFos labeling of regions in each community in adolescent THC- vs. vehicle-treated mice following a reinstatement opioid dose. We found a significant increase in the log fold-change of cFos-labeled cells in THC vs vehicle-treated mice in the purple and blue communities relative to the yellow community, suggesting that the subcortical and cortical modules both showed an overall larger increase in recent activity relative to mid/hindbrain regions in THC-treated mice (One-way ANOVA F3,254 = 6.638, p = 0.0002; multiple comparisons-corrected unpaired t-tests purple vs. yellow 95% CI 0.08193 to 0.5109, p = 0.0024; blue vs. yellow 95% CI 0.1467 vs. 0.5937, p = 0.0002; Fig. 3G).

Following this analysis, we wanted to assess if these community-specific changes in cFos activation were accompanied by any changes in the functional relationships of different brain regions. To quantitatively assess functional modularity, we built a brain-wide functional correlation matrix for adolescent THC-treated and vehicle-treated mice based on their cFos labeling (Figs. 3H–J, S4). We performed hierarchical clustering on both correlation matrices across a range of parameters. Across a wide parameter range, adolescent THC-treated mice exhibited fewer recent activity-defined clusters than vehicle-treated mice, reflecting an overall reduced modularity in brains from adolescent THC-treated mice (Fig. 3H). We then compared a specific clustering in this range to see how various structures and known circuits are organized. We focused specifically on the VTA and frontal cortex regions, given their interconnectivity and known contributions to substance misuse [43,44,45,46]. While in vehicle-treated mice the VTA and frontal cortex regions separated into different clusters, reflecting their separation into different networks (Fig. 3I), in THC-treated mice the IL, PL, ACC, and AI all co-clustered with the VTA, indicating a higher level of correlated activity amongst these regions in adolescent THC-treated mice (Fig. 3J). These findings overall show that the increased activation of cortical and subcortical pathways in adolescent THC-treated vs vehicle-treated mice after reinstatement is accompanied by a collapse in modularity of pathways in adolescent THC-treated mice, including a tighter clustering of the VTA with frontal cortex regions.

Adolescent THC exposure changes connectivity to ventral tegmental area dopamine cells

The VTA is a critical brain region for the development of a variety of drug-induced behavioral adaptations, including drug reward, sensitization, and withdrawal. To test whether adolescent THC exposure modified the input control to VTADA cells, we mapped inputs to these cells using the rabies virus (RABV) monosynaptic input mapping method [19, 47]. RABV-labeled input cells were observed throughout the brain, as in previous studies (Fig. 4A–C) [19, 47]. Overall, the input labeling pattern was significantly different between vehicle-treated and THC-treated mice (2-way ANOVA interaction F21,176 = 2.058, p = 0.0062; Fig. 4D). While several brain regions showed a visible difference in connectivity, for example elevated connectivity from cortical regions comprising the anterior cortex (infralimbic cortex, prelimbic cortex, anterior insula cortex, anterior cingulate cortex, and the anterior motor cortex) [19, 43, 47, 48], parabrachial nucleus (PBN) and deep cerebellar nuclei (DCN), only the anterior cortex was significantly different following correction for multiple comparisons (multiple comparisons-corrected unpaired t-test 95% CI 0.2923 to 10.66, corrected p-value p = 0.029).

Fig. 4: Adolescent THC exposure changes input connectivity to VTADA cells.

A Schematic of experiments. B Representative image of the ventral midbrain of a DAT-Cre mouse showing starter cells in the VTA and medial SNc. Scale, 1 mm. C Representative images of input cell populations in several brain sites, including the anterior cortex, NAcMed, NAcCore, NAcLat, MHb, LHb, and PBN. Scale, 250 μm. D Bar graph plot showing the percentage of RABV-labeled inputs in control vs. THC-treated mice. Two-way ANOVA, interaction p = 0.0062. Only the anterior cortex was significantly different following correction for multiple comparisons (corrected p-value p = 0.029).

This traditional approach to assessing differences in connectivity is statistically conservative, and only assesses potential differences in connectivity, one input at a time. To reveal further patterns in the data, we have recently employed a dimensionality reduction approach that enables us to detect larger-scale differences in input patterns between brains of mice undergoing a variety of treatments [47, 49]. We therefore first used this approach to assess which inputs best differentiated THC-treated vs. control mice using principal component analysis (PCA). We found that 3 PCs were sufficient to explain approximately 75% of the variance in the data, and thus we focused only on these 3 PCs (Fig. 5A, B). We found that PC1 completely separated THC-treated and control mice, whereas PC2 and PC3 did not (Fig. 5C–D). Control mice had high PC1 values, which were driven predominately by strong contributions from nucleus accumbens medial shell (NAcMed), core (NAcCore), ventral pallidum (VP), septum, lateral habenula (LHb) and medial habenula (MHb), while THC-treated mice had low PC1 values, which were driven by the bed nucleus of the stria terminalis (BNST), extended amygdala area (EAM), entopeduncular nucleus (EP), paraventricular nucleus of the hypothalamus (PVH), lateral hypothalamus (LH), and PBN (Fig. 5E). To quantify these differences, we plotted each brain’s location on each PC, and performed t-tests. Differences between THC-treated and control groups were strongly significant along PC1 (unpaired t-test t8 = 6.559, p = 0.0002), and not different along PCs 2 and 3, as expected (unpaired t-test PC2 t8 = 0.1407, p = 0.8916; unpaired t-test PC3 t8 = 0.7424, p = 0.4791; Fig. 5F–H). To further quantify relationships between brains and conditions, the Euclidean distance between each brain in the PC1 versus PC2 coordinate space was calculated, and the relationships plotted as a correlogram. THC-treated and control brains separated largely by condition, as expected (Fig. 5I).

Fig. 5: Identification of inputs driving the difference in connectivity to VTADA cells following adolescent THC exposure.

A Schematic and explanation for how PCA analysis was carried out on the RABV tracing dataset. B Proportion of cumulative variance explained by each principal component. C Plot of PC1 and PC2 from brains of control and THC-treated mice. D Plot of PC1 and PC3 from brains of control and THC-treated mice. E Heatmap of the contributions of each brain region, or feature, in the data to PCs 1 through 3. Comparison of total PC values for each brain along (F) PC 1, (G) PC 2, and (H) PC 3. F Control 3.3 vs. THC −2.2, p = 0.0002. G Control 0.13 vs. THC −0.09, p = 0.89. H Control 0.57 vs. THC −0.38, p = 0.48. I To further quantify relationships between brains and conditions, the Euclidean distance between each brain from control and THC-treated mice in the PC1 versus PC2 coordinate space was calculated. These distances were then plotted in a heatmap with brains being organized by similarity assessed by hierarchical clustering.