Molecular Psychiatry www.nature.

com/mp

SYSTEMATIC REVIEW OPEN

Systematic review and meta-analysis on the effects of chronic

peri-adolescent cannabinoid exposure on schizophrenia-like
behaviour in rodents
Zhikun Li 1,7, Diptendu Mukherjee 2,7, Bea Duric 3, Isabelle Austin-Zimmerman1, Giulia Trotta1,4, Edoardo Spinazzola 1,4
,
✉
Diego Quattrone 1,4, Robin M. Murray 4,5 and Marta Di Forti 1,4,6

© The Author(s) 2024

BACKGROUND: The link between cannabis use and schizophrenia is well-established in epidemiological studies, especially among
adolescents with early-onset use. However, this association in rodent models is less clear. This meta-analysis examined the effects of
adolescent cannabinoid exposure on distinct schizophrenia-like behaviours in rodents and how experimental variations influence
outcomes.
METHODS: Following a pre-registered protocol (CRD42022338761), we searched PubMed, Ovid Medline, Embse and APA PsychInfo
1234567890();,:

for English-language original studies until May 2024. We synthesised data from experiments on schizophrenia-like behaviour in rats
and mice after repeated peri-pubertal (onset between P23-P45) cannabinoid exposure. Risk of bias was assessed using the SYRCLE’s
tool.
RESULTS: We included 359 experiments from 108 articles across 9 behavioural tests. We found meta-analytic evidence supporting
that CB1R agonists, both natural and synthetic, elicited broad schizophrenia-like behavioural alterations, including impaired
working memory [g = −0.56; (CI: −0.93, −0.18)], novel object recognition [g = −0.66; (CI: −0.97, −0.35)], novel object location
recognition [g = −0.70; (CI: −1.07, −0.33]), social novelty preference [g = −0.52; (CI: −0.93, −0.11)], social motivation [g = −0.21;
(CI: −0.42, −0.00)], pre-pulse inhibition [g = −0.43; (CI: −0.76, −0.10)], and sucrose preference [g = −0.87; (CI: −1.46, −0.27)]. By
contrast, effects on novelty-induced locomotion were negligible. Subgroup analyses revealed similar effects across sexes and
species. Substantial variance in the protocols and moderate-to-high heterogeneity in behavioural outcomes were observed. We
found CBD may enhance fear memory recall, but data was limited.
DISCUSSION: This is the first meta-analysis to comprehensively assess the link between cannabinoids and schizophrenia-like
behaviours in rodents. Our results support epidemiological links between early cannabis use and schizophrenia-like phenotypes,
confirming the utility of animal models. Standardising protocols will optimise models to strengthen reproducibility and
comparisons, our work provides a framework for refining rodent models to elucidate biological pathways linking cannabis and
schizophrenia.
Molecular Psychiatry (2025) 30:285–295; https://doi.org/10.1038/s41380-024-02668-5

INTRODUCTION between frequent cannabis use and psychosis [5, 6]. Recent
Cannabis is one of the oldest and most widely used psychoactive evidence showed that daily cannabis users had a three-fold higher
substances in human history [1]. The latest UN report estimated risk of developing psychosis than non-users. The risk increases
that the global number of cannabis users reached 209 million in with the use of high-potency cannabis that contains high levels of
2020, representing a 23% increase from 2010 [2]. The increasing tetrahydrocannabinol (THC), the main psychoactive ingredient in
popularity of cannabis use may be influenced by several factors, cannabis [7]. Furthermore, evidence suggested that earlier
such as the legalisation of medicinal and recreational cannabis in adolescent cannabis initiation may also confer greater psychosis
some countries, the increased availability and social acceptance, vulnerability [8]. Despite this knowledge, the biological mechan-
and the perception that cannabis has low health risks [3, 4]. isms underlying this relationship, and the role of genetics have not
However, epidemiological studies have consistently shown a link been conclusively elucidated [9].

1
Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London SE5 8AF, UK. 2MRC Centre for
Neurodevelopmental Disorders, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, London SE1 1UL, UK. 3GKT School of Medical Education,
King’s College London, London SE1 1UL, UK. 4South London and Maudsley NHS Mental Health Foundation Trust, London, UK. 5Department of Psychosis Studies, Institute of
Psychiatry, King’s College London, De Crespigny Park, Denmark Hill, London SE5 8AF, UK. 6National Institute for Health Research (NIHR) Mental Health Biomedical Research Centre
at South London and Maudsley NHS Foundation Trust and King’s College London, London, UK. 7These authors contributed equally: Zhikun Li, Diptendu Mukherjee.
✉email: [email protected]

Received: 6 October 2023 Revised: 2 July 2024 Accepted: 5 July 2024

Published online: 2 August 2024
 Z. Li et al.
286
Human studies on cannabis and psychosis face considerable

Anhedonia

preference

preference
challenges in controlling for genetics, cannabis type, consumption

Sucrose

Sucrose

index
patterns and social contexts. In contrast, rodent models enable
controlled examination of cannabis exposure effects on the brain
and behaviour. Given the conserved nature of brain circuits
between humans and rodents [10, 11], animal models hold

Attentional
set-shifting

criterion in
Number of
promise for probing the pathophysiological mechanisms under-

Cognitive
flexibility

shift trial
trials to
lying cannabis effects relevant to humans.

reach
Over the years, a growing body of research has investigated the
impact of cannabinoids on schizophrenia-like behaviours in
rodents [12–16]. While providing useful preliminary evidence,

memory

platform
these studies have substantial variability in factors like cannabi-

Time to
Spatial

hidden
Morris
water

reach
maze
noid type and dosage, timing of exposure, sex, and species of
animals used [14]. Additionally, while numerous behavioural
aspects have been reported to be impacted by cannabis exposure
in rodents, the validity and sensitivity of the individual tests used

conditioning
Associative

freezing in
to model complex schizophrenia-like behaviour in rodents remain

recall trial
memory
unclear.

%Time
To address these questions, we conducted a systematic review

Fear
and meta-analysis of rodent experiments that modelled chronic
cannabis use and assessed its link with schizophrenia-like
behaviours. We focused on studies that administered cannabi-

inhibition
Pre-pulse
Sensori-
noids during adolescence, a critical neurodevelopmental period

gating
motor

%PPI
with heightened vulnerability to substance impacts [17–21].
Through this meta-analysis, we aimed to 1) summarise existing
behavioural data of rodent experiments that modelled adolescent

location
cannabis exposure, 2) compare the impacts of distinct cannabi-

object
Novel
noids, particularly THC and cannabidiol (CBD), 3) explore the

Discrimination Index
Short-term memory
potential moderating factors of sex, species, time lapse between
treatment and assessment (short-term vs long-term), and 4)
discuss the implications for future research and identify open

recognition
questions in the field of rodent models of cannabinoids exposure.

object
Novel
METHODS
This systematic review and meta-analysis followed the PRISMA

%Correct alternations
T maze

(Preferred Reporting Items for Systematic Reviews and Meta-

Working memory

Analyses) guidelines (Supplementary Appendix 1). The protocol

was pre-registered at PROSPERO on 21st June 2022, protocol
number: CRD42022338761.
Behavioural tests and primary outcome measures for data extraction.

Y maze

Search strategy
The literature search was conducted on 5th May 2024 across
electronic databases including PubMed, EMBASE, MEDLINE and
APA PsycINFO, using the following keywords: 1) cannabis, 2)
preference

preference

animal, and 3) adolescence. (Full search terms in Supplementary

novelty
Social

Social

appendix 2). We included peer-reviewed original studies written in

index

English.
Social withdrawal

Study screening and eligibility criteria

motivation

motivation

The titles and abstracts of 2806 articles retrieved from the

Relevant schizophrenia symptoms

preliminary search were screened and cross-checked by two

Social

Social

index

independent reviewers (ZL and BD) and discrepancies were

resolved through discussion with a third researcher (DM).
Subsequently, the two reviewers evaluated the full texts of the
Total distance travelled
stimulant-

remaining articles using specific inclusion criteria adapted from

induced
Psycho-
Locomotion in open
Positive symptoms

the PICO method [22] (Additional details in Supplementary

in the open field

Appendix 3). We pre-specified 12 behavioural tests [Table 1]

relevant to the three core domains of schizophrenia symptoms:
positive symptoms, negative symptoms and cognitive impair-
induced
novelty

ments [23–26]. We pooled data from each behavioural test into

field

separate meta-analyses.

Data extraction strategy and quality assessment

Two reviewers (ZL and BD) independently extracted data using a
outcome
measure
Primary
Table 1.

standardised data extraction form (Supplementary Data Collection

Test

Sheet). Importantly, we regarded a comparison between a control

and a cannabinoid treatment group as a single experiment, and

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
287
an effect size was calculated for each such comparison/experi- While the majority of the studies used wildtype animals,
ment. Therefore, if an article included multiple control-treatment 8 studies used genetically modified mice [Fig. 1F, a descriptive
comparisons, each comparison was treated as one separate summary of gene-environment interaction findings in Supple-
experiment, and multiple effect estimate data from that article mentary appendix 4]. Regarding substances, the effects of THC
were calculated and pooled. (n = 58) have been extensively profiled, followed by the
Quality assessment was performed following the SYRCLE’s risk synthetic full CB1R agonists WIN-55212-2 (n = 26) and
of bias tool for animal studies [27], which evaluated each article for CP55,940 (n = 5) [Fig. 1G]. We also observed substantial variance
their risk of bias across 10 items, assigning a rating of high risk, low in the tested THC doses (ranging from 0.2 mg/kg to 15 mg/kg),
risk, or unclear for each item. plus 77 experiments adopting an escalating dose paradigm (e.g.,
2.5–5–10 mg/kg). Furthermore, non-contingent methods of drug
Statistical analysis delivery [31] such as intraperitoneal (n = 92) or subcutaneous
Analyses were performed in R, using packages meta, metafor and injection (n = 9) were more common than voluntary self-
RevMan. From each experiment, we calculated the effect sizes as administration [Fig. 1H].
Hedge’s g. The inverse variance-weighted random effects model Corresponding to the schizophrenia symptoms in humans, we
with Knapp-Hartung adjustment was used to calculate overall specified a list of 12 behavioural tasks. From all the experiments
effect sizes. The Restricted Maximum-Likelihood (REML) estimator (K = 359) synthesised from the 108 articles, we registered 9
was used for τ2. behavioural tests that were reported in more than 4 experiments.
The direction of effect sizes reflects the numerical change of Morris water maze (K = 10), attentional set-shifting (K = 3), and
effect in the experiment group compared to the control group. psychostimulant-induced hyperactivity in an open field (K = 5)
Data are presented as Hedge’s g ± 95% confidence intervals. were excluded due to data inaccessibility or lack of suitable
Results were regarded as significant when the confidence interval outcome measure. Amongst the 9 behavioural tests, novelty-
entirely excluded zero and corresponded to a p value lower than induced locomotor activity in an open field test was most
0.05 in Cochran’s Q test. A summary effect was considered valid examined (K = 114), followed by novel-object recognition (K = 70)
only when at least 4 individual effect estimates could be pooled. [Fig. 1I]. Within our specified range of administration onset around
Heterogeneity was assessed using the I2 statistics [28]. We puberty (P21-56), most studies began treatment in week 5 (P28-
regarded an I2 of 0–40%: might not be important; 30–60%: may P34). Finally, while a small portion of the experiments (K = 86)
represent moderate heterogeneity; 50–90%: may represent examined the short-term effects of adoslecent cannabinoid
substantial heterogeneity; 75–100%: considerable heterogeneity treatments, most studies performed experiments (K = 427) after
[29]. an abstinent period of more than ten days exclusively or
The pre-specified list of subgroup analyses included assess- repeatedly after short-term experiments [Fig. 1J].
ments for the effect of a) species, b) sex, c) substance, d) time
lapse of behavioural assessment post treatment. Cochran’s Q Exposure to CB1R agonists and schizophrenia-like behaviours
statistics was used to assess between-subgroup differences. We in rodents
regarded a subgroup result as valid only when there was a Exposure to CB1R agonists was associated with a detrimental
minimum number of 4 experiments in each subgroup. effect on working memory tasks [g = −0.56; 95% CI = (−0.93;
Publication bias was inspected qualitatively by assessing the −0.18), p = 0.005; n = 28] and short-term memory tests including
asymmetry of funnel plots and quantitatively by the Egger’s test novel object recognition [g = −0.66; 95% CI = (−0.97; −0.35),
[30]. Sensitivity analyses were conducted to evaluate the p < 0.0001, n = 63], novel object location [g = −0.70; 95% CI =
robustness of the overall effect estimates when excluding studies (−1.07; −0.33), p < 0.005, n = 25], social novelty preference
a) with high-risk level of bias and b) reported alternative outcome [g = −0.52; 95% CI = (−0.93; −0.11), p < 0.05, n = 25], as well as
measures. sensorimotor gating assessed through pre-pulse inhibition
[g = −0.43; 95% CI = (0.76; −0.1), p < 0.05, n = 33]. Rodents
exposed to CB1R agonists also displayed behaviour related to
RESULTS negative symptoms, such as reduced social motivation [g = −0.21;
A total of 2806 records were identified through database search, 95% CI = (−0.42; −0.005), p < 0.05, n = 18]. There was also
amongst which 118 studies matched our inclusion criteria. 10 reduced sucrose preference behaviour [g = −0.87; 95% CI =
articles were subsequently excluded for insufficient data, resulting (−1.46; −0.27), p < 0.05, n = 14] and a trend towards impaired
in n = 108 studies published between 2003 and 2024 being used 24 h fear memory recall [g = −0.45; 95% CI = (−0.91; 0.00),
for quantitative data synthesis [Fig. 1A, C; Supplementary Data p = 0.051, n = 11]. Notably, novelty-induced locomotion in an
Sheet]. open field was the only behaviour displaying no significant
Risk of bias assessment [Fig. 1B] revealed all included articles change following adolescent CB1R agonists treatment [g = −0.1;
pre-specified outcome measures. Most articles adequately con- 95% CI = (−0.24; 0.04), p = 0.15, n = 99].
trolled and reported baseline animal characteristics (93.5%) and
provided sources of funding/conflict of interest statements (87%). Comparing behavioural effects of THC and synthetic CB1R
However, few articles disclosed sufficient details on randomisation agonists
procedures and blinding methods as specified in the SYRCLE Synthetic Cannabinoids (SCs) encompass a broad class of artificial
assessment tool. Therefore, the majority of studies were labelled compounds that mimic the effects of phytocannabinoids such as
“unclear risks” for these domains. THC, but often at multiple times higher potency and binding
To determine the diversity of experimental paradigms applied, affinity [32]. In our meta-analysis, the SC agonists identified
we categorised studies according to the genetic background, included WIN55,212-2, CP55,940, AB-PINACA, AB-FUBINACA, 5-
species, sex, substance administered, route of drug administration, MDMB-PICA, HU-210 and JWH-018. In a subgroup analysis, we
behavioural measures, and the timing of behavioural assessment. compared behavioural impacts of these SCs versus THC. Overall,
These data are presented as pie charts [Fig. 1D–K]. We found that THC and SCs produced similar be effects [Fig. 2]. The differences,
most studies (n = 73) used rats as the model organism, while comparing THC and SCs, were non-significant for for novelty-
n = 34 studies used mice, and one study characterised both induced locomotion in an open field (Q = 0.11, p = 0.74), novel
species [Fig. 1D]. Over half of the studies included only one sex, object recognition (Q = 0.12, p = 0.73), novel object location
with male animals (n = 59) being preferred over females (n = 9) (Q = 0.15, p = 0.70), social novelty preference (Q = 0.27, p = 0.60)
[Fig. 1E]. and sucrose preference (Q = 0.16, p = 0.70). However, SCs led to

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
288

Fig. 1 Article screening and general characteristics of the studies included. A Flowchart of the screening strategy for the articles returned
from electronic databases and the number of studies excluded at each step. B Cumulative bar charts displaying risk of bias assessment across
10 items specified in SYRCLE’s assessment tool. C Histogram showing the trend in articles published every year between 2003 and 2024
included in the meta-analysis. Pie charts depicting the proportion and number of studies categorised by species (D), sex (E), genetic
background (F), substance administered (G), route of administration (H), the number of behavioural tests reported (I) and timing of
behavioural tests (short term: 24 h to 10 d after final dose; long term: >10 d after final dose) (J). OF novelty-induced locomotion in an open
field, WM working memory, NOR novel object recognition, NOL novel object location, SNP social novelty preference, PPI prepulse inhibition,
FC fear conditioning, SM social motivation, SP sucrose preference.

greater impairment than THC social motivation (Q = 6.12, was observed for novelty-induced locomotion in the open field test
p = 0.01), pre-pulse inhibition (Q = 5.40, p < 0.05) and fear [g = −0.18; 95% CI = ( − 0.42; 0.06), p = 0.89, n = 15], novel object
conditioning (Q = 18.92, p < 0.0001). We also observed a signifi- recognition [g = −0.05; 95% CI = ( − 1.18; 1.07), p = 0.91, n = 7], pre-
cant sub-group effect in working memory tests (Q = 4.28, pulse inhibition [g = 0.40; 95% CI = ( − 0.60; 1.41), p = 0.34, n = 6] and
p = 0.03), albiet this is likely due to the small sample size and an sucrose preference tests [g = 0.10; 95% CI = ( − 0.75; 0.95), p = 0.77,
outlier in the SC group. n = 5] with CBD.
Subsequently, comparing the effects of CB1R agonists and CBD,
Exposure to CBD and behavioural modulations we found a significant difference for pre-pulse inhibition (Q = 3.86,
In contrast to CB1R agonists, we found few studies assessing CBD’s p < 0.05), sucrose preference (Q = 5.50, p < 0.05) and fear memory
impact that matched our inclusion parameters [Figs. 1G and 3B]. retrieval (Q = 11.32, p < 0.005). In each of these tests, CBD
Therefore, we analysed the effect sizes for a subset of behavioural improved the performance, whereas CB1R agonists worsened it.
parameters with at least 4 effect estimates pooled [Fig. 3A]. We No difference was detected in the novel object recognition test
observed significant effects of CBD in enhancing fear memory retrieval (Q = 1.56, p = 0.21) or novelty-induced locomotion in an open
[g = 0.53; 95% CI = (0.04; 1.02), p < 0.05, n = 8]. No significant effect field (Q = 0.37, p = 0.5).

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
289

Fig. 2 THC and synthetic CB1R agonists are associated with similar schizophrenia-like behavioural modification. Boxplots displaying the
distribution of effect estimates (Hedge’s g) of THC (light blue) and synthetic CB1R agonists (dark blue) for different behavioural tests; Number
of experiments included (N), pooled effect size (ES) and corresponding significance levels are listed on top of each corresponding boxplot.
Significant subgroup differences between THC and SC estimated by Cochran’s Q statistics are denoted by asterisks below the boxplots.
*p < 0.05, **p < 0.005, ***p < 0.0005. Black horizontal lines in the boxplots indicate pooled effect sizes (inverse variance weighted, mixed effect
model), grey horizontal lines indicate medians. Outliers are indicated by grey circles. OF novelty-induced locomotion in an open field, WM
working memory, NOR novel object recognition, NOL novel object location, SNP social novelty preference, PPI prepulse inhibition, FC fear
conditioning, SM social motivation, SP sucrose preference.

Heterogeneity, publication bias and sensitivity analysis Behavioural outcomes of chronic adolescent CB1R agonist
We identified moderate to high levels of heterogeneity for most exposure across rodent species and sexes
behavioural tests, with I2 statistics ranging between 0 and 73% To further explore how variations in experiment design might
[Fig. 3C]. To identify potential sources of heterogeneity, statistical impact behavioural outcomes and contribute to between-study
outlier analyses were conducted. Outliers were defined as studies heterogeneities, we performed a series of subgroup analyses. All
in which the 95% confidence interval of the effect size did not subgroup analyses were performed with data obtained from
overlap with the confidence interval of the pooled effect. studies on CB1R agonists exposure, as CBD studies were sparse.
Statistical outliers were detected in most behavioural tests, We first compared the subgroup effect sizes for mice and rats
accounting for a portion of the heterogeneity identified for each [Fig. 5A]. Significant between-species difference was found only in
of the tests (Supplementary Appendix 5). fear conditioning recall (Q = 7.25, p < 0.05), albeit all experiments
We found significant asymmetry in 5 out of 14 funnel plots (4 out in the rat subgroup came from the same article, which bear the
of 9 behavioural outcomes measured for CB1R agonist and 1 out of risk of being skewed by the same unidentified confounds.
5 for CBD), indicating publication bias [Fig. 4]. These include novel Heterogeneity remained moderate to high in the subgroups,
object recognition [intercept = −3.859, 95% CI = (−5.17; −2.75), indicating that species was not a significant moderator in our
p < 0.001], novel object location [intercept = −4.976; 95% CI = meta-analysis.
(−7.99; −1.96), p < 0.05], social motivation tests [intercept = Figure 5B displays the subgroup effect sizes for each
−2.401, 95% CI = (−4.3; −0.50), p < 0.05], and sucrose preference behavioural test by sex. Notably, 8 out of 9 behavioural tests
test for CB1R agonists [intercept = −5.986, 95% CI = (−7.74; −4.23), revealed no significant differences between subgroups. Further-
p < 0.01] and pre-pulse inhibition for CBD [intercept = 15.371, 95% more, only the female subgroup for social novelty preference
CI = (6.44; 24.31), p < 0.05], indicating the existence of publication showed a substantial reduction in heterogeneity (I2 = 30%)
bias. relative to the overall effect, indicating that sex is not a strong
Furthermore, we performed a sensitivity analysis to assess moderator for most of the behavioural outcomes.
whether the inclusion of alternative outcome measures would
significantly modify the overall effects. Among the 9 behavioural Comparing short-term and long-term effects of CB1R agonist
tests, only open field tests assessing novelty-induced locomotion exposure
included alternative outcome measures (e.g., beam breaks, To distinguish the short-term and protracted effects of CB1R
number of floor squares entered) in addition to the pre- agonists, we stratified the data based on the timing of the
specified primary measure (total distance travelled) for locomotor behavioural tests. Specifically, we categorised tests performed
activity. The analysis result showed no significant change in between 24 h to 10 days after the final dose as short-term, and
pooled effect size before and after removing alternative measures. those conducted after an abstinence period of more than 10 days
[EStotal = −0.12, 95% CI = (−0.24; 0.28); ESprimary = −0.16, 95% as long-term. As shown in Fig. 1K, most studies assessed
CI = (−0.33; 0.01)]. Based on the risk of bias assessment results, we behaviour only in the long term. Consequently, we conducted
performed sensitivity analyses by comparing effect sizes before subgroup analyses only for novelty-induced locomotion, novel
and after removing data points with unclear or high risk. We object recognition, and prepulse inhibition, which each contained
focused on assessment items including random sequence more than four short-term outcome data points. Substantial
generation, blinding to experimenter, blinding to outcome between-subgroup difference was found for novelty-induced
assessor and addressing incomplete outcome data. Our results locomotion (Q = 4.89, p < 0.05), where CB1R agonists administra-
sugggest that studies of high and unknown risk had limited tion was linked with significantly reduced locomotion in the short-
impact on the overall effects (see Supplementary Appendix 6 for term [g = −0.50, 95% CI = (−0.86; −0.13), n = 19, p < 0.005,
summary effect sizes including only low risk experiments). I2 = 50%] but not in the long-term [g = −0.08, 95% CI = (−0.22;

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
290

Fig. 3 Chronic adolescent exposure to CB1R agonists and CBD differently modulates schizophrenia-like behaviour. A Boxplots displaying
the distribution of effect sizes (Hedge’s g) of CB1R agonists (blue) and CBD (red) across nine behavioural tests; Number of experiments
included (N), values of pooled effect sizes (ES) and corresponding significance levels (pES) are listed on top of each corresponding boxplot.
Cochran’s Q statistics to compare effect sizes between drug interventions are denoted by asterisks below the boxplots. *p < 0.05, **p < 0.005,
***p < 0.0005. Black horizontal lines in the boxplots indicate pooled effect sizes (inverse variance weighted, random effect model), grey
horizontal lines indicate medians. Outliers are indicated by grey dots. B Percentage of experiments analysing CB1R agonists (blue) and CBD
(red) in each behavioural test. C Bar charts displaying heterogeneity of each analysis represented by I2 statistic and associated p values shown
within each bar. OF novelty-induced locomotion in an open field, WM working memory, NOR novel object recognition, NOL novel object
location, SNP social novelty preference, PPI prepulse inhibition, FC fear conditioning, SM social motivation, SP sucrose preference. Forest plots
for each behavioural test are available in Supplementary Figs. S3–16.

0.07), n = 88, p = 0.32, I2 = 47%]. By contrast, there was a Our meta-analysis revealed a robust association between
significant long-term effect in the prepulse inhibtion test adolescent exposure to natural and synthetic CB1R agonists and
[g = −0.50, 95% CI = (−0.89; −0.12), n = 28, p < 0.05, I2 = 76%], impaired schizophrenia-related behavioural phenotypes in
but not in the short term [g = −0.15, 95% CI = (−0.63; 0.33), n = 6, rodents. We report that exposure to CB1R agonists is associated
p = 0.51, I2 = 21%]. Similarly, performance in novel object with prominent cognitive deficits and pronounced behaviour
recognition was shown significantly impaired in the long-term changes similar to negative symptoms of schizophrenia. Notably,
[g = −0.71, 95% CI = (−1.06; −0.36), n = 58, p < 0.005, I2 = 73%] these effects were persistent even after long-term abstinence. This
but not in the short-term [g = −0.89, 95% CI = (−1.92; 0.14), n = 8, suggests that adolescent exposure to CB1R agonists may cause
p = 0.08, I2 = 73%]. However, the subgroup comparisons of short- lasting disruptions to the brain and impaired behaviour that
term and long-term effects did not reach statistical significance for extends into adulthood in rodents. Our results are consistent with
novel object recognition (Q = 0.14, p = 0.70) or prepulse inhibition the conclusion from existing reviews on rodent literature
tests (Q = 1.77, p = 0.18). [15, 39, 40].

Locomotor hyperactivity as a proxy for positive symptoms of

DISCUSSION schizophrenia
Previous meta-analyses have synthesised preclinical evidence for Novelty-induced hyperactivity in an open field is often considered
cannabinoids effects on rodent behaviour related to nociception a proxy for positive symptoms of schizophrenia [41, 42] due to the
[33], sleep [34] anxiety and depression [35, 36], and some narrative well-established connection between dopamine and movement
systematic reviews have addressed aspects of schizophrenia- control [43–46]. Indeed, enhanced striatal and subcortical
related behaviour such as social behaviour [37] and cognitive dopaminergic activity has been reported in schizophrenia patients
function [38]. However, to our knowledge, this is the first [47–50] and rodents [51, 52]. Although dopaminergic dysfunction
systematic review to meta-analyse results from a comprehensive is not the only pathophysiological mechanism affected, it is
battery of tests for schizophrenia-like behaviour, focusing on proposed to be a final common pathway where multiple effector
adolescent cannabinoids exposure. pathways converge [53, 54] Therefore, modelling a behaviour that

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
291

Fig. 4 Contour-enhanced funnel plots for assessing potential publication bias. In each plot, effect sizes (Hedge’s g) are plotted on the x-axis
and standard errors on an inverted y-axis. Each coloured dot on the plot represents a single effect size data point. The dashed lines form an
idealised funnel shape that indicates the expected distribution of studies. The vertical line in the middle of the funnel represents the average
effect size. The contours from light to dark grey signify the significance levels of the data points, with darker shades epresenting higher
significance levels (p < 0.01, p < 0.05, and p < 0.1). A Funnel plots displaying the distribution of individual effect estimates from experiments
assessing CB1R agonists effects across nine behavioural tests. B Funnel plots displaying the distribution of individual effect estimates from
experiments assessing CBD effects across five behavioural tests.

is susceptible to dopamine-dependent changes provides con- of cannabis use under controlled and stressed conditions
struct validity to this paradigm. [62, 63]. These observations would suggest that using a simple
However, we found that adolescent exposure to CB1R agonists novelty-induced locomotion test to model positive symptoms in
did not have a significant overall effect on novelty-induced cannabis-treated rodents oversimplifies a complex relationship.
locomotor activity, and this was overall not modified by sex or Therefore, we suggest refined protocols to study positive
species. In a subgroup analysis separating short-term versus long- symptoms are necessary. For example, experiments that
term effects, we found a significant locomotor suppressing effect performed a pharmacological challenge with psychostimulants
when the test was performed more recently to treatment such as amphetamine and cocaine revealed a significantly
cessation. This could be due to some residual effects of the acute augmented difference in the locomotor activity between animals
drug-induced hypoactivity commonly observed immediately after treated with CB1R agonists and controls [64–67]. However, our
drug administration, which subsided during abstinence. current study did not identify sufficient data from these
However, the lack of evidence of a long-term effect on experiments to conduct a quantitative analysis of the overall
locomotion does not necessarily negate the hypothesis that effects. Over the years, several other methods for assessing
CB1R activation modifies dopaminergic signalling [51, 55, 56]. In positive symptom-like behaviour have also been developed,
fact, chronic long-term exposure to CB1R agonists could have a such as models of “altered reality testing” through Pavlovian
paradoxical effect on dopamine. Similar to other drug depen- conditioning [68, 69], artificial manipulation of perceptual
dence models such as with psychostimulants or opioids [57, 58], decision making [70, 71], and a more recent development which
it has been shown that chronic cannabis users often display probes experimentally controlled auditory hallucinations in
suppressed dopamine release [59–61]. Furthermore, striatal rodents [72]. Together, these improved behavioural paradigms
dopamine levels are also found to be reduced in patients with hold great potential as more robust tools for modelling positive
dual diagnoses, including schizophrenia patients with a history symptoms in rodents.

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
292

Fig. 5 Subgroup analyses revealed similar effects of CB1R agonists across sexes and species. Boxplots displaying the distribution of effect
sizes (Hedge’s g) for CB1R agonists in each behavioural test sub-grouped by A. species and B. sex. Number of experiments (N), pooled effect
size values (ES), corresponding I2statistics and P values are shown on the top of each corresponding boxplot. Black horizontal lines in the
boxplots indicate pooled subgroup effect sizes (inverse variance weighted, mixed effect model), grey horizontal lines indicate medians.
Outliers are indicated by grey dots. Significant between-subgroup differences are denoted by asterisks below the boxplots. Comparing the
effect sizes between rats and mice: OF novelty-induced locomotion in an open field: Q = 0.13, p = 0.72; WM Working memory: Q = 0.19,
p = 0.66; NOR Novel object recognition: Q = 2.14, p = 0.14; NOL Novel object location: Q = 0.69, p = 0.40; SNP Social novelty preference:
Q = 1.17, p = 0.28; PPI Prepulse inhibition: Q = 1.36, p = 0.24; FC Fear conditioning: Q = 7.25, p < 0.05; SM Social motivation, SP sucrose
preference. Comparing the effect sizes between males and females: OF- Q = 1.00; p = 0.31, WM- Q = 1.45; p = 0.22, NOR- Q = 0.12; p = 0.73,
NOL- Q = 0.03; p = 0.87, SNP- social novelty preference- Q = 3.33; p = 0.07, PPI- Q = 0.75; p = 0.39, SM- Q = 0.26, p = 0.61; SP- Q = 4.15; p = 0.04.
*p < 0.05, **p < 0.005, ***p < 0.0005.

CBD in schizophrenia result should be interpreted with caution, as the uneven data
In this meta-analysis, we identified 9 publications that examined distribution and low statistical power of our subgroup analysis
the effects of CBD on schizophrenia-like behaviour in wildtype could explain the lack of sex-mediated effect. We also noticed that
animals. We found that chronic adolescent treatment of CBD was although sexually dependent effects were frequently observed
associated with a moderate but significant effect on improving among individual studies, the direction and magnitude often
fear memory recall in the fear conditioning task. However, results varied greatly, sometimes contradictory. Therefore, an average
from other tests were non-significant and some likely effect of these studies might not be informative. Herein, we echo
underpowered. the calls for more inclusion of female animals in preclinical
While the effects of CBD alone might be less prominent, some research [82, 83], as this is essential for improving its clinical
preclinical studies show that CBD reduced hyperlocomotion translation.
induced by psychotomimetic agents like amphetamine and
ketamine [73], as well as in glutamatergic dysfunction models of Limitations and challenges
schizophrenia [74]. Social and sensorimotor deficits were also This meta-analysis provides a rigorous and comprehensive
shown to be alleviated by chronic CBD treatment in MK801- assessment of the effects of adolescent cannabinoid exposure
induced schizophrenia models [74] and in stress-induced models on rodent behaviour. However, we must consider the limitations
[75, 76]. Along the same lines, it has been proposed that CBD may of our study. Firstly, the great disparity in the experimental
mitigate the psychotic effects of CB1R agonists by acting as a settings of the animal studies, such as the age of exposure and the
negative allosteric modulator of CB1R activity [77, 78], and/or treatment protocols, could profoundly influence the animals’
modifying downstream signalling [79]. Here, we identified a need response to drugs, which in turn increases heterogeneity and
for more research to elucidate the impact of CBD and THC co- reduces the validity and generalizability of our findings. To
administration during adolescence, as existing evidence is limited circumvent this, we had initially planned additional subgroup
and inconsistent. analyses to explore the impact of dose and age of onset. However,
such an effort was restricted due to the uneven distribution of
Sex-mediated effect of cannabinoids in rodents data across subgroups, and no conclusive association could be
Preclinical and clinical studies have consistently reported sex- observed. For reference data, see Supplementary Figs. S1 and S2.
dimorphic effects of cannabinoids [80, 81]. However, subgroup Second, many studies did not report essential information on
analyses from the current meta-analysis did not reveal sex as a experimental design and outcome data adequately. This hindered
significant mediator in 8 of 9 behavioural tests included. This the risk of bias assessment and sensitivity analysis, and the

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
293
outcome data had to be frequently extracted using a digital ruler REFERENCES
software from graphical representations. Third, we could not 1. Long T, Wagner M, Demske D, Leipe C, Tarasov PE. Cannabis in Eurasia: origin of
include three behavioural tests for schizophrenia-like behaviour in human use and Bronze Age trans-continental connections. Veg Hist Archaeobot.
the quantitative synthesis, because they either had insufficient 2017;26:245–58.
data reported, or they lacked standardised protocol and outcome 2. UNODC World Drug Report 2022 - World | ReliefWeb. https://reliefweb.int/report/
measurement. Fourth, some of our subgroups included data from world/unodc-world-drug-report-2022. Accessed June, 2024.
3. Martins SS, Segura LE, Levy NS, Mauro PM, Mauro CM, Philbin MM, et al. Racial
multiple experiments within the same study rather than from
and Ethnic Differences in Cannabis Use Following Legalization in US States With
different studies. This could potentially impact the accuracy of the Medical Cannabis Laws. JAMA Netw Open. 2021;4:e2127002.
overall effect estimates due to uncontrolled study-specific factors. 4. Mennis J, McKeon TP, Stahler GJ. Recreational cannabis legalization alters asso-
Finally, the publication bias observed in our meta-analyses may ciations among cannabis use, perception of risk, and cannabis use disorder
indicate selective reporting of significant data, which could treatment for adolescents and young adults. Addictive Behav. 2023;138:107552.
potentially result in inflated or biased summary effects. 5. Ganesh S, D’Souza DC. Cannabis and Psychosis: Recent Epidemiological Findings
Continuing the “Causality Debate”. Am J Psychiatry. 2022;179:8–10. https://
doi.org/10.1176/appi.ajp.2021.21111126.
Future directions 6. Hasan A, von Keller R, Friemel CM, Hall W, Schneider M, Koethe D, et al. Cannabis
use and psychosis: a review of reviews. Eur Arch Psychiatry Clin Neurosci.
In conducting this meta-analysis, we found that the definition of
2020;270:403–12.
adolescence in rodents varied greatly, with a window ranging 7. Di Forti M, Quattrone D, Freeman TP, Tripoli G, Gayer-Anderson C, Quigley H, et al.
from P23 to P45. Adolescence marks a critical period for The contribution of cannabis use to variation in the incidence of psychotic dis-
neurodevelopmental changes, when the brain structure and order across Europe (EU-GEI): a multicentre case-control study. Lancet Psychiatry.
function change immensely, and are highly susceptible to the 2019;6:427–36.
effects of drugs [40, 84, 85]. Therefore, we highlight that 8. Arseneault L. Cannabis use in adolescence and risk for adult psychosis: long-
standardising the adolescent period would improve compar- itudinal prospective study. BMJ. 2002;325:1212–3.
ability and interpretability. Another key challenge in studying the 9. Gillespie NA, Kendler KS. Use of Genetically Informed Methods to Clarify the
effects of cannabis on schizophrenia-like behaviour is to devise a Nature of the Association Between Cannabis Use and Risk for Schizophrenia.
JAMA Psychiatry. 2021;78:467–8.
drug delivery method that accurately reflects human cannabis
10. Beauchamp A, Yee Y, Darwin BC, Raznahan A, Mars RB, Lerch JP. Whole-brain
use. Most of the current studies rely on the intraperitoneal route, comparison of rodent and human brains using spatial transcriptomics. Elife.
which is convenient but may not capture the complexity and 2022;11:e79418.
variability of human cannabis consumption patterns. We are 11. Marín O. Developmental timing and critical windows for the treatment of psy-
encouraged by recent studies utilising more translational relevant chiatric disorders. Nat Med. 2016;22:1229–38.
models like inhaled cannabis vapour [86], but such studies remain 12. Levine A, Clemenza K, Rynn M, Lieberman J. Evidence for the Risks and Con-
uncommon. Moving forward, characterising cross-species phar- sequences of Adolescent Cannabis Exposure. J Am Acad Child Adolesc Psychiatry.
macodynamic/pharmacokinetic correlations for cannabinoids 2017;56:214–25.
would enhance model validity and reliability. Investigating 13. Rubino T, Parolaro D. The Impact of Exposure to Cannabinoids in Adolescence:
Insights From Animal Models. Biol Psychiatry. 2016;79:578–85.
dose-response relationships and drug interactions would also
14. Stringfield SJ, Torregrossa MM. Disentangling the lasting effects of adolescent
inform model optimisation. Such efforts to improve translational cannabinoid exposure. Prog Neuropsychopharmacol Biol Psychiatry.
relevance will accelerate insights into mechanisms linking human 2021;104:110067.
cannabis exposure and schizophrenia phenotypes, informing 15. Arnold JC, Boucher AA, Karl T. The Yin and Yang of Cannabis-induced Psychosis:
prevention and treatment. the Actions of 9-Tetrahyd-rocannabinol and Cannabidiol in Rodent Models of
While our meta-analysis indicates that chronic adolescent canna- Schizophrenia. Curr Pharm Des. 2012;18:5113–30.
binoid exposure in rodents leads to behavioural impairments relevant 16. Dunn AL, Michie PT, Hodgson DM, Harms L. Adolescent cannabinoid exposure
to schizophrenia, it is important to recognise that these impairments interacts with other risk factors in schizophrenia: A review of the evidence from
are not exclusive to schizophrenia-spectrum disorders. Cognitive and animal models. Neurosci Biobehav Rev. 2020;116:202–20.
17. Arseneault L, Cannon M, Witton J, Murray RM. Causal association between can-
social deficits, for example, can also be present in other conditions
nabis and psychosis: examination of the evidence. Br J Psychiatry.
such as depressive disorders [87] and autism-spectrum disorders [88]. 2004;184:110–7.
Therefore, the behavioural tests used in this study, while indicative of 18. Hall W. The Mental Health Risks of Adolescent Cannabis Use. PLoS Med.
schizophrenia-like symptoms, may also reflect broader psychopatho- 2006;3:e39.
logical processes. Our careful phrasing of these symptoms as 19. Chadwick B, Miller ML, Hurd YL. Cannabis Use during Adolescent Development:
‘schizophrenia-related’ or ‘schizophrenia-like’ underscores this Susceptibility to Psychiatric Illness. Front Psychiatry. 2013;4:129.
broader implication but also reflects the aims of the studies included 20. Blest-Hopley G, Colizzi M, Giampietro V, Bhattacharyya S. Is the Adolescent Brain
in the meta-analysis. Future research should aim to develop at Greater Vulnerability to the Effects of Cannabis? A Narrative Review of the
behavioural tasks that capture more specific effects of cannabinoid Evidence. Front Psychiatry. 2020;11:859.
21. Scheyer AF, Laviolette SR, Pelissier A-L, Manzoni OJJ. Cannabis in Adolescence:
exposure across different psychopathological domains.
Lasting Cognitive Alterations and Underlying Mechanisms. Cannabis Cannabi-
In conclusion, despite variation in experimental protocols and noid Res. 2023;8:12–23.
paradigms, the results of this meta-analysis confirm that chronic 22. Aslam S, Emmanuel P. Formulating a researchable question: A critical step for
exposure to CB1R agonists during adolescence is associated with facilitating good clinical research. Indian J Sex Transm Dis AIDS. 2010;31:47.
the expression of several schizophrenia-like behaviours in rodents. 23. Powell CM, Miyakawa T. Schizophrenia-Relevant Behavioral Testing in Rodent
This supports findings from human epidemiological studies. Models: A Uniquely Human Disorder? Biol Psychiatry. 2006;59:1198–207.
Moving forward, standardisation of protocols, consideration of 24. Miyamoto Y, Nitta A. Behavioral Phenotypes for Negative Symptoms in Animal
developmental periods and sex differences, and inclusion of Models of Schizophrenia. J Pharm Sci. 2014;126:310–20.
diverse cannabinoid combinations will facilitate cross-species 25. Rajagopal L, Massey B, Huang M, Oyamada Y, Meltzer H. The Novel Object
Recognition Test in Rodents in Relation to Cognitive Impairment in Schizo-
translation to elucidate the mechanisms linking adolescent
phrenia. Curr Pharm Des. 2014;20:5104–14.
cannabis use and schizophrenia phenotypes. 26. Gill KM, Miller SA, Grace AA. Impaired contextual fear-conditioning in MAM
rodent model of schizophrenia. Schizophr Res. 2018;195:343–52.
27. Hooijmans CR, Rovers MM, de Vries RB, Leenaars M, Ritskes-Hoitinga M, Lan-
gendam MW. SYRCLE’s risk of bias tool for animal studies. BMC Med Res Meth-
DATA AVAILABILITY odol. 2014;14:43.
The data supporting the findings of this study are available within this article and 28. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat
the Supplementary Data Sheet. Med. 2002;21:1539–58.

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
294
29. Chapter 10: Analysing data and undertaking meta-analyses. Cochrane Training. 55. Kesner AJ, Mateo Y, Abrahao KP, Ramos-Maciel S, Pava MJ, Gracias AL, et al.
https://training.cochrane.org/handbook/current/chapter-10. Accessed June, Changes in striatal dopamine release, sleep, and behavior during spontaneous Δ-
2023. 9-tetrahydrocannabinol abstinence in male and female mice. Neuropsycho-
30. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a pharmacology. 2022;47:1537–49.
simple, graphical test. BMJ. 1997;315:629–34. 56. Behan ÁT, Hryniewiecka M, O’Tuathaigh CMP, Kinsella A, Cannon M, Karayiorgou
31. Kuhn BN, Kalivas PW, Bobadilla A-C. Understanding Addiction Using Animal M, et al. Chronic Adolescent Exposure to Delta-9-Tetrahydrocannabinol in COMT
Models. Front Behav Neurosci. 2019;13:262. Mutant Mice: Impact on Indices of Dopaminergic, Endocannabinoid and
32. Walsh KB, Andersen HK. Molecular Pharmacology of Synthetic Cannabinoids: GABAergic Pathways. Neuropsychopharmacology. 2012;37:1773–83.
Delineating CB1 Receptor-Mediated Cell Signaling. Int J Mol Sci. 2020;21:6115. 57. Volkow ND, Fowler JS, Wang G-J, Swanson JM, Telang F. Dopamine in Drug Abuse
33. Soliman N, Haroutounian S, Hohmann AG, Krane E, Liao J, Macleod M, et al. and Addiction. Arch Neurol. 2007;64:1575.
Systematic review and meta-analysis of cannabinoids, cannabis-based medicines, 58. Koob GF, Volkow ND. Neurobiology of addiction: a neurocircuitry analysis. Lancet
and endocannabinoid system modulators tested for antinociceptive effects in Psychiatry. 2016;3:760–73.
animal models of injury-related or pathological persistent pain. Pain. 59. van de Giessen E, Weinstein JJ, Cassidy CM, Haney M, Dong Z, Ghazzaoui R, et al.
2021;162:S26–S44. Deficits in striatal dopamine release in cannabis dependence. Mol Psychiatry.
34. Suraev AS, Marshall NS, Vandrey R, McCartney D, Benson MJ, McGregor IS, et al. 2017;22:68–75.
Cannabinoid therapies in the management of sleep disorders: A systematic 60. Bloomfield MAP, Morgan CJA, Egerton A, Kapur S, Curran HV, Howes OD.
review of preclinical and clinical studies. Sleep Med Rev. 2020;53:101339. Dopaminergic Function in Cannabis Users and Its Relationship to Cannabis-
35. Reuveni N, Carlson CA, Schwartz S, Meter D, Barrett TS, Freeman SM. The anti- Induced Psychotic Symptoms. Biol Psychiatry. 2014;75:470–8.
depressant and anxiolytic effects of cannabinoids in chronic unpredictable stress: 61. Volkow ND, Wang G-J, Telang F, Fowler JS, Alexoff D, Logan J, et al. Decreased
a preclinical systematic review and meta-analysis. Transl Psychiatry. 2022;12:217. dopamine brain reactivity in marijuana abusers is associated with negative
36. Iglesias LP, Bedeschi L, Aguiar DC, Asth L, Moreira FA. Effects of Δ 9 -THC and Type-1 emotionality and addiction severity. Proc Natl Acad Sci. 2014;111:E3149–56.
Cannabinoid Receptor Agonists in the Elevated Plus Maze Test of Anxiety: A Sys- 62. Mizrahi R, Kenk M, Suridjan I, Boileau I, George TP, McKenzie K, et al. Stress-
tematic Review and Meta-Analysis. Cannabis Cannabinoid Res. 2023;8:24–33. Induced Dopamine Response in Subjects at Clinical High Risk for Schizophrenia
37. Ahmed M, Best LM, Pereira CF, Boileau I, Kloiber S. Effects of endocannabinoid with and without Concurrent Cannabis Use. Neuropsychopharmacology.
system modulation on social behaviour: A systematic review of animal studies. 2014;39:1479–89.
Neurosci Biobehav Rev. 2022;138:104680. 63. Thompson JL, Urban N, Slifstein M, Xu X, Kegeles LS, Girgis RR, et al. Striatal
38. Cohen K, Weinstein A. The Effects of Cannabinoids on Executive Functions: Evi- dopamine release in schizophrenia comorbid with substance dependence. Mol
dence from Cannabis and Synthetic Cannabinoids—A Systematic Review. Brain Psychiatry. 2013;18:909–15.
Sci. 2018;8:40. 64. Gomes FV, Guimaraes FS, Grace AA. Effects of Pubertal Cannabinoid Adminis-
39. Renard J, Rushlow WJ, Laviolette SR. What Can Rats Tell Us about Adolescent tration on Attentional Set-Shifting and Dopaminergic Hyper-Responsivity in a
Cannabis Exposure? Insights from Preclinical Research. Can J Psychiatry. Developmental Disruption Model of Schizophrenia. Int J Neuropsychopharmacol.
2016;61:328–34. 2015;18:1–10.
40. Rubino T, Parolaro D. Cannabis abuse in adolescence and the risk of psychosis: A 65. Aguilar DD, Giuffrida A, Lodge DJ. Adolescent Synthetic Cannabinoid Exposure
brief review of the preclinical evidence. Prog Neuropsychopharmacol Biol Psy- Produces Enduring Changes in Dopamine Neuron Activity in a Rodent Model of
chiatry. 2014;52:41–4. Schizophrenia Susceptibility. Int J Neuropsychopharmacol. 2018;21:393–403.
41. Jones C, Watson D, Fone K. Animal models of schizophrenia. Br J Pharm. 66. Friedman AL, Meurice C, Jutkiewicz EM. Effects of adolescent Δ9-
2011;164:1162–94. tetrahydrocannabinol exposure on the behavioral effects of cocaine in adult
42. van den Buuse M. Modeling the Positive Symptoms of Schizophrenia in Sprague-Dawley rats. Exp Clin Psychopharmacol. 2019;27:326–37.
Genetically Modified Mice: Pharmacology and Methodology Aspects. Schizophr 67. Stollenwerk TM, Hillard CJ. Adolescent thc treatment does not potentiate the
Bull. 2010;36:246–70. behavioral effects in adulthood of maternal immune activation. Cells.
43. Ma L, Day-Cooney J, Benavides OJ, Muniak MA, Qin M, Ding JB, et al. Locomotion 2021;10:3503.
activates PKA through dopamine and adenosine in striatal neurons. Nature. 68. McDannald M, Schoenbaum G. Toward a Model of Impaired Reality Testing in
2022;611:762–8. Rats. Schizophr Bull. 2009;35:664–7.
44. Howe MW, Dombeck DA. Rapid signalling in distinct dopaminergic axons during 69. Busquets-Garcia A, Soria-Gómez E, Redon B, Mackenbach Y, Vallée M, Chaouloff F,
locomotion and reward. Nature. 2016;535:505–10. et al. Pregnenolone blocks cannabinoid-induced acute psychotic-like states in
45. Dodson PD, Dreyer JK, Jennings KA, Syed ECJ, Wade-Martins R, Cragg SJ, et al. mice. Mol Psychiatry. 2017;22:1594–603.
Representation of spontaneous movement by dopaminergic neurons is cell-type 70. Carandini M, Churchland AK. Probing perceptual decisions in rodents. Nat Neu-
selective and disrupted in parkinsonism. Proc Natl Acad Sci. 2016;113:E2180–8. rosci. 2013;16:824–31.
46. Freed CR, Yamamoto BK. Regional Brain Dopamine Metabolism: A Marker for the 71. Robbins TW. Animal Models of Hallucinations Observed Through the Modern
Speed, Direction, and Posture of Moving Animals. Science. 1985;229:62–65. Lens. Schizophr Bull. 2017;43:24–26.
47. Abi-Dargham A, van de Giessen E, Slifstein M, Kegeles LS, Laruelle M. Baseline and 72. Schmack K, Bosc M, Ott T, Sturgill JF, Kepecs A. Striatal dopamine mediates
Amphetamine-Stimulated Dopamine Activity Are Related in Drug-Naïve Schizo- hallucination-like perception in mice. Science. 2021;372:eabf4740.
phrenic Subjects. Biol Psychiatry. 2009;65:1091–3. 73. Moreira FA, Guimarães FS. Cannabidiol inhibits the hyperlocomotion induced by
48. Kegeles LS, Abi-Dargham A, Frankle WG, Gil R, Cooper TB, Slifstein M, et al. psychotomimetic drugs in mice. Eur J Pharm. 2005;512:199–205.
Increased Synaptic Dopamine Function in Associative Regions of the Striatum in 74. Gururajan A, Taylor DA, Malone DT. Effect of cannabidiol in a MK-801-rodent
Schizophrenia. Arch Gen Psychiatry. 2010;67:231. model of aspects of Schizophrenia. Behavioural Brain Res. 2011;222:299–308.
49. Howes OD, Williams M, Ibrahim K, Leung G, Egerton A, McGuire PK, et al. Mid- 75. Stark T, Ruda-Kucerova J, Iannotti FA, D’Addario C, Di Marco R, Pekarik V, et al.
brain dopamine function in schizophrenia and depression: a post-mortem and Peripubertal cannabidiol treatment rescues behavioral and neurochemical
positron emission tomographic imaging study. Brain. 2013;136:3242–51. abnormalities in the MAM model of schizophrenia. Neuropharmacology.
50. Bossong MG, van Berckel BN, Boellaard R, Zuurman L, Schuit RC, Windhorst AD, 2019;146:212–21.
et al. Δ9-Tetrahydrocannabinol Induces Dopamine Release in the Human Stria- 76. Peres FF, Diana MC, Levin R, Suiama MA, Almeida V, Vendramini AM, et al.
tum. Neuropsychopharmacology. 2009;34:759–66. Cannabidiol Administered During Peri-Adolescence Prevents Behavioral
51. Renard J, Szkudlarek HJ, Kramar CP, Jobson CEL, Moura K, Rushlow WJ, et al. Abnormalities in an Animal Model of Schizophrenia. Front Pharm. 2018;9:901.
Adolescent THC Exposure Causes Enduring Prefrontal Cortical Disruption of 77. Chung H, Fierro A, Pessoa-Mahana CD. Cannabidiol binding and negative allos-
GABAergic Inhibition and Dysregulation of Sub-Cortical Dopamine Function. Sci teric modulation at the cannabinoid type 1 receptor in the presence of delta-9-
Rep. 2017;7:11420. tetrahydrocannabinol: An In Silico study. PLoS One. 2019;14:e0220025.
52. Loureiro M, Kramar C, Renard J, Rosen LG, Laviolette SR. Cannabinoid Transmis- 78. Jakowiecki J, Abel R, Orzeł U, Pasznik P, Preissner R, Filipek S. Allosteric Mod-
sion in the Hippocampus Activates Nucleus Accumbens Neurons and Modulates ulation of the CB1 Cannabinoid Receptor by Cannabidiol—A Molecular Modeling
Reward and Aversion-Related Emotional Salience. Biol Psychiatry. 2016;80:216–25. Study of the N-Terminal Domain and the Allosteric-Orthosteric Coupling. Mole-
53. Howes OD, Shatalina E. Integrating the Neurodevelopmental and Dopamine cules. 2021;26:2456.
Hypotheses of Schizophrenia and the Role of Cortical Excitation-Inhibition Bal- 79. Szkudlarek HJ, Rodríguez-Ruiz M, Hudson R, De Felice M, Jung T, Rushlow WJ,
ance. Biol Psychiatry. 2022;92:501–13. et al. THC and CBD produce divergent effects on perception and panic beha-
54. McCutcheon RA, Krystal JH, Howes OD. Dopamine and glutamate in schizo- viours via distinct cortical molecular pathways. Prog Neuropsychopharmacol Biol
phrenia: biology, symptoms and treatment. World Psychiatry. 2020;19:15–33. Psychiatry. 2021;104:110029.

Molecular Psychiatry (2025) 30:285 – 295

 Z. Li et al.
295
80. Fattore L, Fratta W. How important are sex differences in cannabinoid action? Br J by the Biomedical Research Centre of the South London and Maudsley NHS Foundation
Pharm. 2010;160:544–8. Trust.
81. Rubino T, Parolaro D. Sex-Dependent Vulnerability to Cannabis Abuse in Ado-
lescence. Front Psychiatry. 2015;6:56.
82. Klein SL, Schiebinger L, Stefanick ML, Cahill L, Danska J, de Vries GJ, et al. Sex COMPETING INTERESTS
inclusion in basic research drives discovery. Proc Natl Acad Sci. 2015;112:5257–8. The authors declare no competing interests.
83. Beery AK, Zucker I. Sex bias in neuroscience and biomedical research. Neurosci
Biobehav Rev. 2011;35:565–72.
84. Hurd YL, Manzoni OJ, Pletnikov MV, Lee FS, Bhattacharyya S, Melis M. Cannabis
ADDITIONAL INFORMATION
and the Developing Brain: Insights into Its Long-Lasting Effects. J Neurosci.
2019;39:8250–8. Supplementary information The online version contains supplementary material
85. Ruiz CM, Torrens A, Castillo E, Perrone CR, Cevallos J, Inshishian VC, et al. Phar- available at https://doi.org/10.1038/s41380-024-02668-5.
macokinetic, behavioral, and brain activity effects of Δ9-tetrahydrocannabinol in
adolescent male and female rats. Neuropsychopharmacology. 2021;46:959–69. Correspondence and requests for materials should be addressed to Marta Di Forti.
86. Coleman JR, Madularu D, Ortiz RJ, Athanassiou M, Knudsen A, Alkislar I, et al.
Changes in brain structure and function following chronic exposure to inhaled Reprints and permission information is available at http://www.nature.com/
vaporised cannabis during periadolescence in female and male mice: A multi- reprints
modal MRI study. Addiction Biol. 2022;27:e13169.
87. Wang Q, Timberlake MA, Prall K, Dwivedi Y. The Recent Progress in Animal Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims
Models of Depression. Prog Neuropsychopharmacol Biol Psychiatry. 2017. https:// in published maps and institutional affiliations.
doi.org/10.1016/j.pnpbp.
88. Ey E, Leblond CS, Bourgeron T. Behavioral Profiles of Mouse Models for Autism
Spectrum Disorders. Autism Res. 2011. https://doi.org/10.1002/aur.175.

Open Access This article is licensed under a Creative Commons

Attribution 4.0 International License, which permits use, sharing,
AUTHOR CONTRIBUTIONS adaptation, distribution and reproduction in any medium or format, as long as you give
MDF conceived and supervised the study. ZL took the lead in data extraction and appropriate credit to the original author(s) and the source, provide a link to the Creative
analysis with support from BD, DM and IA. ZL and DM equally contributed to the Commons licence, and indicate if changes were made. The images or other third party
writing of the manuscript. GT, ES, DQ and RMM contributed to the interpretation of material in this article are included in the article’s Creative Commons licence, unless
the results. All authors provided critical feedback and contributed to the final version indicated otherwise in a credit line to the material. If material is not included in the
of the manuscript. article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this licence, visit http://
creativecommons.org/licenses/by/4.0/.
FUNDING
This study was supported by a Senior Clinical Fellowship (grant number MR/T007818/1)
from the Medical Research Council to Dr Di Forti. Professor Robin Murray was supported © The Author(s) 2024

Molecular Psychiatry (2025) 30:285 – 295