Comparative analysis of 105 datasets across species and tissues reveals higher variability in transcriptomic responses to THC than CBD

Comparative analysis of 105 datasets across species and tissues reveals higher variability in transcriptomic responses to THC than CBD | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Comparative analysis of 105 datasets across species and tissues reveals higher variability in transcriptomic responses to THC than CBD Ruoshui Liu, Thomas Kowal, Caden Chow, Tyler Olson, Emily Nguyen, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6857614/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 16 Dec, 2025 Read the published version in Journal of Cannabis Research → Version 1 posted 9 You are reading this latest preprint version Abstract Background: Cannabis use is on the rise yet the systematic molecular impact of key cannabinoid components on various tissues in diverse organisms remains incompletely understood. We aim to systematically elucidate the molecular pathways and networks affected by delta-9-tetrahydrocannabinol (THC) and cannabidiol (CBD) across species and tissue types. Methods: We curated 105 THC- and CBD-related RNA sequencing (RNAseq) and microarray datasets from Gene Expression Omnibus (NCBI GEO) with a focus on mammalian species (human, non-human primate rhesus macaque, mouse, rat). Differentially expressed genes (DEGs) were identified using limma for microarrays and DESeq2 for RNAseq data. DEGs were analyzed for pathway enrichment using EnrichR, network regulation using Mergeomics key driver analysis, and disease associations using Mergeomics Marker Set Enrichment Analysis. Comparative analyses were conducted across compounds, datasets, species, and tissues. Results: CBD transcriptomic signatures demonstrated greater stability and consistency across species and experimental conditions compared to THC. CBD datasets clustered more tightly by route of administration and species and were more frequently enriched for pathways related to zinc homeostasis, inflammation suppression, and cell cycle regulation. In contrast, THC signatures were more heterogeneous and did not exhibit consistent clustering, although a small number of consistently altered genes associated with antioxidant activity, neuronal myelination, and synaptic signaling were identified across datasets. THC altered endocannabinoid signaling genes more often in brain tissues while CBD affected this pathway more heavily in both central and peripheral tissues. Disease enrichment analyses revealed significant associations of CBD DEGs with lipid metabolism and body composition traits, while DEGs of both compounds showed links to neuropsychiatric disorders and type 2 diabetes. Conclusions: THC and CBD demonstrated distinct and largely non-overlapping transcriptomic responses, with CBD showing more coherent molecular effects across biological systems. Our results underscore the potential therapeutic relevance of CBD to metabolic and psychiatric regulation, highlight the variability of THC’s molecular actions, and offer molecular insights into the therapeutic and side effects of cannabinoids. delta-9-tetrahydrocannabinol THC cannabidiol CBD endocannabinoids system transcriptomics cannabinoids mammalian multispecies analysis cross-tissue analysis Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 INTRODUCTION Cannabis use for recreational and medicinal purposes has increased substantially over the past several years in the United States, likely due to broadening legalization of cannabis. By December 2018, medical cannabis was legalized in 33 states and D.C., while recreational use was permitted in 10 states and D.C. ( 1 ). In 2022, 30.7% of US high school seniors reported cannabis use in the past year, with 6.3% using it daily ( 2 ). Among adults, cannabis use increased from 7.59–15.11% in 2013–2022 ( 3 ). Delta-9-tetrahydrocannabinol (THC) and cannabidiol (CBD) are the two most prominent cannabinoids in Cannabis sativa , comprising up to 40% of the plant’s extract ( 4 , 5 ). These compounds interact with the endocannabinoid system, a complex network of receptors and signaling molecules that regulate pain, mood, inflammation, and immune responses ( 6 ). THC, the primary psychoactive component, exhibits therapeutic effects as an analgesic for cancer-related chronic pain and has demonstrated anti-invasive and anti-metastatic properties in cancer treatment ( 7 ). CBD is non-psychoactive but still influences the brain and nervous system ( 8 , 9 ), is recognized for its potential in managing epilepsy, anxiety, and neurodevelopmental disorders, and is valued for its anti-inflammatory and antioxidant properties ( 10 ). While cannabinoids have demonstrated therapeutic potential, concerns regarding their safety in terms of physical health, mental health, public safety, and the side effects have tempered their widespread acceptance ( 10 , 11 ). For instance, while some studies suggest that chronic daily use of up to 1500 mg/day is well tolerated in humans, others report both physical and mental side effects such as cognitive impairment, cardiovascular complications, and respiratory issues ( 11 ). Additionally, cannabinoids have been linked to immune suppression, resulting in increased susceptibility to human immunodeficiency virus − 1 (HIV-1) infections and disease progression. Mental health effects are particularly concerning, as cannabinoids may exacerbate bipolar disorder in predisposed individuals and increase the risk of temporary psychosis. In addition, cannabinoids also pose broader public health and safety concerns. Driving under the influence is associated with impaired motor function and a higher incidence of motor vehicle accidents. Frequent and heavy cannabis use during adolescence increases the risk of developing cannabis use disorder (CUD) ( 12 ), while addiction and cannabis dependence may contribute to lower income, unemployment, and reduced life satisfaction ( 11 ). Therefore, understanding the precise molecular and biological mechanisms underlying the beneficial and adverse effects of THC and CBD is crucial for guiding safe use and developing strategies to mitigate health concerns. At the transcriptomic level, studies have shown that cannabinoid-induced gene expression changes vary depending on tissue type, sex, age, and genetic backgrounds, such as gene mutations ( 12 – 15 ). This variability has made it challenging to establish precision targets that differentiate the therapeutic vs adverse effects. This study systematically analyzes over 100 transcriptomic datasets related to THC and CBD across species and tissue types from the Gene Expression Omnibus (GEO) ( 16 ) (Fig. 1 ). Using transcriptomic data, we identified differentially expressed genes (DEGs) for each dataset. We further investigated global regulatory patterns across datasets through clustering and correlation assessment as well as cannabinoid-specific effects through pathway and disease association analysis, and network modeling. We also investigated the effects of THC and CBD on the endocannabinoid system across the datasets. These integrative approaches allowed us to uncover consistent and unique patterns of gene regulation across studies to partition potential targets underlying beneficial vs. adverse effects of THC and CBD. MATERIALS & METHODS Data curation Transcriptomic data from previous studies were curated from the National Center for Biotechnology Information’s Gene Expression Omnibus (NCBI GEO) ( 16 , 17 ). To identify THC-specific datasets, we queried GEO using the keywords “THC” and “Delta-9-tetrahydrocannabinol,” while CBD-specific datasets were retrieved using “CBD” and “Cannabidiol.” Studies were included if they met the following criteria: 1) the model organism was human, mouse, rat, or rhesus macaque to focus on mammalian species with higher translational potential; 2) the dataset was generated using cDNA microarrays or RNA sequencing; 3) a minimum sample size threshold of n = 3 per group for in vivo studies and n = 2 per group for in vitro studies was applied, allowing us to include most available studies while balancing data quality and comprehensiveness; 4) the study was not part of a subseries or superseries on GEO to avoid duplicate data. For studies testing multiple conditions, samples were grouped based on their matching physiological and pathological background conditions to ensure accurate differential expression analysis. Specifically, only control and treatment samples with shared conditions, such as prior chemical exposures, genetic mutations, and preexisting diseases, were analyzed together. This approach allowed for the assessment of treatment effects within comparable biological contexts while minimizing confounding factors. A total of 47 datasets for THC and 58 datasets for CBD met the criteria and were curated (Table S1 ). To validate that our data curation and analysis procedures can identify THC and CBD-specific transcriptomic effects, we also included additional well-studied, non-cannabinoids chemicals, including bisphenol A (BPA), perfluorooctanoic acid (PFOA), and estradiol, for comparison. Our previous studies of BPA and PFOA have revealed stable and consistent gene signatures for PFOA but more variable gene expression changes in response to BPA across studies, making them potential reference points to assess gene signature stability vs variability ( 18 – 20 ). We also included estradiol, an endogenous bioactive molecule with well-known biology, as a control to assess whether our analytical pipeline retrieves known biology. Gene expression data for these chemicals were obtained from GEO and processed using the same methods as described for THC and CBD. A total of 50 BPA datasets, 39 estradiol datasets, and 14 PFOA datasets were curated (Table S2 ). Data download and preprocessing Microarray data were downloaded from GEO using the GEOquery package ( 21 ). As microarray data submitted to GEO are pre-processed and quality-controlled, we verified normalization and applied log2 transformation before downstream analysis. Raw RNA sequencing (RNA-seq) datasets were retrieved from NCBI’s Sequence Read Archive (SRA), quality-checked, and processed ( 22 ). FASTQ files were downloaded using the parallel-fastq-dump wrapper, followed by quality control and preprocessing. Trim Galore (v0.6.6) was used to remove low-quality bases from the 3’ end of reads ( 23 ). Cutadapt (v2.1.0) ( 24 ) was employed to remove adapter sequences. Reads shorter than 20 base pairs after trimming and adapter removal were filtered out to ensure data quality. Reads were mapped to appropriate species-specific reference genomes using Salmon (v 1.9.0) ( 25 ). The reference genome assemblies used were GRCh38.108 for Homo sapiens , GRCm39.108 for Mus musculus , mRat.BN7.2.108 for Rattus norvegicus , and Mmul_10.108 for Macaca mulatta . Reference genomes for each species were indexed using the Salmon index tool to optimize alignment efficiency. Finally, transcript-level quantification results from Salmon were imported and summarized using the tximport package (v1.32.0) ( 26 ). Differentially expressed gene (DEG) analysis For each dataset, the treatment group (THC or CBD) was compared to its corresponding control group to identify DEGs using Linear Models for Microarray Data ( LIMMA ) for cDNA microarray datasets ( 27 ) and DESeq2 for RNA-seq datasets ( 28 ). Multiple testing was corrected using the Benjamini-Hochberg method to obtain false discovery rate (FDR). For studies with multiple doses, baseline conditions (e.g. prior chemical exposures, genetic mutations, and preexisting diseases), or time points, each variation was treated as an independent dataset to extract distinct DEG signatures for each dose, condition, and timepoint. DEGs were considered significant at an FDR < 5%. All gene labels were converted to their human orthologs to facilitate cross-species comparisons. Clustering and correlation analysis of DEG gene signatures across datasets To compare DEGs across studies, we combined the gene expression log fold changes from individual studies from the differential gene expression analysis. Since RNAseq and microarray experiments can produce log fold changes on different scales, we applied a rank-based normalization method to standardize values within a range of -1 (downregulated in the treatment group) to 1 (upregulated in the treatment group) for each dataset. Specifically, we separately ranked positive and negative log fold changes within each dataset. Positive values were ranked in ascending order and scaled between 0 and 1 (no change to most upregulated), while negative values were ranked in descending order and scaled between − 1 and 0 (most downregulated to no change). This approach preserves the relative magnitude of gene expression changes among genes while making the datasets comparable across studies and transcriptome platforms. To assess similarity and differences across datasets for THC and CBD through cluster analysis, we used the top 2,500 most variable genes. For the analysis across all chemicals (THC, CBD, PFOA, BPA, estradiol), given the large number of datasets, we first selected genes present in at least 70% of the 208 datasets before selecting the top 2,500 most variable genes. Missing values in the DEG fold change table were imputed using the missForest R package ( 29 ). We then applied UMAP for dimensionality reduction using Euclidean distance to visualize the clustering patterns across chemicals and datasets. As an alternative approach to assess similarity and differences across datasets, we used Spearman’s rank correlation analysis of the normalized gene expression fold change data to compute pairwise correlations among datasets. The pairwise correlation coefficients were further used to group datasets into clusters with similar gene regulation patterns. To identify consistent DEGs across datasets within each cluster in response to THC or CBD, we performed a meta-analysis using the Robust Rank Aggregation (RRA) package (v1.2.1) ( 30 ). Rank aggregation was conducted separately for up-regulated and down-regulated DEGs in each dataset cluster. Genes with an RRA score < 0.05 were considered robust, consistent DEGs, reflecting consistent differential expression across datasets within each cluster. The RRA score measures the probability of a gene achieving its observed ranking pattern across datasets by chance, with lower scores indicating greater consistency. Pathway enrichment analysis Pathway enrichment analysis was performed on the identified DEGs using EnrichR ( 31 ). DEGs were compared against the Gene Ontology Biological Process (GOBP) ( 32 ) databases to identify significantly enriched pathways. Pathways with an FDR < 5% and at least 5 overlapping DEGs were considered statistically significant. Weighted Key Driver Analysis (wKDA) for brain network modeling of THC and CBD DEGs To identify potential gene regulatory networks and network key drivers (KDs) underlying cannabis-induced brain effects, we applied Weighted Key Driver Analysis from the Mergeomics pipeline ( 33 , 34 ). We conducted the analysis separately from THC- and CBD-treated brain datasets. Using a previously constructed brain Bayesian network based on large human and animal model omics studies in Mergeomics, we identified nodes (genes) whose immediate subnetworks were enriched for the THC or CBD DEG sets at FDR < 0.05 as statistically significant KDs. Network visualizations were performed using Cytoscape ( 35 ). Marker Set Enrichment Analysis (MSEA) for disease/trait association assessment MSEA in the Mergeomics package was used to identify the enrichment of THC or CBD DEGs for genetic associations with 101 genome-wide association studies (GWAS) of diseases and phenotypic traits ( 33 , 34 ). Disease-associated genes were mapped using full summary statistics from the GWAS Catalog ( 36 ), with single nucleotide polymorphisms (SNPs) assigned to genes within a 50 kb distance. MSEA employs a chi-square-like statistic with multiple quantile thresholds to assess whether a DEG set shows enrichment of disease SNPs compared to random chance. 10,000 permuted gene sets were generated for each DEG set. As detailed in Shu et al. ( 34 ), the enrichment statistics from the permutations were used to approximate a Gaussian distribution from which enrichment p-values were determined. FDR was estimated using the Benjamini-Hochberg (BH) correction. DEG sets were determined to be statistically significant for a given disease or trait if FDR < 5%. RESULTS Curation of transcriptomic datasets on cannabinoids across species and tissues To investigate the impact of cannabinoids on gene expression, we obtained 108 transcriptomic datasets from GEO ( 16 , 21 ) (Table S1 ). Specifically, we obtained 3 datasets for overall cannabis use, 58 for CBD, and 47 for THC, spanning 32 human ( Homo sapiens ) studies, 8 non-human primate ( Macaca mulatta ) studies, 50 mouse ( Mus musculus ) studies, and 18 rat ( Rattus norvegicus ) studies across 15 broad tissue categories (e.g. blood, brain, cancer, digestive, heart, immune, kidney, liver, lung, muscle, oral, placenta, skin, stem cell, and vasculature). Since the “overall cannabis use” datasets involved either cannabis smoke exposure or observational studies in which other botanical compounds from the cannabis plant were not controlled for, we excluded them from our analysis to maintain our specific focus on the compounds THC and CBD. Notably, reflecting the focus on neurological effects of cannabinoids, brain-related datasets accounted for nearly half of all studies. Many studies included various physiological and pathological background conditions in both control and treatment groups, such as genetic mutations (e.g., Cox15 , Ndufs4 ), preexisting diseases (e.g., SARS-CoV-2, Simian Immunodeficiency Virus (SIV) infection), and prior chemical exposures (e.g., 2'-3'-cGAMP, formoterol/budesonide treatment). Additionally, the datasets covered variations in study design, including differences in sex distribution, dose, route and duration of administration. The broad coverage of diverse conditions allows us to not only determine specific transcriptomic signatures for each condition but also assess the consistency and generalizability of cannabinoid-specific effects across datasets. After downloading and preprocessing the gene expression data, we conducted differential gene expression analysis. We then performed pathway enrichment and disease association analyses to characterize the broader transcriptomic effects of cannabinoids (Fig. 1 ). THC and CBD DEGs are distinct from other chemicals To evaluate whether cannabinoids THC and CBD show distinct transcriptomic signatures compared to other chemicals, we incorporated 103 datasets from three well-characterized chemicals, Bisphenol A (BPA), perfluorooctanoic acid (PFOA), and estradiol, which have been associated with metabolic and reproductive functions, respectively. Both BPA and PFOA are endocrine-disrupting chemicals that influence cardiometabolic disease-related pathways ( 34 – 36 ). Estradiol, a primary estrogen hormone, plays a role in reproductive and sexual function by regulating gene expression through estrogen receptors ( 37 , 38 ). We selected these as comparative datasets because we expect them to be functionally distinct from THC and CBD, which can be reflected by different downstream genes and pathways. In total, we included 50 BPA datasets, 39 estradiol datasets, and 14 PFOA datasets from GEO (Table S2 ) and processed them using the same methodology as cannabinoids. We first compared the chemical-induced expression changes for all genes across all THC, CBD, BPA, estradiol, and PFOA datasets based on the similarities in the normalized log fold changes in gene expression changes between treatment and control groups using clustering analysis and visualized the datasets with UMAP. Agreeing with our hypothesis, most BPA, estradiol, and PFOA datasets each formed distinct clusters, supporting biological coherence within each chemical and differences between chemicals (Fig. 2 A). These results suggest that comparative transcriptomics analysis captures gene expression patterns that likely reflect the different molecular mechanisms of different classes of chemicals. Agreeing with our previous observation on higher stability of PFOA gene signatures ( 18 ), most PFOA datasets clustered tightly in UMAP. In comparison, while BPA and estradiol datasets also formed distinct clusters in UMAP, the distance between datasets for BPA or estradiol was larger than that for PFOA. Compared to these reference chemicals, THC and CBD datasets were more spread out in UMAP. These results suggest that the molecular mechanisms of the reference chemicals are more coherent and consistent across datasets and species, with PFOA showing the highest stability, whereas THC and CBD datasets showed fewer stable gene signatures in comparison. Additionally, we did not observe a distinct separation between datasets from RNA sequencing and microarray methods, indicating a minimal effect of the technical platform on gene signatures (Figure S1 A). We also observed that many tightly clustered datasets for the reference chemicals originated from the same study. For example, several BPA and estradiol microarray datasets were derived from GSE50705, where datasets shared the same experimental setup but differed in dosage. These datasets are clustered in the upper left corner of the UMAP plot (Fig. 2 A). To mitigate the contribution of study-specific batch effects to the tight clustering of the chemicals, we averaged the log fold changes of each gene across datasets for each chemical from the same study, then performed dimensionality reduction again (Figure S1 B). The clustering patterns remained consistent with the original analysis, confirming that study-related effects did not drive the observed trends in Fig. 2 A. Specifically, when accounting for study effects, we found a stability ranking among the chemicals: PFOA exhibited the highest stability, followed by estradiol and BPA, with CBD and THC being highlighted as having the greatest variability. CBD gene signatures demonstrate tighter clustering and higher stability across species and routes of administration compared to THC To better assess the similarities and differences between THC and CBD datasets, we further clustered THC and CBD datasets excluding BPA, estradiol, and PFOA datasets. There was a subtle separation between THC and CBD datasets along the first UMAP dimension (UMAP1), indicating that their global transcriptional effects exhibit certain differences although not strongly distinguishable (Fig. 2 B). Notably, datasets from both RNAseq and microarray platforms were mixed, again supporting that the technological platform did not induce major differences in transcriptomic signatures (Figure S1 C, S1D, S1E). Within the THC datasets, we found no distinct clustering patterns based on species (Fig. 2 C), route of administration (Figure S1 F), or tissue type (Figure S1 G), suggesting that these factors did not significantly influence global transcriptional profiles. In contrast, CBD datasets exhibited weak species-specific clustering. Along the second UMAP dimension (UMAP2), Homo sapiens and Mus musculus datasets showed subtle shifts, whereas Rattus norvegicus datasets clustered more tightly (Fig. 2 D). However, these trends were not strong enough to indicate robust species-driven transcriptional signatures. Regarding the route of administration, in vitro (in medium) datasets tended to cluster at higher UMAP1 values, whereas in vivo (injection or ingestion) datasets were mainly positioned at lower UMAP1 values (Figure S1 H). In the tissue-based UMAP (Figure S1 I), no clear clustering trends emerged across tissue types. However, datasets from the same study, such as the two heart datasets and two kidney datasets, clustered more closely, suggesting that study-specific effects also contributed to their proximity. Overall, the weak clustering patterns across species, tissues, and routes of administration suggest that these factors did not play a dominant role in shaping the global transcriptomic shifts induced by THC or CBD. However, CBD datasets displayed slightly greater clustering and stability compared to THC, indicating more consistent transcriptional responses across different conditions. CBD studies demonstrate different correlation patterns than THC datasets In addition to analyzing dataset clustering based on global gene expression changes through dimensional reduction using UMAP, we assessed correlation patterns in gene expression changes across all THC, CBD, BPA, estradiol, and PFOA datasets. A positive correlation (red) indicates that most genes are upregulated or downregulated concordantly between two datasets, whereas a negative correlation (blue) suggests discordance or opposite directions in gene regulation patterns. Results from this analysis are consistent with our UMAP clustering results in that the molecular patterns of THC and CBD are less consistent compared to the reference chemicals (Figure S2 ). Next, we carried out clustering analyses separately for THC and CBD datasets based on the correlation patterns. In the THC correlation heatmaps, two major clusters of positively correlated datasets were observed: THC Cluster 1 in the upper left quadrant (11 datasets) and THC Cluster 2 in the lower right quadrant (6 datasets), both largely composed of brain-related datasets spanning multiple species. However, these two clusters exhibited moderate negative correlations with each other (Fig. 3 ). This result suggests that THC gene signatures from the brain could be partitioned into distinct patterns with each pattern showing consistency across studies. Notably, some datasets with the same experimental designs but differing in sex were assigned to different clusters. For instance, in GSE273695, which analyzed rat blood samples with daily intraperitoneal injections escalating from 2.5 mg/kg to 10 mg/kg, the male dataset clustered in THC Cluster 1, while the female dataset appeared in THC Cluster 2. Additionally, multiple mouse brain datasets from GSE189821 (daily 10 mg/kg intraperitoneal injection) were distributed across both clusters: specifically, the male amygdala dataset was placed in THC Cluster 2, while the female amygdala dataset fell into THC Cluster 1. However, this separation of male and female samples into different clusters was not observed in all datasets. For example, rat prefrontal cortex datasets of both sexes in GSE273695 were all assigned to THC Cluster 1. While this finding suggests that sex may influence THC-induced transcriptomic responses, this sex-based separation was not consistently observed across all THC datasets and hence any potential sex-specific effects are likely more nuanced and depending on tissue type or other conditions to be elucidated. CBD datasets formed four distinct clusters, with the two largest being CBD Cluster 1 in the upper left quadrant (17 datasets) and CBD Cluster 4 in the lower right quadrant (15 datasets) (Fig. 4 ). These clusters were primarily composed of in vitro experiments and included datasets from three species: human, mouse, and rat. CBD Cluster 1 featured a mixture of tissues, with prominent representation from immune, cancer, and brain. In contrast, approximately half of the datasets in CBD Cluster 4 were from the skin. Two smaller clusters were observed in the central region of the plot, representing study-specific clusters: CBD Cluster 2 included 7 mouse hypothalamus cell line datasets from GSE270378, and CBD Cluster 3 contained 5 mouse liver datasets from GSE261716, where datasets varied by CBD dosage and exposure time. CBD induces more transcriptional changes than THC To go beyond global transcriptomic patterns and focus on individual genes exhibiting significant changes in response to THC and CBD, we derived DEGs at FDR < 5%. Overall, CBD induced more significant DEGs than THC (Fig. 5 A, 5 B). Most CBD datasets contain hundreds or thousands of DEGs, whereas THC datasets predominantly have fewer than 100 DEGs (Table S3 ). Kolmogorov-Smirnov (KS) tests of the distributions of DEG counts across datasets confirmed that both upregulated and downregulated gene counts differed significantly between THC and CBD. Other factors, including transcriptome platform, species, tissue, exposure type, duration, dosage, and sex, also influenced DEG counts. RNA-seq datasets generally yield more DEGs than microarray datasets (Fig. 5 C). In vitro exposure studies of CBD exhibited higher numbers of DEGs than THC, while in vivo datasets show similar DEG distributions between CBD and THC (Fig. 5 D). Examining species effects in CBD datasets, human in vitro cell line studies had the highest number of DEGs (Fig. 5 E). For THC datasets, where the most DEG-rich studies are in vivo , mouse studies generate the largest number of significant DEGs, followed by non-human primate and rat studies. At the tissue level, CBD studies show the highest median number of DEGs in lung, muscle, cancer, immune, and kidney tissues (Fig. 5 F). For THC, while no tissue exhibits a notably high median DEG count, the brain had the greatest variation in DEG numbers across studies, likely due to brain region-specific effects (Fig. 5 G). Additionally, as expected, the number of DEGs in CBD studies increases with both exposure time and dosage (Figure S3 A, Figure S3 B, Figure S3 C for different studies). For THC, no study specifically examined both time points and dosage variations. Unlike time and dosage, sex had a notable but inconsistent effect on DEG counts. Among 10 THC datasets with identical experimental designs differing only by sex, females exhibited more differentially expressed genes (DEGs) in five datasets, including the mouse brain amygdala, dorsolateral striatum, ventral tegmental area (VTA), prefrontal cortex, and rat peripheral blood mononuclear cells (PBMCs) under LPS stimulation. In contrast, males showed higher DEG counts in four datasets: the rat brain orbitofrontal cortex (under both LPS and saline treatments), rat PBMCs with saline, and the mouse brain nucleus accumbens (Figure S3 D). This variability highlights the complexity of sex-specific factors in cannabinoid responses. CBD demonstrates higher consistency at gene and pathway levels across datasets To evaluate the consistency of differentially expressed genes (DEGs) across datasets, we applied two complementary approaches: 1. identifying consistent DEGs within correlation-based clusters, and 2. assessing individual DEG recurrence and directionality consistency. Cluster-Level Consistency: We used rank aggregation across datasets and identified consistently upregulated and downregulated DEGs within each correlation cluster (Fig. 3 , 4 ) ( 30 ). THC Cluster 1 contained relatively few DEGs, with 29 consistently upregulated genes, 18 consistently downregulated genes, and no significantly enriched pathways (Fig. 3 , Table S4 A). Many consistent DEGs are related to neurodevelopment and brain functions. For example, upregulated genes include TBR1 , a key regulator of cortical development ( 37 ); NPTX2 , neuronal pentraxin for synaptic plasticity ( 38 ); neuropeptide S receptor NPSR1 in neuroendocrine cells ( 39 ); ADRA2B , which regulates neurotransmitter norepinephrine ( 40 ); ADORA2A , linked to anxiety, arousal, and sleep regulation ( 41 ). Downregulated genes include FEZF1 for neuronal differentiation ( 42 ); LHX5 associated with mammillary body development ( 43 ); GPR50 for neural progenitor cell differentiation ( 44 ); CACNA1B , a calcium channel subunit for neurotransmitter release ( 45 ). THC Cluster 2 showed even lower consistency, with only two consistently downregulated genes, B Cell receptor CD72 ( 46 ) and presynaptic receptor GRM8 ( 47 ), and no significant pathways (Fig. 3 , Table S4 B). In contrast, CBD Cluster 1 exhibited robust transcriptional changes, with 1,005 consistently upregulated DEGs and 997 consistently downregulated DEGs (Fig. 4 , Table S5 A). The upregulated pathways included negative regulation of growth, cellular response to zinc ions, response to ER stress, and inflammatory response. In contrast, the downregulated pathways were primarily related to development, ECM organization, and sprouting angiogenesis, and skin development. CBD Cluster 4 had 251 consistently upregulated DEGs and 283 downregulated DEGs (Fig. 4 , Table S5 B), but only the upregulated DEGs showed significantly enriched pathways, including negative regulation of growth and cellular response to zinc ions. The absence of strong downregulated pathways in Cluster 4 may explain the separation between the two large clusters and their weak negative correlation. CBD Clusters 2 and 3 displayed more limited consistency. CBD Cluster 2 only had 2 consistent DEGs: cellular zinc sensor MTF1 upregulated ( 48 ) and SENP3 in Wnt signaling downregulated ( 49 ) (Fig. 4 , Table S5 C). CBD Cluster 3 had 73 consistently upregulated DEGs, 77 downregulated DEGs, and two enriched upregulated pathways: cellular response to lipopolysaccharide and inflammatory response (Fig. 4 , Table S5 D). Consistency in DEGs across Datasets: When evaluating consistency in DEGs across datasets, again we found CBD induced broader and more reproducible gene expression changes compared to THC (Fig. 6 A, Table S6 A, Table S6 B). Among the 30 most frequently detected DEGs, 22 exhibited (> 70%) a consistent direction of regulation, meaning that they were detected as either consistently upregulated or downregulated across datasets (Fig. 6 A, Table S6 A). For example, SLC30A1 , a zinc transporter ( 50 ), was significantly upregulated in 23 CBD datasets and not downregulated in any. HMOX1 (heme oxygenase 1), a gene involved in oxidative stress response ( 51 ), was significantly upregulated in 19 datasets while downregulated in only 4 CBD datasets. Some other frequent DEGs play a role in ion metabolism, such as downregulation of SLC39A10 for zinc transport ( 52 ) and upregulation of MT2A and MT1E for metal detoxification and protection against oxidative stress ( 53 ). In contrast, DEGs associated with THC were less frequently replicated across datasets, and the most recurrent genes appeared only three times across 46 datasets (Fig. 6 A, Table S6 B). PRXL2A (peroxiredoxin-like 2A), an antioxidant protein that protects cells from oxidative stress ( 54 ), was downregulated in three datasets. Other THC-associated upregulated genes included BPGM in red blood cell metabolism, COL1A2 for collagen synthesis, microtubule-binding protein FRY , a potassium channel gene KCNQ5 , circadian regulator PER1 , RALGAPA2 for intracellular signaling, and XRCC2 for DNA repair. We also performed pathway enrichment analysis of the significant DEGs for each dataset and summarized the frequency of significant pathways across datasets. Again, THC studies exhibited minimal pathway-level consistency (Fig. 6 B, 6 C, Table S7 B), but CBD datasets showed recurring enriched pathways (Table S7 A). Consistent CBD pathways include the upregulation of the “metabolic pathway” across 17 studies (Fig. 6 B) and the “cellular response to zinc ion” pathway across nine studies (consistent with the reproducibility of DEGs SLC30A1 and MT2A ), and the downregulation of cell cycle, mRNA export, cancer, and mismatch repair pathways (Fig. 6 C). Gene and pathway-level evidence for the endocannabinoid system in CBD and THC transcriptomic signatures Given the well-established interaction between cannabinoids and the endocannabinoid system (ECS) ( 55 , 56 ), we anticipated enrichment of ECS-associated genes in the transcriptional signatures of both CBD and THC. To systematically evaluate this, we curated a list of 135 ECS-associated genes using prior literature and established databases, including Gene Ontology Biological Process (GOBP) annotations ( 32 ) and the DisGeNET database ( 57 , 58 ) (Table S9 ). This gene set includes metabolic enzymes involved in endocannabinoid degradation (i.e. FAAH , MGLL , ABHD6 , ABHD12 , PTGS2 ) and biosynthesis (i.e. NAPEPLD , DAGLA , DAGLB ) of key ligands, such as 2-arachidonoylglycerol (2-AG) and anandamide (AEA). It also encompasses cannabinoid receptors (i.e. CRN1 , CRN2 , GPR55 , TRPV1 ), neurotransmitter receptors (i.e. DRD2 , DRD4 , HTR1A , SSTR4 ), or nicotinic acetylcholine receptors (i.e. CHRNA2 , CHRNA3 , CHRNA5 , CHRNA6 , CHRNA7 , CHRNAB3 ), endocannabinoid transporters (i.e. FABP1 , FABP3 , FABP5 , FABP7 ), and transcriptional regulators (i.e. PPARA , EP300 , FOXP2 ). We examined the regulation of ECS genes identified as DEGs and found that more CBD datasets had significantly regulated ECS genes compared to THC (Fig. 8 A, Table S10 ). In addition, THC datasets generally exhibited fewer ECS DEGs than CBD datasets. Notably, most THC datasets with significant ECS gene regulation were brain-derived, whereas CBD datasets predominantly originated from peripheral tissues. A cluster on the left side of Fig. 8 A, composed mainly of peripheral CBD datasets, exhibited some consistency in regulatory direction. For instance, consistently upregulated genes included SQSTM1 , SLC6A9 , and BEST1 , while consistently downregulated genes included CALM2 , KIAA2013 , and SMC2 . This limited degree of coherence suggests weak gene-level evidence for consistent ECS involvement in the transcriptomic effects of CBD and THC, although context-dependent, tissue-specific regulation of individual genes remains possible. To evaluate whether ECS pathway genes were overrepresented (enriched) among differentially expressed genes (DEGs), we performed Fisher’s Exact Tests comparing the DEG lists to the ECS gene set. Significance was defined as p < 0.05, with p < 0.10 considered suggestive (Fig. 8 B, 8 C, Table S10 ). Overall, CBD showed more datasets with significant or suggestive enrichment compared to THC. Notably, all THC datasets showing significant and suggestive ECS gene set enrichment originated from brain regions, including the rat orbitofrontal cortex, mouse amygdala, mouse dorsolateral striatum, and mouse medial prefrontal cortex (Table S10 ). In contrast, approximately half of the significantly or suggestively enriched CBD datasets came from peripheral tissues and cell lines, including the mouse liver, human liver cell line, monocyte cell line, and microglial cell line. These findings support that THC’s regulatory effects are brain-specific, whereas CBD exhibits more pleiotropic effects across central and peripheral systems. In addition, we investigated pathways related to pain and inflammation, which are regulated by the endocannabinoid system ( 6 ). However, no significant enrichment was found for the GOBP “sensory perception of pain pathway” and “response to pain pathway” ( 32 ). However, the inflammatory response pathway was enriched among DEGs from two CBD datasets, both derived from monocyte-derived dendritic cell lines. THC- and CBD-induced brain network modeling identifies unique key drivers We next focused on the brain datasets on THC and CBD to identify key similarities and differences in the regulation of cannabis-induced transcriptomic alterations in the brain. We applied weighted key driver analysis (wKDA) to DEG sets from THC- and CBD- treated brain datasets to identify potential key drivers (KDs) using a Bayesian gene regulatory network previously constructed from dozens of human and mouse brain datasets ( 33 , 34 ). KDs with a false discovery rate (FDR) < 0.05 were considered statistically significant. For CBD-treated brain datasets, 14 significant KDs were identified (top 5 shown in Fig. 7 A; complete list in Table S8 ). The KDs in the CBD network were derived from multiple brain cell lines, including mouse brain microglia, rat primary hippocampal neurons, and human neuroblastoma cells (Fig. 7 A). Most KDs were involved in neural development and synaptic functions, such as SH3GL2 , DPYSL5 , JPH3 , SLC17A7 , and GRIN3B ( MMT00076709 ). Other KDs were associated with signal transduction and kinase activity ( PRKCA , TAOK1 , and ADRBK1 ), membrane transport ( SLC9A3R1 ), metabolism ( PPP1R3B ), gene regulation ( USF1 ), and immune responses and cancer ( SLAMF8 , SRC , and KIAA0100 ). Despite the lack of consistent DEGs and pathways across THC-treated brain datasets, we identified a total of 176 significant KDs based on DEGs from individual datasets (top 5 shown in Fig. 7 B; complete list in Table S8 ), among which SLC17A7 and TAOK1 were shared KDs between THC and CBD. Notably, most THC KDs were found to be significant in only a single DEG dataset rather than across multiple datasets (Fig. 7 B), suggesting that THC may exert brain region-specific effects on gene regulation. Key drivers (KDs) in the THC network were identified from multiple animal models, including the mouse hippocampus, mouse amygdala, mouse microglia, and rat orbitofrontal cortex. A majority of THC-associated KDs were involved in neuronal functions, signal transduction, structural integrity, and metabolic functions. Notably, some KDs with known neuronal roles include SLC17A7 , RTN4R , RTN4RL1 , RTN4RL2 , NEUROD2 , NEUROD6 , NGB , CAMK2A , ADCY1 , SYT1 , DLGAP1 , DLGAP3 , NTS , RGS14 , GAD2 , and CNIH3 . CBD and THC DEGs are associated with neuropsychiatric disorders while CBD is associated with more metabolic traits The endocannabinoid system plays a role in various health outcomes, including mood disorders, cardiovascular disease, stroke, cancer, diabetes, autoimmune conditions, and neurological disorders ( 59 ). To further investigate the connection between cannabinoids and human health, we analyzed publicly available summary statistics from large-scale Genome-Wide Association Studies (GWAS) for more than 100 diseases or phenotypic traits using Marker Set Enrichment Analysis (MSEA) via the Mergenomics package ( 33 , 34 ). This approach enabled us to assess whether signature genes from THC and CBD datasets were enriched for human disease/phenotype variants. DEGs from both cannabinoids showed significant associations with Schizophrenia, type 2 diabetes (T2D), and depressive symptoms (Fig. 9 ). In addition, CBD transcriptomic signatures exhibited broader associations with lipid metabolism, including low-density lipoprotein (LDL), high-density lipoprotein (HDL), total cholesterol (TC), and triglycerides (TG), as well as with body composition traits such as waist circumference (WCadjBMI), hip circumference (HIPadjBMI), waist-to-hip ratio (WHRadjBMI), and height. In contrast, THC was linked only to WCadjBMI. Beyond metabolic and anthropometric traits, CBD was associated with Crohn’s disease (CD) and coronary artery disease (CAD). To identify the genes most strongly linked to disease-associated SNPs, we examined the top genes from the significantly enriched datasets (Figure S4 A, S4B, S4C, S4D). In the CBD datasets, key associations for CD included tumor suppressor CYLD ( 60 ) and cytosolic receptor NOD2 (Figure S4 A). Mutations or altered expression of NOD2 have been observed in CD patients ( 61 ), and genetic studies have identified NOD2 locus polymorphisms and an independent involvement of the neighboring gene CYLD in CD ( 62 ). For T2D, a key association for CBD DEGs was the Wnt signaling pathway transcription factor TCF7L2 (Figure S4 C). Carrying two copies of a common variant in this gene is associated with an approximately twofold increase in T2D risk ( 63 , 64 ). CBD DEGs CDKN2A and CDKN2B were also among the top CAD-associated GWAS candidates, which are located adjacent to the lead CAD-linked SNP on the 9p21 locus ( 65 ) (Figure S4 D). In contrast, THC showed more moderate levels of association between DEGs and disease-related SNPs. Of note, this analysis mainly focuses on overlaps between CBD/THC DEGs with genetic associations of diseases and does not implicate increases or decreases in disease risks. DISCUSSION In this study, we systematically analyzed over 100 publicly available transcriptomic datasets to understand the molecular effects of THC and CBD across species and tissue types. By examining both global and significant gene expression changes from microarray and RNA-sequencing data, we found highly variable gene signatures for these two cannabinoids compared to other exposures such as PFOA, BPA, and estradiol. We also identified DEGs, pathways, network key drivers (KDs), and diseases/phenotypes associated with the significant DEGs. Our findings highlight significant variability in THC and CBD-induced transcriptional responses, with CBD demonstrating better consistency across studies, species, and central and peripheral tissue types compared to THC, which shows more dataset-specific effects in brain regions. Our broad dimensionality reduction and correlation analyses revealed that CBD and THC datasets were more scattered in contrast with other well-studied chemicals PFOA, BPA, and estradiol, which displayed stronger internal consistency across datasets. In addition, neither THC nor CBD datasets exhibited strong clustering in the UMAP or formed larger clusters in the correlation analyses across species and tissues. This finding suggests that cannabinoid-induced gene expression changes are highly variable depending on combinations of experimental conditions, genetic or disease backgrounds, and prior chemical exposures. These results highlight the challenge of defining a universal and predictable transcriptional signature for cannabinoids and align with previous literature indicating that cannabinoids effects depend on tissue, developmental stage, sex, and exposure conditions ( 12 , 14 , 15 , 66 ). Our analyses on the DEGs and enriched pathways across datasets indicate that CBD has a more stable gene regulatory signature than THC. This regulatory stability agrees with CBD’s well-characterized antioxidant, anti-inflammatory, and neuroprotective effects ( 67 , 68 ). In particular, two major CBD-associated gene clusters (CBD Cluster 1 and CBD Cluster 4) exhibited hundreds to thousands of consistent DEGs, with both CBD clusters exhibiting upregulation in two pathways: negative regulation of growth and cellular response to zinc ions. CBD was associated with higher frequency of consistent DEGs across datasets, such as HMOX1 , SLC30A1 , MT2A , and SLC39A10 . Previous in vitro studies have identified HMOX1 as the most upregulated gene and protein following CBD treatment. The proposed mechanism involves CBD-induced nuclear export and proteasomal degradation of transcriptional repressor BACH1( 51 ). Another study suggests that CBD may influence HMOX1 levels through the Nrf2 pathway, which regulates antioxidant defenses ( 69 ). These indicate that HMOX1 is linked to antioxidant and anti-inflammatory properties under the impact of CBD ( 70 ). Notably, pathways related to metabolic regulation and cellular response to zinc ion were recurrently upregulated. Supporting this finding, a previous study in BV-2 microglial cells reported that CBD upregulated zinc-related genes ( Mt2 , Ndrg1 , Mmp23 ) and zinc transporters SLC30A1 and SLC39A4 while downregulating SLC39A10 and Zfp472 ( 71 ). Another study in autoimmune T cells found that CBD suppressed pro-inflammatory genes and enhanced oxidative stress-related genes, including SLC30A1 ( 72 ). We now expand this finding to a much broader range of tissues and experimental conditions. Together, these results suggest that modulation of zinc homeostasis may be a key mechanism through which CBD exerts its antioxidant and anti-inflammatory effects. The cell cycle and cancer pathway were top downregulated pathways across CBD-associated datasets. Prior studies have shown that CBD can arrest the cell cycle in the G0/G1 phase and promote cell death in gastric cancer cell lines, such as SGC-7901 ( 73 ). Our study of around 50 datasets highlighting this pathway across studies further highlight the potential anti-cancer role of CBD in addition to its metabolic and immunomodulatory effects. In contrast, THC datasets exhibited minimal gene-level and pathway-level consistency in dataset clusters and globally, reinforcing its more variable and context-dependent effects. The weaker transcriptional stability of THC may stem from its complex pharmacodynamics, particularly its biphasic effects on neural and immune signaling. Our analysis revealed several diverse central and peripheral nervous system targets of THC that may ultimately lead to these downstream differences in pathway enrichment. For example, among the most consistently enriched DEGs induced by THC was EGR2 , which is associated with neuropathy and congenital hypomyelination of peripheral neurons ( 74 ). A number of THC target DEGs were also associated with synaptic signaling in the brain, and the onset of epilepsy and intellectual deficits, including RPH3A ( 75 ), KCNQ5 ( 76 ), and FRY ( 77 ). Further, we also found that PER1 was among our most consistently detected DEGs, which is a primary circadian pacemaker which has implications on human behavior and cognition ( 78 ). Some of these genes, such as FRY and EGR2 , are associated with neuronal irregularities in region-specific targets (e.g. the cerebellum for FRY and peripheral neurons for EGR2 ). One possible explanation for the highly variable effects of THC lies in its high lipid solubility ( 79 ), which may influence its distribution across datasets depending on adipose content and tissue composition. Additionally, physiological barriers such as the blood-brain and blood-testicular barrier restrict THC accumulation in the brain and testes during acute exposure, and similar protective mechanisms may exist in other tissues as well ( 80 ). These factors together may contribute to the heterogeneous effects of THC across individuals and biological contexts. We also observed heterogeneity in the brain network through KDA. While the CBD network based on DEGs from individual datasets had only 14 KDs, the THC network contained over 100 KDs. However, most of these THC KDs were significant in only one DEG dataset, further supporting the idea that THC induces region-specific effects in the brain. Additionally, KDA revealed that the molecular mechanisms activated in the brain differ between THC and CBD. Although both networks featured KDs involved in neuronal function and signal transduction, only two KDs, SLC17A7 and TAOK1 , were shared between them. SLC17A7 encodes vesicular glutamate transporter 1 ( VGLUT1 ), which facilitates glutamatergic neurotransmission ( 81 ). Cannabinoids have been shown to influence glutamatergic signaling: CBD reduces neuronal activation in VGLUT + neurons ( 82 ), while cannabinol upregulates genes associated with glutamatergic synaptic function ( 83 ). TAOK1 encodes a serine/threonin-protein that functions as a MAP kinase kinase kinase (MAP3K) and regulates MAPK signaling cascade ( 84 ). Although direct evidence linking cannabis treatment to TAOK1 regulation is lacking, THC has been shown to modulate microRNAs that target mRNAs of proteins involved in MAPK signaling, including TAOK1 ( 85 ). The limited overlap in KDs between the two networks highlights the distinct molecular and cellular mechanisms through which THC and CBD exert their effects in the brain. This divergence is consistent with our broader findings that THC and CBD regulate largely non-overlapping gene sets and pathways across different tissue types. While the endocannabinoid system (ECS) is widely recognized as a primary target of phytocannabinoids such as THC and CBD ( 55 ), our multi-dataset analysis revealed that ECS-associated genes were regulated inconsistently across studies, both in direction and magnitude. These inconsistencies might reflect several underlying factors, including the tissue-specific expression of ECS components as well as distinct pharmacological profiles of THC and CBD. For example, in terms of binding affinity, THC acts as a partial agonist at both cannabinoid receptors, while CBD exhibits negligible affinity for cannabinoid receptor 1 (CB1) and functions as a partial agonist at cannabinoid receptor 2 (CB2) ( 86 , 87 ). Despite this gene-level variability, pathway-level enrichment analysis demonstrated that the ECS-related gene set were significantly enriched in several datasets, comparable in frequency to some of the top-ranking Gene Ontology Biological Process (GOBP) and KEGG pathways (Fig. 6 B, 6 C). Importantly, a clearer divergence between THC and CBD emerged when ECS enrichment was stratified by tissue type. ECS enrichment associated with THC was exclusively observed in brain-derived datasets, where CBD-associated ECS enrichment was distributed more evenly across both central and peripheral systems, including immune-related cell types and liver tissue. This divergence likely reflects differences in receptor affinity and expression patterns. CB1 is abundantly expressed in the brain and to a lesser extent in select peripheral tissues, whereas CB2 is predominantly found in immune cells and exhibits limited expression in the central nervous system ( 88 ). Thus, THC’s higher binding affinity to CB1 compared to CB2 explains its brain-focused transcriptomic effects ( 88 ), while CBD’s lack of CB1 binding and moderate activity at CB2 likely underlies its broader effects across peripheral and immune contexts. The observed significant ECS enrichment in CBD-treated immune and liver datasets aligns with this pharmacological profile. To further investigate the potential health implications of cannabis exposure, we performed Marker Set Enrichment Analysis (MSEA) to assess whether gene expression profiles associated with THC and CBD were enriched for genetic markers of human diseases. CBD-responsive genes showed significant enrichment for markers related to cholesterol and lipid metabolism traits, aligning with prior studies indicating that CBD can modulate lipid profiles and holds therapeutic potential for managing lipid disorders ( 89 ). CBD was also significantly associated with anthropometric traits, such as waist circumference, hip circumference, and height. A systematic review across multiple databases and registries reported that while cannabis use is generally associated with reductions in weight, waist circumference, and BMI, CBD specifically has been linked to increased body fat ( 90 ). Both CBD and THC show association with multiple diseases, such as Schizophrenia, type 2 diabetes (T2D), and depressive symptoms. The relationship between cannabis use and depression is particularly complex and potentially bidirectional. Studies relying on self-reported depressive symptoms suggest mixed short-term effects of cannabis use: a minority (20%) with increased depression and a majority (64%) with decreased depression ( 91 ). However, a meta-analysis of 15 studies found that even a single THC administration induces depression and anxiety with large effect sizes ( 92 ). Conversely, extended cannabis abstinence has been associated with significant improvements of depressive symptoms ( 93 ). Longitudinal data also suggest a bidirectional or reinforcing relationship, as baseline depression has been significantly associated with increased THC use in e-cigarettes 12 months later ( 94 ). CBD and other non-psychoactive cannabinoids have been investigated in human clinical trials for their potential therapeutic benefits in T2D and Schizophrenia. Indeed, a randomized clinical trial showed that CBD reduced circulating resistin, a hormone associated with insulin resistance, and increased glucose-dependent insulinotropic peptide (GIP), which plays a role in preserving pancreatic β-cell function in T2D patients ( 95 ). CBD has also demonstrated beneficial effects and a safety profile in Schizophrenia, where patients treated with CBD for six weeks demonstrated lower levels of positive psychotic symptoms ( 96 ). In another randomized clinical trial comparing CBD with the antipsychotic amisulpride, both treatments led to significant clinical improvement; however, CBD had a markedly superior side-effect profile ( 97 ). Thus, our MSEA results, supported by human clinical evidence, highlight the broad therapeutic potential of cannabinoids, especially CBD, in metabolic and psychiatric diseases and underscore the relevance of cannabinoid-responsive genes in the genetic architecture of complex traits. Here, we have conducted a comprehensive, multispecies, multitissue investigation of the metabolic effects of THC and CBD. One of the primary strengths of our study was the use of highly heterogeneous datasets, with variations in experimental design, dosing regimens, and sample characteristics. This allowed us to consider cannabinoid exposure in a number of biological contexts, including in co-occurrence with diseases, on top of exposure to other relevant chemical agents, and in different tissues or regions within those tissues (e.g. several different brain regions were covered by our datasets). Meta-analysis across species allows us to consider the maximal amount of data from translatable model organisms so we can draw human-relevant conclusions. Despite the comprehensive nature of our analysis, however, some important limitations should be acknowledged. First, while we applied normalization techniques to minimize batch effects, residual variability may have influenced our findings. However, the same procedures were applied to the datasets from the reference chemicals, where coherent clustering and coherence across datasets were found. Therefore, our findings are less likely due to technical artifacts but are more likely the results of intrinsic activities of the compounds examined. Second, the sample sizes for cannabinoid studies tend to be small, with most groups consisting of only around three replicates. This limitation reflects the current state of the field and underscores the need for larger, more coordinated genomic studies. To mitigate the impact of limited sample size, we adopted a meta-analytic strategy to integrate signatures across studies, enhancing robustness through aggregated evidence. Third, the use of bulk transcriptomic data precludes cell type-specific resolution, which is critical for understanding the differential effects of cannabinoids on distinct cell populations. Future studies employing single-cell RNA sequencing or spatial transcriptomics, when datasets are available, could provide deeper insights into the cellular specificity of cannabinoid-induced gene expression changes. Additionally, while we focused on gene expression data, post-transcriptional and epigenetic modifications likely play a crucial role in cannabinoid-mediated effects. Integrating multi-omics approaches, including proteomics and metabolomics, could enhance our understanding of cannabinoid biology. Lastly, our study highlights the need for controlled, well-designed longitudinal studies to assess the long-term impact of THC and CBD exposure on human health. In summary, this study provides a comprehensive transcriptomic analysis of THC and CBD across species and tissue types. Our findings demonstrate that CBD exhibits greater transcriptional stability compared to THC, with more consistent effects on metabolic, cell cycle, and inflammatory pathways. In contrast, THC-induced gene expression changes are highly variable, complicating the identification of robust molecular signatures. These insights have important implications for the therapeutic use of cannabinoids and highlight the less predictive nature of the long-term biological effects of THC. As the legalization and medicinal use of cannabis continues to expand, a deeper understanding of its molecular mechanisms will be critical for optimizing its clinical applications while minimizing potential risks. Declarations Funding Declaration The authors received no specific funding for this work. Data Availability Declaration Data are publicly available on the open-access Gene Expression Omnibus (GEO). Competing Interest Declaration The authors declare that they have no competing interests. Author Contribution Declaration R.L., T.K., X.Y., and M.B. designed and directed the project. R.L, T.K., C.C, T.O., and E.N. curated the data. R.L., T.K., and S.Y. conducted the analysis and prepared figures. R.L., T.K., J.L., X.Y., and M.B. wrote the manuscript. All authors read and approved the final manuscript. Acknowledgments Not applicable. Ethics Declaration Not applicable. Consent to Participate Declaration Not applicable. References Smart R, Pacula RL. Early evidence of the impact of cannabis legalization on cannabis use, cannabis use disorder, and the use of other substances: Findings from state policy evaluations. Am J Drug Alcohol Abuse. 2019;45(6):644–63. Miech RA, Johnston LD, Patrick ME, O’Malley PM, Bachman JG, Schulenberg JE. Monitoring the Future National Survey Results on Drug Use, 1975–2022: Secondary School Students. Inst Soc Res. 2023. Mattingly DT, Richardson MK, Hart JL. Prevalence of and trends in current cannabis use among US youth and adults, 2013–2022. Drug Alcohol Depend Rep. 2024;12:100253. Maroon J, Bost J. Review of the neurological benefits of phytocannabinoids. Surg Neurol Int. 2018;9. Pourseyed Lazarjani M, Torres S, Hooker T, Fowlie C, Young O, Seyfoddin A. Methods for quantification of cannabinoids: a narrative review. J Cannabis Res. 2020;2:1–10. Pertwee RG. The pharmacology of cannabinoid receptors and their ligands: an overview. Int J Obes. 2006;30(1):S13–8. Pagano C, Navarra G, Coppola L, Avilia G, Bifulco M, Laezza C. Cannabinoids: therapeutic use in clinical practice. Int J Mol Sci. 2022;23(6):3344. Batalla A, Bos J, Postma A, Bossong MG. The impact of cannabidiol on human brain function: a systematic review. Front Pharmacol. 2021;11:618184. Malabadi RB, Kolkar K, Chalannavar R. ∆9-Tetrahydrocannabinol (THC): The major psychoactive component is of botanical origin. Int J Innov Sci Res Rev. 2023;5(3):4177–84. Machado Bergamaschi M, Helena Costa Queiroz R, Waldo Zuardi A, Crippa AS. Safety and side effects of cannabidiol, a Cannabis sativa constituent. Curr Drug Saf. 2011;6(4):237–49. Sachs J, McGlade E, Yurgelun-Todd D. Safety and toxicology of cannabinoids. Neurotherapeutics. 2015;12(4):735–46. Zuo Y, Iemolo A, Montilla-Perez P, Li HR, Yang X, Telese F. Chronic adolescent exposure to cannabis in mice leads to sex-biased changes in gene expression networks across brain regions. Neuropsychopharmacology. 2022;47(12):2071–80. Puighermanal E, Luna-Sánchez M, Gella A, van der Walt G, Urpi A, Royo M, et al. Cannabidiol ameliorates mitochondrial disease via PPARγ activation in preclinical models. Nat Commun. 2024;15(1):7730. Shapira G, Israel-Elgali I, Grad M, Avnat E, Rachmany L, Sarne Y, et al. Hippocampal differential expression underlying the neuroprotective effect of delta-9-tetrahydrocannabinol microdose on old mice. Front Neurosci. 2023;17:1182932. Bilkei-Gorzo A, Albayram O, Draffehn A, Michel K, Piyanova A, Oppenheimer H, et al. A chronic low dose of ∆ 9-tetrahydrocannabinol (THC) restores cognitive function in old mice. Nat Med. 2017;23(6):782. Edgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30(1):207–10. Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, et al. NCBI GEO: archive for functional genomics data sets—update. Nucleic Acids Res. 2012;41(D1):D991–5. Zamora Z, Wang S, Chen YW, Diamante G, Yang X. Systematic transcriptome-wide meta-analysis across endocrine disrupting chemicals reveals shared and unique liver pathways, gene networks, and disease associations. Environ Int. 2024;183:108339. Diamante G, Cely I, Zamora Z, Ding J, Blencowe M, Lang J, et al. Systems toxicogenomics of prenatal low-dose BPA exposure on liver metabolic pathways, gut microbiota, and metabolic health in mice. Environ Int. 2021;146:106260. Chen Y, Shu L, Qiu Z, Lee DY, Settle SJ, Que Hee S, et al. Exposure to the BPA-substitute bisphenol S causes unique alterations of germline function. PLoS Genet. 2016;12(7):e1006223. Davis S, Meltzer PS. GEOquery: a bridge between the Gene Expression Omnibus (GEO) and BioConductor. Bioinformatics. 2007;23(14):1846–7. Leinonen R, Sugawara H, Shumway M. International Nucleotide Sequence Database Collaboration. The sequence read archive. Nucleic Acids Res. 2010;39(suppl1):D19–21. Krueger F. Trim Galore! A wrapper around Cutadapt and FastQC to consistently apply adapter and quality trimming to FastQ files, with extra functionality for RRBS data. Babraham Inst. 2015. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J. 2011;17(1):10–2. Patro R, Duggal G, Love MI, Irizarry RA, Kingsford C. Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods. 2017;14(4):417–9. Soneson C, Love MI, Robinson MD. Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences. F1000Research. 2016;4:1521. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47–47. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:1–21. Stekhoven DJ, Bühlmann P. MissForest—non-parametric missing value imputation for mixed-type data. Bioinformatics. 2012;28(1):112–8. Kolde R, Laur S, Adler P, Vilo J. Robust rank aggregation for gene list integration and meta-analysis. Bioinformatics. 2012;28(4):573–80. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;14:1–14. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. Nat Genet. 2000;25(1):25–9. Ding J, Blencowe M, Nghiem T, Ha S, min, Chen YW, Li G, et al. Mergeomics 2.0: a web server for multi-omics data integration to elucidate disease networks and predict therapeutics. Nucleic Acids Res. 2021;49(W1):W375–87. Shu L, Zhao Y, Kurt Z, Byars SG, Tukiainen T, Kettunen J, et al. Mergeomics: multidimensional data integration to identify pathogenic perturbations to biological systems. BMC Genomics. 2016;17:1–16. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498–504. MacArthur J, Bowler E, Cerezo M, Gil L, Hall P, Hastings E, et al. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res. 2017;45(D1):D896–901. Hevner RF, Shi L, Justice N, Hsueh YP, Sheng M, Smiga S, et al. Tbr1 regulates differentiation of the preplate and layer 6. Neuron. 2001;29(2):353–66. Zhou J, Wade SD, Graykowski D, Xiao MF, Zhao B, Giannini LA, et al. The neuronal pentraxin Nptx2 regulates complement activity and restrains microglia-mediated synapse loss in neurodegeneration. Sci Transl Med. 2023;15(689):eadf0141. Pulkkinen V, Ezer S, Sundman L, Hagström J, Remes S, Söderhäll C, et al. Neuropeptide S receptor 1 (NPSR1) activates cancer-related pathways and is widely expressed in neuroendocrine tumors. Virchows Arch. 2014;465:173–83. Xie W, Cappiello M, Meng M, Rosenthal R, Zhang W. ADRA2B deletion variant and enhanced cognitive processing of emotional information: A meta-analytical review. Neurosci Biobehav Rev. 2018;92:402–16. Hohoff C, Kroll T, Zhao B, Kerkenberg N, Lang I, Schwarte K, et al. ADORA2A variation and adenosine A1 receptor availability in the human brain with a focus on anxiety-related brain regions: modulation by ADORA1 variation. Transl Psychiatry. 2020;10(1):406. Shimizu T, Nakazawa M, Kani S, Bae YK, Shimizu T, Kageyama R, et al. Zinc finger genes Fezf1 and Fezf2 control neuronal differentiation by repressing Hes5 expression in the forebrain. Development. 2010;137(11):1875–85. Heide M, Zhang Y, Zhou X, Zhao T, Miquelajáuregui A, Varela-Echavarría A, et al. Lhx5 controls mamillary differentiation in the developing hypothalamus of the mouse. Front Neuroanat. 2015;9:113. zahid Khan M, He L. The emerging role of GPR50 receptor in brain. Biomed Pharmacother. 2016;78:121–8. Szymanowicz O, Drużdż A, Słowikowski B, Pawlak S, Potocka E, Goutor U, et al. A Review of the CACNA gene family: Its role in neurological disorders. Diseases. 2024;12(5):90. Wu HJ, Bondada S. Positive and negative roles of CD72 in B cell function. Immunol Res. 2002;25:155–66. Scherer SW, Soder S, Duvoisin RM, Huizenga JJ, Tsui LC. The human metabotropic glutamate receptor 8 (GRM8) gene: a disproportionately large gene located at 7q31. 3–q32. 1. Genomics. 1997;44(2):232–6. Andrews GK. Cellular zinc sensors: MTF-1 regulation of gene expression. Zinc Biochem Physiol Homeost Recent Insights Curr Trends. 2001;37–51. Wang P, Yang L, Guo Y, Qi S, Liang J, Tian G, et al. SENP3 mediates the activation of the Wnt/β-catenin signaling pathway to accelerate the growth and metastasis of oesophagal squamous cell carcinoma in mice. Funct Integr Genomics. 2024;24(2):40. Ryu MS, Lichten LA, Liuzzi JP, Cousins RJ. Zinc transporters ZnT1 (Slc30a1), Zip8 (Slc39a8), and Zip10 (Slc39a10) in mouse red blood cells are differentially regulated during erythroid development and by dietary zinc deficiency. J Nutr. 2008;138(11):2076–83. Casares L, García V, Garrido-Rodríguez M, Millán E, Collado JA, García-Martín A, et al. Cannabidiol induces antioxidant pathways in keratinocytes by targeting BACH1. Redox Biol. 2020;28:101321. Ren X, Feng C, Wang Y, Chen P, Wang S, Wang J, et al. SLC39A10 promotes malignant phenotypes of gastric cancer cells by activating the CK2-mediated MAPK/ERK and PI3K/AKT pathways. Exp Mol Med. 2023;55(8):1757–69. Schulkens IA, Castricum K, Weijers EM, Koolwijk P, Griffioen AW, Thijssen VL. Expression, regulation and function of human metallothioneins in endothelial cells. J Vasc Res. 2014;51(3):231–8. Chen YF, Wei YY, Yang CC, Liu CJ, Yeh LY, Chou CH, et al. miR-125b suppresses oral oncogenicity by targeting the anti-oxidative gene PRXL2A. Redox Biol. 2019;22:101140. Tan Y, Xia H, Song Q. Research mapping of cannabinoids and endocannabinoid system in cancer over the past three decades: insights from bibliometric analysis. Front Pharmacol. 2025;16:1540619. Wu J. Cannabis, cannabinoid receptors, and endocannabinoid system: Yesterday, today, and tomorrow. Acta Pharmacol Sin. 2019;40(3):297–9. Piñero J, Bravo À, Queralt-Rosinach N, Gutiérrez-Sacristán A, Deu-Pons J, Centeno E et al. DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants. Nucleic Acids Res. 2016;gkw943. Piñero J, Ramírez-Anguita JM, Saüch-Pitarch J, Ronzano F, Centeno E, Sanz F, et al. The DisGeNET knowledge platform for disease genomics: 2019 update. Nucleic Acids Res. 2020;48(D1):D845–55. Lowe H, Toyang N, Steele B, Bryant J, Ngwa W. The endocannabinoid system: a potential target for the treatment of various diseases. Int J Mol Sci. 2021;22(17):9472. Massoumi R. CYLD: a deubiquitination enzyme with multiple roles in cancer. Future Oncol. 2011;7(2):285–97. Negroni A, Pierdomenico M, Cucchiara S, Stronati L. NOD2 and inflammation: current insights. J Inflamm Res. 2018;49–60. Elding H, Lau W, Swallow DM, Maniatis N. Dissecting the genetics of complex inheritance: linkage disequilibrium mapping provides insight into Crohn disease. Am J Hum Genet. 2011;89(6):798–805. del Bosque-Plata L, Martínez-Martínez E, Espinoza-Camacho MÁ, Gragnoli C. The role of TCF7L2 in type 2 diabetes. Diabetes. 2021;70(6):1220–8. Weedon M. The importance of TCF7L2. Diabet Med. 2007;24(10):1062–6. Zivotić I, Djurić T, Stanković A, Milasinovic D, Stankovic G, Dekleva M, et al. CDKN2B gene expression is affected by 9p21. 3 rs10757278 in CAD patients, six months after the MI. Clin Biochem. 2019;73:70–6. Martin BR. Cellular effects of cannabinoids. Pharmacol Rev. 1986;38(1):45–74. Al-Khazaleh AK, Zhou X, Bhuyan DJ, Münch GW, Al-Dalabeeh EA, Jaye K, et al. The neurotherapeutic arsenal in Cannabis sativa: Insights into anti-neuroinflammatory and neuroprotective activity and potential entourage effects. Molecules. 2024;29(2):410. Atalay S, Jarocka-Karpowicz I, Skrzydlewska E. Antioxidative and anti-inflammatory properties of cannabidiol. Antioxidants. 2019;9(1):21. Ekiner SA, Gęgotek A, Skrzydlewska E. The molecular activity of cannabidiol in the regulation of Nrf2 system interacting with NF-κB pathway under oxidative stress. Redox Biol. 2022;57:102489. Baswan SM, Klosner AE, Glynn K, Rajgopal A, Malik K, Yim S et al. Therapeutic potential of cannabidiol (CBD) for skin health and disorders. Clin Cosmet Investig Dermatol. 2020;927–42. Juknat A, Rimmerman N, Levy R, Vogel Z, Kozela E. Cannabidiol affects the expression of genes involved in zinc homeostasis in BV-2 microglial cells. Neurochem Int. 2012;61(6):923–30. Kozela E, Juknat A, Gao F, Kaushansky N, Coppola G, Vogel Z. Pathways and gene networks mediating the regulatory effects of cannabidiol, a nonpsychoactive cannabinoid, in autoimmune T cells. J Neuroinflammation. 2016;13:1–19. Pagano C, Navarra G, Coppola L, Bifulco M, Laezza C. Molecular mechanism of cannabinoids in cancer progression. Int J Mol Sci. 2021;22(7):3680. Nagarajan R, Svaren J, Le N, Araki T, Watson M, Milbrandt J. EGR2 mutations in inherited neuropathies dominant-negatively inhibit myelin gene expression. Neuron. 2001;30(2):355–68. Pavinato L, Stanic J, Barzasi M, Gurgone A, Chiantia G, Cipriani V, et al. Missense variants in RPH3A cause defects in excitatory synaptic function and are associated with a clinically variable neurodevelopmental disorder. Genet Med. 2023;25(11):100922. Wei AD, Wakenight P, Zwingman TA, Bard AM, Sahai N, Willemsen MH, et al. Human KCNQ5 de novo mutations underlie epilepsy and intellectual disability. J Neurophysiol. 2022;128(1):40–61. Paulraj P, Bosworth M, Longhurst M, Hornbuckle C, Gotway G, Lamb AN, et al. A novel homozygous deletion within the FRY gene associated with nonsyndromic developmental delay. Cytogenet Genome Res. 2019;159(1):19–25. Lim AS, Chang A, Shulman JM, Raj T, Chibnik LB, Cain SW, et al. A common polymorphism near PER1 and the timing of human behavioral rhythms. Ann Neurol. 2012;72(3):324–34. Chayasirisobhon S. Mechanisms of action and pharmacokinetics of cannabis. Perm J. 2020;25:19–200. Huestis MA. Human cannabinoid pharmacokinetics. Chem Biodivers. 2007;4(8):1770. Reimer RJ. SLC17: a functionally diverse family of organic anion transporters. Mol Aspects Med. 2013;34(2–3):350–9. Nedelescu H, Wagner GE, De Ness GL, Carroll A, Kerr TM, Wang J, et al. Cannabidiol produces distinct U-shaped dose-response effects on cocaine-induced conditioned place preference and associated recruitment of prelimbic neurons in male rats. Biol Psychiatry Glob Open Sci. 2022;2(1):70–8. Trainito A, Muscarà C, Gugliandolo A, Chiricosta L, Salamone S, Pollastro F, et al. Cannabinol (CBN) Influences the Ion Channels and Synaptic-Related Genes in NSC-34 Cell Line: A Transcriptomic Study. Cells. 2024;13(18):1573. Fang CY, Lai TC, Hsiao M, Chang YC. The diverse roles of TAO kinases in health and diseases. Int J Mol Sci. 2020;21(20):7463. Simon L, Song K, Vande Stouwe C, Hollenbach A, Amedee A, Mohan M, et al. ∆9-Tetrahydrocannabinol (∆9-THC) promotes neuroimmune-modulatory microRNA profile in striatum of simian immunodeficiency virus (SIV)-infected macaques. J Neuroimmune Pharmacol. 2016;11:192–213. Zou S, Kumar U. Cannabinoid receptors and the endocannabinoid system: signaling and function in the central nervous system. Int J Mol Sci. 2018;19(3):833. Shahbazi F, Grandi V, Banerjee A, Trant JF. Cannabinoids and cannabinoid receptors: The story so far. IScience. 2020;23(7). Schönke M, Martinez-Tellez B, Rensen PC. Role of the endocannabinoid system in the regulation of the skeletal muscle response to exercise. Curr Opin Pharmacol. 2020;52:52–60. Wiciński M, Fajkiel-Madajczyk A, Kurant Z, Gryczka K, Kurant D, Szambelan M, et al. The Use of Cannabidiol in Metabolic Syndrome—An Opportunity to Improve the Patient’s Health or Much Ado about Nothing? J Clin Med. 2023;12(14):4620. Reis MG, Ferreira AJ, Sohouli MH, Taimeirão DR, Vieira RAL, Guimarães NS. Effect of cannabis and subproducts on anthropometric measures: a systematic review and meta-analysis. Int J Obes. 2024;48(1):44–54. Li X, Diviant JP, Stith SS, Brockelman F, Keeling K, Hall B, et al. The effectiveness of cannabis flower for immediate relief from symptoms of depression. Yale J Biol Med. 2020;93(2):251. Hindley G, Beck K, Borgan F, Ginestet CE, McCutcheon R, Kleinloog D, et al. Psychiatric symptoms caused by cannabis constituents: a systematic review and meta-analysis. Lancet Psychiatry. 2020;7(4):344–53. Lucatch AM, Kloiber SM, Meyer JH, Rizvi SJ, George TP. Effects of extended cannabis abstinence in major depressive disorder. Can J Addict. 2020;11(3):33–41. Clendennen SL, Smith J, Sumbe A, Chen B, Wilkinson AV, Harrell MB. Symptoms of depression and anxiety and subsequent use of nicotine and THC in electronic cigarettes. Subst Use Misuse. 2023;58(5):591–600. Jadoon KA, Ratcliffe SH, Barrett DA, Thomas EL, Stott C, Bell JD, et al. Efficacy and Safety of Cannabidiol and Tetrahydrocannabivarin on Glycemic and Lipid Parameters in Patients With Type 2 Diabetes: A Randomized, Double-Blind, Placebo-Controlled, Parallel Group Pilot Study. Diabetes Care. 2016;39(10):1777–86. McGuire P, Robson P, Cubala WJ, Vasile D, Morrison PD, Barron R, et al. Cannabidiol (CBD) as an adjunctive therapy in schizophrenia: a multicenter randomized controlled trial. Am J Psychiatry. 2018;175(3):225–31. Leweke F, Piomelli D, Pahlisch F, Muhl D, Gerth C, Hoyer C, et al. Cannabidiol enhances anandamide signaling and alleviates psychotic symptoms of schizophrenia. Transl Psychiatry. 2012;2(3):e94–94. Additional Declarations No competing interests reported. Supplementary Files THCsupplementaryfigures.pptx SupplementTable1.xlsx SupplementTable2.xlsx SupplementTable3.xlsx SupplementTable4B.xlsx SupplementTable4A.xlsx SupplementTable5A.xlsx SupplementTable5B.xlsx SupplementTable5C.xlsx SupplementTable5D.xlsx SupplementTable6A.xlsx SupplementTable6B.xlsx SupplementTable7A.xlsx SupplementTable7B.xlsx SupplementTable8.xlsx SupplementTable9.xlsx SupplementTable10.xlsx Cite Share Download PDF Status: Published Journal Publication published 16 Dec, 2025 Read the published version in Journal of Cannabis Research → Version 1 posted Editorial decision: Revision requested 15 Sep, 2025 Reviews received at journal 15 Sep, 2025 Reviewers agreed at journal 15 Sep, 2025 Reviews received at journal 05 Sep, 2025 Reviewers agreed at journal 30 Jul, 2025 Reviewers invited by journal 11 Jun, 2025 Editor assigned by journal 11 Jun, 2025 Submission checks completed at journal 11 Jun, 2025 First submitted to journal 09 Jun, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6857614","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":469949658,"identity":"12b8a32a-3834-4193-b6be-12979f3670ff","order_by":0,"name":"Ruoshui Liu","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Ruoshui","middleName":"","lastName":"Liu","suffix":""},{"id":469949663,"identity":"3b553bc5-781d-4311-af5c-49648852e746","order_by":1,"name":"Thomas Kowal","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Kowal","suffix":""},{"id":469949667,"identity":"77d8e196-9f4e-4569-a083-10e9f2332e4b","order_by":2,"name":"Caden Chow","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Caden","middleName":"","lastName":"Chow","suffix":""},{"id":469949669,"identity":"1f25a715-385e-44e4-a0ed-04648315a288","order_by":3,"name":"Tyler Olson","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Tyler","middleName":"","lastName":"Olson","suffix":""},{"id":469949675,"identity":"9b0197ff-fc2d-4db1-9cec-98345b67a143","order_by":4,"name":"Emily Nguyen","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Emily","middleName":"","lastName":"Nguyen","suffix":""},{"id":469949676,"identity":"22f73857-72af-4da6-bc36-55b6e7676719","order_by":5,"name":"Sen Yang","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Sen","middleName":"","lastName":"Yang","suffix":""},{"id":469949677,"identity":"4d9941f8-c103-44fb-a28d-579beee90d3c","order_by":6,"name":"Jimin Lee","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Jimin","middleName":"","lastName":"Lee","suffix":""},{"id":469949678,"identity":"d9f2c3cb-2db3-4aeb-804c-e72d2c4b0979","order_by":7,"name":"Xia Yang","email":"","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":false,"prefix":"","firstName":"Xia","middleName":"","lastName":"Yang","suffix":""},{"id":469949679,"identity":"1f524100-9122-4353-9b17-d69ee92dc9a9","order_by":8,"name":"Montgomery Blencowe","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6UlEQVRIiWNgGAWjYLACxgYLKKuCgYEPRPMQ0sDYIAFln2FgYCNNC2MbEVp023uPP+bdISFnzn74mMTHeYfl2dgPMD5424Zbi9mZc4nNvGckjC170tIkZ247bNjGk8BsOBeflhs5hs28bRKJG27wmN3m3XY4gY0hgU2alwgt9WAtf+cAtfA/YP9NjJYEA5AWxgagFokENma8Ws6cMZw594yE4YYzaek/e46lG7ZJPGyWnHMOj5bjPQYf3u6wkTc4fviwwY8aa3l+/uSDH96U4dYCAkxoscDYgF89SMkPgkpGwSgYBaNgRAMAyZ1R1YLf6hEAAAAASUVORK5CYII=","orcid":"","institution":"University of California, Los Angeles","correspondingAuthor":true,"prefix":"","firstName":"Montgomery","middleName":"","lastName":"Blencowe","suffix":""}],"badges":[],"createdAt":"2025-06-09 22:53:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6857614/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6857614/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s42238-025-00361-0","type":"published","date":"2025-12-16T15:57:40+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":84568776,"identity":"df7f8d1b-7580-4e3f-9bed-a71eb032f109","added_by":"auto","created_at":"2025-06-13 14:53:22","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":332057,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverall study design. \u003c/strong\u003eDatasets from THC and CBD exposure studies were curated from GEO and analyzed according to our established pipelines depending on whether they were cDNA microarray or RNAseq studies to detect differentially expressed genes (DEGs) across 4 mammalian species. Control chemicals, including PFOA, BPA, and estradiol, were processed similarly. Gene expression signatures were clustered using dimensionality reduction analysis and correlated using heatmaps derived from Spearman’s rank correlation coefficients. DEGs were then analyzed for biological pathway enrichment, key regulators in brain gene regulatory networks, and disease enrichment using Mergeomics.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/f6ce8948455a720e3173b804.png"},{"id":84568768,"identity":"1a8372d4-5d8f-4951-9f00-0fefc01b9386","added_by":"auto","created_at":"2025-06-13 14:53:21","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":179705,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eUMAP clustering reveals instability in DEG signatures across tissues in THC and CBD. \u003c/strong\u003eGene expression signatures were clustered using UMAP by the normalized log2 fold change induced in each gene between the treatment and control groups for THC and CBD, as well as the reference endocrine disrupting chemicals PFOA, BPA, and estradiol. Across all these chemicals, the top 2,500 most variable genes by log2 fold change were included in this analysis. Dots in the plot correspond to individual DEG signatures. In (\u003cstrong\u003eA-B)\u003c/strong\u003e, the color of the dot corresponds to the chemical, and the shape of the dot corresponds to the species. \u003cstrong\u003eA) \u003c/strong\u003ePFOA (purple), BPA (green), and estradiol (yellow) all form relatively tight clusters on the left-hand side of the plot. THC (blue) and CBD (pink) signatures are more widely spread, indicating a higher degree of variability in the DEG signatures curated for this study. \u003cstrong\u003eB) \u003c/strong\u003eTHC (blue) and CBD (pink) exhibit subtle separation along the UMAP1 dimension, indicating weak differences between these two chemicals in their DEG signatures. In (\u003cstrong\u003eC-D)\u003c/strong\u003e, the color of the dot corresponds to the species. \u003cstrong\u003eC) \u003c/strong\u003eWithin THC signatures, signatures derived from experiments on all four model organisms clustered together with no sharp distinction among any of them, revealing that none of the species has a species-specific response in their THC-associated DEG signatures. \u003cstrong\u003eD) \u003c/strong\u003eWithin CBD signatures, mouse datasets and human datasets show subtle shifts along the UMAP2 dimension, and rat datasets cluster tightly, revealing that CBD exhibits weak species-specific effects on gene signatures.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/e28c619171a65f5313055604.png"},{"id":84568763,"identity":"eebc74c1-7fdb-4b32-8479-569e7f065ca6","added_by":"auto","created_at":"2025-06-13 14:53:21","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":313505,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHierarchical clustering reveals clusters of datasets with concordant gene expression signatures in response to THC exposure. \u003c/strong\u003eThe log2 fold changes across the genome for each THC dataset across were used to calculate Spearman’s correlation coefficients and perform hierarchical clustering analysis. The results are shown in the heatmap. Red hues indicate concordant gene signatures (e.g. both datasets have their genes changing in the same direction and by a similar magnitude) while blue hues indicate discordant gene signatures. Two distinct clusters of concordant THC datasets are indicated, with cluster 1 containing 11 datasets in the upper left quadrant and cluster 2 containing 6 datasets in the lower right quadrant. Both clusters are composed mostly of brain-related datasets and contain datasets derived from multiple species. These two clusters exhibited moderate negative correlations with each other.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/0c65c643216bc926a15f1fbe.png"},{"id":84568779,"identity":"ef782fc7-2bdd-48a4-a48a-e6dde1f1083e","added_by":"auto","created_at":"2025-06-13 14:53:22","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":429227,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHierarchical clustering reveals clusters of datasets with concordant gene expression signatures in response to CBD exposure. \u003c/strong\u003eThe log2 fold changes across the genome for each CBD dataset across were used to calculate Spearman’s correlation coefficients and perform hierarchical clustering analysis. The results are shown in the heatmap. Red hues indicate concordant gene signatures (e.g. both datasets have their genes changing in the same direction and by a similar magnitude) while blue hues indicate discordant gene signatures. Four distinct clusters of concordant CBD datasets are indicated, with cluster 1 containing 17 datasets in the upper left quadrant, cluster 2 containing 7 datasets in the center-left region, cluster 3 containing 5 datasets in the central right region, and cluster 4 containing 15 datasets in the lower right quadrant. Both cluster 1 and cluster 4 contained datasets from multiple species and tissue types. Cluster 1 is purely \u003cem\u003ein vitro\u003c/em\u003e datasets where cluster 4 has a mixture of multiple exposure routes and a high representation of skin-derived datasets. Clusters 2 and 3 were composed of datasets derived from one murine hypothalamus study and one murine liver study, respectively.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/4a3f75365c9809e165f6b0a0.png"},{"id":84568317,"identity":"8d13d020-f983-4a5f-8159-989e96a93e29","added_by":"auto","created_at":"2025-06-13 14:45:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":281383,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDEG results indicate CBD induces more DEGs across tissues and identify factors affecting the number of observed DEGs across studies. \u003c/strong\u003eThe number of DEGs observed across datasets was quantified according to chemical exposure and the potential confounding variables of sequencing platform, route of exposure, species, and tissue. \u003cstrong\u003eA) \u003c/strong\u003eA bar plot showing the number of DEGs detected in each THC dataset analyzed. Four datasets had hundreds of DEGs and approximately half of the datasets had at least ten, while very low or zero DEG counts were observed in approximately half of the studies. \u003cstrong\u003eB) \u003c/strong\u003eA bar plot showing the number of DEGs detected in each CBD dataset analyzed. Nearly one third of datasets demonstrated more than 1,000 DEGs, and a significant portion of datasets also resulted in hundreds of DEGs. A small number of datasets with ten or fewer DEGs was also observed. \u003cstrong\u003eC) \u003c/strong\u003eA dot plot showing the amount of DEGs observed in each study for CBD (left) and THC (right), colored according to the sequencing platform used to derive each dataset. In general, RNAseq datasets are more sensitive to DEGs, and more are observed in these studies than in cDNA microarray studies of the same chemical exposures. \u003cstrong\u003eD) \u003c/strong\u003eA dot plot showing the amount of DEGs observed in each study for CBD (left) and THC (right), colored according to whether the study design used an \u003cem\u003ein vivo\u003c/em\u003e or \u003cem\u003ein vitro\u003c/em\u003eexposure model. CBD datasets with \u003cem\u003ein vitro\u003c/em\u003e exposure paradigms exhibit more DEGs while THC datasets with \u003cem\u003ein vivo\u003c/em\u003e exposures exhibit more DEGs. \u003cstrong\u003eE) \u003c/strong\u003eA dot plot showing the amount of DEGs observed in each study of CBD (left) and THC (right), colored according to the species used in each study. CBD human and mouse datasets exhibited higher numbers of DEGs than rat datasets, while THC mouse datasets exhibited the most DEGs with slightly fewer in rat and macaque models. \u003cstrong\u003eF-G) \u003c/strong\u003eDot plots showing the number of DEGs for CBD \u003cstrong\u003e(F)\u003c/strong\u003e and THC \u003cstrong\u003e(G)\u003c/strong\u003e datasets, colored according to the tissue examined in each dataset. CBD exhibited high numbers of DEGs in all tissues except the heart and kidney, with consistently high numbers of DEGs observed in the liver, lung, immune cells, and cancer cell line models across tissues. THC consistently exhibited high numbers of DEGs in brain datasets with far fewer DEGs in the rest of the tissues, suggesting brain specific gene perturbation effects.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/be4d49c33849293e13f233d1.png"},{"id":84568294,"identity":"303a3b89-3c26-40cc-acd5-1e0bf7eb3211","added_by":"auto","created_at":"2025-06-13 14:45:21","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":289531,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMeta-analysis reveals highly represented gene and pathway perturbations across cannabinoid exposure signatures. A) \u003c/strong\u003eRecurrent genes that were detected as DEGs across CBD and THC exposure datasets are plotted on this bar graph. CBD tended to induce more DEGs and thus has recurrent genes that appear more often across datasets. THC had more variable gene perturbation effects and thus fewer genes were identified across multiple studies. Numbers of studies in which the genes appear as up- or downregulated DEGs are shown in each bar according to the color legend. \u003cstrong\u003eB-C) \u003c/strong\u003eRecurrent pathways that were detected based on the enrichment analysis of the DEGs for CBD \u003cstrong\u003e(B) \u003c/strong\u003eand THC \u003cstrong\u003e(C)\u003c/strong\u003e. The bars are colored based on whether the pathway was up- or downregulated according to the color legend, and the length of the bars indicates the number of exposure datasets in which the pathway was significantly enriched.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/2c9e7c248ce460282ba922b5.png"},{"id":84568319,"identity":"fcde302c-f0cc-4270-bbc8-960d0c9d137b","added_by":"auto","created_at":"2025-06-13 14:45:22","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":839168,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKey driver analysis identifies key regulatory genes in the perturbation effects of CBD and THC. \u003c/strong\u003eGene regulatory networks were constructed using key driver analysis from Mergeomics for CBD \u003cstrong\u003e(A) \u003c/strong\u003eand THC \u003cstrong\u003e(B)\u003c/strong\u003e brain exposure datasets. The colors of the network nodes indicate datasets in which the gene was a DEG, as described by the color legends. \u003cstrong\u003eA) \u003c/strong\u003eBrain dataset key drivers for CBD were primarily related to neural development and synaptic functions and were consistently detected across datasets from multiple model organisms and brain regions. \u003cstrong\u003eB) \u003c/strong\u003eBrain dataset key drivers for THC were primarily related to neuronal functions, signal transduction, structural integrity, and metabolic functions but were less consistently detected across different datasets, as most were uniquely DEGs in specific studies.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/e5234a56dafd5d5a4586bcb3.png"},{"id":84570182,"identity":"1587d78f-fc40-438b-88d2-db40193cd2f3","added_by":"auto","created_at":"2025-06-13 15:09:21","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":305340,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelation and overlap analysis reveals the extent of ECS gene involvement in CBD- and THC-induced transcriptome perturbations. A) \u003c/strong\u003eThe normalized log2 fold changes of the ECS genes detected as significant DEGs in each study are shown in this heatmap, with more highly upregulated genes shown in red hues and more highly downregulated genes shown in blue hues. Clusters of studies with similar directional changes of each gene indicate that the genes are consistently affected by CBD and THC exposure across datasets, and these results indicate ECS gene involvement in CBD- and THC-induced gene perturbations across studies. \u003cstrong\u003eB-C) \u003c/strong\u003eFisher’s exact test was performed to determine if the ECS gene set is significantly enriched in each CBD and THC dataset. The number of datasets derived from brain and peripheral tissue datasets that showed significant enrichment of the ECS gene set at a significant (p \u0026lt; 0.05, \u003cstrong\u003eB\u003c/strong\u003e) and suggestive (p \u0026lt; 0.10, \u003cstrong\u003eC\u003c/strong\u003e) threshold is shown in each table.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/dd235b83a72380503063d901.png"},{"id":84569311,"identity":"a8cc86a4-94ad-48af-8375-30c7b311669f","added_by":"auto","created_at":"2025-06-13 15:01:21","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":233847,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMSEA reveals associations between cannabinoid exposure and psychiatric, metabolic diseases. \u003c/strong\u003eMarker set enrichment analysis\u003cstrong\u003e \u003c/strong\u003e(MSEA) from Mergeomics was performed on the DEGs from each dataset to identify disease associations based on marker genes for each disease. Only diseases or traits with at least one significantly associated dataset are shown. More highly associated diseases are shown with redder hues, and significant associations are starred. Psychiatric conditions, including schizophrenia and depression, and metabolic conditions, such as type 2 diabetes, were detected as significant associations with both CBD and THC exposure.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/e9f1f9ae42c499f5668f59cd.png"},{"id":98813927,"identity":"3e76e959-1d6c-4d07-b129-8fbec5d769c6","added_by":"auto","created_at":"2025-12-22 16:07:51","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4620460,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/1775e134-8866-4649-a3b0-d56a7ecfaa1a.pdf"},{"id":84569310,"identity":"73ac392b-abe6-47fe-9562-ab1c91dfdda8","added_by":"auto","created_at":"2025-06-13 15:01:21","extension":"pptx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2793557,"visible":true,"origin":"","legend":"","description":"","filename":"THCsupplementaryfigures.pptx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/94524052caca5ef81aa498b3.pptx"},{"id":84568762,"identity":"bee4a24b-08ed-4b81-84e0-aa8f14243cc9","added_by":"auto","created_at":"2025-06-13 14:53:21","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":18739,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/99aedc3313611079c3d46e5a.xlsx"},{"id":84569314,"identity":"d8f7e0cc-f90c-44ca-8680-d1867964dd85","added_by":"auto","created_at":"2025-06-13 15:01:22","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":78602,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/829dc91b95ed71c2c12f4408.xlsx"},{"id":84568771,"identity":"f37f9539-446c-474c-8bd6-45dfb2b2eacf","added_by":"auto","created_at":"2025-06-13 14:53:21","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":11325,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/2347da97116dab3c838eadcf.xlsx"},{"id":84568769,"identity":"852a3d5c-d5e4-4574-a024-e1f5d1f864a9","added_by":"auto","created_at":"2025-06-13 14:53:21","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":10223,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable4B.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/07bc591ce693bab98ea71291.xlsx"},{"id":84568775,"identity":"fbc232c0-d04c-480d-a093-31560e5e6917","added_by":"auto","created_at":"2025-06-13 14:53:22","extension":"xlsx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":12294,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable4A.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/271c7f7279426faddd91d20a.xlsx"},{"id":84568778,"identity":"14738010-0049-46e4-80da-a3921888d5ea","added_by":"auto","created_at":"2025-06-13 14:53:22","extension":"xlsx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":78133,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable5A.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/dfa1ee3ccebac51242ae5a68.xlsx"},{"id":84568296,"identity":"6173dc02-4189-49ee-8433-8ffc61e34078","added_by":"auto","created_at":"2025-06-13 14:45:21","extension":"xlsx","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":27589,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable5B.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/59254ac47b9b16175ce66109.xlsx"},{"id":84568310,"identity":"0f3774e8-e215-46fd-9d38-0fc5f9b4e682","added_by":"auto","created_at":"2025-06-13 14:45:22","extension":"xlsx","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":10985,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable5C.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/e6200688e3889c4776ae9429.xlsx"},{"id":84568307,"identity":"43b96e8c-e8e2-4581-83b9-820969778bfe","added_by":"auto","created_at":"2025-06-13 14:45:21","extension":"xlsx","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":16379,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable5D.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/6447f3a7532fcf159cdf17eb.xlsx"},{"id":84569313,"identity":"ef555647-e441-4202-9561-a666f08b7f8f","added_by":"auto","created_at":"2025-06-13 15:01:22","extension":"xlsx","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":708533,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable6A.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/b9552b72e8f0ee2ba419db78.xlsx"},{"id":84568336,"identity":"b05c66b1-159a-4b36-824c-56c19d5afe13","added_by":"auto","created_at":"2025-06-13 14:45:22","extension":"xlsx","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":99090,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable6B.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/b48f7af648b8dcef6795560e.xlsx"},{"id":84568784,"identity":"91fad38a-f64d-4021-b3b8-1db18f785f2e","added_by":"auto","created_at":"2025-06-13 14:53:22","extension":"xlsx","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":51110,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable7A.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/57c28266df4f5a6e073993df.xlsx"},{"id":84568805,"identity":"9f85c02b-a839-4317-a791-00af69ee90a5","added_by":"auto","created_at":"2025-06-13 14:53:23","extension":"xlsx","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":14361,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable7B.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/87ff79c30969541203cc49c5.xlsx"},{"id":84568782,"identity":"c1caf6fd-2844-4732-8446-fea305dfc32c","added_by":"auto","created_at":"2025-06-13 14:53:22","extension":"xlsx","order_by":15,"title":"","display":"","copyAsset":false,"role":"supplement","size":55724,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable8.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/d4de7e40e7e0ddd195a7ee5a.xlsx"},{"id":84568350,"identity":"91fec676-11df-4776-9537-b6bd62a70078","added_by":"auto","created_at":"2025-06-13 14:45:23","extension":"xlsx","order_by":16,"title":"","display":"","copyAsset":false,"role":"supplement","size":13227,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable9.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/b3f9fd418efb8a7925e781d7.xlsx"},{"id":84568344,"identity":"7faf4db3-bca8-45bd-ae76-1d4eb4099099","added_by":"auto","created_at":"2025-06-13 14:45:22","extension":"xlsx","order_by":17,"title":"","display":"","copyAsset":false,"role":"supplement","size":17121,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementTable10.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6857614/v1/e4ac2f2e626d21506b37d19e.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Comparative analysis of 105 datasets across species and tissues reveals higher variability in transcriptomic responses to THC than CBD","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eCannabis use for recreational and medicinal purposes has increased substantially over the past several years in the United States, likely due to broadening legalization of cannabis. By December 2018, medical cannabis was legalized in 33 states and D.C., while recreational use was permitted in 10 states and D.C. (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). In 2022, 30.7% of US high school seniors reported cannabis use in the past year, with 6.3% using it daily (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Among adults, cannabis use increased from 7.59\u0026ndash;15.11% in 2013\u0026ndash;2022 (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Delta-9-tetrahydrocannabinol (THC) and cannabidiol (CBD) are the two most prominent cannabinoids in \u003cem\u003eCannabis sativa\u003c/em\u003e, comprising up to 40% of the plant\u0026rsquo;s extract (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). These compounds interact with the endocannabinoid system, a complex network of receptors and signaling molecules that regulate pain, mood, inflammation, and immune responses (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). THC, the primary psychoactive component, exhibits therapeutic effects as an analgesic for cancer-related chronic pain and has demonstrated anti-invasive and anti-metastatic properties in cancer treatment (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). CBD is non-psychoactive but still influences the brain and nervous system (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e), is recognized for its potential in managing epilepsy, anxiety, and neurodevelopmental disorders, and is valued for its anti-inflammatory and antioxidant properties (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWhile cannabinoids have demonstrated therapeutic potential, concerns regarding their safety in terms of physical health, mental health, public safety, and the side effects have tempered their widespread acceptance (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). For instance, while some studies suggest that chronic daily use of up to 1500 mg/day is well tolerated in humans, others report both physical and mental side effects such as cognitive impairment, cardiovascular complications, and respiratory issues (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Additionally, cannabinoids have been linked to immune suppression, resulting in increased susceptibility to human immunodeficiency virus \u0026minus;\u0026thinsp;1 (HIV-1) infections and disease progression. Mental health effects are particularly concerning, as cannabinoids may exacerbate bipolar disorder in predisposed individuals and increase the risk of temporary psychosis. In addition, cannabinoids also pose broader public health and safety concerns. Driving under the influence is associated with impaired motor function and a higher incidence of motor vehicle accidents. Frequent and heavy cannabis use during adolescence increases the risk of developing cannabis use disorder (CUD) (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e), while addiction and cannabis dependence may contribute to lower income, unemployment, and reduced life satisfaction (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTherefore, understanding the precise molecular and biological mechanisms underlying the beneficial and adverse effects of THC and CBD is crucial for guiding safe use and developing strategies to mitigate health concerns. At the transcriptomic level, studies have shown that cannabinoid-induced gene expression changes vary depending on tissue type, sex, age, and genetic backgrounds, such as gene mutations (\u003cspan additionalcitationids=\"CR13 CR14\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). This variability has made it challenging to establish precision targets that differentiate the therapeutic vs adverse effects. This study systematically analyzes over 100 transcriptomic datasets related to THC and CBD across species and tissue types from the Gene Expression Omnibus (GEO) (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Using transcriptomic data, we identified differentially expressed genes (DEGs) for each dataset. We further investigated global regulatory patterns across datasets through clustering and correlation assessment as well as cannabinoid-specific effects through pathway and disease association analysis, and network modeling. We also investigated the effects of THC and CBD on the endocannabinoid system across the datasets. These integrative approaches allowed us to uncover consistent and unique patterns of gene regulation across studies to partition potential targets underlying beneficial vs. adverse effects of THC and CBD.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"MATERIALS \u0026 METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData curation\u003c/h2\u003e \u003cp\u003eTranscriptomic data from previous studies were curated from the National Center for Biotechnology Information\u0026rsquo;s Gene Expression Omnibus (NCBI GEO) (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). To identify THC-specific datasets, we queried GEO using the keywords \u0026ldquo;THC\u0026rdquo; and \u0026ldquo;Delta-9-tetrahydrocannabinol,\u0026rdquo; while CBD-specific datasets were retrieved using \u0026ldquo;CBD\u0026rdquo; and \u0026ldquo;Cannabidiol.\u0026rdquo; Studies were included if they met the following criteria: 1) the model organism was human, mouse, rat, or rhesus macaque to focus on mammalian species with higher translational potential; 2) the dataset was generated using cDNA microarrays or RNA sequencing; 3) a minimum sample size threshold of n\u0026thinsp;=\u0026thinsp;3 per group for \u003cem\u003ein vivo\u003c/em\u003e studies and n\u0026thinsp;=\u0026thinsp;2 per group for \u003cem\u003ein vitro\u003c/em\u003e studies was applied, allowing us to include most available studies while balancing data quality and comprehensiveness; 4) the study was not part of a subseries or superseries on GEO to avoid duplicate data. For studies testing multiple conditions, samples were grouped based on their matching physiological and pathological background conditions to ensure accurate differential expression analysis. Specifically, only control and treatment samples with shared conditions, such as prior chemical exposures, genetic mutations, and preexisting diseases, were analyzed together. This approach allowed for the assessment of treatment effects within comparable biological contexts while minimizing confounding factors. A total of 47 datasets for THC and 58 datasets for CBD met the criteria and were curated (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo validate that our data curation and analysis procedures can identify THC and CBD-specific transcriptomic effects, we also included additional well-studied, non-cannabinoids chemicals, including bisphenol A (BPA), perfluorooctanoic acid (PFOA), and estradiol, for comparison. Our previous studies of BPA and PFOA have revealed stable and consistent gene signatures for PFOA but more variable gene expression changes in response to BPA across studies, making them potential reference points to assess gene signature stability vs variability (\u003cspan additionalcitationids=\"CR19\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). We also included estradiol, an endogenous bioactive molecule with well-known biology, as a control to assess whether our analytical pipeline retrieves known biology. Gene expression data for these chemicals were obtained from GEO and processed using the same methods as described for THC and CBD. A total of 50 BPA datasets, 39 estradiol datasets, and 14 PFOA datasets were curated (Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eData download and preprocessing\u003c/h3\u003e\n\u003cp\u003eMicroarray data were downloaded from GEO using the \u003cem\u003eGEOquery\u003c/em\u003e package (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). As microarray data submitted to GEO are pre-processed and quality-controlled, we verified normalization and applied log2 transformation before downstream analysis.\u003c/p\u003e \u003cp\u003eRaw RNA sequencing (RNA-seq) datasets were retrieved from NCBI\u0026rsquo;s Sequence Read Archive (SRA), quality-checked, and processed (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). FASTQ files were downloaded using the parallel-fastq-dump wrapper, followed by quality control and preprocessing. Trim Galore (v0.6.6) was used to remove low-quality bases from the 3\u0026rsquo; end of reads (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Cutadapt (v2.1.0) (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e) was employed to remove adapter sequences. Reads shorter than 20 base pairs after trimming and adapter removal were filtered out to ensure data quality.\u003c/p\u003e \u003cp\u003eReads were mapped to appropriate species-specific reference genomes using Salmon (v 1.9.0) (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). The reference genome assemblies used were GRCh38.108 for \u003cem\u003eHomo sapiens\u003c/em\u003e, GRCm39.108 for \u003cem\u003eMus musculus\u003c/em\u003e, mRat.BN7.2.108 for \u003cem\u003eRattus norvegicus\u003c/em\u003e, and Mmul_10.108 for \u003cem\u003eMacaca mulatta\u003c/em\u003e. Reference genomes for each species were indexed using the Salmon index tool to optimize alignment efficiency. Finally, transcript-level quantification results from Salmon were imported and summarized using the \u003cem\u003etximport\u003c/em\u003e package (v1.32.0) (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eDifferentially expressed gene (DEG) analysis\u003c/h3\u003e\n\u003cp\u003eFor each dataset, the treatment group (THC or CBD) was compared to its corresponding control group to identify DEGs using Linear Models for Microarray Data (\u003cem\u003eLIMMA\u003c/em\u003e) for cDNA microarray datasets (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e) and \u003cem\u003eDESeq2\u003c/em\u003e for RNA-seq datasets (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). Multiple testing was corrected using the Benjamini-Hochberg method to obtain false discovery rate (FDR). For studies with multiple doses, baseline conditions (e.g. prior chemical exposures, genetic mutations, and preexisting diseases), or time points, each variation was treated as an independent dataset to extract distinct DEG signatures for each dose, condition, and timepoint. DEGs were considered significant at an FDR\u0026thinsp;\u0026lt;\u0026thinsp;5%. All gene labels were converted to their human orthologs to facilitate cross-species comparisons.\u003c/p\u003e\n\u003ch3\u003eClustering and correlation analysis of DEG gene signatures across datasets\u003c/h3\u003e\n\u003cp\u003eTo compare DEGs across studies, we combined the gene expression log fold changes from individual studies from the differential gene expression analysis. Since RNAseq and microarray experiments can produce log fold changes on different scales, we applied a rank-based normalization method to standardize values within a range of -1 (downregulated in the treatment group) to 1 (upregulated in the treatment group) for each dataset. Specifically, we separately ranked positive and negative log fold changes within each dataset. Positive values were ranked in ascending order and scaled between 0 and 1 (no change to most upregulated), while negative values were ranked in descending order and scaled between \u0026minus;\u0026thinsp;1 and 0 (most downregulated to no change). This approach preserves the relative magnitude of gene expression changes among genes while making the datasets comparable across studies and transcriptome platforms. To assess similarity and differences across datasets for THC and CBD through cluster analysis, we used the top 2,500 most variable genes. For the analysis across all chemicals (THC, CBD, PFOA, BPA, estradiol), given the large number of datasets, we first selected genes present in at least 70% of the 208 datasets before selecting the top 2,500 most variable genes. Missing values in the DEG fold change table were imputed using the \u003cem\u003emissForest\u003c/em\u003e R package (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). We then applied UMAP for dimensionality reduction using Euclidean distance to visualize the clustering patterns across chemicals and datasets.\u003c/p\u003e \u003cp\u003eAs an alternative approach to assess similarity and differences across datasets, we used Spearman\u0026rsquo;s rank correlation analysis of the normalized gene expression fold change data to compute pairwise correlations among datasets. The pairwise correlation coefficients were further used to group datasets into clusters with similar gene regulation patterns. To identify consistent DEGs across datasets within each cluster in response to THC or CBD, we performed a meta-analysis using the Robust Rank Aggregation (RRA) package (v1.2.1) (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). Rank aggregation was conducted separately for up-regulated and down-regulated DEGs in each dataset cluster. Genes with an RRA score\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered robust, consistent DEGs, reflecting consistent differential expression across datasets within each cluster. The RRA score measures the probability of a gene achieving its observed ranking pattern across datasets by chance, with lower scores indicating greater consistency.\u003c/p\u003e\n\u003ch3\u003ePathway enrichment analysis\u003c/h3\u003e\n\u003cp\u003ePathway enrichment analysis was performed on the identified DEGs using EnrichR (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e). DEGs were compared against the Gene Ontology Biological Process (GOBP) (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) databases to identify significantly enriched pathways. Pathways with an FDR\u0026thinsp;\u0026lt;\u0026thinsp;5% and at least 5 overlapping DEGs were considered statistically significant.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eWeighted Key Driver Analysis (wKDA) for brain network modeling of THC and CBD DEGs\u003c/h2\u003e \u003cp\u003eTo identify potential gene regulatory networks and network key drivers (KDs) underlying cannabis-induced brain effects, we applied Weighted Key Driver Analysis from the Mergeomics pipeline (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). We conducted the analysis separately from THC- and CBD-treated brain datasets. Using a previously constructed brain Bayesian network based on large human and animal model omics studies in Mergeomics, we identified nodes (genes) whose immediate subnetworks were enriched for the THC or CBD DEG sets at FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05 as statistically significant KDs. Network visualizations were performed using Cytoscape (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMarker Set Enrichment Analysis (MSEA) for disease/trait association assessment\u003c/h3\u003e\n\u003cp\u003eMSEA in the Mergeomics package was used to identify the enrichment of THC or CBD DEGs for genetic associations with 101 genome-wide association studies (GWAS) of diseases and phenotypic traits (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). Disease-associated genes were mapped using full summary statistics from the GWAS Catalog (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e), with single nucleotide polymorphisms (SNPs) assigned to genes within a 50 kb distance. MSEA employs a chi-square-like statistic with multiple quantile thresholds to assess whether a DEG set shows enrichment of disease SNPs compared to random chance. 10,000 permuted gene sets were generated for each DEG set. As detailed in Shu et al. (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e), the enrichment statistics from the permutations were used to approximate a Gaussian distribution from which enrichment p-values were determined. FDR was estimated using the Benjamini-Hochberg (BH) correction. DEG sets were determined to be statistically significant for a given disease or trait if FDR\u0026thinsp;\u0026lt;\u0026thinsp;5%.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCuration of transcriptomic datasets on cannabinoids across species and tissues\u003c/h2\u003e \u003cp\u003eTo investigate the impact of cannabinoids on gene expression, we obtained 108 transcriptomic datasets from GEO (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e) (Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Specifically, we obtained 3 datasets for overall cannabis use, 58 for CBD, and 47 for THC, spanning 32 human (\u003cem\u003eHomo sapiens\u003c/em\u003e) studies, 8 non-human primate (\u003cem\u003eMacaca mulatta\u003c/em\u003e) studies, 50 mouse (\u003cem\u003eMus musculus\u003c/em\u003e) studies, and 18 rat (\u003cem\u003eRattus norvegicus\u003c/em\u003e) studies across 15 broad tissue categories (e.g. blood, brain, cancer, digestive, heart, immune, kidney, liver, lung, muscle, oral, placenta, skin, stem cell, and vasculature). Since the \u0026ldquo;overall cannabis use\u0026rdquo; datasets involved either cannabis smoke exposure or observational studies in which other botanical compounds from the cannabis plant were not controlled for, we excluded them from our analysis to maintain our specific focus on the compounds THC and CBD. Notably, reflecting the focus on neurological effects of cannabinoids, brain-related datasets accounted for nearly half of all studies.\u003c/p\u003e \u003cp\u003eMany studies included various physiological and pathological background conditions in both control and treatment groups, such as genetic mutations (e.g., \u003cem\u003eCox15\u003c/em\u003e, \u003cem\u003eNdufs4\u003c/em\u003e), preexisting diseases (e.g., SARS-CoV-2, Simian Immunodeficiency Virus (SIV) infection), and prior chemical exposures (e.g., 2'-3'-cGAMP, formoterol/budesonide treatment). Additionally, the datasets covered variations in study design, including differences in sex distribution, dose, route and duration of administration. The broad coverage of diverse conditions allows us to not only determine specific transcriptomic signatures for each condition but also assess the consistency and generalizability of cannabinoid-specific effects across datasets. After downloading and preprocessing the gene expression data, we conducted differential gene expression analysis. We then performed pathway enrichment and disease association analyses to characterize the broader transcriptomic effects of cannabinoids (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eTHC and CBD DEGs are distinct from other chemicals\u003c/h2\u003e \u003cp\u003eTo evaluate whether cannabinoids THC and CBD show distinct transcriptomic signatures compared to other chemicals, we incorporated 103 datasets from three well-characterized chemicals, Bisphenol A (BPA), perfluorooctanoic acid (PFOA), and estradiol, which have been associated with metabolic and reproductive functions, respectively. Both BPA and PFOA are endocrine-disrupting chemicals that influence cardiometabolic disease-related pathways (\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). Estradiol, a primary estrogen hormone, plays a role in reproductive and sexual function by regulating gene expression through estrogen receptors (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e). We selected these as comparative datasets because we expect them to be functionally distinct from THC and CBD, which can be reflected by different downstream genes and pathways. In total, we included 50 BPA datasets, 39 estradiol datasets, and 14 PFOA datasets from GEO (Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e) and processed them using the same methodology as cannabinoids.\u003c/p\u003e \u003cp\u003eWe first compared the chemical-induced expression changes for all genes across all THC, CBD, BPA, estradiol, and PFOA datasets based on the similarities in the normalized log fold changes in gene expression changes between treatment and control groups using clustering analysis and visualized the datasets with UMAP. Agreeing with our hypothesis, most BPA, estradiol, and PFOA datasets each formed distinct clusters, supporting biological coherence within each chemical and differences between chemicals (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). These results suggest that comparative transcriptomics analysis captures gene expression patterns that likely reflect the different molecular mechanisms of different classes of chemicals. Agreeing with our previous observation on higher stability of PFOA gene signatures (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e), most PFOA datasets clustered tightly in UMAP. In comparison, while BPA and estradiol datasets also formed distinct clusters in UMAP, the distance between datasets for BPA or estradiol was larger than that for PFOA. Compared to these reference chemicals, THC and CBD datasets were more spread out in UMAP. These results suggest that the molecular mechanisms of the reference chemicals are more coherent and consistent across datasets and species, with PFOA showing the highest stability, whereas THC and CBD datasets showed fewer stable gene signatures in comparison. Additionally, we did not observe a distinct separation between datasets from RNA sequencing and microarray methods, indicating a minimal effect of the technical platform on gene signatures (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe also observed that many tightly clustered datasets for the reference chemicals originated from the same study. For example, several BPA and estradiol microarray datasets were derived from GSE50705, where datasets shared the same experimental setup but differed in dosage. These datasets are clustered in the upper left corner of the UMAP plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). To mitigate the contribution of study-specific batch effects to the tight clustering of the chemicals, we averaged the log fold changes of each gene across datasets for each chemical from the same study, then performed dimensionality reduction again (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eB). The clustering patterns remained consistent with the original analysis, confirming that study-related effects did not drive the observed trends in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA. Specifically, when accounting for study effects, we found a stability ranking among the chemicals: PFOA exhibited the highest stability, followed by estradiol and BPA, with CBD and THC being highlighted as having the greatest variability.\u003c/p\u003e \u003cp\u003e \u003cb\u003eCBD gene signatures demonstrate tighter clustering and higher stability across species and routes of administration compared to THC\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTo better assess the similarities and differences between THC and CBD datasets, we further clustered THC and CBD datasets excluding BPA, estradiol, and PFOA datasets. There was a subtle separation between THC and CBD datasets along the first UMAP dimension (UMAP1), indicating that their global transcriptional effects exhibit certain differences although not strongly distinguishable (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Notably, datasets from both RNAseq and microarray platforms were mixed, again supporting that the technological platform did not induce major differences in transcriptomic signatures (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eC, S1D, S1E).\u003c/p\u003e \u003cp\u003eWithin the THC datasets, we found no distinct clustering patterns based on species (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC), route of administration (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eF), or tissue type (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eG), suggesting that these factors did not significantly influence global transcriptional profiles.\u003c/p\u003e \u003cp\u003eIn contrast, CBD datasets exhibited weak species-specific clustering. Along the second UMAP dimension (UMAP2), \u003cem\u003eHomo sapiens\u003c/em\u003e and \u003cem\u003eMus musculus\u003c/em\u003e datasets showed subtle shifts, whereas \u003cem\u003eRattus norvegicus\u003c/em\u003e datasets clustered more tightly (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). However, these trends were not strong enough to indicate robust species-driven transcriptional signatures. Regarding the route of administration, \u003cem\u003ein vitro\u003c/em\u003e (in medium) datasets tended to cluster at higher UMAP1 values, whereas \u003cem\u003ein vivo\u003c/em\u003e (injection or ingestion) datasets were mainly positioned at lower UMAP1 values (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eH). In the tissue-based UMAP (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eI), no clear clustering trends emerged across tissue types. However, datasets from the same study, such as the two heart datasets and two kidney datasets, clustered more closely, suggesting that study-specific effects also contributed to their proximity.\u003c/p\u003e \u003cp\u003eOverall, the weak clustering patterns across species, tissues, and routes of administration suggest that these factors did not play a dominant role in shaping the global transcriptomic shifts induced by THC or CBD. However, CBD datasets displayed slightly greater clustering and stability compared to THC, indicating more consistent transcriptional responses across different conditions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eCBD studies demonstrate different correlation patterns than THC datasets\u003c/h2\u003e \u003cp\u003eIn addition to analyzing dataset clustering based on global gene expression changes through dimensional reduction using UMAP, we assessed correlation patterns in gene expression changes across all THC, CBD, BPA, estradiol, and PFOA datasets. A positive correlation (red) indicates that most genes are upregulated or downregulated concordantly between two datasets, whereas a negative correlation (blue) suggests discordance or opposite directions in gene regulation patterns. Results from this analysis are consistent with our UMAP clustering results in that the molecular patterns of THC and CBD are less consistent compared to the reference chemicals (Figure \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNext, we carried out clustering analyses separately for THC and CBD datasets based on the correlation patterns. In the THC correlation heatmaps, two major clusters of positively correlated datasets were observed: THC Cluster 1 in the upper left quadrant (11 datasets) and THC Cluster 2 in the lower right quadrant (6 datasets), both largely composed of brain-related datasets spanning multiple species. However, these two clusters exhibited moderate negative correlations with each other (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). This result suggests that THC gene signatures from the brain could be partitioned into distinct patterns with each pattern showing consistency across studies. Notably, some datasets with the same experimental designs but differing in sex were assigned to different clusters. For instance, in GSE273695, which analyzed rat blood samples with daily intraperitoneal injections escalating from 2.5 mg/kg to 10 mg/kg, the male dataset clustered in THC Cluster 1, while the female dataset appeared in THC Cluster 2. Additionally, multiple mouse brain datasets from GSE189821 (daily 10 mg/kg intraperitoneal injection) were distributed across both clusters: specifically, the male amygdala dataset was placed in THC Cluster 2, while the female amygdala dataset fell into THC Cluster 1. However, this separation of male and female samples into different clusters was not observed in all datasets. For example, rat prefrontal cortex datasets of both sexes in GSE273695 were all assigned to THC Cluster 1. While this finding suggests that sex may influence THC-induced transcriptomic responses, this sex-based separation was not consistently observed across all THC datasets and hence any potential sex-specific effects are likely more nuanced and depending on tissue type or other conditions to be elucidated.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eCBD datasets formed four distinct clusters, with the two largest being CBD Cluster 1 in the upper left quadrant (17 datasets) and CBD Cluster 4 in the lower right quadrant (15 datasets) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). These clusters were primarily composed of \u003cem\u003ein vitro\u003c/em\u003e experiments and included datasets from three species: human, mouse, and rat. CBD Cluster 1 featured a mixture of tissues, with prominent representation from immune, cancer, and brain. In contrast, approximately half of the datasets in CBD Cluster 4 were from the skin. Two smaller clusters were observed in the central region of the plot, representing study-specific clusters: CBD Cluster 2 included 7 mouse hypothalamus cell line datasets from GSE270378, and CBD Cluster 3 contained 5 mouse liver datasets from GSE261716, where datasets varied by CBD dosage and exposure time.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eCBD induces more transcriptional changes than THC\u003c/h2\u003e \u003cp\u003eTo go beyond global transcriptomic patterns and focus on individual genes exhibiting significant changes in response to THC and CBD, we derived DEGs at FDR\u0026thinsp;\u0026lt;\u0026thinsp;5%. Overall, CBD induced more significant DEGs than THC (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). Most CBD datasets contain hundreds or thousands of DEGs, whereas THC datasets predominantly have fewer than 100 DEGs (Table \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003e). Kolmogorov-Smirnov (KS) tests of the distributions of DEG counts across datasets confirmed that both upregulated and downregulated gene counts differed significantly between THC and CBD.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOther factors, including transcriptome platform, species, tissue, exposure type, duration, dosage, and sex, also influenced DEG counts. RNA-seq datasets generally yield more DEGs than microarray datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). \u003cem\u003eIn vitro\u003c/em\u003e exposure studies of CBD exhibited higher numbers of DEGs than THC, while \u003cem\u003ein vivo\u003c/em\u003e datasets show similar DEG distributions between CBD and THC (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). Examining species effects in CBD datasets, human \u003cem\u003ein vitro\u003c/em\u003e cell line studies had the highest number of DEGs (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE). For THC datasets, where the most DEG-rich studies are \u003cem\u003ein vivo\u003c/em\u003e, mouse studies generate the largest number of significant DEGs, followed by non-human primate and rat studies. At the tissue level, CBD studies show the highest median number of DEGs in lung, muscle, cancer, immune, and kidney tissues (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF). For THC, while no tissue exhibits a notably high median DEG count, the brain had the greatest variation in DEG numbers across studies, likely due to brain region-specific effects (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eG). Additionally, as expected, the number of DEGs in CBD studies increases with both exposure time and dosage (Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eA, Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eB, Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eC for different studies). For THC, no study specifically examined both time points and dosage variations. Unlike time and dosage, sex had a notable but inconsistent effect on DEG counts. Among 10 THC datasets with identical experimental designs differing only by sex, females exhibited more differentially expressed genes (DEGs) in five datasets, including the mouse brain amygdala, dorsolateral striatum, ventral tegmental area (VTA), prefrontal cortex, and rat peripheral blood mononuclear cells (PBMCs) under LPS stimulation. In contrast, males showed higher DEG counts in four datasets: the rat brain orbitofrontal cortex (under both LPS and saline treatments), rat PBMCs with saline, and the mouse brain nucleus accumbens (Figure \u003cspan refid=\"MOESM3\" class=\"InternalRef\"\u003eS3\u003c/span\u003eD). This variability highlights the complexity of sex-specific factors in cannabinoid responses.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eCBD demonstrates higher consistency at gene and pathway levels across datasets\u003c/h2\u003e \u003cp\u003eTo evaluate the consistency of differentially expressed genes (DEGs) across datasets, we applied two complementary approaches: 1. identifying consistent DEGs within correlation-based clusters, and 2. assessing individual DEG recurrence and directionality consistency.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eCluster-Level Consistency:\u003c/h2\u003e \u003cp\u003eWe used rank aggregation across datasets and identified consistently upregulated and downregulated DEGs within each correlation cluster (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e). THC Cluster 1 contained relatively few DEGs, with 29 consistently upregulated genes, 18 consistently downregulated genes, and no significantly enriched pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eA). Many consistent DEGs are related to neurodevelopment and brain functions. For example, upregulated genes include \u003cem\u003eTBR1\u003c/em\u003e, a key regulator of cortical development (\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e); \u003cem\u003eNPTX2\u003c/em\u003e, neuronal pentraxin for synaptic plasticity (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e); neuropeptide S receptor \u003cem\u003eNPSR1\u003c/em\u003e in neuroendocrine cells (\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e); \u003cem\u003eADRA2B\u003c/em\u003e, which regulates neurotransmitter norepinephrine (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e); \u003cem\u003eADORA2A\u003c/em\u003e, linked to anxiety, arousal, and sleep regulation (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). Downregulated genes include \u003cem\u003eFEZF1\u003c/em\u003e for neuronal differentiation (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e); \u003cem\u003eLHX5\u003c/em\u003e associated with mammillary body development (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e); \u003cem\u003eGPR50\u003c/em\u003e for neural progenitor cell differentiation (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e); \u003cem\u003eCACNA1B\u003c/em\u003e, a calcium channel subunit for neurotransmitter release (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e). THC Cluster 2 showed even lower consistency, with only two consistently downregulated genes, B Cell receptor \u003cem\u003eCD72\u003c/em\u003e (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e) and presynaptic receptor \u003cem\u003eGRM8\u003c/em\u003e (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e), and no significant pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Table \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003eIn contrast, CBD Cluster 1 exhibited robust transcriptional changes, with 1,005 consistently upregulated DEGs and 997 consistently downregulated DEGs (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Table \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eA). The upregulated pathways included negative regulation of growth, cellular response to zinc ions, response to ER stress, and inflammatory response. In contrast, the downregulated pathways were primarily related to development, ECM organization, and sprouting angiogenesis, and skin development. CBD Cluster 4 had 251 consistently upregulated DEGs and 283 downregulated DEGs (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Table \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eB), but only the upregulated DEGs showed significantly enriched pathways, including negative regulation of growth and cellular response to zinc ions. The absence of strong downregulated pathways in Cluster 4 may explain the separation between the two large clusters and their weak negative correlation. CBD Clusters 2 and 3 displayed more limited consistency. CBD Cluster 2 only had 2 consistent DEGs: cellular zinc sensor \u003cem\u003eMTF1\u003c/em\u003e upregulated (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e) and \u003cem\u003eSENP3\u003c/em\u003e in \u003cem\u003eWnt\u003c/em\u003e signaling downregulated (\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Table \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eC). CBD Cluster 3 had 73 consistently upregulated DEGs, 77 downregulated DEGs, and two enriched upregulated pathways: cellular response to lipopolysaccharide and inflammatory response (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Table \u003cspan refid=\"MOESM5\" class=\"InternalRef\"\u003eS5\u003c/span\u003eD).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eConsistency in DEGs across Datasets:\u003c/h2\u003e \u003cp\u003eWhen evaluating consistency in DEGs across datasets, again we found CBD induced broader and more reproducible gene expression changes compared to THC (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA, Table \u003cspan refid=\"MOESM6\" class=\"InternalRef\"\u003eS6\u003c/span\u003eA, Table \u003cspan refid=\"MOESM6\" class=\"InternalRef\"\u003eS6\u003c/span\u003eB). Among the 30 most frequently detected DEGs, 22 exhibited (\u0026gt;\u0026thinsp;70%) a consistent direction of regulation, meaning that they were detected as either consistently upregulated or downregulated across datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA, Table \u003cspan refid=\"MOESM6\" class=\"InternalRef\"\u003eS6\u003c/span\u003eA). For example, \u003cem\u003eSLC30A1\u003c/em\u003e, a zinc transporter (\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e), was significantly upregulated in 23 CBD datasets and not downregulated in any. \u003cem\u003eHMOX1\u003c/em\u003e (heme oxygenase 1), a gene involved in oxidative stress response (\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e), was significantly upregulated in 19 datasets while downregulated in only 4 CBD datasets. Some other frequent DEGs play a role in ion metabolism, such as downregulation of \u003cem\u003eSLC39A10\u003c/em\u003e for zinc transport (\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e) and upregulation of \u003cem\u003eMT2A\u003c/em\u003e and \u003cem\u003eMT1E\u003c/em\u003e for metal detoxification and protection against oxidative stress (\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn contrast, DEGs associated with THC were less frequently replicated across datasets, and the most recurrent genes appeared only three times across 46 datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA, Table \u003cspan refid=\"MOESM6\" class=\"InternalRef\"\u003eS6\u003c/span\u003eB). \u003cem\u003ePRXL2A\u003c/em\u003e (peroxiredoxin-like 2A), an antioxidant protein that protects cells from oxidative stress (\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e), was downregulated in three datasets. Other THC-associated upregulated genes included \u003cem\u003eBPGM\u003c/em\u003e in red blood cell metabolism, \u003cem\u003eCOL1A2\u003c/em\u003e for collagen synthesis, microtubule-binding protein \u003cem\u003eFRY\u003c/em\u003e, a potassium channel gene \u003cem\u003eKCNQ5\u003c/em\u003e, circadian regulator \u003cem\u003ePER1\u003c/em\u003e, \u003cem\u003eRALGAPA2\u003c/em\u003e for intracellular signaling, and \u003cem\u003eXRCC2\u003c/em\u003e for DNA repair.\u003c/p\u003e \u003cp\u003eWe also performed pathway enrichment analysis of the significant DEGs for each dataset and summarized the frequency of significant pathways across datasets. Again, THC studies exhibited minimal pathway-level consistency (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB, \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC, Table \u003cspan refid=\"MOESM7\" class=\"InternalRef\"\u003eS7\u003c/span\u003eB), but CBD datasets showed recurring enriched pathways (Table \u003cspan refid=\"MOESM7\" class=\"InternalRef\"\u003eS7\u003c/span\u003eA). Consistent CBD pathways include the upregulation of the \u0026ldquo;metabolic pathway\u0026rdquo; across 17 studies (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB) and the \u0026ldquo;cellular response to zinc ion\u0026rdquo; pathway across nine studies (consistent with the reproducibility of DEGs \u003cem\u003eSLC30A1\u003c/em\u003e and \u003cem\u003eMT2A\u003c/em\u003e), and the downregulation of cell cycle, mRNA export, cancer, and mismatch repair pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eGene and pathway-level evidence for the endocannabinoid system in CBD and THC transcriptomic signatures\u003c/h2\u003e \u003cp\u003eGiven the well-established interaction between cannabinoids and the endocannabinoid system (ECS) (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e), we anticipated enrichment of ECS-associated genes in the transcriptional signatures of both CBD and THC. To systematically evaluate this, we curated a list of 135 ECS-associated genes using prior literature and established databases, including Gene Ontology Biological Process (GOBP) annotations (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) and the DisGeNET database (\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e, \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e) (Table \u003cspan refid=\"MOESM9\" class=\"InternalRef\"\u003eS9\u003c/span\u003e). This gene set includes metabolic enzymes involved in endocannabinoid degradation (i.e. \u003cem\u003eFAAH\u003c/em\u003e, \u003cem\u003eMGLL\u003c/em\u003e, \u003cem\u003eABHD6\u003c/em\u003e, \u003cem\u003eABHD12\u003c/em\u003e, \u003cem\u003ePTGS2\u003c/em\u003e) and biosynthesis (i.e. \u003cem\u003eNAPEPLD\u003c/em\u003e, \u003cem\u003eDAGLA\u003c/em\u003e, \u003cem\u003eDAGLB\u003c/em\u003e) of key ligands, such as 2-arachidonoylglycerol (2-AG) and anandamide (AEA). It also encompasses cannabinoid receptors (i.e. \u003cem\u003eCRN1\u003c/em\u003e, \u003cem\u003eCRN2\u003c/em\u003e, \u003cem\u003eGPR55\u003c/em\u003e, \u003cem\u003eTRPV1\u003c/em\u003e), neurotransmitter receptors (i.e. \u003cem\u003eDRD2\u003c/em\u003e, \u003cem\u003eDRD4\u003c/em\u003e, \u003cem\u003eHTR1A\u003c/em\u003e, \u003cem\u003eSSTR4\u003c/em\u003e), or nicotinic acetylcholine receptors (i.e. \u003cem\u003eCHRNA2\u003c/em\u003e, \u003cem\u003eCHRNA3\u003c/em\u003e, \u003cem\u003eCHRNA5\u003c/em\u003e, \u003cem\u003eCHRNA6\u003c/em\u003e, \u003cem\u003eCHRNA7\u003c/em\u003e, \u003cem\u003eCHRNAB3\u003c/em\u003e), endocannabinoid transporters (i.e. \u003cem\u003eFABP1\u003c/em\u003e, \u003cem\u003eFABP3\u003c/em\u003e, \u003cem\u003eFABP5\u003c/em\u003e, \u003cem\u003eFABP7\u003c/em\u003e), and transcriptional regulators (i.e. \u003cem\u003ePPARA\u003c/em\u003e, \u003cem\u003eEP300\u003c/em\u003e, \u003cem\u003eFOXP2\u003c/em\u003e).\u003c/p\u003e \u003cp\u003eWe examined the regulation of ECS genes identified as DEGs and found that more CBD datasets had significantly regulated ECS genes compared to THC (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eA, Table \u003cspan refid=\"MOESM10\" class=\"InternalRef\"\u003eS10\u003c/span\u003e). In addition, THC datasets generally exhibited fewer ECS DEGs than CBD datasets. Notably, most THC datasets with significant ECS gene regulation were brain-derived, whereas CBD datasets predominantly originated from peripheral tissues. A cluster on the left side of Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eA, composed mainly of peripheral CBD datasets, exhibited some consistency in regulatory direction. For instance, consistently upregulated genes included \u003cem\u003eSQSTM1\u003c/em\u003e, \u003cem\u003eSLC6A9\u003c/em\u003e, and \u003cem\u003eBEST1\u003c/em\u003e, while consistently downregulated genes included \u003cem\u003eCALM2\u003c/em\u003e, \u003cem\u003eKIAA2013\u003c/em\u003e, and \u003cem\u003eSMC2\u003c/em\u003e. This limited degree of coherence suggests weak gene-level evidence for consistent ECS involvement in the transcriptomic effects of CBD and THC, although context-dependent, tissue-specific regulation of individual genes remains possible.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo evaluate whether ECS pathway genes were overrepresented (enriched) among differentially expressed genes (DEGs), we performed Fisher\u0026rsquo;s Exact Tests comparing the DEG lists to the ECS gene set. Significance was defined as p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, with p\u0026thinsp;\u0026lt;\u0026thinsp;0.10 considered suggestive (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eB, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eC, Table \u003cspan refid=\"MOESM10\" class=\"InternalRef\"\u003eS10\u003c/span\u003e). Overall, CBD showed more datasets with significant or suggestive enrichment compared to THC. Notably, all THC datasets showing significant and suggestive ECS gene set enrichment originated from brain regions, including the rat orbitofrontal cortex, mouse amygdala, mouse dorsolateral striatum, and mouse medial prefrontal cortex (Table \u003cspan refid=\"MOESM10\" class=\"InternalRef\"\u003eS10\u003c/span\u003e). In contrast, approximately half of the significantly or suggestively enriched CBD datasets came from peripheral tissues and cell lines, including the mouse liver, human liver cell line, monocyte cell line, and microglial cell line. These findings support that THC\u0026rsquo;s regulatory effects are brain-specific, whereas CBD exhibits more pleiotropic effects across central and peripheral systems.\u003c/p\u003e \u003cp\u003eIn addition, we investigated pathways related to pain and inflammation, which are regulated by the endocannabinoid system (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). However, no significant enrichment was found for the GOBP \u0026ldquo;sensory perception of pain pathway\u0026rdquo; and \u0026ldquo;response to pain pathway\u0026rdquo; (\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). However, the inflammatory response pathway was enriched among DEGs from two CBD datasets, both derived from monocyte-derived dendritic cell lines.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eTHC- and CBD-induced brain network modeling identifies unique key drivers\u003c/h2\u003e \u003cp\u003eWe next focused on the brain datasets on THC and CBD to identify key similarities and differences in the regulation of cannabis-induced transcriptomic alterations in the brain. We applied weighted key driver analysis (wKDA) to DEG sets from THC- and CBD- treated brain datasets to identify potential key drivers (KDs) using a Bayesian gene regulatory network previously constructed from dozens of human and mouse brain datasets (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). KDs with a false discovery rate (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant.\u003c/p\u003e \u003cp\u003eFor CBD-treated brain datasets, 14 significant KDs were identified (top 5 shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003eA; complete list in Table \u003cspan refid=\"MOESM8\" class=\"InternalRef\"\u003eS8\u003c/span\u003e). The KDs in the CBD network were derived from multiple brain cell lines, including mouse brain microglia, rat primary hippocampal neurons, and human neuroblastoma cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003eA). Most KDs were involved in neural development and synaptic functions, such as \u003cem\u003eSH3GL2\u003c/em\u003e, \u003cem\u003eDPYSL5\u003c/em\u003e, \u003cem\u003eJPH3\u003c/em\u003e, \u003cem\u003eSLC17A7\u003c/em\u003e, and \u003cem\u003eGRIN3B\u003c/em\u003e (\u003cem\u003eMMT00076709\u003c/em\u003e). Other KDs were associated with signal transduction and kinase activity (\u003cem\u003ePRKCA\u003c/em\u003e, \u003cem\u003eTAOK1\u003c/em\u003e, and \u003cem\u003eADRBK1\u003c/em\u003e), membrane transport (\u003cem\u003eSLC9A3R1\u003c/em\u003e), metabolism (\u003cem\u003ePPP1R3B\u003c/em\u003e), gene regulation (\u003cem\u003eUSF1\u003c/em\u003e), and immune responses and cancer (\u003cem\u003eSLAMF8\u003c/em\u003e, \u003cem\u003eSRC\u003c/em\u003e, and \u003cem\u003eKIAA0100\u003c/em\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eDespite the lack of consistent DEGs and pathways across THC-treated brain datasets, we identified a total of 176 significant KDs based on DEGs from individual datasets (top 5 shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003eB; complete list in Table \u003cspan refid=\"MOESM8\" class=\"InternalRef\"\u003eS8\u003c/span\u003e), among which \u003cem\u003eSLC17A7\u003c/em\u003e and \u003cem\u003eTAOK1\u003c/em\u003e were shared KDs between THC and CBD. Notably, most THC KDs were found to be significant in only a single DEG dataset rather than across multiple datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003eB), suggesting that THC may exert brain region-specific effects on gene regulation. Key drivers (KDs) in the THC network were identified from multiple animal models, including the mouse hippocampus, mouse amygdala, mouse microglia, and rat orbitofrontal cortex. A majority of THC-associated KDs were involved in neuronal functions, signal transduction, structural integrity, and metabolic functions. Notably, some KDs with known neuronal roles include \u003cem\u003eSLC17A7\u003c/em\u003e, \u003cem\u003eRTN4R\u003c/em\u003e, \u003cem\u003eRTN4RL1\u003c/em\u003e, \u003cem\u003eRTN4RL2\u003c/em\u003e, \u003cem\u003eNEUROD2\u003c/em\u003e, \u003cem\u003eNEUROD6\u003c/em\u003e, \u003cem\u003eNGB\u003c/em\u003e, \u003cem\u003eCAMK2A\u003c/em\u003e, \u003cem\u003eADCY1\u003c/em\u003e, \u003cem\u003eSYT1\u003c/em\u003e, \u003cem\u003eDLGAP1\u003c/em\u003e, \u003cem\u003eDLGAP3\u003c/em\u003e, \u003cem\u003eNTS\u003c/em\u003e, \u003cem\u003eRGS14\u003c/em\u003e, \u003cem\u003eGAD2\u003c/em\u003e, and \u003cem\u003eCNIH3\u003c/em\u003e.\u003c/p\u003e \u003cp\u003e \u003cb\u003eCBD and THC DEGs are associated with neuropsychiatric disorders while CBD is associated with more metabolic traits\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe endocannabinoid system plays a role in various health outcomes, including mood disorders, cardiovascular disease, stroke, cancer, diabetes, autoimmune conditions, and neurological disorders (\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e). To further investigate the connection between cannabinoids and human health, we analyzed publicly available summary statistics from large-scale Genome-Wide Association Studies (GWAS) for more than 100 diseases or phenotypic traits using Marker Set Enrichment Analysis (MSEA) via the Mergenomics package (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). This approach enabled us to assess whether signature genes from THC and CBD datasets were enriched for human disease/phenotype variants.\u003c/p\u003e \u003cp\u003eDEGs from both cannabinoids showed significant associations with Schizophrenia, type 2 diabetes (T2D), and depressive symptoms (Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e). In addition, CBD transcriptomic signatures exhibited broader associations with lipid metabolism, including low-density lipoprotein (LDL), high-density lipoprotein (HDL), total cholesterol (TC), and triglycerides (TG), as well as with body composition traits such as waist circumference (WCadjBMI), hip circumference (HIPadjBMI), waist-to-hip ratio (WHRadjBMI), and height. In contrast, THC was linked only to WCadjBMI. Beyond metabolic and anthropometric traits, CBD was associated with Crohn\u0026rsquo;s disease (CD) and coronary artery disease (CAD).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo identify the genes most strongly linked to disease-associated SNPs, we examined the top genes from the significantly enriched datasets (Figure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eA, S4B, S4C, S4D). In the CBD datasets, key associations for CD included tumor suppressor \u003cem\u003eCYLD\u003c/em\u003e (\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e) and cytosolic receptor \u003cem\u003eNOD2\u003c/em\u003e (Figure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eA). Mutations or altered expression of \u003cem\u003eNOD2\u003c/em\u003e have been observed in CD patients (\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e), and genetic studies have identified \u003cem\u003eNOD2\u003c/em\u003e locus polymorphisms and an independent involvement of the neighboring gene \u003cem\u003eCYLD\u003c/em\u003e in CD (\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e). For T2D, a key association for CBD DEGs was the Wnt signaling pathway transcription factor \u003cem\u003eTCF7L2\u003c/em\u003e (Figure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eC). Carrying two copies of a common variant in this gene is associated with an approximately twofold increase in T2D risk (\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e). CBD DEGs CDKN2A and CDKN2B were also among the top CAD-associated GWAS candidates, which are located adjacent to the lead CAD-linked SNP on the 9p21 locus (\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e) (Figure \u003cspan refid=\"MOESM4\" class=\"InternalRef\"\u003eS4\u003c/span\u003eD). In contrast, THC showed more moderate levels of association between DEGs and disease-related SNPs. Of note, this analysis mainly focuses on overlaps between CBD/THC DEGs with genetic associations of diseases and does not implicate increases or decreases in disease risks.\u003c/p\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this study, we systematically analyzed over 100 publicly available transcriptomic datasets to understand the molecular effects of THC and CBD across species and tissue types. By examining both global and significant gene expression changes from microarray and RNA-sequencing data, we found highly variable gene signatures for these two cannabinoids compared to other exposures such as PFOA, BPA, and estradiol. We also identified DEGs, pathways, network key drivers (KDs), and diseases/phenotypes associated with the significant DEGs. Our findings highlight significant variability in THC and CBD-induced transcriptional responses, with CBD demonstrating better consistency across studies, species, and central and peripheral tissue types compared to THC, which shows more dataset-specific effects in brain regions.\u003c/p\u003e \u003cp\u003eOur broad dimensionality reduction and correlation analyses revealed that CBD and THC datasets were more scattered in contrast with other well-studied chemicals PFOA, BPA, and estradiol, which displayed stronger internal consistency across datasets. In addition, neither THC nor CBD datasets exhibited strong clustering in the UMAP or formed larger clusters in the correlation analyses across species and tissues. This finding suggests that cannabinoid-induced gene expression changes are highly variable depending on combinations of experimental conditions, genetic or disease backgrounds, and prior chemical exposures. These results highlight the challenge of defining a universal and predictable transcriptional signature for cannabinoids and align with previous literature indicating that cannabinoids effects depend on tissue, developmental stage, sex, and exposure conditions (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eOur analyses on the DEGs and enriched pathways across datasets indicate that CBD has a more stable gene regulatory signature than THC. This regulatory stability agrees with CBD\u0026rsquo;s well-characterized antioxidant, anti-inflammatory, and neuroprotective effects (\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e). In particular, two major CBD-associated gene clusters (CBD Cluster 1 and CBD Cluster 4) exhibited hundreds to thousands of consistent DEGs, with both CBD clusters exhibiting upregulation in two pathways: negative regulation of growth and cellular response to zinc ions. CBD was associated with higher frequency of consistent DEGs across datasets, such as \u003cem\u003eHMOX1\u003c/em\u003e, \u003cem\u003eSLC30A1\u003c/em\u003e, \u003cem\u003eMT2A\u003c/em\u003e, and \u003cem\u003eSLC39A10\u003c/em\u003e. Previous \u003cem\u003ein vitro\u003c/em\u003e studies have identified \u003cem\u003eHMOX1\u003c/em\u003e as the most upregulated gene and protein following CBD treatment. The proposed mechanism involves CBD-induced nuclear export and proteasomal degradation of transcriptional repressor BACH1(\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e). Another study suggests that CBD may influence \u003cem\u003eHMOX1\u003c/em\u003e levels through the Nrf2 pathway, which regulates antioxidant defenses (\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e). These indicate that \u003cem\u003eHMOX1\u003c/em\u003e is linked to antioxidant and anti-inflammatory properties under the impact of CBD (\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e). Notably, pathways related to metabolic regulation and cellular response to zinc ion were recurrently upregulated. Supporting this finding, a previous study in BV-2 microglial cells reported that CBD upregulated zinc-related genes (\u003cem\u003eMt2\u003c/em\u003e, \u003cem\u003eNdrg1\u003c/em\u003e, \u003cem\u003eMmp23\u003c/em\u003e) and zinc transporters \u003cem\u003eSLC30A1\u003c/em\u003e and \u003cem\u003eSLC39A4\u003c/em\u003e while downregulating \u003cem\u003eSLC39A10\u003c/em\u003e and \u003cem\u003eZfp472\u003c/em\u003e (\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e). Another study in autoimmune T cells found that CBD suppressed pro-inflammatory genes and enhanced oxidative stress-related genes, including \u003cem\u003eSLC30A1\u003c/em\u003e (\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e). We now expand this finding to a much broader range of tissues and experimental conditions. Together, these results suggest that modulation of zinc homeostasis may be a key mechanism through which CBD exerts its antioxidant and anti-inflammatory effects. The cell cycle and cancer pathway were top downregulated pathways across CBD-associated datasets. Prior studies have shown that CBD can arrest the cell cycle in the G0/G1 phase and promote cell death in gastric cancer cell lines, such as SGC-7901 (\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e). Our study of around 50 datasets highlighting this pathway across studies further highlight the potential anti-cancer role of CBD in addition to its metabolic and immunomodulatory effects.\u003c/p\u003e \u003cp\u003eIn contrast, THC datasets exhibited minimal gene-level and pathway-level consistency in dataset clusters and globally, reinforcing its more variable and context-dependent effects. The weaker transcriptional stability of THC may stem from its complex pharmacodynamics, particularly its biphasic effects on neural and immune signaling. Our analysis revealed several diverse central and peripheral nervous system targets of THC that may ultimately lead to these downstream differences in pathway enrichment. For example, among the most consistently enriched DEGs induced by THC was \u003cem\u003eEGR2\u003c/em\u003e, which is associated with neuropathy and congenital hypomyelination of peripheral neurons (\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e). A number of THC target DEGs were also associated with synaptic signaling in the brain, and the onset of epilepsy and intellectual deficits, including \u003cem\u003eRPH3A\u003c/em\u003e (\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e), \u003cem\u003eKCNQ5\u003c/em\u003e (\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e), and \u003cem\u003eFRY\u003c/em\u003e (\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e). Further, we also found that \u003cem\u003ePER1\u003c/em\u003e was among our most consistently detected DEGs, which is a primary circadian pacemaker which has implications on human behavior and cognition (\u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e78\u003c/span\u003e). Some of these genes, such as \u003cem\u003eFRY\u003c/em\u003e and \u003cem\u003eEGR2\u003c/em\u003e, are associated with neuronal irregularities in region-specific targets (e.g. the cerebellum for \u003cem\u003eFRY\u003c/em\u003e and peripheral neurons for \u003cem\u003eEGR2\u003c/em\u003e). One possible explanation for the highly variable effects of THC lies in its high lipid solubility (\u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e79\u003c/span\u003e), which may influence its distribution across datasets depending on adipose content and tissue composition. Additionally, physiological barriers such as the blood-brain and blood-testicular barrier restrict THC accumulation in the brain and testes during acute exposure, and similar protective mechanisms may exist in other tissues as well (\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e80\u003c/span\u003e). These factors together may contribute to the heterogeneous effects of THC across individuals and biological contexts.\u003c/p\u003e \u003cp\u003eWe also observed heterogeneity in the brain network through KDA. While the CBD network based on DEGs from individual datasets had only 14 KDs, the THC network contained over 100 KDs. However, most of these THC KDs were significant in only one DEG dataset, further supporting the idea that THC induces region-specific effects in the brain. Additionally, KDA revealed that the molecular mechanisms activated in the brain differ between THC and CBD. Although both networks featured KDs involved in neuronal function and signal transduction, only two KDs, \u003cem\u003eSLC17A7\u003c/em\u003e and \u003cem\u003eTAOK1\u003c/em\u003e, were shared between them. \u003cem\u003eSLC17A7\u003c/em\u003e encodes vesicular glutamate transporter 1 (\u003cem\u003eVGLUT1\u003c/em\u003e), which facilitates glutamatergic neurotransmission (\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e). Cannabinoids have been shown to influence glutamatergic signaling: CBD reduces neuronal activation in VGLUT\u0026thinsp;+\u0026thinsp;neurons (\u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e82\u003c/span\u003e), while cannabinol upregulates genes associated with glutamatergic synaptic function (\u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e83\u003c/span\u003e). \u003cem\u003eTAOK1\u003c/em\u003e encodes a serine/threonin-protein that functions as a MAP kinase kinase kinase (MAP3K) and regulates MAPK signaling cascade (\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e). Although direct evidence linking cannabis treatment to \u003cem\u003eTAOK1\u003c/em\u003e regulation is lacking, THC has been shown to modulate microRNAs that target mRNAs of proteins involved in MAPK signaling, including \u003cem\u003eTAOK1\u003c/em\u003e (\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e85\u003c/span\u003e). The limited overlap in KDs between the two networks highlights the distinct molecular and cellular mechanisms through which THC and CBD exert their effects in the brain. This divergence is consistent with our broader findings that THC and CBD regulate largely non-overlapping gene sets and pathways across different tissue types.\u003c/p\u003e \u003cp\u003eWhile the endocannabinoid system (ECS) is widely recognized as a primary target of phytocannabinoids such as THC and CBD (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e), our multi-dataset analysis revealed that ECS-associated genes were regulated inconsistently across studies, both in direction and magnitude. These inconsistencies might reflect several underlying factors, including the tissue-specific expression of ECS components as well as distinct pharmacological profiles of THC and CBD. For example, in terms of binding affinity, THC acts as a partial agonist at both cannabinoid receptors, while CBD exhibits negligible affinity for cannabinoid receptor 1 (CB1) and functions as a partial agonist at cannabinoid receptor 2 (CB2) (\u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e86\u003c/span\u003e, \u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e87\u003c/span\u003e). Despite this gene-level variability, pathway-level enrichment analysis demonstrated that the ECS-related gene set were significantly enriched in several datasets, comparable in frequency to some of the top-ranking Gene Ontology Biological Process (GOBP) and KEGG pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB, \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC). Importantly, a clearer divergence between THC and CBD emerged when ECS enrichment was stratified by tissue type. ECS enrichment associated with THC was exclusively observed in brain-derived datasets, where CBD-associated ECS enrichment was distributed more evenly across both central and peripheral systems, including immune-related cell types and liver tissue. This divergence likely reflects differences in receptor affinity and expression patterns. CB1 is abundantly expressed in the brain and to a lesser extent in select peripheral tissues, whereas CB2 is predominantly found in immune cells and exhibits limited expression in the central nervous system (\u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e88\u003c/span\u003e). Thus, THC\u0026rsquo;s higher binding affinity to CB1 compared to CB2 explains its brain-focused transcriptomic effects (\u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e88\u003c/span\u003e), while CBD\u0026rsquo;s lack of CB1 binding and moderate activity at CB2 likely underlies its broader effects across peripheral and immune contexts. The observed significant ECS enrichment in CBD-treated immune and liver datasets aligns with this pharmacological profile.\u003c/p\u003e \u003cp\u003eTo further investigate the potential health implications of cannabis exposure, we performed Marker Set Enrichment Analysis (MSEA) to assess whether gene expression profiles associated with THC and CBD were enriched for genetic markers of human diseases. CBD-responsive genes showed significant enrichment for markers related to cholesterol and lipid metabolism traits, aligning with prior studies indicating that CBD can modulate lipid profiles and holds therapeutic potential for managing lipid disorders (\u003cspan citationid=\"CR89\" class=\"CitationRef\"\u003e89\u003c/span\u003e). CBD was also significantly associated with anthropometric traits, such as waist circumference, hip circumference, and height. A systematic review across multiple databases and registries reported that while cannabis use is generally associated with reductions in weight, waist circumference, and BMI, CBD specifically has been linked to increased body fat (\u003cspan citationid=\"CR90\" class=\"CitationRef\"\u003e90\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eBoth CBD and THC show association with multiple diseases, such as Schizophrenia, type 2 diabetes (T2D), and depressive symptoms. The relationship between cannabis use and depression is particularly complex and potentially bidirectional. Studies relying on self-reported depressive symptoms suggest mixed short-term effects of cannabis use: a minority (20%) with increased depression and a majority (64%) with decreased depression (\u003cspan citationid=\"CR91\" class=\"CitationRef\"\u003e91\u003c/span\u003e). However, a meta-analysis of 15 studies found that even a single THC administration induces depression and anxiety with large effect sizes (\u003cspan citationid=\"CR92\" class=\"CitationRef\"\u003e92\u003c/span\u003e). Conversely, extended cannabis abstinence has been associated with significant improvements of depressive symptoms (\u003cspan citationid=\"CR93\" class=\"CitationRef\"\u003e93\u003c/span\u003e). Longitudinal data also suggest a bidirectional or reinforcing relationship, as baseline depression has been significantly associated with increased THC use in e-cigarettes 12 months later (\u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e94\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCBD and other non-psychoactive cannabinoids have been investigated in human clinical trials for their potential therapeutic benefits in T2D and Schizophrenia. Indeed, a randomized clinical trial showed that CBD reduced circulating resistin, a hormone associated with insulin resistance, and increased glucose-dependent insulinotropic peptide (GIP), which plays a role in preserving pancreatic β-cell function in T2D patients (\u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e95\u003c/span\u003e). CBD has also demonstrated beneficial effects and a safety profile in Schizophrenia, where patients treated with CBD for six weeks demonstrated lower levels of positive psychotic symptoms (\u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e96\u003c/span\u003e). In another randomized clinical trial comparing CBD with the antipsychotic amisulpride, both treatments led to significant clinical improvement; however, CBD had a markedly superior side-effect profile (\u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e97\u003c/span\u003e). Thus, our MSEA results, supported by human clinical evidence, highlight the broad therapeutic potential of cannabinoids, especially CBD, in metabolic and psychiatric diseases and underscore the relevance of cannabinoid-responsive genes in the genetic architecture of complex traits.\u003c/p\u003e \u003cp\u003eHere, we have conducted a comprehensive, multispecies, multitissue investigation of the metabolic effects of THC and CBD. One of the primary strengths of our study was the use of highly heterogeneous datasets, with variations in experimental design, dosing regimens, and sample characteristics. This allowed us to consider cannabinoid exposure in a number of biological contexts, including in co-occurrence with diseases, on top of exposure to other relevant chemical agents, and in different tissues or regions within those tissues (e.g. several different brain regions were covered by our datasets). Meta-analysis across species allows us to consider the maximal amount of data from translatable model organisms so we can draw human-relevant conclusions. Despite the comprehensive nature of our analysis, however, some important limitations should be acknowledged. First, while we applied normalization techniques to minimize batch effects, residual variability may have influenced our findings. However, the same procedures were applied to the datasets from the reference chemicals, where coherent clustering and coherence across datasets were found. Therefore, our findings are less likely due to technical artifacts but are more likely the results of intrinsic activities of the compounds examined. Second, the sample sizes for cannabinoid studies tend to be small, with most groups consisting of only around three replicates. This limitation reflects the current state of the field and underscores the need for larger, more coordinated genomic studies. To mitigate the impact of limited sample size, we adopted a meta-analytic strategy to integrate signatures across studies, enhancing robustness through aggregated evidence. Third, the use of bulk transcriptomic data precludes cell type-specific resolution, which is critical for understanding the differential effects of cannabinoids on distinct cell populations. Future studies employing single-cell RNA sequencing or spatial transcriptomics, when datasets are available, could provide deeper insights into the cellular specificity of cannabinoid-induced gene expression changes. Additionally, while we focused on gene expression data, post-transcriptional and epigenetic modifications likely play a crucial role in cannabinoid-mediated effects. Integrating multi-omics approaches, including proteomics and metabolomics, could enhance our understanding of cannabinoid biology. Lastly, our study highlights the need for controlled, well-designed longitudinal studies to assess the long-term impact of THC and CBD exposure on human health.\u003c/p\u003e \u003cp\u003eIn summary, this study provides a comprehensive transcriptomic analysis of THC and CBD across species and tissue types. Our findings demonstrate that CBD exhibits greater transcriptional stability compared to THC, with more consistent effects on metabolic, cell cycle, and inflammatory pathways. In contrast, THC-induced gene expression changes are highly variable, complicating the identification of robust molecular signatures. These insights have important implications for the therapeutic use of cannabinoids and highlight the less predictive nature of the long-term biological effects of THC. As the legalization and medicinal use of cannabis continues to expand, a deeper understanding of its molecular mechanisms will be critical for optimizing its clinical applications while minimizing potential risks.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors received no specific funding for this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData are publicly available on the open-access Gene Expression Omnibus (GEO).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interest Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contribution Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eR.L., T.K., X.Y., and M.B. designed and directed the project. R.L, T.K., C.C, T.O., and E.N. curated the data. R.L., T.K., and S.Y. conducted the analysis and prepared figures. R.L., T.K., J.L.,\u0026nbsp;X.Y., and M.B. wrote the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to Participate Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Not applicable.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSmart R, Pacula RL. Early evidence of the impact of cannabis legalization on cannabis use, cannabis use disorder, and the use of other substances: Findings from state policy evaluations. Am J Drug Alcohol Abuse. 2019;45(6):644\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiech RA, Johnston LD, Patrick ME, O\u0026rsquo;Malley PM, Bachman JG, Schulenberg JE. Monitoring the Future National Survey Results on Drug Use, 1975\u0026ndash;2022: Secondary School Students. Inst Soc Res. 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMattingly DT, Richardson MK, Hart JL. Prevalence of and trends in current cannabis use among US youth and adults, 2013\u0026ndash;2022. Drug Alcohol Depend Rep. 2024;12:100253.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaroon J, Bost J. Review of the neurological benefits of phytocannabinoids. Surg Neurol Int. 2018;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePourseyed Lazarjani M, Torres S, Hooker T, Fowlie C, Young O, Seyfoddin A. Methods for quantification of cannabinoids: a narrative review. J Cannabis Res. 2020;2:1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePertwee RG. The pharmacology of cannabinoid receptors and their ligands: an overview. Int J Obes. 2006;30(1):S13\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePagano C, Navarra G, Coppola L, Avilia G, Bifulco M, Laezza C. Cannabinoids: therapeutic use in clinical practice. Int J Mol Sci. 2022;23(6):3344.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBatalla A, Bos J, Postma A, Bossong MG. The impact of cannabidiol on human brain function: a systematic review. Front Pharmacol. 2021;11:618184.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalabadi RB, Kolkar K, Chalannavar R. ∆9-Tetrahydrocannabinol (THC): The major psychoactive component is of botanical origin. Int J Innov Sci Res Rev. 2023;5(3):4177\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMachado Bergamaschi M, Helena Costa Queiroz R, Waldo Zuardi A, Crippa AS. Safety and side effects of cannabidiol, a Cannabis sativa constituent. Curr Drug Saf. 2011;6(4):237\u0026ndash;49.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSachs J, McGlade E, Yurgelun-Todd D. Safety and toxicology of cannabinoids. Neurotherapeutics. 2015;12(4):735\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZuo Y, Iemolo A, Montilla-Perez P, Li HR, Yang X, Telese F. Chronic adolescent exposure to cannabis in mice leads to sex-biased changes in gene expression networks across brain regions. Neuropsychopharmacology. 2022;47(12):2071\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePuighermanal E, Luna-S\u0026aacute;nchez M, Gella A, van der Walt G, Urpi A, Royo M, et al. Cannabidiol ameliorates mitochondrial disease via PPARγ activation in preclinical models. Nat Commun. 2024;15(1):7730.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShapira G, Israel-Elgali I, Grad M, Avnat E, Rachmany L, Sarne Y, et al. Hippocampal differential expression underlying the neuroprotective effect of delta-9-tetrahydrocannabinol microdose on old mice. Front Neurosci. 2023;17:1182932.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBilkei-Gorzo A, Albayram O, Draffehn A, Michel K, Piyanova A, Oppenheimer H, et al. A chronic low dose of ∆ 9-tetrahydrocannabinol (THC) restores cognitive function in old mice. Nat Med. 2017;23(6):782.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEdgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002;30(1):207\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, et al. NCBI GEO: archive for functional genomics data sets\u0026mdash;update. Nucleic Acids Res. 2012;41(D1):D991\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZamora Z, Wang S, Chen YW, Diamante G, Yang X. Systematic transcriptome-wide meta-analysis across endocrine disrupting chemicals reveals shared and unique liver pathways, gene networks, and disease associations. Environ Int. 2024;183:108339.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDiamante G, Cely I, Zamora Z, Ding J, Blencowe M, Lang J, et al. Systems toxicogenomics of prenatal low-dose BPA exposure on liver metabolic pathways, gut microbiota, and metabolic health in mice. Environ Int. 2021;146:106260.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Y, Shu L, Qiu Z, Lee DY, Settle SJ, Que Hee S, et al. Exposure to the BPA-substitute bisphenol S causes unique alterations of germline function. PLoS Genet. 2016;12(7):e1006223.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDavis S, Meltzer PS. GEOquery: a bridge between the Gene Expression Omnibus (GEO) and BioConductor. Bioinformatics. 2007;23(14):1846\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeinonen R, Sugawara H, Shumway M. International Nucleotide Sequence Database Collaboration. The sequence read archive. Nucleic Acids Res. 2010;39(suppl1):D19\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKrueger F. Trim Galore! A wrapper around Cutadapt and FastQC to consistently apply adapter and quality trimming to FastQ files, with extra functionality for RRBS data. Babraham Inst. 2015.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet J. 2011;17(1):10\u0026ndash;2.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatro R, Duggal G, Love MI, Irizarry RA, Kingsford C. Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods. 2017;14(4):417\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoneson C, Love MI, Robinson MD. Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences. F1000Research. 2016;4:1521.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRitchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLove MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:1\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStekhoven DJ, B\u0026uuml;hlmann P. MissForest\u0026mdash;non-parametric missing value imputation for mixed-type data. Bioinformatics. 2012;28(1):112\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKolde R, Laur S, Adler P, Vilo J. Robust rank aggregation for gene list integration and meta-analysis. Bioinformatics. 2012;28(4):573\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;14:1\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAshburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. Nat Genet. 2000;25(1):25\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDing J, Blencowe M, Nghiem T, Ha S, min, Chen YW, Li G, et al. Mergeomics 2.0: a web server for multi-omics data integration to elucidate disease networks and predict therapeutics. Nucleic Acids Res. 2021;49(W1):W375\u0026ndash;87.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShu L, Zhao Y, Kurt Z, Byars SG, Tukiainen T, Kettunen J, et al. Mergeomics: multidimensional data integration to identify pathogenic perturbations to biological systems. BMC Genomics. 2016;17:1\u0026ndash;16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13(11):2498\u0026ndash;504.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMacArthur J, Bowler E, Cerezo M, Gil L, Hall P, Hastings E, et al. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res. 2017;45(D1):D896\u0026ndash;901.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHevner RF, Shi L, Justice N, Hsueh YP, Sheng M, Smiga S, et al. Tbr1 regulates differentiation of the preplate and layer 6. Neuron. 2001;29(2):353\u0026ndash;66.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou J, Wade SD, Graykowski D, Xiao MF, Zhao B, Giannini LA, et al. The neuronal pentraxin Nptx2 regulates complement activity and restrains microglia-mediated synapse loss in neurodegeneration. Sci Transl Med. 2023;15(689):eadf0141.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePulkkinen V, Ezer S, Sundman L, Hagstr\u0026ouml;m J, Remes S, S\u0026ouml;derh\u0026auml;ll C, et al. Neuropeptide S receptor 1 (NPSR1) activates cancer-related pathways and is widely expressed in neuroendocrine tumors. Virchows Arch. 2014;465:173\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie W, Cappiello M, Meng M, Rosenthal R, Zhang W. ADRA2B deletion variant and enhanced cognitive processing of emotional information: A meta-analytical review. Neurosci Biobehav Rev. 2018;92:402\u0026ndash;16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHohoff C, Kroll T, Zhao B, Kerkenberg N, Lang I, Schwarte K, et al. ADORA2A variation and adenosine A1 receptor availability in the human brain with a focus on anxiety-related brain regions: modulation by ADORA1 variation. Transl Psychiatry. 2020;10(1):406.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShimizu T, Nakazawa M, Kani S, Bae YK, Shimizu T, Kageyama R, et al. Zinc finger genes Fezf1 and Fezf2 control neuronal differentiation by repressing Hes5 expression in the forebrain. Development. 2010;137(11):1875\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeide M, Zhang Y, Zhou X, Zhao T, Miquelaj\u0026aacute;uregui A, Varela-Echavarr\u0026iacute;a A, et al. Lhx5 controls mamillary differentiation in the developing hypothalamus of the mouse. Front Neuroanat. 2015;9:113.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ezahid Khan M, He L. The emerging role of GPR50 receptor in brain. Biomed Pharmacother. 2016;78:121\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSzymanowicz O, Drużdż A, Słowikowski B, Pawlak S, Potocka E, Goutor U, et al. A Review of the CACNA gene family: Its role in neurological disorders. Diseases. 2024;12(5):90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu HJ, Bondada S. Positive and negative roles of CD72 in B cell function. Immunol Res. 2002;25:155\u0026ndash;66.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScherer SW, Soder S, Duvoisin RM, Huizenga JJ, Tsui LC. The human metabotropic glutamate receptor 8 (GRM8) gene: a disproportionately large gene located at 7q31. 3\u0026ndash;q32. 1. Genomics. 1997;44(2):232\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAndrews GK. Cellular zinc sensors: MTF-1 regulation of gene expression. Zinc Biochem Physiol Homeost Recent Insights Curr Trends. 2001;37\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang P, Yang L, Guo Y, Qi S, Liang J, Tian G, et al. SENP3 mediates the activation of the Wnt/β-catenin signaling pathway to accelerate the growth and metastasis of oesophagal squamous cell carcinoma in mice. Funct Integr Genomics. 2024;24(2):40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRyu MS, Lichten LA, Liuzzi JP, Cousins RJ. Zinc transporters ZnT1 (Slc30a1), Zip8 (Slc39a8), and Zip10 (Slc39a10) in mouse red blood cells are differentially regulated during erythroid development and by dietary zinc deficiency. J Nutr. 2008;138(11):2076\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCasares L, Garc\u0026iacute;a V, Garrido-Rodr\u0026iacute;guez M, Mill\u0026aacute;n E, Collado JA, Garc\u0026iacute;a-Mart\u0026iacute;n A, et al. Cannabidiol induces antioxidant pathways in keratinocytes by targeting BACH1. Redox Biol. 2020;28:101321.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRen X, Feng C, Wang Y, Chen P, Wang S, Wang J, et al. SLC39A10 promotes malignant phenotypes of gastric cancer cells by activating the CK2-mediated MAPK/ERK and PI3K/AKT pathways. Exp Mol Med. 2023;55(8):1757\u0026ndash;69.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchulkens IA, Castricum K, Weijers EM, Koolwijk P, Griffioen AW, Thijssen VL. Expression, regulation and function of human metallothioneins in endothelial cells. J Vasc Res. 2014;51(3):231\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen YF, Wei YY, Yang CC, Liu CJ, Yeh LY, Chou CH, et al. miR-125b suppresses oral oncogenicity by targeting the anti-oxidative gene PRXL2A. Redox Biol. 2019;22:101140.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan Y, Xia H, Song Q. Research mapping of cannabinoids and endocannabinoid system in cancer over the past three decades: insights from bibliometric analysis. Front Pharmacol. 2025;16:1540619.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu J. Cannabis, cannabinoid receptors, and endocannabinoid system: Yesterday, today, and tomorrow. Acta Pharmacol Sin. 2019;40(3):297\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePi\u0026ntilde;ero J, Bravo \u0026Agrave;, Queralt-Rosinach N, Guti\u0026eacute;rrez-Sacrist\u0026aacute;n A, Deu-Pons J, Centeno E et al. DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants. Nucleic Acids Res. 2016;gkw943.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePi\u0026ntilde;ero J, Ram\u0026iacute;rez-Anguita JM, Sa\u0026uuml;ch-Pitarch J, Ronzano F, Centeno E, Sanz F, et al. The DisGeNET knowledge platform for disease genomics: 2019 update. Nucleic Acids Res. 2020;48(D1):D845\u0026ndash;55.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLowe H, Toyang N, Steele B, Bryant J, Ngwa W. The endocannabinoid system: a potential target for the treatment of various diseases. Int J Mol Sci. 2021;22(17):9472.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMassoumi R. CYLD: a deubiquitination enzyme with multiple roles in cancer. Future Oncol. 2011;7(2):285\u0026ndash;97.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNegroni A, Pierdomenico M, Cucchiara S, Stronati L. NOD2 and inflammation: current insights. J Inflamm Res. 2018;49\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElding H, Lau W, Swallow DM, Maniatis N. Dissecting the genetics of complex inheritance: linkage disequilibrium mapping provides insight into Crohn disease. Am J Hum Genet. 2011;89(6):798\u0026ndash;805.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003edel Bosque-Plata L, Mart\u0026iacute;nez-Mart\u0026iacute;nez E, Espinoza-Camacho M\u0026Aacute;, Gragnoli C. The role of TCF7L2 in type 2 diabetes. Diabetes. 2021;70(6):1220\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeedon M. The importance of TCF7L2. Diabet Med. 2007;24(10):1062\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZivotić I, Djurić T, Stanković A, Milasinovic D, Stankovic G, Dekleva M, et al. CDKN2B gene expression is affected by 9p21. 3 rs10757278 in CAD patients, six months after the MI. Clin Biochem. 2019;73:70\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartin BR. Cellular effects of cannabinoids. Pharmacol Rev. 1986;38(1):45\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAl-Khazaleh AK, Zhou X, Bhuyan DJ, M\u0026uuml;nch GW, Al-Dalabeeh EA, Jaye K, et al. The neurotherapeutic arsenal in Cannabis sativa: Insights into anti-neuroinflammatory and neuroprotective activity and potential entourage effects. Molecules. 2024;29(2):410.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAtalay S, Jarocka-Karpowicz I, Skrzydlewska E. Antioxidative and anti-inflammatory properties of cannabidiol. Antioxidants. 2019;9(1):21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEkiner SA, Gęgotek A, Skrzydlewska E. The molecular activity of cannabidiol in the regulation of Nrf2 system interacting with NF-κB pathway under oxidative stress. Redox Biol. 2022;57:102489.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaswan SM, Klosner AE, Glynn K, Rajgopal A, Malik K, Yim S et al. Therapeutic potential of cannabidiol (CBD) for skin health and disorders. Clin Cosmet Investig Dermatol. 2020;927\u0026ndash;42.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJuknat A, Rimmerman N, Levy R, Vogel Z, Kozela E. Cannabidiol affects the expression of genes involved in zinc homeostasis in BV-2 microglial cells. Neurochem Int. 2012;61(6):923\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKozela E, Juknat A, Gao F, Kaushansky N, Coppola G, Vogel Z. Pathways and gene networks mediating the regulatory effects of cannabidiol, a nonpsychoactive cannabinoid, in autoimmune T cells. J Neuroinflammation. 2016;13:1\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePagano C, Navarra G, Coppola L, Bifulco M, Laezza C. Molecular mechanism of cannabinoids in cancer progression. Int J Mol Sci. 2021;22(7):3680.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNagarajan R, Svaren J, Le N, Araki T, Watson M, Milbrandt J. EGR2 mutations in inherited neuropathies dominant-negatively inhibit myelin gene expression. Neuron. 2001;30(2):355\u0026ndash;68.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePavinato L, Stanic J, Barzasi M, Gurgone A, Chiantia G, Cipriani V, et al. Missense variants in RPH3A cause defects in excitatory synaptic function and are associated with a clinically variable neurodevelopmental disorder. Genet Med. 2023;25(11):100922.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei AD, Wakenight P, Zwingman TA, Bard AM, Sahai N, Willemsen MH, et al. Human KCNQ5 de novo mutations underlie epilepsy and intellectual disability. J Neurophysiol. 2022;128(1):40\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaulraj P, Bosworth M, Longhurst M, Hornbuckle C, Gotway G, Lamb AN, et al. A novel homozygous deletion within the FRY gene associated with nonsyndromic developmental delay. Cytogenet Genome Res. 2019;159(1):19\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLim AS, Chang A, Shulman JM, Raj T, Chibnik LB, Cain SW, et al. A common polymorphism near PER1 and the timing of human behavioral rhythms. Ann Neurol. 2012;72(3):324\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChayasirisobhon S. Mechanisms of action and pharmacokinetics of cannabis. Perm J. 2020;25:19\u0026ndash;200.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuestis MA. Human cannabinoid pharmacokinetics. Chem Biodivers. 2007;4(8):1770.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReimer RJ. SLC17: a functionally diverse family of organic anion transporters. Mol Aspects Med. 2013;34(2\u0026ndash;3):350\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNedelescu H, Wagner GE, De Ness GL, Carroll A, Kerr TM, Wang J, et al. Cannabidiol produces distinct U-shaped dose-response effects on cocaine-induced conditioned place preference and associated recruitment of prelimbic neurons in male rats. Biol Psychiatry Glob Open Sci. 2022;2(1):70\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTrainito A, Muscar\u0026agrave; C, Gugliandolo A, Chiricosta L, Salamone S, Pollastro F, et al. Cannabinol (CBN) Influences the Ion Channels and Synaptic-Related Genes in NSC-34 Cell Line: A Transcriptomic Study. Cells. 2024;13(18):1573.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFang CY, Lai TC, Hsiao M, Chang YC. The diverse roles of TAO kinases in health and diseases. Int J Mol Sci. 2020;21(20):7463.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSimon L, Song K, Vande Stouwe C, Hollenbach A, Amedee A, Mohan M, et al. ∆9-Tetrahydrocannabinol (∆9-THC) promotes neuroimmune-modulatory microRNA profile in striatum of simian immunodeficiency virus (SIV)-infected macaques. J Neuroimmune Pharmacol. 2016;11:192\u0026ndash;213.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZou S, Kumar U. Cannabinoid receptors and the endocannabinoid system: signaling and function in the central nervous system. Int J Mol Sci. 2018;19(3):833.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShahbazi F, Grandi V, Banerjee A, Trant JF. Cannabinoids and cannabinoid receptors: The story so far. IScience. 2020;23(7).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSch\u0026ouml;nke M, Martinez-Tellez B, Rensen PC. Role of the endocannabinoid system in the regulation of the skeletal muscle response to exercise. Curr Opin Pharmacol. 2020;52:52\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWiciński M, Fajkiel-Madajczyk A, Kurant Z, Gryczka K, Kurant D, Szambelan M, et al. The Use of Cannabidiol in Metabolic Syndrome\u0026mdash;An Opportunity to Improve the Patient\u0026rsquo;s Health or Much Ado about Nothing? J Clin Med. 2023;12(14):4620.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReis MG, Ferreira AJ, Sohouli MH, Taimeir\u0026atilde;o DR, Vieira RAL, Guimar\u0026atilde;es NS. Effect of cannabis and subproducts on anthropometric measures: a systematic review and meta-analysis. Int J Obes. 2024;48(1):44\u0026ndash;54.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi X, Diviant JP, Stith SS, Brockelman F, Keeling K, Hall B, et al. The effectiveness of cannabis flower for immediate relief from symptoms of depression. Yale J Biol Med. 2020;93(2):251.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHindley G, Beck K, Borgan F, Ginestet CE, McCutcheon R, Kleinloog D, et al. Psychiatric symptoms caused by cannabis constituents: a systematic review and meta-analysis. Lancet Psychiatry. 2020;7(4):344\u0026ndash;53.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLucatch AM, Kloiber SM, Meyer JH, Rizvi SJ, George TP. Effects of extended cannabis abstinence in major depressive disorder. Can J Addict. 2020;11(3):33\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClendennen SL, Smith J, Sumbe A, Chen B, Wilkinson AV, Harrell MB. Symptoms of depression and anxiety and subsequent use of nicotine and THC in electronic cigarettes. Subst Use Misuse. 2023;58(5):591\u0026ndash;600.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJadoon KA, Ratcliffe SH, Barrett DA, Thomas EL, Stott C, Bell JD, et al. Efficacy and Safety of Cannabidiol and Tetrahydrocannabivarin on Glycemic and Lipid Parameters in Patients With Type 2 Diabetes: A Randomized, Double-Blind, Placebo-Controlled, Parallel Group Pilot Study. Diabetes Care. 2016;39(10):1777\u0026ndash;86.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcGuire P, Robson P, Cubala WJ, Vasile D, Morrison PD, Barron R, et al. Cannabidiol (CBD) as an adjunctive therapy in schizophrenia: a multicenter randomized controlled trial. Am J Psychiatry. 2018;175(3):225\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeweke F, Piomelli D, Pahlisch F, Muhl D, Gerth C, Hoyer C, et al. Cannabidiol enhances anandamide signaling and alleviates psychotic symptoms of schizophrenia. Transl Psychiatry. 2012;2(3):e94\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-cannabis-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jcan","sideBox":"Learn more about [Journal of Cannabis Research](https://jcannabisresearch.biomedcentral.com/)","snPcode":"42238","submissionUrl":"https://submission.springernature.com/new-submission/42238/3","title":"Journal of Cannabis Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"delta-9-tetrahydrocannabinol, THC, cannabidiol, CBD, endocannabinoids system, transcriptomics, cannabinoids, mammalian, multispecies analysis, cross-tissue analysis ","lastPublishedDoi":"10.21203/rs.3.rs-6857614/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6857614/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e Cannabis use is on the rise yet the systematic molecular impact of key cannabinoid components on various tissues in diverse organisms remains incompletely understood. We aim to systematically elucidate the molecular pathways and networks affected by delta-9-tetrahydrocannabinol (THC) and cannabidiol (CBD) across species and tissue types.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e We curated 105 THC- and CBD-related RNA sequencing (RNAseq) and microarray datasets from Gene Expression Omnibus (NCBI GEO) with a focus on mammalian species (human, non-human primate rhesus macaque, mouse, rat). Differentially expressed genes (DEGs) were identified using \u003cem\u003elimma\u003c/em\u003e for microarrays and \u003cem\u003eDESeq2\u003c/em\u003e for RNAseq data. DEGs were analyzed for pathway enrichment using EnrichR, network regulation using Mergeomics key driver analysis, and disease associations using Mergeomics Marker Set Enrichment Analysis. Comparative analyses were conducted across compounds, datasets, species, and tissues.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eCBD transcriptomic signatures demonstrated greater stability and consistency across species and experimental conditions compared to THC. CBD datasets clustered more tightly by route of administration and species and were more frequently enriched for pathways related to zinc homeostasis, inflammation suppression, and cell cycle regulation. In contrast, THC signatures were more heterogeneous and did not exhibit consistent clustering, although a small number of consistently altered genes associated with antioxidant activity, neuronal myelination, and synaptic signaling were identified across datasets. THC altered endocannabinoid signaling genes more often in brain tissues while CBD affected this pathway more heavily in both central and peripheral tissues. Disease enrichment analyses revealed significant associations of CBD DEGs with lipid metabolism and body composition traits, while DEGs of both compounds showed links to neuropsychiatric disorders and type 2 diabetes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e THC and CBD demonstrated distinct and largely non-overlapping transcriptomic responses, with CBD showing more coherent molecular effects across biological systems. Our results underscore the potential therapeutic relevance of CBD to metabolic and psychiatric regulation, highlight the variability of THC’s molecular actions, and offer molecular insights into the therapeutic and side effects of cannabinoids.\u003c/p\u003e","manuscriptTitle":"Comparative analysis of 105 datasets across species and tissues reveals higher variability in transcriptomic responses to THC than CBD","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-13 14:45:16","doi":"10.21203/rs.3.rs-6857614/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-09-15T09:36:45+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-15T09:26:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"186180555779747432160154845571381307272","date":"2025-09-15T09:20:41+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-05T21:27:33+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"164619097707899206655923702168653321472","date":"2025-07-30T12:29:52+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-11T16:51:14+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-11T08:30:29+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-11T08:28:33+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Cannabis Research","date":"2025-06-09T22:37:15+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-cannabis-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jcan","sideBox":"Learn more about [Journal of Cannabis Research](https://jcannabisresearch.biomedcentral.com/)","snPcode":"42238","submissionUrl":"https://submission.springernature.com/new-submission/42238/3","title":"Journal of Cannabis Research","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8d153875-2049-4efb-91a9-e9a5cafa75dc","owner":[],"postedDate":"June 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-22T16:01:22+00:00","versionOfRecord":{"articleIdentity":"rs-6857614","link":"https://doi.org/10.1186/s42238-025-00361-0","journal":{"identity":"journal-of-cannabis-research","isVorOnly":false,"title":"Journal of Cannabis Research"},"publishedOn":"2025-12-16 15:57:40","publishedOnDateReadable":"December 16th, 2025"},"versionCreatedAt":"2025-06-13 14:45:16","video":"","vorDoi":"10.1186/s42238-025-00361-0","vorDoiUrl":"https://doi.org/10.1186/s42238-025-00361-0","workflowStages":[]},"version":"v1","identity":"rs-6857614","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6857614","identity":"rs-6857614","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}