Skip to main content
NCBI home page
HHS Author Manuscripts logoLink to HHS Author Manuscripts
. Author manuscript; available in PMC: 2025 Dec 30.
Published before final editing as: Arch Toxicol. 2025 Nov 22:10.1007/s00204-025-04236-4. doi: 10.1007/s00204-025-04236-4

Mechanistic study of the cytotoxicity of cannabidiol and its metabolites in HepG2 cells

Si Chen 1,*, Yuxi Li 1, Montserrat Puig 2, Frederic Moulin 3, Supratim Choudhuri 4, Jeremy Gingrich 4, Lei Guo 1
PMCID: PMC12746337  NIHMSID: NIHMS2125271  PMID: 41272177

Abstract

The cannabidiol (CBD)-based drug, Epidiolex, received approval from the U.S. Food and Drug Administration (FDA) for treating seizures in certain childhood-onset epileptic disorders. CBD-associated liver toxicity is a serious side effect listed on the drug label. Our previous studies demonstrated cytotoxicity in primary human hepatocytes and HepG2 cells induced by CBD, 7-hydroxy-CBD, and 7-carboxy-CBD, with cell cycle disturbances, endoplasmic reticulum (ER) stress, and apoptosis identified as the underlying mechanisms. In this study, using a transcriptomic approach with mRNA-sequencing analysis, we found that downregulation of genes associated with oxidative phosphorylation and upregulation of genes associated with mitochondrial dysfunction, autophagy, and ER stress were among the top 10 canonical pathways consistently affected across different CBD concentrations. Direct measurement of the activity of the mitochondrial respiratory complexes that compose the oxidative phosphorylation process, demonstrated that CBD strongly inhibited Complexes IV and V and moderately inhibited Complexes II and III. CBD-induced mitochondrial dysfunction was indicated by a classic glucose-galactose assay, and the loss of mitochondrial membrane potential was confirmed by a JC-1 assay. Additionally, CBD induced autophagy, as evidenced by autophagosome formation and enhanced autophagic flux. Similar to CBD, 7-hydroxy-CBD induced a strong inhibition of Complexes IV and V, mitochondrial dysfunction, and autophagy, while 7-carboxy-CBD induced autophagy, with marginal inhibition on the respiratory complexes and no identified mitochondrial dysfunction. In summary, autophagy emerges as the common mechanism underlying CBD-, 7-hydroxy-CBD-, and 7-carboxy-CBD-induced cytotoxicity. Inhibition of mitochondrial respiratory complexes and mitochondrial dysfunction were observed with CBD and 7-hydroxy-CBD, but not with 7-carboxy-CBD.

Keywords: cannabidiol, CBD, 7-hydroxy-CBD, 7-carboxy-CBD, liver toxicity, RNA-sequencing, ER stress, mitochondrial dysfunction, mitochondria respiratory complexes, autophagy

Introduction

Cannabidiol (CBD), a prominent cannabinoid in the plant Cannabis sativa L, has recently gained attention due to its potential therapeutic effects. The U.S. Food and Drug Administration (FDA) approved Epidiolex, a CBD-based drug, for treating seizures associated with Lennox-Gastaut and Dravet syndromes in patients two years and older (FDA 2018) and for treating seizures associated with tuberous sclerosis complex in patients one year and older (FDA 2020). Apart from CBD’s approved medical applications, CBD is found in numerous non-pharmaceutical consumer products. Despite growing public interest in CBD products, concerns persist regarding unproven claims of beneficial effects (FDA 2019a; FDA 2021), along with potential adverse effects, including alterations in organ weight, developmental and reproductive toxicities, liver injury, immune system suppression, and interference with liver metabolizing enzymes and drug transport proteins (Chen et al. 2024a; Chen et al. 2024b; FDA 2019b; Gingrich et al. 2023; Li et al. 2023a; Li et al. 2025a; Li et al. 2023b; Li et al. 2025b; Li et al. 2024; Li et al. 2022b).

Animal studies in multiple species (mice, rats, and dogs) have indicated that repeated oral administration of CBD induces liver changes, including hepatocellular hypertrophy and elevated levels of alanine aminotransferase (ALT), alkaline phosphatase (ALP), and aspartate aminotransferase (AST) (FDA 2017). Clinically, elevated levels of ALT and AST have also been reported in patients with epilepsy and healthy adults following the use of CBD, either alone or in combination with anti-epileptic drugs, such as valproate or clobazam (Lo et al. 2023; Szaflarski et al. 2018; Taylor et al. 2020; Watkins et al. 2021). Liver problems are listed as a serious side effect on the Epidiolex drug label (EPIDIOLEX® 2023). While these findings suggest the liver may be a target of CBD toxicity, the underlying mechanisms of CBD-induced liver toxicity remain unclear. In addition, while substantial amounts of 7-hydroxy-CBD and 7-carboxy-CBD have been found in animals and humans (FDA 2017; Taylor et al. 2018), less is known about their hepatotoxicity profiles and potential mechanisms of toxicity.

In our previous studies, we demonstrated the cytotoxic effects of CBD, 7-hydroxy-CBD, and 7-carboxy-CBD in both primary human hepatocytes and HepG2 cells. These compounds induced cell cycle disruption, apoptosis, and ER stress in HepG2 cells (Chen et al. 2024a; Li et al. 2025b). In this study, we employed a transcriptomic approach with mRNA-sequencing analysis to investigate comprehensively the changes in signaling pathways associated with CBD-induced cytotoxicity in HepG2 cells and validated the concentration-dependence of these pathway alterations using a series of biochemical and morphological assays. Further, we explored whether cytotoxicity by 7-hydroxy-CBD and 7-carboxy-CBD occurred through the same mechanisms as CBD.

Materials and methods

Test Chemicals

CBD (CAS # 13956-29-1, Batch # NQSS1951, stated purity 100%) was purchased from Purisys™ (Athens, GA). 7-Carboxy-CBD (CAS # 63958-77-0, Batch # BDG11603, stated purity 97.4%) and 7-hydroxy-CBD (CAS # 50725-17-2, Batch # BDG17881.2, stated purity 99.7%) were purchased from BDG Synthesis (Wellington, New Zealand). The identity and purity were confirmed by in-house 1H-nuclear magnetic resonance and mass spectral analyses. Dimethyl sulfoxide (DMSO, Catalog # D8418) was purchased from MilliporeSigma (St. Louis, MO) and used as the vehicle. CBD, 7-hydroxy-CBD, and 7-carboxy-CBD were dissolved in DMSO and stored as 100 mM solutions at −20 °C.

Cell culture

The HepG2 cell line was purchased from the American Type Culture Collection (ATCC, Catalog # HB-8065™, Manassas, VA). The HepG2-GFP-LC3B stable cell line was established previously (Chen et al. 2014). Both cell lines were maintained in Williams’ Medium E complete media (Catalog # W4125, Millipore Sigma) supplemented with 10% fetal bovine serum (FBS, Catalog # S11150, R&D systems, Minneapolis, MN), 100 units/mL penicillin, and 100 μg/mL streptomycin (Catalog # 15140-122, Gibco, Waltham, MA). The passage number was kept below 10.

RNA isolation and mRNA sequencing analysis

HepG2 cells were seeded at a density of 8.3 × 105 cells/well in 6-well plates (Catalog # 3506, Corning, Glendale, AZ) for 24 h prior to treatment with DMSO or 10 – 50 μM CBD for 24 h. Total RNA was extracted from the cells using a RNeasy Mini kit (Catalog # 74106, Qiagen, Germantown, MD). The quality and purity of RNA were measured using a NanoDrop™ 8000 (ThermoFisher Scientific), RNA 6000 Nano kits (Catalog # NC1783726, Agilent Technologies, Santa Clara, CA), and an Agilent 2100 Bioanalyzer (Agilent Technologies). RNA samples with the RNA integrity number > 9 were sent to Qiagen for subsequent mRNA-sequencing analyses. Library preparation was conducted using the QIAseq Stranded mRNA Kit (Qiagen). Sequencing was carried out on the Illumina NovaSeq platform using paired-end 75 bp reads (2×75), achieving a depth of approximately 45 million reads per sample. The resulting mRNA expression profiles were generated using Perl scripts, Cutadapt, HISAT2, and StringTie. Subsequently, the raw gene count matrix was filtered to remove genes with counts ≤ 5 using RStudio 2023.03.0. The resulting 13,833 genes were subjected to normalization using DESeq2 before conducting the differential gene expression analysis with DESeq2 (Anders and Huber 2010). The differentially expressed genes (DEGs) were defined as having an adjusted p-value < 0.01 and absolute fold-change ≥ 2. DEGs obtained from 5 different pair comparisons, including DMSO versus 10, 20, 30, 40, and 50 μM CBD, were input into the Qiagen Ingenuity Pathway Analysis (IPA) for comparison analysis. The IPA’s function allows for the identification of pathways that are consistently affected across different treatment concentrations of CBD by comparing pathway Z-scores. IPA also predicts the direction of pathway changes, i.e., upregulated, downregulated, or unaffected as determined by Z-score (a statistical measure). A Z-score was used to assess the correlations between gene expression data and biological pathways or functions reached statistical significance and was calculated by IPA by comparing the actual expression level to the expected pattern based on previous literature observations. Z-scores with absolute values ≥ 2 and p value < 0.05 were considered statistically significant. In our presentation in this work, blue colors represent down-regulation, and red colors represent up-regulation.

Measurement of mitochondrial Complexes I – V activities

When measuring the activities of mitochondrial complexes, the highest concentrations of CBD and 7-hydroxy-CBD used did not exceed 125 μM because of poor solubility in the complexes’ activity solution (visible precipitates formed at concentrations above this threshold). Therefore, the concentration range for CBD and 7-hydroxy-CBD used was 0.98 – 125 μM. In contrast, no visible precipitates of 7-carboxy-CBD were observed, even at a high concentration of 1000 μM in any of the complexes’ activity solutions, allowing for a greater concentration range of 0.98 – 1000 μM. Bayesian benchmark dose (BMD) analysis (Shao and Shapiro 2018) was utilized to determine the concentration of CBD, 7-hydroxy-CBD, and 7-carboxy-CBD that caused a 50% relative inhibition in complex activities (i.e., BMC50).

To measure the activities of mitochondrial complexes, assay kits were purchased from Abcam (Cambridge, MA), including a MitoTox™ Complex I OXPHOS Activity Assay kit (Catalog # ab109903), a MitoTox™ Complex II OXPHOS Activity Assay kit (Catalog # ab109904), a MitoTox™ Complex III OXPHOS Activity Assay kit (Catalog # ab109905), a MitoTox™ Complex IV OXPHOS Activity Assay kit (Catalog # ab109906), and a MitoTox™ Complex V OXPHOS Activity Assay kit (Catalog # ab109907). The kits were used according to the manufacturer’s protocols as detailed in our previous publication (Chen et al. 2025).

JC-1 assay

Mitochondrial membrane potential was measured using the dual fluorescence dye JC-1 (Catalog # T3168, ThermoFisher Scientific). Cells were seeded at a density of 1.7 × 106 cells per 60-mm dish (Catalog # 628160, Greiner Bio-One GmbH, Frickenhausen, Germany) and incubated for 24 h before treatment with various concentrations of CBD, 7-hydroxy-CBD, 7-carboxy-CBD, or DMSO. After 24 h of treatment, the cells were collected, resuspended in serum-free Williams’ Medium E, and transferred to 5 mL Eppendorf tubes. The cell number of each sample was adjusted to 1 × 106 before centrifugation at 500 g for 5 min to obtain a pellet. Each pellet was resuspended in 1 mL of serum-free Williams’ Medium E containing 2 μM JC-1 dye and incubated at 37°C in darkness for 20 min. Unbound JC-1 dye was removed by adding 3 mL of 1 × phosphate-buffered saline (PBS) to each tube, followed by centrifugation at 500 g for 5 min. Cells were then resuspended in 0.5 mL of 1 × PBS, filtered through a 35 μm nylon mesh, and transferred to Falcon round-bottom polystyrene test tubes (Catalog # 352235, Corning). Samples were immediately analyzed using a FACSCanto™ flow cytometer with FACSDiva™ software (BD Biosciences, San Jose, CA) and processed with FlowJo® software. Cell distribution was visualized using two-dimensional dot plots. JC-1 monomers were detected at an excitation wavelength of 485 nm and an emission wavelength of 535 nm, while JC-1 aggregates were detected at an excitation wavelength of 530 nm and an emission wavelength of 590 nm. Mitochondrial membrane potential was calculated as the ratio of JC-1 aggregates to monomers.

Glucose-galactose assay

Glucose-galactose assay is a commonly used method for detecting mitochondrial dysfunction. HepG2 cells were seeded at a density of 2.5 × 104 cells per well in 96-well plates (Catalog # 3585, Corning) and cultured for 24 h in Williams’ Medium E complete media before treatment with DMSO or test compounds at designated concentrations in either glucose or galactose culture media. Both glucose and galactose culture media were freshly prepared for the glucose-galactose assay. To prepare 100 mL of glucose media, 86 mL of glucose-free Dulbecco’s Modified Eagle Medium (DMEM; Catalog #11966-025, Gibco) was supplemented with 1 mL of 100× HEPES (Catalog #15630-080, Gibco), 1 mL of 100× sodium pyruvate (Catalog #11360-070, Gibco), 1 mL of 100× penicillin-streptomycin (Catalog #15140-122, Gibco), 10 mL of fetal bovine serum (FBS), and 1 mL of 2.5 M glucose (Catalog #G5767, MilliporeSigma), which was dissolved in glucose-free DMEM and sterilized by 0.22 μm membrane filtration. To prepare 100 mL of galactose media, the same components were used, except that 1 mL of 1 M galactose (Catalog #G5388, MilliporeSigma), prepared and sterilized similarly, was used instead of glucose. Cell viability was assessed using the CellTiter-Blue Cell Viability Assay kit (Catalog #G8082, Promega). Following compounds or DMSO treatment, supernatants were aspirated, and 10 μL of CellTiter-Blue reagent, pre-mixed with 90 μL of serum-free Williams’ Medium E, was added to each well. After a 1 h incubation in a tissue culture incubator, fluorescence was measured using a Cytation 5 Multi-Mode Microplate Reader (excitation: 560 nm; emission: 590 nm). Background fluorescence, determined from a set of cell-free wells, were subtracted from the sample readings. Cell viability was calculated by normalizing the fluorescence of treated cells to that of the DMSO control. Cytotoxicity was assessed using the LDH assay as described previously (Li et al. 2024).

Western blot analysis

HepG2 or HepG2-GFP-LC3B cells were seeded at a density of 1.7 × 106 cells per 60-mm dish and cultured for 24 h before treatment with various concentrations of CBD, 7-hydroxy-CBD, 7-carboxy-CBD, or DMSO. After 24 h of treatment, whole-cell lysates were extracted using RIPA buffer (Catalog # 89901, ThermoFisher Scientific) supplemented with Halt™ Protease Inhibitor Cocktail (Catalog # 1861279, ThermoFisher Scientific). Protein concentrations were determined using a Bio-Rad Protein assay (Catalog # 5000006, Bio-Rad, Hercules, CA), and samples were diluted in 4 × Laemmli Sample Buffer (Catalog # 1610747, Bio-Rad) containing 2-mercaptoethanol (Catalog # 1610710, Bio-Rad). The samples were then heated at 95 °C for 10 min to ensure complete protein denaturation. A total of 20 μg of protein was loaded onto a gel, separated by SDS-PAGE, and transferred to a PVDF membrane (Catalog # 1620177XTU, Bio-Rad) for Western blot analysis. Primary antibodies were used to probe LC3B (Catalog # L7543, MilliporeSigma), GFP (Catalog # sc-8334, Santa Cruz Biotechnology, Santa Cruz, CA), and GAPDH (Catalog # 5174, Cell Signaling Technology, Danvers, MA). A secondary mouse anti-rabbit IgG-horseradish peroxidase (HRP) antibody (Catalog # sc-2357, Santa Cruz Biotechnology) was use for detection. Protein band signals were captured using the FluroChem E System and quantified using AlphaView software (ProteinSimple, San Jose, CA).

Measurement of GFP-LC3B puncta formation and total GFP intensity

GFP-LC3B puncta formation and total GFP intensity were used to monitor autophagosome formation and autophagic flux. HepG2-GFP-LC3B cells were seeded at a density of 2.5 × 104 cells per well in 18-well chambered glass coverslips (Catalog # 81817, ibidi GmbH, Martinsried, Planegg, Germany) and cultured for 24 h before treatment with various concentrations of CBD, 7-hydroxy-CBD, 7-carboxy-CBD, or DMSO. Green fluorescence puncta (small, bright spots) and total GFP-LC3B expression were measured following the same protocols as described in our previous study (Chen et al. 2014).

Statistical analyses

Data are expressed as the mean ± standard deviation (S.D.) from at least three independent experiments. Analyses were conducted using GraphPad Prism 9 (GraphPad Software, San Diego, CA). Statistical significance was assessed using one-way analysis of variance (ANOVA) followed by Dunnett’s test for pairwise-comparisons, or two-way ANOVA followed by the Bonferroni post-test. A p-value of less than 0.05 was considered statistically significant.

Results

Analysis of consistently affected canonical pathways and differentially expressed genes (DEGs) in CBD treated HepG2 cells.

Our previous study demonstrated that cell cycle disturbances and apoptosis were the most notable phenotypes, and endoplasmic reticulum (ER) stress was identified as one of the molecular mechanisms contributing to CBD-induced cytotoxicity (Chen et al. 2024a). In this study, to investigate comprehensively the underlying signaling pathways involved in CBD-induced hepatotoxicity, HepG2 cells were treated with CBD at 10 – 50 μM or DMSO, total RNA was isolated, and gene expression profiles were determined by mRNA sequencing. Based on results from DEseq2 analyses, 21, 259, 3748, 4780, and 5780 DEGs were identified from 10, 20, 30, 40, and 50 μM CBD-treated HepG2 cells compared to DMSO-treated cells. Based on these identified DEGs, we applied IPA analyses to identify pathways that demonstrated consistent trends across different concentrations of CBD treatment. Figure 1A and Table 1 show the top 10 canonical pathways consistently affected across concentrations of CBD, ranked by absolute z-scores. A list of genes altered in these top 10 pathways is summarized in Supplemental Table 1.

Figure 1. Transcriptomic analysis of CBD-treated HepG2 cells.

Figure 1.

Open in a new tab

Total RNA was extracted from HepG2 cells after a 24-h treatment with DMSO, or 10 – 50 μM CBD. Gene expression profiles (n = 4/group) were analyzed using mRNA-sequencing. (A) Heatmap illustrating the top 10 canonical pathways consistently affected across CBD concentrations, ranked by absolute z-scores. (B – E) Heatmaps depicting alternations in the top 20 genes associated with (B) oxidative phosphorylation, (C) mitochondria dysfunction, (D) unfolded protein response, and (E) autophagy.

Table 1.

Z-scores of the top 10 canonical pathways consistently affected across different concentrations of CBD.

Rank# Canonical Pathways 0 VS 10 μM CBD 0 VS 20 μM CBD 0 VS 30 μM CBD 0 VS 40 μM CBD 0 VS 50 μM CBD
1 Oxidative phosphorylation Z-score a N/A N/A −6.557 −6.091 −6.008
P-value b N/A N/A 3.49e-11 6.65e-07 2.36e-04
Ratio c N/A N/A 44/111 42/111 41/111
2 DHCR24 signaling pathway Z-score N/A N/A −4.522 −5.078 −4.128
P-value N/A N/A 2.32e-09 3.46e-10 1.17e-03
Ratio N/A N/A 47/137 56/137 46/137
3 LXR/RXR activation Z-score N/A N/A −4.012 −5.000 −4.111
P-value N/A N/A 5.25e-13 1.96e-12 5.87e-06
Ratio N/A N/A 50/123 56/123 49/123
4 Autophagy Z-score N/A N/A 3.207 3.969 4.756
P-value N/A N/A 3.59e-06 1.45e-04 1.41e-06
Ratio N/A N/A 56/216 61/216 78/216
5 Unfolded protein response Z-score N/A 2.000 3.273 3.674 3.900
P-value N/A 1.18e-02 8.46e-05 5.77e-05 1.81e-07
Ratio N/A 4/90 27/90 32/90 42/90
6 Xenobiotic metabolism PXR signaling pathway Z-score N/A 0.447 −3.576 −4.389 −4.296
P-value N/A 4.14e-02 2.62e-08 7.34e-06 3.34e-03
Ratio N/A 5/193 57/193 60/193 59/193
7 Mitochondrial dysfunction Z-score N/A 0 4.538 3.796 4.274
P-value N/A 4.34e-01 1.09e-10 3.86e-06 4.88e-05
Ratio N/A 4/344 94/344 96/344 107/344
8 Superpathway of cholesterol biosynthesis Z-score N/A N/A −4.123 −3.873 −3.464
P-value N/A N/A 3.41e-08 4.07e-05 5.14e-03
Ratio N/A N/A 17/29 15/29 13/29
9 SPINK1 general cancer pathway Z-score −2.828 −2.530 −2.236 −1.964 −1.789
P-value 6.60e-16 1.34e-09 4.02e-04 3.86e-03 6.42E-02
Ratio 8/69 10/69 21/69 22/69 21/69
10 Stearate biosynthesis I Z-score N/A N/A −3.771 −3.674 −3.800
P-value N/A N/A 1.10e-06 3.29e-09 1.85e-06
Ratio N/A N/A 26/69 34/69 33/69
Open in a new tab
a.

In IPA, the Z-score is a statistical measure used to assess the significance of correlations between gene expression data and biological pathways or functions. It is calculated by comparing the observed expression patterns to the expected patterns based on the literature.

b.

P-value < 0.05 and Z-scores with absolute values ≥ 2 were considered statistically significant.

c.

In IPA, the pathway analysis ratio represents the proportion of molecules from the analyzed dataset that are present in a particular pathway, relative to the total number of molecules in that pathway. It is calculated as: (Number of molecules from the CBD dataset in the pathway) / (total number of molecules in the pathway). A higher ratio indicates that a larger proportion of molecules from the dataset are enriched within a specific pathway, suggesting a stronger association.

Among these pathways, the oxidative phosphorylation pathway exhibited the highest absolute Z-score, thus indicating it was the most significantly altered (Figure 1A and Table 1). Oxidative phosphorylation is the process of generating adenosine triphosphate (ATP) by donating electrons to the electron transport chain, which is composed of mitochondria respiratory Complexes I, II, III, IV, and V embedded within the inner membrane of mitochondria. These complexes consist of proteins and lipids coupled with metal cofactors, such as iron and copper, that facilitate the movement of electrons (Chen et al. 2024a). As shown in Figure 1B, the top 20 genes with the highest absolute log2 fold-change were identified in this pathway. All 20 genes were downregulated by CBD treatment, and they are key genes encoding enzyme proteins of Complexes I – V. For example, the NADH: Ubiquinone Oxidoreductase Subunits (NDUFA3, NDUFA7, NDUFA8, NDUFA11, NDUFB1, NDUFB6, NDUFB7, NDUFB10, and NDUFS8) are involved in mitochondrial respiratory chain Complex I assembly and play an important role in transferring electrons from NADH to the respiratory chain. The succinate dehydrogenase complex subunit C (SDHC) encodes a subunit that comprises Complex II. The ubiquinol-cytochrome c reductase, Complex III subunits (UQCR10, UQCR11, and UQCRQ) genes encode components of Complex III. The cytochrome c oxidase copper chaperone (COX11) gene encodes a protein involved in Complex IV formation. The ATP synthase peripheral stalk subunit d (ATP5PD) gene and ATP synthase membrane subunits (ATP5MC1, ATP5MF, and ATP5ME) genes encode mitochondrial ATP synthases (also known as Complex V) to catalyze ATP synthesis. CYB5A and CYB5B genes encode electron transport proteins. The downregulation of these genes indicates a malfunction in Complexes I – V, which may lead to damage of the mitochondrial respiratory chain, and eventually, total mitochondrial dysfunction.

In a tight association with downregulation of the oxidative phosphorylation pathway, activation of the mitochondrial dysfunction pathway was identified as one of the top 10 affected pathways (Figure 1A). The top 20 genes with the highest absolute log2 fold-change identified in this pathway are listed in Figure 1C. Among these genes are NDUFA7, NDUFA8, NDUFB10, UQCR10, and ATP5MC1, which were also mapped to oxidative phosphorylation, demonstrating their important role in encoding mitochondrial respiratory chain complexes. The expression alternations of the remaining 15 genes have been reported to be associated with mitochondrial dysfunction, such as cAMP responsive element binding protein 5 (CREB5), which has been shown to be upregulated if mitochondrial metabolic activity is impaired (Chen et al. 2024a). The main function of uncoupling proteins (UCPs) is to catalyze the transport of protons across the mitochondrial membrane, whose deregulation can result in mitochondrial dysfunction (Mailloux and Harper 2011). The NADH: Ubiquinone Oxidoreductase subunit 4-like 2 (NDUFA4L2) gene encodes a subunit of Complex I; downregulation of NDUFA4L2 may induce mitochondrial dysfunction (Li et al. 2017). Additionally, transcriptional induction of BCL2 binding component 3 (BBC3) has been demonstrated to trigger the mitochondrial apoptosis pathway in response to ER stress (Reimertz et al. 2003). The alternation of these genes upon CBD treatment indicated that CBD treatment may cause mitochondrial dysfunction.

In addition to the mitochondrial dysfunction pathway, the unfolded protein response was identified as another activated pathway (Figure 1A). The unfolded protein response is a signaling network activated by ER stress. The key biomarkers of ER stress, including endoplasmic reticulum to nucleus signaling 1 (ERN1, encoding protein IRE1α), DNA damage inducible transcript 3 (DDIT3, also known as CHOP), eukaryotic translation initiation factor 2A (EIF2A), eukaryotic translation initiation factor 2 alpha kinase 3 (EIF2AK3, also known as PERK), and transcription factor 4 (ATF4), were all consistently increased after CBD treatment (Figure 1D), indicating the occurrence of ER stress. In our previous study, the protein level of CHOP was shown to increase in response to CBD exposure, particularly at early time points (2 h and 5 h) (Chen et al. 2024a). Notably, the induction of CHOP at both the gene (DDIT3) (Figure 1D) and protein level (Chen et al. 2024a) cross-validates the results from the two studies. Moreover, inhibition of ER stress led to a significant attenuation of cytotoxicity in HepG2 cells (Chen et al. 2024a).

Another important finding is the induction of autophagy pathway (Figure 1A). Autophagy is a type of programmed cell death, which is the process of delivering cytoplasmic material of endogenous or exogenous origin to the lysosome for degradation. In this pathway, genes such as CREB5, DDIT3, EIF2AK3, sirtuin (SIRT)1, protein kinase AMP-activated catalytic subunit alpha 2 (PRKAA2), and ERN1 also mapped to mitochondria dysfunction or unfolded protein response, indicating that these pathways were highly interconnected (Figures 1C – 1E).

Taken together, the RNA-sequencing results not only validated our previous findings regarding the induction of ER stress but also suggested the involvement of damage to the mitochondria respiratory chain, mitochondrial dysfunction, and autophagy in CBD-induced cytotoxicity.

CBD inhibits mitochondrial respiratory chain complexes.

The downregulation of the key genes encoding enzyme proteins of Complexes I – V prompted us to investigate whether the activity of each complex is affected by the treatment of CBD. Using immunocapture-based OXPHOS activity assays, we investigated the effects of CBD exposure (up to 125 μM) on individual respiratory complexes. The calculated 50% inhibitory concentration (BMC50) revealed that Complex IV and Complex V were particularly susceptible to CBD exposure, with the BMC50 values at 11.9 μM and 3.4 μM, respectively. Moderate inhibition was observed for Complex II (BMC50 = 47.6 μM) and Complex III (BMC50 = 45.0 μM). Complex I, on the other hand, exhibited only mild inhibition by CBD, with an BMC50 of 124.2 μM (Figure 2). These results indicate that CBD inhibits mitochondrial respiratory chain complexes, with a particularly strong impact on Complexes IV and V.

Figure 2. CBD causes direct inhibition of mitochondrial respiratory chain complexes.

Figure 2.

Open in a new tab

The inhibitory effects of CBD on the activities of Complexes I (A), II (B), III (C), IV (D), and V (E) were assessed in isolated mitochondria. The results presented represent the mean ± SD from three independent experiments. The 50% inhibitory concentration (BMC50) of CBD for each complex activity was calculated from the concentration-response curves.

CBD triggers mitochondrial dysfunction in HepG2 cells.

Inhibition of mitochondrial respiratory chain complex activity is regarded as a major contributor to mitochondrial dysfunction (Fernandez-Vizarra and Zeviani 2021). The inhibitory effects examined by immunocapture-based assays (Figure 2) and activation of mitochondrial dysfunction pathway identified through RNA-sequencing analysis (Figures 1A and 1C) indicate that mitochondrial dysfunction may occur in response to CBD’s treatment. Mitochondrial dysfunction often involves the opening of the mitochondrial permeability transition pore, resulting in the loss of mitochondrial membrane potentials. To monitor changes in mitochondrial membrane potential, a classic fluorescent assay, namely JC-1 staining was used in cells treated with 10 – 50 μM CBD. As shown in Figures 3A and 3B, treatment of HepG2 cells with 10 – 20 μM CBD did not affect the JC-1 fluorescence ratio. However, treatment with 30 – 50 μM CBD for 24 h led to a concentration-dependent decrease in the JC-1 fluorescence ratio, indicating a loss of mitochondrial membrane potential. Another well-established and commonly used method for detecting mitochondrial dysfunction is the “glucose-galactose” assay (Marroquin et al. 2007). This assay involves using galactose to replace glucose in the culture media and comparing cytotoxicity under both conditions. Culturing cells in a galactose-containing medium forces them to switch from glycolysis to oxidative phosphorylation for ATP production, making them more susceptible to mitochondrial toxicants. The “glucose-galactose” assay enhances the detection of mitochondrial dysfunction-mediated liver toxicity. To assess whether CBD treatment can cause mitochondrial dysfunction, HepG2 cells were treated with 10 – 50 μM CBD or DMSO in either glucose- or galactose-containing medium for 24 h. Cell viability was measured using the CellTiter-Blue assay, and cell death was assessed using LDH release. As shown in Figures 3C and 3D, HepG2 cells grown in galactose-containing medium showed greater sensitivity to CBD treatment. At concentrations ≥ 20 μM, cells in the galactose-containing medium exhibited a marked decrease in viability. At concentrations ≥ 30 μM, these cells showed a significant increase in LDH release compared to those cultured in glucose-containing medium under the same treatment conditions. Together, these findings demonstrate that CBD induces concentration-dependent mitochondrial dysfunction in HepG2 cells.

Figure 3. CBD induces mitochondrial dysfunction.

Figure 3.

Open in a new tab

(A and B) Mitochondrial membrane potential was measured using JC-1 staining after CBD treatment for 24 h. (A) Cells were analyzed using flow cytometry. A collapse in mitochondrial membrane potential is indicated by an increase in the number of cells shifting from red to green fluorescence, as shown in the representative dot plots. (B) Bar graphs represent the mean ratio of red aggregates to green monomers ± S.D. (n = 3). (C and D) HepG2 cytotoxicity was assessed using a CellTiter-Blue cell viability assay (C) and a LDH release assay (D) following culturing in glucose or galactose medium, and exposure to 0 (DMSO), 10, 20, 30, 40, or 50 μM CBD for 24 h. The results presented represent the mean ± SD from three independent experiments. *p < 0.05 treatment under galactose condition versus glucose condition.

CBD induces autophagy in HepG2 cells.

Gene expression analysis result showed that activation of the autophagy signaling pathway was among the top altered pathways upon CBD treatment (Figure 1A). To investigate further whether CBD induces autophagy, we utilized well-established biochemical methods commonly used in autophagy research, including the assessment of autophagosome formation and autophagic flux, a dynamic process of autophagosome synthesis (Mizushima et al. 2010). Autophagosome formation is typically determined by the conversion of cytosolic LC3B I to autophagosomal membrane bound LC3B II. An increase in LC3B II protein levels is a widely recognized marker of autophagosome formation (Mizushima et al. 2010; Wan et al. 2021). As shown in Figures 4A and 4B, 30 – 50 μM CBD significantly increased LC3B II protein levels in HepG2 cells, indicating that CBD induces autophagosome formation.

Figure 4. CBD triggers autophagy.

Figure 4.

Open in a new tab

(A) Representative images of Western blots against LC3B I, LC3B II, and GAPDH in whole cell lysates after treatment with 0 (DMSO) or 10 – 50 μM CBD in HepG2 cells. GAPDH was used as an internal control. (B) Quantification of LC3B II from Western blots, expressed as relative band density (n = 3). *p < 0.05 compared to the DMSO control. (C – F) HepG2-GFP-LC3B cells were incubated with the indicated concentrations of CBD for 24 h. (C) Cells with GFP-LC3B punctations (red arrows) were observed under a 60 × objective using a Cytation 5 cell imaging reader. Representative images were obtained from three independent experiments. The scale bar indicates a length of 100 μm. (D) Fluorescence intensity was measured using a Cytation 5 Multi-Mode microplate reader. The results shown are the mean ± SD from three independent experiments. *p < 0.05 compared with the DMSO control. (E) Representative images of Western blots against LC3B-GFP, Free-GFP, and GAPDH using whole cell lysates from HepG2-GFP-LC3B cells. GAPDH was used as an internal control. (F) Quantification of LC3B-GFP and Free-GFP from Western blots, expressed as relative band density (n = 3). *p < 0.05 compared to the DMSO control.

To visualize further autophagosome formation and monitor autophagic flux, we utilized a previously developed stable HepG2-derived cell line that expresses the GFP-LC3B fusion protein (HepG2-GFP-LC3B) (Chen et al. 2014). During autophagy, GFP-LC3B is incorporated into autophagosomal membranes and subsequently degraded upon fusion with lysosomes. As a result, autophagosomes can be visualized as green fluorescence puncta within cells, and autophagic flux can be assessed by measuring the decrease in GFP-LC3B fluorescence using a spectrofluorometer (Mizushima et al. 2010). This reduction of the fluorescence reflects the degradation of GFP-LC3B during autophagic flux, leading to a decrease in total GFP-LC3B levels. However, because GFP is relatively resistant to lysosomal proteolysis, it is cleaved from LC3B and accumulates as a stable free GFP fragment. Thus, the level of free GFP, typically measured by Western blotting using an anti-GFP antibody, serves as a surrogate marker for autophagic flux. Using this HepG2-GFP-LC3B cell line, we observed that CBD at concentrations ≥ 20 μM increased the number of GFP-LC3B puncta in cells (red arrows, Figure 4C) and decreased GFP fluorescence intensity compared to the DMSO control (Figure 4D). Furthermore, Western blot analysis showed a reduction in full-length GFP-LC3B fusion protein and a corresponding increase in free GFP fragments in cells treated with 20 – 50 μM CBD compared to the DMSO control (Figures 4E and 4F). Collectively, these data clearly demonstrate that CBD induced autophagy and enhances autophagic flux in HepG2 cells.

The effect of 7-hydroxy-CBD and 7-carboxy-CBD on mitochondria function

In our previous study, we observed that, similar to CBD, its two main metabolites, 7-hydroxy-CBD and 7-carboxy-CBD, induced cell cycle disturbances, apoptosis, and ER stress (Chen et al. 2024a; Li et al. 2025b). In this study, we investigated whether 7-hydroxy-CBD and 7-carboxy-CBD share the mechanisms of CBD toxicity, such as the inhibition of mitochondrial respiratory chain complexes, mitochondria dysfunction, and autophagy. First, we examined whether 7-hydroxy-CBD and 7-carboxy-CBD induce inhibition of mitochondrial respiratory Complexes I, II, III, IV, and V. Using immunocapture-based OXPHOS activity assays, we examined the effect of 7-hydroxy-CBD at a maximum concentration of 125 μM and 7-carboxy-CBD at 1000 μM on the activity of individual respiratory complexes. As shown in Figure 5A, Complexes IV and V were particularly vulnerable to the exposure of 7-hydroxy-CBD, with the BMC50 values at 15.7 μM and 6.4 μM, respectively. Moderate inhibition was observed for Complex I (BMC50 = 58.2 μM) and Complex II (BMC50 = 25.4 μM). Complex III was not inhibited by 7-hydroxy-CBD, even at the highest tested concentration of 125 μM (Figure 5A). As shown in Figure 5B, 7-carboxy-CBD exhibited no inhibitory effect on Complexes I, II, and III, even at the highest tested concentration of 1000 μM. Mild inhibition was observed for Complexes IV and V, with the BMC50 values at 88.0 μM and 99.3 μM, respectively (Figure 5B).

Figure 5. The effect of 7-hydroxy-CBD and 7-carboxy-CBD on mitochondrial respiratory chain complexes.

Figure 5.

Open in a new tab

The inhibitory effects of 7-hydroxy-CBD and 7-carboxy-CBD on the activities of Complexes I, II, III, IV, and V were assessed in isolated mitochondria. The results presented represent the mean ± SD from three independent experiments. The values of BMC50 of 7-hydroxy-CBD (A) and 7-carboxy-CBD (B) on each complex activity were calculated from the concentration-response curves.

To study further whether or not mitochondrial dysfunction occurs in 7-hydroxy-CBD or 7-carboxy-CBD-induced cytotoxicity, JC-1 staining was used to detect changes in mitochondrial membrane potential in cells treated with 7-hyroxy-CBD or 7-carboxy-CBD. As shown in Figures 6A and 6B, treatment with 7-hydroxy-CBD at 50 μM for 24 h caused a reduction in the JC-1 fluorescence ratio, suggesting a loss in mitochondrial membrane potential. Glucose-galactose assays demonstrated that HepG2 cells cultured in galactose-containing medium displayed a higher vulnerability to 7-hydroxy-CBD treatment. At concentration of 10 μM or higher, 7-hydroxy-CBD treated cells in the galactose-containing medium exhibited a significant reduction in cell viability compared to the cells subjected to the same treatment but grown in a glucose-containing medium (Figure 6C). An increase in LDH release was observed in cells treated with 7-hydroxy-CBD ≥ 30 μM in the galactose-containing medium compared to the cells grown in glucose-containing medium (Figure 6D). These data suggest that 7-hydroxy-CBD induces mitochondrial dysfunction in HepG2 cells. In contrast, 7-carboxy-CBD treatment at 100 – 300 μM did not induce mitochondrial dysfunction, as evidenced by the negative results from JC-1 staining (Figures 6E and 6F) and glucose-galactose assays (Figures 6G and 6H).

Figure 6. The effect of 7-hydroxy-CBD and 7-carboxy-CBD on mitochondrial function.

Figure 6.

Open in a new tab

Mitochondrial membrane potential was measured using JC-1 staining after 7-hydroxy-CBD (A and B) or 7-carboxy-CBD (E and F) treatment for 24 h. (A and E) Cells were analyzed using flow cytometry. A collapse in mitochondrial membrane potential is indicated by an increase in the number of cells shifting from red to green fluorescence, as shown in the representative dot plots. (B) Bar graphs represent the mean ratio of red aggregates to green monomers ± S.D. (n = 3). (C, D, G, and H) HepG2 cells cultured in glucose or galactose medium were exposed to the indicated concentrations of 7-hydroxy-CBD (C and D) or 7-carboxy-CBD (G and H) for 24 h. Cytotoxicity was assessed using a CellTiter-Blue cell viability assay (C and G) and a LDH release assay (D and H). The results presented represent the mean ± SD from three independent experiments. *p < 0.05 for treatment under galactose condition versus glucose condition.

7-hydroxy-CBD and 7-carboxy-CBD induce autophagy

We next investigated whether 7-hydroxy-CBD and 7-carboxy-CBD induce autophagy. We observed that both 30 – 50 μM 7-hydroxy-CBD and 100 – 300 μM 7-carboxy-CBD treatment increased the formation of autophagosome, as confirmed by the conversion of LC3B I to LC3B II in HepG2 cells (Figures 7A and 7B for 7-hydroxy-CBD; Figures 8A and 8B for 7-carboxy-CBD). Interestingly, although a significant increase in LC3B I to LC3B II conversion was observed at all the tested concentrations of 7-carboxy-CBD, the increase peaked at 150 and 200 μM and then declined at 250 and 300 μM (Figures 8A and 8B). The reasons for the decline in LC3B II expression are not fully understood but may be due to a shift in signaling pathways from autophagy to other cellular processes following substantial apoptotic and necrotic cell death at higher concentrations, as indicated by caspase 3/7 activity and LDH release in our previous study (Li et al. 2025b). We also observed an increase in the number of GFP-LC3B puncta in HepG2-GFP-LC3B cells compared to the DMSO control (red arrows, Figures 7C and 8C), which further supports autophagosome formation. In addition, treatment with 7-hydroxy-CBD and 7-carboxy-CBD caused a significant reduction in GFP fluorescence intensity (Figures 7D and 8D) and a decrease in LC3B-GFP fusion protein levels, accompanied by an increase in free GFP generation as detected by Western blot in HepG2-GFP-LC3B cells (Figures 7E, 7F, 8E, and 8F). Collectively, these data suggest that autophagy was induced by both 7-hydroxy-CBD and 7-carboxy-CBD.

Figure 7. 7-Hydroxy-CBD triggers autophagy.

Figure 7.

Open in a new tab

(A) Representative images of Western blots against LC3B I, LC3B II, and GAPDH in whole cell lysates after treatment with 0 (DMSO) or 10 – 50 μM 7-hydroxy-CBD. GAPDH was used as an internal control. (B) Quantification of LC3B II from Western blots, expressed as relative band density (n = 3). *p < 0.05 compared to the DMSO control. (C – F) HepG2-GFP-LC3B cells were incubated with the indicated concentrations of 7-hydroxy-CBD for 24 h. (C) Cells with GFP-LC3B punctations (red arrows) were observed under a 60 × objective using a Cytation 5 cell imaging reader. Representative images were obtained from three independent experiments. The scale bar indicates a length of 100 μm. (D) Fluorescence intensity was measured using a Cytation 5 Multi-Mode microplate reader. The results shown are the mean ± SD from three independent experiments. *p < 0.05 compared with the DMSO control. (E) Representative images of Western blots against LC3B- GFP, Free-GFP, and GAPDH using whole cell lysates from HepG2-GFP-LC3B cells. GAPDH was used as an internal control. (F) Quantification of LC3B-GFP and Free-GFP from Western blots, expressed as relative band density (n = 3). *p < 0.05 compared to the DMSO control.

Figure 8. 7-Carboxy-CBD triggers autophagy.

Figure 8.

Open in a new tab

(A) Representative images of Western blots against LC3B I, LC3B II, and GAPDH in whole cell lysates after treatment with 0 (DMSO) or 100 – 300 μM 7-carboxy-CBD. GAPDH was used as an internal control. (B) Quantification of LC3B II from Western blots, expressed as relative band density (n = 3). *p < 0.05 compared to the DMSO control. (C – F) HepG2-GFP-LC3B cells were incubated with the indicated concentrations of 7-carboxy-CBD for 24 h. (C) Cells with GFP-LC3B punctations (red arrows) were observed under a 60 × objective using a Cytation 5 cell imaging reader. Representative images were obtained from three independent experiments. The scale bar indicates a length of 100 μm. (D) Fluorescence intensity was measured using a Cytation 5 Multi-Mode microplate reader. The results shown are the mean ± SD from three independent experiments. *p < 0.05 compared with the DMSO control. (E) Representative images of Western blots against LC3B-GFP, Free-GFP, and GAPDH using whole cell lysates from HepG2-GFP-LC3B cells. GAPDH was used as an internal control. (F) Quantification of LC3B-GFP and Free-GFP from Western blots, expressed as relative band density (n = 3). *p < 0.05 compared to the DMSO control.

Discussion

Previously, we investigated the hepatotoxic effects of CBD, 7-hydroxy-CBD, and 7-carboxy-CBD in primary human hepatocytes and HepG2 cells (Chen et al. 2024a; Li et al. 2025b). Although different sensitivities were observed, the results showed that all tested compounds induced hepatotoxicity in both primary human hepatocytes and HepG2 cells. The current study aimed to increase our understanding of the mechanisms underlying the cytotoxicity of these compounds. We selected HepG2 cells for this study not only because the liver is a primary target of toxicity for CBD, 7-hydroxy-CBD, and 7-carboxy-CBD, but also due to their suitability for genetic modifications and the development of pathway-specific reporter assays, which enable the detection of key cytotoxicity mechanisms. For instance, the formation of autophagosome puncta, a hallmark of autophagy, can be monitored using the stable HepG2-GFP-LC3B cell line (Chen et al. 2014). While HepG2 cells may not be the optimal choice for studying metabolic activation-associated toxicity due to their lower expression of CYP enzymes compared to primary human hepatocytes, metabolic capacity was not a major concern for this study because our previous research indicated that most of the key CYPs that metabolize CBD did not significantly impact CBD’s cytotoxicity (Chen et al. 2024a). Furthermore, CYP2D6-mediated metabolism of CBD or 7-hydroxy-CBD was found to attenuate their cytotoxic effects (Chen et al. 2024a; Li et al. 2025b). Conversely, the limited metabolic capacity of HepG2 cells allows for a clearer understanding of the inherent toxicity of CBD metabolites and their direct effects on cellular processes.

Clinically, the mean plasma concentration (Cmax) of CBD, 7-hydroxy-CBD, and 7-carboxy-CBD were 1.1, 0.5, and 28.5 μM, respectively, on the morning of day 7 in healthy adult volunteers who received a CBD oral solution (750 mg, twice daily) under fasted conditions for seven consecutive days (Taylor et al. 2018). However, several factors such as a high-fat meal and preexisting liver disease have been shown to increase the Cmax of CBD and its metabolites. For example, when 1500 mg of CBD was administered following a high-fat breakfast, the mean Cmax of CBD, 7-hydroxy-CBD, and 7-carboxy-CBD increased by approximately 5-, 3-, and 2-fold, respectively, compared to administration in the fasted state (Taylor et al. 2018). In individuals with hepatic impairment, CBD Cmax also increased proportionally with the severity of liver dysfunction (Taylor et al. 2019). In this study, significant mitochondrial dysfunction and autophagy were observed at approximately 20 – 30 μM for CBD and 7-hydroxy-CBD, while 7-carboxy-CBD induced autophagy at concentrations above 100 μM. Although these concentrations exceed the reported plasma Cmax values, the primary focus of our study was on hepatotoxicity, and the liver being the first-pass organ for orally administered CBD is likely to be exposed to higher concentrations than those detected in plasma. Although direct measurements of liver concentrations in humans are lacking, an animal study has demonstrated liver accumulation of CBD (Child and Tallon 2022). Furthermore, a physiologically based pharmacokinetic (PBPK) modeling study predicted that the hepatic Cmax of CBD could be up to 16.9-fold higher than plasma Cmax following a 30 mg oral dose in humans (Liu and Sprando 2023). Taken together, these data suggest that the concentrations used in this study may still fall within a physiologically relevant range for liver tissue. However, given the uncertainty surrounding in vivo hepatic exposure, the relevance of the in vitro findings (particularly those observed in cell lines) should be interpreted with caution, as the observed effects may not directly reflect human in vivo outcomes.

Previous investigations employing various approaches have shown that CBD exhibits mitochondrial-damaging potential. For example, a proteome profiling study conducted in the human neuroblastoma cell line SH-SY5Y reported a downregulation of the proteins in the oxidative phosphorylation pathway, following 10 μM CBD treatment for 24 h (Abyadeh et al. 2023). Another study observed a reduction in the oxygen consumption rate in HepG2 cells after exposure to CBD (IC50 = 174 μM) for 1 hour, suggesting its potential as electron transport chain inhibitors (Lakhani et al. 2023). These findings support our results that CBD strongly inhibits mitochondrial respiratory complexes (Figure 2) using a cell-free system. Lakhani et al., also reported IC50 values for electron transport chain inhibition by 7-hydroxy-CBD and 7-carboxy-CBD at approximately 10 μM and 1740 μM, respectively (Lakhani et al. 2023). These findings align with our results, which demonstrated that 7-hydroxy-CBD, but not 7-carboxy-CBD, strongly inhibits mitochondrial respiratory complexes (Figure 5). CBD-induced loss of mitochondrial membrane potential and suppressed oxygen consumption have also been documented in various cell models, including human ovarian cancer cell lines SKOV3 and CAOV3 (Ma et al. 2023), chronic myeloid leukemia cell lines K562, KU812 and MOLM-6 (Maggi et al. 2022), human neuroblastoma cell line BE(2)-M17 and acutely isolated rat brain mitochondria (Drummond-Main et al. 2023), and human gastric cancer cell line AGS (Jeong et al. 2019). Additionally, studies have reported CBD-induced mitochondrial dysfunction and autophagy in human glioma cell lines U251, U87 MG, and LN18 (Huang et al. 2021). Consistent with these findings, our study demonstrates that CBD exposure at certain concentrations resulted in a loss of mitochondrial membrane potential and mitochondrial dysfunction in HepG2 cells (Figure 3). Notably, we demonstrated that 7-hydroxy-CBD, similar to CBD, caused a loss of mitochondrial potential and mitochondrial dysfunction, whereas 7-carboxy-CBD did not (Figure 6). In contrast, CBD has been reported to alleviate mitochondrial dysfunction induced in a number of experimental models: by aluminum phosphide in hearts cells of adult male Albino Wistar rats (Hooshangi Shayesteh et al. 2022), by 1-methyl-4-phenylpyridine (MPP+) in human neuroblastoma cell line SH-SY5Y (Kang et al. 2021), by hemorrhagic shock in Sprague Dawley rats (Lin et al. 2020), and by doxorubicin in heart cells from male C57BL/6J mice (Hao et al. 2015). The underlying mechanisms of these protective effects of CBD are unknown, but interactions between CBD and other compounds may play an important role, warranting further investigation.

Employing multiple methods, we found that CBD induced autophagy in HepG2 cells (Figure 4), which aligns with previous observations of CBD-induced autophagy in numerous other cell types, including human colorectal cancer cell line HCT116 (Wang et al. 2023), human leukemia monocytic cell line THP-1 (Tomer et al. 2022), human breast cancer cell line MDA-MB-231 (D’Aloia et al. 2022; Shrivastava et al. 2011), human keratinocyte cell line HaCaT under UVB irradiation (Li et al. 2022a), human neuroblastoma cell line SH-SY5Y (Vrechi et al. 2021; Wang et al. 2022), human placental trophoblast cell lines HTR-8/SVneo and BeWo (Alves et al. 2021), human adipose tissue-derived mesenchymal stem cell line (Bublitz et al. 2020), human umbilical vein endothelial cell line (Bockmann and Hinz 2020), human glioblastoma multiforme cell line U87MG (Ivanov et al. 2020), and human colonic epithelial cell line Caco-2 (Koay et al. 2014). Importantly, we demonstrated 7-hydroxy-CBD and 7-carboxy-CBD also induced autophagy, despite differences in their activity (Figures 7 and 8), implying these metabolites share similar autophagy-inducing properties with CBD.

Based on our RNA-sequencing pathway analyses, three of the top 10 signaling pathways affected by CBD are associated with cholesterol biosynthesis. These include the 24-dehydrocholesterol reductase (DHCR24) signaling pathway, liver X receptor/retinoid X receptor (LXR/RXR) activation pathway, and super-pathway of cholesterol biosynthesis. DHCR24 catalyzes the conversion of desmosterol to cholesterol, marking the final step in cholesterol biosynthesis (Zerenturk et al. 2013). The LXR/RXR pathway is heavily involved in cholesterol efflux from hepatocytes into bile, intestinal absorption of cholesterol, reverse transport of excess cholesterol from peripheral tissues to the liver, and excretion in the bile and eventually the feces (Afonso et al. 2018; Lee and Tontonoz 2015). Additional data regarding CBD and cholesterol biosynthesis include a study that reported CBD impaired the formation of cholesteryl ester, which can be hydrolyzed by cholesterol esterase to produce cholesterol and free fatty acids (Cornicelli et al. 1981), and a study conducted in the human neuroblastoma cell line SK-N-BE(2), where CBD treatment led to changes in biosynthesis, storage, and uptake of cholesterol as indicated by proteomics and RNA-sequencing analyses (Guard et al. 2022). The findings from these studies, along with our RNA-sequencing results, suggest a potential link between CBD exposure and cholesterol homeostasis that warrants further study. Moreover, cholesterol synthesis takes place in the ER. Given that our previous study demonstrated in vitro CBD exposure can induce ER stress (Chen et al. 2024a), it was not surprising that cholesterol synthesis was inhibited when ER function was impaired. Interestingly, DHCR24 also plays an important role in ER stress signaling and can be deactivated by caspase cleavage during apoptosis (Zerenturk et al. 2013). Therefore, it is reasonable to hypothesize that the downregulation of the DHCR24 signaling pathway observed in this study resulted from CBD-induced apoptosis.

In summary, our investigation indicates that in HepG2 cells CBD and 7-hydroxy-CBD trigger the inhibition of mitochondrial respiratory complexes, leading to mitochondrial dysfunction, ER stress, and autophagy. Compared to CBD and 7-hydroxy-CBD, 7-carboxy-CBD induces ER stress and autophagy at higher concentrations but does not appear to disrupt mitochondrial functions. While the precise mechanisms governing the delicate balance between cell survival and cell death remain to be fully elucidated, it is clear that the cytotoxicity induced by CBD, 7-hydroxy-CBD, and 7-carboxy-CBD shown in this study involve a complex process encompassing multiple molecular events and various signaling pathways.

Supplementary Material

Supplementary tables

Acknowledgments:

This work was supported by U.S. Food and Drug Administration’s intramural grant program.

Footnotes

Conflict of interest: The authors declare no conflict of interest.

Disclaimer: This article reflects the views of the authors and does not necessarily reflect those of the U.S. Food and Drug Administration. Any mention of commercial products is for clarification only and is not intended as approval, endorsement, or recommendation.

References

  1. Abyadeh M, Gupta V, Liu X, et al. (2023) Proteome-Wide Profiling Using Sample Multiplexing of a Human Cell Line Treated with Cannabidiol (CBD) and Tetrahydrocannabinol (THC). Proteomes 11(4) doi: 10.3390/proteomes11040036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Afonso MS, Machado RM, Lavrador MS, Quintao ECR, Moore KJ, Lottenberg AM (2018) Molecular Pathways Underlying Cholesterol Homeostasis. Nutrients 10(6) doi: 10.3390/nu10060760 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Alves P, Amaral C, Teixeira N, Correia-da-Silva G (2021) Cannabidiol disrupts apoptosis, autophagy and invasion processes of placental trophoblasts. Arch Toxicol 95(10):3393–3406 doi: 10.1007/s00204-021-03122-z [DOI] [PubMed] [Google Scholar]
  4. Anders S, Huber W (2010) Differential expression analysis for sequence count data. Genome Biol 11(10):R106 doi: 10.1186/gb-2010-11-10-r106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bockmann S, Hinz B (2020) Cannabidiol Promotes Endothelial Cell Survival by Heme Oxygenase-1-Mediated Autophagy. Cells 9(7) doi: 10.3390/cells9071703 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bublitz K, Bockmann S, Peters K, Hinz B (2020) Cannabinoid-Induced Autophagy and Heme Oxygenase-1 Determine the Fate of Adipose Tissue-Derived Mesenchymal Stem Cells under Stressful Conditions. Cells 9(10) doi: 10.3390/cells9102298 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen S, Dobrovolsky VN, Liu F, et al. (2014) The role of autophagy in usnic acid-induced toxicity in hepatic cells. Toxicol Sci 142(1):33–44 doi: 10.1093/toxsci/kfu154 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chen S, Li X, Wu Q, et al. (2024a) Investigation of cannabidiol-induced cytotoxicity in human hepatic cells. Toxicology 506:153884 doi: 10.1016/j.tox.2024.153884 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chen S, Li Y, Li X, et al. (2025) Mechanistic study of pexidartinib-induced toxicity in human hepatic cells. Chem Biol Interact 419:111641 doi: 10.1016/j.cbi.2025.111641 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chen S, Li Y, Li X, et al. (2024b) Metabolism and liver toxicity of cannabidiol. J Environ Sci Health C Toxicol Carcinog:1–17 doi: 10.1080/26896583.2024.2366741 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Child RB, Tallon MJ (2022) Cannabidiol (CBD) Dosing: Plasma Pharmacokinetics and Effects on Accumulation in Skeletal Muscle, Liver and Adipose Tissue. Nutrients 14(10) doi: 10.3390/nu14102101 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cornicelli JA, Gilman SR, Krom BA, Kottke BA (1981) Cannabinoids impair the formation of cholesteryl ester in cultured human cells. Arteriosclerosis: An Official Journal of the American Heart Association, Inc 1(6):449–454 doi:doi: 10.1161/01.ATV.1.6.449 [DOI] [PubMed] [Google Scholar]
  13. D’Aloia A, Ceriani M, Tisi R, Stucchi S, Sacco E, Costa B (2022) Cannabidiol Antiproliferative Effect in Triple-Negative Breast Cancer MDA-MB-231 Cells Is Modulated by Its Physical State and by IGF-1. Int J Mol Sci 23(13) doi: 10.3390/ijms23137145 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Drummond-Main CD, Ahn Y, Kesler M, et al. (2023) Cannabidiol Impairs Brain Mitochondrial Metabolism and Neuronal Integrity. Cannabis Cannabinoid Res 8(2):283–298 doi: 10.1089/can.2022.0011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. EPIDIOLEX® (2023) EPIDIOLEX® drug label information. In. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=8bf27097-4870-43fb-94f0-f3d0871d1eec
  16. FDA US (2017) Non-clinical Review. Application number: 210365Orig1s000.
  17. FDA US (2018) FDA Approves First Drug Comprised of an Active Ingredient Derived from Marijuana to Treat Rare, Severe Forms of Epilepsy. In. https://www.fda.gov/news-events/press-announcements/fda-approves-first-drug-comprised-active-ingredient-derived-marijuana-treat-rare-severe-forms
  18. FDA US (2019a) FDA warns company marketing unapproved cannabidiol products with unsubstantiated claims to treat cancer, Alzheimer’s disease, opioid withdrawal, pain and pet anxiety. In. https://www.fda.gov/news-events/press-announcements/fda-warns-company-marketing-unapproved-cannabidiol-products-unsubstantiated-claims-treat-cancer.
  19. FDA US (2019b) Safety of CBD in Humans – A Literature Review.
  20. FDA US (2020) FDA Approves New Indication for Drug Containing an Active Ingredient Derived from Cannabis to Treat Seizures in Rare Genetic Disease. In. https://www.fda.gov/news-events/press-announcements/fda-approves-new-indication-drug-containing-active-ingredient-derived-cannabis-treat-seizures-rare
  21. FDA US (2021) FDA Warns Companies Illegally Selling Over-the-Counter CBD Products for Pain Relief. In. https://www.fda.gov/news-events/press-announcements/fda-warns-companies-illegally-selling-over-counter-cbd-products-pain-relief.
  22. Fernandez-Vizarra E, Zeviani M (2021) Mitochondrial disorders of the OXPHOS system. FEBS Letters 595(8):1062–1106 doi: 10.1002/1873-3468.13995 [DOI] [PubMed] [Google Scholar]
  23. Gingrich J, Choudhuri S, Cournoyer P, Downey J, Muldoon Jacobs K (2023) Review of the oral toxicity of cannabidiol (CBD). Food Chem Toxicol:113799 doi: 10.1016/j.fct.2023.113799 [DOI] [PubMed] [Google Scholar]
  24. Guard SE, Chapnick DA, Poss ZC, et al. (2022) Multiomic Analysis Reveals Disruption of Cholesterol Homeostasis by Cannabidiol in Human Cell Lines. Mol Cell Proteomics 21(10):100262 doi: 10.1016/j.mcpro.2022.100262 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hao E, Mukhopadhyay P, Cao Z, et al. (2015) Cannabidiol Protects against Doxorubicin-Induced Cardiomyopathy by Modulating Mitochondrial Function and Biogenesis. Mol Med 21(1):38–45 doi: 10.2119/molmed.2014.00261 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Hooshangi Shayesteh MR, Haghi-Aminjan H, Baeeri M, et al. (2022) Modification of the hemodynamic and molecular features of phosphine, a potent mitochondrial toxicant in the heart, by cannabidiol. Toxicol Mech Methods 32(4):288–301 doi: 10.1080/15376516.2021.1998851 [DOI] [PubMed] [Google Scholar]
  27. Huang T, Xu T, Wang Y, et al. (2021) Cannabidiol inhibits human glioma by induction of lethal mitophagy through activating TRPV4. Autophagy 17(11):3592–3606 doi: 10.1080/15548627.2021.1885203 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Ivanov VN, Grabham PW, Wu CC, Hei TK (2020) Inhibition of autophagic flux differently modulates cannabidiol-induced death in 2D and 3D glioblastoma cell cultures. Sci Rep 10(1):2687 doi: 10.1038/s41598-020-59468-4 [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
  29. Jeong S, Jo MJ, Yun HK, et al. (2019) Cannabidiol promotes apoptosis via regulation of XIAP/Smac in gastric cancer. Cell Death Dis 10(11):846 doi: 10.1038/s41419-019-2001-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kang S, Li J, Yao Z, Liu J (2021) Cannabidiol Induces Autophagy to Protects Neural Cells From Mitochondrial Dysfunction by Upregulating SIRT1 to Inhibits NF-kappaB and NOTCH Pathways. Front Cell Neurosci 15:654340 doi: 10.3389/fncel.2021.654340 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Koay LC, Rigby RJ, Wright KL (2014) Cannabinoid-induced autophagy regulates suppressor of cytokine signaling-3 in intestinal epithelium. Am J Physiol Gastrointest Liver Physiol 307(2):G140–8 doi: 10.1152/ajpgi.00317.2013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lakhani VV, Generaux G, Howell BA, Longo DM, Watkins PB (2023) Assessing Liver Effects of Cannabidiol and Valproate Alone and in Combination Using Quantitative Systems Toxicology. Clin Pharmacol Ther 114(5):1006–1014 doi: 10.1002/cpt.3004 [DOI] [PubMed] [Google Scholar]
  33. Lee SD, Tontonoz P (2015) Liver X receptors at the intersection of lipid metabolism and atherogenesis. Atherosclerosis 242(1):29–36 doi: 10.1016/j.atherosclerosis.2015.06.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Li J, Bai C, Guo J, Liang W, Long J (2017) NDUFA4L2 protects against ischaemia/reperfusion-induced cardiomyocyte apoptosis and mitochondrial dysfunction by inhibiting complex I. Clin Exp Pharmacol Physiol 44(7):779–786 doi: 10.1111/1440-1681.12768 [DOI] [PubMed] [Google Scholar]
  35. Li Y, Hao D, Wei D, et al. (2022a) Photoprotective Effects of Cannabidiol against Ultraviolet-B-Induced DNA Damage and Autophagy in Human Keratinocyte Cells and Mouse Skin Tissue. Molecules 27(19) doi: 10.3390/molecules27196740 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Li Y, Li X, Cournoyer P, Choudhuri S, Guo L, Chen S (2023a) Induction of apoptosis by cannabidiol and its main metabolites in human Leydig cells. Arch Toxicol 97(12):3227–3241 doi: 10.1007/s00204-023-03609-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Li Y, Li X, Cournoyer P, Choudhuri S, Guo L, Chen S (2025a) Comparing the cannabidiol-induced transcriptomic profiles in human and mouse Sertoli cells. Toxicology 512:154068 doi: 10.1016/j.tox.2025.154068 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Li Y, Li X, Cournoyer P, et al. (2023b) Cannabidiol-induced transcriptomic changes and cellular senescence in human Sertoli cells. Toxicol Sci 191(2):227–238 doi: 10.1093/toxsci/kfac131 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Li Y, Li X, Wu Q, et al. (2025b) 7-Hydroxycannabidiol and 7-carboxycannabidiol induced cytotoxicity via apoptosis and endoplasmic reticulum stress in human hepatic cells. Arch Toxicol doi: 10.1007/s00204-025-04001-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Li Y, Wu Q, Li X, et al. (2024) Toxicity of cannabidiol and its metabolites in TM3 mouse Leydig cells: a comparison with primary human Leydig cells. Arch Toxicol 98(8):2677–2693 doi: 10.1007/s00204-024-03754-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Li Y, Wu Q, Li X, et al. (2022b) In vitro effects of cannabidiol and its main metabolites in mouse and human Sertoli cells. Food Chem Toxicol 159:112722 doi: 10.1016/j.fct.2021.112722 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Lin B, Gao Y, Li Z, Zhang Z, Lin X, Gao J (2020) Cannabidiol alleviates hemorrhagic shock-induced neural apoptosis in rats by inducing autophagy through activation of the PI3K/AKT pathway. Fundam Clin Pharmacol 34(6):640–649 doi: 10.1111/fcp.12557 [DOI] [PubMed] [Google Scholar]
  43. Liu Y, Sprando RL (2023) Physiologically based pharmacokinetic modeling and simulation of cannabinoids in human plasma and tissues. J Appl Toxicol 43(4):589–598 doi: 10.1002/jat.4409 [DOI] [PubMed] [Google Scholar]
  44. Lo LA, Christiansen A, Eadie L, et al. (2023) Cannabidiol-associated hepatotoxicity: A systematic review and meta-analysis. J Intern Med doi: 10.1111/joim.13627 [DOI] [PubMed] [Google Scholar]
  45. Ma L, Zhang H, Liu C, et al. (2023) A novel mechanism of cannabidiol in suppressing ovarian cancer through LAIR-1 mediated mitochondrial dysfunction and apoptosis. Environ Toxicol 38(5):1118–1132 doi: 10.1002/tox.23752 [DOI] [PubMed] [Google Scholar]
  46. Maggi F, Morelli MB, Tomassoni D, et al. (2022) The effects of cannabidiol via TRPV2 channel in chronic myeloid leukemia cells and its combination with imatinib. Cancer Sci 113(4):1235–1249 doi: 10.1111/cas.15257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Mailloux RJ, Harper ME (2011) Uncoupling proteins and the control of mitochondrial reactive oxygen species production. Free Radic Biol Med 51(6):1106–15 doi: 10.1016/j.freeradbiomed.2011.06.022 [DOI] [PubMed] [Google Scholar]
  48. Marroquin LD, Hynes J, Dykens JA, Jamieson JD, Will Y (2007) Circumventing the Crabtree effect: replacing media glucose with galactose increases susceptibility of HepG2 cells to mitochondrial toxicants. Toxicol Sci 97(2):539–47 doi: 10.1093/toxsci/kfm052 [DOI] [PubMed] [Google Scholar]
  49. Mizushima N, Yoshimori T, Levine B (2010) Methods in mammalian autophagy research. Cell 140(3):313–26 doi: 10.1016/j.cell.2010.01.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Reimertz C, Kogel D, Rami A, Chittenden T, Prehn JH (2003) Gene expression during ER stress-induced apoptosis in neurons: induction of the BH3-only protein Bbc3/PUMA and activation of the mitochondrial apoptosis pathway. J Cell Biol 162(4):587–97 doi: 10.1083/jcb.200305149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Shao K, Shapiro AJ (2018) A Web-Based System for Bayesian Benchmark Dose Estimation. Environ Health Perspect 126(1):017002 doi: 10.1289/EHP1289 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Shrivastava A, Kuzontkoski PM, Groopman JE, Prasad A (2011) Cannabidiol induces programmed cell death in breast cancer cells by coordinating the cross-talk between apoptosis and autophagy. Mol Cancer Ther 10(7):1161–72 doi: 10.1158/1535-7163.MCT-10-1100 [DOI] [PubMed] [Google Scholar]
  53. Szaflarski JP, Bebin EM, Comi AM, et al. (2018) Long-term safety and treatment effects of cannabidiol in children and adults with treatment-resistant epilepsies: Expanded access program results. Epilepsia 59(8):1540–1548 doi: 10.1111/epi.14477 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Taylor L, Crockett J, Tayo B, Checketts D, Sommerville K (2020) Abrupt withdrawal of cannabidiol (CBD): A randomized trial. Epilepsy Behav 104(Pt A):106938 doi: 10.1016/j.yebeh.2020.106938 [DOI] [PubMed] [Google Scholar]
  55. Taylor L, Crockett J, Tayo B, Morrison G (2019) A Phase 1, Open-Label, Parallel-Group, Single-Dose Trial of the Pharmacokinetics and Safety of Cannabidiol (CBD) in Subjects With Mild to Severe Hepatic Impairment. J Clin Pharmacol 59(8):1110–1119 doi: 10.1002/jcph.1412 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Taylor L, Gidal B, Blakey G, Tayo B, Morrison G (2018) A Phase I, Randomized, Double-Blind, Placebo-Controlled, Single Ascending Dose, Multiple Dose, and Food Effect Trial of the Safety, Tolerability and Pharmacokinetics of Highly Purified Cannabidiol in Healthy Subjects. CNS Drugs 32(11):1053–1067 doi: 10.1007/s40263-018-0578-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Tomer S, Mu W, Suryawanshi G, et al. (2022) Cannabidiol modulates expression of type I IFN response genes and HIV infection in macrophages. Front Immunol 13:926696 doi: 10.3389/fimmu.2022.926696 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Vrechi TAM, Leao A, Morais IBM, et al. (2021) Cannabidiol induces autophagy via ERK1/2 activation in neural cells. Sci Rep 11(1):5434 doi: 10.1038/s41598-021-84879-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Wan DF, Pan SS, Tong YS, Huang Y (2021) Exercise Preconditioning Promotes Autophagy to Cooperate for Cardioprotection by Increasing LC3 Lipidation-Associated Proteins. Front Physiol 12:599892 doi: 10.3389/fphys.2021.599892 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Wang F, Dezfouli AB, Khosravi M, et al. (2023) Cannabidiol-induced crosstalk of apoptosis and macroautophagy in colorectal cancer cells involves p53 and Hsp70. Cell Death Discov 9(1):286 doi: 10.1038/s41420-023-01578-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Wang Z, Zheng P, Chen X, et al. (2022) Cannabidiol induces autophagy and improves neuronal health associated with SIRT1 mediated longevity. Geroscience 44(3):1505–1524 doi: 10.1007/s11357-022-00559-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Watkins PB, Church RJ, Li J, Knappertz V (2021) Cannabidiol and Abnormal Liver Chemistries in Healthy Adults: Results of a Phase I Clinical Trial. Clin Pharmacol Ther 109(5):1224–1231 doi: 10.1002/cpt.2071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Zerenturk EJ, Sharpe LJ, Ikonen E, Brown AJ (2013) Desmosterol and DHCR24: Unexpected new directions for a terminal step in cholesterol synthesis. Progress in Lipid Research 52(4):666–680 doi: 10.1016/j.plipres.2013.09.002 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary tables
NIHMS2125271-supplement-Supplementary_tables.xlsx (107KB, xlsx)

RESOURCES