To the Editor:

Cannabis is one of the most widely consumed drugs in the world, with an estimated 180 million annual users globally [1]. Recent trends of legalisation of medicinal and/or recreational cannabis have led to increased access to cannabis products that will likely lead to increased use over the coming years. Inhalation of combusted cannabis is by far the most prevalent form of cannabis use, reported by 89% of users in the 2018 Canadian Cannabis Survey [2]. Chemically, cannabis smoke includes pharmacologically active components such as tetrahydrocannabinol (THC), cannabidiol (CBD) and polycyclic aromatic hydrocarbons [3]. Many studies have investigated the psychoactive and immunomodulatory properties of THC and CBD respectively, and examined the potential use of cannabis as an intervention for chronic pain, immune disorders and neurological disorders [4, 5]. Despite this body of knowledge and recent rise of systematic reviews on these topics, the effects of inhaled cannabis smoke on the respiratory mucosal immune responses are less clear [6].

Like tobacco smoke, repeated exposure to cannabis smoke has been associated with cough and shortness of breath [7]. However, few studies have examined the effects of cannabis smoke on the respiratory mucosa, the primary physical and immunological barrier to the environment [8]. Tobacco smoke induces oxidative stress and cytokine production while disrupting epithelial barrier function, impacting chronic lung disease pathology and exacerbations [9]. To manage chronic lung diseases, an individual may be prescribed a broad, anti-inflammatory long-acting β-agonist (LABA)/glucocorticoid (GC) combination treatment [10], a therapy that may be compromised by tobacco smoke exposure [11, 12]. Given the similarities between tobacco and cannabis smoke [8], we examined whether cannabis smoke exposure similarly attenuates LABA/GC transcriptomic responses and inflammatory mediator release in human airway epithelial cells.

Cannabis smoke extract (CSE)- and tobacco smoke extract (TSE)-conditioned media were prepared according to previously published methods [12]. TSE was generated from Kentucky Research Grade Cigarettes with an intact filter (lot 3R4F), and CSE from cannabis rolled with cardboard filters and sourced from Jonathan Page (University of British Columbia, Vancouver, Canada) (13% (w/w) tetrahydrocannabinolic acid strain with 0.18% THC, 0.35% tetrahydrocannabivarinic acid and 0.18% cannabigerolic acid; ∼0.7 g dried cannabis). Extracts were prepared by bubbling cannabis or tobacco smoke through 4 mL HEPES-buffered Eagle's minimal essential medium, filtering (0.22 μm) and diluting with fresh medium (10% dilution, optical density at 260 nm 0.04045). Calu-3 lung epithelial cells on Transwells (Corning Inc., Corning, NY, USA) were apically exposed to 10% of either CSE or TSE with and without 10 nM formoterol and 100 nM budesonide (Form/Bud) for 24 h. Total RNA was extracted and sent to The Centre for Applied Genomics at the Hospital for Sick Children (Toronto, Canada) for RNA sequencing. Apical cell supernatants were assessed using a multiplex cytokine/chemokine protein array by Eve Technologies (Calgary, Canada). Oxidative stress was assessed using 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA) and CellROX Green (Thermo Fisher Scientific, Waltham, MA, USA), counterstained using Hoechst 3342 and quantified using a plate reader. All experiments had four biological replicates. Significant differences in cytokines and oxidative stress were identified through permutation ANOVA followed by a Tukey Honest Significant Difference post hoc test using the “lmPerm” package in R (version 3.4.3; The R Foundation, Vienna, Austria). Significant differences between transcriptional expression profiles were identified through analysis of similarities using the “vegan” package in R.

Exposure to 10% CSE or TSE and/or Form/Bud for 24 h did not significantly change cell viability as assessed by lactate dehydrogenase release assays (not shown). Like TSE, CSE significantly elevated proinflammatory cytokine interleukin-8/CXCL8 and epithelial cell repair mediator transforming growth factor-α (figure 1a and b). Intervention with Form/Bud prevented CSE and TSE increases in IL-8/CXCL8 release and, to a lesser extent, TGF-α. We also observed a significant decrease in interferon-γ-induced protein 10/CXCL10 (figure 1c) in response to CSE and TSE exposure, a decrease that was augmented in response to Form/Bud intervention.

We next interrogated RNA sequencing data to address the efficacy of LABA/GC intervention on attenuating CSE- and TSE-induced transcriptomic responses observed in our previous study [8]. First, we observed that the most upregulated gene exclusive to Form/Bud treatment was HSD11B2 (corticosteroid 11-β-dehydrogenase) (vehicle+Form/Bud FC 34.80, CSE+Form/Bud FC 33.99 and TSE+Form/Bud FC 27.32), indicating that our intervention was efficacious on modulating glucocorticoid signalling (data not shown). Second, to identify expression patterns unique to each smoke exposure+LABA/GC intervention, we directly compared these datasets to identify differentially expressed genes and visualised their log₂ fold changes relative to untreated+Form/Bud (figure 1d). Form/Bud intervention in smoke-exposed cells induced a highly correlated transcriptomic response (r=0.771, p<1.0×10⁻¹⁵), with 801 differentially expressed genes in CSE+Form/Bud versus untreated+Form/Bud and 1105 differentially expressed genes in TSE+Form/Bud versus untreated+Form/Bud. Third, NEU1 was the most significantly increased gene in both CSE+Form/Bud and TSE+Form/Bud compared to vehicle+Form/Bud exposed cells, and aryl hydrocarbon receptor (AHR) genes CYP1A1 and CYP1B1 remained upregulated in both CSE+Form/Bud- and TSE+Form/Bud-exposed cells.

To further interrogate genes significantly differentially expressed in both CSE+Form/Bud- and TSE+Form/Bud-exposed cells, we performed a functional enrichment analysis (figure 1e). Interestingly, regulation of apoptotic process (rank 1, 45 out of 816 genes, gene ontology (GO):0042981), neutrophil-mediated immunity (rank 3, 30 out of 108 genes, GO:0002446), and cellular response to oxidative stress (rank 6, 13 out of 488 genes, GO:0034599) were included in the top 10 GO terms ranked by adjusted p-value. To functionally characterise the “cellular response to oxidative stress” term, we performed additional in vitro experiments and analysis of oxidative stress-related gene expression data [13, 14] (figure 1f). Interestingly, several curated genes (e.g. GCLC and GCLM) were increased beyond CSE or TSE exposure with Form/Bud intervention. Furthermore, when we examined overall oxidative stress using CellROX (figure 1g–m) and H₂DCFDA (figure 1n) in vitro imaging assays, we observed reactive oxygen species (ROS) generation in both CSE+Form/Bud- or TSE+Form/Bud-exposed cells.

Tobacco smoke has long been associated with increased symptoms and severity of chronic lung diseases that include asthma, pulmonary fibrosis and COPD, and attenuated GC responses in human lung epithelial cells [12]. Chronic bronchitis and airway remodelling/scarring due to persistent inflammation and oxidative stress from recurring exposure to tobacco smoke are known to be important pathological mechanisms in the progression of COPD and asthma [11, 15]. We demonstrate here for the first time that a Form/Bud intervention of human lung cells exposed to cannabis smoke suppressed increased inflammation (IL-8/CXCL8) and epithelial repair mediators (TGF-α), while expression of oxidative stress genes remained elevated. Although Form/Bud intervention reduced the expression of IL-8/CXCL8, CSE and TSE exposure may still be modulating epithelial immune responses, as seen by the continued enrichment of neutrophil-related GO terms. Furthermore, the persistence of increased generation of ROS and expression of the Nrf2 oxidative stress response genes in the presence of LABA/GC medications used in chronic respiratory disease management may further impact long-term disease development and management.

Our study has several limitations in the experimental design. Our model is a single exposure to smoke-conditioned media and may not accurately represent chronic smoke exposure conditions, which would require repeated exposures over time in vitro. Despite this limitation, we emphasise that our previous publication using this model demonstrated strong significant correlations of gene expression with in situ bronchial brushes from human smokers and primary air–liquid interface cultures exposed to tobacco smoke [8]. Additional experiments using primary human airway epithelial cells grown under air–liquid interface culture conditions and exposed to whole-smoke extract should be explored to complement the present data. We recognise that there are substantial differences in cannabis strains available on the medicinal and recreational market, and therefore chose a representative hybrid strain available on the medicinal cannabis market in Canada. Despite these limitations, the robust changes we observed in our data suggest that cannabis smoke exposure still poses a significant health risk, and warrants ongoing study to build a body of evidence to support public policy, government regulations and individual user practices.