Abstract
Effects of waterpipe smoking on lung pathobiology and carcinogenesis remain sparse despite the worldwide emergence of this tobacco vector. To address this gap, we investigated the effects of chronic waterpipe smoke (WPS) exposure on lung pathobiology, host immunity, and tumorigenesis using an experimental animal model that is prone to tobacco carcinogens and an exploratory observational analysis of human waterpipe smokers and nonsmokers. Mice exhibited elevated incidence of lung tumors following heavy WPS exposure (5 days/week for 20 weeks) compared to littermates with light WPS (once/week for 20 weeks) or control air. Lungs of mice exposed to heavy WPS showed augmented CD8+ and CD4+ T cell counts along with elevated protumor immune phenotypes including increased IL17A in T/B cells, PD-L1 on tumor and immune cells, and the proinflammatory cytokine IL1β in myeloid cells. RNA-sequencing (RNA-seq) analysis showed reduced antitumor immune gene signatures in animals exposed to heavy WPS relative to control air. We also performed RNA-seq analysis of airway epithelia from bronchial brushings of cancer-free waterpipe smokers and nonsmokers undergoing diagnostic bronchoscopy. Transcriptomes of normal airway cells in waterpipe smokers, relative to waterpipe nonsmokers, harbored gene programs that were associated with poor clinical outcomes in patients with lung adenocarcinoma, alluding to a WPS-associated molecular injury, like that established in response to cigarette smoking. Our findings support the notion that WPS exhibits carcinogenic effects and constitutes a possible risk factor for lung cancer as well as warrant future studies that can guide evidence-based policies for mitigating waterpipe smoking.
Potential carcinogenic effects of waterpipe smoking are very poorly understood despite its emergence as a socially acceptable form of smoking. Our work highlights carcinogenic effects of waterpipe smoking in the lung and, thus, accentuate the need for inclusion of individuals with exclusive waterpipe smoking in prevention and smoking cessation studies.
Introduction
Tobacco use poses a major worldwide health concern and accounts for over 8 million deaths yearly, including 1.8 million lung cancer–related deaths (1). While cigarette smoking prevalence has declined globally in recent decades, waterpipe smoking has emerged as a significant vector for tobacco use, particularly among the youth (2). The emergence of waterpipe use globally may be partly fueled by the perception that it is a safer and nonaddictive alternative to cigarette smoking. In addition, attitudes towards waterpipe smoke (WPS) use are generally more accepting than for cigarette smoke (CS; ref. 3). However, contrary to popular belief, during a typical 1 hour use session, the waterpipe emits and delivers high doses of nicotine, cardiovascular and pulmonary toxicants, and carcinogens to the user (4, 5) and surroundings (6). In addition, studies have shown that the manifold “tar” and polycyclic aromatic hydrocarbons yielded by a single waterpipe smoking session exceed those generated by a single cigarette (4, 5).
The extensive research on CS over the past several decades has established clear causal links between CS and lung cancer (7). It is estimated that there are more than 1.3 billion tobacco users worldwide who are at elevated risk for lung disease, including lung adenocarcinoma, the most common lung cancer subtype (8). Similar to CS, exposure to WPS is associated with increased pulmonary inflammation and deleterious effects on both the respiratory (e.g., chronic bronchitis, emphysema) and cardiovascular systems (9, 10). Furthermore, previous reports in murine models showed that WPS activates a prominent inflammatory response in the host lung (11–13). Yet, most of those studies interrogated WPS in the acute or subacute settings, and reports on the effects of chronic waterpipe exposure, remain limited (12–15). Importantly, effects of chronic exposure to WPS on lung carcinogenesis are unknown.
To begin to address this gap, in this study we sought to investigate the effects of chronic WPS exposure on lung pathobiology and host immunity in an animal model and in human participants. We compared lung immune responses and tumorigenesis in mice exposed to a 1 hour WPS session for 20 weeks in a whole-body exposure system, once a week (light WPS) or five times a week (heavy WPS), relative to control mice exposed to ambient air. We also compared the transcriptomes of normal-appearing airway cells from bronchoscopy samples in human waterpipe smokers and nonsmokers.
Materials and Methods
Smoke exposure system
A whole-body WPS exposure system was designed and built for this study at the American University of Beirut (AUB; Beirut, Lebanon) Aerosol Research Lab (Fig. 1). Smoke was drawn intermittently by a positive pressure smoking machine and discharged into a transparent polycarbonate chamber (38 × 25 × 25 cm) containing the animals. During each 1 hour exposure session, each puff was entirely discharged into the chamber. Ambient air was continuously pumped into the chamber at a flow rate of 0.5 liters per minute (LPM), resulting in approximately 1.25 air changes per hour, not including the intermittent smoke flow. A carbon monoxide (CO) analyzer (Bacharach Monoxor III) drawing a sample flow of 0.3 LPM was used to monitor CO levels in the chamber, and a 3-cm diameter fan was suspended from the top to ensure uniform aerosol concentrations throughout the chamber. A glass fiber filter (Gelman Type A/E) was placed upstream of the CO monitor to allow gravimetric determination of the mean total particulate matter concentration during each exposure session. For the control group, the smoking machine was switched off and animals were exposed to ambient air only.
Experimental design and waterpipe exposure apparatus. Schematic design depicting the whole-body waterpipe smoke exposure system. Smoke is drawn intermittently by a positive pressure smoking machine and discharged into a transparent polycarbonate chamber (38 × 25 × 25 cm) containing the animals. During the exposure event, ambient air is continuously piped into the chamber at a flow rate of 0.5 LPM. A CO analyzer (Bacharach Monoxor III) is used to monitor the CO levels in the chamber, and a fan is suspended to ensure uniform aerosol concentrations throughout the chamber. A glass fiber filter (Gelman Type A/E) is placed upstream of the CO monitor to allow gravimetric determination of the mean total particulate matter concentration during each exposure session. For the control group, the smoking machine remains off during each exposure session, and animals are exposed to ambient air only. Bottom, Schematic timeline depicting the three experimental exposure groups composed of 8 weeks–old Gprc5a–/–;Lcn2–/– mice divided into groups of up to 5 mice (per exposure group) and studied at the end of exposure, and 20 weeks after completion of exposure. Mice were exposed to WPS for once a week (light WPS) or five times a week (heavy WPS).
Experimental design and waterpipe exposure apparatus. Schematic design depicting the whole-body waterpipe smoke exposure system. Smoke is drawn intermittently by a positive pressure smoking machine and discharged into a transparent polycarbonate chamber (38 × 25 × 25 cm) containing the animals. During the exposure event, ambient air is continuously piped into the chamber at a flow rate of 0.5 LPM. A CO analyzer (Bacharach Monoxor III) is used to monitor the CO levels in the chamber, and a fan is suspended to ensure uniform aerosol concentrations throughout the chamber. A glass fiber filter (Gelman Type A/E) is placed upstream of the CO monitor to allow gravimetric determination of the mean total particulate matter concentration during each exposure session. For the control group, the smoking machine remains off during each exposure session, and animals are exposed to ambient air only. Bottom, Schematic timeline depicting the three experimental exposure groups composed of 8 weeks–old Gprc5a–/–;Lcn2–/– mice divided into groups of up to 5 mice (per exposure group) and studied at the end of exposure, and 20 weeks after completion of exposure. Mice were exposed to WPS for once a week (light WPS) or five times a week (heavy WPS).
Smoke generation protocol
A waterpipe (53-cm overall height) fitted with a leather hose with an infiltration rate below 2 LPM, as previously measured per Saleh and Shihadeh (16), was used in this study. Prior to each session, the waterpipe was loaded with 10 g of “two apples” flavor, “Nakhla” brand tobacco. A single quick light charcoal disk (“Three Kings” brand) was placed on top of the tobacco mix prior to smoking, and an additional half-disk was added at the 105th puff. The smoking machine was programmed to draw 171 puffs of a 2.6-second duration and 17-second interpuff interval, at a mean flow rate of 12 LPM, as per the Beirut Protocol (17). Further details about the waterpipe preparation and the charcoal application were previously reported (18).
Animal housing and WPS exposure experiments
Animal studies were conducted under institutional animal care and use committee (IACUC)-approved protocols. Gprc5a–/–;Lcn2–/− mice were generated as previously described (8), and 8 to 9 weeks–old mice (males and females) were divided into groups of up to 5 mice (per exposure group and time point) and were exposed (full-body) during a 1 hour session to control air or to light or heavy WPS (one time or five times per week, respectively) for 20 weeks. Different groups of mice were humanely euthanized immediately at the end of exposure for immune analyses or after 20 weeks following end of exposure for sequencing and histopathologic assessment of lungs.
Histopathologic analysis of mouse lung lesions
Lungs were inflated with formalin, excised, and macroscopically examined for surface lesions before being processed for histopathologic assessment. Formalin-fixed lungs were embedded in paraffin, and four deep sections (4-μm each) were acquired from each fixed lung, stained with hematoxylin and eosin (H&E), digitally scanned using the Aperio ScanScope Turbo slide scanner (Leica Microsystems Inc.) at 200X magnification, and visualized by ImageScope software (Leica Microsystems, Inc.). Images were analyzed by a well-experienced pathologist who was blinded to groups to identify the tumors.
Flow cytometry
Lungs were perfused with PBS containing 2% FCS, excised and finely minced, followed by enzymatic digestion for 30 minutes at 37°C in PBS containing 150 U/mL collagenase type IV and 20 U/mL DNase type I. Lung homogenates were suspended in a 20% Percoll gradient and centrifuged. Pellets were then washed, and red blood cells lysed. Cells were incubated in RPMI 10% FCS containing Golgi Plug/Golgi Stop for 2 hours at 37°C. Cells were then stained for 30 minutes with fluorescence-conjugated antibodies (Biolegend or Becton Dickinson) diluted in PBS 2% FCS: anti-CD45 (AF700-conjugated), anti-Ly6G (FITC-conjugated), anti-CD11b (PerCpCy5.5-conjugated), anti-SiglecF (APCCy7-conjugated), anti-CD11c (FITC-conjugated), anti-MHCII (PerCpCy5.5-conjugated), anti-F4/80 (PeCy7-conjugated), anti-CD4 (FITC-conjugated), anti-CD8 (APCCy7-conjugated), anti-B220 (PerCpCy5.5-conjugated), and anti–PD-L1 (PeCy7-conjugated). Cells were washed and fixed, permeabilized, and stained with PE-conjugated antibodies against pro-IL1β (Thermo Fisher Scientific) or IL17A (Biolegend), and analyzed on a BD Aria cell sorter. Flow cytometry analyses were performed using the Diva software.
Collection of human bronchial epithelial brushings
Participants were identified prior to undergoing routine bronchoscopy at the Department of Internal Medicine at AUB Medical Center (AUBMC). Eligible patients were selected from a cohort of prospectively acquired bronchial brushings from consented patients with or without lung lesions (suspect or already cytopathologically confirmed to be malignant) who are undergoing routine diagnostic bronchoscopy at the Department of Internal Medicine at AUBMC under a protocol approved by the Institutional Review Board. Written informed consents were obtained from all participants and the studies were conducted in accordance with the ethical guidelines of the Declaration of Helsinki. Inclusion criteria included: (i) nonsmokers, former, or current smokers (cigarette, cigar, waterpipe) who are scheduled to undergo bronchoscopy; (ii) age ≥ 18 years old; (iii) no intercurrent illness such as another active malignancy, and active infectious bronchitis at time of bronchoscopy. Exclusion criteria included: (i) pregnant females; (ii) contraindications to fiberoptic bronchoscopy including hemodynamic instability, decompensated heart failure, recent myocardial infarction within 6 weeks, uncontrolled arrythmias, high risk for respiratory failure, uncontrolled asthma, severe obstructive airway disease (as determined by spirometry); (iii) contraindications to brushing/biopsy: patients at risk of bleeding (e.g., with known bleeding disorders and abnormal coagulation platelets count less than 50,000; international normalized ratio >1.5, partial thromboplastin time >50 seconds); (iv) suspected active pulmonary tuberculosis; (v) known human immunodeficiency virus (HIV); (vi) patients who are unable to consent. Bronchoscopy and collection of airway epithelia from bronchial brushings were done as previously described (19).
Participants were surveyed for demographic characteristics, bronchoscopy indication, presence of underlying lung disease, malignancy diagnosis (type and date of diagnosis status), occupational/environmental exposures and smoking habits (type, duration, and intensity) using investigator-administered questionnaires. Individuals who have never smoked a waterpipe or have smoked less than 100 waterpipes in their lifetime were categorized as waterpipe nonsmokers. Waterpipe smokers included those who smoked at least 100 waterpipes during their life and are currently smoking and those who quit smoking less than 12 months before enrollment. Individuals who smoked at least 100 waterpipes during their life and quit smoking more than 12 months before enrollment were considered former waterpipe smokers. We studied 5 lung cancer–free individuals who never smoked waterpipe and were categorized as waterpipe nonsmokers and 4 lung cancer–free individuals who smoked at least 100 waterpipes in their lifetime, including during the 12-month period prior to enrollment, categorized as waterpipe smokers. One waterpipe nonsmoker was a former light cigarette smoker for 2 years and quit 31 years prior to enrollment. Age, sex, and smoking history of those 9 participants are summarized in Supplementary Table S1.
ELISA
Bronchoalveolar lavage fluid (BALF) was collected from mice exposed to control air, light, and heavy WPS. IL17A (eBiosciences) levels in the supernatants were measured by ELISA as described previously.
Total RNA isolation from mouse lung tissues and human airway epithelia
Total RNA was isolated from homogenized tissues (mouse lungs and human bronchial airway epithelia) using Qiagen RNeasy mini kit (Qiagen) according to manufacturer's recommendations. Qubit 4 Fluorometer (Thermo Fisher Scientific) was used for measurement of RNA yield, and RNA quality was assessed by NanoDrop 1000 (Thermo Fisher Scientific) as well as the Agilent RNA 6000 Pico Kit (Agilent) for computation of RNA integrity numbers.
RNA sequencing and analysis
At least 10 ng of input RNA were used for generation of barcoded cDNA transcriptome libraries using the SMARTer seq V3, Takara Bio Kit (Takara) adhering to the manufacturer's protocol. Libraries were then sequenced using NovaSeq 6000 platform from Illumina at a target coverage of 40 million reads per sample. All raw data are uploaded to the Gene Expression Omnibus (GEO). Alignment and mapping was performed using STAR (20). Genes with no coverage in more than 50% of samples were filtered out. DESEQ2 (21) was applied for identification of differentially expressed genes (DEG) using P values that were adjusted using Benjamini–Hochberg methods. DEGs were analyzed by pathways analysis or topologically organized into gene networks using Ingenuity Pathways Analysis (IPA; 8).
Pathways and gene network analysis as well as in silico assessment of gene signatures
Following RNA sequencing (RNA-seq) of Gprc5a–/–;Lcn2–/– mouse lung tissues at 20 weeks following end of control air or heavy WPS exposure, DEGs were organized into pathways and topological gene–gene interaction networks using Ingenuity Pathways Analysis as described previously (22). Immune signatures were curated from the previous reports: CTL derived by Jiang and colleagues (Cd8a, Cd8b, Gzma, Gzmb, Prf1; ref. 23); T cell expanded signature by Ayers and colleagues (Expanded signature; Cd3d, Ido1, Ciita, Cd3e, Ccl5, Gzmk, Cd2, Hla-Dra, Cxcl13, Il2rg, Nkg7, Hla-E, Cxcr6, Lag3, Tagap, Cxcl10, Stat1, Gzmb; ref. 24); and T effector signature by McDermott and colleagues (Cd8A, Eomes, Prf1, Ifng, Cd274; ref. 25). We also interrogated the expression of a gene signature that is prognostic in lung adenocarcinoma (LUAD) and previously published by Wang and colleagues (ARNTL2, PLA2G4F, CYP17A1, LOXL2, HMMR, NKX2–5, GJB2, DSG3, SIX1, MC4R, TRPA1, RAET1L, KLF4, HSF2BP, GRIP1, PTPRN, MAPK4, PLOD2; ref. 26) in our RNA-seq cohort of airways of waterpipe smokers and nonsmokers. For each of the signatures, the average of normalized log-transformed values of the signature's gene members were computed and statistically compared between samples using the Wilcoxon rank-sum test. Survival analysis was performed in treatment-naïve patients with lung adenocarcinoma (n = 508) from the publicly available The Cancer Genome Atlas (TCGA) cohort (27), and from the Profiling of Resistance Patterns and Oncogenic Signaling Pathways in Evaluation of Cancers of the Thorax dataset (PROSPECT; n = 152; ref. 28). Patients were dichotomized into upper and lower quartiles based on median values of the prognosis signature. Statistical differences in survival between different groups were evaluated in R using the log-rank test and the Kaplan–Meier method for estimation of survival probability.
Statistical analysis
GraphPad Prism software and the R statistical language and environment (r-project.org) were used for statistical analyses. Pearson correlation coefficients were used to statistically examine correlations between continuous variables. To establish statistical significance of differences in specific contrasts between two groups, we used Student t test or Wilcoxon rank–sum test. To evaluate statistical differences among three or more groups, we used ANOVA or Kruskal–Wallis tests. Statistical significance was achieved with P values less than 0.05.
Data availability
Raw RNA-seq data of lungs of Gprc5a–/–;Lcn2–/– mice at 20 weeks following end of exposure to control air or heavy WPS exposure were deposited into GEO as GSE197994. Raw RNA-seq data of airway epithelia from bronchial brushings of patients undergoing bronchoscopy were deposited into GEO as GSE197653.
Results
Effects of chronic waterpipe exposure on lungs of mice
To begin to understand the impact of chronic WPS exposure on lung cancer development, a whole-body WPS exposure system was exclusively designed for this study. We studied Gprc5a–/–;Lcn2–/– mice, which we previously showed to be highly susceptible to nicotine-specific carcinogen-mediated lung cancer development (8), despite being of the C57/BL6 strain that is resistant to tobacco carcinogens (29). Mice were exposed to whole-body WPS or control ambient air. The smoking machine was programmed per the Beirut protocol (17). Mice were chronically exposed to one (light WPS) or five (heavy WPS) of these smoking sessions per week, and for 20 weeks (Fig. 1). At 20 weeks following end of exposure, histopathologic analysis revealed lung tumor formation in animals exposed to WPS but not in animals exposed to control air (Fig. 2A). Fraction of mice with lung tumor development was higher in mice with heavy (5 days a week) relative to light (once a week) WPS exposure or control air-exposed animals (P < 0.0001 of the χ2 test; Fig. 2B). The higher incidence in the heavy WPS-exposed group was complemented by a higher average total tumor volume compared with light WPS-exposed animals (Supplementary Table S2). Furthermore, we found Ki-67 nuclear staining in cells within lung tumors suggesting a proliferative feature of these lesions (Fig. 2C). These findings suggest that WPS exposure promotes lung tumor formation. They also demonstrate that lung tumor formation is increased with more frequent WPS exposure.
Effects of chronic waterpipe exposure on lungs of mice. A, Representative histopathologic (H&E) images of lungs from control air, light WPS, and heavy WPS-exposed mice with lesions indicated by black arrows. The scale bars denote 1 mm (top) and 100 μmol/L (bottom). B, Fraction of mice developing lung lesions following control air, light WPS, and heavy WPS exposure. C, Representative photomicrographs of positive Ki-67 staining in tumor lung tissues of mice exposed to WPS (indicated by yellow arrows) and PD-L1 staining on epithelial cells in tumors (indicated by green arrows).
Effects of chronic waterpipe exposure on lungs of mice. A, Representative histopathologic (H&E) images of lungs from control air, light WPS, and heavy WPS-exposed mice with lesions indicated by black arrows. The scale bars denote 1 mm (top) and 100 μmol/L (bottom). B, Fraction of mice developing lung lesions following control air, light WPS, and heavy WPS exposure. C, Representative photomicrographs of positive Ki-67 staining in tumor lung tissues of mice exposed to WPS (indicated by yellow arrows) and PD-L1 staining on epithelial cells in tumors (indicated by green arrows).
Exposure to WPS modulates host lung immune responses
We were prompted to examine changes in the lung immune microenvironment by WPS exposure and prior to LUAD formation (Fig. 1). Flow cytometry analysis of mouse lung mononuclear cells revealed a significant decrease in dendritic cells in mice exposed to heavy WPS compared with control air (P < 0.05; Fig. 3A). In contrast to ambient air-exposed mice, littermates with WPS exposure showed significantly increased CD4+ T cell counts (P < 0.05; Fig. 3A). These effects were more pronounced in mice with heavy WPS exposure (Fig. 3A). Mice exposed to light WPS also showed significantly elevated CD8+ T cell counts (P < 0.05; Fig. 3A). We noted no significant differences in neutrophil, macrophage, alveolar macrophage, and B cell counts across the three different exposure groups (Fig. 3A). We then probed the expression of proinflammatory immune markers. We found that IL17A protein expression in CD4+ and CD8+ T as well as B cells was significantly enriched in lungs of mice exposed to WPS compared with control air counterparts (P < 0.05; Fig. 3B). We concordantly found significantly increased levels of IL17A protein in BALF of WPS-exposed mice compared with control air-exposed littermates (P < 0.05 for the heavy WPS group; Fig. 3C). We noted trends for increased IL17A protein in lung and BALF of heavy relative to light WPS-exposed mice (Fig. 3B and C). We also found that levels of the proinflammatory cytokine pro-IL1β in neutrophils (P < 0.05) and dendritic cells (P < 0.01) were significantly increased following chronic exposure to light or heavy WPS (Supplementary Fig. S1A). Intriguingly, expression of the major protumor immune checkpoint programmed cell death marker ligand (PD-L1) was significantly increased in neutrophils and macrophages following light or heavy WPS exposure (P < 0.05; Supplementary Fig. S1B). Furthermore, we also noted PD-L1 staining on epithelial cells in tumors (Fig. 2C).
Exposure to WPS induces an inflammatory response in lungs of mice. A, Flow cytometry analysis of the frequencies of various immune cells after 20 weeks of exposure to control air, light WPS, and heavy WPS in lungs of Gprc5a–/–;Lcn2–/– mice. B, Flow cytometry analysis of IL17A expression in CD4+ T cells, CD8+ T cells and B cells after 20 weeks of exposure to control air, light WPS, and heavy WPS in lungs of Gprc5a–/–;Lcn2–/– mice. C, IL17A ELISA of BALF obtained from Gprc5a–/–;Lcn2–/– cells after 20 weeks of exposure to control air, light WPS, and heavy WPS. Differences between two groups were statistically examined using the Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Exposure to WPS induces an inflammatory response in lungs of mice. A, Flow cytometry analysis of the frequencies of various immune cells after 20 weeks of exposure to control air, light WPS, and heavy WPS in lungs of Gprc5a–/–;Lcn2–/– mice. B, Flow cytometry analysis of IL17A expression in CD4+ T cells, CD8+ T cells and B cells after 20 weeks of exposure to control air, light WPS, and heavy WPS in lungs of Gprc5a–/–;Lcn2–/– mice. C, IL17A ELISA of BALF obtained from Gprc5a–/–;Lcn2–/– cells after 20 weeks of exposure to control air, light WPS, and heavy WPS. Differences between two groups were statistically examined using the Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
We then performed RNA-seq analysis of lungs of Gprc5a–/–;Lcn2–/– mice 20 weeks following exposure to control air or heavy WPS (Fig. 1). We identified 4,262 gene transcripts that were differentially and significantly modulated (based on an adjusted P threshold of 0.01 and a fold change of at least 1.5) between heavy WPS and control air-exposed mice (Fig. 4A). Lungs of mice heavily exposed to WPS exhibited overall markedly increased activation of oxidative phosphorylation genes (P < 0.05; Fig. 4B). Conversely, we noted significant inhibition of phagosome formation in heavy WPS-exposed animals (P < 0.05; Fig. 4B) which was consistent with our observation of reduced counts of dendritic cells by flow cytometry following WPS (Fig. 3A). We then analyzed immune gene sets curated from the literature (23–25). We found significantly attenuated CTL (P < 0.05), expanded immune (P < 0.05), and T-effector (P < 0.001) signatures following heavy WPS exposure compared with control air (Fig. 4C), suggesting that while T cell counts may have been increased in WPS-exposed animals (Fig. 3A), they may be dysfunctional or exhausted (30). Intriguingly, we found, among genes that were differentially expressed in mice heavily exposed to WPS, increased expression (P < 0.001) of Tmprss4, Cd55, and Ace2 which are known to be important for SARS-CoV-2 viral entry into host airway cells and contraction of COVID-19 (Fig. 4D). Our data collectively suggest that WPS exposure suppresses key antitumor immune functions and modulates the host lung immune contexture by skewing it to a protumorigenic inflammatory state.
Waterpipe exposure suppresses key antitumor immune functions and elevates expression of receptors for the novel coronavirus SARS-CoV-2. A, Heatmap showing differentially expressed transcripts between lungs exposed to control air or heavy WPS (HWPS) and identified by RNA-seq. Columns indicate samples and rows denote transcripts (red, relatively upregulated; blue, downregulated). B, Enriched pathways in heavy WPS-exposed lung tissues were identified using IPA. Activation of the pathway is indicated by z-scores (red, upregulated; blue, downregulated). C, The following immune cell signatures indicative of different phenotypes were propagated and computed for each sample as described in Methods: CTL, expanded immune, and T-effector signature. Immune cell signatures were then statistically compared between control air- (blue) and heavy WPS-exposed (red) lung tissues using the Wilcoxon rank sum test. D,Tmprss4, Cd55, and Ace2 mRNA expression levels were statistically compared between control air- (blue) and heavy WPS-exposed (red) mice using the Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Waterpipe exposure suppresses key antitumor immune functions and elevates expression of receptors for the novel coronavirus SARS-CoV-2. A, Heatmap showing differentially expressed transcripts between lungs exposed to control air or heavy WPS (HWPS) and identified by RNA-seq. Columns indicate samples and rows denote transcripts (red, relatively upregulated; blue, downregulated). B, Enriched pathways in heavy WPS-exposed lung tissues were identified using IPA. Activation of the pathway is indicated by z-scores (red, upregulated; blue, downregulated). C, The following immune cell signatures indicative of different phenotypes were propagated and computed for each sample as described in Methods: CTL, expanded immune, and T-effector signature. Immune cell signatures were then statistically compared between control air- (blue) and heavy WPS-exposed (red) lung tissues using the Wilcoxon rank sum test. D,Tmprss4, Cd55, and Ace2 mRNA expression levels were statistically compared between control air- (blue) and heavy WPS-exposed (red) mice using the Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Airways of waterpipe smokers exhibit gene signatures prognostic of lung adenocarcinoma
Previous work from our group and others revealed that cigarette smoke (CS)-exposed normal-appearing airways in patients with LUAD carry alterations that are characteristic of the nearby tumors (31), suggestive of a field of injury that is pertinent to carcinogenesis. Also considering our murine findings on lung tumor development and immune modulation following WPS exposure, we were thus prompted to interrogate normal-appearing airways of human cancer-free waterpipe smokers relative to those who did not smoke waterpipe. To this end, we studied bronchial brushings from patients undergoing diagnostic bronchoscopy including 4 lung cancer–free smokers (smoked at least 100 waterpipes in their lifetimes) and 5 lung cancer–free nonsmokers (never smoked a waterpipe or smoked less than 100 waterpipes). The nonsmoker group comprised CS nonsmokers except for one case who was a light former CS smoker with less than 2 lifetime pack-years who quit 20 years ago. (Supplementary Table S1). We analyzed gene expression of airway epithelia from the bronchial brushings of those subjects by RNA-seq (Fig. 5A). We identified 168 gene transcripts (Supplementary Table S3) that were significantly modulated (based on an adjusted P threshold of 0.05) between waterpipe smokers and nonsmokers (Fig. 5B). Normal bronchial brushings of waterpipe smokers showed increased activation of the NRF-2–mediated oxidative stress response pathway (Fig. 5C), consistent with our findings in mice heavily exposed to WPS (Fig. 4B).
Gene signatures related to lung cancer aggressiveness in airways of human waterpipe smokers. A, Schematic depicting human cohort selection for RNA-seq of bronchial brushings among those undergoing bronchoscopy. B, Heatmap showing differentially expressed transcripts between human airways of waterpipe smokers compared with nonsmokers. Columns indicate patient airways and rows indicate transcripts (red, relatively upregulated; blue, downregulated). C, Topological network analysis of genes associated with NRF-2–mediated oxidative stress in human airways of waterpipe smokers was performed using IPA. Predicated activation of NRF-2–mediated oxidative stress based on the gene set is indicated by the orange color. D, Left: Kaplan–Meier of OS of patients from TCGA and PROSPECT cohorts stratified by median expression of the WPS-associated prognosis signature (red, relatively higher than the median; blue, lower than the median). Right: Lung adenocarcinoma (LUAD) prognosis signature was propagated and computed as described in Methods. The prognosis signature was then statistically compared between nonsmokers (blue) and waterpipe smokers (red) using the Wilcoxon rank-sum test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Gene signatures related to lung cancer aggressiveness in airways of human waterpipe smokers. A, Schematic depicting human cohort selection for RNA-seq of bronchial brushings among those undergoing bronchoscopy. B, Heatmap showing differentially expressed transcripts between human airways of waterpipe smokers compared with nonsmokers. Columns indicate patient airways and rows indicate transcripts (red, relatively upregulated; blue, downregulated). C, Topological network analysis of genes associated with NRF-2–mediated oxidative stress in human airways of waterpipe smokers was performed using IPA. Predicated activation of NRF-2–mediated oxidative stress based on the gene set is indicated by the orange color. D, Left: Kaplan–Meier of OS of patients from TCGA and PROSPECT cohorts stratified by median expression of the WPS-associated prognosis signature (red, relatively higher than the median; blue, lower than the median). Right: Lung adenocarcinoma (LUAD) prognosis signature was propagated and computed as described in Methods. The prognosis signature was then statistically compared between nonsmokers (blue) and waterpipe smokers (red) using the Wilcoxon rank-sum test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
We also interrogated the expression of a previously reported gene signature that is prognostic in LUAD (ARNTL2, PLA2G4F, CYP17A1, LOXL2, HMMR, NKX2–5, GJB2, DSG3, SIX1, MC4R, TRPA1, RAET1L, KLF4, HSF2BP, GRIP1, PTPRN, MAPK4, PLOD2) in our RNA-seq cohort of airways of waterpipe smokers and nonsmokers (26). We found that this gene set was indeed associated with reduced overall survival (OS) when we analyzed LUADs from TCGA (P = 0.0007; Fig. 5D, left panel) and with reduced progression-free interval in our previously reported PROSPECT cohort (ref. 28; P = 0.03; Fig. 5D, middle panel) datasets. Notably, we found that these genes were enriched in airways of waterpipe smokers relative to nonsmokers (P < 0.05; Fig. 5D, right panel). Our analyses underscore transcriptomic alterations in normal-appearing airways of cancer-free waterpipe smokers that are indicative of LUAD pathogenesis. Intriguingly, and similar to our findings in mice exposed to heavy WPS, we found increased expression (P < 0.01) of TMPRSS4 in airway epithelia of waterpipe smokers. We also noted increased levels of CD55 and ACE2 in airways of waterpipe smokers albeit not reaching statistical significance (Supplementary Fig. S2).
Discussion
Research efforts focused on the effects of waterpipe smoking on the lung remain sparse despite its worldwide emergence as a socially acceptable form of smoking, particularly among the youth (2). There is a pressing need to better understand the effects of this form of smoking on lung pathophysiology, particularly lung cancer. Here, we interrogated the effects of WPS exposure on mice known to be susceptible to carcinogen mediated LUAD development. We found that WPS exposure was associated with increased lung tumor formation in mice and was more pronounced following a heavier WPS exposure regimen. Lungs of WPS-exposed mice also exhibited a modulated host lung immune microenvironment that favored a protumorigenic inflammatory state. We further assessed the effects of WPS exposure on normal-appearing airways of cancer-free waterpipe smokers. We found that transcriptomes of normal airway cells from waterpipe smokers, relative to nonsmokers, exhibited pronounced immunomodulation as well as genes that were associated with poor survival among patients with LUAD: thereby supporting the supposition that waterpipe smokers might be at a heightened risk of developing lung cancer.
We found that mice subjected to WPS were more likely to develop lung tumors. Further validation is needed to establish an empirical link between WPS and lung cancer development, in additional animal models. While we found significantly increased CD4 and CD8 T cell counts in lungs of WPS-exposed mice, our RNA-seq analysis showed that signatures of cytotoxic or expanded immune response were, conversely, decreased in animals exposed to heavy WPS relative to control air. In spite that these observations may at first appear contrary to each other, they perhaps suggest that T cell populations in WPS-exposed animals may be dysfunctional or exhausted. Our findings and supposition are in line with previous reports showing that cigarette smoking exerts immunosuppressive effects on the tumor immune microenvironment by hampering cytotoxic T cells and causing a predominant state of T cell anergy (30). Levels of the proinflammatory cytokines IL17, and pro-IL1β, were also increased post-WPS exposure. IL17-mediated inflammatory responses have been shown to be key in the progression of cancer, particularly lung tumors (32). Our findings resonate with trends observed in our murine model which had increased production of IL17A by CD4+ and CD8+ T cells during tumor development upon exposure to chemical-induced carcinogenesis (8). Moreover, evidence suggests a pivotal role for IL1β in inflammation and possibly in lung cancer pathogenesis (33). Indeed, the IL1β antibody canakinumab significantly reduced lung cancer incidence and mortality in a high-risk population (34). Our findings are also in line with previously reported effects of tobacco and waterpipe exposure, such as the activation of host defense inflammatory mechanisms (11–13, 15) with protumor effects (35). Another worthy observation was the increased expression of the major immune checkpoint PD-L1 protein in neutrophils and macrophages following WPS exposure. PD-L1 plays important roles in immune suppression and evasion, and anti–PD-L1 antibodies are approved for immunotherapy of lung cancer (36). Indeed, a study investigating the effects of chronic waterpipe exposure described immune suppressive effects that are comparable to our observations (37). It is thus plausible that WPS exposure leads to early immune suppression and evasion which promote lung carcinogenesis.
When analyzing the transcriptome of airways of cancer-free patients exposed to WPS, we found significant activation of genes related to the NRF-2–mediated oxidative stress response, in line with a previous report probing epigenetic changes in the small airway epithelium of waterpipe smokers (14). This was also concordant with the activation of oxidative phosphorylation pathways in lungs of mice heavily exposed to WPS. Indeed, smoking has been shown to induce the production of reactive oxygen species by oxidative phosphorylation (38). Furthermore, our findings allude to a possible molecular airway field of injury similar to that established, at the transcriptomic level, in response to CS (31, 39, 40). Our data also suggest that transcriptomes of normal airway cells in waterpipe smokers, relative to waterpipe nonsmokers, harbor gene programs that are associated with poor clinical outcome in patients with LUAD, also suggestive of a WPS-associated molecular injury that may inform of LUAD pathogenesis. Thus, we provide evidence, albeit preliminary, that WPS may elicit an airway field of injury that, based on its molecular (e.g., gene signatures) and immune (e.g., proinflammatory milieu) features, favors lung carcinogenesis. It is noteworthy to point that in mice heavily exposed to WPS, we found increased expression of Tmprss4, Cd55, and Ace2, key mediators for the entry of SARS-CoV-2 virus implicated in COVID-19 disease (41). Upregulation of pulmonary ACE2 in smokers has been previously reported and suggested to increase susceptibility for SARS-CoV-2 infection (42). For instance, cigarette smoking is known to induce goblet cells metaplasia (main source of ACE2 in the lungs) by epigenetic alterations (43), partly explaining the increased expression of ACE2 in smokers. Therefore, it is not unlikely that waterpipe smokers might also be at higher risk of infection by mechanisms that are yet to be discovered. This assumption requires further validation to cognize its relevance, especially that the effect of smoking on the development of COVID-19 pneumonia remains controversial (44).
Our study incurred several limitations. Using our mouse model, we show increased lung tumor development following a 1 hour waterpipe session, loaded with a particular type of tobacco and charcoal. Those findings ought not to be underappreciated but should be interpreted cautiously as further studies are warranted to replicate our findings using different types of waterpipe tobacco and charcoal, as well as in other mouse models. Indeed, comparison of WPS exposure to positive controls with well-known carcinogenic effects in the lung (such as tobacco carcinogen as chemical or inhaled cigarette smoke) is also warranted to better corroborate our findings. While we show increased tumor development in mice exposed to waterpipe using our whole-body exposure system, our findings will need to be validated using systems that better simulate human waterpipe smoking. For instance, exposure systems that can ensure enhanced inhalational efficacy while minimizing particulate deposition on animal fur are desired to draw solid conclusions from murine models. Furthermore, behavioral studies on waterpipe consumption show that a single waterpipe session can last longer than an hour (45), making our employed smoking protocol ungeneralizable to all WPS regimens. It is thus plausible to surmise that different smoking protocols using distinct combinations of different waterpipe tobacco and charcoal types can lead to various other molecular (e.g., gene signatures) and immune (e.g., proinflammatory milieu) effects. While we provide support for the hypothesis that WPS exposure increases lung cancer risk in humans, based on lung tumor-associated gene profiles in exposed epithelia, this supposition warrants further validation in a larger cohort of nonsmokers and exclusive waterpipe smokers. In addition, and due to the worldwide spread of waterpipe smoking, more efforts are needed to interrogate WPS cohorts that extend beyond Middle Eastern countries which have constituted the major geographic focus in published WPS reports thus far (9–12, 15, 46, 47). Furthermore, epidemiologic studies suggest that many smokers are exposed to both CS and WPS throughout their lifetime (46, 47). This “concoction” of exposures may lead to an increased susceptibility to various lung diseases, such as lung cancer, compared with exclusive waterpipe or cigarette smoking. Therefore, future studies are warranted to better elucidate the consequences of being exposed to a combination of both waterpipe and CS.
In conclusion, we show that WPS exposure leads to lung tumor formation in mice and elicits molecular and immune features that are informative of the pathogenesis of lung cancer, and, thus, risk for the disease. Additional studies are needed to fully assess procarcinogenic effects of WPS, and to demand a resurgence of attention to this topic in terms of evidence-based science policy enactment, which includes but is not limited to regulation of WPS products’ consumer age and availability, packaging warning labels, and the use of WPS in public spaces (48).
Authors’ Disclosures
H. Chami reports grants from Pfizer outside the submitted work. A.L. Shihadeh reports grants from NIH/FDA during the conduct of the study; personal fees from litigation against the tobacco industry (consultant) outside the submitted work; in addition, A.L. Shihadeh has a patent for WO2017023589A1. H. Kadara reports grants from Johnson and Johnson outside the submitted work. No disclosures were reported by the other authors.
Authors’ Contributions
M. Hassane: Data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. Z. Rahal: Data curation, formal analysis, investigation, visualization, writing–original draft, writing–review and editing. N. Karaoghlanian: Resources, software, investigation, methodology. J. Zhang: Data curation, formal analysis, validation, investigation. A. Sinjab: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. J.W. Wong: Data curation, formal analysis. W. Lu: Data curation, formal analysis. P. Scheet: Data curation, formal analysis. J.J. Lee: Resources, data curation, formal analysis. M.G. Raso: Data curation, formal analysis, investigation. L.M. Solis: Data curation, formal analysis, investigation, methodology. J. Fujimoto: Data curation, formal analysis, validation, investigation, visualization. H. Chami: Resources, data curation, formal analysis, validation, investigation, writing–original draft, project administration, writing–review and editing. A.L. Shihadeh: Resources, data curation, formal analysis, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. H. Kadara: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, writing–original draft, project administration, writing–review and editing.
Acknowledgments
Funded in part by the NCI (grant no. R01CA205608 to H. Kadara) and a medical practice plan (MPP) grant from the American University of Beirut (to H. Chami).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.