Introduction
The use of electronic nicotine delivery systems (ENDS, also called e-cigarettes or vaping) has surged in recent years, with over 10 million adults and 3 million adolescents actively using ENDS [
1]. The FDA commissioner has used the term “epidemic” levels to describe the ubiquity of vaping. However, a recent spate of hospitalizations and deaths in 2019 due to Ecig or vaping-associated acute lung injury (EVALI), as well as combined use with traditional cigarettes, suggests that ENDS may pose unanticipated risks to lung health [
2]. While new products and components are being incorporated into vaping products, very little is known about specific effects of vaping on lung function, particularly impacts to immune function.
It is well known that inhaled toxicants including traditional cigarette smoking have significant impacts on immunity, including damage to respiratory epithelium [
3,
4], as well as increased chronic inflammation with increased susceptibility to viral and bacterial respiratory infections [
5]. Vaping has been associated with increased risk of COVID-19 [
6‐
8], and experimental evidence shows that vaping in animal models increases vulnerability to influenza A virus (IAV) [
9], and predisposes airway epithelial cells to bacterial infection [
10]. However, the precise effects of vaping and ENDS liquids on immune cells and immune function remain understudied.
ENDS products predominately use a base carrier propylene glycol (PG) plus vegetable glycerin (VG) to act as one of the most common carriers for vaping liquids [
11‐
14]. PG and VG are also widely used in the food and cosmetics industries, and effects of PG and/or VG on multiple biological systems including renal and respiratory function have been documented [
15‐
21]. PGVG is used in combination with cutting agents such as Vitamin E acetate (VEA), which has been identified as the likely causal driver of EVALI, and thus its use mostly discontinued [
1,
2,
22,
23]. Alternatives to VEA include Phytol which is a diterpene alcohol that may naturally exist in cannabis products in small amounts, but it was an inexpensive diluent that many manufacturers of cannabis products were considering incorporating into products as a principal ingredient and was found in counterfeit cartridges [
24]. An initial study highlighted significant pulmonary toxicity from phytol in rats and, for the most part, this finding quickly circulated to manufacturers who removed phytol from products [
25]. That study used a high concentration incorporation of phytol and observed overt morbidity and mortality. In the instance of EVALI, vitamin E acetate has similarly been shown to induce severe lung inflammation in rodent models [
22,
26]. While these products have been appropriately removed from commercial products, it remains unclear how these chemicals promote lung injury and, thus, what other similar chemicals may also be unsafe for general inhalation. Agents continuously change, but PGVG remains more consistent, and a better understanding of potential effects of base vaping compounds is needed.
In this study, we developed an exposure system to expose animals to PGVG vapor to study both system as well as lung specific effects of vaping exposure. Using a combination of transcriptome, proteome, metabolome, lipidome, as well as flow cytometry analyses of immune cell populations, we aim to identify novel pathways that are impacted by vaping exposure to PGVG as well as the addition of phytol. We find that these vaping exposures have both lung-specific as well as systemic effects, suggesting that vaping may have broad consequences for immune function.
Discussion
As Ecig use has become more widespread, the term “epidemic” has been applied to the ubiquitous use of Ecigs. Many users switch from nicotine products to Ecigs believing vaping to be a safer alternative, but little is actually known about the effects of vaping on physiological processes. The United Kingdom has largely deemed vaping as a safer alternative to conventional smoking. However, vaping is difficult to study due to the large number of sometimes unknown additives that are contained within unregulated vaping products.
The recent outbreak of vaping-associated acute lung injury (EVALI), traced to the additive Vitamin E acetate, demonstrated the potential harm posed by vaping components. In this study, we focused on the effects of exposure to the base fluid contained in many vaping systems, PGVG, as well as a novel terpene additive, phytol found in counterfeit cartridges [
24]. We find that PGVG and PGVG + 1% phytol elicits both lung-specific and systemic effects on multiple systems, including metabolites, lipids, and specific subsets of immune cells. Our results suggest that PGVG with and without phytol altered lipidomic, metabolomic, and transcriptomic profiles of multiple systems and even modified immune cell populations outside the lung. These results agree well with previous published studies showing that PGVG alone can alter lung function, lipid composition and T cell subsets [
27,
28]. However, when put into the context of systemic and multi-omic pathway integration, the effects of PGVG alone are quite modest. Namely, PGVG alone did not produce a statistically significant multi-omic pathway.
Additives or alternative ingredients—such as phytol added to the base liquid PGVG in vaping products—can combine for unpredictable effects, as was the case for vitamin E acetate [
22,
23]. The addition of phytol modified the lung-omic profiles relative to PGVG alone, but several common outcomes were noted. Both PGVG and PGVG + 1% phytol upregulated AMP and downregulated proline, although they appear to have opposing effects on isobutyric acid, with PGVG increasing isobutyric acid levels while PGVG + phytol decreased levels. Changes in gene transcription and metabolites were quite different between PGVG alone and PGVG + phytol, and the different effects on weight gain between PGVG and PGVG + phytol are likely to result from systemic metabolic and transcriptomic effects. Connective tissue growth factor (CTGF), known for its involvement in wound healing and pathological lung remodeling [
30,
31], was elevated in both exposure groups. On the other hand, 1% phytol led to a significant upregulation of MUC5B, a regulator of mucin production in the airways important for mucociliary clearance and associated with fibrosis in the lung [
32,
33]. In contrast, while we found PGVG did not affect MUC5B, a previous study showed increased Muc5ac in lungs of vaping exposed animals [
28], suggesting that additives and exposure conditions may affect similar lung repair pathways. PGVG did significantly change multiple lipid species including many ceramides, Phosphatidylcholine (PCs), and phosphatidylethanolamine species (PEs), while 1% Phytol + PGVG exposure evinced only modest alteration of lipids. Ceramide increases have been associated with lung injury, particularly COPD [
34,
35], while PEs and PCs are components of lung surfactant, critical for alveolar expansion for gas exchange [
34‐
38]. Lipids have been found to be changed in the plasma of a small group of Ecig users, including similar lipid species to those identified in our studies [
39]. These results point to a complex interplay and unpredictable interactive effects between additives such as phytol with base liquids. Complex addition of nicotine, cannabinoids, flavorings, etc., has the potential to affect multiple physiological systems, including metabolic and immune systems, in a manner that is currently difficult to predict.
Little is known about the novel diterpene alcohol phytol found in Ecig products [
24]. An extremely high dose (100%) of phytol was shown to be lethal to rats [
25], but in our studies we find that even at a very low dose, 1%, phytol can induce significant changes to lung transcription, metabolites, and even function. Using analysis combining our results showing changes in transcriptomics, lipidomics, and metabolomics, we find that phytol increases acetylcholine. Acetylcholine is a potent endogenous bronchoconstrictive agent [
40‐
43]. However, we find that 1% phytol possibly decreases resistance (non-significant trend) and increases lung compliance. Given the extent of lipidomic changes seen in the present study and elsewhere [
9], increased acetylcholine could reflect compensatory mechanisms in vaping-exposed lungs to offset physicochemical changes in the airway surfactant. Furthermore, the upregulation of bradykinin receptor B1 could indicate a request for additional vasodilation and/or bronchoconstriction. Through the linkage to a downregulated palmitic acid, the effects observed could partially describe an inappropriate response to pathogens during sterile conditions, and an increased vulnerability to pathogen insult.
We also show phytol had significant effects on immune genes, including regulators of immune cell migration (CCL19, CCR7), inflammation (IL-1β), and immune regulation (TIMP1). Both PGVG and PGVG + phytol changed the circulating CD4 + T cell population, with decreased numbers in the blood and increased numbers in the spleen, both in total CD4s as well as CD44hi and CD25hi populations. Changes to the CD4 T cell subset were also previously observed in animals exposed to VEA in an EVALI model, suggesting that Ecig exposure may broadly affect the CD4 T cell response [
44]. Interestingly, while we find no changes in overall immune subsets in the lung, one previous study did show some change in CD19 + and CD4 + cells in the lung [
27]. These results suggest that changes to metabolites, lipids, and gene products in the lung may translate into systemic changes, including systemic changes to immune cell subsets. Such alterations to the systemic T cell population may only manifest under the circumstance of a lung immunological challenge (
i.e., a respiratory pathogen). Epidemiological data have been reported showing individuals who vape exhibit increased susceptibility to infectious disease [
6,
45]. Our results provide a potential mechanism tying exposures in the lung to systemic changes in immunity.
These results must be put into context with the operating conditions, which included a specific Ecig device, coil, and wattage. Sub-ohm vaping is not an uncommon practice among users but is more prone to high temperatures at the coil, which can generate numerous pyrolysis products [
46,
47]. We endeavored to keep the puff topography (short puffs, longer intervals between puffs) to avoid overheating conditions, however, we did not explicitly measure products like metals, acrolein, or formaldehyde that could have been generated. Furthermore, additives including triacetin can be found in Ecig liquids, particularly in devices that contain THC or cannabis [
48,
49], which can also lead to formation of acrolein and formaldehyde upon vaping [
50]. Our addition of phytol was a simple individual study permutation that clearly altered the response phenotype. We predict that numerous other ingredients and aerosol generating conditions can also change the responses.
Changes to lipids, metabolites, and genes can lead to a multitude of downstream effects, but in the present study we have limited phenotyping showing modest alterations of lung function. Lung disease is often multifactorial, with even pathologies from long-term cigarette smoking being heavily influenced by genetics and lifestyle. The mulit-omic alterations observed in response to PGVGV and PGVG + phytol may simply amount to moderate homeostatic adjustments to an altered chemistry of the airway lining fluid. The data are concerning, however, with respect to the complex differences with a small amount of terpene added to the base, and in consideration of systemic immune cell population changes. Long-term Ecig use overlaid with a vulnerable genetic profile or in the face of a second hit (e.g., pathogens or toxicants), may lead to more pathological alterations to respiratory health. While these remain to be determined, our study demonstrates the importance of testing the effects of specific compounds in combination due to the complex interplay of toxicological exposures with biological processes.
Materials and methods
Mice and reagents
C57BL/6 from Jackson Laboratories were used. All mice were maintained in a specific pathogen-free environment in barrier facilities at the University of New Mexico School of Medicine in Albuquerque, NM, and conform to the principles outlined by the Animal Welfare Act and the National Institutes of Health guidelines and approved by the IACUC animal use committee. Females were used at between 8–20 weeks. All experimental protocols were approved by the IACUC animal use committee at UNM Health Sciences Center in accordance with relevant guidelines and regulations (IACUC protocol #s: 18-200797-B-HSC; 16-200497-HSC; 19-200892-HSC). All animal work was performed and reported according to ARRIVE guidelines.
Reagents used
Antibodies used include: anti-mouse CD45 APCCY7 (Clone 30-F11 Biolegend Cat #103116) and anti-mouse CD45 Eflour 450 (Clone 30-F11 Invitrogen Cat#48-0451-82); anti-mouse LY6G (Clone 1A8; Biolegend Cat#127623); anti-Mouse CD3 Pacific Blue (Clone 145-2C11 Biolegend Cat#100334); anti-Mouse CD3e, PerCP-Cy5.5 (Clone: 145-2C11, eBioscience, Cat#: 45-0031-82); anti-Mouse CD8a PerCP (Clone Ly-2 BDBiosciences Cat#M037858); anti-Mouse CD4, FITC (Clone: GK1.5; eBioscience, Cat#: 25-0041-82); anti-mouse CD19 PE (Clone eBio1D3 ThermoFisher Cat# 12-0193-83); anti-mouse CD11c APC (Clone N418; Biolegen Cat #117310); anti-mouse CD11b FITC (Clone M1/70.15 Invitrogen Cat #RM2801); anti-mouse CD25 PE (Clone 3C7 Biolegend Cat#101904); anti-mouse CD44 PE (Clone 1M7; Biolegend Cat# 10,008); anti-mouse CD62L APC (Clone Mel-14 Biolegend Cat# 104411).
Exposure
We developed a customized, flexible exposure system built on the InExpose (Scireq, Inc) cigarette exposure platform. Our modular, adjustable system allowed for a variety of commercially-available devices and e-liquids to be incorporated into our study design. The InExpose software/hardware system permitted real-time control of air flow from the vape device through the whole-body exposure chamber. A linear actuator was connected to adjustable timers to alter the duration and intervals of vape puffs. We used a Smok® G-Priv3 Mod, which allowed for adjustment of wattage up to 230W (50W with a 0.17Ω nickel coil was used for all data), used rechargeable batteries, provided a warning when resistance (ohms) or temperatures were out of range, and flexible use of a variety of commercial or laboratory-prepared e-liquids. Our platform was easily modified to incorporate other commercial vape devices and, naturally, accommodate conventional cigarettes. The exposure chamber was monitored in real-time for mass concentration (DustTrak II, TSI), targeting an average of 125 mg/m
3 (Additional file
1: Fig. S1), based largely off of prior work measuring urine cotinine levels in exposed mice (Irfan 200 mg/m
3: [
51,
52]). The system was controlled as a stable concentration from day to day, despite the intermittent nature of the “puff” settings. The frequency and duration of puffs ranged from 2–3 per minute and 2–5 s per puff; these were manually varied throughout the exposure period to ensure comparable overall mass concentration levels. Dilution air was pulled through the exposure chamber (4.3L volume) at a manually adjusted rate of 1–3 lpm. A small fan was mounted in the exposure chamber to enhance mixing and consistency of exposures between mice. Aerosol size distribution (mmad typically 120-250 nm) was also characterized for different e-liquids of device/settings using a Laser Aerosol Spectrometer (TSI). The exposure chamber housed 16 mice. For safety, an O
2-capnograph continuously monitored oxygen and carbon dioxide levels in exposure chambers.
C57BL/6 mice were exposed by inhalation for 2 h/day, 5 days/week × 8 weeks to a vaporized mixture of propylene glycol/vegetable glycerin (PGVG) (50% + 50%) with and without 1% phytol by volume.
Pulmonary function
Post-exposure, a subset of mice from each group were anaesthetized via isoflurane and tracheostomized with a 19-gauge cannula before pulmonary function measurement using a Flexivent system (SCIREQ Scientific Respiratory Equipment), as previously described [
53]. Resistance and compliance measurements were recorded in multiple intervals over a 5-min period. Methacholine challenge was omitted in these studies as we sought to capture baseline changes in respiratory function due to the exposure conditions. Data was compiled using FLEXIWARE software, version 7.6, and statistical analysis was performed in Graphpad Prism8.
Transcriptomic analysis
Following sacrifice, lungs were extracted and placed in ice-cold PBS without calcium or magnesium until batch dissociation could be performed. The Miltenyi Biotec Lung Dissociation Kit (cat. No. 130-095-927) was employed, and the associated protocol followed, as it has been developed to minimize cell death during the single cell suspension steps on the gentleMACS™ Dissociators. One lung lobe per mouse was placed in a gentleMACS C Tube containing an enzymatic digestion solution, and manually cut into sections < 5 mm. Lungs were initially dissociated on the gentleMACS Octo Dissociator under heated conditions using the 37C_m_LDK_1 program. Tubes were rotated for 30 min at 37 °C before reattaching to the Octo Dissociator and running the m_lung_02 program. Sample suspensions were strained (70um pores) to remove large debris, and strainers were washed with a kit buffer. Subsequently, tubes were centrifuged at 300 × g for 10 min to pellet cells and supernatant was aspirated. Cells were resuspended and RNA was extracted using a commercial kit.
Lung Fibrosis and Host Response kits were purchased from Nanostring based on their targeted primer approach. Data was analyzed by ROSALIND® (
https://rosalind.bio/), with a HyperScale architecture developed by ROSALIND, Inc. (San Diego, CA). Read Distribution percentages, violin plots, identity heatmaps, and sample MDS plots were generated as part of the QC step. Normalization, fold changes and p-values were calculated using criteria provided by Nanostring. ROSALIND® follows the nCounter® Advanced Analysis protocol of dividing counts within a lane by the geometric mean of the normalizer probes from the same lane. Housekeeping probes to be used for normalization are selected based on the geNorm algorithm as implemented in the NormqPCR R library1. Abundance of various cell populations is calculated on ROSALIND using the Nanostring Cell Type Profiling Module. ROSALIND performs a filtering of Cell Type Profiling results to include results that have scores with a p-Value greater than or equal to 0.05. Fold changes and pValues are calculated using the fast method as described in the nCounter® Advanced Analysis 2.0 User Manual. P-value adjustment is performed using the Benjamini–Hochberg method of estimating false discovery rates (FDR). Clustering of genes for the final heatmap of differentially expressed genes was done using the PAM (Partitioning Around Medoids) method using the fpc R library2 that takes into consideration the direction and type of all signals on a pathway, the position, role and type of every gene, etc. Hypergeometric distribution was used to analyze the enrichment of pathways, gene ontology, domain structure, and other ontologies. The topGO R library3, was used to determine local similarities and dependencies between GO terms in order to perform Elim pruning correction. Several database sources were referenced for enrichment analysis, including Interpro4, NCBI5, MSigDB6,7, REACTOME8, WikiPathways9. Enrichment was calculated relative to a set of background genes relevant for the experiment.
For metabolite extraction, each tissue sample (~ 20 mg) was homogenized in 200 µL MeOH:PBS (4:1, v:v, containing 1,810.5 μM 13C3-lactate and 142 μM 13C5-glutamic Acid) in an Eppendorf tube using a Bullet Blender homogenizer (Next Advance, Averill Park, NY). Then 800 µL MeOH:PBS (4:1, v:v, containing 1,810.5 μM 13C3-lactate and 142 μM 13C5-glutamic Acid) was added, and after vortexing for 10 s, the samples were stored at -20 °C for 30 min. The samples were then sonicated in an ice bath for 30 min. The samples were centrifuged at 14,000 RPM for 10 min (4 °C), and 800 µL supernatant was transferred to a new Eppendorf tube. The samples were then dried under vacuum using a CentriVap Concentrator (Labconco, Fort Scott, KS). Prior to MS analysis, the obtained residue was reconstituted in 150 μL 40% PBS/60% ACN. A quality control (QC) sample was pooled from all the study samples.
For lipid extraction, each tissue sample (~ 20 mg) was mixed with 200 µL 10x-diluted PBS (4 °C) and 80 µL internal standard solution (PC (17:0/17:0) and PG (17:0/17:0) in MeOH; 50 uM; 4 °C) in an Eppendorf tube (1.5 ml). Then stainless beads were added into the tube, and homogenization (2 min) was performed using a Bullet Blender homogenizer (Next Advance, Averill Park, NY). After homogenization, 400 µL MTBE were added into the sample. Then the sample was vortexed for 30 s, stored under − 20 °C for 30 min, and sonicated in an ice bath for 10 min. After centrifugation (14,000 rpm, 10 min), 300 µL upper MTBE layer was collected into a new Eppendorf tube. The MTBE layer was then dried in a Vacufuge Plus Evaporator. Samples were then reconstituted with 100 μL 1:1 IPA:MeOH. 80 μL of each sample was transferred to a LC-MS vial for analysis, while the remaining 20 μL was pooled to create a quality control (QC) sample. The QC was measured every 10 samples to ensure consistent output by the LC–MS/MS.
The untargeted LC-MS metabolomics method used here was modeled after that developed and used in a growing number of studies [
54‐
57]. Briefly, all LC-MS experiments were performed on a Thermo Vanquish UPLC-Exploris 240 Orbitrap MS instrument (Waltham, MA). Each sample was injected twice, 1 µL for analysis using negative ionization mode and 1 µL for analysis using positive ionization mode. Both chromatographic separations were performed in hydrophilic interaction chromatography (HILIC) mode on a Waters XBridge BEH Amide column (150 × 2.1 mm, 2.5 µm particle size, Waters Corporation, Milford, MA). The flow rate was 0.3 mL/min, auto-sampler temperature was kept at 4 ̊C, and the column compartment was set at 40 ̊C. The mobile phase was composed of Solvents A (10 mM ammonium acetate, 10 mM ammonium hydroxide in 95% H
2O/5% ACN) and B (10 mM ammonium acetate, 10 mM ammonium hydroxide in 95% ACN/5% H
2O). After the initial 1 min isocratic elution of 90% B, the percentage of Solvent B decreased to 40% at t = 11 min. The composition of Solvent B maintained at 40% for 4 min (t = 15 min), and then the percentage of B gradually went back to 90%, to prepare for the next injection. Using mass spectrometer equipped with an electrospray ionization (ESI) source, we will collect untargeted data from 70 to 1050 m/z.
The untargeted LC–MS lipidomics method used here was modeled after that developed and used in a growing number of studies [
58‐
60]. All mass spectrometry experiments were done on a Thermo Vanquish UPLC-Exploris 240 Orbitrap MS system (Waltham, MA). Each sample was run twice; one for positive ion mode and one for negative ion mode. For positive mode, 1 μL was used per injection, whereas 1 μL was used in negative ion mode injections. Both modes used reverse phase chromatography with a Waters XSelect HSS T3 column (150 × 2.1 mm, 2.5 µm particle size; Waters Corporation, Milford, MA). Flow through the column was maintained at 0.3 mL/min. The mobile phase Solvent A was comprised of 10 mM ammonium acetate in 60% H
2O/40% ACN. Solvent B consisted of 10 mM ammonium acetate in 90% IPA/10% ACN. An isocratic elution was used with 50% solvent B for 3 min before its percentage was gradually increased to 100% over the next 12 min. Following 10 min of continued 100% solvent B, at t = 25 min, the percent of B was decreased gradually back to 50% to prepare for the next sample injection. Using mass spectrometer equipped with an electrospray ionization (ESI) source, we will collect untargeted data from 100 to 2000 m/z.
To identify peaks from the MS spectra, we made extensive use of the in-house chemical standards (~ 600 aqueous metabolites and (~ 800 lipids), and in addition, we searched the resulting MS spectra against the HMDB library, Lipidmap database, METLIN database, as well as commercial databases including mzCloud, Metabolika, and ChemSpider. The absolute intensity threshold for the MS data extraction was 1000, and the mass accuracy limit was set to 5 ppm. Identifications and annotations used available data for retention time (RT), exact mass (MS), MS/MS fragmentation pattern, and isotopic pattern. We used the Thermo Compound Discoverer 3.3 software and LipidSearch 4.2 software for aqueous metabolomics and lipidomics data processing, respectively. The untargeted data were processed by the software for peak picking, alignment, and normalization. To improve rigor, only the signals/peaks with CV < 20% across quality control (QC) pools, and the signals showing up in > 80% of all the samples were included for further analysis.
Resultant metabolomic and lipidomic datasets were input into metaboanalyst.ca. Batch runs contained QC samples at intervals of 1 QC per every 5 experimental samples and were used for initial normalization. Sample-wise, normalized by sum to make samples more comparable to each other. Featurewise, log transformed to correct for heteroscedasticity, and mean centered to focus on the differences rather than the similarities of the data. Heatmaps were autoscaled by features, with Euclidean distance, and Ward clustering. Top number of metabolites/lipids/RNA are described per figure legend. PLS-DA components were generated by DiscrMiner (v. 0.1–29) and plotted with a 95% confidence region ellipse. Volcano plots were generated by MeataboAnalystR (v. 2.0.0) for metabolites and lipids by FDR p < 0.1, minimum fold-change of 1.5.
Statistically significant RNA, metabolites, and lipids were compiled into lists and converted from mouse to human using the Rstudio convertid package (v. 0.1.3) based on the requirements of the MetaboAnalystR package. Knowledge-based network generation utilized each input value as seeds to map networks and subnetworks with neighbors. Our parameters (degree min = 2, betweenness min = 1) resulted in a single first-order subnetwork continent when examining metabolite/lipid-gene interactions and 11 subnetworks when examining metabolite/lipid-gene-disease interaction networks. The top subnetwork for each examination was compiled and pruned. I.e. subnetwork 1 from the metabolite/lipid-gene-disease interaction network contained 52 nodes, 54 edges, and 21 seeds, but 2 islands are presented here. The entirety of the metabolite/lipid-gene network is presented. Pruning was based on disease states that could not exist within the lungs, such as Alzheimer’s and schizophrenia connections with PC(16:0/18:2(9Z,12Z)) and acetylcholine, respectively.
Flow cytometry
Single cell suspensions from murine LNs and spleens were stained with conjugated antibodies (listed in Mice and reagents) according to standard protocols. Data was acquired using a BD LSR Fortessa (BD) and analyzed using FlowJo (FlowJo, LLC). Flow cytometric analysis of primary murine cells consisted of gating on lymphocytes population from FSC x SSC, then on the indicated markers as shown.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.