An assessment of vaping-induced inflammation and toxicity: a feasibility study using a 2-stage zebrafish and mouse platform

Abstract

It is currently understood that tobacco smoking is a major cause of pulmonary disease due to pulmonary/lung inflammation. However, due to a highly dynamic market place and an abundance of diverse products, less is known about the effects of e-cigarette (E-cig) use on the lung. In addition, varieties of E-cig liquids (e-liquids), which deliver nicotine and numerous flavor chemicals into the lungs, now number in the 1000s. Thus, a critical need exists for safety evaluations of these E-cig products.

Herein, we employed a “2-stage in vivo screening platform” (zebrafish to mouse) to assess the safety profiles of e-liquids. Using the zebrafish, we collected embryo survival data after e-liquid exposure as well as neutrophil migration data, a key hallmark for a pro-inflammatory response. Our data indicate that certain e-liquids induce an inflammatory response in our zebrafish model and that e-liquid exposure alone results in pro-inflammatory lung responses in our C57BL/6J model, data collected from lung staining and ELISA analysis, respectively, in the mouse. Thus, our platform can be used as an initial assessment to ascertain the safety profiles of e-liquid using acute inflammatory responses (zebrafish, Stage 1) as our initial metric followed by chronic studies (C57BL/6J, Stage 2).

1. Introduction

E-cigarettes (E-cigs) are an emerging form of tobacco products and pose a potentially new danger. Tobacco can be smoked in many different forms, including cigarettes, cigars, pipes, hookah and more recently E-cigs. However, although the deleterious effects of cigarette smoking have been documented widely, significantly less is known about the potential health effects of exposure to the diverse array of chemicals contained within newer products (such as E-cigs) on the lung (1). E-cigs differ from conventional cigarettes in that they contain no combustible tobacco but rather utilize a battery-operated coil to heat and aerosolize the nicotine (if present) in a liquid vehicle (e-liquid; vehicle is composed of propylene glycol (PG) and vegetable glycerin (VG) at varying ratios and very often supplemented with a diverse array of flavors/flavor chemicals) (2) to the lungs (3). This relatively new and fast-growing subset of nicotine users, described as “vapers” rather than smokers, utilize products that are very efficient at delivering high doses of nicotine so that plasma nicotine levels comparable to those observed with conventional tobacco smoking have now been recorded (4–7). Further, with thousands of varieties of different E-cig liquid (e-liquid) flavors commercially available and a highly dynamic market place, a critical need exists for overall safety evaluations of these e-liquid products containing various flavor chemicals. For example, recent work performed on free-base nicotine products needs to be replicated using JUUL products, which have a different composition. However, JUUL has had a recent (2020) diminishing market share due to new FDA regulation, and new disposable devices, for example as “POSH” and “STIG”, are now on the market (3, 8). In addition, as data accumulate indicating the adverse effects of E-cig intake, it is becoming increasingly clear that better assessment and regulation of e-liquid are necessary for population safety. For example, small, but significant, amounts of several carcinogens (e.g., formaldehyde, acetaldehyde and acrolein) have been detected in E-cig vapors (9–14). However, these chemicals are absent in the unvaped liquid.

Traditional tobacco exposure triggers a number of inflammatory responses in the airways, often leading to ailments such as airway inflammation, acute respiratory distress syndrome, acute lung injury, chronic obstructive pulmonary disease (COPD) and lung cancer (15). In addition, E-cig use in particular is also linked with immunosuppression in the upper airways of vapers (16, 17), and inhaled e-liquid aerosols are believed to be deposited in the upper airway areas, that is, in the broncho-alveolar region (18). Thus, neutrophils and macrophages, which are two of the major inflammatory cell types involved in lung defense (19) and situated within these airway regions, are likely affected (20). Indeed, evidence demonstrates that E-cigs exert some pro-inflammatory effects on human alveolar macrophages (AM), i.e., increased levels of ROS production, significantly inhibited phagocytosis and increased levels of several cytokines, e.g., IL-6 and TNF-α (21). A second example is more specific with the popular e-liquid flavor chemical cinnamaldehyde possessing dose-dependent broadly immunosuppressive effects: diminished phagocytic capacity (neutrophils and macrophages), pro-inflammatory cytokine production and a cell-mediated cytotoxic response (NK cells) (22).

We can also point to more recent examples of health risks associated with E-cig use. During late 2019 and early 2020, E-cig use was linked to a mysterious severe pulmonary disease, that is, E-cig or vaping product use-associated lung injury (EVALI) (23–25). More specifically, the Centers for Disease Control and Prevention (CDC) have linked this dangerous, newly identified lung disease, which is characterized by severe and sometimes fatal lung infections, to vaping. In addition, it is likely that unknown risk factors such as vaping may contribute to the severity of COVID-19 lung disease and must, therefore, be further investigated (26, 27). These two more recent examples illustrate some of the associated health risks that potentially accompany E-cig use.

Herein, we have utilized a 2-stage in vivo screening platform for initial assessments of the safety profiles of e-liquids using inflammation and toxicity biomarkers. Of note such a platform (zebrafish and mouse) has been previously employed to evaluate diverse compounds, ranging from nanoparticles, cinnamaldehyde, kinase inhibitors, anti-cancer and anti-seizure drugs (28–32). However, our platform is novel because we evaluate the model’s feasibility to assess vaping-induced inflammation and toxicity.(33)

Since the cytotoxicity, and particularly the inflammatory, effects associated with e-liquid usage are likely underappreciated (34–38), we first utilized an embryo survival assay followed by a neutrophil inflammation assay in the zebrafish, with both assays highly amenable to development for high-throughput use and having proven successful in drug screening campaigns (39, 40). Neutrophils have the ability to migrate to a site of injury or infection quickly and efficiently, and the zebrafish neutrophil migration/motility model is well established for the evaluation of inflammation due to tissue damage or infection (41–45). In addition, the use of zebrafish embryos for toxicity assays has been well established, allowing for toxicological evaluation based on multiple endpoints related to lethal or sub-lethal effects along with developmental defects (39, 46–49). Here, we were able to use zebrafish embryos to evaluate the toxicity of e-liquids with different flavors, each containing a diverse array of chemicals. We found that different flavored e-liquids can be characterized when using various lethal concentrations or pro-inflammatory responses in zebrafish embryos. During inflammation, neutrophils are an extremely important cell type involved in lung injury or COPD pathology and play a critical role in cytokine release as well as the protease release involved in tissue destruction and chronic inflammation (50). A zebrafish study has also demonstrated lung injury due to a folate deficiency-induced swim bladder defect (51). A zebrafish model involving a pro-inflammatory response to e-liquids will enable future studies centered around inflammatory cytokines or protease expression using whole-mount in situ hybridization assays. We then shifted to our mouse model for chronic E-cig exposure with or without a microbial challenge, that is, the murine hepatitis virus (MHV), which is a causative agent of SARS-like pneumonia after intranasal exposure in mice (52). While pathology was evident with chronic E-cig exposure alone, exacerbation was visible with the MHV infection, providing valuable insight as it is currently well known that traditional smoking renders the user more susceptible to pulmonary disease initiation/progression due to microbial challenge. With both models in mind, the goal of this platform is to be able to screen large numbers of e-liquid chemicals in a high-throughput fashion, so as to assess their pro-inflammatory and potential toxicity. The zebrafish model has the advantage of being high-throughput while the mouse is a mammalian model for pulmonary inflammation. It is of course not practical to screen large numbers of compounds in mice, therefore combining the 2 models provides the advantages of both, thus enabling us to perform large-scale screens as well as providing a mammalian model. Potential flavor chemicals that are not identified as the hit molecules but that are of other interest may then still be evaluated using our mouse model, if needed.

2. Materials and Methods

2.1. Chemicals and Reagents.

Unless otherwise noted, all reagents were purchased from either Thermo Fisher Scientific (Waltham, MA, USA) or Sigma-Aldrich (St. Louis, MO, USA) at the highest level of purity available. The utilized JUUL e-liquids/ “pods” were purchased locally from retailers in Durham, NC, USA from November 1, 2020—July 31, 2021. These products were inventoried and stored at room temperature until used. However, when vaped using the JUUL E-cig device, the vaped e-liquid was stored at −20°C. The manufacturer’s label information stated ingredients include only vegetable glycerin (VG), propylene glycol (PG), nicotine, flavoring and benzoic acid with each pod containing 0.7 ml of the flavored fluid at 3% nicotine. In addition, Omaiye et al. (13) have performed detailed chemical analyses (gas chromatography-mass spectrometry (GC–MS)) of these products, providing detailed information (chemical identities and relative concentrations). In addition to these commercial products, the vehicle control (50:50 (vol/vol) PG/VG) and PG/VG + 33 mg/ml nicotine were formulated in-house.

The e-liquid used for the mouse studies was vaped using a previously described (53) apparatus to produce an e-liquid vapor distillate (27, 54). Briefly, the e-liquid was vaped using the commercially available JUUL E-cig device connected to a silicon tubing and to the mouthpiece of the E-cig on one end. The other end was placed in a 50-ml conical tube, in which the distillate condensed, suspended above liquid nitrogen inside a thermal container. The JUUL device was utilized for periods of up to 5 s with at least 10 s between activations to simulate “puffs”. To reduce the chance of “dry puffing” events, only three-fourths of the pod fluid was vaped. The vaped e-liquid condensates were then stored at −20°C until they were used.

2.2. Virus

Coronavirus strain MHV-A59 and delayed brain tumor (DBT) cells were kind gifts from the laboratory of Ralph Baric (The University of North Carolina at Chapel Hill, Chapel Hill, NC, United States). DBT cells were maintained in Eagle’s minimum essential medium supplemented with 10% fetal bovine serum, 0.05 μg/ml of gentamicin and 0.25 μg/ml of kanamycin. DBT cells express a relatively uniform and abundant amount of MHVR, the receptor for MHV-A59 docking and entry into cells. Thus, the virus was both generated and quantified by standard plaque assay to determine plaque-forming units (PFUs), which are also known as infectious units (IUs)(55), using DBT cells for all experiments.

2.3. Zebrafish Maintenance and Treatments.

Zebrafish husbandry:

Zebrafish (Danio rerio) AB wildtype and transgenic lines Tg(lysC:dsRed) have been described previously (56). Identified Tg(lysC:dsRed) adult fish were used, and embryos were staged and maintained according to standard protocols (57) and NCCU IACUC approved guidelines. Zebrafish embryos were incubated in 0.3× Danieau’s solution (19.3 mM NaCl, 0.23 mM KCl, 0.13 mM MgSO4, 0.2 mM Ca(NO3)2, 1.7 mM HEPES, pH 7.4) at 28.5°C. Phenylthiourea (PTU, 30 μg/ml) was added at approximately 8 hpf (75% epiboly stage) to inhibit pigment formation as described previously (58). Following exposure to e-liquids, embryos are washed, dechorionated and anaesthetized before imaging or analysis.

E-liquid exposure:

E-liquids (JUUL “Mint”, “Virginia Tobacco” and “Menthol”) were dissolved in zebrafish medium (0.3× Danieau’s buffer) from 3 days-post-fertilization (dpf) to 5 dpf. The concentration utilized for the zebrafish model was based on preliminary toxicity test data obtained from our previous in vitro toxicity study (59) and then adjusting for a range of concentrations 10X higher and lower than in the in vitro study. Fluorescence imaging of neutrophils Tg(lysC:dsRed) was performed using a VAST BioImager (Union Biometrica) and evaluated by ImageJ or imaged and analyzed using ImageXpress Pico Automated Cell Imaging System (Molecular Devices) (Figs. 2,3).

Figure 2. Neutrophil migration assay in the Tg(lysC:dsRed) transgenic zebrafish.

(A) In the control, neutrophils are localized around the vascular niche of the caudal hematopoietic tissue (CHT) at resting state during normal development in transgenic Tg(lysC:dsRed) zebrafish. Activation of neutrophils, i.e., by the Mint e-liquid promotes migration from the vascular niche of the CHT, a well-known hallmark for a pro-inflammatory response. (B) Pattern analysis, i.e., the summation of all Nearest Neighbor Distances, is an indicator of dispersed neutrophils. Boxes in (A) indicate the posterior trunk/tail region of zebrafish embryos. (C) Quantification of the neutrophil migration assay data from panel (A). n = 10–12 zebrafish embryos per treatment group.

Figure 3. E-liquid exposures result in pro-inflammatory responses in zebrafish embryos.

(A) Pro-inflammatory responses with increases in neutrophil migration in transgenic Tg(lyscC:dsRed) zebrafish embryos are observed after exposures (3 dpf to 5 dpf) to Virginia Tobacco (VT), Menthol and Mint e-liquids or the known pro-inflammatory chemicals DSS or TNBS. Fluorescent neutrophils were imaged and analyzed using ImageXpress Pico. (B) The cell distance indicates the total nearest neighbor distances from neutrophils in the posterior trunk/tail regions of the transgenic Tg(lysC:dsRed) zebrafish at 5 dpf. n = 10–12 zebrafish embryos per treatment group. # p<0.001; * p<0.0001.

Microscope imaging:

Real time imaging of embryos was performed using a MVX10 Macro-View fluorescence microscope (Olympus) equipped with a C9300–221 high-speed digital CCD camera (Hamamatsu). Picture acquisition parameters were kept constant to allow for direct comparisons. Raw data were analyzed using MetaMorph Basic software (Olympus). The fluorescent neutrophils Tg(lysC:dsRed) were pseudo colored using MetaMorph.

Image Cytometry of zebrafish embryos:

Zebrafish embryos were imaged in 96-well plates using ImageXpress Pico (Molecular Devices). The “Stitch Plate Acquisition” application of ImageXpress along with the bright field and TRITC fluorescent setting were selected. Once inside the ImageXpress chamber, plates were subjected to image acquisition according to the manufacturer’s instructions with z-stacking of red fluorescence within the selected range of 5 × 50 μm.

Quantification of neutrophil migration distance (Nearest Neighbor Distance) in ImageJ:

Fluorescent Tg(lysC:dsRed) neutrophils were analyzed after the e-liquid treatment vs. the control groups. The ImageXpress pico stitched images were documented and further subjected to ImageJ for particle analysis using Nearest Neighbor Distances. The dispersion of activated neutrophils was measured using the total Nearest Neighbor Distances of all detected neutrophils vs. the control group with non-dispersed neutrophils localized in the caudal hematopoietic tissue (CHT).

2.4. Mice and Treatments.

Mice (C57-BL/6J) were from Jackson Laboratories (Bar Harbor, MA, USA). Young adult (6- to 8-week-old male and female) mice were used for all experiments (60). After delivery, the mice were allowed to recover from shipping stress for 1 week at the NC Central Univ. Animal Resource Complex, which is accredited by American Association for Accreditation of Laboratory Animal Care. All animal care and use were conducted in accordance with the Guide for the Care and Use of the Laboratory Animals (National Institutes of Health) and approved by the NCCU IACUC. Mice were maintained at 25°C and 15% relative humidity with alternating 12-h light /dark periods.

Once acclimated, mice were anesthetized using isoflurane and a SomnoSuite low-flow anesthesia system (Kent Scientific Corporation, Torrington, CT, USA). The e-liquid distillate, vehicle control (50:50 (vol/vol) PG/VG), PG/VG + 33 mg/ml nicotine or saline (10 μl) was then delivered dropwise intranasal (IN) using a micropipette, which is a previously described (61, 62) method found to be highly effective by us (27, 54) and others (63) for liquid delivery into the lungs, once daily for 5 weeks to each animal in the appropriate treatment group. After the initial 5-week treatment period, half of the mice in each group (4) were sedated with an i.p. injection of ketamine (100 mg/kg) and xylazine (50 mg/kg) and infected IN with 1 × 10⁶ infectious units (IU) of MHV-A59 suspended in 20 μl 1× phosphate buffered saline (PBS), as previously described (27), with the remaining uninfected mice then euthanized for bronchoalveolar lavage (BAL) fluid and lung tissue collection. Body weight and health conditions were monitored daily per IACUC protocols. Humane endpoints were not used for this study as experimental time points of completion were chosen before any significant body weight change or clinical sign of disease was observed. At time points of experimental completion (5 or 7 weeks, Fig. 4A), mice were humanely euthanized using CO2 asphyxiation and cervical dislocation, as per our accepted NCCU animal protocol and NCCU Animal Resource Complex housing guidelines and conditions.

Figure 4. Chronic vaping contributes to increased pulmonary pathology in mice infected with MHV-A59.

(A) Study design. (B) Inflammatory cytokine analysis of BAL fluid from mock-, PG/VG alone-, PG/VG + NIC alone-, “Mint” vape alone-, mock + MHV-, PG/VG + MHV-, PG/VG + NIC + MHV-, “Mint” vape + MHV-treated mice (compared with mock control). ANOVA was performed for multi-group comparisons (Tukey’s multiple comparison tests). Symbols and bars represent the mean ±SEM compared with the mock control (*p<0.05, **p<0.01, ***p<0.005, ****p<0.001). N.S., not significant. (C) H&E staining of lung tissue sections isolated from mock-, PG/VG alone-, PG/VG + NIC alone-, “Mint” vape alone-, mock + MHV-, PG/VG + MHV-, PG/VG + NIC + MHV-, “Mint” vape + MHV-treated mice. Alveolar wall thickening and the infiltration of inflammatory cells into the interstitial spaces were particularly observable in the lungs from the MHV-alone and vape + MHV mice (indicated by arrows). 200× magnification. Images are representative of five images examined per lung. NIC = 33 mg/ml nicotine

2.5. Cytokine Analysis.

At experimental endpoints, BAL fluid was collected as has been previously described (54, 64, 65). In brief, mice were euthanized, and a catheter attached to 1-ml syringe was inserted into the trachea. The syringe was then used to deliver 1× PBS, which was gently pipetted up and down 3× to remove fluid. BAL fluid was then clarified via centrifugation for cytokine analysis. Inflammatory cytokine proteins were evaluated using ELISA (OptEIA, BD Pharmingen) reagents.

2.6. Histopathology.

After being euthanized, lungs were inflated with 1 ml of 10% neutral-buffered formalin, then removed and suspended in 10% formalin for 12 h. Lungs were washed once in PBS and then immersed in 70% ethanol. Tissues were then embedded in paraffin, and three 5-μm sections 200 μm apart per lung were stained with hematoxylin/eosin (H&E) for examination by the NCCU histopathology core (directed by Dr. X. Chen). Sections were evaluated blindly for gross pathology.

2.7. Statistics.

Statistics for analysis were performed using GraphPad Prism (La Jolla, CA, United States) and Microsoft Excel analysis. Appropriate statistical tests (Student’s t-test, ANOVA) were determined after discussion with NCCU biostatistics faculty. All studies were powered to provide for a more than 95% confidence interval when using Fisher’s Exact Test, which were again designed in consultation with NCCU biostatistics faculty. Unless otherwise indicated, the results are shown as the mean ± the standard error of the mean (SEM). A value of p<0.05 was considered as statistically significant.

3. Results

3.1. Effects of e-liquid exposure on the survival of zebrafish embryos

In order to overcome the limited ability to translate in vitro data into relevant safety profiles, we leveraged the zebrafish model (66–68). At present, few studies have utilized the advantages, i.e., a short generation time, high fecundity and significant physiological and genetic homology to humans (69, 70), of the zebrafish model to evaluate the developmental effects or toxicology associated with e-cigarette exposure (71–76). Here, we began by making an assessment of the overall effects of resting e-liquid exposure on the survival of zebrafish embryos (Fig. 1). As illustrated, it quickly became obvious that exposure to these e-liquids was highly detrimental to development during 3 to 5 dpf, even at low concentrations, i.e., 0.125% (v/v) or less. Strikingly doses higher than 0.075%, 0.05% and 0.025% for “Menthol”, “Virginia Tobacco” and “Mint”, respectively, were demonstrated as lethal for zebrafish embryo survival (Fig. 1). However, these initial dosing experiments did provide us valuable insight into the range of e-liquid concentrations to be used moving forward.

Figure 1. E-liquid exposure affects the survival of zebrafish embryos.

Survival rate of embryos after e-liquid exposure (Menthol, Virginia Tobacco (VT) or Mint) from 3 to 5 days-post-fertilization (dpf). CTL is embryo medium only. Sodium trimethylsilylpropanesulfonate (DSS) and trinitrobenzenesulfonic acid (TNBS) are known pro-inflammatory reagents. PG/VG is the vehicle control for the e-liquid. n = 8–10 embryos per group.

3.2. Neutrophil inflammation assay using 3 to 5 dpf zebrafish exposed to e-liquids

While the zebrafish has been used for developmental e-cigarette exposure studies, to our knowledge no zebrafish study has focused on developing an inflammatory model of e-liquid exposure using the zebrafish. Based on the recent literature, it is likely that the cytotoxicity and inflammatory effects associated with e-liquid usage are unappreciated (34–38). For example, alveolar macrophages (AM) have been demonstrated to present with a phagocytosis defect within the lung when exposed to cigarette smoke (77), and more recently a study utilizing human AMs exposed to an e-liquid vapor distillate revealed increased levels of ROS production, significantly inhibited phagocytosis and increased levels of several cytokines, e.g., IL-6 and TNF-α (21). Thus, e employed a neutrophil inflammation assay to make such assessments.

Transgenic zebrafish that express fluorescent proteins Tg(mpo:GFP) and Tg(lysC:dsRed) within their neutrophils have been established (56, 78). These fluorescent reporter lines provide us the ability for live cell imaging to observe and characterize inflammation in real-time within a living vertebrate (Fig. 2) (79). In addition, the feasibility of zebrafish embryos for high-throughput chemical screening has been demonstrated (80, 81). These assays are also well established as a high-throughput chemically-induced inflammation assay (82–84). Therefore, we utilized the Tg(lysC:dsRed) line to better examine the outcomes of e-liquid exposure beyond simply embryo survival from 3 to 5 dpf using first the e-liquid that had the most profound effect on embryo survival, i.e., the “Mint”, in a neutrophil migration assay (Fig. 2).

Here, the “Mint” e-liquid induced a well-known hallmark for a pro-inflammatory response in zebrafish embryos, resulting in neutrophil migration from the vascular niche of the caudal hematopoietic tissue (CHT), whereas control embryos possessed neutrophils at the resting stage that localized primarily along the CHT region (Fig. 2A). In fact, the “Virginia Tobacco”, “Mint” and “Menthol” e-liquids all induced pro-inflammatory responses and led to neutrophil migration (Fig. 3). Therefore, quantification of the pro-inflammatory response (Figs. 2,3) in zebrafish and toxicological response (Fig. 1) can be used to prioritize various flavored e-liquids for pro-inflammation and embryo toxicity, which can facilitate this 2-stage model for the assessment of inflammation and toxicity biomarkers to move forward into future assays/model organisms.

3.3. A chronic model of vaping indicates cytotoxicity and pro-inflammatory responses, which are exacerbated by MHV-A59 infection

We next evaluated the toxicity of our e-liquids using the mouse. For these studies, we first vaped, then collected and condensed the e-liquid vapor for a more physiologically relevant reagent for our mouse model exposures. To evaluate pathology, we delivered this “vaped” e-liquid to mice [10 μl, containing ∼20 μg of nicotine], PG/VG, PG/VG + NIC, or saline intranasal (IN) for 5 weeks to evaluate chronic pathology associated with vaping. After 5 weeks, half (4) of the mice in each group were euthanized for BAL fluid and lung tissue collection while the remaining mice were infected with 1 × 10⁶ infectious units (IU) of MHV-A59 and vaped for an additional 2 weeks and then euthanized (Fig. 4A). From the BAL fluid, we observed increases, though not statistically significant, in the pro-inflammatory cytokine IL-6 in the uninfected mice exposed to the vaped PG/VG, PG/VG + NIC or e-liquid. However, a more notable and statistically significant increase was observed with the e-liquid-treated and MHV infected mice (Fig. 4B). These data were then reinforced by the H&E staining of lung tissue isolated from vaped vehicle- or vaped e-liquid-treated mice with alveoli thickening and neutrophil influx most apparent in the e-liquid-treated tissue and exacerbated by the MHV infection (Fig. 4C).

4. Discussion

The use of E-cig products has increased tremendously, as evidenced by increases in both sales and popularity, and is especially troubling among middle school- and high school- aged students and young adults (85, 86). Also, as previously referenced, mysterious vaping illnesses and deaths mounted since summer 2019, leading to lung injuries requiring hospitalization across all 50 states. However, high-throughput in vivo model systems to assess the potential toxicity, as well as disease-related or predictive biomarkers, of these hundreds to thousands of flavored e-liquids are lacking. Existing animal models such as the mouse are not feasible for high-throughput screening. The advantage of the transparency of zebrafish embryos and stable transgenic lines with fluorescently labeled immune cells allow for the real-time tracking of individual immune cells (82, 84, 87). In addition, the zebrafish has a highly conserved vertebrate innate immune system (88–90), and a high fecundity with several hundred eggs per female in a single spawning (91). The embryos can also be placed into 96-well plates, allowing for a high-throughput assay of hundreds to thousands of compounds (92–94), thereby reducing time and costs in chemical screening. This model also utilizes inflammatory responses using well-known pro-inflammatory chemicals, i.e., positive controls, such as the compounds DSS (95–98) and TNBS (99–102), which are well-established models of mucosal inflammation and pro-inflammatory cytokine induction. These responses add impact to our study and also agree with our previous study wherein we determined that the “Mint” e-liquid is the most inflammatory and that “Menthol” and “Virginia Tobacco” were less inflammatory when utilizing CALU-3 and A549 human bronchial epithelial cell lines (59). Importantly, these findings are consistent with our zebrafish findings as well as our mouse experiments. Finally, these assays are highly powered with the zebrafish toxicity study assessing end points of survival using 8–10 embryos per group, which are values utilized based upon our preliminary testing that varying the concentration of these chemicals gave reliable survival data (Fig. 1). The neutrophil migration assay is also adequately powered using 10–12 embryos, again based upon preliminary testing that the number of embryos was sufficient to provide for a more than 95% confidence interval in power analysis. Therefore, it is a very desirable system to use to prioritize candidates in a large-scale screen (33, 103, 104).

Our 2-stage in vivo screening platform, i.e., zebrafish embryo survival followed by the neutrophil inflammation assay and a lung-related injury model in the mouse, is able to prioritize large numbers of flavored e-liquids. This model allows for the initial investigation of e-liquids using these medium- to high-throughput techniques as our initial metrics followed by long-term, or potentially acute, mouse studies. In addition, this model is highly innovative in that our assessment of e-liquids uses inflammation and not traditional toxicity alone as a key metric. From these studies, we are able to differentiate, for the first time, among varying responses to diverse e-liquid products and how these multivariate responses may correlate to inflammatory pulmonary disease, which we believe is a valid measure of in vivo toxicity, as well as embryo toxicity, lung histopathology, chemokine and cytokine analysis and will facilitate the future discovery of toxicity profiles and biomarkers related to vaping. Further, this study serves as a useful framework/model for evaluating potential outcomes/disease states that may arise from E-cig use, which will be a new platform in contrast to the pathologic outcomes of traditional cigarette smoking, which are well-described.

This 2-stage system also has the potential to assist in better regulation of current and future commercially available e-liquid products. This is an extremely important point because of the wide variety of different e-liquid flavors available, i.e., there are currently >7000 different flavored e-liquids that are commercially available (2). In addition, the number of flavor chemicals composing any given e-liquid is highly variable, for example, JUUL uses a relatively small number (<20) of different flavor chemicals in their e-liquid “pods” (13) while some of the more popular e-liquid refill fluids can contain >50 flavor chemicals (12).

However, there are some limitations to our study. For example, direct addition of e-liquid to zebrafish medium is less than ideal but is an accepted practice for this model system and has proven to be extremely effective (39, 72, 73) and has been widely utilized in vitro (59, 105) but does not replicate real world vaping. In addition, should insolubility be an issue with a given e-liquid in future studies, the zebrafish assay can be modified appropriately, e.g., using a microinjection technique, which is a routine procedure used in zebrafish laboratories. However, given that there are thousands of e-liquid commercially available, it yields valuable increased throughput for preliminary experiments, as we have also done. Our use of a condensed vaped e-liquid distillate for the mouse studies also limits the applicability of our study even though it has been previously employed (61, 62). This methodology has also been termed an “intermediate approach” and does overcome some of the shortcomings of direct exposure methods, i.e., unlike cigarette puff topographies, which are well-studied and defined, E-cig topographies are poorly understood and change as new E-cig devices emerge (62). However, there is clearly value to direct exposure studies, and we will develop this model for our future studies. Finally, the sample size for the mouse studies is low. However, from the perspective of a proof-of-principle study they were highly effective. For example, and as noted above, from the BAL fluid, we observed increases, though not statistically significant, in the pro-inflammatory cytokine IL-6 in the uninfected mice exposed to the vaped PG/VG, PG/VG + NIC or e-liquid and even noted a statistically significant increase with the e-liquid-treated and MHV infected mice (Fig. 4B). In future, we will be employing larger cohorts of mice.

In sum, we have developed a highly innovative model for the assessment of e-liquid safety using inflammation, and not traditional toxicity alone, as our key metric, which overcomes some of the limitations of using toxicity alone as the key screening metric, i.e., some potentially toxic e-liquids may be missed due to the sensitivity of the toxicity screen but might then be detected when rescreened using our neutrophil migration/inflammation assay. Inflammation is already known to play a key role in host defense. Acute inflammation is involved in the recruitment and activation of immune cells including neutrophils, leading to a “search-and-destroy” mechanism utilizing reactive oxygen species as a potential “weapon” (106). Therefore, inflammation and oxidative stress work cooperatively, as can be observed in inflammation-generated oxidative stress, and may also lead to cardiovascular diseases, i.e., long term exposure to oxidative stress and inflammation contribute to vascular endothelial cell dysfunction and may result in abnormal angiogenesis and vascular disease (106). As such, clinical studies have produced corroborating data. With one study illustrating that long‐term E-cig use has the potential to increase the proportion of innate and adaptive immune cell subtypes, as well as to elevate cellular oxidative stress (107). A second study then found that healthy young people who regularly smoke tobacco cigarettes or E-cigs developed endothelial dysfunction, as measured by flow-mediated vasodilation (108). These findings are consistent with our own that e-liquid exposure leads to a pro-inflammatory response with an elevated neutrophil migration rate in the zebrafish assay and are an example of a potential mechanism of action, that is, the implication that inflammation begets oxidative stress. However, other investigators have noted smaller increases in oxidative stress and inflammation after acute E-cig use compared to conventional tobacco smoking (109) or after switching to chronic e-cigarette smoking (110–112). To address these other studies, which all utilize unflavored e-liquids, we do have additional theories.

As has been described by us and others, the various chemicals present in some flavored e-liquids affect cell growth/viability and also increase cytoplasmic Ca²⁺ levels in the airway epithelia, with the more chemicals present in an e-liquid, the more likely the elicitation of the Ca²⁺ response (54, 59, 113). Our results herein illustrate that e-liquids containing different flavoring chemicals exhibited different degrees of embryo toxicity. For example, the Mint e-liquid exhibited the most toxicity towards zebrafish embryo survival. Our findings of a pro-inflammatory response induced by e-liquids along with elevated neutrophil migration is therefore consistent with the clinical findings that E-cig vaping also induce cellular oxidative stress and endothelial dysfunction. In our future studies, we will further investigate the relationship of oxidative stress and exposure to various e-liquids using our zebrafish and mouse models. In conclusion, based upon this study, this model will likely have a significant impact when used to assess currently commercially available e-liquids, which now number in the 1000s.

Figure 5.

Model for the 2-Stage in vivo Screening Platform: Zebrafish to Mouse.

Highlights.

• An assessment of the inflammatory effects of e-cigarettes

• A high-throughput 2-stage (zebrafish and mouse) model to interrogate e-liquids

• Useful in assessing large numbers of e-liquid safety profiles