This study was designed to characterize the effects of exposure to flavored e-cig aerosols at single-cell level to understand the immunological changes in the lung microenvironment. To do so, we generated a single-cell profile of e-cig aerosol exposed mouse lungs (n = 2/sex/group). The thus obtained results were then validated with the help of our validation cohort of n = 3/sex/group as shown in Figure 1A. Since all the commercially available e-liquids used in this study contained tobacco-derived nicotine (TDN), we first determined the levels of serum cotinine (a metabolite of nicotine) to prove successful exposure of the mice in each treatment group. As expected, we did not see any traces of cotinine in the serum of air and PG:VG exposed mice. Significant levels of cotinine were detected in the serum of mice exposed to fruit-, menthol-, and tobacco-flavored e-cig aerosols (Figure 1—figure supplement 1B), validating successful exposure of our test animals to TDN containing flavored e-cigs.

Since metals released upon heating of the coils of e-cig devices are a source of toxicity upon vaping (Olmedo et al., 2018; Aherrera et al., 2023), we further monitored the levels of metals in the e-cig aerosols generated during each day of mouse exposures. This acted as an indirect measure for characterizing the chemical properties of the aerosols used for exposure in this study. To monitor the release of metals into the mouse lungs, the aerosol condensate from each day of exposure was collected and the levels of select elements were detected using inductively coupled plasma mass spectrometry (ICP-MS). A detailed account of the concentrations of identified elements/metals is provided in Table 1. Interestingly, we identified flavor-dependent changes in the levels of metals like Ni, Zn, Na, K, and Cu on a day-to-day basis. Note, despite the use of the same wattage and temperature (max of 230°C) for generation of e-cig aerosols, the leaching of each metal varied per day of exposure (Figure 1B). This is a crucial result as it highlights the importance of studying the impact of atomizer, coil composition, and design on the chemical composition of the generated aerosols. These variations might affect the risk and toxicity associated with each of these products, an area that has been recently explored by our group (Effah et al., 2025).

Next, we performed hematoxylin and eosin (H&E) staining on the lung tissue sections to study the morphometric changes in the mouse lungs upon exposure to differently flavored e-cig aerosols. We did not find much evidence of tissue damage or airspace enlargement upon acute exposures in our model, as expected. However, we found evidence of increased alveolar septa thickening in the lungs of both male and female mice exposed to fruit-flavored e-cig aerosols (Figure 1C). This could be a result of lower agarose inflation observed for the mouse lungs in this group. We had challenges/difficulties with inflating these mouse lungs. Owing to the lack of proper inflation, the lungs were in a collapsed state, which could be a probable explanation for our histological observation. Since this was not the prime focus of our study, we did not conduct further experiments to confirm our speculations.

As the principal focus of this study was to identify the flavor-dependent and independent effects upon exposure to commercially available e-cig aerosols at the single-cell level, we performed the scRNA seq on the mouse lungs from exposed and control mice. After quality control filtering (Figure 2—figure supplement 1A, B), normalization and scaling, we generated scRNA seq profiles of 71,725 cells in total. Except for the PG:VG group, all the rest of the treatments had approximately similar cell viabilities, cell capture, and other quality assessments. However, for normalization, equal features/genes were used across all the groups for subsequent analyses. A detailed account of the cell number (single-cell capture) and gene features identified before and after filtering upon QC check of sequenced data is provided in Supplementary file 1a.

Uniform Manifold Approximation and Projection (UMAP) was used for dimensionality reduction and visualization of cell clusters. Cell annotations were performed based on the established cell markers in the Tabula Muris database and available published literature, and we identified 24 distinct cell clusters as shown in Figure 2A. The general clustering of individual cell types based upon the commonly known cell markers was used to identify Endothelial (identified by expression of Cldn5), Epithelial (identified by expression of Sftpa1), Stromal (identified by expression of Col3a1), and Immune (identified by expression of Ptprc) cell populations (Figure 2B). The ‘FindVariableFeatures’ from Seurat was used to identify cell-to-cell variation between the identified clusters, and the top 2000 variable genes identified in each cluster have been elaborated in Supplementary file 1b. We observed minor variations in the cell frequencies as observed through scRNA seq analyses within cell types across different treatment groups (Figure 2—figure supplement 1C). The largest proportion of cells was found in the endothelial cell cluster (43.87% of total per sample), followed by lymphoid (26.43% of total per sample) and myeloid (16.79% of total per sample) clusters. A detailed account of the two-way ANOVA statistics for the general clustering with cell types and treatment groups as independent variables is provided in Supplementary file 1c.

Groupwise comparisons of each treatment group versus air control did not show many changes in the cell clusters, thus showcasing little to no effect on the overall lung cellular compositions in treated and control groups. However, the number of cells for the PG:VG group was very low (8102 cells) as compared to other treatments (~15,710 cells on average). We would like to report that this is an outcome of the low viability of this sample prior to scRNA seq, and not necessarily the effect of the treatment (Figure 2C). Differential gene expression analyses showed dysregulation of genes in all cell populations, but maximum effect was observed on the cells from immune cell clusters. This is not surprising, as the immune system, especially the myeloid cells, forms the frontline of the host’s defenses against external stressors (Del Fresno and Sancho, 2021; Marshall et al., 2018). Compared to air, we observed a dysregulation of 553 genes (338 upregulated; 215 downregulated) in the myeloid cell cluster of mouse lungs exposed to tobacco-flavored e-cig aerosol. We identified 324 and 24 DEGs in the myeloid lung cluster from mouse lungs exposed to menthol- and fruit-flavored e-cig aerosols, respectively, as compared to air control (Figure 2D; Supplementary file 1e–j). For the lymphoid cluster, we observed maximum dysregulation in the lungs exposed to fruit-flavored e-cig aerosols with a total of 112 DEGs. In contrast, 41 and 9 significant DEGs were identified for the lymphoid cluster from lungs exposed to tobacco and menthol-flavored aerosols, respectively (Figure 2D; Supplementary file 1l–q).

It is important to mention here that most of the commercially available e-liquids/e-cig products use PG:VG as the base liquid to generate the aerosol and act as a carrier for flavoring chemicals. Thus, we compared the effect of PG:VG alone in our study to make further comparisons between individual flavoring products with that of PG:VG only. DESeq2 analyses showed dysregulation of 276 genes in mouse lungs exposed to PG:VG alone in the myeloid cell cluster as compared to air controls (Figure 2D; Supplementary file 1d). Contrary to this, exposure to PG:VG aerosols affected 24 genes in the lymphoid cluster as compared to air control (Figure 2D; Supplementary file 1k). Furthermore, when compared to PG:VG, exposure to fruit, menthol, and tobacco-flavored e-cig aerosol dysregulated 262, 873, and 960 genes in the myeloid cluster and 37, 64, and 112 genes in the lymphoid cluster, respectively. A detailed account of the DEGs identified upon comparing fruit, menthol, and tobacco-flavored e-cig aerosol exposed mouse lungs to PG:VG in myeloid and lymphoid clusters has been provided in Supplementary file 1s–x.

To study the specific changes in the innate and adaptive immunity upon acute exposure to differently flavored e-cig aerosols, we first compared the changes in the overall cell population of individual cell types using scRNA seq. The scRNA seq data was validated with the help of flow cytometry using a larger cohort of animals. Since we observed sex-dependent variations in oxidative stress responses upon exposure to e-cig aerosols in previous studies (Lamb et al., 2023; Wang et al., 2019; Lamb et al., 2022), the flow cytometry data was analyzed in a sex-dependent fashion.

We did not observe any changes in the cell frequencies of alveolar macrophages across treatments through scRNA seq analyses. In general, there was a moderate decrease in the cell frequencies of alveolar macrophages in exposed mice as compared to air control independent of the flavor profile of the e-liquid employed (Figure 3A). Flow cytometric analyses confirmed little to no change in the alveolar macrophage (CD45⁺ CD11b⁻ SiglecF⁺) percentages within the lung of exposed versus control mice (Figure 3C, D). Contrarily, we observed a flavor-dependent moderate increase in the cell frequencies of neutrophil clusters in menthol (0.1 ± 0.02) %- and tobacco (0.1 ± 0.06) %-flavored aerosol exposed mouse lungs as compared to controls (0.06 ± 0.02) % (Figure 3B). We used flow cytometry to validate the scRNA seq changes. Flow cytometry analyses showed an increase in the neutrophil (CD45⁺ CD11b⁺ Ly6G⁺) percentages of menthol-flavored e-cig aerosol exposed mouse lung, corroborating with the scRNA seq results. Furthermore, our results showed this increase to be more pronounced in male mice (p = 0.0880) as compared to their female (p = 0.9662) counterparts (Figure 3C, D).

It is important to mention that due to the presence of nicotine in all the e-liquids used as treatments for this study, we added an extra control group: PG:VG +Nic for select experiments. Flow cytometric analyses revealed slight variations in the lung neutrophils and macrophage percentages observed in the PG:VG + Nic group as compared to the PG:VG only. However, none of these changes were significant. Furthermore, the patterns of change observed for the lungs exposed to aerosols from flavored e-liquids were quite distinct from those observed for PG:VG + Nic thus proving that the observed changes are not solely due to the presence of nicotine in these groups (Figure 3—figure supplement 1).

Though we did not find a distinct eosinophil cluster in our scRNA seq analyzed data, interesting sex- and flavor-specific changes were observed in the lung eosinophil (CD45⁺ CD11b^+/− CD11c⁻ Ly6G⁻ SiglecF⁺) population upon analyzing the flow cytometry results. A significant decline in the eosinophil percentage was observed in the lung homogenates from menthol (p = 0.0113) and tobacco (p = 0.0312)-flavored e-cig aerosol exposed male C57BL/6J mice as compared to air control. Contrarily, the eosinophil levels in the lung of male mice exposed to fruit-flavored e-cig aerosols did not show remarkable changes when compared to the levels found in control (Figure 4). These results further emphasize the need to study the sex-specific changes in the immune responses upon exposure to e-cig aerosols in vivo.

We next studied the changes in the lymphoid clusters of treated and control samples. While we did not notice any change in the CD4⁺ T-cell frequencies in treatment and control groups using scRNA seq, our flow cytometry data showed significant increase in the CD4⁺ T cell (CD45⁺ CD11c⁻ Ly6G⁻ CD11b⁻ MHCII⁻ CD4⁺) frequencies in the lungs of tobacco-flavored e-cig aerosol exposed female (p = 0.0492) C57BL/6J mice (Figure 5A, C, D). scRNA data did not show any change in the cell frequencies of CD8⁺ T cells in control and exposed mouse lungs either (Figure 5B). However, flow cytometric analyses presented a different picture. We found a significant sex-independent increase in the CD8⁺ T (CD45⁺ CD11c⁻ Ly6G⁻ CD11b⁻MHCII⁻CD8⁺) cell percentages in the lungs of menthol- and tobacco-flavored e-cig aerosol exposed mice as compared to air control. Contrarily, the CD8⁺ T-cell percentages increased in the fruit-flavored e-cig aerosol exposed male (p = 0.0163) mouse lungs and not in their female counterparts (Figure 5C, D). Of note, we did not observe any changes in the CD4⁺ and CD8⁺ T-cell percentages in the PG:VG + Nic group as compared to control, reiterating that the e-cig flavors are responsible for altered immune responses upon acute exposure in mice (Figure 3—figure supplement 1).

To probe further, we subclustered the myeloid cell populations. Upon subclustering, we identified 14 unique clusters comprising all the major cell phenotypes including neutrophils, alveolar macrophages, interstitial macrophages, monocytes, dendritic cells, and mast cells (Figure 6A). A detailed account of all the cell types identified with their respective marker genes in myeloid subcluster is provided in Supplementary file 1r. On deeper evaluation, we identified two unique phenotypes of neutrophils (identified by S100a8, S100a9, Il1b, Retnlg, Mmp9, and Lcn2) in the mouse lungs. These clusters were named as ‘Ly6G⁺ Neutrophils’ and ‘Ly6G⁻ Neutrophils’ based on the presence or absence of Ly6G marker, respectively (Figure 6B). Ly6G is important for neutrophil migration, maturation, and function within the lung (Lee et al., 2013). Since we did not anticipate identifying such a population of neutrophils following acute e-cig exposure, we did not gather flow cytometry evidence to validate this finding.

We speculated that the Ly6G⁻ population is a population of immature neutrophils. To confirm this possibility, we performed co-immunofluorescence using S100A8 (red; marker for neutrophil activation) and Ly6G (green) specific antibodies for control (air) and tobacco-flavored e-cig aerosol exposed mouse lungs. Co-immunofluorescence results showed a moderate increase (p = 0.4429) in the level of Ly6G⁺ cells in the lungs of tobacco-flavored aerosol exposed mice as compared to air. However, the level of S100A8⁺ cells decreased markedly (p = 0.0571) in e-cig exposed group, thus showing that activation of neutrophils could be affected upon acute exposure to e-cigs (Figure 6—figure supplement 1C, D).

To assess the effects of acute exposures to differently flavored e-cig aerosols on the cell composition in the mouse lungs, we performed differential expression analyses for each flavor in comparison to air and PG:VG controls. The fruit-flavored e-cig aerosol exposure had the mildest effect on the cell compositions and gene expression as compared to the controls in our study. We did not observe major dysregulation in the gene expression for myeloid cell cluster in fruit-flavored e-cig exposed mouse lungs as compared to air controls. A total of 24 genes (17 upregulated; 7 downregulated) were significantly (p < 0.05) differentially expressed in the myeloid clusters in the lungs of animals exposed to fruit-flavored e-cig aerosols (Figure 7Ai, Supplementary file 1e). GO analyses of differentially expressed upregulated genes showed enrichment of terms like ‘NK cell-mediated immune response to tumor cells’ (fold enrichment = 9.75; p_adj = 0.0003), ‘haptoglobin binding’ (fold enrichment = 7.76; p_adj = 0.0096) and ‘transmembrane-ephrin receptor activity’ (fold enrichment = 7.05; p_adj = 0.014) in (Figure 7Aii, Supplementary file 1f) in fruit-flavored aerosol exposed mice lung as compared to control. Importantly, few gene clusters (Hbb-bs, Hbb-bt, Hba-a2, and Hbb-a1) showed sex-specific changes in the gene expressions in exposed lungs as compared to control. These gene clusters are enriched for ‘erythrocyte development’ and warrant further study.

We further found dysregulation of 112 genes (40 up- and 72 downregulated) in the lymphoid cluster of exposed mice. Upregulation of H2-DMb2, H2-DMb1, Gp5, Pdpn, among others, enriched for ‘MHC class II protein complex assembly’ (fold enrichment = 7.10; p_adj = 0.0009) and ‘positive regulation of platelet activation’ (fold enrichment = 6.08; p_adj = 0.0023) in exposed group as compared to air control. We also observed downregulation of genes like Mcm4, Mcm2, Kif4, Cdca8, Cacna1f, and Cacnb3, in the lymphoid cluster of mice exposed to fruit-flavored e-cig aerosol. GO analyses of the downregulated genes enriched for terms like ‘CMG complex’ (fold enrichment = 10.75; p_adj = 3.39E-07), ‘spindle elongation’ (fold enrichment = 7.72; p_adj = 0.0005) and ‘high voltage-gated calcium channel activity’ (fold enrichment = 7.09; p_adj = 0.0059) in fruit-flavored aerosol exposed mouse lungs as compared to air controls (Figure 7Bi–ii, Supplementary file 1l-m).

We demonstrated an upregulation of 220 genes and a downregulation of 104 genes in the myeloid cluster of mouse lungs exposed to menthol-flavored e-cig aerosol as compared to ambient air. We observed increased expression of inflammatory genes including Il12b, Arid5a, Il12a, Il1b, Cacna1d, and Cacnb2 enriching for terms like ‘T-helper 1 cell cytokine production’ (fold enrichment = 6.21; p_adj = 0.0115) and ‘L-type voltage-gated calcium channel complex’ (fold enrichment = 5.58; p_adj = 0.0191) in the myeloid cluster of menthol-flavored e-cig aerosol exposed mouse lungs as compared to air. We also found a downregulation in the expression of Ovol1, Mapk15, Erbb2, Nrg4, Katnal2, Hspa1b, and Hspa1a in the myeloid cells of flavored e-cig aerosols. GO analyses of the DEGs, thus, showed enrichment of terms like ‘regulation of meiotic cell cycle phase transition’ (fold enrichment = 6.48; p_adj = 0.0051), ‘ERBB4 signaling pathway’ (fold enrichment = 5.40; p_adj = 0.025), and ‘protein-containing complex destabilizing activity ’ (fold enrichment = 6.36; p_adj = 0.0095) as the top hits as shown in Figure 8Ai, ii (Supplementary file 1g, h).

Contrary to the responses observed for exposure to fruit-flavored e-cig aerosols, we found significant (p < 0.05) upregulation in the expression of Cdk8 and Camk1d genes in the lymphoid cell population for menthol-flavored aerosol exposed mouse lungs (Figure 8Bi, Supplementary file 1n). Cdk8 (cyclin-dependent kinase 8) is a transcriptional regulator that has a role in the cell cycle progression (Dannappel et al., 2018). Whereas Camk1d (calcium/calmodulin-dependent protein kinase ID) functions to regulate calcium-mediated granulocyte function and respiratory burst within the cells (Jin et al., 2022). GO analyses of up- and downregulated genes showed enrichment of terms including ‘regulation of toll-like receptor 9 signaling pathway’ (fold enrichment = 5.96; p_adj = 0.034), ‘transforming growth factor beta activation’ (fold enrichment = 5.42; p_adj = 0.044), ‘mitotic DNA replication’ (fold enrichment = –7.98; p_adj = 2.51E−05) and ‘outer kinetochore’ (fold enrichment = –8.58; p_adj = 1.13E−08) in the lymphoid cluster from menthol-flavored e-cig aerosol exposed mouse lungs (Figure 8Bii, Supplementary file 1o).

Like menthol-, tobacco-flavored e-cig aerosol also elicited a significant increase in the expression of 338 genes and a decrease in 215 genes as compared to air controls in the myeloid cell cluster. We observed an increase in the expression of chemokines like Stat4, Il1b, Il1bos, Il18r1, Unc13d, Lgals9, and Nkg7 in the myeloid cells resulting in enrichment of terms like ‘T-helper 1 cell cytokine production’ (fold enrichment = 6.66; p_adj = 0.0011) and ‘natural killer cell degranulation’ (fold enrichment = 6.28; p_adj = 0.0042) in tobacco-flavored e-cig aerosol exposed mouse lungs as compared to air controls.

We further found downregulation of genes including Mcmdc2, Rad51, Spp1, and Slc34a2 which enrich for terms like ‘double-strand break repair involved in meiotic recombination’ (fold enrichment = 6.44; p_adj = 0.0019) and ‘intracellular phosphate ion homeostasis’ (fold enrichment = 6.52; p_adj = 0.0052) in myeloid cluster from tobacco-exposed e-cig aerosols (Figure 9Ai, ii, Supplementary file 1i, j).

We also observed a downregulation of genes responsible for chaperone-mediated protein folding (Cct5, Cct7, and Cct8) in the lymphoid cluster from tobacco-flavored e-cig aerosol exposed mouse lungs. Downregulation of these genes could be indicative of the accumulation of misfolded proteins in these lungs which may lead to enhanced cell death (Vallin and Grantham, 2019; Haeri and Knox, 2012). DEG and GO analyses identified upregulation of Robo1, Trem2, Padi2, Gp5, Gp9, and Pla2g4a and downregulation of Mcm4, Mcm2, Rad51, Cacnb2, and Cacnb3 in the lymphoid cluster from tobacco-flavored e-cig aerosol exposed mouse lungs resulting in enrichment of terms including ‘regulation of chemokine-mediated signaling pathway’ (fold enrichment = 5.43; p_adj = 0.0304), ‘positive regulation of platelet activation’ (fold enrichment = 5.43; p_adj = 0.0089), ‘mitotic DNA replication’ (fold enrichment = –9.6; p_adj = 6.63E−07), and ‘L-type voltage-gated calcium channel complex’ (fold enrichment = –9.12; p_adj = 0.0005) (Figure 9Bi, ii, Supplementary file 1p, q).

Since we showed increased production of cytokines/chemokines, driving the immune responses in mouse lungs exposed to flavored e-cigs, we performed multianalyte assay to determine the levels of these inflammatory cytokines in the lung digests from the exposed animals as shown in Figure 9—figure supplement 1A. Exposure to tobacco-flavored e-cig aerosol resulted in a marked increase in the levels of chemotactic chemokines, including CXCL16, CXCL12, CXC3R, and proinflammatory cytokines, including CCl12, CCL17, CCL24, and Eotaxin, in the mouse lung digests as compared to air control. Interestingly, the fold changes in the PG:VG + Nic group were contrasting to those observed by PG:VG alone and flavored e-cig aerosol exposed mouse lungs, but none of these changes were highly significant.

To identify genes that were commonly altered upon exposure, we generated a list of common genes that were significantly dysregulated in exposure categories (fruit, menthol, and tobacco). We identified nine such target genes – Neurl3, Egfem1, Stap1, Tfec, Mitf, Cirbp, Hist1h1c, Gmds, and Htr2c – that were dysregulated in the myeloid cluster from lungs exposed to differently flavored e-cig aerosol, but not PG:VG. We observed significant upregulation of Neurl3, Stap1, Cirbp, and Hist1h1c and downregulation of Tfec, Mitf, Gmds, and Htr2c in the myeloid cluster of mice exposed to differently flavored e-cig aerosols. Upon analyzing the lymphoid cluster for commonly dysregulated genes, we identified – Klra8 (Killer cell lectin-like receptor 8) and Nfia (nuclear factor I) – that were significantly upregulated in the exposure groups as compared to air-controls (Figure 9—figure supplement 1B). Klra8 is a natural killer cell associated gene, and its upregulation is generally associated with viral infection associated host immune responses within the mouse lungs (Lopes et al., 2022; Pommerenke et al., 2012; Akter et al., 2022). Nfia is a transcriptional activator responsible for regulating Oxphos-mediated mitochondrial responses and proinflammatory pathways (Kong et al., 2023; Hiraike et al., 2023).

Overall, we identified a total of 29 commonly dysregulated gene targets that were identified from five major cell clusters and performed gene enrichment analyses on the identified targets to identify the top hits (Tables 2 and 3). Terms like ‘negative regulation of immune system’ (Hmgb3/Gpam/Scgb1a1/Stap1/Ldlr), ‘positive regulation of lipid biosynthetic pathway’ (Htr2c/Gpam/Ldlr), and ‘receptor recycling’ (Ldlr/Ramp3) were among the top hits in our observations (Figure 9—figure supplement 1C).

Of note, the data presented in this study is a sub-part of a larger study. In addition to the groups mentioned in this manuscript, we also had two additional groups of TDN and tobacco-free nicotine. Though further objectives and experimentations performed in both these studies were varied, common air and PG:VG samples were used for analyses of cytokine/chemokine and cell count data as described in our previous publication (Lamb et al., 2023).

All experiments were designed to ensure rigor and reproducibility through inclusion of appropriate controls, randomization, blinding, and replication. Sample sizes and statistical analyses are described in the Methods and figure legends. All reagents and analytical methods are reported in sufficient detail to allow replication. All the lab-based experiments comprise two technical replicates with a minimum of two to three biological replicates to ensure rigor. The data generated from this study is publicly available for future reference.

All materials used in this study are commercially available or publicly accessible, and no unique materials were generated. The respective catalog numbers and vendor information of the chemicals, kits, and/or antibodies used have been detailed in the respective sections.

We ordered 5-week-old pups of male and female C57BL/6J mice (strain ID: 000664) from Jackson Laboratory to conduct this experiment. Prior to the start of the experiment, mice were housed at the URMC Vivarium for acclimatization. Thereafter, the animals were moved to the mouse Inhalation Facility at URMC for training and exposures.

One week prior to the start of the exposures, mice underwent a 5-day nose-only training to adapt themselves to the mesh restraints of the exposure tower. The mouse restraint durations were increased gradually to minimize the animal’s stress and discomfort. Of note, the mouse sacrifice was performed within 8–12 weeks’ age for each mouse group to ensure that the mouse age corresponds to the age of adolescents (12–17 years) in humans (Dutta and Sengupta, 2016; Jackson et al., 2017). Age and sex-matched animals (n = 2/sex/group) used to perform scRNA seq were considered as the ‘experimental cohort’; whereas another group of age and sex-matched (n = 3/sex/group) mice exposed to air and flavored e-cig aerosol served as ‘validation cohort’ for this study.

We utilized an eVic-VTC mini and CUBIS pro atomizer (SCIREQ, Montreal, Canada) with a BF SS316 1.0-ohm coil from Joyetech (Joyetech, Shenzhen, China) for vaping and the inExpose nose-only inhalation system from SCIREQ (SCIREQ, Montreal, Canada) for mouse exposures. Both air and PG:VG exposed mice groups were considered as controls for this experiment. We used commercially available propylene glycol (PG; EC Blend) and vegetable glycerin (VG; EC Blend) in equal volumes to prepare a 50:50 solution of PG:VG. For flavored product exposures, mice were exposed to three different e-liquids – a menthol flavor ‘Menthol-Mint’, a fruit flavor ‘Mango’ and a tobacco flavor ‘Cuban Blend’. Of note, all the e-liquids were commercially manufactured with 50 mg/ml of TDN. So, all treatments have nicotine in addition to the flavoring mentioned, respectively. Additionally, we used a mixture of PG:VG with 50 mg/ml of TDN as a control for limited experiments to study the effect of nicotine alone in our treatment. This group is labeled PG:VG + Nic for the rest of the manuscript.

Scireq Flexiware software with the InExpose Inhalation system was used for controlling the Joyetech eVic-VTC mini device to perform nose-only mouse exposures. For this exposure, we utilized a puffing profile that mimicked the puffing topography of e-cig users in two puffs per minute with a puff volume of 51 ml, puff duration of 3 s, and an inter-puff interval of 27 s with a 2-l/min bias flow between puffs (Lee et al., 2018a). Figure 1—figure supplement 1A depicts the experimental design and exposure system employed for this study.

Age-matched male and female (n = 5 per sex) mice were used for each group, namely, air, PG:VG, fruit, menthol, and tobacco. To ensure rigor and reproducibility in our work, we have used age- and sex-matched control and treated mice in this study. Confounders like environment and stress were minimized by housing all the cages in environmentally controlled conditions and training all the mice (both control and treated) in nose-only chambers. Each group of mice was exposed to the above-mentioned puffing profile for 1 hr each day (120 puffs) for a total of five consecutive days. Additionally, a group of mice (n = 3/sex) was exposed to PG:VG + Nic for the same duration using similar exposure profile to serve as control to assess the effect of nicotine on the observed changes using selected experiments. Air-exposed mice were exposed to the same puffing profile for a total of five consecutive days to ambient air. We recorded the temperature, humidity, and CO levels of the aerosols generated at the start, mid, and end of the exposure on each day using the Q-Trak Indoor Air Quality Monitor 7575 (TSI, Shoreview, MN). Total particulate matter (TPM) sampling was done from the exhaust tubing of the setup at the 30 min mark of the exposure and at the inlet connected to the nose-only tower (shown in Figure 1—figure supplement 1A) immediately after the culmination of the exposure. Gravimetric measurements for TPM were also conducted to confirm relative dosage to each mouse group daily.

The animals (at 8–10 weeks’ age) were sacrificed immediately after the final exposure. Vascular lung perfusion was performed using 3 ml of saline before harvesting the lung lobes for preparation of single-cell suspension. It is important to mention here that of the 5 lung samples/sex/group; 2/ sex/group were used for histological assessments and scRNA analyses. Here, the left lung lobe was inflated using low-melting agarose and used for histology, while the rest of the uninflated lung lobes were used for preparing the single-cell suspension. We pooled the lung lobes for each sex per group for preparation of the single-cell suspension as depicted in Figure 1A. The lung lobes were weighed and digested using the Liberase method as described earlier (Kaur, 2024). Briefly, lung lobes were weighed and digested using Liberase (Cat# 5401127001; Roche, Basel, Switzerland) enzymatic cocktail with 1% DNase. The tubes were then transferred to the gentleMACS dissociator (Miltenyi, Gaithersburg, MD) and the manufacturer’s protocol for mouse lung digestion was run. The sample tubes were next incubated at 37°C for 30 min with constant rotation, after which the suspension was strained through 70 µm MACS Smart Strainer. Thereafter, the suspension was centrifuged at 500 × g for 10 min at 4°C, the supernatant was discarded, and 0.5 ml of RBC lysis buffer was added to the cell pellet to digest RBCs. The suspension was left on ice for 5 min in RBC lysis buffer and then 4 ml of ice-cold PBS with 10% FBS was added to stop the lysis. The suspension was again centrifuged at 500 × g for 10 min at 4°C. The cell pellet was suspended in 1 ml PBS with 10% FBS, and cell number and viability were checked using AO/PI staining on a Nexcelom Cellometer Auto2000.

The prepared single-cell suspension was sent to the Genomics Research Center (GRC) at URMC for library preparation and single-cell sequencing. Library preparation was performed by control and treatment groups using the 10X single-cell sequencing pipeline by 10X Genomics, and 10,000 cells were captured per sample using the Chromium platform. The prepared library was sequenced on NovaSeq 6000 (Illumina, San Diego, CA) at a mean sequence depth of 30,000 reads per cell. Read alignment was performed to GRCm38 Sequence.

We used the standard Seurat v4.3 analyses pipeline to analyze our data (Hao et al., 2021). In brief, the low-quality cells and potential doublets were excluded from the dataset to create the analyses dataset. The residual features due to the presence of RBCs were corrected before integration. ‘scTransform’ function was used for integration of all the datasets, after which the standard Seurat pipeline was used for data normalization of integrated data. ‘FindVariableGenes’ gene function was used to identify the variable genes for dimensionality reduction using PCA function. UMAP was used for dimensionality reduction and clustering of cells.

To identify the unique features and cell clusters within each cell subtype, we used the sub-setting feature within Seurat. After identifying the five major cell populations (epithelial, endothelial, stromal, myeloid, and lymphoid) in our datasets, each of these cell types was sub-clustered using the ‘subset’ function, normalized, and re-clustered. Cell annotation for each of the subsets was performed with the help of the Tabula Muris database (The Tabula Muris Consortium et al., 2018; Hurskainen et al., 2021). However, some clusters were annotated manually with the help of a literature search.

DESeq2 (V.1.42.1) was used to perform pseudobulk analyses to identify DEGs within each group. Here, genes showing a fold change >0.5 and <–0.5 along with a p_adj value <0.05 were considered significantly dysregulated and plotted as heatmap using GraphPad. The ClusterProfiler R package (V. 4.10.1) (Yu et al., 2012) was employed to perform gene enrichment analyses of the DEGs (fold change >0.5 and <–0.5).

We used a multiplex assay to determine the levels of cytokine/chemokine in the lung homogenates from control and e-cig aerosol exposed mouse lungs using commercially available Bio-Plex Pro Mouse Chemokine Assay (Cat# 12009159, Bio-Rad, Hercules, CA) per the manufacturer’s instructions. Approximately 40 mg of mouse lung lobes were homogenized in 300 µl of 1X RIPA buffer with 0.1% protease and phosphatase inhibitor. The lung homogenate was stored on ice for 30 min. Following incubation, the homogenate was centrifuged at 15,000 rpm for 15 min at 4°C. The supernatant was collected and used for performing the multianalyte assay for determination of cytokine/chemokine levels using Luminex FlexMap3D system. A heatmap after normalization (Z-score) of the measured cytokine/chemokine to the protein amount loaded was plotted.

The left lung lobe of mice used for scRNA seq was inflated with 1% low melting agarose and fixed with 4% neutral buffered PFA. Fixed lungs were dehydrated before being paraffin‐embedded and sectioned (5 μm). H&E staining was performed by the Histology, Biochemistry, and Molecular Imaging Core at URMC. The H&E stain was observed at ×10 magnification using Nikon Elipse‐Ni fluorescence microscope. Ten to fifteen random images were captured per sample.

Flow cytometry was performed on the cells collected from lung homogenates from air and flavored e-cig aerosol exposed mouse lungs. For analyses of immune cell population in the lung, the lung lobes were digested as described earlier (Kaur, 2024). The single-cell suspension thus prepared was used to run flow cytometry using the BD LSRFortessa cell analyzer. Cells were blocked with CD16/32 (Tonbo biosciences 70-0161 u500, 1:10) to prevent nonspecific binding and stained with a master mix of Siglec F (BD OptiBuild Cat# 740280, 1:200), CD11b (Biolegend Cat# 101243, 1:200), Ly6G (BD Horizon Cat# 562700, 1:200), CD45 (Biolegend Cat# 103126, 1:200), CD11c (Biolegend Cat# 117318, 1:200), CD4 (Biolegend Cat# 116012, 1:200), and CD8 (eBiosciences Cat# 17-0081-82, 1:200). 7AAD (eBiosciences Cat# 00-6993-50, 1:10) was used as the nucleic acid dye to detect live and dead cells. The gating strategy used for this assay has been depicted in Figure 3—figure supplement 2.

To understand the levels of metals released during subsequent days of exposure, we performed ICP-MS on the e-cig aerosol condensates collected from each day of exposure using a Perkin Elmer ICP-MS model 2000C. The samples were run using a Total Quant KED protocol with 4 ml/min Helium flow and externally calibrated using a blank and a 100 ppb standard for the 51 elements. The samples were submitted to the Element Analyses facility at URMC, and levels of metals thus detected were plotted.

To determine the various populations of neutrophils in exposed and control groups, FFPE tissue sections from air and tobacco-flavored e-cig aerosol exposed lungs were stained with Ly6G and S100A8. In brief, 2–3 tissue sections per sample were deparaffinized using serial incubation in xylene followed by graded alcohol. Slides were incubated in 1X Citrate Buffer (Cat# S1699, Agilent, Santa Clara, CA) for 10 min at 95°C for antigen retrieval, which was followed by incubation at room temperature for 30 min. The slides were next washed with water and permeabilized using a permeabilization buffer (0.1% Triton-X in 1X TBST) for 10 min. Next, the slides were again washed with 1X TBST and blocked using Blocking buffer (5% goat serum in 1X TBST) for 30 min at room temperature. The blocked slides were incubated overnight at 4°C with Ly6G (Cat# 16-9668-85, Invitrogen, dilution: 1:100) and S100A8 (Cat# 26992-1-AP, Proteintech, dilution: 1:200). The next day, the slides were washed with 1X TBST and incubated for 2 hours at room temperature with goat anti-rabbit Alexa Fluor 594 (Cat# A11012, Invitrogen) and donkey anti-mouse Alexa Fluor 488 (Cat# A21202, Invitrogen) secondary antibody at 1:1000 dilution. Thereafter, the slides were washed and mounted with ProLong Diamond Antifade Mountant with DAPI (Cat# P36962, Invitrogen, Waltham, MA). Six to ten images were captured using Zoe Fluorescent Cell Imaging System (Bio-Rad Laboratories , Hercules, CA) at ×20 magnification. The ImageJ deconvolution was used for quantifying the fluorescence in green (Ly6G⁺) and red (S100A8⁺) channels relative to DAPI (blue channel), and the relative fluorescence was plotted.

We used GraphPad Prism 10.5.0 for all statistical calculations. All the data plotted in this paper are expressed as mean ± SEM. Pairwise comparisons were done using unpaired t test while one‐way analysis of variance (ANOVA) with ad hoc Tukey’s test was employed for multi-group comparisons. To identify sex-based variations in our treatment groups, Tukey post hoc two-way ANOVA was employed.

Data collection was performed with mkfastq pipeline in Cell Ranger’s (v7.0.1). Cell Ranger (v7.0.1) was used for cell and gene counting using the default settings. Single-cell analysis was performed using the Seurat R package (v4.3.0) using the recommended workflow.