Single-cell transcriptome analysis revealed the heterogeneity of ALI-HBE.
Primary HBECs were induced by ALI culture to differentiate into ALI-HBE, which is a pseudostratified bronchial epithelium consisting of ciliated cells, secretory cells (goblet cells and club cells), and basal cells (Fig. 1). The ciliated cells and goblet cells are located on the apical side (exposed to the air) of the ALI-HBE, whereas basal cells are located on the basolateral side (submerged in the liquid culture medium).
The effects of common e-liquid ingredients were evaluated independently by exposing fully differentiated ALI-HBE to vehicle control or Saucy e-liquid, both with or without nicotine. The ALI cultures were subsequently digested into single-cell suspensions and subjected to multiplex scRNA-seq (Fig. 2a). A total of 15874 effective cells were collected to evaluate the cellular heterogeneity of ALI-HBE and its transcriptomic responses to e-vapors. Among these, eight distinct clusters were identified by unsupervised cluster analysis (Fig. 2b). Three predominant cell types were identified by their corresponding marker genes, including basal cells, secretory cells, and ciliated cells (Fig. 2c and Figure S1). Dying and proliferating cells were also identified by high expression of mitochondrial genes (MT high cells) and mitosis-related genes (G2M high cells), respectively. In addition, two small groups of cells were found co-expressing markers of ciliated cells and basal cells (ciliated basal cells, CBCs) or ciliated cells and secretory cells (ciliated secretory cells, CSCs). Comparisons among the cell types revealed that the transcriptomes of basal cells and secretory cells were similar to each other and relatively distinct from the transcriptome of ciliated cells (Fig. 2d). Additionally, comparisons within cell types showed that transcriptome similarity was lowest among the secretory cells and highest among the basal cells (Fig. 2e), suggesting that the secretory cells possessed higher heterogeneity. The proportions of major cell types were variable across different treatments (Fig. 2f), suggesting that acute exposure to flavoring or nicotine may lead to cell type transformation in the bronchial epithelium.
Acute exposure to nicotine e-vapor affected gene expression related to secretory cell differentiation and secretion.
To further elucidate secretory cell heterogeneity and transcriptional changes after exposure to e-vapor, we analyzed the transcriptomes of 4857 secretory cells separately, which revealed three main clusters (Fig. 3a, left). The proportions of each cluster of secretory cells were relatively stable across samples (Fig. 3a, right). Basal-like secretory cells (BSCs) were identified by higher expression of basal cell markers (S100A2 and KRT19) and lower expression of secretory cell markers (MUC5B and SCGB3A1), whereas terminal secretory cells (TSCs) showed the opposite pattern (Fig. 3b). Another distinct cluster, which we named intermediate secretory cells (ISCs), expressed both basal and secretory marker genes. Combination of RNA velocity with diffusion mapping revealed a basal-to-secretory differentiation trajectory, suggesting that the basal-to-secretory transition was a continuous process that was not affected by acute exposure to e-vapors (Fig. 3c).
Next, we performed a differential expression analysis to elucidate potential effects of exposure to nicotine e-vapor with or without flavoring on secretory cells. Exposure to nicotine e-vapor without flavoring resulted in upregulation of 116 genes and downregulation of 104 genes relative to the expression levels in cells exposed to vehicle control without nicotine (Fig. 3d). Exposure to nicotine e-vapor with flavoring resulted in fewer (10) up-regulated genes and more (140) down-regulated genes compared with exposure to nicotine e-vapor without flavoring (Fig. 3e), suggesting that the influence of nicotine on secretory cells might be different under conditions of distinct flavoring. Forty-four genes were significantly co-down-regulated after exposure to nicotine e-vapor with or without flavoring. Among these, eight genes were involved in epithelial cell differentiation (AKR1C1, AKR1C2, CTSB, DHRS9, LGALS3, GSTK1, KRT13, and UPK1B), and six genes were involved in calcium ion binding (S100A10, S100A14, S100A4, S100A6, GSN, and TKT). Notably, among the co-down-regulated genes, 21 genes (51.1%) were related to extracellular exosomes, and 13 (28.9%) were secreted proteins, suggesting that nicotine might impair the secretory function of human airway epithelium. Differential expression tests were also performed separately for each of the three main clusters of secretory cells (Figure S2). Fourteen genes were co-upregulated in all three clusters by exposure to nicotine e-vapor without flavoring, most of which were related to stress response. On the other hand, the genes that were co-down-regulated in all three clusters by exposure to nicotine e-vapor were similar with or without flavoring.
Pathway enrichment analysis revealed that exposure to nicotine e-vapor with or without flavoring resulted in downregulation of similar pathways (Fig. 3h, 3i, and 3j), such as formation of cornified envelope, keratinization degradation of extracellular matrix, assembly of collagen fibrils and other multimeric structures, and neutrophil degranulation, which might be related to the differentiation and secretory function of the cells[23]. Consistent with the observations of differentially expressed genes, more pathways were down-regulated than were up-regulated after exposure to nicotine e-vapor, especially in the presence of flavoring, and no gene ontology (GO) biological process was co-enriched by exposure to nicotine e-vapor with or without flavoring.
Acute exposure to nicotine e-vapor promoted basal-to-secretory transformation.
Next, we analyzed the transcriptomes of 2226 basal cells identified by expression of conventional basal cell markers. Differential expression of specific markers revealed three distinct clusters of basal cells: secretory-like basal cells (SBCs), intermediate basal cells (IBCs), and progenitor-like basal cells (PBCs; Fig. 4a and 4b). SBCs displayed relatively high secretory scores, estimated by the expression of secretory cell markers (LCN2, TSPAN8, BPIFB1, and SCGB1A1). In addition, SBCs showed high expression of S100A9, SLPI, and SERPINB3, which were also highly expressed in secretory cells (Fig. 4b and Figure S3). These top marker genes were highly related to epithelial cell differentiation, and the gene set for this process was enriched in SBCs (Figure S4a, GO:0030855, GSEA p = 0.004). These observations suggested that SBCs were likely to initialize differentiation from basal cells to secretory cells and had begun to show secretory characters, such as the biological process of secretion by tissue (Figure S4b, GO:0032941, GSEA p = 0.005).
Compared with SBCs, PBCs expressed much higher levels of TP63 (Figure S4c), a traditional epithelial basal cell marker, and also presented a high level of progenitor cell regulation. GSEA results showed that PBCs were enriched with expression of genes involved in stem cell differentiation (Fig. 4c, GO:0048863, GSEA p = 0.007). They were also relatively highly enriched with expression of biological processes involved in cell adhesion, cell matrix, fiber assembly and substrate junction, and external encapsulating structure organization (Figure S4f), indicating that the stemness of basal cells plays important roles in maintaining the structure and environment of bronchial epithelium in the ALI model. Additionally, PBCs were enriched with expression of transforming growth factor beta (TGFB) production pathways (Figure S4d and S4e), suggesting they might be involved in cell proliferation, differentiation, and growth. IBCs expressed marker genes of both SBCs and PBCs and showed intermediate levels of secretory and basal scores, suggesting they were in an intermediate state between the other two basal subtypes.
Pseudotime trajectory analysis revealed a path of basal cell differentiation (Fig. 4d). Combined with RNA velocity analysis, the trajectory could be divided into two directions. One represented the basal-to-secretory transition, which was mostly occupied by SBCs with high secretory scores (Fig. 4e). The other branch, which was mostly occupied by PBCs, had higher basal scores (Fig. 4f) and represented basal cells shifting into a stem/progenitor-like status to maintain renewal and proliferation potential. Notably, acute exposure to nicotine e-vapor increased the proportion of SBCs and decreased the proportion of PBCs (Fig. 4g), suggesting that nicotine might promote the transformation of basal cells into secretory cells and thus impair the stemness maintenance of bronchial epithelial basal cells.
Differential expression analysis revealed a series of genes that were significantly up-regulated or down-regulated after exposure to nicotine e-vapor (Fig. 4h). Among the up-regulated genes, bone marrow stromal cell antigen 2 (BST2) belongs to the IFN-stimulated genes (ISGs) family, which is a limited gene set that controls viral infection by inhibiting viral RNA synthesis, viral assembly/egress, and viral entry. In addition, FOS and JUNB are the key members forming the transcription factor complex AP-1, which is related to basal-to-squamous cell carcinoma transition[24]. Similar to the observation in secretory cells, no genes were co-up-regulated after exposure to nicotine e-vapor with or without flavoring, suggesting that the influence of nicotine on basal cells might be different under conditions of distinct flavoring (Fig. 4j). Nevertheless, seven genes were co-down-regulated after exposure to nicotine e-vapor with or without flavoring (Fig. 4k). Three of these genes were involved in negative regulation of cell proliferation (DAVID, GO: 0008285, p = 0.016), suggesting that nicotine might promote basal cell proliferation (especially in SBCs) by down-regulating genes that inhibit cell growth, which is consistent with the observation that nicotine may promote basal-to-secretory transition. Furthermore, a series of biological functions were co-down-regulated by exposure to nicotine e-vapor with or without flavoring (Fig. 4l), including genes involved in stimulus responses, cell communication, cell development, and wound healing.
Acute exposure to nicotine e-vapor with flavoring might promote susceptibility to virus infection.
In addition to nicotine, the effects of flavoring were investigated (Figure S5). Several genes were differentially expressed after acute exposure to e-vapor with flavoring with or without nicotine. Among these, five genes were involved in cytokine signaling in immune response (FOS, JUNB, SOCS3, IFI27, and BST2; Fig. 5a–d). However, the number of genes that were up-regulated by exposure to e-vapor with flavoring was diminished in the presence of nicotine (Fig. 5c). These observations suggested that nicotine might affect how epithelial cells respond to the flavorings. Interestingly, in basal cells, many genes were up-regulated but only a few were down-regulated after exposure to e-vapor with flavoring, whereas ciliated cells showed the opposite pattern (Fig. 5a–d), suggesting that different types of epithelial cells might react distinctly to e-vapor flavoring. To better understand the pathways impacted by the flavoring, we identified significant differentially expressed genes using STRING databases and generated an interaction network (Fig. 5e). Exposure to e-vapor with flavoring increased expression of genes that interact with genes from the MAPK (MAPK9, MAPK8, MAPK14, MAPKAPK2) and JAK-STAT (JAK2 and EPOR) pathways, which might trigger the expression of a wide array of cytokines and growth factors to promote cell differentiation and growth. In addition, the same exposure down-regulated the genes BST2 and IFI27, which interact with the early growth response protein EGR1, further suggesting that the growth pattern of epithelial cells was switched in the presence of e-vapor flavoring.
To further investigate the potential connection between e-cigarette use and virus infection, we investigated five of the most relevant genes in SARS-CoV-2 infection (ACE2, TMPRSS4, TMPRSS2, CTSL, and BSG). The expression of ACE2, the receptor for the spike glycoprotein of human coronavirus, was increased in basal cells after exposure to e-vapor with flavoring (Fig. 5f), suggesting that bronchial epithelial basal cells might be more susceptible to SARS-CoV infection after exposure to e-vapor containing flavoring. In line with this observation, expression of a series of pathways related to SARS-CoV infection was enriched in ALI-HBE basal cells after exposure to e-vapor with flavoring regardless of whether the e-vapor contained nicotine (Fig. 5g and 5h).
Acute exposure to e-vapor affected epithelial cell interactions.
To investigate how the ALI-HBE cells communicate with each other, we constructed cell-to-cell interaction networks based on ligand-receptor pair databases (Fig. 6a). We found that basal cells, especially PBCs and IBCs, which possess high stemness potential, were more likely than other cell types to interact with each other and with other cell types (Fig. 6b and 6c). In addition, PBCs and IBCs were the most self-regulating cells in the ALI culture system, suggesting that they might possess higher potential to adjust to the growth environment. Comparatively, secretory cells showed less participation in cell communication, and ciliated cells showed the least evidence of intracellular communication.
The top four pathways through which basal cells interacted with other basal cells were the MK, LAMININ, MIF, and COLLAGEN pathways (Fig. 6d). In the ALI-HBE model, MIF interacted strongly with ACKR3 and the CD74 and CD44 complex (Figure S6c). The main interactions enriched in secretory cells included the MK, PTN, and EPHA pathways (Fig. 5d; Figure S6a and S6e). MK was also the most extensively interacting pathway in the entire ALI-HBE.
Next, we compared changes in bronchial epithelial cell interactions after acute exposure to e-vapor. The results showed that exposure to nicotine e-vapor without flavoring had a significant effect on cell interactions, which enhanced the ability of secretory cells to secrete signals, whereas PBCs were the least affected (Fig. 6e; Figure S7a). An analysis of signaling pathways showed that the EGF pathway was the most altered pathway after exposure to nicotine e-vapor without flavoring (Fig. 6f; Figure S7b). In the presence of flavoring, the ability of most cell types to express and receive signals after exposure to nicotine e-vapor was reduced, and the CDH and JAM pathways were significantly inhibited (Fig. 6e–h).