Research ArticleCELL BIOLOGY

Characterizing smoking-induced transcriptional heterogeneity in the human bronchial epithelium at single-cell resolution

See allHide authors and affiliations

Science Advances  11 Dec 2019:
Vol. 5, no. 12, eaaw3413
DOI: 10.1126/sciadv.aaw3413
  • Fig. 1 scRNA-Seq of human bronchial cells from never and current smokers.

    (A) Bronchial brushings were procured from the right mainstem bronchus of six never smokers and six current smokers. Bronchial tissue was dissociated, single cells were isolated by fluorescence-activated cell sorting (FACS), and single-cell RNA libraries were prepared and sequenced. (B) t-distributed stochastic neighbor embedding (t-SNE) was performed to illustrate transcriptomic relationships among cells. Donor smoking status (NS, never smoker; CS, current smoker) was visualized for each cell as well as expression of bronchial cell type marker genes [z-normalized transcripts per million (TPM) values] across all cells: KRT5 (basal), FOXJ1 (ciliated), SCGB1A1 (club), MUC5AC (goblet), and CD45 (WBC). (C) An unsupervised analytical approach (LDA) was used to identify distinct cell clusters and sets of coexpressed genes. Cell clusters were defined by unique gene set expression patterns, and never or current smoker cell enrichment was assessed.

  • Fig. 2 Characterization of bronchial cluster transcriptomic profile, cell type, and smoking status.

    (A) Global transcriptomic profiles of 13 bronchial cell clusters were defined by expression of unique combinations of 19 gene sets and visualized by heatmap (z-normalized TPM values). (B) A MetaGene was generated for each gene set (GS-1 to GS-19), and mean cluster-specific expression was designated: high (pink), medium (white), low (light gray), or not expressed (dark gray). (C) Mean expression of marker genes was summarized for each cluster designated: high (pink), medium (white), low (light gray), or not expressed (dark gray). (D) Per-cluster percentage of total cells and the ratio of never and current smoker cells were calculated, and per-cluster statistical enrichment (FDR q < 0.05, indicated in blue) of NS or CS cells was assessed.

  • Fig. 3 A smoking-induced detoxification program was observed in ciliated cells.

    (A) Expression of gene sets GS-2, GS-3, and GS-7 in clusters C-5 and C-11 was visualized by heatmap (z-normalized TPM values). (B) Cluster C-5 was split into never and current smoker subsets, and expression of GS-8 genes was visualized by heatmap. (C) Bronchial tissue procured from an independent cohort of never and current smokers (UMCG cohort, table S2) was immunostained for AKR1B10, Ac-α-Tub, and KRT8. Representative images of never smoker (left) and current smoker (right) tissue were displayed. Arrows specify examples of AKR1B10+ ciliated cells (Ac-α-Tub+). (D) An increase in tissue length (μm)–normalized numbers of AKR1B10+ Ac-α-Tub+ cells was observed in current smokers relative to never smokers [P = 7.4 × 10−7, Wilcoxon rank-sum (WRS) test].

  • Fig. 4 Smoking is associated with increased numbers of goblet cells and decreased numbers of club cells in the bronchial epithelium.

    (A) Expression of gene sets GS-19, GS-17, GS-13, and GS-1 in clusters C-1, C-8, and C-3 was visualized by heatmap (z-normalized TPM values). Bronchial tissue procured from an independent cohort of never and current smokers (UMCG cohort, table S2) was immunostained for MUC5B and MUC5AC. (B) Representative images of never smoker tissue, MN current smoker tissue, and current smoker GCH were displayed. Arrows specify examples of MUC5B+, MUC5B+ MUC5AC+, and MUC5AC+ cells. Changes in tissue length (μm)–normalized numbers of (C) MUC5B+ cells (MN decrease, P = 0.02; GCH decrease, P = 1.8 × 10−5), (D) MUC5B+ MUC5AC+ cells (GCH decrease, P = 0.02), and (E) MUC5AC+ cells (MN increase, P = 1.5 × 10−6; GCH increase, P = 7.4 × 10−7) were observed (WRS test) in current smoker MN and GCH tissue relative to never smokers (WRS test). (F) Average proportions of MUC5B+, MUC5B+ MUC5AC+, and MUC5AC+ cells observed in never smokers, as well as MN and GCH current smoker tissue are displayed.

  • Fig. 5 A previously unidentified subpopulation of PG cells was observed in the airways of smokers.

    (A) Expression of gene sets GS-12, GS-16, GS-15, and MUC5AC in clusters C-3 and C-9 was visualized by heatmap (z-normalized TPM values). (B) t-SNE was used to visualize cluster C-3 and C-9 cells as well as (C) CEACAM5 expression (z-normalized TPM values) across all cells. (D) Bronchial tissue procured from an independent cohort of never and current smokers (UCL cohort, table S3) was immunostained for CEACAM5, KRT8, and MUC5AC. Representative images of never smoker tissue and current smoker GCH were displayed. Arrows specify examples of CEACAM5+ KRT8+ MUC5AC PG cells. (E) A significant increase in tissue length (μm)–normalized numbers of CEACAM5+ KRT8+ MUC5AC cells in current smoker GCH tissue, relative to never smokers, was observed (P = 0.004, WRS test).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/12/eaaw3413/DC1

    Table S1. Bronchial brushings were procured from six never smokers and six current smokers.

    Table S2. Bronchial tissue was obtained by lung resection from four never smokers and four current smokers at the UMCG.

    Table S3. Bronchial tissue was obtained by lung resection from five never smokers and five current smokers at the UCL Hospital.

    Fig. S1. Single bronchial cells were isolated by FACS.

    Fig. S2. scRNA-Seq data quality were evaluated for each donor.

    Fig. S3. Low-quality cells were excluded from downstream analyses.

    Fig. S4. Bronchial brushings reconstructed in silico from single-cell data resemble data generated from bulk bronchial brushings.

    Fig. S5. LDA was used to identify Cell-States and Gene-States.

    Fig. S6. Gene-State and Cell-State model optimization.

    Fig. S7. LDA was used to identify 13 cell clusters.

    Fig. S8. LDA was used to identify 19 gene sets.

    Fig. S9. Gene set expression across cell clusters.

    Fig. S10. T cell receptor genes were detected in CD45+ cell cluster.

    Fig. S11. Cluster 13 cells expressed CFTR.

    Fig. S12. Distributions of cell clusters within each subject.

    Fig. S13. Smoking-associated differential expression of each gene set was analyzed in published bulk bronchial brushing data.

    Fig. S14. Nonciliated cell AKR1B10 expression was uncommon.

    Fig. S15. MN and GCH tissue regions were distributed throughout the bronchial airways of current smokers.

    Fig. S16. Basal cell numbers were not altered in smokers.

    Fig. S17. Increased numbers of indeterminate KRT8+ cells were observed in GCH smoker tissue.

    Fig. S18. PG cells were enriched in regions of GCH within the airways of smokers.

    Fig. S19. Smoking-induced heterogeneity was observed in the human bronchial epithelium.

    Extended table S1. Primer sequences for scRNA-Seq.

    Extended table S2. Statistical modeling results, State Specificity, and State Similarity values for all genes.

    Extended table S3. Functional annotation results for each gene set.

  • Supplementary Materials

    The PDFset includes:

    • Table S1. Bronchial brushings were procured from six never smokers and six current smokers.
    • Table S2. Bronchial tissue was obtained by lung resection from four never smokers and four current smokers at the UMCG.
    • Table S3. Bronchial tissue was obtained by lung resection from five never smokers and five current smokers at the UCL Hospital.
    • Fig. S1. Single bronchial cells were isolated by FACS.
    • Fig. S2. scRNA-Seq data quality were evaluated for each donor.
    • Fig. S3. Low-quality cells were excluded from downstream analyses.
    • Fig. S4. Bronchial brushings reconstructed in silico from single-cell data resemble data generated from bulk bronchial brushings.
    • Fig. S5. LDA was used to identify Cell-States and Gene-States.
    • Fig. S6. Gene-State and Cell-State model optimization.
    • Fig. S7. LDA was used to identify 13 cell clusters.
    • Fig. S8. LDA was used to identify 19 gene sets.
    • Fig. S9. Gene set expression across cell clusters.
    • Fig. S10. T cell receptor genes were detected in CD45+ cell cluster.
    • Fig. S11. Cluster 13 cells expressed CFTR.
    • Fig. S12. Distributions of cell clusters within each subject.
    • Fig. S13. Smoking-associated differential expression of each gene set was analyzed in published bulk bronchial brushing data.
    • Fig. S14. Nonciliated cell AKR1B10 expression was uncommon.
    • Fig. S15. MN and GCH tissue regions were distributed throughout the bronchial airways of current smokers.
    • Fig. S16. Basal cell numbers were not altered in smokers.
    • Fig. S17. Increased numbers of indeterminate KRT8+ cells were observed in GCH smoker tissue.
    • Fig. S18. PG cells were enriched in regions of GCH within the airways of smokers.
    • Fig. S19. Smoking-induced heterogeneity was observed in the human bronchial epithelium.

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Extended table S1 (Microsoft Excel format). Primer sequences for scRNA-Seq.
    • Extended table S2 (Microsoft Excel format). Statistical modeling results, State Specificity, and State Similarity values for all genes.
    • Extended table S3 (Microsoft Excel format). Functional annotation results for each gene set.

    Files in this Data Supplement:

Stay Connected to Science Advances


Editor's Blog

Navigate This Article