If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Microbiology & Immunology, United StatesCenter for Translational Research in Infection and Inflammation, School of Medicine, Tulane University, New Orleans, LA, United States
Corresponding author at: Center for Translational Research in Infection and Inflammation, Tulane School of Medicine, JBJ 375, 333 S. Liberty St, New Orleans, LA 70112, United States.
Polarization of cells dramatically increases cytokine secretion in bronchial epithelial cells.
•
CFTR F508del mutation itself is sufficient to significantly alter the epigenome and transcriptome of epithelial cells.
•
The F508del mutation is closely associated with immunity-related pathways in the epithelial cells in the absence of infection.
•
These data may serve as epithelial biomarkers for CFTR mRNA therapies.
Abstract
Background
Cystic fibrosis (CF) is characterized by chronic inflammation and excessive cytokines secretion in the lung. Isogenic human CF bronchial epithelial (CFBE41o-) cell lines stably expressing wt-CFTR (WTBE) or F508del mutant (CFBE) are widely used tools in understanding responses to stimuli or drugs and CF pathogenesis in vitro. However, the intrinsic cellular differences in culture are unknown.
Methods
We performed integrative analyses of these isogenic cells at the protein, mRNA, and chromatin levels in the submerged and air-liquid interface (ALI) conditions to determine cell intrinsic effects of mutant versus complemented CFTR expression.
Results
CFBE and WTBE cells displayed different cytokine secretion patterns, including IL-6, IL-8, CXCL1, CXCL10, and CCL5. The ALI culture dramatically increased cytokine secretion in both cells. Assay for transposase-accessible chromatin using sequencing (ATAC-seq) result showed different chromatin landscapes upon polarization and CFBE cells, compared to WTBE cells, exhibited higher genome-wide chromatin accessibility under both culture methods. At the transcriptome level, differentially expressed genes identified by mRNA sequencing between two cell lines were highly concentrated in immunity-related pathways.
Conclusions
This multilayered study shows that expression of wild-type CFTR has an epithelial cell intrinsic effect on the cell's epigenome and transcriptome particularly in immunity relevant activities. These data will serve as a resource for the CF community and may serve as epithelial biomarkers for CFTR mRNA therapy.
Cystic fibrosis (CF) is the most common life-threatening genetic disease in people with northern European ancestry. It is caused by mutations in the cystic fibrosis transmembrane regulator (CFTR) gene, which encodes a chloride/bicarbonate channel [
]. According to the Cystic Fibrosis Foundation, today over 2000 mutations have been identified, and 80%−90% of patients have at least one copy of the phenylalanine 508 deletion (F508del). The F508del mutation results in a CFTR misfolding defect and absent functional surface protein, leading to insufficient ion transport primarily in epithelial cells [
]. This disorder involves multiple organs, including the lung, intestine, pancreas, and sweat glands, but pulmonary diseases account for the majority of morbidity and mortality. This is in part due to a complex interplay of chronic pulmonary infection as well as innate and adaptive immune responses to the chronic infection. Additionally, there is evidence that misfolded or absent CFTR may intrinsically affect lung epithelial inflammatory responses [
]. However, there are several caveats of CF primary cells, including varying host genetic differences and the fact that many donors have significant lung disease. Thus, it is difficult in this system to isolate the effect of CFTR complementation per se.
To better characterize tools used for studying CFTR complementation, we undertook a comprehensive transcriptomic and epigenomic analysis on isogenic immortalized CFBE41o- cell lines, which were initially isolated from a CF patient homozygous for the F508del mutation and stably transfected with wild-type or F508del CFTR cDNA using HIV-based lentiviral vectors (named WTBE cells and CFBE cells, respectively, in this study) [
]. These isogenic subclones possess a high level of transgene-derived CFTR mRNA expression. Indeed, the cAMP-activated chloride transport of CFTR in WTBE cells has been restored via exogeneous expression of wt-CFTR, whereas endogenous F508del CFTR mRNA is at a minimal/undetectable level [
]. Stable transfection of CFBE subclones with F508del CFTR provided a vector control to account for the effect of exogenous gene expression on WTBE and CFBE cell lines. Thus, we thought this system would be a good model for epithelial intrinsic changes that may occur upon CFTR complementation in CFTR gene replacement trials.
In this study, we performed an integrative study using RNA-seq to assay the transcriptome as well as ATAC-seq to assay chromatin structure [
]. Additionally, since CF lung disease is characterized by neutrophilic inflammation, we assayed the secretion of specific cytokines and chemokines in cell supernatants and their expression at the DNA and mRNA levels. Lastly, to understand how cell polarity affects the epigenomic and transcriptomic landscape, we conducted this study in submerged (SUB) as well as air-liquid interface (ALI) culture to model loss of polarities such as squamous metaplasia that can occur in COPD and other chronic lung diseases.
2. Materials and methods
2.1 Cell lines
Human bronchial epithelial CFBE41o- cells stably expressing F508del-CFTR (CFBE) or wild-type CFTR (WTBE) were cultured in MEM medium, supplemented with 10% fetal bovine serum (Hyclone), 0.5% Penicillin/Streptomycin and 2 mM L-glutamine. 0.5 µg/mL puromycin (InvivoGen) were added for WTBE while 2 µg/mL puromycin for CFBE cells. Cells were maintained in a humidified chamber with 5% CO2 at 37 °C. For ALI-condition, cells were seeded on Millicell cell culture insert (Millipore) placed in 24-well plates as previously described with modifications [
]. Media on the apical surface was removed daily until complete confluency (dry membrane). Confluent cells were cultured for at least another 10days for polarization, replacing the basal medium every three days while keeping the apical surface dry. In submerged condition, cells had reached confluency before sample collection.
2.2 Luminex analysis
Overnight medium cytokine accumulations (basolateral side for ALI culture) were measured using Milliplex MAP Human Cytokine/Chemokine Magnetic Beads (Millipore) and Bio-Plex Human Multiplex Immunoassays (Bio-Rad) according to the manufacturer's protocols and analyzed using a Bio-Plex Manager software (Bio-Rad). The comparison of cytokine concentrations between SUB and ALI conditions were normalized by culture surface area (insert 0.6cm2 and 24-well plate 1.9cm2) and medium volume.
2.3 Transposase-Accessible chromatin sequencing
ATAC-seq was performed as described with modifications [
]. Typically, 60,000 cells (>95% viability) were trypsinized and lysed in 50μL cold lysis buffer (10 mM Tris-HCl, 10 mM NaCl, 3 mM MgCl2, 0.1% NP40 and Tween-20, and 0.01% Digitonin) and incubated on ice for 5 min. Nuclei were pelleted and resuspended in 50 µL of transposition mixture (TDE1 tagment DNA enzyme and TD buffer, Illumina) and incubated for 30 min at 37 °C. Purified DNA products were pre-amplified for 5 cycles with NEBNext High-Fidelity 2x PCR MasterMix (NEB). Additional cycles were added based on qPCR results. PCR reactions were purified and qualified using an Agilent High Sensitivity DNA Bioanalyzer Kit (Agilent). The libraries were quantified using Qubit (ThermoFisher) and sequenced on NextSeq 550/2000 system. Raw data were uploaded to GEO (GSE186918).
Paired-end reads were aligned to the human hg19 database (GRCh37.p13) using BWA. Sorted bam files were uploaded and analyzed in the Galaxy platform (https://usegalaxy.org/) [
]. The mitochondrial reads and reads with mapQuality <30 were removed. Duplicate reads were cleaned by MarkDuplicates tool. The narrowpeaks were called using MACS2 from the converted BED files. Normalized bigwig files were generated by Wig/BedGraph-to-BigWig tool and visualized on IGV platform. The differentially accessible peaks were analyzed and plotted as principal component analysis (PCA) using DiffBind. Peak annotations and feature distributions were performed by ChipSeeker (Bioconductor), and KEGG pathways based on differentially accessible genes were analyzed by PathfindR in R studio.
2.4 Next-generation RNA sequencing
Total RNA was extracted using RNeasy Mini Kit (Qiagen) and quantified by Qubit. Quality of RNAs was assessed using an Agilent 2100 Bioanalyzer. The library was prepared using Illumina TruSeq Stranded mRNA sample prep kit as described [
] and sequenced on Illumina NextSeq 2000 system. Raw reads were trimmed and mapped on Banana Slug Analytics Platform. Differentially expressed genes were analyzed using EdgeR, and pathway analyses were performed using Advaita iPathway platform. For volcano plots, genes with a p_value beyond the lowest computer calculation were assigned a -Log10p_value of 314. PCA plots and dimensional contributors were conducted using ggfortify and factoextra (CRAN) packages. All raw data were deposited in GEO (GSE186919).
2.5 Actinomycin D treatment and quantitative RT-PCR (qRT-PCR)
Actinomycin D (Sigma) was dissolved in dimethyl sulfoxide (DMSO) and a final concentration of 10μg/mL was added at confluency [
]. RNAs were collected at 0, 1, 2, 4, 6 and 8 h time points using RNeasy Mini Kit. cDNA was prepared using iScript Reverse Transcriptase MasterMix (Bio-Rad). Then qPCR was performed with Bio-Rad CFX96 system using TaqMan Universal PCR Master Mix and premixed Taqman® primers (ThermoFisher). The mRNA decay rates (half-life) were calculated by the relative abundance of each time point to t = 0.
2.6 Statistics
Graphs were analyzed using GraphPad Prism 9.2. All specific statistical tests are defined in the figure legends and reported as means ± standard error of the mean (SEM). The mRNA decay rate was estimated by non-linear regression curve fitting. P values are annotated as follows *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 and ****p ≤ 0.0001.
3. Results
3.1 Comparison of cytokine/chemokine profiles between CFBE and WTBE cells
To assess the intrinsic expression of cytokines/chemokines in both cell lines, we first assayed a panel of 41 cytokines/chemokines in the supernatants of SUB-cultured CFBE and WTBE cells. The level of most cytokines showed no differences between CFBE and WTBE cells (Supplemental Table S1), but the levels of interleukin (IL)-6, IL-8, C-X-C motif chemokine ligand 1 (CXCL1), CXCL10, colony-stimulating factor (CSF)-3, granulocyte-macrophage CSF (GM-CSF), and C-C motif chemokine ligand 5 (CCL5) were significantly increased in the supernatant of CFBE cells (Supplemental Fig. S1), compared with that of WTBE cells. Based on these data, we customized a 7-plex Luminex assay and confirmed that CFBE cells secreted significantly more IL-6, CXCL10, and CCL5 and slightly higher IL-8 and CXCL1 overnight under SUB conditions (Fig. 1A). Next, to study basolateral cytokine/chemokine secretion, CFBE and WTBE cells were cultured in the air-liquid interface (ALI) environment and overnight cytokine secretions were assayed. Results showed that the polarization to ALI dramatically increased cytokine secretions in both WTBE and CFBE cells (Fig. 1A). In contrast to the SUB condition, we observed that ALI WTBE cells secreted significantly greater amounts of CXCL10 and IL-8 but a similar amount of IL-6, CCL5, and CXCL1 under ALI conditions, compared with ALI CFBE cells (Fig. 1A). Similarly, qPCR studies revealed consistent differences in mRNA expression of the above cytokines/chemokines (Fig. 1B).
Fig. 1Relative cytokine production in submerged and ALI-cultured WTBE and CFBE cells. (A) Fold changes of IL-6, IL-8, CXCL1, CXCL10 and CCL5 production from WTBE and CFBE cells relative to WTBE sample. (n = 3/group from 4 experiments). FC, fold change. (B) RT-qPCR validation of cytokine mRNA expressions. Fold changes are shown relative to submerged WTBE. Comparisons were analyzed by Tukey's multiple comparison test, * p <0.05; ** p<0.01; *** p<0.001; ****, p <0.0001.
3.2 Genome-wide chromatin accessibility in CFBE and WTBE cells
To investigate the open chromatin landscape of these cells, we assessed transposase-accessible chromatin in both cells using ATAC-seq. We first examined the correlation of chromatin accessibility under the SUB culture condition. Principal component analysis (PCA) of total ATAC-seq reads showed that the CFBE samples clustered together whereas the WTBE samples clustered separately with no overlap (Fig. 2A), suggesting significant divergence of the chromatin landscape of the two cell lines.
Fig. 2Differences in chromatin accessibility of submerged-cultured WTBE and CFBE cells. (A) Principal component analysis (PCA) plot from bulk ATAC-seq (n = 3). (B) Average distribution of open chromatin regions (OCRs) enriched at transcriptional start site (TSS, −1 kb to +1 kb) in three samples. (C) Genomic distribution of OCRs (Percentages %) and (D) Summary of promoter (<=1 kb) frequency in each cell type, the comparison was performed by one-tailed unpaired t-test. (E) Volcano plots of genes with differentially chromatically accessible peaks and (F) percentages of their corresponding functional regions. (G) Top 10 representative KEGG pathways enriched in differentially chromatically accessible genes of CFBE cells, ranking based on p_value (cutoff: adjusted p<0.001, absolute Log2FC ≥ 1). (H) Representative chromatin accessibility profiles of cytokine genes, including IL6, CXCL8, CXCL1, CXCL10, CCL5, in WTBE (Blue) and CFBE (Rose) cells. Black box: significantly differential enriched loci, along with peak score on the right. Comparisons were analyzed by unpaired t-test, * p <0.05; ** p<0.01. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).
Furthermore, the openness of different genomic regions might have various impacts on mRNA transcription, and thus we conducted unbiased genome-wide chromatin analyses. The read count frequency around transcriptional start sites (TSSs) of all peaks (average 132,372 per WTBE sample and 103,923 per CFBE sample) were analyzed. CFBE samples had noticeably higher average count frequencies (−1 kb to +1 kb to the TSSs) than WTBE samples (Fig. 2B, Supplemental Fig. S2A).The genomic distribution of functional regions also revealed a significant difference in the percentage of peaks in promoter regions (Fig. 2C and D). CFBE cells had considerably more peaks associated with proximal promoters (≤1 kb) but a similar mapping percentage of distal promoters (1kb-3 kb) as well as other functional regions such as untranslated regions (UTRs) and exons. After peak annotation, we found CFBE cells displayed more accessible peaks than WTBE cells (7656 vs 914) while majority of these peaks fell into intron and distal intergenic regions (Fig. 2E and F). Further we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis among those peaks using an adjusted p_value < 0.01 and enrichment fold change of absolute Log2FC ≥ 1. Under these criteria, we identified substantial upregulation of CF-related or/and immunity-associated pathways (Fig. 2G, Supplemental Table S2), including the TGF-β and Wnt signaling pathways.
We next focused on the genes encoding the cytokines shown in Fig. 1 to assess if there were any differences in the accessibility (shown by enrichment peak score) of their open loci in regulatory regions such as promoters and TSSs. We found that the accessibility of IL6 locus was similar in both CFBE and WTBE cells. In contrast, peaks located around the TSS of CXCL10 and in the promoter region of CCL5 were statistically more enriched in CFBE cells (Fig. 2H). Interestingly, we observed a third pattern of openness shown in CXCL8 and CXCL1 genes, where the accessibility of indicated peaks was significantly higher in WTBE cells compared to CFBE cells (Fig. 2H).
3.3 Chromatin accessibility in polarized cells
To understand whether CFBE and WTBE cells maintain the same chromatin features under different culture conditions, we also performed an ATACseq analysis on ALI-cultured cells. After polarization, the PCA plot of narrowpeaks revealed a complete separation between ALI CFBE and ALI WTBE samples (Fig. 3A). The overview of genomic features in ALI-cultured cells exhibited a similar pattern to that of the SUB condition (Fig. 2B–E). The distribution of open chromatin regions (OCRs) near the TSSs (−1 kb to +1 kb) was again notably higher in the CFBE samples compared with WTBE cells (average 158,497 peaks per WTBE sample and 109,604 per CFBE sample) (Fig. 3B, Supplemental Fig. S2B), suggesting the higher open status of chromatin in CFBE cells was maintained after polarization. Moreover, the differences in the genomic distribution of peaks in the promoter region, as well as the percentage of the adjacent promoter (≤1 kb) were also higher in ALI CFBE cells than ALI WTBE cells (Fig. 3C and D). The ALI CFBE cells displayed more accessible peaks than ALI WTBE cells while majority of these peaks were in intron and distal intergenic regions (Fig. 3E and F).The majority of the top 10 enriched pathways in ALI CFBE cells were comparable to the SUB condition, including the TGF-β signaling pathway and the Wnt signaling pathway (Fig. 3G, Supplemental Table S2). Additionally, there was an enrichment of the Th17 cell differentiation pathway in ALI CFBE samples (Fig. 3G).
Fig. 3Differences in chromatin accessibility of ALI-cultured WTBE and CFBE cells. (A) PCA plot of the ATAC-seq triplicates for each cell. (B) Profiles of open chromatin regions (OCRs) enriched at TSS (−1 kb to +1 kb) in each sample (n = 3). (C) Genomic distribution of OCRs (Percentages %) and (D) Summary of promoter (<=1 kb) frequency in each cell type after combining replicates, the comparison was performed by one-tailed unpaired t-test. (E) Volcano plots of differentially chromatically accessible genes and (F) percentages of their corresponding functional regions (Down). (G)Top 10 representatives of KEGG pathway enriched in differentially chromatinally accessible genes in ALI CFBE cells, ranking on p_value (cutoff: adjusted p<0.001, absolute Log2FC ≥ 1).(H) Representative chromatin accessibility profiles of cytokine genes, including IL6, CXCL8, CXCL1, CXCL10, CCL5 in ALI WTBE (Green) and ALI CFBE (Orange) cells. Comparisons were analyzed by unpaired t-test, * p <0.05; ** p<0.01. (I) Comparison of average open chromatin frequency during cell polarization. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).
Since the production of several cytokines—including IL-6, IL-8, CXCL1, CXCL10, and CCL5—was dramatically elevated in both cell lines after polarization (Fig. 1A), we visualized and quantified the narrow peak enrichments in these regions under ALI culture condition. Interestingly, the loci accessibility of these target cytokines did not closely correlate with secreted protein expression. Instead, we observed significantly lower peak enrichment in the CCL5 and CXCL1 loci from ALI CFBE cells compared to ALI WTBE cells but similar chromatin openness in the IL6, CXCL10, and CXCL8 loci (Fig. 3H).
Finally, to understand the overall chromatin accessibility changes during cell polarization, we compared the accessible peak count frequency between submerged- and ALI-cultured cells. To our surprise, we found that, regardless of cell types, polarized cells displayed lower openness than their submerged counterparts (Fig. 3I). This result suggested that polarization lowers the overall accessibility even though the secretion of our target cytokines increased.
To understand whether chromatin accessibility was associated with differential gene expression in WTBE and CFBE cells, we conducted mRNA sequencing of these cells under SUB and ALI culture conditions. Based on the fragments per kilobase of transcript per million mapped reads (FPKM) of the four groups, PCA plots showed that samples were separated by both cell type and the culture conditions (Fig. 4A). The major dimensional contributors on dimension one are shown in the biplot (Supplemental Fig. S3A). Particularly, the majority of contributors explaining the PCA differences in SUB CFBE versus SUB WTBE cells are concentrated in two cellular pathways (Fig. 4B): organization of the cytoskeleton pathway containing thymosin beta 4 X-linked (TMSB4X), beta-actin (ACTB), tumor protein translationally-controlled 1 (TPT1), and myosin light chain 6 (MYL6); and the protein synthesis pathway including ribosomal protein L35, L12, L26, L7A and lateral stalk subunit P1 (RPL35, RPL12, RPL26, RPL7A, and RPLP1). Some of these genes in the two pathways also contributed to differences in the PCA plot of cells under the ALI condition, such as TPT1 and RPL35, suggesting that these genes are differentially expressed independent of culture condition. In addition, we identified several immunity-related genes from the contributors, specifically in ALI culture, including fibronectin1 (FN1), beta-2-microglobulin (B2M), and ferritin heavy chain1 (FTH1) (Fig. 4C).
Fig. 4Bulk RNA-seq analysis of WTBE and CFBE cells under same culture condition. (A) PCA analysis of RNA-seq analysis in submerged and ALI-cultured cells. (B and C) Top10 contributory genes in dimension 1 of the PCA plot in submerged culture conditions (B) and ALI-conditions (C). (D and E) Volcano plots of differentially expressed genes in WTBE and CFBE cells under two conditions. All volcano plots were analyzed by EdgeR and plotted with Padj <0.01. Gene labels: 5 highest enrichment ratio (±) and/or 5 lowest p_value. (F) Venn diagram of differentially regulated genes mapped by RNA-seq and ATAC-seq under both conditions with Padj<0.01. (G and H) Top10 differential pathways in submerged- & transwell- cultured cells. Differentially expressed pathway genes on the intestinal immune network for IgA production were specifically clarified.
Next, we compared the differentially expressed genes between WTBE and CFBE cells. We found differential expression of 6658 genes under the SUB culture condition (Fig. 4D) and 4872 genes under ALI culture conditions (Fig. 4E), using the threshold of adjusted p_value ≤ 0.01. We also found that CFBE cells had upregulated expression of MSN, FAR2, ZNF502 and downregulated expression of ZNF43, FAM71F1, C5orf46 (among the top 5 genes with the greatest fold change) in comparison with WTBE cells, regardless of culture condition. All genes that were significantly differentially expressed are listed in supplemental Table S3. We next integrated the differentially expressed genes measured by RNA-seq with the ones identified from the ATAC-seq data. Venn Diagram analysis of genes under the SUB culture condition revealed that 952 differentially expressed genes were detected both by ATAC-seq and mRNA-seq, while the majority were only reported by either ATAC-seq or mRNA-seq (cut-off: adjusted p_value≤ 0.01) (Fig. 4F). This trend was also true for samples from the ALI condition (Fig. 4F, supplemental Table S4). For instance, CXCL1, CXCL10 differences were only detectable under mRNA-seq analysis. Taken together, these results suggested that chromatin accessibility near a gene locus is not linearly associated with mRNA transcription of that gene.
sdTo uncover potential pathways in the transcriptomic differences between WTBE and CFBE cells, we performed the KEGG pathway analysis of differentially expressed genes. Five of the top ten pathways enriched in CFBE cells were closely associated with immune responses and cytokine secretion, regardless of culture condition (Fig. 4G and H), including cytokine-cytokine receptor interaction, cell adhesion molecules, and viral protein interaction with cytokine and cytokine receptors. Interestingly, the intestinal immune network for IgA production pathway was enriched explicitly in the ALI culture condition, in which human leukocyte antigens (HLA-DQA1, HLA-DOA), polymeric immunoglobulin receptor (PIGR), and CCL28 were highly upregulated in CFBE cells, compared with WTBE cells (Fig. 4H).
3.5 Polarization partially changes the gene expression landscape
To assess the impact of polarization on the transcriptome, we next analyzed our RNA-seq data to compare gene expression differences between polarized and submerged cells for each cell type. Results showed that gene expressions were significantly altered during polarization. 1818 genes of WTBE cells (Fig. 5A) and 2172 genes of CFBE cells (Fig. 5B) were significantly changed, as illustrated on the volcano plots. The gene expression of representative cytokines IL6, IL8, CXCL1, CXCL10, and CCL5 were significantly increased after polarization of each cell type, consistent with higher protein secretion measured by Luminex (Fig. 1A). We again performed pathway analysis using all differentially expressed genes of WTBE cells (Fig. 5C) and CFBE cells (Fig. 5D) and detected mostly immunity-associated pathways among the top ten enriched pathways (7 of 10). These pathways were shared between CFBE and WTBE cells (Fig. 5C and D). Notably, we also found antigen processing and presentation and staphylococcus aureus infection pathways were significantly induced during polarization in both cell lines.
Fig. 5Integrative RNAseq analysis of WTBE and CFBE cells. (A and B) Volcano plots of significantly expressed genes in submerged- versus transwell- cultured WTBE and CFBE cells. Cutoff: Padj<0.01, abs (Log2FC) ≥1. Genes with a p_value beyond the lowest computer calculation were assigned a -Log10p_value of 314. (C and D) Top10 differential pathways between two culture conditions of WTBE cells (C) and CFBE cells (D), shown in bar plots. (E) Venn diagram of the overlap of upregulated genes (Log2FC≥1) in polarization. Bottom left: Top 10 upregulated genes unique in WTBE cells. Bottom middle: Fold change heatmap of top 10 upregulated common genes, ranking based on CFBE Log2FC value. Bottom right: Top 10 upregulated genes unique CFBE cells. (F) The gene expression of IL17RA (Up) and IL17RC (Down) in both cell lines undergoing polarization. FPKM, fragments per kilobase of transcript per million mapped reads. Comparisons were performed using unpaired t-test, ** p<0.01, ***p<0.001. (G)Top10 common upregulated pathways during polarization (Ranking on enrichment). (H) Venn diagram representing the overlap of significantly differential pathways in the meta-analysis of the four RNAseq results. The six common significant differential pathways are shown on the right. (I) Common predicted significant upstream regulators for three comparisons involving transwells-cultured condition, including potential activator (Up) and inhibitors (Down). Adapted from the Advaita iPathway platform.
A Venn diagram analysis was performed to compare genes upregulated during polarization in both cell lines. We found 871 common genes induced in WTBE and CFBE cells (Fig. 5E, Supplemental Table S5). Ranking on the fold changes of expression in polarized CFBE cells, we identified CCL2, S100 calcium-binding proteins (S100A7A, S100A7, S100A8), and peptidoglycan recognition protein 4 (PGLYRP4) were primarily induced in both WTBE and CFBE cells. Noteworthy, we also found the expression of IL17RA and IL17RC was sigifinantly increased during polarization, independent of the cell type (Fig. 5F). Further integrated pathway analysis on the common elevated genes reiterated the significance of IL-17/Th17 signaling and TNF/chemokine signaling pathway during the polarization of those two epithelial cells (Fig. 5G, Supplemental Table S6).
We then performed an integrative meta-analysis to understand how cell types and cell polarization change cellular activities holistically. Six common pathways were significantly impacted by both cell types and polarization status, half of them were immunity-related (Fig. 5H). In order to provide more insights regarding molecular and cell biology between different cell types and polarization status, we also analyzed the common gene ontology (GO) terms, including biological processing, molecular function and cell compartment (Supplemental Fig. 3B–D). Interestingly, molecular function result showed significant differences in the transporter-related activities among four comparisons, which is consistent with CFTR function as a primary chloride/bicarbonate transporter.
Interestingly, no common regulators were found in the meta-analysis of four comparisons. However in any comparisons involving ALI condition, IL17A was predicted as the only upstream activator which again pinpoint the importance of IL-17 signaling pathway during cell polarization (Fig. 5I). Meanwhile, fibroblast growth factor 2 (FGF2), G-protein signaling modulators (GPSMs) 1–3 and Purkinje cell protein 2 (PCP2) were predicted as upstream inhibitors (Fig. 5I).
3.6 Some cytokine mRNAs exhibit culture-dependent decay kinetics
As cytokine/chemokine expression is also regulated by mRNA stability, we examined the mRNA stability of target cytokines/chemokines and assayed their half-lives by inhibiting new transcriptions using actinomycin D. Under the SUB condition, CFBE cells exhibited longer half-lives of IL6, CXCL10, CXCL8 mRNA but significantly shorter CXCL1 decay (Fig. 6A) compared with WTBE cells. Interestingly, CCL5 mRNA showed very little decay in either WTBE or CFBE cells. The decay kinetics of some target mRNAs changed when the cells were polarized. The half-lives of IL6, CXCL8 and CXCL1 mRNA were slightly prolonged especially in WTBE cells after polarization, similar to the ones in CFBE cells (Fig. 6B), which might partially contribute to the higher cytokine secretion (Fig. 1A). Notably, the degradation of CXCL10 mRNA was largely slowed down after polarization (incalculable within our 8hr testing window). The CCL5 mRNA again showed minimal degradation in both polarized cells (Fig. 6B).
Fig. 6RNA stability in WTBE and CFBE cells. mRNA decay curves of representative cytokines in submerged-(A) and transwells-(B) cultured cells were measured using actinomycin D treatment followed by qRT-PCR. Half-life was calculated using one-phase decay (n = 4). The mRNA half-life (Hour) values are in the upper right corner of each plot. Statistical comparisons were measured by the Extra sum-of-squares F test, p_value in the bottom left corner.
This study provides the first comprehensive transcriptome analyses on gene expression and cytokine/chemokine productions in matched isogenic CF model cells, where expression of WT CFTR is the sole difference. In this case, we will be able to interpret the cellular changes truly brought by CFTR mRNA complementation and provide a well-understood cell model for the CF research community. Currently, two sets of stable CFBE41o- subclones are widely used. Both were generated by CFTR cDNA electroporation but using two different vectors: an EBV-based episomal expression vector pCEP4 [
Promoter hypomethylation of Toll-like receptor-2 gene is associated with increased proinflammatory response towardbacterial peptidoglycan in cystic fibrosis bronchial epithelial cells.
]. Both sets of cell lines have been proved to express wt-CFTR or F508del-CFTR mRNA and display transgene-derived phenotypes. The pCEP4 transfected WTBE cells secrete higher intrinsic IL-6 and IL-8 concentrations than isogenic CFBE cells under ALI condition [
]. In contrast, our analyses (lentiviral vector transfected cells) showed an elevated IL-8 but similar IL-6 level in ALI WTBE cells compared with ALI CFBE cells. Although minor changes exist between the cytokine production of these two sets of cells, these data suggested that the F508del mutation in polarized CFBE cells compromises the immune responses, supporting the idea that immunological defects in CF exist prior to infection.
We further demonstrated that polarization using ALI culture has a dramatic effect on the chromatin landscape, mRNA expression, and cytokine secretion in both cell lines. Even though the genome-wide chromatin openness of cells was similar between ALI and SUB conditions, the accessibility of specific genes was different. For instance, the CXCL10 gene in ALI cells (both WTBE and CFBE cells) displayed substantially greater detectable peaks (accessible sites) than their SUB counterparts. The significantly differentiated pathways during polarization in mRNA-seq results were highly overlapped between CFBE and WTBE cells including antigen processing and presentation, cytokine-cytokine receptor signaling, and chemokine signaling pathway. These pathways were widely upregulated, indicating more sensitive responses to antigens or stimuli in polarized cells than the submerged cells. Therefore, culture condition clearly affects cell activities, we need to carefully choose the culture system in vitro and maximize the similarities between cell culture and the real physiological environment.
We also identified three genes, FN1, B2M, and FTH1, which are the top contributors in the PCA analysis among ALI WTBE and CFBE cells. Quaresma et al. found that airway epithelium from CF patients lost a well-defined layer of fibronectin (FN1) and aberrantly activated epithelial-mesenchymal transition (EMT) [
]. Shown by our mRNA-seq data, this loss can be compensated by wt-CFTR complementation as the ALI WTBE cells expressed significantly higher FN1 mRNA than ALI CFBE cells. B2M is a critical component of major histocompatibility complex (MHC) class I, exerting pH-dependent antimicrobial activities in human broncho-alveolar-lavage (BAL) fluid. More specifically, B2M can destroy bacterial membranes under an acidic pH [
]. In our study, CFBE cells had slightly upregulated B2M expression which may contribute to abnormal innate immune system in CF lung disease, where there is an acidic bronchial microenvironment. FTH1, a subunit of the iron storage protein ferritin, is involved in iron accumulation, innate immunity, and the ferroptosis pathway. Lower expression of the FTH1 gene in ALI CFBE cells may lead to a potential loss of intracellular iron storage whereas extracellular iron accumulation in the airway could facilitate biofilm formation of Pseudomonas aeruginosa (P. aeruginosa) [
]. This proposed mechanism is consistent with relative iron deficiency in CF patients, which is directly related to the increased severity of pulmonary pathogenesis [
]. While the close association of these genes with CF suggests their potential as CF markers, additional studies are needed to better define these relationships.
Several cytokine-related pathways were upregulated in CFBE cells suggesting their preexisting proinflammatory status. For example, the enrichment in TGF-β and Wnt signaling pathway in CFBE cells, shown by our ATAC-seq data, are coherent with the conclusion that activation of those pathways drives inflammatory cytokine secretions and exacerbates CF pathology [
]. Among all cytokine-related pathways, the IL-17A associated pathways were compelling, providing considerable evidences about the role of IL-17 signaling pathway in CF lung pathogenesis. Researchers have shown that IL-17A and IL-17F are elevated in the P. aeruginosa colonized sputum of CF patients during pulmonary exacerbation and are reduced under antibiotic therapy [
Role of IL-17A, IL-17F, and the IL-17 receptor in regulating growth-related oncogene-α and granulocyte colony-stimulating factor in bronchial epithelium: implications for airway inflammation in cystic fibrosis.
]. In addition, early inflammation in the CF lung is potentially initiated by Th17 cells, which secrete IL-17 to promote neutrophil recruitment and prime the CFTR dysfunctional airway epithelial cells into a hyperinflammatory state [
]. Our mRNA-seq analyses showed that both WTBE and CFBE cells upregulated the IL-17 signaling pathway (Rank 1 of 10) and Th17 cell differentiation pathway (Rank 5 of 10) during polarization, but upregulation scale was greater in CFBE cells. Indeed, the accessibility of genes in the Th17 cell differentiation pathway was enhanced more than twofold in CFBE cells compared to WTBE cells. Moreover, IL-17A, which was predicted as the most potential upstream activator of all differentially expressed genes between CFBE and WTBE cells, is critical in Th17 effector cell function [
]. Overall, our results re-emphasize the importance of the IL-17 signaling pathway in CF epithelial cells under a physiological environment.
Taken together, we have profiled the gene expression of WTBE and CFBE cells (under SUB and ALI culture) and revealed their culture-dependent changes in gene expression and regulation. Importantly, cells cultured fully submerged or on transwells are not interchangeable. The CFTR mutation in CFBE cells not only led to ion transport defects but also changed the cellular responses in many aspects. One limitation of this paper is that these corrected cells overexpress CFTR which may impact the implications of these findings in cells expressing physiological CFTR levels in non-CF individuals. However, these data may be relevant to subjects undergoing heterologous CFTR expression such as in the context of CFTR mRNA therapy. For instance, our data suggest a combination of upregulation in ZNF43, FAM71F1, C5orf46 and downregulation in FAR2, MSN, ZNF502 may serve as potential biomarkers for CFTR complementation in CFTR null-patients undergoing CFTR mRNA therapy. ZNF43 is a gene that belongs to the C2H2-type zinc finger gene family involved in tissue development. Data in the human protein atlas (www.proteinatlas.org) shows expression in human bronchial epithelium but whether it is regulated by CFTR expression per se is unclear. FAM71F1 has mostly been studied in spermatogenesis as it controls acrosome assembly and male fertility [
] which is a phenotype in CFTR deficient patients and whether there is a direct link of CFTR and FAM71F1 in these tissues remains to be determined.C5orf46 is expressed in epithelial cells and has recently been reported to encode an antimicrobial protein with activity against gram negative bacteria [
]. Overall, these data provided a comprehensive resource for the CF community and support the existence of immunological changes in the CF airway independent of infections.
Funding
This work was supported by the Louisiana Board of Regents Endowed Chairs for Eminent Scholars program, as well as by PHS grant R35HL139930.
CRediT authorship contribution statement
Shiping Lu: Methodology, Formal analysis, Data curation, Validation, Visualization, Writing – original draft, Writing – review & editing. Jay K. Kolls: Conceptualization, Funding acquisition, Supervision, Methodology, Writing – original draft, Writing – review & editing.
Declaration of Competing Interest
All authors declare no conflict of interest.
Acknowledgments
The authors thank Dr. Raymond A Frizzell (University of Pittsburgh, PA) for his generous gift of the WTBE and CFBE cells and MaryJane Jones (Tulane University, LA) for the assistance in operating Bio-Rad Luminex assay. We would like to thank Dr. Kejing Song and Dr. Sylvia Hilliard for assistance in sequencing. We also thank Dr. Janet McCombs for reviewing the paper.
Promoter hypomethylation of Toll-like receptor-2 gene is associated with increased proinflammatory response towardbacterial peptidoglycan in cystic fibrosis bronchial epithelial cells.
Role of IL-17A, IL-17F, and the IL-17 receptor in regulating growth-related oncogene-α and granulocyte colony-stimulating factor in bronchial epithelium: implications for airway inflammation in cystic fibrosis.
00597-5&id=gr1.jpg){kind=link}
00597-5&id=gr2.jpg){kind=link}
00597-5&id=gr3.jpg){kind=link}
00597-5&id=gr4.jpg){kind=link}
00597-5&id=gr5.jpg){kind=link}
00597-5&id=gr6.jpg){kind=link}