Advertisement

Inactivation of CFTR by CRISPR/Cas9 alters transcriptional regulation of inflammatory pathways and other networks

Open ArchivePublished:May 21, 2019DOI:https://doi.org/10.1016/j.jcf.2019.05.003

      Highlights

      • Non-stimulated cells with or without functional CFTR behave similarly
      • TNFα, but not IL1β, evokes differential responses that are CFTR-dependent
      • Transcriptional network regulation is affected by absence of CFTR

      Abstract

      Background

      Individuals with cystic fibrosis (CF) experience elevated inflammation in multiple organs, but whether this reflects an inherent feature of CF cells or is a consequence of a pro-inflammatory environment is not clear.

      Method

      Using CRISPR/Cas9-mediated mutagenesis of CFTR, 17 subclonal cell lines were generated from Caco-2 cells. Clonal lines with functional CFTR (CFTR+) were compared to those without (CFTR) to directly address the role of CFTR in inflammatory gene regulation.

      Results

      All lines maintained CFTR mRNA production and formation of tight junctions. CFTR+ lines displayed short circuit currents in response to forskolin, while the CFTR lines did not. Baseline expression of cytokines IL6 and CXCL8 (IL8) was not different between the lines regardless of CFTR genotype. All lines responded to TNFα and IL1β by increasing IL6 and CXCL8 mRNA levels, but the CFTR lines produced more CXCL8 mRNA than the CFTR+ lines. Transcriptomes of 6 CFTR and 6 CFTR+ lines, before and after stimulation by TNFα, were compared for differential expression as a function of CFTR genotype. While some genes appeared to be differentially expressed simply because of CFTR's absence, others required stimulation for differences to be apparent.

      Conclusion

      Together, these data suggest cells respond to CFTR's absence by modulating transcriptional networks, some of which are only apparent when cells are exposed to different environmental contexts, such as inflammation. With regards to inflammation, these data suggest a model in which CFTR's absence leads to a poised, pro-inflammatory state of cells that is only revealed by stimulation.

      1. Introduction

      Inflammation is a property of numerous tissues in CF, including the airways, the gut, pancreas, liver, and probably others. Elevated cytokines and other inflammatory molecules are characteristic of cystic fibrosis (CF) airways [
      • Nichols D.P.
      • Chmiel J.F.
      Inflammation and its genesis in cystic fibrosis.
      ]. While altered airway surface environment and chronic pulmonary infections certainly contribute to the inflammatory response in the airway, there are studies that suggest this is not specific to airways, such that CF epithelial cells and cell models are inherently prone to hyper-inflammation [
      • Carrabino S.
      • Carpani D.
      • Livraghi A.
      • Di Cicco M.
      • Costantini D.
      • Copreni E.
      • et al.
      Dysregulated interleukin-8 secretion and NF-kappaB activity in human cystic fibrosis nasal epithelial cells.
      ]. Studies of the lungs of mice inoculated with bacteria showed that inflammatory markers rose higher in CF mice relative to wild type animals and failed to resolve inflammation as rapidly as non-CF animals [
      • Heeckeren A.
      • Walenga R.
      • Konstan M.W.
      • Bonfield T.
      • Davis P.B.
      • Ferkol T.
      Excessive inflammatory response of cystic fibrosis mice to bronchopulmonary infection with Pseudomonas aeruginosa.
      ], indicating inherent differences in inflammatory responses. Studies in transformed cell lines suggested this might reflect inherent properties of epithelial cells lacking CFTR [
      • Bartling T.R.
      • Drumm M.L.
      Oxidative stress causes IL8 promoter hyperacetylation in cystic fibrosis airway cell models.
      ]. We previously found reduced histone deacetylase 2 (HDAC2) in nasal epithelium freshly excised from CF patients [
      • Bartling T.R.
      • Drumm M.L.
      Loss of CFTR results in reduction of histone deacetylase 2 in airway epithelial cells.
      ] suggesting that epigenetic differences between CF and non-CF cells are involved. We also found that chromatin of the promoter of CXCL8, encoding IL8, was hyperacetylated in cell lines with reduced or absent CFTR expression [
      • Bartling T.R.
      • Drumm M.L.
      Loss of CFTR results in reduction of histone deacetylase 2 in airway epithelial cells.
      ,
      • Bartling T.R.
      • Drumm M.L.
      Oxidative stress causes IL8 promoter hyperacetylation in cystic fibrosis airway cell models.
      ]. These cell lines, however, were developed using heterologous CFTR expression [
      • Flotte T.R.
      • Solow R.
      • Owens R.A.
      • Afione S.
      • Zeitlin P.L.
      • Carter B.J.
      Gene expression from adeno-associated virus vectors in airway epithelial cells.
      ] or other sequences to alter CFTR function [
      • Kube D.
      • Sontich U.
      • Fletcher D.
      • Davis P.B.
      Proinflammatory cytokine responses to P. aeruginosa infection in human airway epithelial cell lines.
      ] and thus may circumvent endogenous regulatory processes controlling CFTR gene expression.
      Like the airway, the gastrointestinal tract of individuals with CF is also inflamed, coinciding with goblet cell hyperplasia and overproduction of mucus. The role of inflammation in CF gastrointestinal manifestations, such as malabsorption and poor nutritional status are being increasingly recognized [
      • De Lisle R.C.
      • Borowitz D.
      The cystic fibrosis intestine.
      ,
      • Dhaliwal J.
      • Leach S.
      • Katz T.
      • Nahidi L.
      • Pang T.
      • Lee J.M.
      • et al.
      Intestinal inflammation and impact on growth in children with cystic fibrosis.
      ,
      • Peuhkuri K.
      • Vapaatalo H.
      • Korpela R.
      Even low-grade inflammation impacts on small intestinal function.
      ]. In addition, individuals with CF are reported to be at much greater risk for inflammatory bowel disease [
      • Banville N.
      • Broderick A.
      • Fitzgerald R.
      • Drumm B.
      • McDermott M.
      Giant inflammatory polyposis coli as a manifestation of Crohn's disease in patients with coexistent cystic fibrosis.
      ,
      • Bresso F.
      • Askling J.
      • Astegiano M.
      • Demarchi B.
      • Sapone N.
      • Rizzetto M.
      • et al.
      Potential role for the common cystic fibrosis DeltaF508 mutation in Crohn's disease.
      ] and subsequent gastrointestinal cancer (reviewed in [
      • Hou Y.
      • Guan X.
      • Yang Z.
      • Li C.
      Emerging role of cystic fibrosis transmembrane conductance regulator - an epithelial chloride channel in gastrointestinal cancers.
      ] and [
      • De Lisle R.C.
      • Borowitz D.
      The cystic fibrosis intestine.
      ]), both of which are associated with GI inflammation. We have shown previously in mouse models by conditional expression of CFTR that inflammation and goblet cell hyperplasia in the small intestines is due to CFTR's absence from epithelial cells [
      • Hodges C.A.
      • Grady B.R.
      • Mishra K.
      • Cotton C.U.
      • Drumm M.L.
      Cystic fibrosis growth retardation is not correlated with loss of Cftr in the intestinal epithelium.
      ]. However, we could not determine from these studies whether the inflammation is due to dehydrated mucosa, responses to obstruction, altered intracellular signaling caused by CFTR's absence, or some other mechanism.
      Recently available genome editing technologies, particularly the CRISPR/Cas9 system, allow one to generate matched cell populations that differ at a single genomic locus. Unlike heterologous expression, cells engineered this way do not circumvent endogenous regulation of cellular processes. To study the effects of CFTR's absence on other genes and cellular processes such as inflammation, we used the CRISPR/Cas9 genome editing technology to generate subclonal cell lines from single cells of Caco-2 [
      • Fogh J.
      • Wright W.C.
      • Loveless J.D.
      Absence of HeLa cell contamination in 169 cell lines derived from human tumors.
      ], a colonic adenocarcinoma-derived line that expresses high levels of CFTR [
      • Chang E.B.
      • Bookstein C.
      • Vaandrager A.
      • DeJonge H.R.
      • Buse J.
      • Musch M.W.
      Cystic fibrosis transmembrane regulator mRNA expression relative to ion-nutrient transport in spontaneously differentiating human intestinal CaCo-2 epithelial cells.
      ]. Caco-2 cells have been reported to differ in cytokine induction upon siRNA-mediated knockdown of CFTR [
      • Crites K.S.
      • Morin G.
      • Orlando V.
      • Patey N.
      • Cantin C.
      • Martel J.
      • et al.
      CFTR knockdown induces proinflammatory changes in intestinal epithelial cells.
      ] and thus were used here to investigate cytokine regulation in models with the CFTR locus rendered inactive, akin to CF patients' cells.
      The goals of the work presented here were to determine if cells with and without functional CFTR were inherently pro-inflammatory in a non-stimulated context and whether they would respond differently to inflammatory stimuli, either of which might explain CF-associated inflammation. Caco-2 cells have been used to study inflammatory responses in gastrointestinal barrier function and those studies showed Caco-2 cells respond to stimuli such as TNFα by increasing CXCL8 (IL8) production [
      • Bahar B.
      • O'Doherty J.V.
      • Maher S.
      • McMorrow J.
      • Sweeney T.
      Chitooligosaccharide elicits acute inflammatory cytokine response through AP-1 pathway in human intestinal epithelial-like (Caco-2) cells.
      ,
      • Shin H.S.
      • Zhao Z.
      • Satsu H.
      • Totsuka M.
      • Shimizu M.
      Synergistic effect of tumor necrosis factor-alpha and hydrogen peroxide on the induction of IL-8 production in human intestinal Caco-2 cells.
      ,
      • Zhang L.
      • Shen J.
      • Cheng J.
      • Fan X.
      MicroRNA-21 regulates intestinal epithelial tight junction permeability.
      ], a key chemokine contributing to neutrophilia in CF lungs. We tested Caco-2 lines for temporal responses of pro-inflammatory genes to varying doses of inflammation-stimulating molecules such as E. coli lipopolysaccharide (LPS), tumor necrosis factor α (TNFα), and interleukin-1β (IL1β). Clonal lines that retained functional, intact CFTR were compared to those in which all CFTR alleles carried inactivating mutations. We found no difference in IL6 or CXCL8 mRNA expression prior to stimulation, but after stimulation the cells lacking CFTR appeared more sensitive to stimuli and achieved higher levels of CXCL8 mRNA, but not IL6, than cells with functional CFTR. Transcriptome profiling indicated the expression differences were not restricted to CXCL8 but rather a number of other networks were also affected by CFTR's absence.

      2. Methods

      2.1 Cell culture

      Caco-2 cells obtained from the American Type Culture Collection (ATCC) were maintained in Eagle's minimum essential medium with 10% FBS, 1% glutamine, and 1% penicillin-streptomycin. Cells were cultured at 37 °C in an atmosphere of 95% O2–5% CO2.

      2.2 Cell mutagenesis

      Plasmids were obtained from Addgene (pCas9 GFP, #44719; https://www.addgene.org/) or constructed using Invitrogen's GeneArt® Gene synthesis system.
      pCas9 GFP nuclease was used with a guide RNA (gRNA) targeting exon 11 of CFTR and a gRNA targeting the G551 codon in exon 12.
      A template plasmid carrying the G551D mutation was constructed and contained the G551D mutation along with 5 silent substitutions to prevent gRNA cleavage. The sequence for the template plasmid and gRNAs can be found in Table S1.
      Mutagenesis was carried out using the CRISPR/Cas9 system [
      • Mali P.
      • Yang L.
      • Esvelt K.M.
      • Aach J.
      • Guell M.
      • DiCarlo J.E.
      • et al.
      RNA-guided human genome engineering via Cas9.
      ]. To create loss of function mutations, 6 μg of exon 11 gRNA plasmid, 6 μg of exon 12 gRNA plasmid, and 12 μg of Cas9 GFP plasmid were added to the cells with 40 μl of Lipofectamine 2000. To create the G551D mutation, 6 μg of exon 12 gRNA plasmid, 6 μg of Cas9 GFP plasmid, and 12 μg of template plasmid were added with 40 μl of Lipofectamine.

      2.3 Clonal isolation and sequencing

      Forty-eight hours after transfection, transfected cells were isolated by florescence-activated cell sorting of GFP-expressing cells. The sorted cells were cultured using limiting dilution techniques to create subclonal populations. Each clone was grown and expanded. Once expanded, genomic DNA from each clone was isolated and amplified by PCR with primers recognizing sequences in exon 11 and/or exon 12 designed using Integrated DNA Technologies software. The sequences for the PCR primers can be found in Table S2. The most likely potential off-target cleavage sites for each gRNA, predicted by in silico analyses, were also sequenced in the lines and no sequence alterations were found. The use of two independent gRNAs and no detectable off-target changes strongly support the conclusion that cellular changes are the result of CFTR's loss of function.
      PCR products were sequenced and if Sanger sequencing identified mutations of interest, next-generation sequencing was carried out to define alleles. Sequences for next-generation PCR primers can be found in Table S2. Next-Generation sequencing was performed using paired end, 125 base pair sequencing via Illumina MiSeq. >50,000 reads were obtained from each clone and clones were sequenced multiple times after passaging to determine stability. RNA sequencing data are available under GEO accession GSE130226 at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE130226

      2.4 Functional characterization

      Short circuit currents (Isc) were measured with a modified Ussing chamber. Once confluent on transwell filters, cells were placed in asymmetric chloride solutions and 100 μM amiloride was added to block sodium absorption. Changes in transepithelial Isc and conductance in response to 10 μM forskolin were measured. Clones were further characterized by examining the effects of CFTR inhibition with 10 μM Inh172 on Isc induced by forskolin.
      Cells were assessed for CFTR protein by immunoblot. Cells were lysed with RIPA buffer containing protease inhibitors and incubated on ice for 30 min. Protein was quantified by BCA assay (BioRad). Samples were prepared in 2× Leammli sample buffer (BioRad) and incubated 30 min at room temperature. Ten micrograms of protein were loaded in a 7.5% precast polyacrylamide gel (BioRad). Protein was transferred to polyvinylidene fluoride microporous membranes (Millipore) using Trans-Blot Turbo Blotting System (BioRad). Membranes were blocked in 5% nonfat milk prepared in PBS with 0.01% Tween-20 for 2 h at room temperature. Membranes were incubated overnight with anti-CFTR (Antibody 596, 1:2000, CFTR Antibody Distribution Program). Goat anti-mouse IgG antibody, HRP conjugate (Millipore, AP181P) was used as the secondary antibody. Immunoblot signal was detected using a ChemDoc imaging system (BioRad).

      2.5 Cytokine stimulation

      Cells were exposed to either E. coli LPS, IL1β, or TNFα. Two sub-maximal stimulation doses were chosen to stimulate the cells based on dose response curves. The cells did not increase CXCL8 expression in response to LPS. The sub-maximal stimulating doses for IL1β were 0.1 ng/ml and 10 ng/ml and the sub-maximal stimulating doses for TNFα were 10 ng/ml and 100 ng/ml. Time courses were performed to determine time to peak response. Cells were stimulated for 2 h prior to harvesting RNA for RT PCR and RNAseq.

      2.6 RT qPCR and RNAseq

      CFTR, CXCL8 and IL6 mRNA were quantified by Real-Time Polymerase Chain Reaction (RT-PCR). RNA was harvested using RNAeasy kit per manufacturer instructions. One microgram of total RNA was reverse-transcribed into cDNA using qScript cDNA Synthesis Kit per manufacturer instructions. Two microliters of synthesized cDNA was amplified by quantitative PCR. Quantitative PCR was carried out using Taqman gene expression assays (Applied Biosystems). Relative expression was analyzed with the comparative cycle threshold method (2-∆∆CT). For CFTR expression, values were expressed as fold change compared to the average of CFTR+ group. For IL6 and CXCL8 expression, ∆CT was normalized to β-actin for each individual sample and then compared to the individual sample's untreated control.

      2.7 RNA expression analyses

      RNA seq data were analyzed for individual genes (CXCL8, IL6, ACTB) to compare with RT-PCR values and gene set enrichment was assessed to evaluate responses to TNFα, as well as to compare CF and non-CF cells under basal and stimulated conditions. The Molecular Signatures Database Hallmark Gene Set [
      • Liberzon A.
      • Birger C.
      • Thorvaldsdottir H.
      • Ghandi M.
      • Mesirov J.P.
      • Tamayo P.
      The molecular signatures database (MSigDB) hallmark gene set collection.
      ] was analyzed for enrichment of coordinately regulated genes. Those gene sets with nomimal p-values below 0.05 and a conservative false discovery rate q-value of <0.1 were considered significant.

      3. Results

      This study was designed to examine the effects of CFTR's absence on inflammatory processes in epithelial cell models. Caco-2 cells have been used extensively for studies of CFTR mRNA expression and protein function and we found these cells to be particularly amenable to subcloning. Using gRNAs targeting exon 11 and/or exon 12, we generated a panel of clonal lines with all, some, or no CFTR alleles mutated. Caco-2 cells, derived from a colonic carcinoma, are aneuploid as are many such cancer-derived lines [
      • Melcher R.
      • Koehler S.
      • Steinlein C.
      • Schmid M.
      • Mueller C.R.
      • Luehrs H.
      • et al.
      Spectral karyotype analysis of colon cancer cell lines of the tumor suppressor and mutator pathway.
      ]. CFTR genotyping by next generation sequencing showed that CFTR copy number ranged from 2 to 4, indicative of the genomic heterogeneity of this line. Next-generation sequencing was performed over multiple passages to evaluate stability and consistency of the genotype and allele ratios. The genotypes of these clonal lines are given in Table S3 and S4, supplementary data.
      The clonal lines were assessed for several features, including CFTR mRNA, CFTR protein, transepithelial resistance and forskolin-induced short-circuit currents. As Fig. 1 shows, all lines maintained CFTR mRNA expression (Fig. 1A), measured after multiple passages and found to be reproducible. Lines without wild type CFTR alleles (CFTR) had no detectable CFTR protein (Fig. 1B) or forskolin-stimulated function (Fig. 1C), while both were readily detected in lines with wild type CFTR alleles (CFTR+).
      Fig. 1
      Fig. 1Characterization of clonal lines. Clonal lines were tested after multiple passages, to evaluate stability and consistency of phenotypes. “W” denotes CFTR+ clonal lines, “C” denotes CFTR clonal lines. (A) CFTR mRNA expression detected by RT-PCR was present and stable over multiple passages in both CFTR+ and CFTR lines. (B) CFTR protein was detected by Western Blot analysis in all CFTR+ lines and was not detected in any of the CFTR lines. Only clones that were used for RNA profiling are shown here. (C) CFTR+ lines showed an increase in the transepithelial short circuit current after stimulation with forskolin (forsk). The current was due to CFTR, as it was inhibited by Inh172, a selective CFTR inhibitor. CFTR lines showed no response to forskolin.
      To assess if two cytokines commonly elevated in lungs of individuals with CF were inherently elevated as a consequence of CFTR's absence, IL6 and CXCL8 mRNA, encoding IL6 and IL8, respectively, were measured in the clonal lines. CFTR genotype had no detectable effect on baseline expression of these cytokine genes (Fig. 2). It has also been proposed that CF cells are more sensitive to pro-inflammatory stimuli than non-CF cells [
      • Stecenko A.A.
      • King G.
      • Torii K.
      • Breyer R.M.
      • Dworski R.
      • Blackwell T.S.
      • et al.
      Dysregulated cytokine production in human cystic fibrosis bronchial epithelial cells.
      ]. To explore this concept, the clonal lines were exposed to either E. coli lipopolysaccharide (LPS), IL1β, or TNFα. LPS had no detectable effect on IL6 or CXCL8 mRNA expression and was not pursued further. Both IL1β and TNFα induced expression markedly. When exposed to TNFα the CFTR lines appeared more sensitive than CFTR+ lines, producing greater CXCL8 mRNA amounts at each dose, reaching statistical significance at a dose of 100 ng/ml (Fig. 2). IL6 mRNA expression between the CFTR+ and CFTR clonal lines was not different (p > .05).
      Fig. 2
      Fig. 2Cytokine gene expression. Expression of IL6 and CXCL8 mRNA levels were assessed by RT-PCR at baseline and after stimulation with TNFα. Solid circles represent individual CFTR+ clonal lines while open circles represent individual CFTR clonal lines. Each circle represents the average of 3 stimulation assays performed on each clonal line over multiple passages. The median is denoted by the horizontal bar. (A) No statistically significant difference was seen in IL6 mRNA expression between CFTR+ and CFTR lines at baseline or after stimulation with TNFα. (B) No statistically significant difference was seen in CXCL8 mRNA expression between CFTR+ and CFTR clonal lines at baseline; however, CFTR lines appeared more sensitive than CFTR+ lines, producing statistically greater CXCL8 mRNA amounts after stimulation with 100 ng/ml TNFα. Statistical analysis was performed by Mann-Whitney U test.
      The stimulation-dependent, differential expression of CXCL8, but not IL6, suggested epigenetic remodeling as a consequence of CFTR's absence. We previously reported differential acetylation of the CXCL8 promoter in CF model cell lines [
      • Bartling T.R.
      • Drumm M.L.
      Oxidative stress causes IL8 promoter hyperacetylation in cystic fibrosis airway cell models.
      ] and reasoned that such a phenomenon is not likely restricted to a single gene. Transcriptome profiling (RNA-seq) was carried out to compare mRNA from a panel of clonal lines, 6 CFTR+ and 6 CFTR, to address inter-line variation that might exist. RNA profiles pre- and post-TNFα stimulation were compared and relatedness of lines and genotypes were assessed by hierarchical clustering (Fig. 3). Each line was more closely related to itself pre- and post-stimulation than to any other line under the same conditions (Fig. 3A). This likely reflects the heterogeneity of cells within the parent cell line and each subclone's unique genomic complement, underscoring the importance of comparing multiple, independent lines. When grouped by CFTR genotype (Fig. 3B), the CFTR+ lines pre- and post-stimulation were more similar to each other than the CFTR lines, indicating CFTR's absence does influence gene expression of multiple genes.
      Fig. 3
      Fig. 3Gene expression comparisons. RNA-seq was performed on 6 CFTR+ and 6 CFTR lines before and after stimulation with 100 ng/ml TNFα. (A) Hierarchical clustering showed each line was more closely related to itself before and after stimulation than to any other line under the same conditions. Samples did not segregate based on genotype or stimulation status. (B) Hierarchial clustering based on genotype and stimulation status showed CFTR+ lines before and after stimulation were more similar to each other than CFTR lines. CFTR lines before stimulation were more similar to both the unstimulated and stimulated CFTR+ lines than they were to the CFTR stimulated lines. (C) Heat map displaying those genes showing significant, differential expression by genotype or by stimulation. Those in red indicate increased expression and those in green indicate decreased expression. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
      To better understand the expression patterns contributing to the clustering, expression levels of individual genes showing significantly different expression between any two groups were plotted as a heat map. As (Fig. 3C) shows, there are groups of genes that differentiate CFTR+ from CFTR, groups of genes that discriminate between stimulated and non-stimulated and groups of genes that separate both genotype and stimulation status.
      To better understand the mRNA expression patterns and develop hypotheses regarding transcriptional regulatory pathways that might explain the gene expression clustering, we carried out gene set enrichment analyses. These types of analyses provide clues regarding co-regulation and counter-regulation of genes by comparing with expression patterns from other contexts or scenarios [
      • Liberzon A.
      • Birger C.
      • Thorvaldsdottir H.
      • Ghandi M.
      • Mesirov J.P.
      • Tamayo P.
      The molecular signatures database (MSigDB) hallmark gene set collection.
      ,
      • Subramanian A.
      • Tamayo P.
      • Mootha V.K.
      • Mukherjee S.
      • Ebert B.L.
      • Gillette M.A.
      • et al.
      Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles.
      ].
      We first compared each clonal line pre- and post-stimulation by TNFα. As Table 1 shows, TNFα SIGNALING VIA NFκΒ was the most significantly enriched gene set regardless of CFTR genotype, and serves as a control for the process. Overall, 20 gene sets were significantly enriched from stimulation in both CFTR+ and CFTR cells, 6 of which are clearly involved in inflammation and therefore could be expected. Ten other sets were enriched in CFTR cells, but not CFTR+, and two sets enriched in CFTR+ but not CFTR.
      Table 1Gene sets responding to TNFα stimulation.
      Enriched

      by TNFα in

      CFTR+ or

      CFTR
      Gene set# genes in setCF

      p-val
      WT

      p-val
      CF FDR

      q-val
      WT FDR

      q-val
      BothTNFα SIGNALING VIA NFκB195<0.001<0.001<0.001<0.001
      BothAPOPTOSIS155<0.001<0.001<0.001<0.001
      BothINFLAMMATORY RESPONSE195<0.001<0.001<0.001<0.001
      BothUV RESPONSE UP153<0.001<0.001<0.001<0.001
      BothINTERFERON GAMMA RESPONSE195<0.001<0.001<0.0010.004
      BothIL2 STAT5 SIGNALING194<0.001<0.001<0.0010.002
      BothKRAS SIGNALING UP191<0.001<0.001<0.001<0.001
      BothEPITHELIAL MESENCHYMAL TRANSITION195<0.001<0.001<0.0010.008
      BothP53 PATHWAY195<0.001<0.001<0.0010.004
      BothHYPOXIA193<0.001<0.001<0.0010.004
      BothANDROGEN RESPONSE99<0.001<0.001<0.0010.005
      BothINTERFERON ALPHA RESPONSE93<0.001<0.001<0.0010.001
      BothIL6 JAK STAT3 SIGNALING87<0.001<0.001<0.001<0.001
      BothMYC TARGETS V1190<0.001<0.001<0.001<0.001
      BothCHOLESTEROL HOMEOSTASIS71<0.001<0.001<0.0010.002
      BothMTORC1 SIGNALING195<0.001<0.001<0.0010.007
      BothCOMPLEMENT190<0.001<0.0010.0010.008
      BothCOAGULATION134<0.001<0.0010.0070.012
      BothUNFOLDED PROTEIN RESPONSE1100.002<0.0010.0130.008
      BothOXIDATIVE PHOSPHORYLATION1930.007<0.0010.0440.008
      CFTRALLOGRAFT REJECTION197<0.001NS<0.002NS
      CFTRTGF BETA SIGNALING52<0.001NS<0.002NS
      CFTRMITOTIC SPINDLE194<0.001NS0.002NS
      CFTRAPICAL JUNCTION195<0.001NS0.004NS
      CFTRPROTEIN SECRETION930.002NS0.004NS
      CFTRESTROGEN RESPONSE EARLY197<0.001NS0.01NS
      CFTRUV RESPONSE DN1400.002NS0.012NS
      CFTRG2M CHECKPOINT1940.003NS0.018NS
      CFTRADIPOGENESIS1930.007NS0.018NS
      CFTRESTROGEN RESPONSE LATE1970.007NS0.029NS
      CFTR+MYC TARGETS V257NS0.009NS0.01
      CFTR+E2F TARGETS190NS<0.001NS0.01
      NS = not significant.
      Cells with functional CFTR (CFTR+) and with CFTR inactivated (CFTR) were examined for gene expression responses to TNFα exposure, utilizing gene sets as indicators of transcriptional coordination. The number of genes in each set are listed as well as the nominal p values and false discovery rate (FDR) adjusted for the number of genes examined and number of comparisons made.
      To investigate effects of CFTR function, CFTR+ lines were compared to CFTR lines pre- and post-stimulation (Table 2). Of the 6 gene sets enriched in CFTR cells pre-stimulation, 5 were still enriched after stimulation, indicating that the genes involved are sensitive to CFTR genotype. Interestingly, TNFα SIGNALING VIA NFκΒ was enriched in stimulated CFTR lines relative to stimulated CFTR+, even though this gene set was enriched in all cells and suggesting that it is enhanced by absence of CFTR. Three gene sets were enriched in unstimulated CFTR+ cells relative to CFTR cells, but no sets were significantly enriched in stimulated CFTR+ cells relative to stimulated CFTR cells.
      Table 2Gene sets differentially enriched when stratified by CFTR genotype.
      GenotypeGene set# genes in setUnstim

      p-val
      Unstim FDR q-valStim

      p-val
      Stim

      FDR q-val
      CFTRTNFα SIGNALING VIA NFκB195NSNS<0.001<0.001
      CFTRE2F TARGETS190<0.001<0.001<0.001<0.001
      CFTRMYC TARGETS V1190<0.001<0.001<0.001<0.001
      CFTRG2M CHECKPOINT194<0.001<0.001<0.001<0.001
      CFTRMYC TARGETS V257<0.0010.002NSNS
      CFTRMTORC1 SIGNALING1950.0050.03<0.0010.004
      CFTRESTROGEN RESPONSE LATE1970.0020.0640.0050.056
      CFTR+IL6 JAK STAT3 SIGNALING870.0040.118NSNS
      CFTR+INTERFERON GAMMA RESPONSE1950.0020.076NSNS
      CFTR+EPITHELIAL MESENCHYMAL TRANSITION1950.0040.1NSNS
      NS = not significant.
      Gene set comparisons were carried out to assess the effect of CFTR's presence on baseline (unstimulated) mRNA expression of genes in a set and expression in response to TNFα (stimulated). Gene sets significantly different between CFTR+ and CFTR cells are listed here. The number of genes in each set is listed as well as the nominal p values and false discovery rate (FDR) adjusted for the number of genes examined and number of comparisons made.

      4. Discussion

      Cystic fibrosis is a multisystem disease caused by absence of CFTR function in many different cell types and organs. Recognizing that no single model is likely to be representative of all cell types, we compared the effects of removing CFTR function from an epithelioid cell line on inflammatory gene regulation and the broader cellular transcriptome. Multiple clonal lines were examined to evaluate inter-sample variation and establish whether or not CFTR function influences cytokine mRNA regulation.
      Cell attributes were not detectably affected by the mutagenesis and cloning processes, as all subclonal lines maintained CFTR mRNA expression, electrophysiologic properties and response to inflammatory stimuli. Forskolin-stimulated, short-circuit current was the only measured phenotype that changed, and absent only from those lines lacking CFTR. No detectable difference was found between cells with and without CFTR in baseline expression of the cytokine genes tested, while stimulation by TNFα, elicited differential expression. CFTR cells had, on average, higher CXCL8 mRNA and much greater variation in response to TNFα stimulation than CFTR+ cells. These result bear similarity to a report showing increased CXCL8 (IL8) expression in response to siRNA interference of CFTR in Caco-2 cells. However, that report found differential expression even without exogenous stimulation [
      • Crites K.S.
      • Morin G.
      • Orlando V.
      • Patey N.
      • Cantin C.
      • Martel J.
      • et al.
      CFTR knockdown induces proinflammatory changes in intestinal epithelial cells.
      ], a significant difference from the work shown here. Whether the siRNA treatment itself acted as an inflammatory stimulus is unclear, as those were acute experiments, while our cultures were carried without exposure to any exogenous stimuli for baseline expression measurements.
      The differential expression of CXCL8 led us to examine the broader transcriptome of these cells. This analysis identified genes that appear to be differentially expressed simply from CFTR's absence and another set of genes that, like CXCL8, are only apparent as differentially expressed when stimulated. Heatmaps derived from the RNA-seq data show that the patterns of expression are quite different between groups when all expressed genes are examined (data not shown). Restricting the analysis to genes showing significant, differential expression in any of the comparisons, reveals more apparent patterns of genes that may be inherently different and those that that reveal themselves only after stimulation (Fig. 3C). We propose this differential expression may represent a mechanism in which the genome is remodeled in response to CFTR's absence, altering chromatin in a way that alters active expression or puts genes in a poised position to undergo transcription.
      The data presented here provide a model to help explain CF-associated inflammation and show that the presence or loss of CFTR impacts cellular gene expression much more broadly than just inflammation. The data suggest that CF cells are not overtly different in cytokine expression until provoked, an example of a “gene by environment” interaction. This model also implies that there are epigenetic or other transcriptional regulatory processes affected by CFTR's absence. Caradonna et al. [
      • Caradonna F.
      • Cruciata I.
      • Schifano I.
      • La Rosa C.
      • Naselli F.
      • Chiarelli R.
      • et al.
      Methylation of cytokines gene promoters in IL-1beta-treated human intestinal epithelial cells.
      ] showed that DNA methylation of CXCL8 and IL6 promoters in Caco-2 cells decreases and gene expression increases upon stimulation by IL1β, but IL10 is unaffected. We reported previously that CF primary nasal epithelial cells and several CF cell models exhibit reduced promoter acetylation of the CXCL8 gene. Clearly additional studies will be needed to more completely address epigenetic effects.
      Based on data presented here, we propose a threshold model to help explain the chronic inflammation in CF in which absence of CFTR confers changes on the epigenome that influence transcriptional regulation of genes such as CXCL8. For these genes, such chromatin modifications make these genes more sensitive to stimulation. Ex vivo and without stimulation, CF cells do not behave very differently from non-CF cells. Once stimulation is initiated in vivo, however, pro-inflammatory immune cells would be recruited and contribute to the cytokine/inflammatory mediator environment. In a non-CF context, levels of these mediators would diminish as inflammation is resolved, along with neutralization of the inciting stimulus. However, because of the increased sensitivity of the CF cells, amounts of cytokines that would cease to stimulate WT cells and allow resolution of inflammation would still be able to stimulate CF cells and chronic inflammation would result. Future studies, including co-culture and/or in vivo experiments are needed to test this hypothesis.

      Conflict of interest statement

      MD is a consultant for the Cystic Fibrosis Foundation. There are no other potential conflicts.

      Acknowledgments

      This work was supported by NIH grants P01HL128192, T32HL125245 and Cystic Fibrosis Foundation grants DRUMM15R0 and ROESCH15D0.

      Appendix A. Supplementary data

      References

        • Bahar B.
        • O'Doherty J.V.
        • Maher S.
        • McMorrow J.
        • Sweeney T.
        Chitooligosaccharide elicits acute inflammatory cytokine response through AP-1 pathway in human intestinal epithelial-like (Caco-2) cells.
        Mol Immunol. 2012; 51: 283-291
        • Banville N.
        • Broderick A.
        • Fitzgerald R.
        • Drumm B.
        • McDermott M.
        Giant inflammatory polyposis coli as a manifestation of Crohn's disease in patients with coexistent cystic fibrosis.
        Pediatr Dev Pathol. 2006; 9: 25-30
        • Bartling T.R.
        • Drumm M.L.
        Loss of CFTR results in reduction of histone deacetylase 2 in airway epithelial cells.
        Am J Physiol Lung Cell Mol Physiol. 2009; 297: L35-L43
        • Bartling T.R.
        • Drumm M.L.
        Oxidative stress causes IL8 promoter hyperacetylation in cystic fibrosis airway cell models.
        Am J Respir Cell Mol Biol. 2009; 40: 58-65
        • Bresso F.
        • Askling J.
        • Astegiano M.
        • Demarchi B.
        • Sapone N.
        • Rizzetto M.
        • et al.
        Potential role for the common cystic fibrosis DeltaF508 mutation in Crohn's disease.
        Inflamm Bowel Dis. 2007; 13: 531-536
        • Caradonna F.
        • Cruciata I.
        • Schifano I.
        • La Rosa C.
        • Naselli F.
        • Chiarelli R.
        • et al.
        Methylation of cytokines gene promoters in IL-1beta-treated human intestinal epithelial cells.
        Inflam Res. 2017; 67: 327-337
        • Carrabino S.
        • Carpani D.
        • Livraghi A.
        • Di Cicco M.
        • Costantini D.
        • Copreni E.
        • et al.
        Dysregulated interleukin-8 secretion and NF-kappaB activity in human cystic fibrosis nasal epithelial cells.
        J Cyst Fibr. 2006; 5: 113-119
        • Chang E.B.
        • Bookstein C.
        • Vaandrager A.
        • DeJonge H.R.
        • Buse J.
        • Musch M.W.
        Cystic fibrosis transmembrane regulator mRNA expression relative to ion-nutrient transport in spontaneously differentiating human intestinal CaCo-2 epithelial cells.
        J Lab Clin Med. 1991; 118: 377-381
        • Crites K.S.
        • Morin G.
        • Orlando V.
        • Patey N.
        • Cantin C.
        • Martel J.
        • et al.
        CFTR knockdown induces proinflammatory changes in intestinal epithelial cells.
        J Inflamm. 2015; 12
        • De Lisle R.C.
        • Borowitz D.
        The cystic fibrosis intestine.
        Cold Spring Harb Perspect Med. 2013; 3a009753
        • Dhaliwal J.
        • Leach S.
        • Katz T.
        • Nahidi L.
        • Pang T.
        • Lee J.M.
        • et al.
        Intestinal inflammation and impact on growth in children with cystic fibrosis.
        J Pediatr Gastroenterol Nutr. 2015; 60: 521-526
        • Flotte T.R.
        • Solow R.
        • Owens R.A.
        • Afione S.
        • Zeitlin P.L.
        • Carter B.J.
        Gene expression from adeno-associated virus vectors in airway epithelial cells.
        Am J Respir Cell Mol Biol. 1992; 7: 349-356
        • Fogh J.
        • Wright W.C.
        • Loveless J.D.
        Absence of HeLa cell contamination in 169 cell lines derived from human tumors.
        J Natl Cancer Inst. 1977; 58: 209-214
        • Heeckeren A.
        • Walenga R.
        • Konstan M.W.
        • Bonfield T.
        • Davis P.B.
        • Ferkol T.
        Excessive inflammatory response of cystic fibrosis mice to bronchopulmonary infection with Pseudomonas aeruginosa.
        J Clin Invest. 1997; 100: 2810-2815
        • Hodges C.A.
        • Grady B.R.
        • Mishra K.
        • Cotton C.U.
        • Drumm M.L.
        Cystic fibrosis growth retardation is not correlated with loss of Cftr in the intestinal epithelium.
        Am J Physiol Gastrointest Liver Physiol. 2011; 301: G528-G536
        • Hou Y.
        • Guan X.
        • Yang Z.
        • Li C.
        Emerging role of cystic fibrosis transmembrane conductance regulator - an epithelial chloride channel in gastrointestinal cancers.
        World J Gastroint Oncol. 2016; 8: 282-288
        • Kube D.
        • Sontich U.
        • Fletcher D.
        • Davis P.B.
        Proinflammatory cytokine responses to P. aeruginosa infection in human airway epithelial cell lines.
        Am J Physiol Lung Cell Mol Physiol. 2001; 280: L493-L502
        • Liberzon A.
        • Birger C.
        • Thorvaldsdottir H.
        • Ghandi M.
        • Mesirov J.P.
        • Tamayo P.
        The molecular signatures database (MSigDB) hallmark gene set collection.
        Cell Syst. 2015; 1: 417-425
        • Mali P.
        • Yang L.
        • Esvelt K.M.
        • Aach J.
        • Guell M.
        • DiCarlo J.E.
        • et al.
        RNA-guided human genome engineering via Cas9.
        Science. 2013; 339: 823-826
        • Melcher R.
        • Koehler S.
        • Steinlein C.
        • Schmid M.
        • Mueller C.R.
        • Luehrs H.
        • et al.
        Spectral karyotype analysis of colon cancer cell lines of the tumor suppressor and mutator pathway.
        Cytogenet Genome Res. 2002; 98: 22-28
        • Nichols D.P.
        • Chmiel J.F.
        Inflammation and its genesis in cystic fibrosis.
        Pediatr Pulmonol. 2015; 50: S39-S56
        • Peuhkuri K.
        • Vapaatalo H.
        • Korpela R.
        Even low-grade inflammation impacts on small intestinal function.
        World J Gastroenterol. 2010; 16: 1057-1062
        • Shin H.S.
        • Zhao Z.
        • Satsu H.
        • Totsuka M.
        • Shimizu M.
        Synergistic effect of tumor necrosis factor-alpha and hydrogen peroxide on the induction of IL-8 production in human intestinal Caco-2 cells.
        Inflammation. 2011; 34: 440-447
        • Stecenko A.A.
        • King G.
        • Torii K.
        • Breyer R.M.
        • Dworski R.
        • Blackwell T.S.
        • et al.
        Dysregulated cytokine production in human cystic fibrosis bronchial epithelial cells.
        Inflammation. 2001; 25: 145-155
        • Subramanian A.
        • Tamayo P.
        • Mootha V.K.
        • Mukherjee S.
        • Ebert B.L.
        • Gillette M.A.
        • et al.
        Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles.
        Proc Natl Acad Sci U S A. 2005; 102: 15545-15550
        • Zhang L.
        • Shen J.
        • Cheng J.
        • Fan X.
        MicroRNA-21 regulates intestinal epithelial tight junction permeability.
        Cell Biochem Funct. 2015; 33: 235-240