Introduction
Cystic fibrosis (CF) is a life-limiting autosomal recessive genetic disease due to mutations in the
CFTR gene [
1‐
3]. In the lung, the common CF mutation [
F508del] [
4,
5], and the less common [
G551D] mutation [
6‐
9], are both associated with a hyper-proinflammatory phenotype that is due to constitutive activation of the TNFα/NFκB signaling pathway [
10‐
14]. For both mutations, failure to suppress inflammation leads to further reactive inflammation, exacerbation of constitutive secretion of mucin, local airway obstruction and hypoxia, infection, and, finally, complete loss of lung function. Recently, to treat the pulmonary manifestations of CF, a “potentiator” drug, VX-770, has been licensed that induces chloride channel activity in the [
G551D] gating mutant [
15]. A modestly effective “corrector” drug, VX-809, has been licensed to treat the common [
F508del] mutation in combination with VX- 770 [
16]. More recently, a licensed combination of the corrector drug VX-661 with VX-770 has been shown to be safer and possibly more effective in CF patients homozygous [
17] and heterozygous [
18] for the [
F508del] CFTR mutation. However, despite substantial and successful responses in CF patients, there remain other aspects of the disease for which the corrector and potentiator drugs are less effective [
19‐
22]. The need to fill this gap is illustrated by the Cystic Fibrosis Foundation (CFF)‘s recent offer of support for “discovery of ways to dampen the exaggerated immune response that causes chronic inflammation without affecting the body’s natural defenses against infection” [
23].
It is possible that efficacious, adjuvant anti-inflammatory drugs could be combined therapeutically with currently licensed drugs to ameliorate even severe CF disease phenotypes [
24]. Unfortunately, several CF anti-inflammatory adjuvant drugs, including ibuprofen and prednisone, carry significant risk to patients [
25‐
29]. However, there are also several new candidate anti-inflammatory drugs in early stages of development [
30‐
32]. Among these new candidates is the cardiac glycoside drug digitoxin. It was first tested in this context in CF lung epithelial cells
, where it was found to inhibit TNFα/NFκB signaling and downstream IL-8 secretion [
33,
34]. This discovery led to testing digitoxin in CF patients as an anti-inflammatory agent in a Phase 2, dose escalation, placebo-controlled clinical trial (NCT00782288,
clinicaltrials.gov). It was found that mono-therapy digitoxin not only suppressed respiratory adverse events by 69% (
p = 0.0365), but also blocked TNFα/NFκB signaling and
IL-8 gene expression in nasal epithelial cells biopsied from drug-treated CF patients [
35]. The fact that digitoxin could act as an anti-inflammatory agent in vivo, motivated us to consider the use of digitoxin as a CF treatment, either alone or as an adjuvant to licensed VX drugs. However, we reasoned that it was critical to carefully test combinations of digitoxin and VX drugs therapy in vitro, in anticipation of adjuvant use in patients, because drug-drug interferences have been described for VX-809 and VX-770 [
36]
, as well as between VX-770 and CFTR itself [
37].
Here, we show that digitoxin is able to suppress proinflammatory TNFα/NFκB signaling and downstream secretion of IL-8, IL-6 and GM-CSF, when tested alone or in the presence of individual or combinations of licensed corrector and potentiator drugs. By contrast, the VX-drugs are relatively inactive in terms of inhibiting chemokine and cytokine secretion. Expression changes measured by RNA-seq are consistent with this conclusion, and also show that digitoxin, alone or in combination with VX-drugs, causes changes that significantly emulate some of the effects of AAV-[wildtype]CFTR gene therapy in CF lung epithelial cells. Finally, re-examination of nasal epithelial cell mRNA expression data from CF patients treated with digitoxin suggests that there are significant clinical parallels with the in vitro data collected in this study. Taken together, these data suggest adjuvant gene therapy-emulating activities of digitoxin may contribute to enhancing the efficacy of currently licensed correctors and potentiators in CF patients.
Materials and methods
General information is given below. Greater detail is included in Additional file
1.
Cells and drugs
CF lung epithelial IB3–1 cells ([
F508del/W1282X]), and daughter IB3–1/S9 cells stably transfected with AAV-[
wildtype] CFTR [
38], were grown in serum-free LHC-8 medium (Biofluids, Bethesda, MD), formulated without gentamycin,
exactly as previously described [
33]. Digitoxin was obtained from Sigma (Sigma-Aldrich Corp., St. Louis, MO), and dissolved in 100% ethanol prior to dilution in PBS. The final ethanol concentration was 0.001-0.0025%. VX-770, VX-809 and VX-661 were obtained from
Selleckchem.com (Houston, TX), and dissolved in 100% DMSO prior to dilution in PBS.
Fisher rat thyroid (FRT) cells stably expressing human [G551D]CFTR were generously provided by Dr. Eric Sorscher (Emory University, Atlanta, GA) and grown in F-12 Modified Coon’s medium supplemented with 10% FBS, 200 μg/ml hygromycin, 0.23% sodium bicarbonate, 100 units/ml penicillin and 100 μg/ml streptomycin. FRT cells were seeded onto Costar 0.4-mm Snapwell inserts, cultured for 5 days as electrically resistive monolayers, and then treated for 24 h at 37 °C with digitoxin (25 nM) ± VX-770 (3 μM).
Reporter gene assays
IB3–1 cells were seeded in 6 well plates overnight, then subsequently cotransfected overnight (16 h) with NFκB-luc and lacZ plasmids using Lipofectamine 3000 transfection reagent (Invitrogen). The cells were treated with 20 ng/ml TNFα and/or 25 nM digitoxin or/and VX drugs overnight. Cells were harvested and lysed with 1x passive lysis buffer. Luciferase assays were performed with the Promega Luciferase Assay System. The luciferase values were normalized to β-galactosidase activity. Differential statistics were calculated between groups using two-tailed t-tests after normalizing individual experimental values to TNFα-treated controls (n = 4–7 biological replicates for all groups considered).
Measurement of cytokines and chemokines
Cytokines and chemokines were assayed on the SECTOR® S 600 instrument (Meso Scale Discovery, Gaithersburg, MD, USA). The Human Pro-Inflammatory 9 Plex Tissue Culture kit (MesoScale catalog #K15007B-2) was used for the measurement of IL-2, IL-8, IL-12p70, IL-1β, GM-CSF, IFN-γ, IL-6, IL-10, and TNF-α [
39]. Differential statistics were calculated between groups using two-tailed t-tests after normalzing individual experimental values to TNFα-treated controls (
n = 4–7 biological replicates for all groups considered).
Western blot analysis
IB3–1 cells were lysed in M2 buffer (20 mM pH 7.0 Tris, 0.5% NP-40, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 20 mM β-glycerol phosphate, 10 mM, 4-nitrophenyl phosphate disodium salt, 1 mM sodium vanadate, 1 mg/ml of leupeptin). 20 μg of the cell lysate protein from each sample were fractionated by SDS-PAGE and immunoblotted. Blots were visualized with chemiluminescent substrate (Millipore, Rockville, MD) and band densities were measured using the NIH ImageJ program [
40]. Anti-IκBα antibody was from Cell Signaling Technologies (Boston, MA). The antibody for β-actin was from Sigma (Sigma-Aldrich, St. Louis, MO). Differential statistics were calculated between groups using two-tailed t-tests after normalizing individual experimental values to TNFα-treated controls (
n = 3 biological replicates). Westerns were also performed on the WES instrument (Protein Simple, San Jose, CA) according to manufacturer’s instructions.
RNA-seq
Cultured cells were harvested in Qiazol (Qiagen, Germantown, MD) and homogenized using a QIAshredder (Qiagen) before isolation of total RNA using the RNeasy Mini Kit (Qiagen). Total RNA was quantified via a fluorescence dye-based methodology using the Quant-IT RiboGreen RNA Reagent (Thermo Scientific, Waltham, MA) and assay measurement with a Spectramax Gemini XPS plate reader (Molecular Devices, San Jose, CA).
Details for alignment and sample quality assessment, gene quantification and differential expression analysis, gene ontology and clustering analysis, and human exon array expression analysis are given in Additional file
1: Materials and Methods.
Discussion
The need for anti-inflammatory drugs as adjuvants to standard CF therapy has been prominent in the CF literature for decades, and remains a challenge even in this time of licensed corrector and potentiator drugs [
17,
25‐
28]. Here, we show that treatments with clinically relevant concentrations of individual or combination corrector or potentiator drugs are relatively ineffective at suppressing secretion of IL-8, IL-6 and GM-CSF proteins from cultured CF lung epithelial cells. By contrast, additions of digitoxin, alone or in combination with corrector/potentiator drugs, potently suppresses not only IκBα and NFκB activation, but also baseline and TNFα-activated secretion of these three proinflammatory mediators. Importantly, transcriptional studies via RNA-seq strongly support these proteomic results. In addition, we find that the transcriptomes of gene therapy-treated and digitoxin-treated CF cells substantially overlap. Among genes that are significantly modified in parallel by these two treatments, we identified an 84 gene signature for
reduced inflammation, a 49 gene signature for
reduced fibrosis, and an 82 gene signature for
elevated epithelial differentiation. Consistently, digitoxin also
suppressed expression of TGFBR2 protein, the receptor for TGFβ, the master regulator of fibrosis [
44]. Digitoxin also
increased expression of KRT8 protein, a biomarker for epithelial differentiation [
43]. Thus, while digitoxin does not affect trafficking of mutant CFTR, and only modestly opens the [G551D]CFTR channel to potentiate chloride transport across the membrane, this drug is capable of emulating other physiologically important corrections expected from gene therapy. Specifically, these corrections include reduced inflammation and fibrosis and increased epithelial differentiation. By contrast, our transcriptional data indicate that the licensed VX drugs minimally impact these shared gene therapy/digitoxin signatures. Finally, using the biological features uncovered in the in vitro data, we found a consistent relationship with clinically derived in vivo mRNA expression data from nasal epithelial biopsies of digitoxin-treated CF patients. Importantly, this insight was not apparent when the results of our clinical trial were first published [
35]. We therefore suggest that using digitoxin to treat CF, either alone or as an adjuvant drug, is worthy of further consideration.
Insight into the mechanism for these shared functional gene ontology themes may be gained by consideration of early studies showing that digitoxin blocks the interaction between the TNFα/TNFR1 complex and TRADD [
33,
34]. TRADD is the first intracellular adaptor for TNFα signaling to activate NFκB, and therefore downstream
IL-8 expression. We have previously shown that high levels of
TRADD mRNA characterize epithelial brush biopsies of CF patients with the most severe disease, as defined by the rate of FEV1 decline [
45]. More recently, we reported that [wildtype]CFTR also directly interacts with TRADD and tonically directs TRADD for proteosomal destruction [
14]. By contrast, we showed in that report that the mutants [
F508del] and [G551D]CFTR proteins fail to bind TRADD, and thus fail to inhibit downstream activation of NFκB. These two different experiments thus indicate that TRADD is a common inflammation-controlling target for both digitoxin and [wildtype]CFTR. It is thus possible that coincident interactions with TRADD may be the basis for the large scale and highly functional transcriptional overlap between digitoxin and gene therapy. Digitoxin thus differs from other candidate NFκB inhibitors with different mechanisms of action in this way, whose renal and CNS side effects are said to have rendered them generally useless for clinical applications [
46]. Finally, it is possible that digitoxin may also inhibit pro-fibrotic signaling through its action on TRADD. For example, increased fibrosis is also known to be a direct consequence of TNFα-driven activation of TGFβ1 signaling [
47‐
49]. Thus, by blocking expression of both
TGFBR2 mRNA and TGFBR2 protein, the receptor for TGFβ, both digitoxin and gene therapy suppress profibrotic TGFβ signaling. Hence, these data suggest that digitoxin appears to uniquely address the root cause of not only inflammation but also fibrosis in CF.
Elevation of mRNAs associated with epithelial differentiation may also be related to TRADD function. It is possible that the effect of digitoxin and gene therapy on CF cells is to start repair of developmental defects in these cells, and thus reset the differentiated state of the cells. In fact, many of the mRNAs in this category are directly associated with regulating differentiation of stem cells into the more differentiated pseudostratified epithelium lining the airway. Here, the action of digitoxin and gene therapy on TRADD function may provide a clue. In the lung, Activin A, a member of the TGFβ superfamily, initiates the differentiation process from iPSCs (induced pluripotent stem cells) to form definitive endoderm [
50]. To convert definitive endoderm to anterior foregut endoderm, these cells must be incubated in NOGGIN to
reduce levels of TGFβ, and inhibit BMP signaling. In as much as there is significant reduction in
TGFBR2 due to digitoxin blocking TRADD-dependent activation of NFκB, TGFβ signaling would be suppressed, independent of TGFβ availability. Both digitoxin and gene therapy also elevate
KRT8, a biomarker for conversion of basal stem cells into differentiated epithelial cells, as well as several NOTCH and WNT isoforms. Relevantly, following differentiation of stem cells into basal stem cells in the lung, NOTCH signaling takes over as a driver for the more differentiated pseudostratified epithelial cells lining the lung. We therefore suggest that this uniquely elevated functional gene ontology theme may reflect the action of digitoxin or gene therapy to reset and repair CF related defects in the differentiated state of CF cells. This polypharmacy property, on the one hand anti-inflammatory and anti-fibrotic, and on the other pro-differentiation, may uniquely distinguish digitoxin from other candidate drugs being developed for CF.
Finally, to our knowledge these are the first data to characterize changes in mRNA expression caused by licensed CF correctors and potentiators, including VX-770 (Ivacaftor®), [
VX-770/VX-809] (Orkambi®), and [
VX-770/VX-661] (Symdeko®). Based on the ELISA and RNA-seq data,
reduction of TNFα/NFκB-driven inflammation is not among the significant functional gene ontology themes for the VX drugs (see Additional file
6). While mono-therapy with VX-809 alone can reduce expression of
IL-8 mRNA, it does not significantly affect IL-8 protein expression. However, mixing VX-809 with VX-770 results in loss of this effect on
IL-8 mRNA. This result is reminiscent of previous reports of interference between these two drugs [
36]. In addition, none of the VX drugs, alone or in combination, affect IL-6 mRNA or protein expression. Furthermore, none of the VX drugs are characterized by the digitoxin-defined gene ontology themes for
reduced fibrosis or elevated epithelial differentiation. Thus, while these VX drugs may sufficiently modify folding of mutant CFTR to enable correction of trafficking or potentiation of chloride conductance, the drug-dependent modifications in folding are clearly not sufficient to correct other functional deficits. Here digitoxin may make a significant contribution as an adjuvant CF drug, since it addresses the root causes of multiple disease-related dysfunctions in CF.
In conclusion, the motivation for having tested the small molecule digitoxin in these ways has rested on our earlier discovery of its ability to suppress IL-8 secretion from CF lung epithelial cells [
33,
34], and more recently on its suppressive effects on
IL-8,
IL-6 and related mRNAs in CF patients [
35]. Based on a new analysis of the clinical trial data we suggest that either independent use of digitoxin, or adjuvant use with the licensed VX drugs, may confer reduced signaling for inflammation and fibrosis and increased signaling for epithelial differentiation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.