Introduction
Idiopathic pulmonary fibrosis (IPF) has an unknown etiology [
1] and is marked by progressive lung fibrosis leading to respiratory failure. The pathogenic mechanisms involved in its initiation and progression are poorly understood [
2] and there are limited therapeutic options with poor efficacy [
3],[
4]. Prognosis is bleak with a median survival of only 3 years, worse than many cancers [
5]. IPF patients present with a mean age of between 60 to 65 years at diagnosis [
4]. In the USA the overall incidence of IPF is 16 per 100,000 person-years [
2] and the incidence is increasing by 11% annually in the UK [
6]. The most favoured hypothesis regarding its development is that on-going multiple, microscopic, isolated episodes of alveoli epithelial injury lead to an abnormal wound healing response involving fibrotic repair mechanisms [
7].
Fibroblasts are mesenchymal cells that serve a critical role in both normal and fibrotic repair processes, which when activated, become differentiated, highly secretory and contractile smooth muscle-like cells termed myofibroblasts [
8]. Expression of alpha smooth muscle actin (αSMA) and αSMA-containing stress fibres is the hallmark of these cells [
9]-[
12]. IPF evolves from dysfunctional interactions between the injured epithelium and fibroblasts which lead to pathologic lesions called fibroblast foci, which are comprised of activated myofibroblasts [
13]. In their activated state, myofibroblasts are the primary cell responsible for the synthesis, secretion and remodelling of the extracellular matrix in IPF [
14]. The human lung myofibroblast (HLMF) is therefore an attractive target for the treatment of IPF.
αSMA is a key protein expressed by HLMFs as compared to quiescent fibroblasts [
15], and contributes to the formation of characteristic HLMF contractile stress fibres [
8],[
16],[
17]. αSMA expression and stress fibre formation in myofibroblasts is regulated in part by the TGFβ1/Smad signalling pathway [
18],[
19]. Smads are intracellular proteins which transduce TGFβ1-dependent signals. Following binding of TGFβ1 to the TGFβRII, Smad2/3 are phosphorylated and form hetero-oligomeric complexes with Smad 4, leading to nuclear translocation and the regulation of gene transcription [
20]. They therefore regulate many biological effects in HLMFs that are under the control of TGFβ1, including collagen secretion, proliferation, differentiation and contraction [
18]-[
21].
Ion channels are attractive therapeutic targets for many chronic diseases including fibrosis. Activated intermediate conductance Ca
2+-activated K
+ channels promote several pro-fibrotic processes in HLMFs such as basic fibroblast growth factor (bFGF)-dependent proliferation, and TGFβ1-dependent wound healing, collagen secretion and contraction [
22]. K
Ca3.1 activity was also shown to contribute to the upregulation of αSMA in response to TGFβ1 through the enhancement of Smad phosphorylation [
23], and contributed to diabetic [
24] and surgically-induced kidney fibrosis in rodents [
25]. However, much of the research published to-date has focussed on the activity of myofibroblasts following TGFβ1 stimulation, and there are few studies investigating basal signalling differences between non fibrotic control (NFC) and IPF-derived myofibroblasts. Previously, IPF-derived HLMFs demonstrated significantly higher constitutive αSMA [
26] and functional K
Ca3.1 channel expression [
22] compared to NFC-derived cells. However, whether these basal increases in αSMA in IPF-derived HLMFs are due to altered constitutive Smad pathway signalling or K
Ca3.1 activity is not known.
We therefore hypothesized that Smad2/3 signalling is increased at baseline in IPF-derived HLMFs compared to NFC cells, and if true, that this would be susceptible to KCa3.1 channel inhibition. We therefore investigated constitutive Smad2/3 nuclear localisation and αSMA expression in NFC- and IPF-derived HLMFs, and the role of KCa3.1 ion channels in these processes.
Materials and methods
Human lung myofibroblast isolation and culture
Non-fibrotic control (NFC) HLMFs were derived from healthy areas of lung from patients undergoing lung resection for carcinoma at Glenfield Hospital, Leicester, UK. No morphological evidence of disease was found in the tissue samples used for HLMF isolation. IPF HLMFs were derived from patients undergoing lung biopsy for diagnostic purposes at the University of Pittsburgh Medical Center, USA, and were shown to have UIP on histological examination. Myofibroblasts were grown from explanted lung tissue from both sources under identical conditions, using DMEM supplemented with 10% FBS, antibiotic/antimycotic agents and non-essential amino acids [
27],[
28]. The cells were cultured at 37°C in 5% CO
2/95% air. Cells were studied at passages 4–5 for functional studies. HLMFs were characterised as previously described [
22]. All NFC patients gave informed written consent and the study was approved by the National Research Ethics Service (references 07/MRE08/42 and 10/H0402/12). Written informed consent was also obtained from all IPF subjects, in accordance with the responsible University of Pittsburgh Institutional Review Board.
Human myofibroblast characterisation
Human myofibroblasts were harvested from 80-90% confluent monolayers with 0.1% trypsin/0.1% EDTA. Cells were seeded into 8-well chamber slides, grown to confluence, and to confirm a myofibroblast population the cells were immunostained using the following antibodies: FITC-conjugated mouse monoclonal anti-α-smooth muscle actin (αSMA) (F3777, 10 μg/ml, Sigma-Aldrich, Poole, Dorset, UK); mouse monoclonal anti-fibroblast surface protein(FSP)(F4771, 4 μg/ml, Sigma-Aldrich); mouse monoclonal anti-fibroblast antigen THY-1 (CP28, 3 μg/ml, Calbiochem, San Diego, CA; rabbit polyclonal collagen type 1 (550346, 20 μg/ml, Millipore, Watford, UK), isotype control rabbit IgG (20 μg/ml). To detect the presence of contaminating cells the following antibodies were used; monoclonal mouse CD68 antibody (6.4 μg/ml, Dako) to detect the presence of macrophages, mast cells and monocytes; mouse monoclonal CD3 (4.5 mg/ml Dako) to detect T-cells, and mouse monoclonal CD34 R-PE antibody (0.5 μg/ml, Catlag) was also used to detect progenitor cells. Secondary antibodies labelled with FITC or R-PE (F0313, Dako) were applied and the cells counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich). All isotype controls were negative. Cells were mounted with fluorescent mounting medium and cover-slipped. The results confirmed that the cells isolated were of myofibroblast-rich population of cells (99%), with no contaminating cells identified. Full details, results and accompanying images have been previously published in Roach et al. [
22].
Immunofluorescence
HLMFs were grown on 8-well chamber slides and serum-starved for 24 hours prior to the experiment. The cells were then treated for 24 hours with either 0.1% DMSO, TRAM-34 (20 and 200 nM) or ICA-17043 (10 and 100 nM). Cells were then immunostained as described previously [
22] using FITC-conjugated mouse monoclonal anti-α-SMA (F3777, 10 μg/ml, Sigma-Aldrich, Poole, Dorset, UK) and isotype control FITC-conjugated mouse IgG
2a (X0933, 10 μg/ml, Dako, Ely, UK). K
Ca3.1 expression was examined by immunofluorescence using rabbit polyclonal anti-K
Ca3.1 (AV35098, 5 μg/ml, Sigma) and appropriate isotype control. Secondary antibodies labelled with FITC (F0313, Dako) were applied and the cells counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich). Cells were mounted with fluorescent mounting medium and cover-slipped. Original images were captured on an epifluorescent microscope (Olympus BX50, Olympus UK Ltd, Southend–on-sea); grey scale intensity was examined using Cell F imaging software (Olympus UK Ltd). Matched exposures were used for isotype controls.
Actin stress fibres were calculated using a specialised macro on image J designed by Dr Kees Straatman, University of Leicester. The macro is capable of providing a quantitative, unbiased score of the number of stress fibres per individual cell by determining the fluctuations of grey scale intensity created by the αSMA staining within the stress fibres.
Flow cytometry
Cells were grown on T25 flasks and serum-starved for 24 hours prior to the experiment. The myofibroblasts were incubated for 24 hours in the presence of 0.1% DMSO control, TRAM-34 (200 nM) or ICA-17043 (100 nM). Cells were detached using 0.1% trypsin/0.1% EDTA, washed then fixed and permeabilised in 4% paraformaldehyde plus 0.1% saponin (Sigma) respectively for 20 minutes on ice. Myofibroblasts were labelled with FITC-conjugated mouse monoclonal anti-α-smooth muscle actin (sigma) or isotype control FITC-conjugated mouse IgG2a. Secondary antibodies labelled with FITC (F0313, Dako) were applied. Analysis was performed using single colour flow cytometry on a FACScan (BD, UK).
Smad nuclear localisation
HLMFs were grown on 8-well chamber slides and serum-starved for 24 hours prior to the experiment. The cells were then stimulated with TGFβ1 (10 ng/ml) in the presence of either 0.1% DMSO control, TRAM-34 (200 nM), ICA-17043 (100 nM) or Ca2+ free media. After 1 hour cells were immunostained using rabbit monoclonal anti-Smad2/3 (0.174 μg/ml, Cell Signalling). Secondary antibody labelled with FITC (F0313, Dako) was applied and the cells counterstained with DAPI (Sigma-Aldrich). Cells were mounted with fluorescent mounting medium and cover-slipped. Images were analysed as above. The intensity of nuclear Smad2/3 staining was quantified by measuring the grey scale intensity of DAPI positive nuclei to whole cell staining.
Western blot for Smad proteins
Cells were grown in T75 flasks, serum starved for 24 hours, and then incubated with either 0.1% DMSO control, TRAM-34 (200 nM), ICA-17043 (100 nM) or Ca2+ free media for 1 hour. Cells were detached with 0.1% Trypsin/EDTA and washed. Protein was isolated using the RIPA buffer lysis system (Santa Cruz, Germany) and total protein concentration was determined using the DC Bio-Rad protein Assay (Bio-Rad, UK). 30 μg of protein was resolved using 10% Mini-Protean TGX precast gels (Bio-Rad) and then transferred to an immunobilon-P polyvinylidene difluoride membrane, using Trans-blot Turbo transfer packs (Bio-Rad). Membranes were blocked with 5% milk and incubated with rabbit monoclonal anti-phospho-Smad2/3 (0.231 μg/ml, Cell Signalling, USA), rabbit polyclonal anti-Smad2/3 (0.0087 μg/ml, Cell Signalling), mouse monoclonal anti-TATA binding protein (TBP) (1 μg/ml, Abcam), or mouse monoclonal anti-β-actin antibody (0.2 μg/ml, SantaCruz). Protein bands were identified by horseradish peroxidase-conjugated secondary antibody and enhanced chemiluminescence reagent (Amersham, UK). Immunolabelled proteins were visualized using ImageQuant LAS 4000 (GE Healthcare Life Sciences, UK).
Nuclear fraction
NFC and IPF-derived HLMFs were grown in T75 flasks, serum starved for 24 hours, then detached with 0.1% Trypsin/EDTA and washed. Nuclear and cytoplasmic extracts were isolated using the Nuclear Extract Kit (Abcam, ab113474). Protein concentrations of both nuclear and cytoplasmic extract was then was isolated using the RIPA buffer lysis system determined using the DC Bio-Rad protein Assay. Western blot was then performed as described above.
qRT-PCR
Myofibroblast RNA was isolated using the RNeasy Plus Kit (Qiagen, West Sussex, UK) according to the manufacturer’s instructions. Primers were designed for Smad2 , forward CGTCCATCTTGCCATTCACG and reverse CTCAAGCTCATCTAATCGTCCTG, product size 182 bp from NCB1 Reference sequence NM_005901.5; and Smad3, forward GCGTGCGGCTCTACTACATC and reverse GCACATTCGGGTCAACTGGTA product size 233 bp from reference sequence NM_005902.3 β-actin primers were analysed using gene-specific Quantitect Primer Assay primers (Qiagen, Germany), HS_ACTB_1_SG. All expression data were normalized to β-actin and corrected using the reference dye ROX. Gene expression was quantified by real-time PCR using the Brilliant SYBR Green QRT-PCR 1-Step Master Mix (Strategene, The Netherlands). PCR products were run on a 1.5% agarose gel to confirm the product amplified was the correct size, and each of the products were sequenced to confirm the specificity of the primers.
Statistical analysis
Experiments from an individual donor were performed either in duplicate or triplicate and a mean value was derived for each condition. Cells from 8 IPF and 8 NFC donors were used, and the number of donors used for each experimental condition is stated in the text and/or figure legends. Data distribution across donors was tested for normality using the Kolmogorov-Smirnov test. For parametric data the 1-way ANOVA or repeated measures ANOVA for across-group comparisons was used followed by the appropriate multiple comparison post hoc test; otherwise an unpaired or paired t-test was used. For non-parametric data the Friedman test was used for across group comparisons followed by the appropriate multiple comparison post hoc test, or the Mann Whitney U test was used where there were two unpaired groups. GraphPad Prism for windows (version 6, GraphPad Software, San Diego California USA) was used for these analyses. A value of P < 0.05 was taken to assume statistical significance and data are represented as mean (± SEM) or median (IQR).
Discussion
We have demonstrated that IPF-derived HLMFs constitutively overexpress αSMA and actin stress fibres together with an increase in constitutive Smad2/3 nuclear localisation when compared to NFC cells. The increased constitutive Smad2/3 nuclear localisation was heavily reliant on the presence of extracellular Ca2+, and was inhibited by selective KCa3.1 ion channel blockers. Taken together, these data suggest that increased basal Ca2+- and KCa3.1-dependent processes facilitate “constitutive” Smad2/3 signalling in IPF-derived fibroblasts, and thus promote fibroblast to myofibroblast differentiation. Importantly, inhibiting KCa3.1 channels reversed this process.
Phenotypically both NFC and IPF-derived human lung parenchymal fibroblasts in culture demonstrate the classic fibroblast-like elongated morphology [
21],[
22],[
37]. Previous work has shown that healthy lung parenchymal fibroblasts express high levels of αSMA and display upregulated genes associated with actin binding and cytoskeletal organisation together with up-regulation of activated Smad2 and Smad3 when compared to airway fibroblasts from the same donors [
21]. These cultured lung parenchymal fibroblasts therefore have features of a myofibroblast phenotype, hence we describe them as HLMFs. Of note, IPF-derived HLMFs express higher concentrations of αSMA than cells from healthy lung [
26],[
38], a finding we have confirmed here. In addition to this, we found that IPF-derived cells exhibited increased actin stress fibre formation and cytoskeletal organisation compared to NFC cells. We also found that IPF-derived HLMFs, without TGFβ1 stimulation, have significantly higher constitutive total Smad2/3 protein expression, greater Smad2/3 expression within the nucleus and higher Smad2 and Smad3 mRNA expression when compared to NFC control cells. Smad3 in particular was implicated in driving the myofibroblast phenotype in healthy parenchymal HLMFs [
21], and while we found that both Smad2 and Smad3 mRNA were increased in IPF-derived HLMFs, Smad3 mRNA was significantly higher than Smad2 in both NFC and IPF-derived cells. This supports the work of Zhou et al. [
21], but shows that differential Smad2 versus Smad3 mRNA expression persists in IPF-derived cells. Thus while healthy parenchymal lung fibroblasts exhibit features of a myofibroblast phenotype, this is significantly more pronounced in IPF-derived myofibroblasts.
The mechanism behind this increased constitutive myofibroblast differentiation in IPF-derived cells is unclear and potentially multi-factorial. That such a phenotypic difference persists for several passages in cell culture suggests that genetic or epigenetic factors may contribute [
39]. Whether pre-programmed by genetics, or re-programmed by epigenetics, IPF-derived HLMFs clearly have the potential to perpetuate the fibrotic process. Reversing this pro-fibrotic phenotype is therefore an important goal therapeutically.
Irrespective of the point from where pathological HLMF differentiation occurs in IPF, we have shown that K
Ca3.1 channels play a key role in this process. By patch clamp electrophysiology we have shown previously that functional K
Ca3.1 ion channels are increased in IPF-derived HLMFs [
22], and we found increased K
Ca3.1 ion channel protein in this study. Importantly, we have shown here that blocking K
Ca3.1 channels has the unique ability to de-differentiate HLMFs towards a fibroblast phenotype as indicated by a marked reduction in αSMA protein and stress fibre formation. This appears to operate via Ca
2+-dependent processes as the reduction in constitutive nuclear Smad2/3 expression induced by K
Ca3.1 blockers was mimicked by removing extracellular Ca
2+, and K
Ca3.1 channel activity is known to influence intracellular Ca
2+ concentrations in many cell types [
34],[
35],[
40]-[
42]. For example, we demonstrated previously that K
Ca3.1 activity is required for a rise in intracellular Ca
2+ that occurs following exposure of HLMFs to TGFβ1 [
22].
Previous studies have focused mainly on TGFβ1-dependent Smad2/3 processes and have been contradictory with regards to the role of Ca
2+. A study on oesteoblasts showed that a TGFβ1-dependent Ca
2+ signal was not required for Smad2 phosphorylation [
43] where as in kidney fibroblasts, TGFβ1-dependent Smad function was controlled directly by Ca
2+-calmodulin [
44]. We found that constitutive Smad2/3 nuclear localisation is highly dependent on Ca
2+ and K
Ca3.1 channel activity. This suggests that the contribution of Ca
2+ signals and K
Ca3.1 activity to Smad2/3 signalling may be cell-specific. This well documented heterogeneity in fibroblast biology highlights the importance of studying cells from the relevant species and tissue of interest. Thus these data from primary IPF-derived HLMFs demonstrate the potential of K
Ca3.1 as a target for IPF.
The increased nuclear localisation of Smad2/3 in IPF-derived HLMFs in association with increased αSMA actin and K
Ca3.1 expression, and the parallel reductions in Smad2/3 nuclear localisation and αSMA expression following K
Ca3.1 blockade, suggest that Smad2/3 signalling is likely to be increased constitutively in IPF-derived cells. However, we were unable to observe phosphoSmad2/3 in the cytoplasm or nucleus. Under these basal conditions, phosphoSmad2/3 might be below the limit of detection, and dephosphorylation within the nucleus may be relatively dominant without a strong exogenous stimulus. For Smad2/3 to enter the nucleus, prior phosphorylation is a requirement [
20],[
45], so it likely that there is increased constitutive Smad2/3 activation in IPF-derived HLMFs. Further experiments will be required to prove this definitively. Similarly, it is likely that K
Ca3.1 is operating to control αSMA expression via regulation of Smad2/3 signalling, but further experiments involving Smad2/3 downregulation would be required to confirm this.
We used two distinct and selective K
Ca3.1 blockers at the IC
50 (20 nM for TRAM-34 and 10 nM for ICA-17043) and at 10× the IC
50 where >95% of channels will be blocked. These concentrations are physiologically relevant, and furthermore, two structurally similar drugs without channel blocking activity, TRAM-7 and TRAM-85, were without effect here and in previous experiments [
22]. Importantly, the concentration of ICA-17043 used here can be achieved in vivo in humans with oral dosing [
46], indicating that the targeting of K
Ca3.1 in IPF is feasible.
Conclusion
K
Ca3.1 channel inhibition attenuates many TGFβ1- and bFGF-dependent profibrotic activities in HLMFs [
22]. However, as shown here, blocking K
Ca3.1 channels promotes the de-differentiation of IPF-derived HLMFs towards a quiescent fibroblast phenotype. This suggests that K
Ca3.1-dependent cell processes may be a common denominator in IPF pathophysiology. K
Ca3.1 knockout animals are relatively healthy, and the K
Ca3.1 blocker ICA-17043 (Senicapoc), when delivered orally, was well tolerated for 12 months in a phase III clinical trial of sickle cell disease [
46]. Targeting K
Ca3.1 may therefore provide a novel and effective approach for the treatment of IPF and there is the potential for the rapid translation of K
Ca3.1-directed therapy to the clinic.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
KMR performed all of the laboratory studies on human lung myofibroblasts under the supervision of PB, analyzed the data and drafted the manuscript. CF-B recruited and characterised patients with IPF for inclusion in the study, advised on myofibroblast cell culture, reviewed the data, and revised the manuscript critically for intellectual content. HW provided TRAM-34 for the in vitro studies, reviewed the data, and revised the manuscript critically for intellectual content. PB and YA conceived and designed the research, supervised KMR, analyzed the data, and drafted the manuscript. All authors approved the final manuscript.