TGF-β1 Impairs Vitamin D-Induced and Constitutive Airway Epithelial Host Defense Mechanisms Open Access

Patients with chronic obstructive pulmonary disease (COPD) suffer more frequently from respiratory infections than ex- or non-smokers, and this may contribute to exacerbations and to the further progression of the disease [1, 2]. This increased susceptibility to infections can be explained by impaired mucociliary clearance and decreased host defense [3], that may in part result from persistent exposure to cigarette smoke (CS) or to other noxious gases [4-6]. The airway epithelium serves as the front line in the lung’s host defense by preventing microbes from entering the tissue and bloodstream. Its contribution to this important function is mediated by a combination of mechanisms including (but not limited to) the maintenance of a physical barrier supported by its tight- and adherens junctions, mucociliary clearance, and secretion of both inducible and constitutively expressed host defense peptides and proteins (HDPs), reactive oxygen- and nitrogen species, interferons, chemokines, and cytokines [7]. In addition to broad-spectrum antimicrobial activity, HDPs also have the ability to modulate immune responses and promote wound repair [8]. Under homeostasis, inducible HDPs such as human β-defensin-2 and hCAP18/LL-37 are expressed at low levels and their expression can be increased upon, for example, activation of pattern recognition receptors [9], cytokine and growth factor receptors, and by other mediators such as vitamin D (VD), whereas constitutively expressed HDPs do not require such stimuli for their expression [9-11].

Whereas VD is classically known for its function in the regulation of calcium homeostasis and bone metabolism, multiple studies have shown that it also acts as an important regulator of host defense and immunity, including respiratory host defense [12, 13]. This was supported by 2 clinical trials that showed that VD supplementation reduces the exacerbation rate in VD-deficient COPD patients [14, 15] and a recent meta-analysis that demonstrated that VD supplementation protects against acute respiratory tract infections [16]. Various mechanisms may contribute to this protective effect of VD, including direct effects such as VD-mediated increases of hCAP18/LL-37, and/or indirect effects via promotion of CFTR expression or its ability to reduce oxidative stress [11, 17-19]. In the airway epithelium, the main circulating form of VD (25[OH)]D₃) is hydroxylated to generate the active form of VD (1,25[OH]₂D₃) by α1-hydroxylase (cytochrome P [CYP] 27B1) [20]. Next, 1,25(OH)₂D₃ binds the nuclear VD receptor (VDR) and heterodimerizes with the retinoic acid receptor to interact with VD response elements to initiate gene expression of more than 900 genes, including CYP24A1, which converts both 25(OH)D₃ and 1,25(OH)₂D₃ into inactive metabolites [12, 21]. The airway epithelium is an important site of local VD metabolism and we and others have shown that the expression or the activity of CYP27B1, CYP24A1, and VDR can be modulated by several inflammatory mediators such as TNF-α/IL-1β, IL-17A, non-typeable Haemophilus influenzae (NTHi), IL-13, the viral analog poly(I:C) and CS, which have been implicated in the pathogenesis of chronic inflammatory lung diseases [5, 20, 22, 23].

Studies in prostate cancer and stromal cells and hepatocytes suggest that the positive effects of VD might be modulated by TGF-β1 [24, 25]. This may be relevant for COPD, since elevated levels of TGF-β1 expression were found in the airways of COPD patients [26-28], although this was not shown by all studies [29]. TGF-β1 is a multifunctional cytokine that is produced and activated upon injury, through CS-exposure or by inflammation [30-33]. When this injury persists, continued release of TGF-β1 contributes to tissue remodeling, a process that may be driven in part by epithelial-to-mesenchymal transition (EMT) [34]. Several studies have shown that VD might counteract TGF-β1-mediated effects on fibrosis, as demonstrated by its ability to inhibit TGF-β1-induced EMT in both mouse models of asthma and fibrosis and in airway epithelial cell lines [17, 35-37]. In addition to its role in fibrosis and EMT, TGF-β1 affects respiratory host defense by impairing anti-viral interferon type I and III responses, but also by restricting the expression of constitutively expressed host defense mediators such as secretory leukocyte protease inhibitor (SLPI) and polymeric immunoglobulin receptor (pIgR) [31, 32, 38, 39].

Despite this insight into the role of TGF-β1 in the pathogenesis of COPD and other chronic inflammatory lung disease, it is not known whether exposure to TGF-β1 also affects respiratory host defense by affecting VD-metabolism and VD-mediated expression of the HDP hCAP18/LL-37. Moreover, it is currently unknown whether VD modulates TGF-β1-mediated repression of constitutively expressed HDPs. We therefore aimed to study the interaction between VD and TGF-β1 on airway epithelial cell host defense mechanisms. To this end, we first investigated the effects and underlying mechanisms of TGF-β1 on VD-metabolism on VD-mediated hCAP18/LL-37 expression. Next, we studied the effects of VD on TGF-β1-induced changes in epithelial composition and on expression of a group of constitutively expressed host defense mediators, as well as on antibacterial activity.

Primary Bronchial Epithelial Cells (PBEC) were obtained from tumor-free bronchial lung tissue from anonymous donors during lung resection surgery for lung cancer at LUMC. Cells were cultured, as previously described, with some adaptations [5, 40]. Briefly, cultures of bronchial epithelial cells (passage 1) were first expanded in T75 culture flasks, pre-coated with a mixture of 30 μg/mL Purecol (Advanced BioMatrix, San Diego, CA, USA), 5 μg/mL stabilized fibronectin (Alfa Aesar, Thermo Fisher scientific, Landsmeer, The Netherlands) and 10 μg/mL BSA (Sigma Aldrich, -Zwijndrecht, The Netherlands). Next, cells were seeded at a density of 5,000 cells per well for submerged cultures of (S)-PBEC or 40,000 cells per insert (passage 2) on pre-coated 24-well plates (Corning Costar, Cambridge, MA, USA) and semi-permeable Transwell inserts, respectively (12 mm, 0.4 μm pore-size, Corning Costar). The cells were cultured in BEpiCM-b:DMEM (B/D)-medium (1: 1; ScienCell Research Laboratories, Uden, The Netherlands and STEMCELL Technologies, Köln, Germany, respectively), supplemented with Bronchial Epithelial Cell Growth Supplement (ScienCell Research Laboratories), and additional 1 nM EC-23 (Tocris, Bio-techne Ltd, Abington, UK; for the submerged phase of PBEC cultures on inserts only), 25 mM HEPES (Cayman Chemical, Hamburg, Germany), 100 U/mL penicillin and 100 µg/mL streptomycin (ScienCell Research Laboratories). After the cells growing on inserts had reached confluence (after 5–7 days), apical medium was removed and the cells were cultured at the air-liquid interface (ALI); the medium was changed 3 times a week with (B/D)-medium supplemented with Bronchial Epithelial Cell Growth Supplement and additional 50 nM EC-23. During refreshment, the apical surface was washed with PBS to remove mucus. After 14 days of air-exposed culture, the cells produced mucus and developed cilia, and cultures were used for experiments. S-PBEC were cultured until they reached 50–70% confluence after 4–5 days and subsequently cultured for 24 h in B/D-medium supplemented with Bronchial Epithelial Cell Growth Supplement without BSA, BPE, EGF and hydrocortisone before stimulation (B/D-starvation medium).

PBEC that had been differentiated for 14 days at the ALI were exposed to various concentrations of TGF-β1 (0.2, 1 and 5 ng/mL) and 100 nM 25(OH)D₃ and/or 1,25(OH)₂D₃ (Millipore B.V., Amsterdam, The Netherlands) for 24 h to assess changes in gene expression, or for 48 h to assess SLPI levels in apical washes, airway epithelial antibacterial activity, and hCAP18/LL-37 release by Western blot or immunofluorescence. S-PBEC cultures were used to elucidate the mechanism of action of TGF-β1-reduced expression of hCAP18/LL-37. To assess the role of TGF-β1-mediated induction of CYP24A1 or canonical TGF-β1-Smad signaling, S-PBEC were treated with 5 ng/mL TGF-β1 and 100 nM 1,25(OH)₂D₃ in the presence or absence of 10 μM TGF-β1-Smad signaling inhibitor SB431542 (Sigma-Aldrich) or 10 μM of the antifungal ketoconazole (KTZ) that acts as an inhibitor of CYP-450 (Sigma-Aldrich) for 24 h.

S-PBEC were used to determine if the TGF-β1-reduced expression of hCAP18/LL-37 was mediated by the transcription factor CCAAT/enhancer-binding protein-α. S-PBEC were refreshed with B/D-starvation medium containing 5 ng/mL TGF-β1 and 100 nM 1,25(OH)₂D₃ and transfected using 3 μL/well RNAiMAX -SilentFect transfection reagent (Thermo Fisher scientific) containing 20 mM CEBPα (CEBPA)- or negative control (CNTRL)-siRNA (Ambion, Thermo Fisher scientific) and incubated for 24 h.

Cells were lysed in RNA lysis buffer (Promega Benelux B.V., Leiden, The Netherlands). The total RNA was robotically extracted using the Maxwell tissue RNA extraction kit (Promega) and quantified using the Nanodrop ND-1000 UV-Vis Spectrophotometer (Nanodrop technologies, Wilmington, DE, USA). For cDNA synthesis, 1 µg of total RNA was reverse transcribed using oligo dT primers (Qiagen Benelux B.V., Venlo, The Netherlands) and M-MLV Polymerase (Thermo Fisher scientific) at 42°C. All quantitative PCR (qPCR) reactions were performed in triplicate on a CFX-384 Real-Time PCR detection system (Bio-Rad Laboratories, Veenendaal, The Netherlands), using primers shown in Table 1 and IQ SYBRGreen supermix (Bio-Rad). The relative standard curve method was used to calculate arbitrary gene expression using CFX-manager software (Bio-Rad). Two reference genes, selected using the “Genorm method” (Genorm, Primer design, Southampton, UK), were included to calculate the normalized gene expression.

For Western blot analysis of hCAP18/LL-37 release, basal medium was applied to Oasis HLB 1cc extraction cartridges (Waters Chromatography, Etten-Leur, The Netherlands) and the eluate was dried by vacuum centrifugation (CHRIST RVC2–25 Vacuüm system) [41]. Lyophilized protein samples were resuspended in 100 µL reducing SDS-PAGE sample buffer, heated for 5 min at 100°C and applied on a 16.5% Tris-Tricine gel, as previously described [5]. Next, proteins were blotted on a Polyvinylidene fluoride membrane and non-specific binding sites were blocked in PBS containing 5% (v/v) heat-inactivated new born calf serum, 5% (w/v) skimmed milk in PBS. Membranes were probed with 1/200 diluted mouse monoclonal anti-hCAP18/LL-37 (clone 1.1.C12; Hycult Biotech, Uden, The Netherlands) in blocking buffer. Next, the membranes were incubated in 1/1,000 diluted rabbit-anti-mouse-HRP (Cell Signaling Technology, Leiden, The -Netherlands) in blocking buffer. SuperSignal West Pico ECL Substrate (Thermo Fisher scientific) was used to visualize hCAP18/LL-37 protein using The ChemiDoc^TM Touch imager in combination with Image Lab^TM software (Biorad).

Cells were fixed on Transwell inserts in 1% paraformaldehyde (Millipore B.V.) in PBS for 10 min on ice and washed with ice-cold PBS. Next, cells were permeabilized with methanol for 10 min at 4°C, washed in PBS, and blocked with PBS/1% (w/v) BSA/0.3% (v/v) Triton-X-100 (PBT) for 30 min at 4°C. Next, cells were treated for 30 min with SFX-signal enhancer (Thermo Fisher scientific) followed by incubation with primary antibodies in PBT for 1 h at reverse transcription (RT; Table 2). After washing in PBS, cells were incubated with an Alexa Fluor 488 labeled secondary antibody (1/200, Alexa Fluor 488 goat anti-rabbit IgG; Thermo Fisher scientific) and Alexa Fluor 568 goat anti-mouse IgG together with DAPI (Sigma Aldrich) in PBT for 30 min at RT in the dark. Finally, cells were mounted in ProLong^TM Gold Antifade Mountant (Thermo Fisher scientific) and images were acquired using a TCS SP5 Confocal Laser Scanning Microscope (Leica Microsystems B.V., Eindhoven, The Netherlands) and LAS AF Lite software (Leica Microsystems B.V.).

Apical washes were obtained by washing the apical surface of the stimulated ALI-PBEC with 200 μL warm PBS for 10 min at 37°C. SLPI in apical washes of the treated ALI-PBEC was measured using an ELISA, as previously described [42].

Antibacterial activity was assessed by applying log-growing cultures of NTHi on the apical surface of the treated cells, as previously described, with a few modifications [40]. NTHi strain D1 was cultured in Tryptone soya broth containing X and V-factor (TSB XV, Mediaproducts BV, Groningen, the Netherlands) while shaking overnight at 37°C [2]. Next, 2 mL of the overnight culture was transferred into fresh 10 mL TSB XV medium and incubated for 4 h at 37°C while shaking to obtain mid log phase growing bacteria. Before applying the washed bacterial suspensions to the apical surface of ALI-PBEC, excess mucus was removed by washing both treated and untreated cells with 200 µL 10 mM sodium phosphate buffer for 15 min at 37°C 6 h before the assay. We applied approximately 1 multiplicity of infection NTHi in 20 μL sodium phosphate buffer + 1% v/v TSB XV per insert for 2 h. Next, membranes containing the cells with bacteria were dissected from the inserts and placed into tubes containing sterile glass beads and 1% TSB in PBS, and cells were disrupted by using a Minilys personal homogenizer (Bertin Instruments, Montigny-le-Bretonneux, France) for 2 times 30 s and kept on ice between treatments. Serial dilutions of bacterial suspensions were plated on chocolate agar plates (Biomerieux, Zaltbommel, The Netherlands), and incubated overnight at 37°C to assess surviving bacteria by colony forming unit determination.

Statistical analysis was conducted using GraphPad Prism 7 (GraphPad Software Inc., La Jolla, CA, USA). To analyze the qPCR results, fold change in gene expression of the stimuli compared to CNTRL was first calculated, followed by log-transformation. All data were analyzed using a two-way ANOVA and the Bonferroni post-hoc test. Differences at p values < 0.05 were considered statistically significant.

We have previously shown that exposure to pro-inflammatory stimuli impairs VD-induced expression and release of the HDP hCAP18/LL-37 in differentiated ALI-PBEC [43]. To investigate if TGF-β1 also affects the expression of the VD-responsive HDP hCAP18/LL-37 (CAMP), we exposed differentiated ALI-PBEC to various concentrations of TGF-β1 for 24–48 h in the presence and absence of 25(OH)D₃ (this inactive form of VD is converted in PBEC by CYP27B1 into active 1,25[OH]₂D₃)). We first confirmed that both 25(OH)D₃ and 1,25(OH)₂D₃ clearly increased the expression of CAMP mRNA in ALI-PBEC after 24–48 h of incubation (Fig. 1a). TGF-β1 dose-dependently limited the VD-increased expression of CAMP at both time points, whereas all concentrations of TGF-β1 decreased the baseline expression of CAMP after 24 h and – only at the highest dose – after 48 h (Fig. 1a). To verify TGF-β1-mediated repression of CAMP at the protein level, we exposed ALI-PBEC to the highest dose TGF-β1 (5 ng/mL) in the presence and absence of 25(OH)D₃ and/or 1,25(OH)₂D₃ for 48 h, and assessed hCAP18/LL-37 secretion in basal medium using Western blot analysis. Using Western blot analysis, both 1,25(OH)₂D₃ and 25(OH)D₃ clearly increased the release of hCAP18/LL-37 in basal medium, which was reduced by TGF-β1, in line with gene expression data (Fig. 1b). These data demonstrate that TGF-β1 interferes with baseline and VD-mediated signaling, resulting in reduced expression and release of the HDP hCAP18/LL-37.

To investigate if the effects of TGF-β1 on VD-mediated expression of hCAP18/LL-37 were mediated by changes in the VD-metabolic pathway, we assessed the effects of TGF-β1 on the expression of the VD-degrading enzyme CYP24A1, the VD-activating enzyme CYP27B1, and the VDR. ALI-PBEC were exposed to various concentrations of TGF-β1 for 24–48 h in the presence and absence of 25(OH)D₃ and 1,25(OH)₂D₃. As expected, both forms of VD increased the CYP24A1 expression (Fig. 2a, left side) compared to CNTRL treated cells. In the absence of VD, TGF-β1 markedly increased the -CYP24A1 expression, and even caused a further small increase in the presence of both forms of VD (Fig. 2a). Furthermore, we also observed a minor dose-dependent change in VDR expression and no effect of TGF-β1 on the expression of CYP27B1 (Fig. 2a). To verify these effects of TGF-β1 on CYP24A1 at the protein level, we performed immunofluorescence staining using CYP24A1 antibodies and confirmed the ability of TGF-β1 to increase the CYP24A1 expression at the protein level in the presence and absence of 1,25(OH)₂D₃ (Fig. 2b). Together these data indicate that TGF-β1 affects VD metabolism by increasing the expression of the VD-degrading enzyme -CYP24A1.

We used S-PBEC to further elucidate the underlying mechanisms of TGF-β1-mediated decreases of hCAP18/LL-37 and increases of CYP24A1 (Fig. 3a, b), after observing that modulation of CAMP and CYP24A1 expression by TGF-β1 was similar in S-PBEC and ALI-PBEC (data not shown). We first investigated if the TGF-β1-mediated repression of VD-induced hCAP18/LL-37 was fully mediated by increases of CYP24A1 expression. To this end, S-PBEC were exposed for 24 h to TGF-β1 and 1,25(OH)₂D₃ in the presence or absence of the CYP-inhibitor KTZ that is known to inhibit CYP24A1 and CYP27B1 activity [44, 45]. To circumvent the effect of KTZ-mediated inhibition of the CYP27B1-mediated hydroxylation of inactive 25(OH)D₃ into active 1,25(OH)₂D₃, we used only 1,25(OH)₂D₃ in these experiments. Whereas KTZ significantly increased 1,25(OH)₂D₃-mediated expression of CAMP in both the presence and absence of TGF-β1, CAMP levels were lower in 1,25(OH)₂D₃-treated cells in the presence of TGF-β1 than in the absence of TGF-β1 (Fig. 3a). This suggests that the VD-mediated reduction of CAMP by TGF-β1 was not fully explained by increased CYP24A1, suggesting involvement of mechanisms other than the VD-metabolic pathway in the observed effects of TGF-β1.

In addition to VD-mediated expression of CAMP, the baseline expression of CAMP was also reduced by TGF-β1 and this could not be restored by inhibition of CYP24A1 (Fig. 3a). TGF-β1 signals through either the canonical-Smad signaling pathway or the non-canonical (MAPK/NF-κB)-pathway [46]. To investigate if the canonical TGF-β-Smad signaling pathway was involved in the inhibition of baseline CAMP expression as well as the increase in CYP24A1, we exposed S-PBEC to TGF-β1 and 1,25(OH)₂D₃ in the presence or absence of SB431542 (an inhibitor of TGF-β type I receptor activin receptor-like kinase and further downstream signaling, that is, via receptor regulated (R)-Smads) [46]. We found that treatment with SB431542 fully reversed the effects of TGF-β1 on both CAMP and CYP24A1 expression, both in the absence or in the presence of 1,25(OH)₂D₃ (Fig. 3b). These results indicate that the canonical TGF-β-Smad signaling pathway mediates the reduction of CAMP and the promotion of CYP24A1 expression by TGF-β1.

In addition to VD-responsive elements that bind the VDR-RXR-1,25(OH)₂D₃-complex, the promotor of CAMP also contains a binding site for the transcription factor C/EBPα, which was shown to be required for the induction of CAMP independent of VD [47]. A study by Li et al. [48] showed that TGF-β-Smad signaling inhibits the expression of C/EBPα mRNA in mesenchymal stem cells, which suggests the involvement of C/EBPα in the suppressive effect of TGF-β1 on CAMP expression. We first demonstrated that TGF-β1 also reduced the mRNA CEBPA expression in S-PBEC after 24 h (Fig. 4a). To investigate the involvement of C/EBPα in baseline and VD-induced CAMP expression, we used siRNA. S-PBEC were transfected with CEBPA-specific siRNA or negative CNTRL siRNA and next cells were exposed to TGF-β1 and 1,25(OH)₂D₃ for 24 h. siRNA caused a marked suppression of CEBPA expression, and both basal- and 1,25(OH)₂D₃-mediated expression of CAMP was reduced (Fig. 4b). The siRNA-induced reductions in both CEBPA and CAMP expression were more pronounced in TGF-β1-treated cells (Fig. 4b). We next measured the expression of CYP24A1, to exclude possible non-specific effects of CEBPA siRNA transfection. Unexpectedly, -CYP24A1, a gene without any known binding sides for C/EBPα in its promotor, was also decreased (Fig. 4b). Since the CYP24A1-promotor does contain binding sides for C/EBPβ (CEBPB) [49], we considered the possibility that CEBPA siRNA also had reduced CEBPB mRNA levels. We therefore assessed the expression of CEBPB and found that this was not affected, suggesting that inhibition of C/EBPα might have indirectly affected the expression of CYP24A1 (Fig. 4b). Collectively, these data show that TGF-β1-mediated changes in CAMP or CYP24A1 expression are mediated through repression of the transcription factor (C/EBP)α by TGF-β1.

Since we demonstrated reduced expression levels of the inducible HDP hCAP18/LL-37 in the presence of VD and TGF-β1, we were interested in investigating whether a similar decrease was also observed in the expression of constitutively expressed host defense mediators such as the HDP SLPI and pIgR, since previous studies showed that they were repressed by TGF-β1 alone in ALI-PBEC [31, 50]. Investigating effects of the combination of TGF-β1 and VD is especially relevant, since VD reduces the effects of TGF-β1-mediated EMT in airway epithelial cell lines [35]. We therefore investigated in ALI-PBEC if VD affected the TGF-β1-induced repression of a selected group of constitutively expressed host defense mediators such as SLPI, s/lPLUNC and pIgR. To this end, differentiated ALI-PBEC were exposed to both TGF-β1 and 1,25(OH)₂D₃ for 24–48 h, and the expression of these mediators was assessed. TGF-β1 decreased the expression of all 3 selected HDPs after 24 h (Fig. 5a), which continued up to 48 h (data not shown). However, 1,25(OH)₂D₃ did not prevent the TGF-β1-meditiated repression of mRNA expression of these constitutively expressed HDPs (Fig. 5a). We additionally verified these effects of TGF-β1 at the protein level using confocal immunofluorescence, and using ELISA to detect SLPI in apical secretions (Fig. 5b, c). TGF-β1-treatment reduced both the staining intensity as well as the number of SLPI- and sPLUNC-positive cells, whereas only the number of pIgR-positive cells was reduced upon TGF-β1-treatment. In cultures from some donors, pIgR was relocated from the cell membrane towards the cytoplasm (Fig. 5b).

We furthermore investigated the functional consequences of the reduced expression of the HDPs by TGF-β1 in the presence and absence of VD by studying the effects of TGF-β1 and VD on antibacterial activity. Differentiated ALI-PBEC were exposed to TGF-β1 for 48 h in the presence and absence of 1,25(OH)₂D₃ and a killing assay was performed by applying log-growing NTHi on the apical surface of 48 h-exposed cells for 2 h. The cell lysates were next diluted and incubated on agar plates overnight to determine surviving bacteria. In contrast to what was previously observed using a different type of antibacterial assay [5], we found that treatment with 1,25(OH)₂D₃ did not increase the antibacterial activity against NTHi. Although not significant, there was a small trend towards reduction of antibacterial activity by TGF-β1 (online suppl. Fig. S1; for all online suppl. material, see www.karger.com/doi/10.1159/000497415).

To investigate whether the ability of TGF-β1 to reduce the expression and release of the constitutively expressed HDPs SLPI, s/lPLUNC and pIgR was explained by TGF-β1-induced changes in epithelial differentiation, effects on differentiation markers were assessed. These experiments were prompted by our recent finding that chronic exposure to CS also impairs the expression and release of these HDPs, accompanied by an impairment of end-stage airway epithelial cell differentiation towards club and goblet cells, being the main source of these HDPs [40]. Furthermore, it was shown that TGF-β1 directs epithelial cells to dedifferentiate, and we therefore investigated if specialized epithelial cells were reduced by TGF-β1 and if VD might counteract this reduction [51]. We therefore exposed differentiated ALI-PBEC to TGF-β1 and 1,25(OH)₂D₃ for 24–48 h and observed that TGF-β1 caused a clear reduction in both mRNA expression of the club and goblet cell markers SCGB1A1 and CLCA1, respectively, as well as in the number of CC16 (club cell) and MUC5AC (goblet cell)-positive cells observed by confocal immunofluorescence staining. mRNA expression of the ciliated cell marker (FOXJ1) as well the number of ciliated cells were unaffected by TGF-β1 (Fig. 6a, b). 1,25(OH)₂D₃ alone did not affect mRNA or protein expression of these cell markers, and did not prevent the TGF-β1-induced decreases in SCGB1A1 and CLCA1 mRNA (Fig. 6a; immunofluorescence data not shown). This indicates that TGF-β1 impairs the expression of luminal expressed HDPs by reducing the number of secretory cells that express these HDPs, which is not modulated by VD.

Here we demonstrate that TGF-β1 affects VD metabolism by increasing the expression of the VD-degrading enzyme CYP24A1 and reduces the VD-mediated expression of the HDP hCAP18/LL37 in both submerged undifferentiated PBEC and in differentiated ALI-PBEC. Moreover, TGF-β1 also reduces the baseline expression of hCAP18/LL-37 via its ability to reduce the expression of C/EBPα. In fully differentiated ALI-PBEC, TGF-β1 represses the expression of constitutively expressed HDPs such as SLPI, s/lPLUNC and pIgR, which might in part be attributed to decreases in the number of secretory club and goblet cells. Treatment with VD did not counteract these effects of TGF-β1 on HDP expression and epithelial differentiation.

To the best of our knowledge, we are the first to show that TGF-β1 affects VD metabolism by increasing the expression of the VD-degrading enzyme CYP24A1, although an earlier study using hepatic cells did show an association between CYP24A1 and TGF-β1 expression [24]. The involvement of CYP24A1 in the TGF-β1-induced decrease in VD-mediated expression of hCAP18/LL-37 was confirmed by inhibiting its activity by KTZ. Since KTZ also inhibits other CYP-enzymes such as CYP27B1, we used 1,25(OH)₂D₃ to avoid the inhibitory effects of KTZ on the conversion of 25(OH)D3 into 1,25(OH)₂D₃ by CYP27B1. These TGF-β1-mediated effects on CYP241 help to explain the inhibitory effects of TGF-β1 on VD-induced expression of hCAP18/LL-37 in both undifferentiated and differentiated PBEC. We observed no changes in CYP27B1 expression and minor changes in VDR expression after TGF-β1 treatment, whereas a study in colon cancer cells showed that VDR expression was repressed by Snail1 and Snail2 that are known TGF-β1-inducible transcription factors [52]. Prior to our findings, Kulkarni et al. [53] demonstrated that TGF-β1 reduces baseline expression as well as phenylbutyrate-mediated increases in CAMP mRNA expression in a bronchial epithelial cell line, which was reversed by SB421543, an inhibitor of the canonical TGF-β-Smad signaling pathway. In line with this, we confirmed that SB421543 also reduced the TGF-β1-mediated increase in CYP24A1 expression. Furthermore, VD-mediated expression of CAMP was even further enhanced when SB421543 was added, suggesting that VD-mediated CAMP expression was negatively affected by endogenous TGF-β activity. We examined the underlying mechanisms of TGF-β1-mediated inhibition of baseline CAMP expression and demonstrated that TGF-β1 repressed mRNA expression of the transcription factor CEBPA. We furthermore demonstrated the relevance of this transcription factor by inhibition of CEBPA using siRNA, which resulted in a decrease of CAMP. TGF-β1 further repressed the expression of CEBPA and CAMP in siRNA treated cells. The importance of C/EBPα in the VD-independent induction of hCAP18/LL-37 (CAMP) expression was previously demonstrated by Park et al. in keratinocytes[47]. Unexpectedly, siRNA-mediated inhibition of CEBPA also inhibited the expression of CYP24A1. By excluding the possibility that CEBPB was also targeted by siRNA, we conclude that CYP24A1 expression may have been indirectly affected by TGF-β1.

In addition to alterations in VD metabolism and effects, TGF-β1 was also found to reduce the number of luminal secretory cells, possibly via initiation of EMT, and impaired expression of constitutively expressed HDPs SLPI and pIgR, which is in line with previous findings [31, 39, 50]. We are, however, the first to report that TGF-β1 decreases the expression of the constitutively expressed HDP s/lPLUNC. We have recently demonstrated that expression of these constitutively expressed host defense mediators was also impaired following chronic CS-exposure, which was accompanied by a selective reduction of differentiation into specialized airway epithelial cells [40]. Since CS-exposure is also known to increase the expression of TGF-β1 in airway epithelial cells [30, 31], we consider the possibility that the CS-induced repression of constitutive expressed HDPs in airway epithelial cells is in part mediated via the induction of TGF-β1. Interestingly and in line with our finding that TGF-β1 decreases the number of club and goblet cells, Gohy et al. reported correlations between a TGF-β1-mediated decrease of pIgR, epithelial dedifferentiation, and increased expression of mesenchymal markers[31]. It is important to consider that TGF-β1 only alters differentiation markers at concentrations above 0.5 ng/ml, as shown by Harrop et al [54]. Additionally, they demonstrated that expression of MUC5AC and MUC5B in differentiated PBEC was decreased by TGF-β2, which is another TGF-β-isoform and uses the same receptors [54]. In contrast to other studies showing inhibition of TGF-β1-mediated effects on EMT by VD [35, 36], we have not observed any ameliorating effects of VD on TGF-β1-mediated effects on the expression of SLPI, s/lPLUNC and pIgR, nor on the expression of secretory cell markers. This may be explained by the fact that we used differentiated primary airway epithelial cells, where VD activity may be more efficiently inhibited by TGF-β1, or by the possibility that autocrine expression or processing of secreted immature TGF-β1 might be more efficient in differentiated primary epithelial cells than in bronchial epithelial cell lines. The underlying mechanism for this difference needs to be further elucidated, for example, by comparing CYP24A1 levels, autocrine production of TGF-β1, expression of TGF-β-receptors or extracellular activation of TGF-β1 between cell lines and differentiated primary cells.

One of the strengths of this study is that we used differentiated primary airway epithelial cells that were obtained from multiple donors, instead of tumor-derived or immortalized airway epithelial cell lines, thereby increasing the relevance of our findings. It needs to be noted that PBEC used in this study were not derived from healthy donors, but from healthy parts of lung tissue derived from patients who have often smoked and underwent lung resection surgery for lung cancer. However, we have previously demonstrated that these cells differentiate into ciliated-, club and goblet cells similar to healthy individuals, develop a strong epithelial barrier and that expression of constitutively expressed HDPs and airway epithelial differentiation between donors with and without COPD do not differ [40]. This was in contrast to the findings by Gohy et al. [51], who showed that features of EMT persist in PBEC derived from COPD patients, most likely as they used material from patients with more severe stages of COPD.

This study also has a few limitations. First, we were not able to detect hCAP18/LL-37 peptide in apical washes of the stimulated PBEC using Western blot analysis, which is in line with our previous study [5]. Therefore, we used pooled and concentrated basal medium, and in line with our previous report detected a hCAP18/LL-37 immunoreactive peptide of 14 kDa size, and did not detect the 4.5 kDa mature antimicrobial peptide LL-37. Levels of this peptide were increased by VD and reduced by TGF-β1. Another limitation of this study is that we were not able to fully confirm TGF-β1-mediated decreases in HDPs at the functional level by measuring the antibacterial activity against NTHi. As hCAP18/LL-37, SLPI and s/lPLUNC also have other activities such as anti-biofilm, immunomodulatory, and anti-protease activities in addition to their antibacterial activities [55-57], a more complex culture model using a combination of immune cells and airway epithelial cells might be relevant as an alternative approach to establish the consequences of these changes on host defense.

To extend the relevance of our findings to the situation in vivo, further studies are required to compare lung tissue levels of CYP24A1, 1,25(OH)₂D₃ and expression of these HDPs in healthy donors and donors with chronic inflammatory lung disease or fibrosis. We showed that TGF-β1 reduces the number of club cells, which is in line with the observations in diseases associated with both elevated TGF-β1-levels and reduced numbers of club cells such as COPD, asthma, and bronchiolitis obliterans syndrome [46, 58, 59]. A contribution of TGF-β1 to the pathogenesis of COPD is supported by findings of 2 studies showing that TGF-β1 levels correlated with disease severity and airflow limitation in COPD patients [27, 60]. In addition, TGF-β1 expression was higher in airway epithelium of smokers with COPD compared to smokers without COPD [26, 27]. To date, no treatment is available for COPD patients that selectively targets the harmful effects of TGF-β1, without affecting the beneficial effects of TGF-β1. Clinical trials that investigated the use of global inhibitors of TGF-β signaling in oncology or in idiopathic pulmonary fibrosis showed that these compounds are frequently associated with adverse (health) effects and limited clinical benefit [61]. Drugs such as pirfenidone that block downstream TGF-β pathways without affecting the immune system might be a better approach, and showed promising results in clinical trials in idiopathic pulmonary fibrosis [61, 62]. Our study additionally suggests that VD is not a candidate to ameliorate the negative effects of TGF-β1 on airway host defense.

In conclusion, we have shown that TGF-β1 reduces the host defense of airway epithelial cells by impairing the VD-mediated expression of HDPs as well as constitutively expressed luminal HDPs such as SLPI and s/lPLUNC and pIgR. We have additionally shown that TGF-β1 reduces the number of secretory club and goblet cells, which might have additional consequences for host defense. We furthermore conclude that TGF-β1 reduces the VD-mediated expression of hCAP18/LL-37 via a dual mechanism: directly by reducing the expression of an important transcription factor for hCAP18/LL-37 and indirectly via increasing of CYP24A1 that promotes the degradation of VD. These findings may have implications on our understanding of the role of TGF-β1 in COPD by extending the range of mechanisms affected by TGF-β1.

We would like to thank the departments of Microbiology and Infectious Diseases for their help and allowing us to use their laboratories for the culturing of NTHi and for performing antibacterial assays, respectively.

Use of lung tissue that became available for research within the framework of patient care was in line with the “Human Tissue and Medical Research: Code of conduct for responsible use” (2011; www.federa.org) that describes the no-objection system for coded anonymous further use of such tissue. Therefore, individual written or verbal consent is not applicable.

The authors have no conflicts of interest to declare.

This study was supported by a grant from the Lung Foundation Netherlands (grant # 5.1.13.033) and a Marie Curie Intra--European Fellowship (grant #622815).

Conception and design: J.S., A.D., P.S.; Analysis and interpretation: J.S., D.N., A.D., P.S.; Drafting the manuscript for important intellectual content: J.S., A.D., P.S.; all authors have reviewed the manuscript and agree with its submission.