Deficient repair response of IPF fibroblasts in a co-culture model of epithelial injury and repair

The extent of re-epithelialization was investigated by wound closure of PKH26-labeled A549 cells in mono-cultures or in co-cultures with NHLFs or IPF fibroblasts over a 96-h time-course (Figure 1). Epithelial wound closure was significantly enhanced in the presence of IPF fibroblasts by 48 h compared to NHLF co-cultures and by 72 h compared to A549 mono-cultures. This difference increased further by 72 h and 96 h. Compared to A549 mono-cultures, co-culture with NHLFs significantly inhibited wound re-epithelialization by A549 cells at 72 h and 96 h post injury. These results were consistent when using multiple NHLF and IPF donors (Figure 2A, B).

Next we investigated how epithelial cells and fibroblasts contributed to epithelial repair. NHLFs and IPF fibroblasts responded differentially to epithelial wounding. While NHLFs sensed and migrated in to wound area progressively in response to epithelial injury, IPF fibroblasts appeared to be characterized by a reduced migratory phenotype. By 48 to 96 h, the entire wound area was populated with NHLFs in NHLF-A549 co-cultures leading to the formation of a fibroblast plug. The number of NHLFs populating the epithelial wound area remained significantly higher compared to IPF fibroblast-A549 co-cultures between 48 and 96 h. IPF fibroblasts failed to populate the scratch wound area and these co-cultures exhibited complete re-epithelialization of the wound by the A549 cells by 72 to 96 h (Figures 3 and 4A; Figure 4B; Additional file 1). Presence of sub-epithelial fibroblasts and their phenotypes also had a marked effect on the rate of epithelial wound closure. NHLFs prevented while IPF fibroblasts promoted the rate of wound closure compared to A549 mono-cultures (Figure 2A). These differences were significant from 48 h for IPF co-cultures and at the 72 h time point for NHLF co-cultures (Figure 2B).

Cell-cell communication via direct contact or through soluble factors may explain the differential behavior we observe. To investigate if this was the case, the levels of secreted growth factors were analyzed. There was no significant difference in the level of bFGF, HGF, transforming growth factor beta-1 (TGF-β1), and platelet-derived growth factor-AA (PDGF-AA) in conditioned medium (CM) from either NHLF or IPF fibroblast mono-cultures (Additional file 2). In contrast, in co-culture experiments we detected significant differences. In CM from IPF co-cultures, the level of bFGF was significantly elevated (five- to seven-fold) at 48 h compared to A549 alone and NHLF co-cultures. This difference was even more pronounced at 72 h with a nearly 20-fold increase in bFGF in IPF co-cultures (Figure 5A). There was also a two-fold increase in the level of PDGF-AA in IPF co-cultures compared to A549 mono-cultures or NHLF co-cultures at 72 h (Figure 5C). On the other hand, the level of HGF was significantly higher in NHLF co-cultures at both time points compared to IPF co-cultures and to A549 mono-cultures (Figure 5B). There was no difference in the level of total TGF-β1 in CM among the different culture conditions (Figure 5D). In the absence of A549, the level of TGF-β1 was extremely low (Additional file 2), suggesting that A549 is the major source of this growth factor. The levels of epidermal growth factor (EGF), insulin like growth factor 1 (IGF-1), and platelet-derived growth factor-BB (PDGF-BB) were below detection threshold.

We investigated the effect of epithelial-fibroblast crosstalk on vimentin expression in the repairing A549 cells. A previous study has demonstrated that increased vimentin expression in A549 cells can enhance migration and repair in vitro[15]. There were significantly fewer e-cadherin and vimentin co-expressing A549 cells in NHLF co-cultures compared to A549 mono-cultures or in co-cultures with IPF fibroblasts (Figure 6A, B). This difference was already striking at 24 h even before the formation of the ‘fibroblast plug’ as seen in the case of NHLF co-cultures (Figure 3). Compared to A549 mono-cultures, the number of vimentin expressing A549 cells peaked earlier in IPF co-cultures. This was significantly greater at 48 h post scratch, reducing gradually thereafter. By 96 h, when re-epithelialization was complete, there were few or no vimentin co-expressing A549 cells in the IPF co-cultures (Figure 6B).

Both epithelial migration and proliferation can contribute to re-epithelialization [16]. We therefore sought to establish if A549 cell proliferation mediated by secreted factors in the CM was an underlying mechanism driving the differences we observed in the response of A549 cells to injury. There was no difference in A549 cell proliferation between cells treated with CM from either epithelial cell mono-cultures or co-cultures with NHLFs or IPF fibroblasts (Figure 6C).

To investigate if differential receptor expression contributed to the altered response of IPF fibroblasts to epithelial injury, PDGFRα, PDGFRβ, and FGFR1 expression was analyzed in the two fibroblast phenotypes. PDGFRα expression was consistently reduced in all IPF fibroblasts compared to NHLF donors (Figure 7A). PDGFRβ expression was reduced to a lesser extent and presented a more heterogeneous expression across the IPF donors. The expression of FGFR1 was largely similar between IPF fibroblasts and NHLFs with a slight trend towards increased expression in IPF fibroblasts (Figure 7A).

Fibroblasts can populate epithelial wounds through migration and also proliferation of migrated fibroblasts [17]. We therefore investigated if receptor expression impacted the migratory potential of IPF fibroblasts and NHLFs. We did not see an effect of fibroblast phenotypes on baseline migration in serum-free medium. Furthermore, in the absence of stimuli, the number of fibroblasts migrating through the membrane remained extremely low. PDGF-AA strongly induced a directional migration in both normal and IPF fibroblasts; however, the extent of this response was significantly lower in the case of IPF fibroblasts (Figure 7B) suggesting a key role for PDGFRα in this response. bFGF significantly induced migration of both fibroblast phenotypes compared to baseline. However, no significant difference was seen in bFGF-induced migration between NHLFs and IPF fibroblasts. It is well documented that both PDGF-AA and bFGF can act as chemo-attractants as well as mitogens for fibroblasts [18, 19]. In line with this, NHLFs stimulated with PDGF-AA demonstrated six-fold increase in proliferation in our study. Consistent with the reduced PDGFRα levels, IPF fibroblasts were less responsive to PDGF-AA-mediated proliferation with only two-fold induction. Both NHLFs and IPF fibroblasts were equally responsive to bFGF-mediated proliferation with a five- to six-fold induction (Figure 7C).

To further ascertain if PDGF-driven fibroblast recruitment is a key mechanism driving the differential wound repair response, we assessed the effect of PDGF neutralization on fibroblast migration at 72 h post injury (Figure 8, Additional file 3). Neutralization of PDGF isoforms did not have an effect on IPF fibroblast migration which remained mostly undetectable. On the contrary, PDGF-AA neutralization resulted in a significant reduction in NHLF invasion. There was a three-fold reduction in the number of NHLFs populating the epithelial scratch wound. Recruited fibroblasts occupied a smaller area and also appeared less dense compared to the IgG treated control. PDGF-BB neutralization did not significantly affect migration of NHLFs (Figure 8A, 8C). Neutralization of PDGF-AA significantly inhibited epithelial wound closure in A549 mono-cultures and IPF co-cultures compared to IgG treated controls. Consistent with reduced NHLF invasion the effect of PDGF-AA was reversed in the case of NHLF co-cultures, where the PDGF-AA neutralization led to an increase in epithelial wound closure as the NHLFs failed to fill the epithelial wound area. No significant effect of PDGF-BB neutralization was seen on epithelial closure in any of the three cultures (Figure 8A, 8B). Neutralization using pan-PDGF antibody had a similar effect as PDGF-AA neutralization alone (Additional file 3). These data suggest that PDGF signaling impacts wound repair in our model system by acting both on fibroblast recruitment and on epithelial migration/proliferation and that this effect is primarily driven by the PDGF-AA isoform.

A549 cells were obtained from ATCC, primary human lung fibroblasts (NHLFs) from Promocell (Heidelberg, Germany). IPF fibroblasts were obtained from surgical lung biopsies of patients and cultured as described previously [48]. All cell types were maintained in DMEM/F12 medium supplemented with 10% FBS, 1% sodium pyruvate, and 1% antibiotics (Invitrogen, Paisley, UK) at 37°C with 5% CO₂. Primary fibroblasts were used between passages 4 to 7.

Trans-well inserts 24 mm in size with 8 μm pore size (VWR, Leicestershire, UK) were coated with collagen-I at on both sides. A total of 100,000 NHLFs or IPF fibroblasts were allowed to attach to inverted trans-wells for 2 to 3 h. Trans-wells were placed in six well plates with 3 mL complete medium in the lower chamber. The upper chamber of each trans-well was incubated with 2 mL of complete medium containing 500,000 A549 cells or medium only (in case of fibroblast only controls) for a further 4 to 5 h. A549 cells were labeled with the fluorescent dye PKH26 (Sigma). The medium in both chambers was replaced with starvation medium containing 0.05% FBS for 16 h. Three 0.4-0.5 mm parallel scratches were made in the A549 cell layer using a sterile pipette tip. Cells were washed and medium was replaced. For PDGF-AA, PDGF-BB, and Pan-PDGF neutralization, medium was supplemented with 0.5 μg/mL of neutralizing antibody or isotype control (R & D systems).

Experiments were carried out in triplicates with NHLFs (n = 2-4) and IPF fibroblasts (n = 3-5). CM was stored at -80°C for up to 3 days before a single thaw and further analysis and trans-wells fixed with 5% neutral buffered formalin solution (Sigma) for immune-fluorescent staining.

The bottom side of trans-well membranes was wiped clean using moist cotton buds to ensure that only cells on the epithelial side remained. Staining was carried out by incubating with primary antibodies: rabbit anti-e-cadherin (Abcam, Cambridge, UK); mouse anti-α-SMA (Sigma); or mouse anti-vimentin (Dako, Ely, UK) followed by incubation with DAPI/AlexaFluor anti-rabbit488 or anti-mouse647 secondary antibodies (Invitrogen). Stained trans-wells were analyzed using LEICA, TCS SP2 confocal microscope (Leica, Heerbrugg, Switzerland).

After seeding fibroblast as described earlier followed by the 16-h starvation, medium in the top chamber was replaced with 2 mL starvation medium containing 20 ng/mL PDGF-AA or 20 ng/mL bFGF (R & D Systems) for directional migration or starvation medium for non-directional baseline migration.

Wound closure was monitored in real time by visualizing the fluorescently labeled A549 cells using an inverted microscope (Olympus IX71) equipped with a Nikon camera. Digitized images of three representative areas for every trans-well were captured at 0 h, 24 h, 48 h, 72 h, and 96 h. Images were analyzed using Image-J software (Image-Pro Plus, Version 6.3, Datacell).

To analyze immuno-fluorescently stained trans-wells, 40× images were taken at 10 random points along a centrally placed scratch. At time points where the field of view could not encompass both wound edges lining the denuded area (0 h and 24 h), images were taken to include one wound edge and the denuded area ensuring similar coverage of scratch for all trans-wells. The number of positive staining cells was counted and averaged for all fields. Data were expressed as the number of cells per High Power Field (HPF). Fibroblasts were identified as elongated cells expressing DAPI/α-SMA or DAPI/vimentin. Epithelial cells (A549) were identified as cells expressing E-cadherin/DAPI. A549 cells transiently expressing mesenchymal markers were identified by the presence of triple staining for e-cadherin/vimentin/DAPI.

CM was analyzed for growth factors. bFGF, HGF, and TGFβ-1 were assessed by using single-plex assays from Meso Scale Discovery (Rockville, MD, USA). Levels of (EGF), IGF-1, PDGF-AA, and PDGF-BB were assessed using Quantikine ELISA (R & D Systems, Abingdon, UK). All assays were performed as per manufacturer’s protocol. For ELISA, CM from four biological replicates were pooled and concentrated 10 times using Amicon Centrifugal Filtration Units (Millipore, Watford, UK). Growth factor concentrations shown in this work fell within the dynamic range of the assays used.

A549 cells were plated at 5,000 cells per well in a 96-well assay plate, starved for 48 h, and then stimulated for a further 48 h with CM. Proliferation of fibroblasts in response to PDGF-AA (150 ng/mL) and bFGF (2.4 ng/mL) was assessed in a similar way. Proliferation was assessed using the DELFIA Assay System (Perkin Elmer, Cambridge, UK).

A549 cells were allowed to reach 100% confluence followed by 24 h starvation. Scratch wounds of 0.3 to 0.4 mm were generated across the wells that were then replenished with either starvation medium or starvation medium containing PDGF-AA (20 ng/mL), bFGF (10 ng/mL). Images were acquired and analyzed as described earlier.

Cells were lysed using RIPA buffer (20 mM Tris-HCl -pH 7.5, 150 mM NaCl, 1 mM EDTA). Proteins were quantified using the Pearce BCA assay kit (VWR). After separation on NUPAGE gels, proteins were transferred to a PVDF membrane and probed overnight with primary antibodies against PDGFRα, PDGFRβ, FGFR1, and GAPDH at 4°C followed by appropriate secondary antibody incubation for 1 h at room temperature. All antibodies were purchased from Cell Signaling Technologies (Boston, MA, USA).