Characterization of intercellular communication and mitochondrial donation by mesenchymal stromal cells derived from the human lung

Bone marrow-derived mesenchymal stromal cells (BM-MSCs) are capable of repairing wounded lung epithelial cells by donating cytoplasmic material and mitochondria. Recently, we characterized two populations of human lung-derived mesenchymal stromal cells isolated from digested parenchymal lung tissue (LT-MSCs) from healthy individuals or from lung transplant recipients’ bronchoalveolar lavage fluid (BAL-MSCs). The aim of this study was to determine whether LT-MSCs and BAL-MSCs are also capable of donating cytoplasmic content and mitochondria to lung epithelial cells. Cytoplasmic and mitochondrial transfer was assessed by co-culturing BEAS2B epithelial cells with Calcein AM or Mitotracker Green FM-labelled MSCs. Transfer was then measured by flow cytometry and validated by fluorescent microscopy. Molecular inhibitors were used to determine the contribution of microtubules/tunnelling nanotubes (TNTs, cytochalasin D), gap junctions (carbenoxolone), connexin-43 (gap26) and microvesicles (dynasore). F-actin microtubules/TNTs extending from BM-MSCs, LT-MSCs and BAL-MSCs to bronchial epithelial cells formed within 45 minutes of co-culturing cells. Each MSC population transferred a similar volume of cytoplasmic content to epithelial cells. Inhibiting microtubule/TNTs, gap junction formation and microvesicle endocytosis abrogated the transfer of cytoplasmic material from BM-MSCs, LT-MSCs and BAL-MSCs to epithelial cells. In contrast, blocking connexin-43 gap junction formation had no effect on cytoplasmic transfer. All MSC populations donated mitochondria to bronchial epithelial cells with similar efficiency. Mitochondrial transfer was reduced in all co-cultures after microtubule/TNT or endocytosis inhibition. Gap junction formation inhibition reduced mitochondrial transfer in BM-MSC and BAL-MSC co-cultures but had no effect on transfer in LT-MSC co-cultures. Connexin-43 inhibition did not impact mitochondrial transfer. Finally, bronchial epithelial cells were incapable of donating cytoplasmic content or mitochondria to any MSC population. Similar to their bone marrow counterparts, LT-MSCs and BAL-MSCs can donate cytoplasmic content and mitochondria to bronchial epithelial cells via multiple mechanisms. Given that BM-MSCs utilize these mechanisms to mediate the repair of damaged bronchial epithelial cells, both LT-MSCs and BAL-MSCs will probably function similarly.


Background
Recent studies suggest that adult stem cells reside within niches where they coordinate organ homeostasis and repair after injury. In the bone marrow, mesenchymal stromal cells (MSCs) and haematopoietic stem cells (HSCs) reside proximal to adrenergic nerve fibres to form the HSC niche [1]. Bone marrow-derived MSCs (BM-MSCs) are an essential component of this structure, forming gap junctions with HSCs and expressing HSC maintenance genes which directly contribute to the HSC volume and capacity to mobilize to extramedullary sites during haematopoiesis [1]. Emerging evidence suggests that the function of MSCs within stem cell niches, irrespective of organ, is to provide support for and coordinate the behaviour of parenchymal progenitor cells [1][2][3][4]. In order to accomplish this, BM-MSCs must be able to effectively communicate and interact with their target cells. Known mechanisms already characterized for BM-MSCs include microtubules [5], tunnelling nanotubes (TNTs) [3,[6][7][8], gap junctions [9], microvesicles [9][10][11][12] and exosomes [13,14]. While the transferred material is yet to be fully defined, it is known to include growth factors [1,7], microRNA [14], immuno-modulatory molecules [15][16][17] and mitochondria [5,6,8,9]. Notably, mitochondrial transfer is emerging as an important mechanism by which MSCs can stabilize or enhance the bioenergetic potential of target cells, which includes alveolar epithelial cells [9] and alveolar macrophages [18]. These mechanisms are thought to contribute to the therapeutic benefit of MSCs by enhancing epithelial proliferation [5,9], modulating T-lymphocyte proliferation [19,20] and inducing the polarization of macrophages towards the M2 phenotype [18,21,22]. As such, MSCs have been demonstrated to promote tissue homeostasis and tissue repair in multiple models of lung injury including heat shock [5,23], cigarette smoke-mediated wounding [6], acute lung injury [8,9], experimental asthma [8,24,25] and bleomycin-induced pulmonary fibrosis [26,27]. Despite this knowledge, whether these modes of behaviour exist in MSC populations which reside endogenously within the human lung remains to be elucidated.
The relationship between MSCs and parenchymal progenitor cells is yet to be fully described in the human lung, but is probably analogous to that of the bone marrow. We suspected that human lung MSCs promote tissue homeostasis and repair by donating cytoplasmic material and mitochondria to parenchymal cells such as bronchial epithelium. Evidence to support this hypothesis comes from murine studies which demonstrate that lung-derived MSCs produce growth factors such as fibroblast growth factor-10 and hepatocyte growth factor, which promote the proliferation and propagation of lung epithelial progenitor cells into complex airway and alveolar structures [2,28]. Recently, we described two populations of human lung MSCs isolated from digested parenchymal lung tissue (LT-MSCs) from healthy individuals or from lung transplant recipients' bronchoalveolar lavage fluid (BAL-MSCs) [30]. The aim of this study was to test the hypothesis that, like their bone marrow counterparts, human lung-derived MSCs donate cytoplasmic content and mitochondria to human bronchial epithelial cells, in vitro, and to characterize the underlying mechanisms.

Cell lines and ethics
Human BM-MSCs were kindly donated by Cell and Tissue Therapies, WA, Australia. BEAS2B epithelial cell lines were purchased from American Type Tissue Culture Collection.

LT-MSC isolation
Parenchymal lung tissue was obtained from patients undergoing a blebectomy to treat recurrent pneumothorax. Samples of macroscopically normal lung were mechanically digested into 1 mm 3 pieces and seeded onto plastic culture dishes at a density of 1 g per dish (10 cm diameter). Cells were maintained in Dulbecco's modified Eagle medium (DMEM; Gibco), supplemented with 10 % fetal calf serum and penicillin/streptomycin/ glutamine (Gibco). Between 7 and 14 days after seeding, colony-forming units comprised of LT-MSCs were observable. MSC colonies were isolated and transferred to individual flasks where they were allowed to clonally expand up to 60-80 % confluence, after which they were cryopreserved in liquid nitrogen. All cells used were between passages 2 and 4 and were preserved in liquid nitrogen prior to use.

BAL-MSC isolation
We and others have reported previously that BAL-MSCs are rarely isolated from healthy individuals [29,30]. Bronchoalveolar lavage (BAL) fluid was therefore collected from lung transplant recipients undergoing a post-transplant bronchoscopy. BAL was performed by wedging the bronchoscope in the middle lobe or lingula, infusing 100 ml of 0.9 % saline via the working channel and aspirating the effluent. BAL samples were immediately centrifuged and seeded into six-well plates at a density of (0.3-0.4) × 10 6 cells/well. Colony-forming units comprised of BAL-MSCs were observable within 14-21 days after seeding. Cells were expanded and maintained as already detailed.

MSC characterization
As described previously [29,31], BAL-MSCs and LT-MSCs are classified as MSCs based on the diagnostic criteria outlined by The Internationally Society for Cellular Therapy [32] and transcriptional profiling [33]. Briefly, both populations are plastic adherent, express the cell surface proteins CD105, CD73 and CD90, and lack CD45 expression. Cells also have a demonstrable capacity for osteoblast, chondroblast and adipocyte differentiation [29,30,31]. Finally, both populations qualify as MSCs according to transcriptional profiling [30,33]. Additionally, BAL-MSCs lack characteristic fibrocyte markers and express a HLA serotype consistent with that of the lung donor.
Cytoplasmic exchange and mitochondrial donation assays BEAS2B lung epithelial cells were labelled with 5 μl of DiD labelling solution (Life Technologies) for 20 minutes at 37°C. MSCs were labelled with 1 mM Calcein AM (Life Technologies) for 35 minutes at 37°C or 40 nM Mitotracker Green (Life Technologies) for 20 minutes at 37°C. Cells were cultured in 24-well sized plates precoated with a solution of 0.1 mg/ml fibronectin (Sigma Aldrich), 0.03 mg/ml bovine collagen type 1 (Worthington) and 0.01 mg/ml bovine serum albumin (Bovogen). Then 3 × 10 4 BEAS2B cells were seeded and allowed to adhere overnight, prior to the addition of 3 × 10 4 MSCs. Cells were co-cultured for 4 hours in a 1:1 mixture of basal epithelial growth medium (BEGM; Lonza) and DMEM containing 10 % fetal calf serum. Cells were then washed with phosphate-buffered saline (PBS) and detached by incubating cells in 0.25 % trypsin (Gibco) for 5 minutes at 37°C. Prior to flow cytometric assessment, cellular viability was assessed using 4 % trypan blue exclusion and light microscopy. Cellular viability was always greater than 95 %. For transwell assays, all details are as already listed except MSCs were seeded into 0.4 μM transwell inserts (MilliCell). Specific dye transfer was measured by gating on DiD Pos events, excluding doublets (forward scatter width vs height) and measuring fluorescence in the fluorescein channel (Additional file 1: Figure S1 and Additional file 2: Figure S2). The mean fluorescent intensity (MFI) is expressed as a ratio to unstained controls. Flow cytometry was performed on a FACS CantoA (BD) and data were analysed using FloJo (Tree Star). Each data point shown in Figs. 2, 3, 4 and 5 represents the average of two replicates.
Inhibitory compounds (or vehicle controls; Additional file 3: Table S1) used to determine the mechanisms underlying cytoplasmic/mitochondrial transfer were added to culture simultaneously with MSCs and remained in culture for the duration of the assay.

Fluorescent microscopy
Co-cultures were fixed with 4 % paraformaldehyde (BD), permeabilized with 1 % Triton X (Scharlau) in PBS and blocked for 1 hour in 1 % bovine serum albumin (Bovogen). Co-cultured cells were stained with Phalloidin (AbCam CytoPainter Kit) for 90 minutes for microtubules/TNTs, as per the manufacturer's instructions, and α-cytokeratin-FITC (15 μG; Milltenyi) for epithelial cells, washed with PBS and mounted in ProLong Gold (containing DAPI; Life Technologies). Fluorescent labelling was visualized using an Axiolmager Z1 microscope (Zeiss) and Axiovision image acquisition software. For live cell confocal microscopy, MSCs were labelled with Mitotracker Green FM and BEAS2B cells with DiD, as already described. Cells were seeded onto 10 mm diameter glass-bottom culture dishes (Ibidi). Dishes were imaged at 37°C at a rate of 1 image per 10 seconds using a 63× oil immersion lens, coupled with a 10× eyepiece lens and a differential interference contrast filter using a Leica TCS SP2 microscope.

Statistics
Statistical analysis was performed using Stata v11 (Stata-Corp) with the Mann-Whitney U test or Wilcoxon signed-rank test, as appropriate. All data are expressed

Patient demographics
Healthy parenchymal lung tissue was collected from 10 patients (six male) with a median age of 24.9 (21.7-27.9) years. BAL-MSCs were collected from 14 lung transplant recipients (10 male) with a median donor age of 30 (28-54) years and median time post transplant of 2.78 (0.27-4.48) years. Patient demographics for BAL-MSCs are presented in Table 1.

BM-MSCs, LT-MSCs and BAL-MSCs connect with BEAS2B epithelial cells via microtubules and TNTs
Microtubules and TNTs were first observed connecting MSCs and BEAS2B epithelial cells after 45 minutes of co-culturing cells. These projections stained strongly with phalloidin indicating that they are comprised of F-actin, consistent with previous reports for BM-MSCs [8] (Fig. 1).

BM-MSCs, LT-MSCs and BAL-MSCs have a similar efficiency for uni-directional cytoplasmic transfer to BEAS2B epithelial cells
We aimed to determine whether the observed interaction between MSCs and epithelial populations mediated the transfer of cytoplasmic material. Calcein AM transfer from MSCs to epithelial cells was rapid and first observed after 4 hours of co-culture, as shown by an increase in BEAS2B cell Calcein AM fluorescence (Fig. 2) and a corresponding decrease in MSC Calcein AM fluorescence (Additional file 4: Figure S3A).
Cytoplasmic transfer occurred at a similar level for all MSC populations (Fig. 2d). After co-culture with BM-MSCs, the Calcein AM MFI in BEAS2B cells was increased 223.5-fold (172.2-565.9, n = 14). Similarly, BEAS2B Calcein AM MFI was increased 252.1-fold . MSCs were prelabelled with Calcein AM (green) and co-cultured with BEAS2B epithelial cells labelled with DiD (red). BEAS2B cells containing MSC-derived Calcein AM appear yellow/orange. Images shown at 630× magnification. Scale bars represent 10 μM (150.9-252.1, n = 9) after co-culture with LT-MSCs and was increased 192.3-fold (156.9-513.4, n = 12) after coculture with BAL-MSCs. We did not observe Calcein AM transfer when the cells were separated by transwell culture (Fig. 2c). Finally, cytoplasmic transfer was unidirectional and could not be observed to occur from BEAS2B cells to any MSC population (Fig. 2b).

Cytoplasmic transfer from BM-MSCs, LT-MSCs and BAL-MSCs to BEAS2B epithelial cells is mediated by contact-dependent and contact-independent mechanisms
We next sought to characterize the mechanisms of cytoplasmic transfer between MSCs and epithelial cells (Fig. 3). In line with a previous report [9], we used the dynamin inhibitor, Dynasore, to inhibit BEAS2B cells endocytosing MSC-produced microvesicles. Endocytosis inhibition reduced cytoplasmic transfer by 61. The greatest reduction in cytoplasmic transfer was observed using the non-specific gap junction blocker carbenoxolone [35,36] (Fig. 3). Cytoplasmic transfer from BM-MSCs to BEAS2B cells was reduced by 90.8 % (88.9-93.2, p < 0.05, n = 6) after carbenoxolone treatment. Similarly, transfer was reduced by 85.0 % (79.4-87.5, p < 0.05, n = 6) and 88.5 % (85.6-90.9, p < 0.05, n = 6) in LT-MSC and BAL-MSC co-cultures respectively. To determine the contribution of connexin-43-dependent gap junctions, which have previously been demonstrated to mediate cytoplasmic transfer in murine studies [9], we employed the inhibitor gap26 [37,38]. Blocking connexin-43 did not prevent cytoplasmic transfer with any MSC cell type.
BM-MSCs, LT-MSCs and BAL-MSCs share similar efficiency for uni-directional mitochondrial transfer to BEAS2B epithelial cells BM-MSCs can promote the regeneration of damaged lung epithelial cells through the donation of mitochondria using similar mechanisms to those described for cytoplasmic transfer [5,9]. We found that all MSC populations can donate mitochondria to BEAS2B epithelial cells, as indicated by increased BEAS2B Mitotracker Green fluorescence (Fig. 4) and decreased MSC Mitotracker Green fluorescence (Additional file 4: Figure  S3B) after co-culture. When visualized using live cell confocal microscopy, mitochondria localized within MSC microtubules or TNTs, tracked towards BEAS2B cells and crossed the cell membrane border. In addition, individual mitochondrion could be seen within endosomes (Fig. 4e, white arrowhead) and microvesicles (Fig. 4e, f, white arrows). Following co-culture with BM-MSCs, BEAS2B Mitotracker Green MFI increased 15.9-fold (10.7-22.4, n = 10). Similarly, BEAS2B Mitotracker Green MFI increased 26.7-fold (12.1-44.6, n = 10) after co-culture with LT-MSCs and increased 15.9-fold (12.7-25.2, n = 9) after co-culture with BAL-MSCs (Fig. 4d). Mitochondrial transfer was not observed when BEAS2B cells were co- were co-cultured with BEAS2B epithelial cells and treated with either dynasore to inhibit endocytosis, cytochalasin D to inhibit microtubule/TNT formation, carbenoxolone to inhibit all gap junction formation or GAP26 to inhibit connexin-43 gap junction formation. Y axis represents this change in mean fluorescent intensity (MFI, %) relative to paired, uninhibited control. Horizontal line represents median. Groups were compared using the Mann-Whitney U test. *p < 0.05 compared with baseline cultured with any MSC populations using a transwell set-up (Fig. 4c). Additionally, we did not observe any mitochondrial transfer from BEAS2B cells to any MSC populations (Fig. 4b).  (Fig. 5). When all inhibitory compounds were added together we observed greater attenuation of mitochondrial transfer (Fig. 5), suggesting synergy Images are shown at 630× magnification with an oil immersion lens. White arrows point to microvesicles containing mitochondria, white arrowhead points to an endocytosed MSC-derived mitochondria between microtubule/TNT, endocytosis and gap junction mediated mechanisms.

Discussion
MSCs exhibit therapeutic effects in multiple models of acute and chronic lung epithelial injury and represent a promising therapeutic approach for the treatment of end-stage lung diseases including idiopathic pulmonary fibrosis [39] and acute lung injury [40] in humans.
These studies have utilized MSCs isolated from tissue sources such as the bone marrow, adipose and placenta. Comparative studies have demonstrated that, irrespective of source, MSCs from any two tissues are strikingly similar [41,42]. Despite this, each population possesses subtle nuances, specific to their source [30,43], which may confer additional benefit or efficiency in a therapeutic context. One of the mechanisms by which MSCs support epithelial cell repair is thought to be via direct transfer of cytoplasmic contents [9,35], but the nature of the transferred material, mechanisms of transfer, effects on target cells and effects of the tissue microenvironment have yet to be fully defined. In this study, we demonstrate that human MSCs isolated from bone marrow, parenchymal lung tissue and BAL fluid donate cytoplasmic contents and mitochondria to bronchial epithelial cells. Transfer was rapid, uni-directional and mediated by multiple mechanisms including microvesicles, microtubules/TNTs and non-connexin-43 gap junctions. Similar mechanisms contributed to mitochondrial donation in BM-MSCs and BAL-MSCs, whereas microvesicles and microtubules/TNTs, but not gap junctions, contributed to mitochondrial transfer by LT-MSCs. Inhibiting these pathways did not completely prevent mitochondrial donation, however, suggesting the existence of additional mechanisms. These findings add to a growing body of literature which demonstrates MSCs isolated from different sources are similar but exhibit subtle differences which may be reflective of adaption to a specific tissue. Our data provide further support to the idea that human lungderived mesenchymal populations contribute to tissue were cultured with either dynasore to inhibit endocytosis, cytochalasin D to inhibit microtubule/nanotube formation, carbenoxolone to inhibit all gap junction formation or GAP26 to inhibit connexin-43 gap junction formation. Y axis represents this change in mean fluorescent intensity (MFI, %) relative to a paired, uninhibited control. Horizontal line represents median. Groups were compared using the Mann-Whitney U test. *p < 0.05 compared with baseline homeostasis and repair through direct interaction with epithelial cell populations. BAL-MSCs are a unique population of transplant donor-derived cells that are generally only recoverable from lung allografts [29,30]. Their absence from healthy individuals and correlation with disease has led to the suggestion that BAL-MSCs are either a causal factor for lung disease [44,45] or an epiphenomenon. We have previously postulated that BAL-MSCs are derived from an in-tissue source which has mobilized into the alveolar space to provide support for injured epithelial cells [30]. This is supported by previous studies demonstrating that BAL-MSCs administered to mice preferentially engraft in close proximity to alveolar progenitor cells and donate keratinocyte growth factor via connexin-43 gap junctions [35]. Our findings expand on this by demonstrating that cytoplasmic content is transferred from human LT-MSCs and BAL-MSCs to human bronchial epithelial cells and by identifying a capacity for mitochondrial donation. Unlike previous studies, we were unable to identify a role of connexin-43 in the transfer of cytoplasmic content or mitochondria. These findings provide a foundation for future investigations into the reparative capacity of BAL-MSCs and LT-MSCs.
Although it is now clear that LT-MSCs and BAL-MSCs transfer cytoplasmic material to the lung epithelium, the composition of transferred material remains to be characterized. We hypothesize that LT-MSCs and BAL-MSCs transfer material analogous to that produced by BM-MSCs. While this material remains incompletely defined, it is known to include growth factors [26,46], immuno-modulatory molecules [16,47], transcription factors [12] and micro-RNAs [11,14]. Human BAL-MSCs are thought to transfer keratinocyte growth factor to murine type 2 alveolar epithelial cells, in vivo [35]. Additionally, murine LT-MSCs have been demonstrated to produce fibroblast growth factor-10 and hepatocyte growth factor which drives the differentiation of epithelial progenitor cells into complex airway and alveolar structures, in vitro [2]. Together, these findings support the notion that the role of MSCs within tissue is to provide support, be it by the production of paracrine factors or mitochondrial donation to the surrounding parenchymal populations.
Our data suggest that human BM-MSCs and BAL-MSCs employ a non-connexin-43 gap junction in the transfer of cytoplasmic content and mitochondria to bronchial epithelial cells. This finding was surprising as connexin-43 has previously been demonstrated to mediate both cytoplasmic and mitochondrial transfer from BM-MSCs to alveolar epithelial cells [9]. Together, these findings suggest MSCs utilize gap junctions in a regionspecific manner. To date, no other connexins have been implicated in the transfer of cytoplasmic content or mitochondria. The most logical alternative candidate is connexin-26 which is expressed by basal cells [48]. Notably, basal cells serve as a progenitor cell for the bronchial epithelium [49] and interaction between basal cells and local MSCs would fit with the developing narrative of the pulmonary stem cell niche.
There are several limitations to this study. Firstly, our study used a human bronchial epithelial cell line which may limit the generalisability of our findings to primary cells. This may explain why we were unable to demonstrate a role for connexin-43 in cytoplasmic or mitochondrial donation. However, healthy human lung epithelium is difficult to obtain and requires a heavily supplemented growth medium, which in itself can lead to artefacts. Secondly, the isolation and ex-vivo expansion of MSCs can lead to functional alterations, including their interaction with lung epithelial cells [2]. Therefore, we used MSCs at the lowest passage feasibly possible and passage-matched all samples.

Conclusion
LT-MSCs and BAL-MSCs share a similar capacity for cytoplasmic and mitochondrial transfer to bronchial epithelial cells. Transfer is rapid, ubiquitous and unidirectional from MSCs to the epithelium. These findings offer novel insight into the potential role of lung MSCs in organ homeostasis and repair.

Additional files
Additional file 1: Figure S1. Showing the gating strategy for measuring BEAS2B Calcein AM fluorescence after co-culture. MFI of Calcein AM within