Re-endothelialization of rat lung scaffolds through passive, gravity-driven seeding of segment-specific pulmonary endothelial cells

Abstract

Effective re-endothelialization is critical for the use of decellularized scaffolds for ex vivo lung engineering. Current approaches yield insufficiently re-endothelialized scaffolds that haemorrhage and become thrombogenic upon implantation. Herein, gravity-driven seeding coupled with bioreactor culture facilitated widespread distribution and engraftment of endothelial cells throughout rat lung scaffolds. Initially, human umbilical vein endothelial cells were seeded into the pulmonary artery by either gravity-driven, variable flow perfusion seeding or pump-driven, pulsatile flow perfusion seeding. Gravity seeding evenly distributed cells and supported cell survival and re-lining of the vascular walls while perfusion pump-driven seeding led to increased cell fragmentation and death. Using gravity seeding, rat pulmonary artery endothelial cells and rat pulmonary vein endothelial cells attached in intermediate and large vessels, while rat pulmonary microvascular endothelial cells deposited mostly in microvessels. Combination seeding of these cells led to positive vascular endothelial cadherin staining. In addition, combination seeding improved barrier function as assessed by serum albumin extravasation; however, leakage was observed in the distal portions of the re-endothelialized tissue suggesting that recellularization of the alveoli is necessary to complete barrier function of the capillary–alveolar network. Overall, these data indicate that vascular recellularization of rat lung scaffolds is achieved through gravity seeding. Copyright © 2016 John Wiley & Sons, Ltd.

1. Introduction

Tissue engineering of lungs provides an alternative for overcoming the shortage of organs available for transplantation and the complications associated with the long-term performance of grafts. Natural, three-dimensional, acellular, extracellular matrix (ECM) organ scaffolds derived through decellularization are ideal for tissue engineering because they contain organ-specific vascular and parenchymal architecture (Badylak et al., 2011; Cortiella et al., 2010; Ott et al., 2010, 2008; Petersen et al., 2010; Price et al., 2010; Soto-Gutierrez et al., 2011; Uygun et al., 2010; Wagner et al., 2013). Although reconstituting a functional vasculature prior to implantation has been shown to be essential (Robertson et al., 2014), complete re-endothelialization of decellularized lungs has not yet been demonstrated.

Re-endothelialization of rodent lung scaffolds has been attempted through the seeding of human umbilical vein endothelial cells (HUVECs) or pulmonary microvascular endothelial cells (MVECs) by either gravity-driven or pump-driven seeding methods (Ott et al., 2010; Petersen et al., 2010). Following orthotopic transplantation, perfusion of these scaffolds resulted in the formation of barriers with some degree of function but limited by the presence of microthrombi and haemorrhage of blood into the airspace due to inadequate revascularization (Ott et al., 2010; Petersen et al., 2010; Song et al., 2011). Thus, the re-establishment of the vasculature is critical for generating functional, transplantable, tissue-engineered lungs.

Efforts to optimize various procedural parameters for re-endothelialization of lung scaffolds are currently underway. Ren et al. (2015) reported that gravity-driven seeding of endothelial cells into the pulmonary artery (PA) and pulmonary veins (PVs) led to a more homogeneous distribution and better vascular coverage than seeding into only the PA. Furthermore, coseeding of endothelial cells with perivascular supporting cells led to ~75% endothelial coverage of the vasculature of rat lung scaffolds relative to native lungs (Ren et al., 2015). In a recent report, positioning a rat lung scaffold in a supine position rather than in an upright position was shown to allow better distribution and retention of MVECs seeded by continuous perfusion with a syringe pump (Stabler et al., 2016). To build on these steps toward a preferred re-endothelialization strategy, gravity-driven seeding was compared to pump-driven seeding. In addition, coseeding of segment-specific pulmonary endothelial cells was evaluated to determine whether endothelial specification is necessary for complete re-endothelialization. The data herein demonstrate that variable flow, gravity-driven seeding promotes better deposition of endothelial cells throughout the vascular tree than pulsatile flow, perfusion pump-driven seeding. Moreover, the data suggest that a preferred re-endothelialization strategy will likely be account to for segment-specific cells, step-wise cell infusion, and the fluid mechanics of the seeding and culture system.

2. Materials and methods

2.1. Experimental design

The goal was to derive a preferred re-endothelialization strategy for decellularized rat lung scaffolds that would promote widespread, even distribution of cells throughout the scaffold, a healthy cell morphology and the best possible functional outcomes. For endothelial cell seeding, we hypothesized that a variable flow, gravity-driven seeding method (as described by Ott et al., 2010) would be superior to a pulsatile flow, perfusion pump-driven seeding method (as described by Petersen et al., 2010). For gravity-driven seeding, hydrostatic pressure was generated by suspending a column of liquid 20 cm above the lung scaffold. For pulsatile flow seeding, a perfusion pump set to a flow rate of either 1 ml/min or 3 ml/min drove seeding within a bioreactor. Seeded lungs were statically incubated for 0, 1 or 24 h and then cultured in a bioreactor for 3 days with recirculation of media at a flow rate of either 1 or 3 ml/min. For lungs seeded by the perfusion pump-driven seeding method, the flow rate used for bioreactor culture was the same as the flow rate used for seeding. After determining a preferred recellularization method using HUVECs, this strategy was evaluated with segment-specific rat pulmonary endothelial cells. Herein, the term re-endothelialization was reserved for data in which the following were achieved: (1) adhesion of endothelial cells to the underlying vascular ECM, (2) spreading of cells along the wall of blood vessels, and (3) maturation of the endothelial monolayer by demonstrating the formation of tight junctions. The term recellularization was used more broadly.

2.2. Animals

All studies were approved by the Institutional Animal Care and Use Committee (IACUC) at Tulane University. Lung scaffolds were generated from male Sprague–Dawley rats weighing 250–350 g (Charles River Laboratories, Portage, MI, USA).

2.3. Extraction and decellularization of lungs

Rats were anaesthetized and exsanguinated, and lungs were extracted as previously described (Scarritt et al., 2014). Lungs were decellularized using a procedure originally described by Price et al. (2010) with modifications described by Scarritt et al. (2014). To facilitate bioreactor culture and to prevent contamination, additional modifications were made to the decellularization procedure including: cannulation of the apex of the left atrium to access the PVs for seeding; inclusion of 500 U/ml penicillin, 500 μg/ml streptomycin, and 1.25 μg/ml amphotericin B (anti-anti; Thermo Scientific Thermo Fisher Scientific, Waltham, MA, USA) in all decellularization solutions except sodium deoxycholate; sterile filtration of all decellularization solutions using 0.2 μm vacuum-driven filter units; and execution of the last 2 days of the decellularization procedure using sterile equipment in a tissue culture hood.

2.4. Cells and cell culture

HUVECs were chosen for initial optimization due to the promising success reported by Ott and colleagues when using this cell type in lung and kidney vascular recellularization (Ott et al., 2010; Song et al., 2013). Cryopreserved, normal HUVECs isolated from a single donor were purchased from Lonza (Allendale, NJ, USA). HUVECs were cultured in endothelial growth media 2 (EGM-2) composed of endothelial basal media 2 (500 ml) supplemented with EGM-2 SingleQuots (10 ml fetal bovine serum, 0.2 ml hydrocortisone, 2 ml human fibroblastic growth factor-B, 0.5 ml vascular endothelial growth factor, 0.5 ml R3-insulin-like growth factor-1, 0.5 ml ascorbic acid, 0.5 ml human epidermal growth factor, 0.5 ml gentamycin-1000, and 0.5 ml heparin).

MVECs, PA endothelial cells (PAECs) and PV endothelial cells (PVECs) were graciously provided by Dr Diego Alvarez at the University of South Alabama (Mobile, AL, USA) who had previously characterized these cells (Alvarez et al., 2008; King et al., 2004). In brief, PAECs and PVECs were obtained by scraping the tunica intima of conduit vessels (arteries for PAECs and veins for PVECs). MVECs were obtained via a pleural cut technique. Cells were transferred to cell culture-treated plastic-ware and manually selected by morphological homogeneity. The cells were incubated with a modified D-valine media enriched with endothelial growth factors to promote the death of nonendothelial cells. Fluorescence-activated cell sorting of stained cells was used to confirm expression of canonical endothelial markers including von Willebrand factor, CD144, CD31 and endothelial nitric oxide synthase as well as the segment-specific lectins Glycine max, Griffonia simplicifolia I, Helix pomatia, Sambucas nigra I and Ulex europeaus. Fluorescence-activated cell sorting was also used to confirm that the cells did not express mesenchymal/fibroblast marker Thy-1. Highly pure colonies of MVECs, PAECs and PVECs were expanded in Dulbecco’s modified Eagle medium (DMEM; Life Technologies, Grand Island, NY, USA) containing 10% fetal bovine serum (Atlanta Biologicals, Flowery Branch, GA, USA) and 100 U/ml penicillin, 100 mg/ml streptomycin, and 0.25 μg/ml amphotericin B (complete culture medium, CCM). Cells were grown on 15 cm tissue culture treated plates (Nunc, Waltham, MA, USA). Media was changed every other day and cells were cultured at 37 °C and 5% CO₂ in a tissue culture incubator.

2.5. Preparation of cells and lung scaffolds

Prior to seeding, decellularized lungs were washed with 50 ml of sterile PBS containing anti/anti through the PA cannula and tracheal cannula, and then washed with 50 ml of the appropriate cell culture media. The lungs were immersed in media and warmed in a water bath to 37 °C until just before cells were seeded.

Passage 6 HUVEC cultures at 70% confluence were harvested using 0.25% trypsin, pelleted by centrifugation at 300xg for 7 min, and resuspended in EGM-2 at a density of 500× 10⁶ cells/l (density was chosen based on Calle et al., 2011). For each cell seeding experiment utilizing HUVECs, approximately 11.36 ± 0.33 × 10⁶ cells were delivered.

MVECs, PAECs, and PVECs at 70–100% confluence were harvested with 0.05% trypsin, pelleted by centrifugation, and resuspended to a density of 500 × 10⁶ cells/l in CCM. MVECs were used at passage 18–19, PAECs at passage 12–14, and PVECs at passage 17–18. For seeding of MVECs only, 13.29 ± 1.97 × 10⁶ cells were instilled via both the PA and PVs sequentially for a total of 26.58 ± 3.94 × 10⁶ cells per lung. For seeding of PAECs only, 12.54 ± 4.87 × 10⁶ cells were seeded into the PA. For seeding of PVECs only, 11.06 ± 5.33 × 10⁶ cells were seeding through the PVs. For combination seeding of MVECs, PAECs, and PVECs, three methods were evaluated (described in section 2.6 below). For Sequential 1 day (Seq1d), 24.88 ± 3.16 × 10⁶ MVECs, 15.33 ± 1.34 × 10⁶ PAECs and 17.85 ± 4.24 × 10⁶ PVECs ere seeded. For Sequential 2 day (Seq2d), 24.75 ± 2.15 × 10⁶ MVECs, 16.92 ± 3.83 × 10⁶ PAECs, and 11.62 ± 3.19 × 10⁶ PVECs were seeded. For Mixed, 22.72 ± 3.48 × 10⁶ MVECs, 18.49 ± 2.17 × 10⁶ PAECs, and 16.95 ± 2.57 × 10⁶ PVECs were seeded.

Prior to seeding, all cell suspensions were passed through a 40 μm cell strainer to remove any cell clumps that could inhibit seeding by occluding vessels. Cells were seeded using either a gravity-driven apparatus or perfusion pump-driven apparatus (Figure S1A, B).

2.6. Gravity seeding

The lungs were seeded in a supine position. Gravity seeding was accomplished using a gravity-driven apparatus (a 10 ml syringe suspended 20 cm above the lung using a ring stand; Figure S1A). Cell suspension was transferred to the syringe and delivered to the vasculature of the lungs passively by gravity-driven hydrostatic pressure. The total seeding volume was approximately 20 ml and, because the seeding reservoir was a 10 ml syringe, the syringe was refilled mid-way through seeding. Seeding time ranged from 5 to 30 min depending on the seeding route and cell number. In general, seeding cells into the PV was quicker than seeding into the PA, and seeding a greater cell number required more time. After the cell suspension was delivered, an additional 5 ml of media was instilled into the vasculature.

For initial optimization experiments, HUVEC cells were seeded into the PA cannula via this procedure. After determining a preferred seeding method, various seeding iterations were introduced. For seeding HUVECs into the pulmonary venous system, the cells were delivered to the PV via the cannulated left atrium. For dual seeding of the arterial and venous systems, HUVECs were delivered first to the PA and then to the PV.

For seeding of PAECs, cells were seeded into the PA only. For seeding of PVECs, cells were seeded into the PV only. For MVEC seeding, cells were seeded into the PA first and then immediately into the PV.

For combination seeding, three methods were assessed: (1) Seq1d: MVECs were seeded into both routes, then PAECs were seeded into the PA, and finally PVECs were seeded into the PV; (2) Seq2d: MVECs were seeded into both routes; on the following day, PAECs were seeded into the PA, then PVECs were seeded into the PV; and (3) Mixed: MVECs, PAECs and PVECs were mixed together into the same cell suspension and seeded into the PA and then the PV.

2.7. Pulsatile flow, perfusion pump-driven seeding (perfusion seeding)

Perfusion seeding was accomplished using a peristaltic perfusion pump (Cole Parmer, Vernon Hills, IL, USA) in series with a bioreactor. The bioreactor was built with modifications to a previously described design (Calle et al., 2011). Briefly, the cell suspension was transferred to a vascular seeding reservoir (Figure S1B). The lung was submerged in cell culture media during seeding. Cells were delivered from the seeding reservoir to the vasculature of the lung via the perfusion pump at a flow rate of either 1 ml/min or 3 ml/min. These flow rates were chosen based on the work of Petersen et al. (2010) who perfused the vasculature of seeded rat lungs at 1–5 ml/min, Calle et al. (2011), who recommend endothelial cell seeding at 3 ml/min, and Robertson et al. (2014) who utilized flow rates of 1–3 ml/min for seeding of rat heart scaffolds. The pressure amplitude and frequency differed for pulsatile perfusion at 1 ml/min vs. 3 ml/min. The 1 ml/min seeding strategy had an amplitude of 2–6 mmHg and a frequency of approximately 3.7 s. The 3 ml/min seeding strategy had an amplitude of 5–13 mmHg and a frequency of approximately 1.2 s.

2.8. Bioreactor design

The bioreactor was based on the design described by Calle et al. (2011) with slight modifications to the size of the tubing and fittings to reduce turbulence during perfusion. L/S14 tubing, which has an inner diameter of 1.6 mm, was used throughout the bioreactor. To compliment this, all fittings (luer connectors, y-connectors, etc.) were of a uniform inner diameter of ~1.6 mm. Additionally, rather than using a y-connector for cannulation of the PA (as described by Calle et al., 2011), a straight connector was used. Although the Calle et al. (2011) bioreactor includes an airway seeding/culture reservoir, this aspect was not required for vascular seeding and culture; therefore, the bioreactor chamber was built to accommodate connection to an airway reservoir for future work, but an airway reservoir was not used.

2.9. Bioreactor-based whole organ culture

Approximately 200 ml of media was added to the bioreactor and the lungs were installed via connection of the PA cannula to the vascular loop (Figure S1C). Once the bioreactor was sealed, media was pulled into the vascular loop via the injection port using a syringe fitted with an 18-gauge needle. All air bubbles were removed from the vascular loop via this injection port. The bioreactor was transferred to a tissue culture incubator containing a perfusion pump. The vascular loop was installed into the perfusion pump and media was circulated from the bioreactor chamber to the vasculature of the lung at a rate of either 1 ml/min or 3 ml/min (Calle et al., 2011; Petersen et al., 2010; Robertson et al., 2014). Although the physiological arterial flow for a rat lung is upwards of 80 ml/min, previous reports in which native rat lungs were cultured in a bioreactor indicated that - without ventilation - high flow leads to arterial hypertension (Petersen et al., 2011). In addition, for bioreactor-based lung tissue engineering ex vivo, the lung does not need to oxygenate the entire blood supply nor provide nutrients to the entire body (Petersen et al., 2011).

For 3-day cultures, a half-media (100 ml) change was conducted 24 h and 48 h after seeding by removing and adding media via the injection port in the bioreactor chamber. Any air bubbles in the vascular loop were removed from the tubing during media changes. For 7- day cultures, half-media changes were conducted at 24, 48, 96 and 120 h, and full media changes were conducted at 72 and 144 h. As a control for assessing the necessity of perfusion culture for nutrient delivery to seeded cells, lungs were installed in the bioreactor but were not perfusion fed (static culture).

2.10. Cell viability and cell growth

To evaluate the effect of seeding method on cell viability and growth kinetics, HUVECs that were not retained in the vasculature during seeding were collected from the perfusate (media outflow from the PV). Outflow cells were counted with a haemocytometer and analysed by flow cytometry to determine cell viability.

Cell viability was determined using an Annexin V-FITC Apoptosis Detection Kit according to the manufacturer’s protocol (Sigma St. Louis, MO, USA). Briefly, cells were suspended in 1x binding buffer at 1 × 10⁶ cells/ml. Next, 5 μL of Annexin V-FITC (AV) conjugate and 10 μL of propidium iodide (PI) were added to the cell suspension and incubated at room temperature, protected from light, for 10 min. Fluorescence was determined using a flow cytometer. Viable cells were considered PI⁻AV⁻ while necrotic cells were considered PI⁺AV⁻. PI⁻AV⁺ staining denoted early apoptosis while PI⁺AV⁺ staining denoted late apoptosis (Vermes et al., 2000, 1995).

Alternatively, outflow cells were collected after seeding scaffolds and were plated at 50,000 viable cells per well of a 6-well plate for analysis of growth kinetics and cell viability after 24, 48 or 72 h. This was conducted to determine if mechanical damage caused during seeding had long-lasting effects on cell viability. Media was collected from each well and cells were harvested using 0.25% trypsin for 3 min. Trypsin was inactivated with an equal volume of media. Trypsin-media-cell solution was added to the media collected from each well (so that dead/unattached cells were included in the analysis). Cells were counted with a haemocytometer and then analysed by flow cytometry as described above. Media was not changed during the (up to) 72 h culture period so that dead cells were included in the analysis; however, additional media was added to wells at 48 h to ensure that attached cells had adequate nutrition. All experiments were done in triplicate. Nonseeded HUVECs harvested from the same plates as seeded HUVECs were used as controls.

2.11. Pressure measurements

Pressure transducers (Harvard Apparatus, Holliston, MA, USA) coupled with BIOS software (Harvard Apparatus) recorded pressure during seeding and culture. The pressure transducers were placed in series with the seeding or bioreactor culture apparatus immediately prior to the PA cannula. During gravity seeding, a syringe acted as a reservoir for the cell suspension and was suspended 20 cm above the lung (i.e. 20 cmH₂O). The outflow from the PV was at 0 cmH₂O. Perfusion seeding occurred within a bioreactor where the lungs were submerged in media approximately 2–3 cm above the bottom of the main chamber with inflow to the PA and outflow from the PV. The bioreactor was sealed at atmospheric pressure and was fitted with an air filter to allow atmospheric pressure to be maintained during perfusion seeding and culture.

The pressure tracings from gravity-seeded tissues were used to determine the approximate flow rates during gravity seeding. The initial flow rate was determined as the time required to seed 10 ml of cell suspension (half of the total seeding amount). The flow rate for the last half of seeding was calculated as the time required to seed the remaining 10 ml of cell suspension.

2.12. Histological characterization and scoring

Lungs were inflation-fixed as previously described (Scarritt et al., 2014). Three transverse sections of the left lung lobe were used for histological analyses. One of the sections was taken at the hilum and the other sections were taken proximal and distal to the hilum.

Lung sections were embedded in paraffin, cut into 5 μm sections, and mounted onto Fisherbrand glass microscope slides. Sections were stained with haematoxylin and eosin (H&E) by the Center for Stem Cell Research and Regenerative Medicine Histology Core Facility using standard protocols. Stained slides were scanned using the Aperio ScanScope (Aperio, Vista, CA, USA) at an initial magnification of 40×. Images were then captured and analysed using the Aperio ImageScope program (available through Leica Biosystems, Buffalo Grove, IL, USA).

Histological scoring of cell morphology, distribution, and recellularization were made according to Table S1. This scoring system was developed based on the criteria that the preferred method for vascular recellularization would strike a balance between distributing cells throughout the tissue, reducing the physical damage cells experience due to perfusion pressure and shearing forces, and influencing cell attachment and organization. Scoring was performed in triplicate for each experimental condition by three blinded observers.

2.13. Cell count and nuclear size determination using ImageScope

Aperio’s ImageScope software default macro Nuclear v9 was colour calibrated for rat lung scaffolds seeded with endothelial cells and stained with H&E (Olson, 2006). The calibrated macro was then optimized to create a custom algorithm that consistently recognized cell nuclei without falsely identifying artefacts and debris. This custom algorithm was designed under the advisement of Leica personnel who provided on-site training on Aperio’s ImageScope program. H&E-stained native lung tissue sections were used to confirm the algorithm’s parameters. All cells were counted; attached and unattached cells were not discriminated based on this algorithm. The adjusted parameters included a minimum nuclear size of 10 μm², a minimum roundness and elongation of 0.1 (a measure of the ratio of length to width), a weak (+1) threshold of 210, a moderate (+2) threshold of 200, and a strong (+3) threshold of 188. The entire tissue section was analyzed using the algorithm and the number of +3 positive cells as well as the average nuclear size (μm²) were recorded. Each experimental condition was assessed using biological triplicates.

2.14. Ki67 and TUNEL staining

Tissue sections were stained with an anti-Ki67 antibody (Catalog #ab15580, Abcam, Cambridge, MA, USA) for assessing proliferation or with an In Situ Cell Death Detection Kit with Fluorescein (Catalog #11684795910, Roche, Indianapolis, IN, USA) for terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) as previously described (Scarritt et al., 2014).

Images were captured using the Leica DMRXA 2 deconvolution inverted fluorescence microscope (Leica Microsystems, Buffalo Grove, IL, USA) fitted with the Cooke SensiCAM camera/controller and Slidebook software. Each tissue was analysed using three tissue sections taken from the left lung lobe (proximal, medial, and distal portions). For each section, 15 pictures were taken at 10× magnification. For each picture, the total number of 4ʹ, 6-diamidino-2-phenylindole (DAPI)-stained cells and the number of Ki-67 or TUNEL stained cells were counted to determine the overall percentage of cells proliferating or undergoing apoptosis, respectively. Each recellularization strategy was analysed with three biological replicates, and each biological replicate was calculated using the average of three technical replicates.

2.15. Vascular endothelial-cadherin staining

Staining was performed on 5-μm lung sections using standard protocols. Briefly, slides were deparaffinized, incubated in Sub-X solvent, and rehydrated. Antigen retrieval was performed enzymatically by incubating tissue sections for 15 min at 37 °C with proteinase K (400 μg/ml in 20 mM Tris pH 8.0) in a humidified chamber. Slides were blocked with Seablock Blocking Buffer (Thermo Fisher Scientific Waltham, MA, USA) for 2 h at room temperature. The slides were stained with rabbit anti-vascular endothelial (VE)-cadherin (Thermo Fisher Scientific; Cat. #36–1900) diluted 1:50 in Tris-buffered saline (TBS) overnight at 4 °C in a humidified chamber. After washing the slides in TBS containing 0.05% Tween-20, the tissue sections were incubated for 1 h at room temperature in a humidified chamber protected from light with AlexaFluor 594 goat anti-rabbit IgG (H+L) (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) diluted 1:200 in TBS containing 1% bovine serum albumin (BSA). After washing with TBS, slides were mounted with Prolong Gold Antifade Reagent with DAPI (Invitrogen) and coverslipped. Images were captured using the Leica DMRXA 2 deconvolution inverted fluorescence microscope (Leica Microsystems, Buffalo Grove, IL, USA) fitted with the Cooke SensiCAM camera/controller and Slidebook software.

2.16. Cell labelling

When seeding via multiple routes, cell labelling was employed. Vybrant Dil and DiO lipophilic carbocyanine dyes from Life Technologies, Thermo Fisher Scientific, Waltham, MA, USA were used according to the manufacturer’s instructions. Briefly, cells were harvested, pelleted at 300xg for 7 min, and re-suspended to 1 × 10⁶ cells/ml in prewarmed (37 °C) serum-free culture medium. Protected from direct light, 5 μL of Vybrant Dil or DiO cell-labelling solution was added per ml of cell suspension and incubated for 20 min at 37 °C in a tissue culture incubator. After incubation, the cells were pelleted at 300xg for 5 min. The cells were gently re-suspended in warm media (containing serum). This wash procedure was repeated twice more before seeding labelled cells into a lung scaffold. Cells were visualized immediately after seeding, as well as 72 h after seeding, using an ImageQuant LAS4000 (GE Healthcare Life Sciences, Pittsburgh, PA, USA). After bioreactor culture, slices of lung tissue (~1 cm thick) were submerged for 5 min in three changes of Tissue-Tek OCT (optimal cutting temperature) compound to allow complete penetration of OCT into the tissue. Lung slices were embedded in fresh OCT and rapidly frozen in 2- methylbutane cooled on dry ice. Tissue sections (5 μm) were cut on a cryotome, mounted and coverslipped with Prolong Gold Antifade Reagent containing DAPI (Invitrogen Thermo Fisher Scientific, Waltham, MA, USA) before photographing on a Nikon Eclipse E800 fitted with a Qlmaging Retiga 2000R camera and Slidebook software.

2.17. Vessel counts

The average size of blood vessels reseeded with cells was determined based on a technique initially described by Robertson et al. (2014). For seeding of scaffolds with PVECs, PAECs or MVECs, H&E-stained tissue sections were scanned with an Aperio ScanScope at an initial magnification of 40×. Ten random images were then captured at 10× magnification, and the number of cells within each vessel, as well as the short axial diameter of the vessel, was recorded. Three biological replicates were assessed per cell type. In order to better represent the distribution of cells to large blood vessels, which are less frequent than small blood vessels, data is presented herein as the percentage of cells contained within vessels of each size. Data are also supplied as number of cells per mm² tissue.

2.18. Evans blue barrier function assay

To assess vascular competency after recellularization, BSA was bound to Evans blue for perfusion through the vasculature. Extravasation of the BSA-Evans blue from the vasculature was measured as a means of determining endothelial barrier function since BSA should be retained within a competent vasculature (Alvarez et al., 2016; Lowe et al., 2010).

To generate BSA-Evans blue, 12 mg of Evans blue was mixed thoroughly with 12 ml of 5% BSA and filtered through a 0.2 μm nylon syringe filter unit. To remove any unbound Evans blue, the mixture was transferred into a prehydrated 3500 MWCO Slide-A-Lyzer Dialysis Cassette (Thermo Scientific, Waltham, MA, USA). Suspended by a Slide-A-Lyzer Buoy, the cassette was immersed in 5% BSA in dH₂O. The solution was stirred overnight at 4 °C to facilitate dialysis of unbound Evans blue from the solution in the cassette to the solution outside of the cassette. The next day, the BSA-Evans blue solution was removed from the cassette and stored shortterm at 4 °C.

For perfusion of freshly isolated native, decellularized or recellularized rat lungs, BSA-Evans blue solution was diluted 1:20 with prewarmed, serum-free DMEM and perfused through the tissue for 15 min at a rate of 1 ml/min. During perfusion, the trachea was held above the lung to prevent drainage of extravasated perfusate from the airways. The circulation was then rinsed with Earle’s balanced salt solution containing 4% BSA for 5 min at a flow rate of 1 ml/min. The lung was blot and air dried for 1 h and then weighed. The tissue was minced and incubated in 1 ml of 37% formamide (#F8775; Sigma Aldrich, St. Louis, MO, USA) for 48 h at 54 °C. The next day, tissue debris was pelleted at 14,000 × g for 30 min. The supernatant was removed to measure the retention of Evans blue dye (i.e. BSA-Evans blue that had escaped from the vasculature into the airspace). As an index of leakage, the optical density of the supernatant was measured at 620 nm. A standard curve of BSA-Evans blue diluted with formamide was used to determine the amount of Evans blue that extravasated per mg tissue.

2.19. Statistical analyses

Cell viability and cell growth analyses for comparing seeding methods were assessed using two-way analysis of variance (ANOVA) with Bonferroni posthoc tests between groups. Histological scoring, cell counts, nuclear size, Ki-67 staining, and TUNEL staining were analysed for statistical significance using three two-way ANOVAS to analyse the interaction of each set of variables with a Bonferroni posthoc test between groups. Individual experiments were compared using a Student t test. A p- value of < 0.05 (indicated by *) WAS considered significant; p-values of < 0.01 and p < 0.001 are denoted with two and three asterisks, respectively, to indicate high significance. Where applicable, posthoc test p-values are denoted with #, ##, ### for significance of <0.05, < 0.01, and <0.001, respectively. NS was used to denote that there was no significant difference. Values are reported as mean ± standard error of the mean (SEM); n ≥ 3 for all experiments.

3. Results

3.1. Seeding method affects arterial pressure but not initial cell retention

Sprague–Dawley rat lungs were decellularized and HUVECs were used for vascular seeding. HUVECs were instilled via the PA by either: (1) variable flow gravity seeding under hydrostatic pressure (Figure S1A); or (2) perfusion seeding with a pulsatile perfusion pump set to a flow rate of either: (a) 1 ml/min; or (b) 3 ml/min (Figure S1B). To determine initial cell retention, HUVECs were collected from the effluent of the PVs and counted. All three methods achieved >87% cell retention (Figure 1A).

Figure 1.

Seeding method does not affect initial cell retention but does influence pressure, cell growth and cell death. (A) Cells that were not retained in the scaffold after seeding were collected from the effluent of the PVs and counted. By one-way ANOVA, seeding method did not affect initial cell retention. (B) Pressure was measured at the PA during seeding. Perfusion seeding at 1 ml/min generated significantly lower pressures than gravity seeding or perfusion seeding at 3 ml/min. Unattached cells from the PV effluent were compared to nonseeded cells (control) for (C) cell growth and (D–G) cell viability/death immediately after seeding or 24, 48 and 72 h after plating. Perfusion seeding at 3 ml/min led to significant increases in early and late apoptosis. Two-way ANOVAwas used to determine statistical significance withp-values (*) depicted horizontally for time and vertically for seeding method. The p-values for the interaction of time and seeding method (not shown) were not significant across the board. The p-values from Bonferroni’s posthoc analysis (#) for (C) compare perfusion seeding at 1 ml/min to control cells, and for (F) compare perfusion seeding at 3 ml/min to control cells. Data are means ± SEM, n ≥ 3.

Pressure was measured at the PA during seeding. Gravity seeding caused an immediate increase in pressure to a maximum of 13.31 ± 0.34 mmHg followed by a gradual decrease and plateau at 8.86 ± 0.62 mmHg (Figure S2A). During gravity seeding, the cell suspension was transferred to a 10 ml reservoir suspended 20 cm above the lung. Because the volume of the cell suspension exceeded the volume of the seeding reservoir, the reservoir was refilled at least once during seeding (Figure S2A, black arrows). During perfusion seeding at 1 ml/min, the pressure gradually increased to a maximum of 4.33 ± 1.52 mmHg (Figure S2B) and displayed an oscillating pattern consistent with the pulsatile perfusion pump (Figure S2B, inset). Seeding at 3 ml/min caused a steeper increase to 18.85 ± 4.12 mmHg (Figure S2C) with occasional pressure spikes (Figure S2D). Peak perfusion pressure at 3 ml/min was significantly higher than perfusion seeding at 1 ml/min or gravity seeding (Figure 1B). Peak pressure during gravity seeding was significantly higher than perfusion seeding at 1 ml/min.

While flow rate was set to 1 ml/min or 3 ml/min for perfusion seeding, flow rate during gravity seeding changed over time. The flow rate was rapid (19.97 ± 2.81 ml/min) when gravity seeding was initiated, but toward the end of seeding after cells had been deposited, the flow rate decreased to 1.70 ± 0.38 ml/min (Figure S2E). Overall, this suggests that gravity seeding allows variable flow, which prevents the seeding pressure from exceeding 15 mmHg (~20 cmH₂O, which is equivalent to the column height used for gravity seeding).

3.2. Seeding method influences the growth kinetics and survival of cells that have passed through the lung vasculature

After seeding, cells were collected from the outflow of the PVs to determine the effect of seeding method on effluent cell fate (Figure 1C). Seeding method significantly affected the viability and growth rate of cells that had passed through the vasculature. There was a significant decrease in cell growth after perfusion seeding, while cells seeded by gravity grew similarly to control cells.

The percentage of viable cells and necrotic cells did not differ between seeding methods (Figure 1D, E). The frequency of early apoptosis increased slightly immediately after seeding regardless of the seeding method used; however, when seeding by perfusion at 3 ml/min, a considerable increase in apoptosis—up to 9 ± 4%—was apparent after 24 h and was sustained at 48 h (Figure 1 F). Perfusion seeding at 3 ml/min also led to substantially higher percentages of cells in late apoptosis at 0 h (13 ± 10%) and 24 h (13 ± 2%; Figure 1G). These data suggest that perfusion seeding at 3 ml/min generated damaging stimuli that increased cell death and led to an overall impairment of growth. Therefore, gravity seeding was predicted to be a preferred cell delivery method for vascular recellularization of rat lung scaffolds.

3.3. Better cell distribution, morphology and recellularization is achieved following gravity seeding

To verify that gravity seeding would lead to better recellularization outcomes than perfusion seeding, bioreactor culture was investigated. After seeding, lungs were statically incubated for 0, 1, or 24 h to permit cell attachment prior to initiating antegrade media perfusion in a bioreactor at either 1 ml/min or 3 ml/min via a pulsatile perfusion pump (Figure S1C). Each of the 12 experimental conditions (see Table S2) was assessed using H&E staining (Figure 2). H&E sections were scored based on cell distribution, morphology and recellularization (relining of blood vessel walls vs. aggregating in the lumen of vessels; Figure 3A–C). Scores were from 0 to 5 with 0 indicating poor outcomes and 5 indicating excellent outcomes (see Table S1 for scoring parameters).

Figure 2.

HUVECs appear healthier in gravity-seeded lungs than in perfusion-seeded lungs after 3 days in bioreactor culture. Lung scaffolds were seeded with HUVECs via the PA using either variable flow, gravity seeding or pulsatile flow seeding with a perfusion pump. After seeding, lung scaffolds were statically incubated for 0, 1 or 24 h prior to circulating media through the lung vasculature in a bioreactor for 3 days at a flow rate of either 1 ml/min or 3 ml/min (see Table S2 for tabular representation of conditions). H&E staining of each experimental condition revealed healthy cells in gravity-seeded lungs and fragmented cells (black arrows) in perfusion-seeded lungs. A 24-h incubation led to cell fragmentation and cells clumped in the lumen. Representative images from one of three experiments are shown.

Figure 3.

Gravity seeding leads to better cell distribution, morphology, and recellularization scores than perfusion seeding. H&E-stained sections from three replicate experiments were evaluated using a histological scoring system from 0 to 5 (poor to excellent) grading cell distribution, cell morphology, and recellularization (see Table S1 for grading criteria). Two-way ANOVA was used to determine significance. (A) Gravity seeding led to better cell distribution throughout the vasculature of lung scaffolds than perfusion seeding. (B) Recellularization was defined as re-lining of the vascular walls with minimal cell aggregation in the lumen or leakage into the airspace. Tissues that were incubated for 24 h led to cell aggregation and very low scores for each condition. Gravity seeding led to better recellularization than perfusion seeding. (C) Cell morphology was best in gravity-seeded tissues, while perfusion seeding frequently led to cell fragmentation. Asterisks (*) indicate results of two-way ANOVAs while # indicates results of Bonferroni’s posthoc test. p < 0.05 was considered significant and values are reported as mean ± SEM, n =3.

Gravity seeding distributed HUVECs throughout the lung vasculature (Figure 3A; S3A), while perfusion seeding often led to an uneven distribution with some areas containing no cells, particularly in the distal portions of the lung (Figure S3B). This disparity was especially evident for perfusion at 1 ml/min, which had the lowest distribution scores (Figure 3A) suggesting that the pressure generated via this method insufficiently distributed cells into the smaller, distal vasculature. Indeed, by two-way ANOVA, seeding method significantly impacted distribution scores for scaffolds cultured at 1 ml/min.

HUVECs in gravity-seeded tissues spread across the vascular matrix, relining the walls of small and large blood vessels leading to high recellularization and morphology scores (FigureS 3B, C and S3C–F). Cells in gravity-seeded tissues displayed a healthy cell morphology as denoted by a clearly stained central nucleus with no signs of cell fragmentation. Although some HUVECs appeared along vessel walls in perfusion-seeded scaffolds, many resided within the lumen or appeared fragmented. Moreover, some cells leaked into the airspace, potentially due to cell aggregates blocking the vasculature and leading to a pressure build-up that caused the matrix of the vessel to rupture (Figure S3G, H). Both seeding method and incubation time were found to be significant effectors of recellularization scores. Histological scores for recellularization and morphology were higher in gravity-seeded scaffolds than in perfusion-seeded tissues.

Regardless of the other experimental conditions, a 24-h static incubation period prior to bioreactor culture led to cell clusters in vessel lumens (Figure 2). This resulted in recellularization scores averaging <1 (Figure 3B). Cell fragmentation was also more common in tissues incubated statically for 24 h as evidenced by low morphology scores (Figure 3C).

Control tissues that were in a bioreactor without any media perfusion for 3 days displayed frequent cell fragmentation and poor recellularization (Figure S3I for a gravity-seeded lung and Figure S3J for a perfusion-seeded lung) indicating the importance of using a bioreactor culture system to circulate nutrient-rich media within lung scaffolds.

3.4. Gravity-seeded lungs contain more intact cells than perfusion-seeded lungs after 3 days of bioreactor culture

H&E sections were evaluated using a customized algorithm that recognized and counted haematoxylin-stained cell nuclei. Cell number was significantly higher in gravity-seeded lungs than in perfusion-seeded lungs regardless of the flow rate used for bioreactor culture (Figure 4A). To account for the frequency of fragmented cells, which may artificially increase the cell number, the size of the nuclei was also determined for each experimental condition (Figure 4B). Nuclear size was higher, denoting less cell fragmentation, in gravity-seeded tissues than in perfusion-seeded tissues. For comparison, native rat lung tissues (parenchymal cells included) had an average nuclear size of 39 ± 3 μm².

Figure 4.

Cell count and cell size are highest in gravity-seeded tissues while a 24 h static incubation leads to frequent apoptosis. H&E-stained tissue sections were evaluated with an algorithm from Leica/Aperio’s imaging program ImageScope. Haematoxylin-stained nuclei were recognized by the algorithm and used to determine (A) cell number and (B) average area of the nuclei. (A) Cell count was higher in gravity-seeded tissues than in perfusion-seeded tissues. (B) Average nuclear size was lower in perfusion-seeded tissues than in gravity-seeded tissues suggesting more cell fragmentation in perfusion-seeded lungs. To investigate cell fate after 3 days of bioreactor culture, tissue sections were stained with (C) Ki67 proliferative marker or (D) TUNEL for assessing apoptosis. (C) Ki67 staining was <2% across the board, indicating very low levels of proliferation regardless of the experimental parameters. (D) Gravity-seeded tissues had lower levels of apoptosis than perfusion-seeded tissues. TUNEL staining was highest in tissues that underwent a 24 h static incubation. Asterisks denote results of two-way ANOVA while # denotes results of Bonferroni’s posthoc test. p < 0.05 was significant with values reported as mean ± SEM, n = 3.

Terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) was used to detect cell fragmentation resulting from apoptosis after 72 h of bioreactor culture (see Figure S3K for an example). The percentage of TUNEL-positive cells was higher in perfusion-seeded tissues than in gravity-seeded tissues (Figure 4C). Apoptosis was highest in tissues that underwent a 24 h static incubation confirming histological observations.

To measure proliferation, tissue sections were stained with an antibody for Ki67 proliferative marker (see Figure S3L for an example). The percentage of cells undergoing proliferation after 72 h of bioreactor culture was very low across the board (<2% on average; Figure 4D). Despite high levels of apoptosis in tissues incubated statically for 24 h, the remaining cells were able to proliferate. No differences were seen based on seeding method; however, in gravity-seeded lungs, increasing the bioreactor culture flow rate from 1 ml/min to 3 ml/min significantly decreased the number of Ki67+ cells.

Overall, experiments that employed gravity seeding yielded better outcomes than perfusion seeding. Gravity- seeded lungs with no static incubation prior to bioreactor culture at 1 ml/min had the lowest level of apoptosis (9 ± 3%), the highest cell number (205 ± 19 cells/mm2), the highest average nuclear area (37 ± 3 μm²), and the best cell distribution, morphology, and recellularization scores (4.6 or higher). Thus, the gravity seeding, 0 h, 1 ml/min strategy (condition 1 in Table S2) was determined to be the preferred method for vascular recellularization and was used for all subsequent experiments.

3.5. Gravity-seeded HUVECs persist in bioreactor culture for at least 7 days

After developing a preferred recellularization method, the bioreactor culture period was increased from 3 to 7 days to assess the effect of extended culture time (Figure 5A). Adherent cells persisted in culture for 7 days without significant changes in cell distribution, morphology or recellularization compared to a 3-day culture period (termed preferred method; Figure S4A–C). Extended culture decreased the cell number in comparison to a 3- day culture, but did not significantly affect cell size, proliferation, or apoptosis (Figure S4D–G). The decrease in cell number may indicate that unattached cells were removed or that cells detached from the vessel walls during the extended culture period. Indeed, the average number of viable cells collected from the media during media changes was greater during 7-day culture (996,000 viable cells) than during 3-day culture (197,000 viable cells), although this was not significant (p = 0.1541).

Figure 5.

Culture time canbe extendedwithout detrimentto recellularizationbutretrograde seedingwas not as effective. (A) H&E-stained tissue from seeded lungs cultured for 7 days showed similar histology to 3-day culture. (B) Seeding through the PVs delivered cells to the venous system, but cell aggregation and fragmentation were apparent. (C) Dual seeding of HUVECs through both the pulmonary arteries and veins led to similar histology as seeding through the PA alone. In dual-seeded lungs, HUVECs labelled with DiI or DiO dyes were imaged with an ImageQuant LAS4000 gel imager (D, E) immediately after seeding and (F, G) after 72 h of culture. Arterial (PA) seeding led to deposition of cells throughout the lung including (E) the distal ends. Cells seeded via the veins deposited cells in the proximal area of the lung. (F) After 3 days of bioreactor culture at 1 ml/ min, the lungs retained their initial seeding pattern. (G) Separation of the lung lobes confirmed that each lobe received cells. (H) Microscopy of fluorescently labelled cells seeded via the arteries (red) and veins (green) showed them in separate vascular compartments. In some portions, cells appeared to reside near alveoli, which may suggest a microvasculature/capillary location. Representative photos from one experiment.

3.6. Gravity seeding through the pulmonary arteries and veins

To evaluate recellularization of the venous system, lung scaffolds were gravity-seeded retrograde into the PVs (Figure 5B). Seeding was immediately followed by 3 days of bioreactor culture with perfusion of media antegrade at 1 ml/min. Histological scoring confirmed that venous seeding led to similar cell distribution as arterial seeding; however, there was a decrease in cell morphology and recellularization scores (Figure S4A–C). In addition, venous seeding led to lower cell counts and average nuclear size (Figure S4D, E). A moderate increase in TUNEL staining suggested that seeding via the veins led to more apoptosis than seeding via the arteries (Figure S4G).

Seeding cells through the PA and then additionally through the PVs (dual seeding; Figure 5C) was evaluated based on reports that seeding via both routes increased cell distribution and coverage in rat lung scaffolds (Ren et al., 2015). Cell distribution, morphology and recellularization after dual seeding were similar to seeding via the PA alone (Figure S4A–C). Dual seeding facilitated an increased deposition of cells with an average of 308 ± 58 cells/mm², but this increase was not found to be significantly different than seeding via the PA only, which deposited 205 ± 19 cells/mm² (p = 0.1661; Figure S4D). This probably reflected a lack of enhanced cell engraftment when seeding via the PVs as evidenced by no observed difference in nuclear size, Ki67 staining, or TUNEL staining when compared to seeding via the PA only (Figure S4E–G).

3.7. HUVECs are deposited in different locations based on the seeding route

To trace the deposition of cells, gravity-seeded HUVECs labelled with DiI or DiO lipophilic carbocyanine dyes were imaged immediately after dual seeding or after 72 h of bioreactor culture (Figure 5D–G). Cells seeded through the PA were diffusely deposited throughout the scaffold down to the distal portions (Figure 5D, E) while cells seeded through the PVs were mainly seen in the proximal regions (Figure 5D). After 72 h, the cells were still dispersed throughout the lungs (Figure 5F, G). Fluorescence microscopy coupled with bright field microscopy indicated that dual seeding of cells into the arteries and veins deposited cells in adjacent vessels but did not appear to deposit cells into the same vessels (Figure 5H). This suggests that the cells did not pass through the capillary bed, an observation also made by Ren et al. (2015) using fluorescently-labelled microspheres.

3.8. Recellularization with segment-specific rat pulmonary endothelial cells

To investigate vascular recellularization further, we evaluated seeding of rat endothelial cells isolated from different pulmonary vascular segments (Alvarez et al., 2008; King et al., 2004). In traditional two-dimensional cultures, PAECs formed tight cobblestone-like colonies; PVECs grew in less-compact colonies and were slow-growing, while MVECs had a homogenous appearance and rapid growth (Figure S5A–C). PAECs, PVCECs and MVECs were also characterized based on expression of endothelial markers and lectin binding ability (Figure S6). All three cell types expressed canonical endothelial markers and demonstrated segment-specific lectin binding.

These three pulmonary endothelial cell phenotypes were individually seeded into the appropriate vascular compartment of rat lung scaffolds. PAECs were seeded into the PA. PVECs were seeded into the PVs. MVECs were dual-seeded to reach both sides of the microvasculature. After seeding, lungs were immediately cultured in a bioreactor for 3 days with antegrade perfusion at 1 ml/min. PAECs were found along the vascular walls, often in large-sized blood vessels and rarely in capillaries (Figure 6A). PVECs mainly lined the walls of small and mid-sized blood vessels (Figure 6B). MVECs attached profusely throughout the vascular tree (Figures 6C, S5D–G) including capillaries (Figure 6C, inset). On average, 97 ± 2% of PAECs, 61 ± 12% of PVECs, and 94 ± 1% of MVECs engrafted during gravity seeding (Figure 6D). Significantly fewer PVECs engrafted than PAECs or MVECs. In addition, fewer PVECs were retained after seeding in comparison to HUVECs seeded via the PVs.

Figure 6.

Rat pulmonary endothelial cells readily attach within the vasculature of rat lung scaffolds. (A) Rat PAECs seeded into the arterial system were found lining many large arteries, often forming a complete monolayer. (B) PVECs seeded into the venous system flattened across the surface of the vascular wall of small and medium-sized veins. (C) MVECs seeded into both the arterial and venous compartments dispersed throughout the matrix in vessels of every size including microvasculature and capillaries (inset). (D) Cell retention during retrograde gravity seeding of PVECs was significantly lower than when seeding HUVECs or PAECs antegrade or when dual seeding MVECs. There was a trend indicating that significantly fewer PVECs were retained than retrograde-seeded HUVECs. (E) The distribution of cells within vessels of set diameters was determined in lungs seeded with HUVECs (optimal method), PAECs, PVECs or MVECs. Because MVECswere seeded in greater numbers thanthe other cell types, the data are displayed as a percentage of total cells (cells/mm is provided inset). Although larger vessels are less numerous than smaller vessels, significantly more PAECs localized to vessels 50–75 compared to PVECs or MVECs. The majority of HUVECs, PAECs and PVECs were located in vessels of 11–25 μm. MVECs, by contrast, were significantly more.

Histological scoring indicated that PAECs were healthy and were moderately distributed through the scaffold (Figure S7A–C). Similarly, PVECs displayed a distribution score of 2.8, which indicates moderate distribution throughout the scaffold. Scaffolds seeded with MVECs yielded higher distribution, morphology and recellularization scores, which may be due, in part, to dual seeding. On average, the cell count algorithm determined the presence of 128 ± 39 PAECs/mm², 65 ± 26 PVECs/mm² and 682 ± 104 MVECs/mm² (Figure S7D). There were significantly more MVECs/mm² than PAECs or PVECs. However, because MVECs were seeding into both the arteries and the veins, nearly twice as many MVECs were seeded as PAECs and PVECs. Half the number of MVECs (386 ± 52 cell/mm²) was also compared to the number of PAECs and PVECs and was still significantly greater (p = 0.0303 and p = 0.0089, respectively).

PAECs displayed greater levels of Ki67 staining than PVECs but not MVECs (Figure S7E). TUNEL staining revealed that PVECs had greater levels of apoptosis compared to PAECs and MVECs (Figure S7F). All rat pulmonary endothelial cells had higher levels of proliferation and lower levels of apoptosis compared to HUVECs, which may reflect a species-specific, organ- specific or vessel segment-specific advantage. For comparison, native rat lungs (parenchymal cells included) were 1.2 ± 0.5% TUNEL-positive. Furthermore, the expected turnover rate for pulmonary endothelial cells is approximately 1% of their total cell number (Bowden, 1983).

To examine further the distribution of PAECs, PVECs and MVECs within the vascular tree, the number of cells seeded within blood vessels of varying diameters was determined as a percentage of total cells (Figure 6E) and as the number of cells per mm² tissue (Figure 6E, inset). Vessel distribution was also examined for HUVECs seeded via the PA. Most HUVECs, PAECs, and PVECs were located within vessels 11–25 μm in diameter, while MVECs were most abundant in microvessels of 10 μm or less. A significantly greater number and percentage of PAECs attached to vessels with a diameter of 50–75 μm than either PVECs or MVECs. Additionally, the percentage of PAECs that attached to intermediate vessels (26–50 μm) was significantly greater than that of MVECs. Overall, HUVECs, PAECs and PVECs localized to intermediate and large vessels while the vast majority of MVECs attached in microvessels.

To determine whether vascular recellularization with HUVECs or segment-specific endothelial cells resulted in the formation of a mature endothelium, tissue sections were stained with VE-cadherin to detect the formation of tight junctions between endothelial cells (Figure S8). Although intermittent positive VE-cadherin staining was seen, the majority of cells failed to express VE-cadherin suggesting that a monolayer of mature endothelium had not yet formed.

3.9. Combination seeding of PAECs, PVECs and MVECs increases cell number and apoptosis

To determine whether multiple endothelial cell types are necessary to promote complete vascular recellularization and the formation of tight junctions, delivery of all three rat pulmonary endothelial cell types was evaluated using three combination seeding methods: (1) sequential seeding on the same day (Seq1d); (2) sequential seeding over the course of 2 days (Seq2d); and (3) simultaneous seeding of a mixture of all three cell types (Mixed; Figure 7 A–C). Each method used a gravity-driven apparatus to deliver the cells. Sequential seeding began with dual seeding of MVECs. This was proceeded immediately (Seq1d) or 24 h later (Seq2d) by seeding of PAECs into the PA and then PVECs in the PVs. MVECs, PAECs, and PVECs were combined and instilled simultaneously in the PA and then the PV for Mixed seeding. For the Mixed seeding method, because all three cell types were instilled as a common mixture, the deposition of cells was not site-specific (meaning that some PAECs were instilled into the PVs and some PVECs were instilled into the pulmonary arteries). After seeding, the lungs were cultured for 3 days in a bioreactor with pulsatile perfusion of media at 1 ml/min.

Figure 7.

Combination seeding of the three rat endothelial cell types increases barrier function. (A) Sequential seeding of MVECs, PAECs, and then PVECs led to good cell distribution and morphology, but cells seen in the lumen of vessels were occasionally fragmented. (B) Sequential seeding on consecutive days (Seq2d) also led to good cell distribution and morphology with re-lining of vessels but also some aggregation in the lumen. (C) Cells delivered by Mixed seeding re-lined much of the vasculature but cell aggregation and fragmentation were common. (D) Albumin tagged with Evans blue dye was perfused into the PA of native, decellularized and combination-seeded lungs. Extravasation of dye into the parenchyma was used to determine barrier function. Decellularized lungs displayed significantly worse barrier function in comparison to native lungs and combination-seeded lungs. There were no statistically significant differences among the combination methods or in comparison to native lung. However, as shown in (E), re-endothelialized lungs did display some extravasation of dye, particularly near the distal portions of the tissue. p-values (*) were determined using t tests. Data are means ± SEM, n = 3.

Each combination seeding method led to cells lining the vessel walls (Figure 7A–C, insets) and yielded excellent cell distribution and morphology scores with no differences amongst the methods (Figure S7A, C). Nonetheless, many cells were also observed within the vessel lumen, perhaps because more cells were seeded via the combination methods than when seeding a single cell type. This observation translated to lower recellularization scores (Figure S7B). Tissues seeded using the Seq1d method had an average of 315 ± 144 cells/mm², while Seq2d had an average of 543 ± 72 cell/mm², and Mixed had 389 ± 73 cells/mm² indicating no significant differences (Figure S7D). For comparison, native rat lung tissues had an average of 1511 ± 140 cell/mm² (parenchymal cells included). Endothelial cells comprise approximately 50% of the cells of the lung (Calle et al., 2014) suggesting that the target endothelial cell number would be approximately 750 endothelial cells/mm².

Unlike seeding of a single endothelial cell type, combination seeding, particularly the Seq2d method, resulted in positive VE-cadherin staining of tight junctions between cells suggesting the formation of a mature endothelium (Figure S8). However, combination seeding also resulted in decreased Ki67 staining with proliferation below 10% for each of the methods and no differences among them (Figure S7E). There were no significant differences in TUNEL staining among the seeding methods, although combination seeding did increase apoptosis in comparison to seeding of just once cell type (Figure S7F). In combination-seeded tissues, TUNEL staining was frequently observed in cell-dense clusters within the lumen of blood vessels (Figure S9).

3.10. Combination seeding improves endothelial barrier function

Because positive VE-cadherin staining was observed after combination seeding, the formation of tight junctions and functional barriers was assessed. BSA conjugated to Evans blue dye was perfused through the PA of lungs that had been re-endothelialized by a combination seeding method. Quantification of the extravasation of albumin into the parenchyma was used as an index of barrier function (Figure 7D). As expected, decellularized lungs exhibited extensive extravasation of BSA-dye (10.6 ± 2.6 μg Evans blue/g tissue) compared to native lungs (0.3 ± 0.1 μg/g). Evans blue dye was seen exiting the PV cannula of native lungs but not decellularized lungs, which permitted the dye to fill the airspace. Regardless of the method used, combination seeding reduced BSA-dye extravasation in comparison to decellularized lungs. There were no significant differences between the combination seeding methods; however, by visual observation, it was apparent that dye leaked from some of the vessels of the lungs, particularly those seeded by the Mixed or Seq1d methods (Figure 7E). Effluent from the PVs was observed in the lungs re-endothelialized using the Seq2d method but not the other two methods. Overall, the data suggested that the Seq2d method led to the formation of the most mature endothelium.

4. Discussion

The generation of a competent, patent vasculature within organ scaffolds is critical for promoting incorporation and survival of an engineered organ upon transplantation. Therefore, optimization of vascular recellularization strategies is necessary for advancing the field of organ engineering (Scarritt et al., 2015; Stabler et al., 2015). Herein, it is demonstrated that gravity seeding confers greater cell distribution and recellularization outcomes than seeding with a pulsatile perfusion pump at a set flow rate. Seeding of endothelial cells under gravity-driven hydrostatic pressure probably allowed for self-regulating flow during seeding, while pulsatile perfusion pump-driven seeding may have forced cells into the vasculature at a set flow rate leading to cell damage.

To determine whether the pressure and shear stress generated during seeding constituted damaging physical stimuli, cells were collected from the pulmonary effluent after seeding to analyse cell growth kinetics and viability. Gravity-seeded HUVECs displayed similar growth and viability to control HUVECs. Perfusion seeding, particularly at 3 ml/min, led to increased apoptosis which, in turn, led to a lag in initial cell growth. Moreover, after transferring seeded lungs to bioreactors for 3 days of perfusion culture, cell fragmentation and apoptosis was observed to a greater degree in perfusion-seeded tissues than in gravity-seeded tissues. These data suggest that cell death was due to shearing forces in the high-pressure environment induced by perfusion pump-driven seeding. Certainly, the pressure during seeding was greatest during perfusion seeding at 3 ml/min and occasional pressure spikes were observed. These observations may indicate that areas of the vasculature become transiently resistant to seeding, perhaps due to the aggregation of cells as they are deposited. Leakage of cells into the parenchyma was observed in perfusion-seeded lungs and may be the result of a pressure build-up that ruptured the vasculature. Flow rate was rapid at the beginning of gravity seeding but decreased 10-fold toward the end of seeding. Thus, for gravity seeding, the decrease in flow was probably in response to cell deposition and may have prevented mechanical damage to cells. It is important to note that cell number and density was held constant for all experiments and was not varied or optimized in this study. Optimization of cell density and seeding volume may reduce cell aggregation. An additional consideration is that gravity-based flow – although variable in rate – was continuous, while perfusion pump-driven flow was pulsatile (correlating with the rotation of the pump head). Further work is necessary to determine the effect of pulsatile perfusion vs. continuous perfusion (via a syringe pump).

Although pump-driven perfusion seeding at 1 ml/min led to significantly lower pressures than gravity seeding or perfusion seeding at 3 ml/min, this pressure may not have been sufficient to seed cells into the distal portions of the scaffold. Indeed, perfusion seeding at 1 ml/min displayed lower distribution and recellularization scores than perfusion seeding at 3 ml/min or gravity seeding. This suggests that there is an ideal pressure range for seeding with a minimum pressure needed to seed the smaller vessels in the distal portions of the tissue and a maximum pressure above which cells are damaged due to fluid shear stress.

Ott et al. (2010) employed gravity seeding of cells into the vasculature of rat lung scaffolds followed by pressure- regulated perfusion culture in a bioreactor. They reported that a pressure range of 10–15 mmHg facilitated vascular recellularization such that seeded lungs could be orthotopically implanted for 2 weeks with no observed thrombosis, although the grafts eventually failed due to incomplete recellularization (Song et al., 2011). The current data indicate that seeding of HUVECs by gravity fell within this range at 13.31 ± 0.34 mmHg, while seeding by perfusion at 3 ml/min exceeded it. Petersen et al. (2010) employed a perfusion flow rate between 1 and 5 ml/min for bioreactor culture of seeded rat lung scaffolds. Their regenerated lung constructs could be implanted, but thrombosis and haemorrhage into the airspace was noted. In a subsequent publication, Calle et al. (2011) emphasized the importance of monitoring the pressures applied to the vasculature during decellularization and seeding, and they recommended the use of gravity flow when possible to prevent pressure from exceeding 15 mmHg. Under pump-driven seeding, pressure increases may cause damage to the cells or the scaffold, whereas gravity seeding or strict pressure-controlled seeding would allow for adjustment of flow to avoid damage.

To evaluate vascular recellularization methods further, seeded lungs were cultured in a bioreactor. The use of a static incubation period to enhance cell attachment prior to initiating bioreactor culture was investigated. A 24 h static incubation led to extensive apoptosis and low recellularization scores. The lack of perfusion during the 24 h incubation allowed cells to aggregate in the lumen rather than bind to the vascular walls. With a 0 or 1 h static incubation period, we observed higher recellularization. Therefore, a long static period is unnecessary for cell attachment and adequate vascular recellularization. Indeed, gravity seeding immediately followed by bioreactor culture at 1 ml/min led to consistently higher cell counts, nuclear size, and histological scores than any other methods. Interestingly, increasing bioreactor flow rate from 1 ml/min to 3 ml/min led to a decrease in Ki67 staining. This may indicate that higher perfusion flow rates generate greater laminar shear stress, which is known to inhibit endothelial cell proliferation (Akimoto et al., 2000; Levesque et al., 1990); however, shear stress has also been shown to mediate phenotypic specification of endothelial cells (Sivarapatna et al., 2015). Because 3 ml/min is a much lower flow rate than what occurs physiologically for rats, this may indicate that the endothelium formed by the HUVECs is still immature at this stage of recellularization.

Ott et al. (2010) demonstrated that HUVECs seeded antegrade and retrograde deposited cells into both the arterial and venous systems. Furthermore, Ren et al. (2015) reported that seeding cells into both routes increased cell distribution and coverage. In this hands, HUVECs were frequently fragmented after seeding through the PVs suggesting that further optimization is required to develop methods for recellularizing the venous system. It should be taken into consideration that physiologically the PVs experience lower pressures than the arteries (which receive the entire cardiac output). Lowering the height of the seeding reservoir during gravity seeding of the PVs may decrease the pressure and damaging stimuli the cells experience during venous seeding. Use of an optimal venous seeding method will probably augment dual seeding outcomes.

Retrograde seeding of rat PVECs led to higher recellularization and morphology scores than retrograde seeding of HUVECs. PAECs and MVECs readily recellularized lung scaffolds displaying higher proliferation than HUVECs. The data indicate that these organ-specific, segment-selective endothelial cells may be preferable for complete lung vascular recellularization. It is important to note that a lower percentage of trypsin was used to harvest segment-specific endothelial cells than HUVECs; this may have allowed better initial adhesion of the segment-specific endothelial cells, although this was not the case for PVECs. Regardless, MVECs seeded by gravity were well distributed throughout the vascular tree, and the majority were deposited in microvessels (<10 μm). PAECs and PVECs frequently attached in the larger blood vessels in the proximal portions of the lung rather than the smaller vessels. These differences in segment-specific endothelial cell deposition may be due to variations in passage number and cell size. Further work is necessary to confirm that these cells truly adhered in a segment-specific fashion, particularly because MVECs were dual-seeded while PAECs and PVECs were seeded into only one route. Additional investigation may reveal that these cell types have a binding affinity for the segment-specific ECM of the larger arteries and veins, as zip codes in which cells localize to their area of origin (Rossi et al., 2013). Indeed, others have demonstrated that stem cells and endothelial progenitor cells have integrin-binding profiles that allow preferential adhesion to specific ECM components (Bonvillain et al., 2012; Caiado and Dias, 2012; Daly et al., 2011; Frith et al., 2012; Lecht et al., 2014). Additionally, it has been demonstrated that there are differences in gene expression that distinguish the endothelial cells of large vessels from microvascular endothelial cells (Chi et al., 2003). Because initial retention of PVECs after seeding was significantly lower than the other cell types, it is possible that the integrin expression profile of the PVECs reduced initial attachment; alternatively, decellularization of the PV ECM may have damaged the binding sites for these cells. Future studies will evaluate the integrin expression of the segment-specific cells, the notion of vascular ECM- specific zip codes and delineate a mechanism of action by which segment-specific cells attach and mature in their segments of origin vs. nonspecific segments.

Seeding of a single cell type did not lead to the formation of a mature endothelium with VE-cadherin+ tight junctions. It was hypothesized that combination seeding of PAECs, PVECs and MVECs would lead to the formation of a mature endothelium by re-establishing the complex physiological heterogeneity of the pulmonary endothelium. Indeed, endothelial cells in the pulmonary arteries, the microvasculature and the PVs experience drastically different physical and chemical environments that influence their phenotype, secretion and proliferation (Aird, 2006, 2007a, b; Alvarez et al., 2008; Brody et al., 1968). Combination seeding, particularly the Seq2d method, resulted in the expression of VE-cadherin and barrier function similar to native lungs. Unfortunately, recellularization scores were lower for combination seeding methods than for seeding of MVECs alone, an effect that could be explained by cell aggregation in the vessel lumen and subsequent anoikis. Therefore, optimization of combination seeding is needed and would probably require reduction in cell seeding number as well as a better understanding of the haemodynamic conditions in lung scaffolds during bioreactor culture. Cell number is indeed an important consideration for re-endothelialization of lung scaffolds. Endothelial cells, particularly microvascular endothelial cells, constitute approximately 50% of the cells of the lung; this translates to 100 billion endothelial cells in a human lung or 500 × 10⁶ cells in a rat lung (Calle et al., 2014; Stone et al., 1992). Considering the high cell count and the partial restoration of barrier properties observed under the haemodynamic conditions performed (e.g. low flow), it is suggested that future studies should focus on optimizing the Seq2d seeding approach with addition of other vascular cells, smooth muscle cells, perivascular cells and/or fibroblasts as well as airway-based introduction of alveolar type I cells to line the alveoli in order to reconstitute a vascular environment reminiscent of the in situ conditions. Indeed, a recent report from Ren et al. (2015) demonstrated that rat scaffolds coseeded with both endothelial and perivascular cells resulted in 75% endothelial coverage and remained patent for 3 days after orthotopic transplantation. Utilization of cell combinations seeded in the correct sequence and proportion will be indispensable to complete re-endothelialization. Because PAECs and PVECs seem to have a proclivity for larger blood vessels, which are often difficult to recellularize, the inclusion of these cell types in re-endothelialization strategies will probably increase vascular coverage and barrier function.

In conclusion, we have demonstrated that gravity- driven seeding is the preferred method for delivery of cells to the vasculature of rat lung scaffolds. Both human and rat endothelial cells lined the vasculature of rat lung scaffolds when seeded by gravity and subsequently cultured in a bioreactor. Culture time could be extended from 3 days to 7 days without detriment to recellularization. The introduction of an additional seeding route did not enhance recellularization suggesting that venous seeding requires further optimization with particular attention paid to the pressure during seeding. PVECs showed promising results, particularly for recellularization of larger vessels. Similarly, PAECs were efficient at lining the vascular walls of larger vessels and also exhibited high percentages of proliferating cells. MVECs attached throughout the vascular tree and localized to the microvasculature with high proficiency. Combined seeding of PAECs, PVECs and MVECs enhanced barrier function suggesting that multiple, segment-specific cell types are necessary for complete re-endothelialization of lung scaffolds. Future studies will focus on identifying the mechanisms behind segment-specific endothelial cell binding, particularly in regard to integrin expression profiles.