Xenogeneic cross-circulation for extracorporeal recovery of injured human lungs

Abstract

Patients awaiting lung transplantation face high waitlist mortality, as injury at the time of donation precludes the use of the majority of donor lungs. Ex vivo lung perfusion (EVLP) has been developed to recover donor organs prior to transplantation. However, EVLP has only demonstrated efficacy in recovering marginal quality donor lungs, and extending normothermic support beyond 6 hours has been challenging. We demonstrate that acutely injured human lungs declined for transplantation, including a lung that failed to recover on EVLP, can be recovered by cross-circulation of whole blood between explanted human lungs and a Yorkshire swine. This xenogeneic cross-circulation platform provided explanted human lungs a supportive, physiologic milieu and systemic regulation that resulted in functional and histological recovery after 24 hours of normothermic support. Our findings suggest that cross-circulation could be developed as a complementary approach to clinical EVLP to enable recovery of injured donor lungs that could not otherwise be utilized for transplantation. Xenogeneic cross-circulation may also serve as a translational research platform for immunomodulation and advanced organ bioengineering.

INTRODUCTION

Medical, surgical, and technological advancements in organ transplantation continue to expand life-saving treatment options for patients with end-stage lung disease, but transplantation remains limited by the low availability of donor organs. As chronic respiratory disease is the third leading cause of death worldwide_¹, the need for innovative solutions to reduce associated morbidity and mortality is imperative. In 2020, lung transplantation remains the only definitive cure for end-stage lung disease, which has poor prognosis due to disease severity, insufficient donor lung availability_^2,3, and high rates of chronic allograft dysfunction_^4,5.

Strategies to increase the organ supply involve utilizing extended criteria lungs_⁶, developing new technologies to recover more donor lungs for transplantation_^7,8, and generating transplantable organs from xenogeneic sources such as genetically-engineered swine_⁹. Currently, the most utilized approach is to recover donor lungs by normothermic ex-vivo lung perfusion (EVLP)_^10,11. However, clinical EVLP devices have failed to demonstrate functional maintenance of an isolated organ beyond 6 hours, or to show evidence of cellular regeneration. Therefore, clinical impact has been limited to recovery of marginal quality donor lungs. To increase the quality and duration of extracorporeal lung support, our group previously reported the development of an organ support platform in a swine model_¹². Cross-circulation of whole blood between explanted swine lungs and a swine host resulted in robust maintenance of healthy lungs outside the body for 4 days_¹³, and enabled significant functional recovery and cellular regeneration of severely damaged swine lungs_¹⁴.

In this study, explanted human lungs declined for transplantation were supported using a cross-circulation platform. We hypothesized that ‘xenogeneic’ cross-circulation between explanted human lungs and a swine host could provide critical systemic regulation, maintain tissue structure and integrity, and support functional lung recovery for 24 hours. A regimen of immunosuppression drugs used in clinical lung transplantation was administered in combination with recombinant cobra venom factor to limit innate and adaptive immune responses. Throughout 24 hours of xenogeneic cross-circulation, human lungs were subjected to multi-scale longitudinal analyses of: (i) physiological and biochemical parameters, (ii) respiratory function and lung histomorphology, (iii) immunologic activity and inflammatory response, and (iv) cellular viability, phenotype, and function.

RESULTS

Characterization of xenogeneic cross-circulation system

A circuit to perfuse explanted human lungs was established with components routinely used for clinical perfusion: polyvinyl chloride flexible tubing, tubing connectors, and a centrifugal pump. A swine host was anesthetized, cannulated through the internal jugular veins, and connected to the circuit as previously described_^12–14. Explanted human lungs were connected to the circuit, marking the start of xenogeneic cross-circulation (Fig. 1a). Clinical characteristics of human lung donors are in Supplementary Table 1. Human lungs were declined for transplantation due to pulmonary contusions with extensive multi-lobar hemorrhage (lung 1), pulmonary consolidation and edema (lung 2, 4, 5, 6), and aspiration pneumonitis with parenchymal opacities (lung 3; Fig. 1b). Notably, lung 5 was procured for xenogeneic cross-circulation after failure to recover during clinical ex vivo lung perfusion (EVLP). The initial quality and function of human lungs in this study were inferior to those used in previous ex vivo lung perfusion studies_^15–18 (Supplementary Table 10).

Fig. 1 |. Maintenance of explanted human lungs using a xenogeneic cross-circulation platform.

a, Human lungs deemed unsuitable for transplantation were procured in standard fashion, transported to the site of experimentation, and cannulated prior to initiation of cross-circulation with a swine host. b, Characteristics of human lungs subjected to xenogeneic cross-circulation. c, Experimental timeline with immunosuppression regimen through 24 hours of normothermic xenogeneic cross-circulation. Induction immunosuppression regimen administered prior to initiation of cross-circulation is described in Supplementary Figure 1. d, Experimental setup for xenogeneic cross-circulation procedures. 1, thermography; 2, perfusion data acquisition system; 3, ventilator; 4, perfusion data display; 5, explanted human lungs; 6, video bronchoscopy; 7, time-lapse photography; 8, swine host. EVLP, ex vivo lung perfusion; LIJ, left internal jugular vein; PA, pulmonary artery; PV, pulmonary vein; RIJ, right internal jugular vein.

As an immunological control, xenogeneic cross-circulation of explanted human lungs was performed without immunosuppression (n=1), which rapidly led to hemodynamic instability, significant increases in serum and airway inflammatory cytokines, pulmonary edema, diffuse alveolar hemorrhage, decreased respiratory function, thrombosis, and irreversible graft dysfunction, consistent with hyperacute rejection (Supplementary Fig. 3). A regimen of immunosuppression drugs used for clinical lung transplantation in combination with recombinant cobra venom factor (a non-toxic protein from cobra venom that depletes complement activity; Fig. 1c, Supplementary Fig. 1b), prevented acute rejection and enabled robust maintenance of explanted human lungs (n=5) throughout 24 hours of xenogeneic cross-circulation (Fig. 1d).

A perfusion and ventilation strategy previously shown to preserve the structural and functional integrity of explanted lungs was implemented_¹⁴. Throughout all procedures, mean pressure at the pulmonary artery was 15.7±4.8 mmHg, while mean pressure at the pulmonary veins was 3.4±2.0 mmHg. During normothermic perfusion (mean perfusate temperature: 35.4±0.2°C), flow rates in the pulmonary artery and veins matched (Supplementary Fig. 1). Swine hosts remained hemodynamically and biochemically stable (mean heart rate, 102±2 beats min₋₁; systolic blood pressure, 88±3 mmHg; pH, 7.43±0.04; lactate; 1.6±0.09 mmol L₋₁) throughout 24 hours of xenogeneic cross-circulation (Supplementary Table 2), and swine host organs had normal histomorphology after 24 hours of xenogeneic cross-circulation (Supplementary Fig. 2).

Functional maintenance of explanted human lungs

Functional parameters are reported both as mean values ± standard deviation for all lungs (Fig. 2), and as separate values for each set of lungs (Supplementary Fig. 5d–k, 6d–k, 7d–k, 8d–k, 9d–k). After 24 hours of xenogeneic cross-circulation, human lungs with fraction of inspired oxygen of 40% had a mean PaO₂/FiO₂ value of 320±124 mmHg, corresponding to a mean increase from baseline of 135 mmHg (Fig. 2a). Mean dynamic compliance was maintained at 15–45 mLcmH₂O₋₁, and consistently increased throughout 24 hours of xenogeneic cross-circulation (Fig. 2b). Oxygenation and ventilation were assessed by changes in hemogases between the pulmonary artery and vein. Mean oxygenation (ΔpO₂) and ventilation (ΔpCO₂) steadily increased over 24 hours (Fig. 2c). Peak inspiratory pressures decreased over 24 hours of xenogeneic cross-circulation, from 19.4±6.7 cmH₂O to 15.6±5.8 cmH₂O, while mean airway pressures were maintained at 8.5±2.4 cmH₂O (Fig. 2d). Notably, after 24 hours of xenogeneic cross-circulation, all lungs ventilated at the target tidal volume of 8 mLkg₋₁ had peak inspiratory pressures within the range deemed acceptable for transplantation^₁₉.

Fig. 2 |. Human lung function throughout 24 hours of xenogeneic cross-circulation.

a, PaO₂/FiO₂. b, Dynamic compliance. c, Changes in pO₂ and pCO₂. Δ = |PA − PV|. d, Peak inspiratory pressure (PIP) and mean airway pressure. e, Lung weight. f, Transpulmonary pressure gradient, the vascular pressure difference between pulmonary artery and pulmonary vein. g, Lactate. h, pH. i-j, Changes in physiological and functional parameters between baseline and 24 hours of xenogeneic cross-circulation. All graphs represent data for human lungs (n = 5). All values represent mean ± standard deviation. PIP, peak inspiratory pressure; P_mean, mean airway pressure; PA, pulmonary artery; PV, pulmonary vein. TPG, transpulmonary pressure gradient.

Following an increase over the first 6 hours of xenogeneic cross-circulation, lung weight, an indicator of pulmonary vascular integrity, had a mean decrease from baseline of 1.1±13.2% (Fig. 2e). Transpulmonary pressure gradient, a measure of pulmonary vascular resistance, was maintained within the target range of 5–15 mmHg (Fig. 2f), and mean lactate at the pulmonary veins remained below 3 mmol L₋₁ (Fig. 2g). The pH was maintained at 7.43±0.04 throughout all procedures (Fig. 2h). Overall, mean changes in functional parameters over 24 hours confirmed that the xenogeneic cross-circulation platform supported recovery of human lung respiratory function: mean PaO₂/FiO₂ increased by 117%, mean dynamic compliance increased by 185% (Fig. 2i), and mean peak inspiratory pressure decreased by 16%, with minimal changes in lung weight, transpulmonary pressure gradient, and lactate (Fig. 2j).

Multi-scale analyses of explanted human lungs

Lung integrity is reported by representative images and mean values ± standard deviation for all lungs (Fig. 3), and individual images and values for each set of lungs (Supplementary Fig. 5a–c, 6a–c, 7a–c, 8a–b, 9a–b). Gross appearance of human lungs at initiation of xenogeneic cross-circulation revealed varying but extensive consolidation, pulmonary edema, and hemorrhage. After 24 hours, lungs showed reduced consolidation and improved aeration (Fig. 3a, Supplementary Video 2). Radiography enabled periodic non-invasive assessment of each lobe and revealed focal and diffuse consolidation prior to xenogeneic cross-circulation. Progressive clearance of regional consolidation was observed after 24 hours, with increased radiolucency and extensive recovery of ventilating lung volume (Fig. 3b). Prior to xenogeneic cross-circulation, video bronchoscopy revealed mild secretions with airway inflammation. However, by 12 hours, large airways appeared normal without evidence of airway secretions, edema, or erythema (Fig. 3c).

Fig. 3 |. Multi-scale analyses of human lungs throughout 24 hours of xenogeneic cross-circulation.

a, Gross appearance. b, Radiographs. c, Bronchographs. Hematoxylin and eosin staining of d, small airways and e, lung parenchyma. f, Scanning electron microscopy demonstrating alveolar structure. g, Transmission electron microscopy demonstrating alveolar–capillary barrier. h, Histopathologic lung injury scores. i, Composite lung injury scores. j, Periodic acid-Schiff staining of bronchoalveolar lavage (BAL) fluid smear. Inset: gross images of BAL fluid. k, Total protein in BAL fluid. l, Quantification of inflammatory cytokines in BAL fluid. * p < 0.05. All images are representative of human lungs from one donor. All graphs represent data for human lungs (n = 5). All values represent mean values or mean ± standard deviation. Representative images of all human lungs are in Supplementary Fig. 5 – 12. Quantifications of cytokines in BAL fluid from all human lungs are in Supplementary Table 4 – 8. Lung injury scores of all human lungs are in Supplementary Fig. 13. PMN, polymorphonuclear cells.

Histologic analyses confirmed maintenance of conducting airway structures, including intact pseudostratified epithelium in segmental and sub-segmental bronchi after 24 hours of xenogeneic cross-circulation (Fig. 3d; Supplementary Fig. 10c, 11c, 12c). Edema, intimal thickening, and cellular derangement in lung parenchyma were minimal (Fig. 3e; Supplementary Fig. 8c; 9c; 10a,b, 11a,b, 12a,b), and extensive alveolar recruitment and aeration were consistent with diagnostic radiography. Scanning electron microscopy revealed that alveolar architecture was initially perturbed, with vascular exudate and/or alveolar hemorrhage. After 24 hours, alveolar architecture appeared substantially restored, with minimal evidence of extravasation or structural degradation (Fig. 3f; Supplementary Fig. 10e, 11e, 12e). Transmission electron microscopy enabled detailed assessment of the alveolar-capillary barrier, which appeared damaged or disrupted prior to initiation of xenogeneic cross-circulation, but nevertheless retained the blood–gas interface throughout 24 hours (Fig. 3g; Supplementary Fig. 10f, 11f, 12f).

Histopathology and airway inflammatory response

Lung injury was evaluated by blinded histopathologic assessment, which included quantification of airway and alveolar polymorphonuclear cells, alveolar and interstitial edema, interstitial infiltrate, and early (caspase 3₊) and late (TUNEL₊) apoptotic cells (Supplementary Table 3). Injury scores at 0, 12, and 24 hours of xenogeneic cross-circulation, visualized by radar plot, indicated reduction from baseline across multiple lung injury categories (Fig. 3h, Supplementary Fig. 13). Although alveolar and interstitial edema combined score increased slightly above baseline (0 h, 0.44; 24 h, 0.69), early and late apoptosis combined score decreased by 63.6% (0 h, 2.26; 24 h, 0.82). Airway and alveolar polymorphonuclear cells combined score decreased by 51.7% (0 h, 1.09; 24 h, 0.53), and the composite lung injury score decreased by 45.1% (0 h, 5.04±0.70; 24 h, 2.76±0.65) over 24 hours of xenogeneic cross-circulation (Fig. 3i).

Inflammation of pulmonary airways was assessed by analyses of bronchoalveolar lavage (BAL) fluid, including staining of BAL fluid smears and quantification of airway inflammatory cytokines. At initiation of xenogeneic cross-circulation, BAL fluid samples appeared reddish and turbid, suggesting the presence of blood, mucous, and cellular debris. After 24 hours, with periodic airway lavage, BAL fluid appeared clear, consistent with cytologic smears that showed reduced cellular debris between 0 hours and 12 hours, and normal pH (5.5) by 24 hours (Fig. 3j). In BAL fluid, total protein concentration, a measure of airway secretions and leakage of serum proteins across the epithelial barrier, remained constant throughout 24 hours, suggesting no degradation of the blood–gas barrier (Fig. 3k). Mean concentrations of inflammatory cytokines IFNγ, IL-1α, IL-1β, IL-1ra, IL-2, IL-8, IL-12, IL-16, 1L-17a, IL-18, TNFα, and TNFβ decreased in BAL fluid, while mean concentrations of IL-4, IL-5, IL-6, and IL-10 increased (Fig. 3l, Supplementary Table 4–8). Notably, in BAL fluid the only statistically significant increase in mean cytokine concentration was IL-6 (4.7-fold), while mean concentrations of IL-1β, IL-1ra, and IL-18 significantly decreased (9.5, 5.4, 2.7-fold).

Pulmonary endovascular integrity and immune responses

Endothelial cells are the first to interact with non-self antigens in organ transplantation, and serve as key regulators of immune response and coagulation. After 24 hours of xenogeneic cross-circulation, angiography confirmed patency and preservation of the pulmonary vascular tree throughout all lobes and peripheral regions (Fig. 4a). Pentachrome staining revealed that endothelial surfaces of large pulmonary arteries and veins were free of fibrin deposition, and elastic fibers in arterial and venous vessel walls retained stereotypical organization and distribution (Fig. 4b). Arterioles and venules retained normal endothelial cell morphology (Fig. 4c). Transmission electron microscopy of pulmonary capillaries confirmed retention of intact alveolar–capillary barrier without evidence of microthrombi (Fig. 4d).

Fig. 4 |. Endovascular integrity and immunologic response during xenogeneic cross-circulation.

a, Angiogram at 24 hours demonstrating patent pulmonary vasculature. b, Pentachrome staining and c, hematoxylin and eosin staining of pulmonary vascular tree. d, Transmission electron microscopy (pseudocolored) demonstrating red blood cells (red), endothelial cell membrane (blue) and nucleus (orange). e, Immunohistochemical staining of fibrin and glycoprotein IIb/IIIa (GpIIb/IIIa) showing lack of thrombotic deposition in large vessels. Star indicates vessel lumen. Positive control staining performed on thrombus formed on glass slide. f, Platelets. g, Quantification of hemolytic and injury markers. h, Uptake of viability marker carboxyfluorescein succinimidyl ester (CFSE) by vascular endothelium. Star indicates vessel lumen. i, Immunohistochemical staining of endothelial cell marker vascular endothelial (VE)-Cadherin. j, Uptake of acetylated low-density lipoprotein (aLDL) by vascular endothelial cells. k, Immunohistochemical staining of vascular smooth muscle cell marker alpha smooth muscle actin (αSMA). l, Vasoresponsiveness to phenylephrine after 24 hours of xenogeneic cross-circulation. m, Relative expression of endothelial genes in human lungs at 0 and 24 hours of xenogeneic cross-circulation, compared with normal human lung from Genotype-Tissue Expression datasets. n, Quantification of circulating serum cytokines. Star indicates statistical significance (p < 0.05). o, Quantification of complement activity (CH50%) after administration of cobra venom factor 4 hours prior to start of xenogeneic cross-circulation. (XC). p, Immunohistochemical staining of swine and human CD45⁺ leukocytes. q, Quantification of swine and human CD45⁺ leukocytes in human lungs. Quantification of r, neutrophil elastase⁺ cells and s, CD163⁺ cells in alveolar and interstitial spaces. All images are representative of human lungs from one donor. a – e, h – k represent data after 24 hours of xenogeneic cross-circulation. All graphs represent data for human lungs (n = 5). All values represent mean values or mean ± standard deviation. aLDL, acetylated low-density lipoprotein; αSMA, alpha smooth muscle actin; CFSE, carboxyfluorescein succinimidyl ester; GpIIb/IIIa, glycoprotein IIb/IIIa; Hgb, hemoglobin; hpf, high-powered field; VE, vascular endothelial.

After 24 hours of xenogeneic cross-circulation, complete blood count and immunostaining for clotting (fibrin) and platelet activation (glycoprotein IIb/IIIa) markers showed no evidence of thrombotic deposition in large vessels (Fig. 4e) or thrombocytopenia (Fig. 4f). Mean hematocrit was consistently maintained above 28% (Supplementary Table 2), indicating minimal hemolysis. Markers of platelet activation, including fibrinogen (clotting precursor), D-dimer (fibrin degradation product), and plasma free hemoglobin (hemolysis marker) remained within normal ranges (Fig. 4g). Remarkably, P-selectin, a marker of endothelial injury and platelet/leukocyte activation, decreased from 22.8 ng mL₋₁ to 4.9 ng mL₋₁ (Fig. 4g). Altogether, human lungs on xenogeneic cross-circulation retained integrity of the delicate pulmonary vascular endothelium without evidence of significant injury, activation, or thrombotic microangiopathy associated with acute immunologic rejection.

Viability and phenotypic maintenance of pulmonary vascular endothelial cells were confirmed, respectively, after 24 hours by pervasive uptake of cell viability marker carboxyfluorescein succinimidyl ester (CFSE, Fig. 4h) and normal distribution of vascular endothelial (VE)-cadherin (Fig. 4i). Uptake of acetylated low-density lipoprotein confirmed endothelial cell function in small and large vessels (Fig. 4j). Immunostaining for alpha smooth muscle actin indicated myocyte preservation in pulmonary vascular smooth muscle (Fig. 4k), which displayed an expected increase in arterial pressure after intravascular injection of phenylephrine (Fig. 4l). To assess transcriptomic changes in endothelial-associated genes, differential gene expression analysis was performed between 0 and 24 hours, and compared to expression in normal human lungs from the Genotype-Tissue Expression Consortium. Throughout xenogeneic cross-circulation, explanted human lungs demonstrated no statistically significant difference in expression of 19 endothelial-associated genes across timepoints or compared to normal lungs (Fig. 4m, Supplementary Fig. 14a), suggesting that endothelial cell phenotype was not significantly altered during xenogeneic cross-circulation.

Serum cytokines serve as major regulators of immune response after organ transplantation. Quantification by multiplex array of circulating swine cytokines revealed that from baseline to 24 hours, concentrations of IL-1α, IL-1β, IL-1ra, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, and IL-18 changed minimally, between –2.9-fold and +5.3-fold (Fig. 4n, Supplementary Table 9). Fold change increases in serum concentrations were greatest for IFNγ (+76.8), TNFα (+23.5), GM-CSF (+7.5), and IL-1ra (+5.3), although only TNFα and IL-12 had statistically significant changes between baseline and 24 hours. Innate immune activity was assessed by total complement activity (CH50) assay, which confirmed depletion of complement activity prior to the start of xenogeneic cross-circulation (Fig. 4o). Without immunosuppression, complement components C3b and C5b-9 demonstrated deposition throughout lung tissue within 30 minutes of initiating xenogeneic cross-circulation (Supplementary Fig. 3o,p). Suppression of cell-mediated immune response was achieved by therapeutic doses of methylprednisolone, mycophenolate, and tacrolimus (FK506; Supplementary Table 2).

To assess retention of leukocytes in human lungs, CD45₊ leukocytes were quantified by immunostaining for swine galactose-α-1,3-galactose (αGal) and human nuclear antigen (HNA). Representative images revealed the presence of human (HNA₊) leukocytes after 24 hours (Fig. 4p), although quantitative analysis indicated progressive clearance of human CD45₊ leukocytes from the lungs over 24 hours of xenogeneic cross-circulation. After an initial increase at the start of cross-circulation, swine CD45₊ leukocytes in human lungs also decreased over 24 hours (Fig. 4q). Neutrophils (human and swine), responders to ischemia–reperfusion injury and acute inflammation, were initially elevated, but by 12 hours were drastically reduced in alveolar and interstitial compartments (Fig. 4r). Similarly, CD163₊ macrophages (swine and human), key regulators of immune response, were initially more abundant in alveolar spaces, likely in response to donor injuries, but by 24 hours were reduced. The number of CD163₊ macrophages in the interstitium remained constant across 24 hours of xenogeneic cross-circulation (Fig. 4s).

Cellular integrity, phenotype, and function

After 24 hours of xenogeneic cross-circulation, lung regions that were initially consolidated were recruited (Fig. 5a). Pentachrome staining confirmed outstanding maintenance of gross structure in main and lobar bronchi (Fig. 5b). Pseudostratified airway epithelium with mucus-containing goblet cells (Fig. 5c) and ciliated airway epithelial cells with intact ciliated brush border (Fig. 5d) were observed throughout large airways. Histomorphology of airway secretory glands (Fig. 5e) and airway smooth muscle myocytes (Fig. 5f) appeared normal. Notably, regenerative airway stem cell populations, including p63₊ basal cells (Fig. 5g) and αSMA₊ submucosal gland epithelial cells (Fig. 5h), were retained after 24 hours, suggesting that the potential for endogenous repair of airways was preserved through xenogeneic cross-circulation.

Fig. 5 |. Maintenance of pulmonary airway and alveolar–capillary barrier after 24 hours of xenogeneic cross-circulation.

a, Gross appearance of human lung 5 demonstrating right lower lobe recruitment (dotted line). Pentachrome staining of b, large airway structures. Star indicates airway lumen. c, goblet cells containing mucin d, intact cilia e, submucosal glands f, airway smooth muscle. g, Immunohistochemical staining for airway p63⁺ basal cells. h, Airway submucosal gland alpha smooth muscle actin⁺ myoepithelial cells. Small airway structures and function: i, Polyurethane airway cast of preserved small airways and respiratory bronchioles. j, Immunohistochemical staining of small airway columnar epithelium and smooth muscle. k, Bronchoresponsiveness of airway smooth muscle to albuterol and methacholine. Immunohistochemical staining of l, alpha-tubulin⁺ ciliated cells m, CC10⁺ club cells n, PGP9.5⁺ neuroendocrine cells o, Mucin5B⁺ goblet cells. p, Uptake of viability marker carboxyfluorescein succinimidyl ester (CFSE) by parenchymal cells. q, Metabolic activity of parenchymal cells. r, Relative expression of epithelial genes in human lungs at 0 and 24 hours of xenogeneic cross-circulation, compared with normal human lung from Genotype-Tissue Expression datasets. Maintenance of alveolar–capillary barrier by immunohistochemical staining: s, CD31⁺ endothelial cells and ZO-1⁺ microvascular tight junctions. t, NG2⁺ pericytes. u, Aquaporin 5⁺ type I pneumocytes. v, Caveolin-1⁺ type I pneumocytes. w, ZO-3⁺ epithelial tight junctions. x, HT-280⁺ type II pneumocytes. y, SPC⁺ type II pneumocytes. z, Uptake of fluorescent BODIPY-SPB by live type II pneumocytes. Images are representative of human lungs from one donor. Dotted line indicates airway basement membrane. All graphs represent data for human lungs (n = 5) and represent mean ± standard deviation. αSMA, alpha smooth muscle actin; Aqp5, aquaporin 5; Cav-1, caveolin-1; CFSE, carboxyfluorescein succinimidyl ester; EpCAM, epithelial cell adhesion molecule; NG2, neural/glial antigen 2; OD, optical density; PEEP, positive end expiratory pressure; PIP, peak inspiratory pressure; SPB, surfactant protein B; SPC, surfactant protein C; ZO-1, zonula occludens-1; ZO-3, zonula occludens-3.

To assess the structural integrity of small airways in explanted human lungs, airway casts were prepared and confirmed maintenance of bronchiolar architecture and respiratory bronchioles (Fig. 5i). Immunostaining for epithelial cell adhesion molecule (EpCAM) revealed undisrupted columnar respiratory epithelium with intact underlying airway smooth muscle (Fig. 5j), which was responsive after 24 hours of xenogeneic cross-circulation to nebulized administration of bronchomotor regulatory agents (Fig. 5k). Primary respiratory epithelial cell phenotypes were preserved throughout small airways, including ciliated cells (Fig. 5l), club cells (Fig. 5m), neuroendocrine cells (Fig. 5n), and goblet cells (Fig. 5o). Ubiquitous uptake of viability marker carboxyfluorescein succinimidyl ester (CFSE) in distal lung regions confirmed parenchymal cell viability (Fig. 5p), and mitochondrial activity assay indicated that explanted human lungs maintained relatively constant metabolic activity throughout 24 hours of xenogeneic cross-circulation (Fig. 5q).

Expression of genes associated with the pulmonary epithelium was assessed by RNA sequencing, which revealed no significant difference in expression of 27 epithelial-associated genes across timepoints or compared to normal human lungs (Fig. 5r, Supplementary Fig. 14b). Pulmonary microvascular CD31₊ endothelial cells exhibited typical co-localization with tight junction protein ZO-1 (Fig. 5s), confirming retention of tight junctions in the pulmonary capillary plexus. Pericytes, critical support cells for alveolar capillaries, were also maintained throughout the distal lung (Fig. 5t). Type I pneumocytes were visualized by aquaporin 5 (Fig. 5u) and caveolin 1 (Fig. 5v) amid normal distribution of epithelial tight junction protein ZO-3 (Fig. 5w), consistent with lung weight and electron microscopy that confirmed maintenance of respiratory epithelium in distal lung regions after 24 hours. Type II pneumocytes retained expression of HT-280 (Fig. 5x) and surfactant protein C (Fig. 5y), and internalized fluorescently-labeled surfactant protein B (SPB-BODIPY; Fig. 5z, Supplementary Video 3), confirming viability and uptake function of type II cells after 24 hours of xenogeneic cross-circulation. Additional representative micrographs of human lungs at baseline and 24 hours of xenogeneic cross-circulation are in Supplementary Fig. 10g–l, 11g–l, 12g–l.

DISCUSSION

We describe a xenogeneic cross-circulation system that supports the viability and functional recovery of human donor lungs declined for transplantation (Supplementary Video 1). Despite compromised respiratory function (PaO₂/FiO₂ < 300 mmHg) at baseline due to common injury etiologies, including (i) traumatic lung injury (lung 1), (ii) severe pulmonary consolidation and edema (lungs 2, 4, 5), and (iii) aspiration pneumonitis (lung 3), human lungs consistently demonstrated functional and histological improvements throughout 24 hours of xenogeneic cross-circulation (Fig. 2i, 3i). Remarkably, lung 5, which was declined for transplantation after 5 hours of clinical EVLP due to persistent consolidation of the right lower lobe, demonstrated significant recruitment and recovery after 24 hours of xenogeneic cross-circulation support. These results suggest that cross-circulation could serve as a complementary therapy when lungs fail to recover within the 6 hour window clinically approved for normothermic EVLP.

Compared to human lungs subjected to EVLP_^15–18, lungs in this study were more severely injured and subjected to longer periods of ischemia prior to extracorporeal support, yet showed greater functional improvements (Supplementary Table 10) and lower concentrations of inflammatory cytokines_^20–22 (Supplementary Table 11) after 24 hours of xenogeneic cross-circulation. Moreover, the greatest improvements in respiratory function occurred after 12 hours of xenogeneic cross-circulation, suggesting that durations of normothermic support beyond those approved for EVLP devices may be required to recover severely injured lungs. Except for slight increases in edema (likely due to the static prone positioning of lungs throughout cross-circulation), histopathologic scores steadily decreased throughout 24 hours of xenogeneic cross-circulation, and approached those of transplanted human lungs in multiple categories (Supplementary Fig. 13d). Previously, mechanisms of airway and alveolar repair after acute lung injury have been shown to involve proliferation and differentiation of endogenous lung progenitor cells (e.g., differentiation of type II to type I pneumocytes to restore the alveolar epithelial barrier)_^23,24. Thus, retention of p63₊ airway basal cells_²⁵, αSMA₊ submucosal epithelial cells_²⁶, and HT-280₊ type II cells_²⁷ after 24 hours of xenogeneic cross-circulation suggests that multiple endogenous airway and alveolar progenitor cell populations critical for tissue regeneration were preserved. In the future, therapeutic strategies aimed at promoting endogenous repair could be employed during multi-day cross-circulation to accelerate recovery of more lungs and expand the supply of available organs.

Research efforts in xenotransplantation have aimed to create an unlimited supply of swine organs for human transplantation._^9,28,29 Despite advances in genetically-engineered swine to eliminate acute xenograft rejection_⁹, clinical translation remains distant, as chronic xenoantigen exposure leads to antibody-mediated rejection and graft failure._³⁰ Notably, xenogeneic cross-circulation could circumvent the translational barriers of xenotransplantation because: (i) during xenogeneic cross-circulation, human lung exposure to circulating swine xenoantigens is transient, mitigating risk of chronic rejection,_^9,31,32 and (ii) human lungs recovered via xenogeneic cross-circulation would be used for allograft transplantation of a human organ into a human recipient, precluding risks associated with xenograft antibody-mediated rejection. Previously, dialysis_^33,34 and liver failure_^35–37 patients who experienced acute exposure to swine xenoantigens developed anti-swine xenoantibodies, but subsequently underwent successful transplantation of kidney or liver allografts. A review of eight studies investigating sensitization to swine antigens concluded that ‘after sensitization to pig antigens, there is no evidence of antibody-mediated or accelerated cellular rejection of a subsequent allograft’_³⁸. These findings suggest that a human organ transiently exposed to swine xenoantigens during xenogeneic cross-circulation could be tolerated by a transplant recipient.

To reduce risk of acute rejection during the procedure, an immunosuppression regimen informed by practices in clinical lung transplantation_³⁹ was implemented: triple therapy of tacrolimus (calcineurin inhibitor), mycophenolate (antiproliferative agent), and methylprednisolone (corticosteroid), in combination with cobra venom factor (complement-activating protein that depletes complement activity_^40–43). Interestingly, cobra venom factor has been shown to eliminate acute xenograft rejection_^44,45, preserve tight junction integrity_⁴¹, and decrease ischemia–reperfusion injury associated with transplantation_^41,46–48. Accordingly, after 24 hours of xenogeneic cross-circulation, explanted human lungs demonstrated patent microvasculature, intact blood–gas barrier, and decreased serum P-selectin, with no evidence of clotting or angiopathy.

In this study, an immunologic control without immunosuppression (lung 6) resulted in severe immunologic rejection within 1 hour of xenogeneic cross-circulation, validating the need for an immunosuppression regimen. We also report the first transcriptomic analyses of human lungs during xenogeneic cross-circulation (Supplementary Fig. 14), and establish benchmarks of inflammatory cytokines in BAL fluid and circulating serum. IL-1β and IL-8, markers of leukocyte infiltration_^49,50 and primary graft dysfunction following lung transplantation_^51,52, showed minimal changes from baseline in both BAL fluid and serum. While mean serum concentrations of IFNγ and TNFα increased, the only BAL fluid cytokine that demonstrated a statistically significant increase was IL-6, which has been reported to both increase inflammatory response_²⁹ and protect early graft survival_^54,55. Altogether, these findings suggest that after 24 hours of xenogeneic cross-circulation with immunosuppression, human lungs experienced minimal endothelial activation_⁵⁶ and immune response_^57–59, and showed no clinical signs of acute rejection.

The experimental use of Yorkshire swine for xenogeneic cross-circulation resulted in the localization of swine cells (e.g., αGal₊ CD45₊ leukocytes) in human lungs, indicating a degree of human–swine chimerism. While chimerism has been shown to induce graft tolerance^₆₀, further research is required to: (i) elucidate the immunological mechanisms in organs recovered by xenogeneic cross-circulation, (ii) develop strategies to reduce non-human cells in human lungs prior to transplantation, and (iii) evaluate lungs recovered by xenogeneic cross-circulation after transplantation in a non-human primate model. Additionally, this proof-of-feasibility study will inform future research evaluating longer durations of extracorporeal support and deeper characterizations of the immune response. Future studies will also investigate strategies aimed at limiting the adaptive immune response, including the use of humanized^_61–63 or immunodeficient swine^₆₄, and targeted lymphodepletion therapies^_65–68. Importantly, medical-grade swine free of zoonoses^_56,57 will be required for clinical translation.

Based on the results of this study, we envision clinical applications wherein human organs declined for transplantation due to reversible injuries (e.g., pulmonary contusion, gastric aspiration), or lungs failing to recover on EVLP, could be recovered using xenogeneic cross-circulation (Supplementary Fig. 15). Alternatively, patients who await transplantation on extracorporeal membrane oxygenation support could serve as allogeneic cross-circulation ‘hosts’ to enable functional assessment and graft recovery of reversibly injured lungs prior to transplantation. Thus, cross-circulation could enable assessment and recovery of high-risk lungs without the concurrent stress of a surgical transplantation procedure. Recovered lungs would be subsequently transplanted into the host-recipient.

Beyond clinical applications, the xenogeneic cross-circulation platform may also serve as a basic and translational research tool to further investigate outstanding immunological questions and enable studies of extracorporeal organ recovery, regeneration, bioengineering_⁷¹, and advanced therapeutic interventions_⁸. Modifications to the xenogeneic cross-circulation circuit could enable investigation and recovery of other human organs, including livers, hearts, kidneys, and limbs. Ultimately, we envision that xenogeneic cross-circulation could be utilized as both a translational research platform to augment transplantation research, and as a biomedical technology to help address the organ shortage by enabling the recovery of previously unsalvageable donor organs.

ONLINE METHODS

Study design.

The study was designed as a proof-of-feasibility study (n = 6) to assess the ability of the xenogeneic cross-circulation system to maintain and recover the quality and function of explanted human lungs for 24 hours. As explanted human lungs are susceptible to injury, dysregulation, and loss of respiratory function without adequate support, 24 hours was deemed a sufficient duration of extracorporeal support to assess the feasibility of xenogeneic cross-circulation. Our hypothesis was that xenogeneic cross-circulation between explanted human lungs and a swine host could provide a systemic environment that supports lung maintenance and recovery for 24 hours. The study included two experimental groups: (i) with immunosuppression (n = 5), and (ii) without immunosuppression (n = 1 as an immunological control), and was conducted with the minimum number of animals to demonstrate feasibility and reproducibility between lungs and across experimental time points. Data from study can be used to conduct power analyses for subsequent investigations. All samples were collected and analyzed in triplicate.

Animals.

Yorkshire swine (n = 6) were 5 – 8 months of age, with a mean weight of 62.3 ± 1.4 kg (range: 61.4 – 64.0 kg). As experimental xenogeneic cross-circulation procedures were conducted at sites in Columbia University and Vanderbilt University, the study received approvals from the Institutional Animal Care and Use Committee (IACUC) at both Columbia University and Vanderbilt University. All animal care and procedures were conducted in accordance with the US National Research Council of the National Academies Guide for the Care and Use of Laboratory Animals, Eighth Edition.

Immunosuppression.

To assess the feasibility of conducting xenogeneic cross-circulation without immunosuppression, immunosuppressive drugs were not administered before or during xenogeneic cross-circulation (n = 1). Lack of immunosuppression resulted in severe acute rejection in less than 1 hour, and practically prohibited xenogeneic cross-circulation. Consequently, to reduce risk of immunologic rejection and enable xenogeneic cross-circulation, an immunosuppression regimen informed by established protocols_^13,14 and current practices in clinical lung transplantation_³⁹ was implemented, and immunosuppression drugs were administered before and during xenogeneic cross-circulation (n = 5). Induction immunosuppression consisted of a conventional combination of calcineurin inhibitor (tacrolimus) and antiproliferative agent (mycophenolate). Mycophenolate (500 mg; Roche) and tacrolimus (0.4 mg kg₋₁; Sandoz) were administered per os twice a day for at least 14 days prior to initiation of xenogeneic cross-circulation. Enrofloxacin (5 mg kg₋₁; Henry Schein) and cephalexin (25 mg kg₋₁; Zoetis) were also administered to reduce risk of infection during induction of immunosuppression. Four hours prior to initiation of xenogeneic cross-circulation, swine hosts were sedated, intubated, and administered intravenous diphenhydramine (50 mg; Henry Schein), methylprednisolone (1 g; Henry Schein), and cobra venom factor (1 mg; Sigma-Aldrich) to deplete complement activity_^40–42. Immediately prior to the start of xenogeneic cross-circulation, tacrolimus (5 mg; Sandoz), methylprednisolone (125 mg; Henry Schein), and mycophenolate (500 mg; Roche) were administered intravenously to inhibit innate and adaptive immune responses. Methylprednisolone was subsequently re-administered at 8 hours and 16 hours. Tacrolimus and mycophenolate were re-administered at 12 hours (Supplementary Fig. 1b). Blood samples were collected every 1 – 4 hours to assess circulating levels of immunosuppression drugs. Complement activity was quantified by CH50 assay (EZ Complement CH50 Test; Diamedix). Tacrolimus (FK506) levels were quantified (Architect System; Abbott) to ensure maintenance of therapeutic levels throughout xenogeneic cross-circulation.

Human lung donors.

Donor lungs (n = 6) unacceptable for clinical transplantation and consented for use in research were procured in coordination with local organ procurement organizations under protocols approved by the Institutional Review Board at Columbia University Medical Center and Vanderbilt University Medical Center. Deidentified donor data were obtained through the United Network for Organ Sharing under the approved protocols. Donors were excluded based on the following criteria: severe purulent secretions on bronchoscopy; history of infection with human immunodeficiency virus (HIV), hepatitis B/C, and methicillin-resistant staphylococcus aureus (MRSA).

Procurement of human lungs.

When human lungs acceptable for the study were identified, a team was dispatched from Columbia University Medical Center or Vanderbilt University Medical Center to procure lungs in standard fashion. All lung procurements were performed in coordination with other teams dedicated to procurement of organs (e.g., heart, kidney, liver) for clinical transplantation. At the time of procurement, a bolus of heparin (30,000 U) was administered intravenously, the aorta was clamped, and cold low-potassium dextran solution (Perfadex; Vitrolife) with alprostadil (25 mg kg₋₁, Prostin VR Pediatric; Pfizer) was administered through the pulmonary artery as an anterograde flush. The chest was packed with sterile ice slush to achieve topical cooling, lungs were inflated to an airway pressure of 15 cmH₂O, and the trachea was stapled (DST TA; Medtronic). Lungs were then explanted and placed in a sterile isolation bag with 500 mL organ preservation solution (Perfadex; Vitrolife) at 4°C. The bag was placed in a second sterile isolation bag containing 1 L normal saline at 4°C, and subsequently placed on ice.

Cannulation of explanted human lungs.

After explant, lungs were transported on ice to the research operating room at Columbia University Medical Center or Vanderbilt University Medical Center, and placed on the back table on ice for preparation. Notably, because the heart was procured for clinical transplantation from three donors (Donor 1, 3, 6), and the contralateral lung was procured for clinical transplantation from two donors (Donor 4, 5), segments of the structures connecting the heart and lungs, i.e., the left atrial cuff and main pulmonary artery, were procured for transplantation, resulting in minimal (1 – 3 mm) circumferential left atrial cuff and truncated or no main pulmonary artery. Consequently, reconstruction of the left atrial cuff and/or pulmonary artery was required in some cases (Supplementary Fig. 4). Donor pericardium, descending thoracic aorta, or polyethylene terephthalate grafts (Dacron; Maquet) were used to facilitate cannulation of explanted human lungs at the pulmonary artery and left atrial cuff. After reconstruction, the pulmonary artery and left atrial cuff were cannulated as previously described_^12–14, and cold sterile isotonic electrolyte-balanced solution was flushed through the lungs to prime arterial and venous cannulas. A basin lined with two sterile isolation bags and filled with warm saline was placed in the organ preservation chamber, which was equipped with a heater and humidifier to maintain normothermic physiologic conditions similar to those in the thoracic cavity. A top-loading balance (Denver Instrument Company) was inserted beneath the basin to enable lung weight measurements. After cannulation, lungs were placed in the basin in prone position, and cannulas were secured.

Xenogeneic cross-circulation between explanted human lungs and swine host.

All swine hosts (n = 6) underwent general anesthesia by intramuscular induction with tiletamine/zolazepam (5 mg kg₋₁, Telazol; Zoetis). Anesthesia was maintained with continuous intravenous infusions of fentanyl citrate (0.1 mg kg₋₁ h₋₁; West Ward Pharmaceuticals), midazolam (1.5 mg kg₋₁ h₋₁; Akorn), and inhaled isoflurane (1 – 5% in oxygen; Henry Schein). Cefazolin (30 mg kg₋₁; Sandoz) and enrofloxacin (5 mg kg₋₁; Henry Schein) were administered prior to skin incision and re-administered every 8 hours and 24 hours, respectively. An auricular or femoral arterial line (Arrow International) was placed for hemodynamic monitoring and periodic blood sampling. Bilateral neck cut-downs exposed the left and right internal jugular veins. A heparin bolus (30,000 U) was administered, and cannulas (18 French; Medtronic) were placed in the internal jugular veins using the Seldinger technique. Calcium chloride (1 g; Henry Schein) was then administered, and explanted human lungs were connected to the circuit via cannulas secured in the pulmonary artery and vein, thereby marking the start of xenogeneic cross-circulation. Initial flow rates were maintained within 5 – 10% of estimated cardiac output of the swine host, with target pulmonary artery pressure < 15 mmHg, and pulmonary vein pressure 3 – 5 mmHg. Perfusion circuit elements consisted of a main console (Jostra HL-20 pump console; Maquet), disposable pump (Rotaflow centrifugal pump; Maquet), softshell reservoir (Maquet), and three-eighths inch flexible tubing (Smart coated tubing; LivaNova). Pressures (PA, PV), flows (PA, PV), and perfusate temperatures were continuously monitored and recorded (VIPER clinical interfacing software G2 v1.26.4; Spectrum Medical). Throughout the duration of xenogeneic cross-circulation, swine hosts were maintained with a continuous heparin infusion (initial rate: 25 U kg₋₁ h₋₁). Activated clotting time was measured using a whole blood micro-coagulation system (Hemochron; Accriva Diagnostics), and the heparin drip was adjusted to maintain activated clotting times within the target range of 250 – 350 s. Physiological parameters of the swine host, including heart rate, electrocardiogram, blood pressure (cuff and arterial line pressure), mean arterial pressure (MAP), oxygen saturation (SpO₂), end-tidal CO₂, temperature, and respiratory rate, were continuously monitored and recorded (Supplementary Table 2) using a multi-parameter vital signs monitor (SurgiVet).

Ventilation of explanted human lungs.

Ventilation strategy.

The circuit and lungs were allowed to acclimate to normothermic temperature, and ventilation was initiated within the first 10 minutes of cross-circulation, with the following initial settings: volume control mode; respiratory rate, 6 – 8 breaths min₋₁; tidal volume, 6 mL kg₋₁; positive end-expiratory pressure (PEEP), 5 cmH₂O; fraction of inspired oxygen (FiO₂), 40% (Oxylog 3000 plus; Dräger). Continuous time-lapse photography (1 frame min₋₁) was captured with a high-resolution camera (Hero4 Black 4K; GoPro, Supplementary Video 1).

Alveolar recruitment.

Atelectatic lung regions were recruited by gradually increasing tidal volume by 15 – 20%, increasing positive end-expiratory pressure (PEEP) up to 10 cmH₂O, and performing inspiratory holds (sustained inflation) with peak inspiratory pressures up to 25 cmH₂O. If ventilator adjustments were insufficient to recruit lung regions, manual recruitment maneuvers were performed by an experienced lung transplant surgeon in a manner consistent with standard practices used during clinical lung transplantation. During inspiration, gentle manual compression was applied to areas of aerating lung, thereby directing ventilation toward areas of non-aerating atelectatic lung to transiently increase the local airway pressure gradient, resulting in recruitment of non-aerated lung volume and increased vital capacity while avoiding barotrauma and the deleterious effects of hyperinflation. Alveolar recruitment was performed as needed, and lungs underwent a maximum recruitment time of approximately 30 minutes, during which peak inspiratory pressures were monitored and maintained below 30 cmH₂O.

Functional analyses of explanted human lungs.

Blood samples were collected from the main pulmonary artery cannula (blood entering left and right lungs), main PV cannula (blood exiting left and right lungs) every 6 hours, and hemogas analysis was performed using a point-of-care blood analysis system (epoc; Siemens Healthineers). Dynamic compliance (C_dyn = TV/(PIP − PEEP) and PaO₂/FiO₂ were calculated every 6 hours. Lung weight was obtained every 6 hours using a top-loading balance (Denver Instrument Company) inside the organ chamber. The organ basin and contents were tared at each time point to ensure accurate measurements. Radiographs of explanted human lungs were acquired using a portable X-ray unit (PXP-16HF; United Radiology Systems) at 2.2 mAs and 90 kVp.

Histopathologic analysis.

Tissue sample collection.

Prior to the start of each experiment, a lung map with pre-determined regions (8 regions per left or right lung) was generated, and each region was assigned an arbitrary number, as previously described._¹² Accordingly, to avoid sampling bias, the location of lung wedge tissue sample collections was randomized using a random number generator (www.random.org). A surgical stapler (GIA Auto Suture; Covidien) with medium/thick reloads was used to obtain lung samples at 0, 12, and 24 hours of xenogeneic cross-circulation. Specimens were immediately fixed in cold 4% paraformaldehyde for 48 – 72 hours, embedded in paraffin, and sectioned at 3 μm or 5 μm thickness. Comparative specimens of deidentified human lung tissue trimmed at the time of transplantation were obtained as paraffin-embedded blocks from the Department of Anatomical Pathology under a protocol approved by the Institutional Review Board at Columbia University Medical Center. All sections were stained with hematoxylin and eosin and examined under light microscopy. Additional sections were stained with silver reticulin, Alcian blue (pH 2), and pentachrome by the histology service in the Department of Molecular Pathology at Columbia University Medical Center.

Blinded pathologic review.

Pathologic review was performed by an experienced lung transplant pathologist blinded to the study protocol. All slides (H&E, immunohistochemical staining for neutrophil elastase and caspase 3, and TUNEL) were randomized, arbitrarily numbered, and delivered to the pathologist without reference to experimental time points or conditions. An established lung injury scoring rubric_^12–14 was used to score each sample (Supplementary Table 3). Scoring criteria were based on histologic analyses of high-power fields (hpf; × 40, 0.2 mm₂). Airway polymorphonuclear cells were evaluated as bronchi and bronchioles containing any neutrophils per hpf (0 ⇒ 0%, 1 ⇒ 1–25%, 2 ⇒ 26–50%, 3 ⇒ >50%). Alveolar polymorphonuclear cells were evaluated as alveoli more than half-filled with neutrophils per hpf (0 ⇒ 0%, 1 ⇒ 1–25%, 2 ⇒ 26–50%, 3 ⇒ >50%). Alveolar edema was evaluated as alveoli with edema per hpf (0 ⇒ <5%, 1 ⇒ 6–25%, 2 ⇒ 26–50%, 3 ⇒ >50%). Interstitial infiltrate was evaluated as the number of lymphocytes and neutrophils in the interstitium around vessels and airways and in alveolar septa and pleura per hpf (0 ⇒ 0, 1 ⇒ <50, 2 ⇒ 50–100, 3 ⇒ >100). Interstitial edema was evaluated as expansion of perivascular and peribronchial spaces with edematous fluid, relative to the width of vessel media, per hpf (0 ⇒ none, 1 ⇒ 1× width vessel media, 2 ⇒ ≥2× width vessel media). Apoptotic cells were evaluated as the percent of early (caspase 3₊) and late (TUNEL₊) apoptotic cells relative to the total number of cells per hpf (0 ⇒ 0–5, 1 ⇒ 6–10, 2 ⇒ 11–15, 3 ⇒ >15).

Immunohistochemical staining.

Human lung sections were de-paraffinized, incubated in boiling citrate buffer (pH 6.0) for antigen retrieval for 15 minutes, and blocked with 10% normal goat serum in phosphate-buffered saline for 2 hours at room temperature. Primary antibodies were added and incubated for 12 hours at 0034°C, or 4 hours at 25°C. For all immunostains, the secondary antibody was diluted 1:200 and incubated for 1 hour at 25°C. Sections were mounted in mounting medium (Vectashield Mounting Medium; Vector Laboratories) with DAPI, coverslipped, and imaged with a fluorescence microscope (DMi8; Leica). Immunostaining was performed by Molecular Pathology Histology Services in the Herbert Irving Comprehensive Cancer Center at Columbia University Medical Center. A list of antibodies and dilutions used is in Supplementary Table 12.

Electron microscopy.

Scanning electron microscopy.

Lung tissue samples were obtained at 0, 12, and 24 hours of xenogeneic cross-circulation, fixed in formalin, rinsed in 70% ethanol, frozen, and lyophilized. Samples were imaged using a scanning electron microscope (GeminiSEM 300; Zeiss) with accelerating voltage of 2.5 kV.

Transmission electron microscopy.

Lung tissue samples were obtained at 0, 12, and 24 hours of xenogeneic cross-circulation, fixed with 2.5% glutaraldehyde, 4% paraformaldehyde, and 0.02% picric acid in 0.1M sodium cacodylate buffer (pH 7.2). Samples were then post-fixed with 1% OsO4 in Sorenson’s buffer for 1 hour, dehydrated, and embedded in Lx-112 (Ladd Research Industries). Sections (thickness: 60 nm) were prepared using an ultramicrotome (PowerTome PT XL; RMC), stained with uranyl acetate and lead citrate, and examined with a transmission electron microscope (JEM-1200 EXII; JEOL). Images were captured with a digital camera (ORCA-HR; Hamamatsu) and recorded with imaging software (Image Capture Engine v.602.569; AMT).

BAL fluid analyses.

BAL fluid sample collection.

BAL fluid samples were collected every 6 hours by wedging a 3.8 mm flexible video bronchoscope (aScope 3; Ambu) into a subsegmental bronchus of the left and right lower lobes of each set of lungs. Sterile normal saline (5 mL) was injected, aspirated, and collected in a sterile specimen trap (Busse Hospital Disposables). Gross photographs of BAL fluid samples were acquired using a digital camera (Sony). The pH of BAL fluid samples was measured with a glass double junction microelectrode (Fisher Scientific). BAL fluid samples were then centrifuged at 3500 rpm for 10 minutes at 4°C. Supernatants were collected and stored at −80°C until further processing.

Cytopathology.

Cellular contents of BAL fluid samples were visualized by periodic acid-Schiff staining of BAL fluid smears.

Molecular pathology.

Total protein concentrations in BAL fluid were quantified by protein assay (Pierce Coomassie Bradford Protein Assay Kit; ThermoFisher Scientific) according to the manufacturer’s instructions. Inflammatory markers in BAL fluid collected from explanted left and right human lungs (Supplementary Table 4–8) were analyzed in triplicate by multiplex cytokine array (Human Cytokine/Chemokine 65-Plex Panel; Eve Technologies).

Immunologic analyses.

Blood analyses.

Blood samples were collected every 6 hours from an auricular arterial line (or femoral arterial line if auricular placement was unsuccessful). Blood gas analysis was performed using a point-of-care blood analysis system (epoc; Siemens Healthineers). Blood samples were collected in blood collection tubes (BD Vacutainer), and analyzed by a diagnostic laboratory (Antech Diagnostics) for complete blood count, comprehensive metabolic panel, and coagulation panel (Supplementary Table 2). Serum inflammatory markers (GM-CSF, IFNγ, IL-1α, IL-1β, IL-1ra, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, IL-18, TNFα) were analyzed in triplicate by multiplex cytokine array (Discovery Assay Pig Cytokine Array; Eve Technologies; Supplementary Table 9). Hemolytic markers (D-dimer, fibrinogen, plasma free hemoglobin) and P-Selectin were analyzed in triplicate by ELISA. A list of ELISA kits used is in Supplementary Table 13.

Lung tissue analyses.

Tissue samples were collected from left and right human lungs at 0, 12, and 24 hours of xenogeneic cross-circulation with and without immunosuppression.\ For quantitative cellular analyses, tissue samples were immunostained for αGal (LS-C63415; LS Bio), CD45 (ab0559; Abcam), CD163 (ab87099; Abcam), human nuclear antigen (ab190710; Abcam), and neutrophil elastase (ab68672; Abcam). Slides were randomized, arbitrarily numbered, and delivered without reference to experimental time points or conditions to a reviewer blinded to the study protocol. Quantification of cells was based on cell counts per high-power field (× 40, 0.2 mm₂)

Functional tests and assays.

Bronchoresponsiveness test.

The effects of methacholine and albuterol on airway smooth muscle tone were assessed after 24 hours of xenogeneic cross-circulation. To ensure that responses were definitively from explanted human lungs, all bronchoresponsiveness tests were conducted shortly after human lungs were disconnected from cross-circulation with swine host and maintained on isolated ex vivo lung perfusion. Methacholine (100 mg kg₋₁; Methapharm) and albuterol (3 mg; Nephron) were administered to explanted human lungs by nebulizer. Airway pressures were continuously recorded to assess changes in peak inspiratory pressures in response to cholinergic or β1/2 agonistic effects of methacholine or albuterol, respectively.

Vasoresponsiveness test.

The effect of selective α₁-adrenergic receptor activation resulting in vasoconstriction of pulmonary vasculature was assessed after 24 hours of xenogeneic cross-circulation. To ensure that responses were definitively from explanted human lungs, vasoresponsiveness tests were conducted shortly after human lungs were disconnected from cross-circulation with swine host and maintained on isolated ex vivo lung perfusion. Phenylephrine (4 mg; International Medication Systems) was administered intravenously into the pulmonary artery cannula of explanted human lungs, and vascular pressures and flows were continuously recorded to assess the changes in pulmonary artery and vein pressures.

CFSE uptake assay.

To assess viability of the pulmonary endothelium in explanted human lungs, carboxyfluorescein succinimidyl ester (CFSE; Affymetrix) was reconstituted in dimethyl sulfoxide at a concentration of 1.06 M, protected from light, and delivered via intravascular injection with a 20 Ga needle after 24 hours of xenogeneic cross-circulation. After 15 minutes of incubation, lung tissue samples were washed five times with phosphate-buffered saline, fixed in cold phosphate-buffered 4% paraformaldehyde for 48 hours, fixed, embedded, de-paraffinized, and imaged as previously described.

Acetylated low-density lipoprotein uptake assay.

To assess function of the pulmonary endothelium in explanted human lungs, biopsies (4 mm) of the left and right pulmonary arteries and pulmonary veins were collected and placed in a 96-well plate (BD Falcon). Acetylated low-density lipoprotein (LDL) with fluorescent Alexa Fluor 594 conjugate (L35353; ThermoScientific) was diluted 1:200 in 1× DMEM/F12K cell culture media (Corning). Wells containing pulmonary vascular tissue biopsies from each lung received 150 μL media with acetylated LDL reagent, or media alone (negative control). The multi-well plate was covered with aluminum foil and incubated at 37°C with gentle shaking for 4 hours. After incubation, samples were washed five times with phosphate-buffered saline, fixed in cold phosphate-buffered 4% paraformaldehyde for 48 hours, embedded in paraffin, sectioned at 5 μm thickness, de-paraffinized, stained with DAPI, and imaged using a fluorescence microscope (FSX100; Olympus).

Surfactant-BODIPY uptake assay.

To assess viability and functional uptake of type II pneumocytes, fluorescent BODIPY-labeled surfactant protein B (SPB-BODIPY) was delivered into distal regions of explanted human lungs after 24 hours of xenogeneic cross-circulation using a flexible bronchoscope and microcatheter system (Renegade; Boston Scientific). After incubation for 30 minutes, a surgical stapler with medium/thick reloads (Medtronic) was used to collect lung tissue samples. Samples were dissected, rinsed in Dulbecco’s phosphate-buffered saline, and imaged immediately with a fluorescence microscope (FSX100; Olympus). Additional images were obtained by incubating endpoint lung tissue samples (2 mm₃) with SPB-BODIPY (20 ng mL₋₁) for 30 minutes at room temperature. Specimens were then stained with plasma-membrane stain (CellMask Deep Red; ThermoFisher Scientific) for 10 minutes, followed by five washes with Dulbecco’s phosphate-buffered saline. Images were acquired with a two-photon confocal laser scanning microscope (TCS SP8; Leica). Visualization of fluorescence signal in punctate patterns indicated surfactant uptake and storage in lamellar bodies of type II pneumocytes.

Metabolic activity assay.

To assess changes in metabolic activity of explanted human lungs, parenchymal lung tissue samples were collected at 0, 12, and 24 hours of xenogeneic cross-circulation. Tissue samples (~250 μL; n = 3 per lung) were dissected in a sterile fashion, finely minced, gently homogenized, and placed in a 96-well plate. Cell viability reagent (AlamarBlue Assay; ThermoFisher) was diluted 1:10 in Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine serum. Wells containing lung sample homogenates from each lung received 100 μL AlamarBlue reagent. Wells containing no lung homogenate (negative control) also received 100 μL AlamarBlue reagent. The multi-well plate was covered with aluminum foil and incubated at 37°C with gentle shaking for 2 hours. After incubation, well supernatants were transferred into new 96-well plates, and absorbance was measured at 570 nm and normalized to 600 nm. To normalized metabolic activity for comparison across samples and timepoints, DNA content of each sample was quantified using a DNA quantification assay (Quanti-iT PicoGreen dsDNA Assay; Invitrogen) according to the manufacturer’s instructions. Samples were digested in papain (250 μg mL₋₁) at 60°C for 4 hours, and mixed with PicoGreen reagent. Fluorescence emission was measured at 520 nm with excitation at 480 nm, and DNA was quantified using a standard curve.

Airway casts.

After 24 hours of xenogeneic cross-circulation, explanted human lungs were disconnected from the cross-circulation system, and the airway compartment was cast with polyurethane (Underwriters Laboratories). Casting material was prepared as a foam mixture of polyurethane and acetone (4:1), with 1 mL organic dye (Polytek Development Corp), and introduced through the endotracheal tube into explanted human lungs using a syringe. Casts were cured overnight at 4°C. Lungs with airway casts were subsequently submerged in 30% sulfuric acid solvent for 15 days to remove surrounding lung tissues from the cast. Sulfuric acid solvent was replaced every 5 days. After 15 days, airway casts were rinsed with water, allowed to dry at 25°C, and imaged with a digital camera (Sony).

RNA sequencing.

Explanted human lung tissue samples (~100 mg, n = 3 per time point) were collected, snap frozen in liquid nitrogen, and stored at −80°C. Samples were shipped on dry ice to an RNA sequencing service provider (Genewiz) for RNA extraction and library preparation with polyA selection. RNA samples were sequenced with 2×150bp configuration and single index per lane (HiSeq; Illumina). Low quality reads and sequencing artifacts were eliminated. Sequencing reads were trimmed (Trimmomatic v.0.36; Usadel Lab)_⁷² to remove possible adapter sequences and nucleotides of poor quality, then aligned to human reference genome GRCh38/hg38 (STAR aligner v.2.5.2b)_⁷³. Raw read counts were normalized by variance stabilizing transformation (DESeq2)_⁷⁴, and differential gene expression was analyzed. Heatmaps with Euclidian-based distance were generated to visualize gene expression (Pheatmap v.1.08 in R v.3.6.0), and a volcano plot was generated to visualize differentially expressed genes (EnhancedVolcano package in R v.3.6.0; Bioconductor)_⁷⁵. Gene ontology analysis was performed to assess changes in gene clusters (GeneSCF v.1.1-p2)_⁷⁶. To enable transcriptomic comparison of explanted human lungs subjected to 24 hours of xenogeneic cross-circulation with normal human lungs, RNA sequencing of tissue samples from representative normal human lungs (n = 50, randomly selected) was accessed through the Genotype-Tissue Expression (GTEx) project_⁷⁷ via the dbGaP Portal using the dbGaP Accession phs000424.v7.p2.

Randomization of sampling.

To avoid sampling bias, the location of lung wedge tissue sample collections was randomized using a random number generator (www.random.org). Prior to the start of each experiment, a lung map with pre-determined regions (8 regions per left or right lung) was generated, and each region was assigned an arbitrary number, as previously described₁₂. Throughout all procedures, samples were collected at each time point according to the lung map regions that corresponded to the randomly generated numbers.

Blinded review.

All analytical assessments were blinded to the maximum practical extent. All pathologic analyses were performed by an independent expert to eliminate bias.

Statistical analyses.

No data were excluded from analysis. One-way ANOVA and Student’s t-tests were performed using statistical analysis software (Prism 8.2.1; GraphPad), and p < 0.05 was considered statistically significant.

DATA AVAILABILITY

The authors declare that all data supporting the findings of this study are available within the manuscript, figures, and Supplementary Information.