Ad J.J.C. Bogers Department of Cardio-Thoracic Surgery, Erasmus MC, Rotterdam, The Netherlands; , Niels P Van der Kaaij Department of Cardio-Thoracic Surgery, Erasmus MC, Rotterdam, The Netherlands; , Jolanda Kluin Department of Cardio-Thoracic Surgery, Erasmus MC, Rotterdam, The Netherlands; , Jack J Haitsma Department of Anethesiology, Erasmus MC, Rotterdam, The Netherlands; , Michael A den Bakker Department of Pathology, Erasmus MC, Rotterdam, The Netherlands; , Bart N Lambrecht Department of Pulmonary Medicine, Erasmus MC, Rotterdam, The Netherlands; , Burkhard Lachmann Department of Anaesthesiology, Erasmus MC, Rotterdam, The Netherlands; , Ron W F de Bruin Department of Surgery, Erasmus MC, Rotterdam, The Netherlands;

Abstract
Background: Lung ischemia-reperfusion injury (LIRI) is suggested to be a major risk factor for development of primary acute graft failure (PAGF) following lung transplantation, although other factors have been found to interplay with LIRI. The question whether LIRI exclusively results in PAGF seems difficult to answer, which is partly due to the lack of a long-term experimental LIRI model, in which PAGF changes can be studied. In addition, the long-term effects of LIRI are unclear and a detailed description of the immunological changes over time after LIRI is missing. Therefore our purpose was to establish a long-term experimental model of LIRI, and to study the impact of LIRI on the development of PAGF, using a broad spectrum of LIRI parameters including leukocyte kinetics. 
Methods: Male Sprague-Dawley rats (n=135) were subjected to 120 minutes of left lung warm ischemia or were sham-operated. A third group served as healthy controls. Animals were sacrificed 1, 3, 7, 30 or 90 days after surgery. Blood gas values, lung compliance, surfactant conversion, capillary permeability, and the presence of MMP-2 and MMP-9 in broncho-alveolar-lavage fluid (BALf) were determined. Infiltration of granulocytes, macrophages and lymphocyte subsets (CD45RA⁺, CD5⁺CD4⁺, CD5+CD8+) was measured by flowcytometry in BALf, lung parenchyma, thoracic lymph nodes and spleen. Histological analysis was performed on HE sections. 
Results: LIRI resulted in hypoxemia, impaired left lung compliance, increased capillary permeability, surfactant conversion, and an increase in MMP-2 and MMP-9. In the BALf, most granulocytes were found on day 1 and CD5⁺CD4⁺ and CD5⁺CD8⁺-cells were elevated on day 3. Increased numbers of macrophages were found on days 1, 3, 7 and 90. Histology on day 1 showed diffuse alveolar damage, resulting in fibroproliferative changes up to 90 days after LIRI. 
Conclusions: The short-, and long-term changes after LIRI in this model are similar to the changes found in both PAGF and ARDS after clinical lung transplantation. LIRI seems an independent risk factor for the development of PAGF and resulted in progressive deterioration of lung function and architecture, leading to extensive immunopathological and functional abnormalities up to 3 months after reperfusion.

Background
Lung transplantation is currently an accepted treatment option for patients with end-stage pulmonary diseases, even though the outcome remains limited ^[1]. Development of primary acute graft failure (PAGF) occurs in 15-30% of lung transplant recipients and is the main cause for early morbidity and mortality after lung transplantation, resulting in a one-year survival rate of approximately 80% [1-3]. Lung ischemia reperfusion injury (LIRI) has been suggested to be a major risk factor for PAGF, although other factors like donor brain death, mechanical ventilation, pneumonia, hypotension, aspiration, donor trauma and allo-immunity have been found to interplay with LIRI in PAGF development ^[1-4]. The clinical expression of LIRI may range from mild hypoxemia and mild pulmonary edema on chest X-ray to PAGF, which is the most severe form of injury ^[1]. Symptoms of PAGF usually develop within 72 hours after reperfusion and consist of hypoxemia, which cannot be corrected by supplemental oxygen, non-cardiogenic pulmonary edema, increased pulmonary artery pressure, and decreased lung compliance ^{[1, 3-5]}.

Even though a positive correlation between cold ischemia time and PAGF development has been suggested ^{[3, 6-8]}, other studies found that duration of cold ischemia did not predict outcome after lung transplantation and suggested that other factors interplay with LIRI in PAGF development ^[9-14]. The question whether LIRI is an independent risk factor for the development of PAGF seems difficult to answer. In clinical studies, often multiple interfering factors are examined simultaneously. Furthermore, a long-term experimental LIRI model, in which PAGF changes can be studied, is missing. The majority of experimental studies use ex vivo LIRI models, like the Langendorff system, which is a nonphysiological model and in which it is impossible to investigate reperfusion times beyond the first hours. In addition, an experimental lung transplantation model with the induction of cold ischemia is technically difficult in rodents. Thus, the purpose of this study was to establish an in vivo model of unilateral severe LIRI and to determine whether symptoms resembling PAGF after clinical lung transplantation could be induced. Although the use of warm rather than cold ischemia seems controversial, it has been demonstrated that there are no major differences between short periods of warm and longer periods of cold ischemia ^[15]. Moreover, warm ischemia has been used extensively in IRI models of liver and kidney as an accelerated model of clinically relevant cold IRI ^[16-19]. Since most studies have only investigated the early hours of reperfusion ^[19-32], the effect of severe LIRI up to months after reperfusion is unknown. Furthermore a detailed description of the subset of leukocytes and the time course of infiltration on both short and long term after LIRI is currently missing. Therefore, we have investigated a broad spectrum of LIRI parameters, including lung function, capillary permeability, matrix metallo proteinase (MMP) production, surfactant conversion, and histological changes on the short (days) and long-term (months) after LIRI and we have described leukocyte kinetics. Finally, in the case of single lung transplantation, the changes in the native lung after transplantation of the contralateral side are not well established, especially on the long term. Therefore, we also assessed changes in nonischemic right lung in animals undergoing left-sided LIRI.

Methods

Experimental design
The experimental protocol was approved by the Animal Experiments Committee under the Dutch National Experiments on Animals Act and complied with the 1986 directive 86/609/EC of the Council of Europe. Male Sprague- Dawley rats (n= 135, weighing 295±4 grams) (Harlan, The Netherlands) were randomised into the experimental LIRI (n=75), sham-operated (n=50) or unoperated (n=10) group. LIRI (n=15 per time point) and sham-operated (n=10 per time point) animals were killed on day 1, 3, 7, 30 or 90 postoperatively. Animals in the LIRI group were subjected to 120 minutes of warm ischemia of the left lung. Sham-operated animals underwent the same protocol as LIRI animals without applying left lung ischemia; unoperated controls were killed without any intervention.

Surgical procedure
Animals were anesthetized with 60 mg/kg of ketaminhydrochloride intraperitoneally and a gas mixture (1,5-3% isoflurane, 57% NO₂ and 40% O₂), whereafter they were intubated and pressure control ventilated on a Siemens Servo 900C ventilator (Maquet Critical Care AB, Solna, Sweden) (14 cm H₂O peak inspiratory pressure (PIP), 4 cm H₂O positive end expiratory pressure (PEEP), frequency 40 breaths/minute, fraction of inspired oxygen (FiO₂) 0.4). Following a left dorsolateral thoracotomy in the fourth intercostal space, the inferior pulmonary ligament was divided. The left lung was mobilized atraumatically, and lung ischemia was induced by clamping the bronchus, pulmonary artery and vein of the left inflated lung using a single noncrushing microvascular clamp. At reperfusion, the lung was recruited by a stepwise increase of PIP and PEEP (maximum respectively 50 and 18 cm H₂O) until the lung was visually expanded. Recruitment was also performed in sham-operated animals. The thorax was closed and the animals received 5 ml of 5% glucose intraperitoneally and 0.1 mg/kg of buprenorphinhydrochloride (0.3 mg/ml) intramuscularly and were weaned from the ventilator. Body temperature was kept within normal range with a heating pad. All animals recovered with additional oxygen during the first 12 hours.

Blood gas values
At the end of the experiment (at day 1, 3, 7, 30 or 90), animals were anesthetized with 20 mg/kg intraperitoneally administered pentobarbital (60 mg/ml) and a gas mixture (3% isoflurane, 64% NO₂ and 33% O₂). After weighing the animals, a polyethylene catheter (0.8 mm outer diameter) was inserted into the carotid artery and a metal cannula was inserted into the trachea. Thereafter, anesthesia was continued with 20 mg/kg pentobarbital intraperitoneally and 0.7 mg/kg pancuronium bromide (2 mg/ml) intramuscularly, whereafter animals were ventilated for 5 minutes (12 cm H₂O PIP, 2 cm H₂O PEEP, frequency 30 breaths/minute and FiO₂ 1.00). Blood gas values were recorded in 0.3 ml heparinized blood taken from the carotid artery (ABL555 gas analyzer, Radiometer, Copenhagen, Denmark). Animals were exsanguinated and euthanised by an overdose of pentobarbital (200 mg/kg), administered intravenously.

Static compliance
The thorax and diaphragm were opened to eliminate the influence of chest wall compliance and abdominal pressure and a static pressure-volume curve (PVC) of the left and right lung together and left lung separately was recorded as described previously ^[33]. The PVC of the individual left lung was conducted by clamping the contralateral hilum. Maximal compliance (C_max) was determined as the steepest part of the lung deflation curve. Maximal lung volume (V_max), corrected for body weight, was recorded at a pressure of 35 cm H₂O.

Broncho-alveolar lavage
Left and right lung were lavaged separately five times with 5 ml sodium chloride containing 1.5 mM CaCl₂. Total recovered volume of BALf was noted. Cell suspensions were centrifuged at 400g and 4^oC for 10 minutes to pellet the cells. Supernatant of BALf was taken and stored at -20^oC for surfactant analysis and the amount of alveolar serum protein.

Cell collection
Left and right lung, thoracic lymph nodes (TLN), and spleen were collected, smashed and suspended in NaCl. Cell suspensions were centrifuged at 400g and 4^oC for 10 minutes to pellet the cells. Red blood cells were lysed with erythrocyte lysis buffer, whereafter the suspension was washed with murine FACS buffer (MFB) (phosphate buffered saline (PBS), 0.05% weight/volume (w/v) sodium azide and 5% w/v bovine serum albumin (BSA)), centrifuged and resuspended in MFB. Cells were counted with a Bürker-Turk cell counter (Erma, Tokyo, Japan).

Flow Cytometry
Pelleted cells (max 1*10⁶ cells per well) were incubated on ice with 2% volume/volume (v/v) normal rat serum (NRS) in MFB for 15 minutes to prevent non-specific binding of Fc-receptors with primary antibodies. Hereafter, cells were washed, centrifuged and surface-stained for 30 minutes at 4^oC in the dark with the following primary mouse anti rat antibodies: biotin conjugated CD5 (OX19¹), phycoerythrin (PE) labelled CD8 (OX8²), fluorescein-isothiocyanate (FITC) labelled CD4 (OX38²), CD45RA-PE (OX33¹), and HIS48¹. After centrifuging and washing, primary staining of the HIS48 and OX-19-Biotin antibody was revealed by secondary staining with respectively goat anti mouse IgM, conjugated to PE (STAR86PE¹) and streptavidin RPE-Cy5 (phycoerythrincychrome) (STAR89¹) for 30 minutes at 4^oC in the dark. Antibodies were obtained commercially from Serotec¹ (Kidlington, United Kingdom) and BD² (Franklin Lakes, New Jersey, USA).

Cellular differentiation was calculated based on morphology (Side SCatter (SSC) for granularity, Forward SCatter (FSC) for size), autofluorescence and specific positive antibody staining. Cells were identified as follows: Lymphocytes low FSC, low SSC, no autofluorescence, and expressing either CD45RA⁺ (Blymphocytes), CD5⁺ (T-lymphocytes), CD5⁺CD4⁺ (helper T-lymphocytes), and CD5⁺CD8⁺ (cytotoxic T-lymphocytes); neutrophils low FSC, intermediate SSC and HIS48+; macrophages as high SSC and FSC and autofluorescent ^[34]. Data were acquired on a FACS Calibur flowcytometer (BD, Franklin Lakes, New Jersey, USA) and were analyzed using CellQuest (BD, Franklin Lakes, New Jersey, USA) and FlowJo software (Tree Star, Ashland, Oregon, USA).

SA/LA ratio
Supernatant of BALf was centrifuged at 4^oC for 15 minutes at 40.000g to separate surface-active surfactant pellet (large aggregate (LA)) from a nonsurface active supernatant fraction (small aggregate (SA)). LA was resuspended in 2ml NaCl, whereafter the phosphorus concentration of LA and SA was determined by phospholipid extraction, followed by phosphorus analysis ^[35].

Protein concentration
The supernatant was further used to determine alveolar protein concentration using the Bio-Rad protein assay (Bio-Rad, Hercules, California, USA) using a Beckmann DU 7400 photospectrometer with a wavelength set at 595 nm (Beckmann, Fullerton, California, USA) ^[36]. Bovine serum albumin was used as standard.

Determination of matrix-metallo-proteinase activity
To determine the activity of MMP-2 and MMP-9, gelatin zymography was performed on BALf of the left lung (n=6 per group, randomly assigned). Zymography was conducted on 10% SDS-polyacrylamide gels containing 1% w/v porcine skin gelatin (Sigma-Aldrich, St. Louis, Missouri, USA). The samples were 1:1 mixed with SDS-PAGE sample buffer (0.25M Tris HCl, pH 6.8, 2% w/v SDS, 20% v/v glycerol, 0.01% v/v bromofenol blue), heated for 3 minutes at 55^oC and subjected to standard electrophoretic analysis at room temperature using the protean II system (Bio-Rad, Hercules, California, USA). After electrophoresis, gels were washed two times for 15 minutes with 2.5% Triton X-100 buffer to renature MMPs by removal of SDS. Hereafter, gels were incubated with development buffer (5mM CaCl₂, 50mM Tris HCl, pH 8.8, 0.02% w/v NaN₃, aquadest) for 20 hours and proteins were fixated for 15 minutes using 45% v/v methanol and 10% v/v acetic acid. Gelatinolytic activity was visualized as clear zones after staining with 0.1% w/v Coomassie Brilliant Blue R-250 in 45% v/v methanol and 10% v/v acetic acid and subsequent destaining in the same solution without Coomassie Brilliant Blue. Gels were scanned (Kodak image station 440cf; Kodak, Rochester, New York, USA) and quantified (Kodak image analysis software). A control sample was used in all gels to be able to compare the various blots. After measuring the band intensity of all blots, values were multiplied by a correction factor, determined by the values of the control sample.

Histology
Histological assessment was performed in 3 animals per group per time point. The heart and lungs were excised en bloc, whereafter the lungs were fixated at a pressure of 10 cm H₂O in 4% paraformaldehyde for 24 hours and embedded in paraffin wax. Sections were cut and stained with haematoxylin and eosin (HE). A histopathologist (MdB), blinded for the treatment, performed histological examination on the following parameters: intra-alveolar and septal edema, hyaline membrane formation, inflammation (classified as histiocytic, lymphocytic, granulocytic, and mixed), fibrosis, atelectasis, intra-alveolar hemorrhage, and overall classification. Each parameter was ranked as mild/scattered, moderate/occasional, or severe/frequent. Sections were overall classified as 1) normal, if no abnormalities were seen, 2) exsudative, if pulmonary edema and/or hyaline membranes were present, 3) fibroproliferative, if activated fibroblasts and/or proliferating alveolar type II cells were found, and 4) resolving, if injury was on return to normal.

Slides were scored on a Leica DMLB light microscope and photographes were taken using a Leica DC500 camera (Leica Microsystems AG, Wetzlar, Germany).

Statistical analysis
The results in text, tables and figures are presented as mean ± standard error of the mean (SEM). Data were analysed using SPSS version 11.1 statistical software (SPSS Inc., Chicago, Illinois, USA). If an overall difference between groups was found by the Kruskal-Wallis test, Mann-Whitney U tests were performed for intergroup comparison. Difference in mortality rate was assessed by the Fisher’s exact test. P values <0.05 were considered to be significant.