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;

Results

Survival and weight loss
All sham-operated animals survived the experimental period. LIRI resulted in a mortality rate of 25% (0/50 in sham-operated animals versus 19/75 after LIRI, P<0.0001). Non-surviving LIRI animals died shortly after weaning due to the development of pulmonary edema. Surviving LIRI animals had lost more weight on day 3 as compared to sham-operated rats (-34.91±3.86g versus - 21.10±2.86g, P=0.01). From day 7 on these differences had disappeared.

PaO₂ & PaCO₂
Arterial oxygenation was lower in LIRI animals than in unoperated and sham-operated controls on day 1, 3, and 7 (Table 1). On day 30 and 90, these differences had disappeared. An elevated PaCO₂ was found 1 day after LIRI, as compared to unoperated and sham-operated animals.

Static compliance of the left lung
LIRI had detrimental effects on both the Cmax and Vmax of the left ischemic lung as compared to control lungs (Table 2). Up to 90 days after LIRI, V_max and C_max of the left lung remained lower than in sham-operated and unoperated rats.

Capillary permeability
The alveolar serum protein level of the ischemic left lung, as parameter for capillary permeability, was increased 1 day after reperfusion as compared to controls (Table 3). On day 3 the amount of alveolar serum protein in left BALf of LIRI animals was still higher than in unoperated rats. From day seven on, no differences were present.

Matrix metalloproteinase activity
MMP-2 is expressed constitutively in all animals (Figure 1 and 2). However, the total amount of pro- and active MMP-2 and MMP-9 per microliter BALf is increased in LIRI animals on day 1 (Figure 2) (recovered volume did not differ between the groups). MMP activity per microgram protein in the BALf, does not differ between the groups (data not shown), which indicates that the increased activity after LIRI must be due to elevated alveolar serum proteins. After day 3, no differences were demonstrable between the groups.

Surfactant small and large aggregates
While an increase in SA was found in the BALf of the left lung of shamoperated animals on day 1, a higher level was measured in LIRI lungs (Figure 3). After LIRI, an elevated amount of SA was also found in the right lung on day 1. The amount of LA in the left lung was decreased from day 3 until day 30 following LIRI, whereafter the LA level returned to normal on day 90.

Infiltrating cells

Neutrophils
Sham operation resulted in some infiltration of neutrophils in the first days after the operation, as demonstrated by an elevated percentage in left and right BALf and lung tissue. However, after LIRI even more neutrophils were measured in predominantly the left, but also the right BALf (Figure 4A) and lung tissue (Figure 4C). Hereafter the number of neutrophils gradually decreased, and could not be measured anymore on days 30 and 90.

Macrophages
Macrophage occurrence followed similar kinetics in sham-operated and ischemic lungs, but more macrophages were present on day 1 and 3 in ischemic lung tissue and on day 3 and 7 in BALf (Figure 4B and 4D). LIRI also led to an increase in macrophages in the BALf of the contralateral lung on day 3 and 7 as compared to sham and unoperated animals (Figure 4B). Although in sham-operated and LIRI animals macrophages had returned to normal on day 30 in left BALf, they were again elevated on day 90 (Figure 4B).

Lymphocytes
Sham operation did not result in infiltration of lymphocytes in BALf (Figure 5A-C). After LIRI, an infiltration of mainly CD5⁺CD4⁺ and CD5⁺CD8⁺ and to a lesser extent CD45RA⁺-lymphocytes occurred in mainly the left, but also right BALf. Lymphocyte infiltration peaked on day 3, with levels decreasing thereafter (Figure 5A-C).

Although lymphocytes in right lung tissue of LIRI animals followed the same kinetics as in sham-operated animals, demonstrated by a decreased number on day 1 (Figure 5D-F), more CD5⁺CD4⁺ and CD5⁺CD8⁺-cells were found in left lung tissue on day 1 and 3 as compared to sham-operated and unoperated animals (Figure 5D-E). On day 1 also more CD45RA⁺-cells were present in the left lung of LIRI animals (Figure 5F). On day 90, the level of CD5⁺CD4⁺, CD5⁺CD8⁺, and CD45RA⁺ lymphocytes in left lung tissue of LIRI animals had decreased as compared to controls (Figure 5D-F).

No differences were found between groups in percentage or total number of cells within the spleen (data not shown). However, more CD5⁺CD4⁺, and CD5⁺CD8⁺-cells were measured in TLN on day 3 (Figure 6A-C). Whereas CD5⁺CD4⁺ and CD5⁺CD8⁺-cells remained higher in LIRI animals than in unoperated animals up to day 90, CD45RA⁺-cells had returned to preoperative values on day 90.

Histology
LIRI resulted in diffuse alveolar damage consisting of severe intra-alveolar edema up to day 3, septal edema, which was mild on day 1 and increased to moderate on day 3, and intra-alveolar hemorrhages (Figure 7). The overall classification of LIRI animals changed from exsudative on day 1 to proliferative from day 3 to day 90. Although no atelectasis and fibrosis were seen on day 1 following LIRI, mild fibrosis and mild to severe atelectasis were seen from day 3 up to day 90 after LIRI (Figure 8). Identification of infiltrating cells confirmed the flowcytometry measurements. A mild inflammatory pattern consisting of histiocytes was found on day 3 and 7 in sham-operated animals.

LIRI caused moderate to severe inflammation, which changed from mixed (granulocytic, lymphocytic, and histiocytic) inflammation on day 1 to a histiocytic and lymphocytic pattern from day 3 to 90 (Figure 9). No major differences between unoperated, shamoperated and LIRI animals were found in the right lung (data not shown).

Discussion
This study describes the effect of warm LIRI on a broad spectrum of LIRI parameters, such as lung function, capillary permeability, MMP production, surfactant conversion, and histology on the short and long term after LIRI. Furthermore, a detailed description of the subsets of leukocytes and the time course of infiltration on both short and long term after LIRI is given. LIRI has been suggested to be a major risk factor for PAGF. The clinical course of PAGF symptomatically resembles the acute respiratory distress syndrome (ARDS) and can be characterized by different stages, each with their specific clinical, histological and immunological changes ^[37]. The acute, exsudative phase is featured by a sudden onset of hypoxemia, decreased lung compliance, increased pulmonary artery pressure, and development of non cardiogenic pulmonary edema ^{[37, 38]}. Experimental studies have shown that abnormalities in, and depletion of pulmonary surfactant contribute to these symptoms of LIRI ^{[26-28, 39-41]}. Histological analysis of LIRI shows diffuse alveolar damage with atelectasis, inflammation, intra-alveolar hemorrhage, formation of hyaline membranes and protein-rich edema ^{[1, 37]}. Finally, production of matrix-metalloproteinases (MMP) is thought to be important in the acute phase of PAGF since MMPs increase the microvascular permeability and thereby enable extravasation of inflammatory cells ^[42-45].

We found hypoxemia, impaired left lung compliance, a mortality rate of 25% due to development of severe pulmonary edema, and an increase in SA subtype surfactant as early effects of LIRI. Conversion of highly surface active LA into poor surface active SA occurs shortly after reperfusion and is partly due to increased capillary permeability, resulting in influx of serum proteins into the alveolus, as confirmed in our study ^{[28, 46]}. Serum proteins inhibit surfactant in a dose-dependent manner by competing with surfactant components at the alveolo-capillary barrier ^[47]. HE slides confirmed extensive alveolar and septal edema, intra-alveolar bleeding, atelectasis and inflammation, which are all indicative of diffuse alveolar damage. The increase in alveolar proteins and neutrophils on day 1 occurred simultaneously with an increased MMP activity.

MMP-2 is usually constitutively expressed by endothelial cells, vascular smooth muscle cells and fibroblasts, fitting with the observation in our study of high levels of both pro-, and active MMP-2 found in unoperated animals ^{[42-45, 48]}. Yano et al demonstrated that LIRI resulted in MMP-9 induction, but not MMP-2 expression 24 hours after LIRI ^[45]. A higher concentration of MMP-9 was also found in lung edema fluid of ARDS patients ^[49]. Nevertheless, we found that levels of both pro- and active MMP-9 and MMP-2 per microliter BALf are elevated 24 hours after LIRI, which correlates well with the presence of neutrophils in left lung BALf. Thus 120 minutes of warm ischemia resulted in our experimental study in a mortality rate of 25%, hypoxemia, early impaired left lung compliance, surfactant conversion, diffuse alveolar damage on HE slides and MMP production, which are all features of the exsudative phase of PAGF.

Human lung transplant patients surviving the acute phase of LIRI may either recover from injury or enter a ‘chronic’ fibroproliferative state, which develops within 4-7 days after the onset of symptoms ^{[37, 38]}. The progression from an exsudative phase to a ‘chronic’ fibroproliferative state within one week after LIRI is supported in our study by the presence of fibroproliferative changes on HE slides in the first week after LIRI and an increased number of macrophages, which are important mediators in the regulation of fibroblast function. Importantly, LIRI induced progressive changes resulting in extensive pulmonary injury up to 3 months after reperfusion. This is demonstrated by a decreased number of lymphocytes found in lung tissue on day 30 and 90, impaired left lung compliance up to day 90, extensive atelectasis on HE slides, and a decreased surfactant recycling and secreting capacity of alveolar type II cells reflected by the decreased LA surfactant subtype ^[50]. Although extensive left pulmonary injury was found on the long-term, hypoxemia was demonstrated up to day 7. Thereafter, no differences in PaO2 were measured between LIRI animals and controls. This discrepancy may be explained by the fact that PaO₂ was dependent on both lungs, so that the loss of left lung function was compensated by the right lung. Furthermore, even though MMPs are important mediators of pulmonary remodeling, no changes in activity on the long-term were found. It is questionable however whether MMPs are present in the BALf of severe atelectatic and fibrotic lungs. Therefore, in future studies, measurement of MMP activity should also be performed in homogenized lung tissue.

Thus, 120 minutes of warm ischemia in this model induces injury comparable to PAGF and ARDS in clinical lung transplantation on the short, but also on the long-term. Nevertheless, this experimental model has several shortcomings we wish to address. First of all, we used warm ischemia to induce LIRI, whereas in the clinical setting cold ischemic time is associated with PAGF. However, it has been demonstrated that there are no major differences between short periods of warm and longer periods of cold ischemia and warm ischemia has been used extensively in IRI models of liver and kidney as an accelerated model of clinically relevant cold IRI ^[15-19]. Another disadvantage is that a rather long period of 120 minutes warm ischemia has been used. Shorter periods of warm ischemia have been investigated in a pilot study to setup our model (data not shown) and we found that 120 minutes of warm ischemia is necessary in our hands to induce symptoms comparable to PAGF. This finding is supported by a clinical study by Thabut et al, which shows that the relationship between cold graft ischemic time and survival appears to be of cubic form with a cut off value of 330 minutes ^[3]. Thereafter short-term mortality increases rapidly mainly due to development of PAGF ^[3].

Another goal of this study was to describe leukocyte kinetics following LIRI. The immunologic effects of LIRI have only been studied up to hours after reperfusion, whereas we investigated leukocyte kinetics after LIRI up to 90 days post-reperfusion. Several studies have shown that macrophages are activated during ischemia, followed by the recruitment of neutrophils within hours after the start of reperfusion ^{[16, 19, 20, 22, 24, 25, 29, 51, 52]}. We now add that neutrophil infiltration lasted for 3 days after reperfusion, thereby strengthening the theory that neutrophils are important in perpetuating LIRI. The extended presence of neutrophils after LIRI may be also explained by the fact that phagocytes are important elements of the repair process after LIRI by clearing apoptotic cells and necrotic debris ^[29]. Nevertheless, since ischemia-reperfusion injury still develops in neutropenic models, it is questionable whether neutrophils are pivotal in LIRI ^{[21, 32]}.

In this regard, other studies suggest an early role for T cells as important mediators of ischemia induced injury. An infiltration of CD4⁺-T-cells occurred in these studies from 1 until 12 hours after reperfusion and disappeared hereafter ^{[16, 19, 20, 51]}, whereas we found elevated numbers of mainly CD5⁺CD4⁺ and CD5⁺CD8⁺ T-lymphocytes 3 and 7 days after reperfusion in the left lung and TLN. Infiltration and activation of T cells has been classically attributed to the presence of antigen; our findings in an autologous setting may be explained by different mechanisms. First, LIRI induced upregulation of adhesion molecules may attract T cells, such as effector and memory T-cells, which are able to proliferate in an antigen independent fashion by cytokines produced locally (bystander effect), in contrast with naïve T-cells which require antigen presentation by antigen presenting cells ^[53]. Moreover, antigen specific T-cells may be attracted by released self-antigens ^{[54, 55]}. Myosin, heat shock proteins and type V collagen, which is released after LIRI in BALf comparable to the level observed in allografted lungs, have been shown to be capable of inducing a Thelper- cell reaction within days after reperfusion ^{[54, 55]}. The latter is supported by the finding of Waddell and colleagues, who demonstrated upregulation of major histocompatibility II complex after LIRI ^[56]. Finally, the elevated levels of CD5⁺CD4⁺ and CD5⁺CD8⁺ may be explained by their possible role in the pathogenesis of lung fibrosis, which is also supported by the presence of macrophages in the BALf and lung parenchyma of ischemic animals 3, 7, and 90 days after reperfusion, since they are also thought to be important mediators in the regulation of fibroblast function ^{[37, 38, 57, 58]}.

Our study furthermore demonstrates an immunosuppressive effect of operation, as measured by the decreased number of lymphocytes in lung tissue on day 1. Although it is very well known that major surgery may cause a short lasting decrease in blood circulating lymphocytes ^[59], we now additionally report that thoracotomy causes a one day decrease in the number of lymphocyte subset in lung parenchyma, while the number of lymphocytes in the BALf of sham-operated animals is close to normal. The immunosuppressive effect of surgery may be due to reduced T-cell proliferation and reduced secretion of interleukin-2 (IL-2), IL-4, and gamma interferon by T-lymphocytes, which may be the effect of inhibitory factors secreted by mononuclear phagocytic cells as a result of injury ^[60]. Moreover, altered migration of memory and activated effector T cells to injury sites may have also contributed to the decreased level of measured cells ^[61].

Finally, another interesting point arising from this study is the effect of left LIRI on right-sided pulmonary injury. While no major changes were seen on HE slides of the right lung, the inflammatory profile of the right BALf resembled that of the left, although it was less severe. Also, an increased amount of SA was measured in these parts of the lung, demonstrating that the right lung has sustained injury. Since the right lung did not sustain ischemia, induction of systemic components, similar to that seen after mesenteric artery ischemia, may have caused right lung injury. In this regard, induction of high-mobility group-1 protein, a downstream pro-inflammatory cytokine produced by necrotic cells ^{[62, 63]}, and production of uric acid could explain this phenomenon ^[64].

Furthermore, activated neutrophils lose their ability to deform, so that they might have plugged the capillaries of the right lung and may have subsequently caused lung injury ^[65]. However, LIRI of the left lung did not result in long-term damage in the right lung.

Conclusions
The short and long-term changes after LIRI in this model resemble those found in both PAGF and ARDS after clinical lung transplantation. Thus LIRI seems a major risk factor for PAGF in the absence of other influencing factors, such as alloimmunity. Importantly, LIRI resulted in progressive deterioration of lung function and architecture, leading to extensive immunopathological and functional abnormalities up to 3 months after reperfusion. Immunologically, LIRI caused neutrophil infiltration early after reperfusion, followed by T lymphocytes and macrophages. The non-ischemic lung also showed signs of inflammation on the short-term, but to a lesser extent, and long-term changes were not found in the right lung.

Abbreviations
ARDS=Acute Respiratory Distress Syndrome, BALf=Broncho Alveolar Lavage Fluid, BSA=Bovine Serum Albumine, C_max= Maximal Compliance of the expiration curve, FACS=Fluorescence Activated Cell Sorter, FiO₂=Fraction of inspired Oxygen, FITC=Fluorescein-IsoThioCyanate, FSC=Forward Scatter, HE=Haematoxylin and Eosin, IL=InterLeukin, LA=Large Aggregate surfactant subtype, LIRI=Lung Ischemia Reperfusion Injury, MFB=Murine FACS Buffer, MMP= Matrix Metallo Proteinase, NRS=Normal Rat Serum, PAGF=Primary Acute Graft Failure, PaO₂=Arterial Oxygen Pressure, PBS=Phosphate Buffered Saline, PE=PhycoErythrin, PE-Cy5=PhycoErythrin-Cychrome 5, PEEP=Positive End Expiratory Pressure, PIP=Peak Inspiratory Pressure, PVC=Pressure Volume Curve, SA=Small Aggregate surfactant subtype, SSC=Side Scatter, TLN=Thoracic Lymph Nodes, V/V=Volume/Volume, V_max=Maximal Lung Volume at a pressure of 35 cm H₂O, W/V=Weight/Volume

Competing interests
All authors declare that they have no competing interests. Furthermore, no external funding was provided.

Authors’ contributions
All authors were involved in the experimental design, interpretation of the data and in the preparation of this manuscript. Furthermore, all authors read and approved the final manuscript. NPvdK operated the animals, collected and analyzed the data, and prepared the manuscript. JK and AJJCB participated in the cardiothoracic approach of this model. JJH and BL took care of the anaesthetic part of this model and performed the surfactant and protein analysis of the supernatant. MAdB performed all histological analysis. BNL contributed to the immunological analysis of LIRI and was of essential help in analysis of the FACS data. RWFdB performed all MMP measurements.

Acknowledgement
The authors thank Laraine Visser-Isles for correcting the English language.