Reverse-migrated neutrophils regulated by JAM-C are involved in acute pancreatitis-associated lung injury

Junctional adhesion molecule-C (JAM-C) plays a key role in the promotion of the reverse transendothelial migration (rTEM) of neutrophils, which contributes to the dissemination of systemic inflammation and to secondary organ damage. During acute pancreatitis (AP), systemic inflammatory responses lead to distant organ damage and typically result in acute lung injury (ALI). Here, we investigated the role of rTEM neutrophils in AP-associated ALI and the molecular mechanisms by which JAM-C regulates neutrophil rTEM in this disorder. In this study, rTEM neutrophils were identified in the peripheral blood both in murine model of AP and human patients with AP, which elevated with increased severity of lung injury. Pancreatic JAM-C was downregulated during murine experimental pancreatitis, whose expression levels were inversely correlated with both increased neutrophil rTEM and severity of lung injury. Knockout of JAM-C resulted in more severe lung injury and systemic inflammation. Significantly greater numbers of rTEM neutrophils were present both in the circulation and pulmonary vascular washout in JAM-C knockout mice with AP. This study demonstrates that during AP, neutrophils that are recruited to the pancreas may migrate back into the circulation and then contribute to ALI. JAM-C downregulation may contribute to AP-associated ALI via promoting neutrophil rTEM.

5 mg/kg) immediately after the final injection of caerulein. In this setting, pancreatic inflammation and lung injury associated with AP became more severe than single use of caerulein (100 μ g/kg) ( Fig. 2A-D, Table 1). Definite ALI was observed at both 12 h and 18 h after the administration of caerulein (Fig. 3A, Table 2). The proportion of neutrophils that had undergone rTEM in murine peripheral blood was also significantly elevated at the same time points (Fig. 3B,C). Higher proportion of rTEM neutrophils was also observed when compared with the group of single use of caerulein (Fig. 2E). Histological scores showed that observed lung injuries became increasingly severe as the proportion of rTEM neutrophils increased (Fig. 3D).  Together, these results indicate that elevated proportion of rTEM neutrophil in peripheral blood were positively correlated with the severity of lung injuries in both human and murine AP.

Expression of JAM-C is downregulated in caerulein and LPS-induced pancreatitis.
Previous study showed JAM-C was downregulated and increased neutrophil rTEM in ischemia-reperfusion injury 25 . To determine the role of JAM-C in neutrophil rTEM during experimental pancreatitis, we investigated the expression of JAM-C in the pancreas in caerulein and LPS-induced pancreatitis. Double staining for the endothelial marker PECAM-1 and JAM-C showed that JAM-C is expressed in pancreatic blood vessels (Fig. 4A), consistent with the results of a previous study 26 . After the induction of AP, remarkable interstitial oedema and necrosis of acinar cells were observed, together with significant inflammatory cell infiltration at 12 h and 18 h after the induction of AP (Fig. 4B). JAM-C expression levels decreased at both 12 h and 18 h after the induction of AP compared with control animals (Fig. 4C-E, Supplementary Figure S1). These findings suggest that JAM-C expression in the pancreas is downregulated in caerulein and LPS-induced pancreatitis.
Decreased JAM-C expression is associated with lung injury and increased neutrophil rTEM. After characterizing the changes in reverse-migrated neutrophils and the expression of JAM-C in AP-associated ALI, we investigated the relationship between the expression level of JAM-C and rTEM neutrophil. As shown in Fig. 4F, JAM-C downregulation was associated with elevated peripheral rTEM neutrophil. Reduced levels of JAM-C expression were also associated with increased histological scores for lung injury (Fig. 4G). As an increased level of rTEM neutrophil correlated with the severity of lung injury, these results suggest that during SAP, a decreased level of JAM-C expression may result in a high level of rTEM neutrophils in circulation, which contributes to ALI.

JAM-C deficiency does not affect pancreatic injury in murine experimental acute pancreatitis.
To further explore the role of JAM-C in the regulation of neutrophil rTEM, we established experimental AP in JAM-C-deficient (JAM-C −/− ) mice. Firstly, murine AP was induced by caerulein and LPS. Unexpectedly, no significant difference was found in the extent of pancreatic damage or in the degree of neutrophil infiltration between JAM-C −/− mice and wild-type mice (Fig. 5A, Table 3). No significant difference in serum levels of amylase and lipase were observed between JAM-C −/− and wild-type mice (Fig. 5B).
Leukocytes in inflammatory tissue become activated and produce a series of inflammatory cytokines that contribute to local tissue damage. Thus, we performed qRT-PCR on pancreatic tissue samples to evaluate the levels of TNF-α and IL-6 mRNA. In agreement with the histological findings, no significant difference in the levels of inflammatory cytokines was observed between JAM-C −/− and wild-type mice (Fig. 5C).
In order to exclude specific differences between models, we also established L-arginine (L-Arg)-induced AP. Again, unchanged histological scores, serum concentrations of amylase and lipase and levels of TNF-α and IL-6 mRNA were found between JAM-C −/− and wild-type mice (Fig. 5).

JAM-C deficiency aggravates lung injuries and systemic inflammation in experimental pancreatitis.
Interestingly, we found that lung injuries were more severe in the JAM-C −/− mice than in the wild-type mice following two models of AP. The severity of lung damage and the extent of leukocyte infiltration were significantly increased in JAM-C −/− mice (Fig. 6A, Table 4). The levels of inflammatory cytokines and MPO activity in the lungs were also significantly higher in JAM-C −/− mice (Fig. 6B,C). Because local inflammation is accompanied by transient leukocytosis in the blood, we then examined the numbers of circulating leukocytes after AP induction. The results showed the numbers of circulating leukocytes were significantly increased in JAM-C −/− mice (Fig. 6D). In addition, the serum levels of TNF-α and IL-6 were also significantly higher in JAM-C −/− mice (Fig. 6E), indicating that systemic inflammation is more severe after the loss of JAM-C.
An increase in neutrophil rTEM is enhanced in JAM-C-deficient mice. We have found that JAM-C downregulation may result in increased neutrophil rTEM in peripheral blood during AP. In agreement with this finding, significantly higher proportion of reverse-migrated neutrophils in peripheral blood was observed in JAM-C-deficient mice than in wild-type mice in response to AP (Fig. 7A,B). Similar results were obtained in neutrophils from pulmonary vascular washouts (Fig. 7A,C).
Previous study reported LPS stimulated the expression of ICAM-1 on neutrophils in vitro 27,28 . To explore in our study whether the very high proportion of ICAM-1 high neutrophils in the circulation is a result of activation by LPS, we observed the proportion of ICAM high CXCR1 low neutrophils by single use of LPS. In this experiment, mice were given ten hourly intraperitoneal injections of saline instead of caerulein first, LPS (5 mg/kg) was administered immediately after final injection of saline. Mice were sacrificed 12h after the first saline injection. The results showed that single use of LPS could only slightly increase the number of ICAM high CXCR1 low neutrophils which was far less than caerulein and LPS-induced pancreatitis or single use of caerulein in both wild-type and JAM-C −/− mice. Besides, no difference was observed in the number of ICAM high CXCR1 low neutrophils between the two groups of mice (Fig. 7D). In addition, single use of LPS was not able to induce distinctive pancreatic Alveolar thickening 0 ± 0 0.67 ± 0.23 1.78 ± 0.23* 1.83 ± 0.24* Inflammation 0 ± 0 0.58 ± 0.23 1.72 ± 0.29* 1.80 ± 0.32* Table 2. Histological evaluation of lung injuries following caerulein and LPS-induced pancreatitis. * p < 0.05 versus 6 h group.
or pulmonary damage in both wild-type and JAM-C −/− mice (Fig. 8). These results indicated the majority of ICAM high CXCR1 low neutrophils from the blood may not due to the use of LPS. Taken together, these results suggest that in the murine model of AP, JAM-C downregulation may promote an increase in the rTEM of neutrophils from the inflamed pancreas and these neutrophils re-enter the circulation and contribute to AP-associated ALI.

Discussion
During the response to infection or injury, neutrophils are rapidly recruited from the circulation to sites of inflammation. However, the neutrophils in inflamed tissues can migrate in a retrograde direction across endothelial cells; these cells, moreover, constitute a population of tissue-experienced neutrophils with a distinct surface phenotype (ICAM-1 high CXCR1 low ) and are referred as reverse-migrated neutrophils. It has been reported that neutrophils that have undergone reverse transendothelial migration (rTEM) are unlikely to be able to re-enter inflamed tissue. The apoptosis of rTEM neutrophils is delayed, which means that these cells are "long-lived" in the circulation 24 . More importantly, rTEM neutrophils have been demonstrated to have a greater ability to generate reactive oxygen species (ROS), which contribute to local tissue damage. The characteristics of rTEM neutrophils suggest the possibility that these cells can contribute to the systemic dissemination of inflammation in some types of inflammatory diseases. In acute pancreatitis (AP), acute lung injury (ALI) is the main factor contributing to early death 5,29,30 . Several studies have shown that ALI is the consequence of the systemic inflammatory response 31,32 . Although the role of neutrophils in AP-associated lung injury has been of significant interest, the exact molecular mechanism underlying the effects of these cells remains unclear. In the present study, we provided evidence for the first time that rTEM neutrophils may play an important role in the development of AP-associated lung injury. Firstly, through an analysis of the peripheral blood from patients with AP, we found the levels of rTEM neutrophils were significantly increased in patients with ALI compared with both healthy volunteers and patients with mild acute pancreatitis (MAP). As the disorder progressed, the proportion of rTEM neutrophils in the circulation gradually increased, along with the severity of the lung injuries. These results suggest that during severe acute pancreatitis (SAP), neutrophils that are recruited to the pancreas may migrate back into the circulation, enter the pulmonary vessels and contribute to lung damage. Reverse-migrated neutrophils have also been reported in peripheral blood from patients with active RA and patients with severe atherosclerotic disease 24 . In the murine cremaster ischemia-reperfusion (I-R) model, a high percentage of rTEM neutrophils were found in the pulmonary vasculature, and this finding was associated with I-R-induced lung inflammation 25 . Together, these results suggest that reverse-migrated neutrophils are important in the pathology of some diseases and may play a critical role in the persistence of inflammation in humans, such as in AP-associated ALI.
However, how neutrophils undergo reverse migration is poorly understood. Recently, it was reported that JAM-C, a type of glycoprotein that is primarily expressed in vascular endothelial cells, regulates this process 25 . One study reported that in a murine cremaster I-R model, neutrophil rTEM occurred where JAM-C expression at EC junctions was reduced. When the interaction of JAM-C with JAM-B (another member of the JAM family which maintains JAM-C at endothelial cell junctions) is blocked, the number of neutrophils that reverse migrated increased. The same phenomenon has also been observed in endothelial cell-specific JAM-C-deficient mice 25 . In addition, another study showed that anti-JAM-C blocking mAbs induced the reverse transmigration of monocytes in vitro and in vivo 33 .
In order to make the pancreatitis more severe, we established the AP model with a higher dose of caerulein (100 μ g/kg) plus LPS (5 mg/kg). In this setting, pancreatic inflammation and lung damage was apparent. JAM-C expression began to decrease 12 h after the induction of pancreatitis, which is associated with the extent of lung injury. However, Vonlaufen and colleagues reported that JAM-C is upregulated in single caerulein-induced (50 μ g/kg) AP 26 . In our pilot study, single use of caerulein (50 μ g/kg) was not able to induce apparent pancreatic inflammation (data not shown). We hypothesize that the extent of inflammation leads to different patterns of JAM-C expression. Further studies are needed to clarify the mechanism behind this phenomenon and by which JAM-C regulates neutrophil migration in cases of pancreatitis of different severity.
To further investigate the mechanisms of rTEM neutrophils in AP with ALI, we examined a JAM-C knockout mouse model. JAM-C is currently considered to be an adhesion molecule that controls the transmigration of leukocytes. Antibodies against JAM-C and soluble JAM-C both efficiently block the recruitment of leukocytes in several inflammatory models in vivo, including cutaneous inflammation, peritonitis and lung inflammation [34][35][36] . It was reported that blocking JAM-C with anti-JAM-C antibody blocks the pancreatic leukocyte infiltration of caerulein-induced AP in mice 26 . However, in our study, the deletion of JAM-C did not alleviate murine experimental pancreatitis. Similar results were reported in JAM-C-deficient mice with LPS-induced lung injury 37   cells 38,39 . Moreover, a compensatory increase in the levels of those other molecules involved in the recruitment of leukocytes could exist in JAM-C-deficient mice. Further research would be necessary to address this question. Importantly, we found that JAM-C deficiency leads to more severe lung injury when AP is induced. A significantly higher percentage of rTEM neutrophils were detected in both peripheral blood and pulmonary vascular washout in JAM-C-deficient mice compared with wild-type mice. Similar results were also observed in L-arginine induced-AP. These results provide evidence that rTEM neutrophils play an important role in AP-associated ALI. Neutrophils that are recruited to inflamed pancreatic tissue may undergo reverse transmigration when JAM-C expression falls, as occurs in SAP.
LPS was able to induce the expression of ICAM-1 on neutrophils in vitro 27,28 . However, we found that single use of LPS by intraperitoneal injection can only slightly increase the number of rTEM neutrophils, which was far less than caerulein and LPS or arginine-induced pancreatitis in both wild-type and JAM-C-deficient mice. So the majority of rTEM neutrophils from the blood in murine AP models may not due to the use of LPS. In addition, lung or pancreas injury was not observed in single use of LPS, which suggest pulmonary damage resulted from pancreatitis itself.
Moreover, we found that circulating leukocyte counts were increased in JAM-C-deficient mice, which may in part be due to the elevated numbers of reverse-migrated neutrophils in the peripheral blood. This result is also consistent with reports that have shown that JAM-C-deficient mice were more likely to have increased numbers of circulating neutrophils compared with their wild-type littermates 37,40 . The phenomenon may result from a defect in leukocyte migration 40 and a low level of CXCR4 surface expression on granulocytes after an inflammatory challenge 37 in these mice.
Cytokines play a major role in the pathogenesis of respiratory complications that follow AP 6 . The neutralization and blockade of cytokines reduces the severity of AP and attenuates the systemic stress response in this disorder 9,[41][42][43] . In our study, the loss of JAM-C also resulted in significantly higher levels of serum TNF-α and IL-6 in experimental pancreatitis. TNF-α and IL-6 are secreted by neutrophils, monocytes, macrophages and lymphocytes 44,45 . Leukocytosis in JAM-C-deficient mice may contribute to the high levels of inflammatory cytokines in the circulation. Together, these results indicate that in SAP, the downregulation of JAM-C results in a severe systemic inflammatory response that may lead to ALI. We have reported the occurrence of rTEM neutrophils under the conditions of reduced expression of JAM-C at endothelial cell junctions in a model of severe experimental pancreatitis. These neutrophils may contribute to ALI during AP and thus may represent a novel mechanism of AP-associated ALI. Notably, neutrophils that have undergone rTEM are also found in the blood of patients with AP. JAM-C may be considered a potential target for clinical applications, and thus, it would be useful to explore the role of these neutrophils in human AP further in future studies.

Methods
Mice. JAM-C-deficient (JAM-C −/− ) mice on a C57BL/6 background were purchased from the Jackson Laboratory. The generation of JAM-C −/− mice has been previously described 46 . Male knockout and wild-type mice that weighed 25-30 g at 8-10 weeks of age were used to investigate experimental pancreatitis. All Wildtype and JAM-C −/− mice used in the study were weight matched. All animal experiments were approved by and were performed in accordance with the guidelines of the Animal Care and Use Committee of Shanghai Tongji University.
Induction of experimental pancreatitis. Two AP models were applied in this study.
Caerulein-pancreatitis was induced as previously described 26 . Briefly, mice were given ten hourly intraperitoneal injections of a supramaximal dose of caerulein (Sigma-Aldrich,St. Louis, Missouri, USA, 100 μ g/kg). Lipopolysaccharide (LPS, Sigma-Aldrich, St. Louis, Missouri, USA, 5 mg/kg) was administered by intraperitoneal injection immediately after the 10th injection of caerulein. Mice were sacrificed 6h, 12h and 18h after the first caerulein injection. Arginine-pancreatitis was induced as previously reported 47 . Mice were intraperitoneally injected with L-arginine solution (Sigma-Aldrich, St. Louis, Missouri, USA, 8%, pH = 7.0) with an interval of 1h at a dose of 4 g/kg. Mice were sacrificed 3 days after the induction of acute pancreatitis. Peripheral blood, pulmonary vascular washout fluid, pancreatic tissue and lung tissue were collected.
Leukocyte counts. Peripheral blood leukocyte counts were obtained for blood samples collected after retro-orbital puncture using heparinised capillary tubes. Blood samples were diluted 1:1 with PBS and 1.25 mM EDTA and counted using a Sysmex xe2100 haematology counter (Digitana, AG) as previously described 37 .
Measurements of serum amylase and lipase. The serum activities of amylase and lipase were measured by enzyme dynamics chemistry using commercial kits according to the manufacturer's protocols in a Roche/ Hitachi modular analytics system (Roche, Mannheim, Germany).

Measurement of serum TNF-α and IL-6.
The levels of TNF-α and IL-6 in murine serum were analyzed using a commercially available ELISA kit (eBioscience, San Diego, CA, USA) according to the manufacturer's instructions.
Measurement of Myeloperoxidase Activity. Neutrophil sequestration in the lung tissue was quantified by measuring tissue MPO activity as previously reported 47 .
Histological examination of the pancreas and lung. Tissues from the pancreas and lungs were fixed in 4% phosphate-buffered formaldehyde and stained with haematoxylin and eosin as we previously reported 48 . The samples were examined using light microscopy by two experienced observer who were blinded to the sample identities. The pancreas sections were scored for necrosis, oedema, and inflammation 48 and the lung sections were scored for alveolar thickening and inflammation 49 on a scale of 0 to 3 (0: the least severe and 3: the most severe) for each parameter.
Immunohistochemistry. Paraformaldehyde-fixed, paraffin-embedded samples were sectioned at a thickness of 5 μ m. Rat anti-mouse Ly-6G antibody (BioLegend) was used for immunohistochemistry according to established procedures 50 . The areas in all specimens that stained positive for Ly-6G were examined using a microscope (CTR 6000; Leica, Wetzlar, Germany). Flow cytometry. Peripheral blood was obtained from patients with AP, healthy volunteers and mice. Murine pulmonary washout was obtained as previously described 25 . Samples from patients were stained with antibodies against CD66b, CD16, ICAM-1 and CXCR1 or the respective isotype control antibodies. Murine samples were stained with antibodies against Ly-6G, ICAM-1 and CXCR1 or the respective isotype control antibodies. After incubation, erythrocytes were lysed in a lysis buffer (BD Biosciences), and the surface expression of molecules of interest was measured by flow cytometry, leukocytes were identified by forward and side light scatter characteristics. Neutrophils were identified in mice based on Ly-6G positivity and in humans based on positivity for both CD16 and CD66b. For each sample, 1× 10 4 neutrophils were analyzed for the expression of ICAM-1 and CXCR1. The following antibodies were used for flow cytometry: PE-conjugated rat anti-mouse Ly6G (BD Biosciences), FITC-conjugated hamster anti-mouse ICAM-1 (BD Biosciences), mouse monoclonal anti-CXCR1 (Abcam), PE/ Cy7-conjugated goat anti-mouse IgG2a (Abcam), PE-conjugated mouse anti-human CD16 (BD Biosciences), PerCP-Cy5.5-conjugated mouse anti-human CD66b (BD Biosciences), APC-conjugated mouse anti-human ICAM-1 (BD Biosciences), FITC-conjugated mouse anti-human CXCR1 (BD Biosciences) and the respective isotype control antibodies (BD Biosciences).
Immunofluorescence. Cryostat sections that were 5 μ m thick were fixed in methanol and air-dried. A rabbit anti-mouse PECAM-1 (Abcam) and goat anti-mouse JAM-C (R&D Systems) antibodies were used for immunofluorescence as previously reported 26 . Pictures were acquired using a Zeiss LSM510 laser scanning confocal microscope. Quantitative analysis of staining for JAM-C was performed by Image Pro Plus 6.0.
Western blotting. Frozen pancreatic tissue samples were ground to a powder in liquid nitrogen and then reconstituted in ice-cold RIPA buffer with 1 mmol/L phenylmethanesulphonyl fluoride and protease inhibitors (Sigma-Aldrich). The tissue homogenates were centrifuged and the supernatants collected. Samples (80 μ g per lane) were analyzed by gel electrophoresis, using a 10% acrylamide gel in Tris-HCl buffer, and were then transferred electrophoretically to PVDF membranes. Murine JAM-C was detected using a goat anti-mouse JAM-C antibody (R&D System). Labelled proteins were visualised with an HRP-coupled secondary antibody. The optical densities of the protein lanes were measured using an Odyssey scanning system, and the JAM-C values were adjusted relative to the levels of actin expression.
Real-time quantitative reverse transcription PCR. Total RNA was extracted from each tissue sample using the acid guanidinium/phenol/chloroform method as previously described 48 . RNA was reverse transcribed and subjected to real-time PCR using gene-specific, intron-spanning primers ( Table 5). qRT-PCR was performed in triplicate for each gene of interest under each experimental condition using an ABI Prism 7900HT Sequence Detection System (Applied Biosystems, CA, USA). Fold changes and gene expression levels were calculated using the comparative CT (2 −ΔΔCT ) method.

Patients with AP.
Of the patients who were admitted to the Gastroenterology Department of the Shanghai Tenth People's Hospital from December 2013 to June 2014, those who met the diagnostic criteria of AP 51 were included in the study. Patients with chronic or other acute inflammatory diseases were excluded. Healthy volunteers were assessed as the control group. The severity of AP was assessed on the basis of the revised Atlanta criteria 51 . Arterial oxygen pressures and the inspired oxygen concentration ratio (PaO2/FiO2) were recorded to evaluate the extent of lung injury; a value below 300 defines the presence of ALI. The entire study design and procedures involved were in accordance with the Declaration of Helsinki. Informed written consent was obtained from all AP patients and healthy volunteers. The study protocol was approved by the medical ethics committee of Shanghai Tenth People's Hospital. The methods regarding human subjects were carried out in accordance with approved guidelines and regulations.
Data presentation and statistics. The data are presented as the mean ± SEM. Comparisons between the group means were performed using the Mann-Whitney non-parametric U test. Correlation analyses were performed by linear regression. Values of p < 0.05 were considered statistically significant.