|Home | About | Journals | Submit | Contact Us | Français|
Acute respiratory infections are responsible for more than 4 million deaths each year. Neutrophils play an essential role in the innate immune response to lung infection. These cells have an armamentarium of pattern recognition molecules and antimicrobial agents that identify and eliminate pathogens. In the setting of infection, neutrophil triggering receptor expressed on myeloid cells 1 (TREM-1) amplifies inflammatory signaling. Here we demonstrate for the first time that TREM-1 also plays an important role in transepithelial migration of neutrophils into the airspace. We developed a TREM-1/3–deficient mouse model of pneumonia and found that absence of TREM-1/3 markedly increased mortality following Pseudomonas aeruginosa challenge. Unexpectedly, TREM-1/3 deficiency resulted in increased local and systemic cytokine production. TREM-1/3–deficient neutrophils demonstrated intact bacterial killing, phagocytosis, and chemotaxis; however, histologic examination of TREM-1/3–deficient lungs revealed decreased neutrophil infiltration of the airways. TREM-1/3–deficient neutrophils effectively migrated across primary endothelial cell monolayers but failed to migrate across primary airway epithelia grown at the air-liquid interface. These data define a new function for TREM-1 in neutrophil migration across airway epithelial cells and suggest that it amplifies inflammation through targeted neutrophil migration into the lung.
Acute respiratory infections have been described as the forgotten pandemic, responsible for 4.5 million deaths annually. Effective neutrophil migration into the intrapulmonary airspace is essential for pathogen elimination in the lung. Neutrophils possess diverse tools, including ROS intermediates, antimicrobial peptides, neutrophil extracellular traps, and numerous pattern recognition receptors (PRRs), that facilitate the response to de novo infections (1, 2). Moreover, neutrophils participate in B cell maturation and differentiation, Th17 responses, dendritic cell maturation, natural killer cell responses, and IL-10 production, revealing a regulatory role for neutrophils (2). Like other cells of the innate immune system, neutrophils express several large genetically encoded receptor families that act to detect and eliminate pathogens. These families include the TLR family, C-type lectin receptors, and the triggering receptor expressed on myeloid cells (TREM) family (3–5).
TREM-1, which was initially discovered on human neutrophils and monocytes, has been studied in patients with pneumonia and sepsis (6, 7). Increased TREM-1 levels in bronchoalveolar lavage (BAL) fluid strongly correlate with ventilator-associated pneumonia, and systemic levels of TREM-1 in the serum serve as a marker of sepsis (8, 9). TREM-1 has been shown to amplify the inflammatory cascade in response to infection (10). Surface expression of TREM-1 is increased in response to LPS and other microbial products in vitro. Antibody-mediated crosslinking of TREM-1 induces modest cellular activation and proinflammatory cytokine secretion. Furthermore, when TREM-1 acts in concert with PRRs, there is a synergistic increase in inflammatory signaling (10–13). Such data suggest that TREM-1 functions as an amplifier of the immune response in the context of microbial infection.
During the process of pathogen recognition and elimination, excessive cytokine and chemokine release can lead to extensive tissue damage and even death (14, 15). Therapeutic agents designed to temper the deleterious consequences of host-pathogen interactions result in a survival benefit in some experimental models of endotoxemia. On the other hand, it is clear that some inflammation is necessary for bacterial killing in vivo. Accordingly, numerous clinical examples of impaired neutrophil function are linked to increased rates of infection and death, including chronic granulomatous disease (CGD), neutropenic fever, and lymphocyte adhesion deficiency. However, animal studies indicate that effective bacterial clearance can be achieved in the setting of decreased inflammation (16, 17). In the case of TREM-1, blockade of TREM-1 has been achieved by administering soluble forms of TREM-1, small molecule blocker (LP-17), or RNA silencing (10, 18, 19). All three of these interventions result in decreased systemic cytokine production and improved survival in mice. Taken together, these animal studies of TREM-1 blockade suggest that it may be possible to modulate the TREM-1 pathway to decrease inflammation without sacrificing bacterial control.
We hypothesized that TREM-1 deficiency would decrease inflammation and improve survival in a murine pneumonia model. Paradoxically, we found that TREM-1/3 deficiency resulted in a marked increase in mortality. Our data suggest that impaired neutrophil transepithelial migration is the mechanism underlying the failure to clear lung bacteria, resulting in increased mortality with TREM-1/3 deficiency. These findings identify a novel role for TREM-1 in neutrophil epithelial transmigration in addition to its established role in amplifying inflammatory signaling.
The goal of these studies was to elucidate the role of TREM-1 in pneumonia using a mouse model of TREM-1 deficiency. In mice, the Trem1 gene is adjacent to a highly homologous gene, Trem3, which likely arose from a duplication event. The two receptors have similar cellular distribution and associate with DAP12. Further suggesting that these two proteins have redundant functions in the mouse, functional studies in our laboratory and others show that both molecules are amplifiers of inflammatory signaling (20). In contrast, Trem3 is a pseudogene in humans, and therefore no functional overlap exists. Thus, to model the effect of blocking TREM-1 in humans, we generated a TREM-1/3–deficient mouse. To abrogate expression of functional TREM-1 and TREM-3 proteins, we made a targeting construct in which an 8.8-kb fragment containing most of exon 3 and exon 4 of Trem1 as well as exon 1 of Trem3 was replaced with a MC1neopA gene flanked by loxP sites (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI64181DS1). This effectively deleted the transmembrane region of TREM-1 in addition to the start site and first domain of TREM-3. Chimeras generated from a correctly targeted clone were bred to transgenic mice expressing Cre under the CMV promoter on a C57BL/6 background to delete the MC1neopA gene. Homozygote TREM-1/3–deficient mice failed to express TREM-1 and TREM-3 proteins (Supplemental Figure 2). Inheritance of this mutation followed expected Mendelian distribution, with 25% homozygotes, 21% WT, and 54% heterozygotes assayed (n = 229). Flow cytometric analysis using lineage markers revealed no significant differences in the percentages of circulating leukocyte populations as a result of Trem1 and Trem3 deletion (Supplemental Figure 3).
To examine the role of TREM-1 in survival in response to pneumonia, we challenged TREM-1/3–deficient mice, WT littermate controls, or WT C57BL/6 mice intratracheally with P. aeruginosa. Survival analysis performed using Gehan-Breslow-Wilcoxon statistical tests showed a significant increase in mortality in TREM-1/3–deficient mice as compared with WT mice, with χ2 = 18.75 and P < 0.0001 (Figure (Figure1A).1A). Additional clinical indicators of sepsis including body temperature and evidence of end-organ damage were monitored. TREM-1/3–deficient animals were hypothermic (Figure (Figure1B,1B, P = 0.0001), consistent with the development of systemic disease. TREM-1/3–deficient animals also showed evidence of end-organ damage in the form of transaminitis as measured by serum alanine aminotransferase (ALT) levels, whereas WT animals had no elevation (Figure (Figure1C).1C). These data indicate that TREM-1/3 expression is essential for host survival following pulmonary challenge with P. aeruginosa. Challenge with heat-killed P. aeruginosa resulted in no mortality in WT or TREM-1/3–deficient mice (n = 4 in each group, data not shown), suggesting that the difference in survival observed in the TREM-1/3–deficient mice was due to a decrease in the ability to control active infection and not increased inflammation in the setting of endotoxin challenge.
To assess bacterial clearance, we performed bacterial sampling of BAL fluid using whole lung lavage on anesthetized animals at either 4 or 14 hours after infection. At 4 hours after infection, both TREM-1/3–deficient and WT BAL fluid (Figure (Figure2A)2A) showed growth of P. aeruginosa. However, at 14 hours after infection, P. aeruginosa CFU were higher in the lungs of TREM-1/3–deficient animals (Figure (Figure2B,2B, P = 0.011). In addition, P. aeruginosa CFU were also significantly higher in cultures obtained from whole lung suspensions generated 14 hours after infection in TREM-1/3–deficient as compared with WT mice (Figure (Figure2C,2C, P = 0.036). Next, blood was obtained by cardiac puncture 14 hours after infection to assess systemic bacteremia (Figure (Figure2D).2D). TREM-1/3–deficient animals had a higher incidence of bacteremia (53% TREM-1/3–deficient vs. 27% WT). Thus, TREM-1/3–deficient mice fail to eradicate P. aeruginosa infection locally, resulting in systemic dissemination of bacteria and death.
To assess whether impaired neutrophil function was the mechanism for poor local and systemic control of P. aeruginosa in the TREM-1/3–deficient animals, we measured phagocytosis and in vitro neutrophil killing of P. aeruginosa. First, phagocytosis was measured using Texas red–zymosan. Neutrophils were isolated from TREM-1/3–deficient and WT bone marrow (pooled 3–5 per group). There were no significant differences in the percentage of neutrophils that ingested particles (Figure (Figure3A)3A) or in the number of particles phagocytized per neutrophil as determined by the geometric mean of Ly6G-positive cells (Figure (Figure3B).3B). To distinguish between intracellular and extracellular particles, we performed fluorescence quenching with trypan blue (21). No significant difference in quenching was detected between WT and TREM-1/3–deficient neutrophils (Supplemental Figure 4). Next, to examine the capacity of murine neutrophils to kill P. aeruginosa, we isolated neutrophils and co-incubated them with bacteria (22). At 30 minutes, more than 90% of the bacteria were ingested. Non-ingested bacteria were killed with gentamicin, as previously described (23). Intracellular bacterial killing was assessed at 30, 60, and 90 minutes after exposure to gentamicin. There was no significant difference in the ability of TREM-1/3–deficient and WT neutrophils to kill P. aeruginosa (Figure (Figure3C).3C). Finally, to determine how TREM-1/3 deficiency affects neutrophil ROS production, we isolated neutrophils from WT and TREM-1/3–deficient animals and stimulated them with opsonized zymosan and PMA. In the presence of no agonist, neither set of neutrophils produced ROS (Supplemental Figure 5), indicating that the neutrophils had not been activated in the isolation process. However, TREM-1/3–deficient neutrophils had a 50% reduction in ROS production in response to opsonized zymosan as compared with WT neutrophils (Supplemental Figure 5), whereas ROS production in response to PMA was intact (Supplemental Figure 5). It is well established that neutrophils eliminate P. aeruginosa through a number of O2-independent pathways, including bactericidal/permeability-increasing protein (BPI), proteases, lipases, and defensins (24). In fact, neutrophils isolated from patients with CGD (an inherited defect in NADPH oxidase resulting in severely impaired ROS production) effectively kill P. aeruginosa, indicating that ROS production is not required in order to effectively kill this organism (24). Therefore, given the in vitro killing data, the striking in vivo phenotype in the TREM-1/3–deficient mice is likely not due to an intrinsic neutrophil killing defect.
Histologic examination of lungs harvested from live mice at 14 hours after infection showed an unexpected difference. Lung tissue from 5 TREM-1/3–deficient animals and 5 WT animals was analyzed by a veterinary pathologist unaware of sample identities. The degree of inflammation was quantified as described in the Methods. TREM-1/3–deficient animals had less evidence of inflammation/pneumonia as compared with WT animals (Figure (Figure4,4, A and B). Moreover, TREM-1/3–deficient animals had reduced leukocyte infiltration by morphometry (P = 0.016) and fewer necropurulent foci (P = 0.05) present in the lungs as compared with WT animals (Figure (Figure4,4, C and D, respectively). Taken together, these data indicate that, despite increased bacterial load and higher mortality, TREM-1/3–deficient animals have impaired pulmonary inflammation in response to P. aeruginosa challenge.
To assess leukocyte migration into the lung, first we asked whether there was a defect in wbc translocation into the lung. Single-cell suspension of lungs from WT and TREM-1/3–deficient animals were prepared to quantitate total lung wbc. TREM-1/3–deficient animals had similar numbers of eosinophils, macrophages, NK cells, CD4+ T cells, CD8+ T cells, and B cells (Figure (Figure5A,5A, n = 15). Unexpectedly, the number of total lung neutrophils was significantly higher in TREM-1/3–deficient animals (Figure (Figure5A,5A, P = 0.001). Therefore, despite having more total lung neutrophils, TREM-1/3–deficient mice failed to develop pneumonia with consolidation and were unable to clear P. aeruginosa lung infection, resulting in a marked increase in mortality.
Next, we determined using BAL whether TREM-1/3–deficient mice have defective neutrophil translocation into the airspace. More than 96% of all cells from the BAL fluid of both groups were Ly6G positive and CD115 negative (data not shown), consistent with murine neutrophil identity. TREM-1/3–deficient mice had a 50% reduction in the number of airway neutrophils (Figure (Figure5B,5B, P = 0.004), suggesting impaired migration of neutrophils into the airspace. We then used a stereological approach to further localize neutrophils in the compartments of the lung by specifically examining the interstitium, the airway lumen (terminal bronchioles), and the alveolar space. To assess the location of neutrophils in the distal airways, we identified terminal bronchioles and a sequestration index was calculated to assess the relative number of neutrophils in the interstitium versus the distal airway lumen. Terminal bronchioles from TREM-1/3–deficient mice had a higher sequestration index as compared with WT mice (Figure (Figure5C,5C, n = 13, P = 0.001), indicating that a higher proportion of neutrophils was present in the interstitial space as compared with the lumen and suggesting a neutrophil migratory block into the bronchiolar airway. There was no detectable difference in the number of neutrophils in the alveolar compartment in WT and TREM-1/3–deficient mice (Figure (Figure5D).5D). The features of consolidation and alveolar collapse were at times confounding issues, and it remains unclear whether a migratory block at the alveolar level exists in TREM-1/3–deficient mice. Taken together, the histologic data and BAL data indicate that, with TREM-1/3 deficiency, there is a defect in neutrophil migration into the airspace.
Because TREM-1 acts synergistically with TLR and other PRRs to augment proinflammatory cytokine production, we hypothesized that a blunted cytokine response to infection with P. aeruginosa would be responsible for decreased airway neutrophils in TREM-1/3–deficient animals. Cytokines were sampled in the plasma, BAL fluid, and total lung homogenates. To avoid survivor bias, all samples were collected at 14 hours following infection, when all animals in both groups were alive. First, to ascertain whether decreased neutrophil migration was the result of a deficiency in local chemokine production, we obtained BAL fluid from infected mice. CXCL1/KC, CXCL2/MIP-2, and CCL2/MCP were elevated in TREM-1/3–deficient as compared with WT mice, consistent with sufficient local production of chemoattractants (Figure (Figure6,6, BAL: P = 0.048, P = 0.017, and P = 0.004, respectively). In separate experiments, total tissue cytokine levels were measured in lungs harvested from anesthetized TREM-1/3–deficient and WT mice after perfusion with PBS from the right ventricle to remove intravascular blood (Figure (Figure6,6, Lung homogenates). TREM-1/3–deficient mice had higher levels of IL-1α (P = 0.038), IL-1β (P = 0.038), IL-6 (P = 0.015), IL-10 (P = 0.05), IL-17 (P = 0.022), and TNF-α (P = 0.007) and chemokines CXCL1/KC (P = 0.02), CXCL2/MIP-2 (P = 0.012), and CCL2/MCP-1 (P = 0.035). Based on these results and those shown in Figures Figures44 and and5,5, we conclude that despite elevated local cytokines and chemokines, TREM-1/3–deficient animals have ineffective accumulation of neutrophils in the airway lumen.
Additionally, systemic cytokine levels were measured (Figure (Figure6).6). The TREM-1/3–deficient mice had statistically significant increases in the proinflammatory cytokines G-CSF, IL-6, IL-1α, CCL5/RANTES, TNF-α, IL-12p40 (P = 0.004, data not shown), and IL12-p70 (P = 0.020, data not shown); increased levels of chemoattractants CXCL1 /KC, CCL2/MCP-1, CXCL2/MIP-2; and increased anti-inflammatory cytokine IL-10. These data indicate that the cytokine production was not impaired in the setting of infection despite the loss of the TREM-1/3 amplifying pathway. Moreover, these data indicate that impaired cytokine production was not responsible for the decrease in neutrophils in the airway lumen of TREM-1/3–deficient animals.
To investigate whether the diminished number of neutrophils in the airways of TREM-1/3–deficient animals was due to increased cell death, we isolated neutrophils from TREM-1/3–deficient and WT animals and determined cell death by propidium iodide uptake as previously described (25). We observed no significant difference in the percentage of neutrophils undergoing cell death spontaneously or in response to LPS or IFN-γ in the absence or presence of serum (Figure (Figure7,7, A and B, respectively). Therefore, impaired neutrophil survival is likely not the mechanism of decreased neutrophil accumulation in the airways of TREM-1/3–deficient animals.
Next, we examined neutrophil migration using the TX migration plate (see Methods) to determine whether the defect in neutrophil accumulation in the airway of TREM-1/3–deficient mice was the result of an intrinsic defect in the ability of TREM-1/3–deficient neutrophils to respond to chemoattractants. Neutrophils were placed in the upper compartment of chambers separated from the lower compartment by a filter with 3-μm pores. Zymosan-activated serum (ZAS) or CXCL-1/KC was added to the lower chamber as a chemoattractant. There was no significant difference in migration between WT and TREM-1/3–deficient cells (Figure (Figure8),8), indicating that deficiency in TREM-1/3 does not alter the migratory response to chemoattractants.
The first step in neutrophil migration into infected airways is neutrophil migration from the vasculature across the pulmonary capillary endothelial cell monolayer. We hypothesized that TREM-1/3 may interact with endothelial cell receptors to facilitate neutrophil migration across the endothelial cell barrier. Therefore, we examined whether TREM-1/3–deficient neutrophils had an impaired ability to cross endothelial cell monolayers in vitro. Murine endothelial cells lines (SVEC4-10 and murine brain endothelial cells [MBECs]) were seeded on 3-μm-pore-size Transwell filters and grown to confluence. Freshly isolated bone marrow–derived neutrophils were loaded on the apical side of the endothelial monolayers, and ZAS (10%), a neutrophil chemoattractant, was added to the opposite side. There was no defect in the ability of TREM-1/3–deficient neutrophils to migrate across murine endothelial cell monolayers as compared with WT neutrophils (Figure (Figure9).9). In addition, we repeated the experiments using primary murine endothelial cells cultured as previously described (26). Consistent with the cell line data, we observed similar migration of TREM-1/3–deficient and WT neutrophils across primary murine endothelial cell monolayers (Figure (Figure9).9). These data suggest that impaired transendothelial migration is not the mechanism of decreased neutrophils in the airway of TREM-1/3–deficient mice following P. aeruginosa challenge.
In vivo, once neutrophils migrate from the capillaries, they must cross a pseudo-stratified epithelium to enter the luminal side of the airway. Airway epithelial cells (AECs) are polarized, and neutrophil transepithelial migration occurs from the basolateral side of AEC to the apical side of AEC into the airway lumen. Therefore, the ability of TREM-1/3–deficient neutrophils to migrate across primary polarized AECs was examined. While TREM-1 and TREM-3 are not known to be expressed on AECs, to eliminate an AEC defect in TREM-1/3–deficient mice as the source of impaired migration, we cultured both WT and TREM-1/3–deficient primary murine tracheal AECs on Transwells using previously described methods (27). The differentiated AECs from WT and TREM-1/3–deficient mice established transepithelial resistance greater than 500 Ω•cm2, indicating the presence of tight junctions. Transmission electron micrograph (TEM) images of the murine airway epithelial monolayers demonstrate ciliate cell features (Figure (Figure10A).10A). Figure Figure10B10B depicts a neutrophil attached to underside of the Transwell filter before migration. Freshly isolated bone marrow–derived neutrophils (pooled from 3 mice per group) were loaded on the basolateral side of AECs by inversion of the Transwells. CXCL1/KC was added as a chemoattractant to the culture medium on the apical side of the epithelium. Migration of WT neutrophils was similar across WT and TREM-1/3–deficient AECs (P = 0.19). By contrast, TREM-1/3–deficient neutrophils demonstrated a significant migratory defect in response to CXCL1/KC across both WT (Figure (Figure10C)10C) and TREM-1/3–deficient AECs (Figure (Figure10D).10D). Treatment of WT neutrophils with a TREM-1 polyclonal antibody but not a TREM-3 monoclonal antibody decreased migration across WT AECs (data not shown). Together, these data suggest that TREM-1 functions in neutrophil migration across AECs and that impaired neutrophil migration is the defect responsible for the susceptibility of TREM-1/3–deficient mice to P. aeruginosa pneumonia.
More than 2 million people die of pneumonia annually worldwide, most of them children under the age of 5 (28). These deaths occur at extremes of the immune response: too much inflammation and its associated lung injury can cause death, and insufficient inflammation and failure to clear infection can lead to the same result. Here, we identify TREM-1 as a novel receptor required for neutrophil transepithelial migration into the airway lumen. Our data suggest a mechanism by which impaired migration results in decreased neutrophil recruitment and ineffective bacterial clearance. Subsequently, the TREM-1–deficient animals die due to excessive cytokine release and bacterial dissemination. Thus, potential therapeutic agents to treat pneumonia, such as TREM-1 inhibitors, must achieve a balance between pathogen control and host tissue damage.
The neutrophil response is titrated not only through regulated cytokine and antimicrobial product release but also via controlled migration into the lung. Indeed, recent studies using intravital microscopy document the complexity of neutrophil migration (29, 30). These cells can immediately cross an endothelial cell barrier, pause and decide not to cross, or even cross and then reverse course to return to the vessel lumen, suggesting that the neutrophil is capable of integrating complex environmental signals (30). Determining how neutrophil migration is influenced by these disparate signals is critical for understanding the pathogenesis of pneumonia, acute respiratory distress syndrome, and other inflammatory diseases of the lung. The role of TREM-1 in cytokine augmentation has been well described, and our data demonstrate that TREM-1 is also involved in the process of neutrophil transepithelial migration.
Neutrophil migration into infected tissues has been studied extensively (31, 32). In the systemic circulation, the initial site of emigration is the postcapillary venule. Here, the neutrophil undergoes capture and rolling on the endothelial surface, and selectin family members play a critical role (33–35). Once firm adhesion has been established, transendothelial migration occurs under these laminar flow conditions. In contrast, neutrophil migration in the lung occurs in the pulmonary capillaries. Here a complex web of anastomosing capillaries requires the neutrophil to deform in order to pass through the vessels (36–38). Spatial constraints in the capillary most likely supplant the need for rolling and capture in the pulmonary circulation (39), thus providing an explanation for the fundamental difference in neutrophil migration in the lung compared with the systemic circulation. On the pulmonary endothelium, firm adhesion is primarily mediated by the interaction between the β2 integrins CD18/CD11a (also LFA-1 or αLβ2) and CD18/CD11b (also Mac-1 or αMβ2) expressed on the neutrophils and ICAM-1 (also CD54), ICAM-2, and ICAM-3 expressed on the endothelial cell surface (40–42). Both homotypic and heterotypic interactions between PECAM-1 (CD31), CD177, β2 integrins, ICAM-1, and junctional adhesion molecule (JAM) family members facilitate transendothelial migration (43–46). Because we observed decreased neutrophil accumulation in the lung airspace in TREM-1/3–deficient mice, we initially hypothesized that TREM-1/3 deficiency would cause defective transendothelial migration. However, TREM-1/3–deficient neutrophils migrated effectively across both murine endothelial cell lines and primary murine endothelial cells, likely via these previously described pathways, suggesting that TREM-1/3 is not required for transendothelial migration.
Compared with transendothelial migration, the molecular details of transepithelial migration are less well understood, including the extent to which neutrophils infiltrate the lung at the level of the large airway, the small airway, and/or the alveolar epithelia. Moreover, transepithelial migration is a dynamic process, and neutrophil passage through the alveolar interstitium is difficult to quantitate macroscopically. Our in vitro data using large airway epithelia grown at the air-liquid interface support a defect in neutrophil transepithelial migration. However, a limitation of our study is that we are unable to culture alveolar epithelia at the air-liquid interface for in vitro testing. In our model of TREM-1/3 deficiency, we detected differential compartmentalization of neutrophils in the interstitial space at the level of the small airway in vivo, consistent with a defect in transepithelial migration.
The integrity of the airway epithelium is achieved via tight junctions and adherens junction protein interactions. Paracellular migration is presumably achieved through alterations in tight junction protein interactions. Neutrophil adherence to the pulmonary epithelial cell layer is mediated by β2 integrins, in particular CD18/CD11b and ICAM, though CD11a and CD11c may play some role (47–55). The role of JAMs in intestinal transepithelial migration has also been studied. For example, JAM C but not JAM A has been implicated in intestinal neutrophil transepithelial migration, with studies demonstrating that JAM C antibodies decrease the rate of neutrophil transepithelial migration but do not block migration (56). Recently a role for junctional adhesion molecule–like protein (JAML), which is expressed on neutrophils, was described in intestinal transepithelial migration (57). JAML binds to epithelial cell via the coxsackie and adenovirus receptor (CAR), and soluble forms of either of these molecules impede neutrophil transepithelial migration across cultured intestinal epithelium. CD47, which is expressed on both epithelial cells and neutrophils, is also required for intestinal neutrophil transepithelial migration, though the mechanism is still being characterized and no studies have been conducted in the lung epithelium (58). To date there is no evidence for an interaction between TREM-1 and any of these adhesion molecules, suggesting that alternative mechanisms exist by which TREM-1 mediates neutrophil adhesion and migration across epithelial cells.
In this study, we demonstrate that loss of TREM-1/3 expression results in impaired neutrophil transepithelial migration in a murine pneumonia model. Gibot and colleagues examined the role of TREM-1 in bacterial pneumonia using a different approach (17). Their study involved instillation of a TREM-1–blocking peptide, LP-17, into the rat trachea following pneumonia challenge. Interestingly, this resulted in decreased lung inflammation and improved survival. Blockade of TREM-1–mediated amplification of cytokine production was thought to be the mechanism behind improved survival. However, our data suggest that LP-17 may play a role in blocking neutrophil migration. We propose that a partial blockade of neutrophil migration into the airways may allow a sufficient quantity of neutrophils to clear the infection but prevent extensive lung injury by attenuating transepithelial migration.
In summary, our results implicate TREM-1 as a novel player in neutrophil pulmonary transepithelial migration. We hypothesize that the TREM-1 interaction with the pulmonary epithelium provides a second novel regulatory pathway in vivo for TREM-1 to amplify inflammation through targeted migration into lung. In the absence of TREM-1, in vivo neutrophil migration to the airway lumen is impaired. This leads to ineffective bacterial killing and increased mortality. Further study of the role of TREM-1/3 molecules in transepithelial migration may identify therapeutic targets to modulate neutrophil influx, thereby altering the local inflammatory response not through classic cytokine signaling but through controlled migratory access.
To abrogate expression of functional TREM-1 and TREM-3 proteins, we made a targeting construct in which an 8.8-kb fragment containing most of exon 3 and exon 4 of TREM-1 as well as exon 1 of TREM-3 was replaced with an MC1neopA gene flanked by loxP sites (Supplemental Figure 1). This effectively deleted the transmembrane region of TREM-1 in addition to the start site and first domain of TREM-3. The targeting construct was linearized and transfected into E14.1 (129P2/OlaHsd) ES cells. 450 clones were chosen and expanded in duplicate; genomic DNA isolated from one set was digested with KpnI (K) and hybridized with a 3′ external probe (B) on Southern blots (endogenous, 13.5 kb; targeted, 17.8 kb). One correctly targeted clone was identified and confirmed by Southern blot analysis using multiple digestions and 5′ (A), 3′ (B), and Neo probes. This clone was injected into C57BL/6 blastocysts; chimeras were bred to transgenic mice expressing Cre under the CMV promoter (59) on a C57BL/6 background to delete the MC1neopA gene (TREM-1/3 del). Transmission of the TREM-1/3 deletion in agouti offspring was determined by PCR using tail DNA as a template and the following oligonucleotides: TREM-1/3 28279, TCTTGCCGCTGATTGGTTCA; TREM-1/3 27096, TCCCAAGAGCAGGCACAAGA; TREM-1/3 19239, TCTCTCCATCTATGCATCCACCC; endogenous, 373 bp, deletion, 270 bp. Heterozygous mice were intercrossed to generate mice homozygous for the deletion.
We backcrossed the TREM-1/3 mutation onto a C57BL/6 background facilitated by genome-wide simple sequence length polymorphism (SSLP) typing at 10-cm intervals for each generation (done by the Speed Congenics Facility supported by the Rheumatic Diseases Core Center at Washington University School of Medicine). The mice used for these experiments were >97% C57BL/6.
C57BL/6 mice were originally purchased from NCI. All animals used in this study were bred in a barrier facility at the University of Iowa. Mice used in pneumonia challenge studies were littermates or WT mice bred in our facility. The age of pneumonia experimental mice (55–75 days) was matched in WT and TREM-1/3–deficient mice. Mice used in all other experiments were <100 days of age.
P. aeruginosa, strain PA103, was grown overnight in LB broth and subcultured to mid-log phase and diluted to 2 × 105 CFU/ml. Mice were anesthetized with ketamine/xylazine, followed by intubation with a 24G catheter, and 1 × 104 CFU P. aeruginosa was instilled into the trachea. Mice recovered from anesthesia and were monitored for symptoms and temperature and euthanized when body temperature fell below 26°C or if they were moribund. Temperatures were determined using an Exergen Corp. Veterinary Infrared Tympanic Temperature Scanner, Model VTTS-1000. Survival studies were performed on separate animals. Mice in survival studies received no additional treatments.
Pneumonia was induced as before in separate groups of mice. At 14 hours after infection, mice were anesthetized, and blood was collected in tubes containing 10 mM EDTA. Tubes were centrifuged at 1,000 g for 10 minutes at 4°C. Supernatant plasma was aspirated and stored in multiple aliquots at –80°C until analysis. For histology or organ cytokine assays, lungs and liver were perfused with PBS through a catheter placed in the right ventricle. An opening was made in the right atrium to allow fluids to escape. Mouse lungs were hand homogenized in a buffer containing 15 mM Tris, 0.5% Triton X-100, 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, pH 7.4. Supernatants from homogenates were centrifuged at 1,000 g for 10 minutes at 4°C and stored in multiple aliquots at –80°C. Thawed lung homogenate supernatants were filtered through a 0.22-μm filter prior to analysis. For single-cell lung suspension analyses, perfused lungs were cut into small pieces and digested in 4 ml HBSS supplemented with 60 U/ml DNase I (Sigma-Aldrich) and 125 U/ml collagenase (Invitrogen) for 30 minutes at 37°C. Lung tissue was pressed through a wire mesh screen (Cellector, Bellco Glass) following digestion to create a single-cell suspension.
Mice were euthanized, and bone marrow was collected from femurs and tibias. Cells were isolated on a discontinuous Percoll gradient (MB Biomedical) composed of 81%, 62%, 55%, and 50%, and cells between the 62% and 81% layers were collected. Red blood cells were lysed with RBC Lysis Buffer (eBioscience) and neutrophils suspended in HBSS, without Ca2+ and Mg2+, and held on ice prior to testing. Neutrophil preparations were routinely >70% pure as determined by Cytospin.
At 14 hours after infection with P. aeruginosa, mice were anesthetized and perfused with 20 ml PBS and 20 ml 4% formaldehyde, and then lungs were placed in vials containing 4% formaldehyde. Lungs were placed under vacuum overnight, paraffin embedded, and stained with H&E. Histologic analysis was performed by a veterinary pathologist unaware of sample identities. Bilateral lungs from 5 animals in each group were examined. The scoring system was as follows — for cellularity (most severe foci assessed in lung): 1, mild, scattered leukocytes in interstitium and airspaces; 2, moderate, scattered leukocytes forming aggregates in interstitium and perivascular spaces and extending into airspaces; 3, moderate to severe prominent leukocyte infiltration in interstitium extending into airspaces often with prominent perivascular aggregates; for edema: 0, none; 1, mild, wispy eosinophilic material detectable in rare foci; 2, mild to moderate amounts of eosinophilic material detectable in scattered to coalescing foci; 3, moderate to extensive amounts of eosinophilic material filling alveoli. For necrotic foci, distinct necrotic cellular foci (often severely degenerated neutrophils with uncommon parenchymal components) in parenchyma were enumerated in the most severe region of each lung using ×200 magnification fields, 2 samples/slide × 2 slides (4 samples total).
Right and left lungs were sectioned in the coronal plane. Using a stereotactic approach, grids were used to segment the parenchyma into squares, and random numbers were used to identify regions for analysis, except when anatomic-specific regions were assessed. All examination, morphometry, and scoring were done in manner blinded to treatment group. For bronchiole neutrophils, terminal bronchioles (TBs) were identified in sections of lung as small airways opening into alveolar ducts, and in each TB the number of neutrophils within the lumen and those in the submucosa (exterior to airway epithelium basement membrane) were enumerated. These were scored based on the relative proportion of neutrophils in each TB as follows: 1, interstitial neutrophils < luminal neutrophils; 0.5, interstitial neutrophils = luminal neutrophils; and 0, interstitial neutrophils > luminal neutrophils. For alveolar neutrophils, random sections of lung were assessed (×400 magnification, BX51 microscope, Olympus) for the number of neutrophils within alveoli, and these numbers were standardized as neutrophils per 100 alveoli (60).
At 14 hours after infection, mice were anesthetized, and the chest cavity was opened. The trachea was exposed, an incision made, and tubing inserted. The lungs were flushed with 3 × 1 ml PBS. Fluids were collected and volumes recorded. For bacterial counts, 100 μl fluid was removed and plated on LB agar as described below. The remaining fluid was centrifuged to remove cells, and supernatant was stored in aliquots at –80°C.
Bacterial load was determined by dilution and plate count on LB agar following standard microbiological methods. CFU were counted, and CFU/ml was calculated.
Heparinized whole blood was stained, and rbc were lysed with 150 mM NH4Cl, 10 mM KHCO3, 0.1 mM EDTA. BAL fluids were centrifuged and pelleted cells resuspended in 2% FBS in PBS for staining. BAL cells were incubated with mouse Fc Block (BD 553141) prior to additional staining. Blood and BAL were stained with anti-CD115 (eBioscience 12-1152) and anti-Ly6G (BD 551460), which is specific for Ly6G without Ly6C cross-reactivity, per the manufacturer’s instructions. Cell type was determined by FACS on a BD LSR II with BD FACSDiva software. For single lung suspension analysis, a total of 2 × 106 lung cells were stained in a 96-well round-bottom plate. Lung homogenates were surface stained with mAbs specific to CD45, Ly6c, Ly6g, Siglec F (BD Biosciences), CD11c, F4/80, MHC II, CD4, CD8, B220, CD19, DX5, and CD3 for 30 minutes at 4°C and fixed with FACS lysing solution (BD Biosciences) for 10 minutes at room temperature. All monoclonal antibodies were purchased from BioLegend unless otherwise stated. Stained cells were run on a BD LSRFortessa. The total numbers of eosinophils (CD11cintSiglec F+), neutrophils (Ly6c+Ly6ghi), macrophages (CD11c+F4/80+), NK cells (CD3+DX5+), CD4+ T cells (CD3+CD4+), CD8+ T cells (CD3+CD8+), and B cells (CD19+B220+) were quantified using FlowJo (Tree Star) software.
Plasma and BAL fluids were analyzed using the Milliplex Bead Assay (Millipore). Cytokines/chemokines were all quantified using a customized multiplex magnetic bead panel kit (catalog MCYTOMAG-70K, Millipore). The manufacturer’s instructions were followed for assay parameters and protocol, with the following minor modification: 50 μl BAL fluid was used instead of the recommended 25 μl. G-CSF levels in mice subjected to bacterial challenge were extremely high compared with those of all other cytokines. Thus, G-CSF was not included in the multiplex panel but assayed independently at a 20-fold dilution using the same magnetic bead procedures. A Bio-Plex Pro II plate washer (Bio-Rad) was used to wash magnetic beads. Fluorescence intensity was measured on a Bio-Plex 200 system using Bio-Plex Manager software. Data collection parameters were set to 50 beads per region, 60-second timeout, bead map of 100 regions, sample size of 100 μl, and gate setting of 5,000–25,000.
ALT was determined from plasma collected as described above using Abaxis VetScan, Prep Profile II.
Murine bone marrow neutrophils were isolated as previously described and held on ice in HBSS, without Ca2+ and Mg2+. Texas red–zymosan particles (Molecular Probes, Invitrogen) were sonicated for 5 minutes, opsonized in normal mouse serum in the dark for 30 minutes at 37°C, washed, and resuspended at 0.4 M/ml in HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 1% human serum albumin. 500,000 neutrophils were mixed with opsonized Texas red–zymosan at MOI of 1, 5, and 25 in 100 μl HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 1% human serum albumin. The mixtures were incubated at 37°C for 30 minutes, then placed on wet ice for 5 minutes, stained with Ly6G-FITC (BD Biosciences — Pharmingen) for 20 minutes on ice, and washed. FACs was performed on a BD LSR II with BD FACSDiva software and analyzed using Cellquest software.
Neutrophils were isolated from bone marrow as described and resuspended at 6 × 105/ml in HBSS without Ca2+ and Mg2+ and held on ice. An overnight culture of P. aeruginosa strain PA103 was subcultured until OD600 was 0.6. One milliliter of culture was pelleted at 600 g for 5 minutes. The pellet was resuspended in 1 ml HBSS with Ca2+ and Mg2+, 0.5% BSA, and 10% pooled human serum and opsonized for 20 minutes at 37°C, with rotation. Bacteria were washed twice in ice-cold PBS and resuspended in 1 ml HBSS with Ca2+ and Mg2+ and 0.5% BSA. Bacteria were diluted to 2.5 × 106/ml in HBSS with Ca2+ and Mg2+ and 0.5% BSA. Neutrophils were mixed with bacteria at 1:1 for an MOI of 1:4 and incubated together at 37°C/5% CO2, with rotation for 45 minutes. Gentamicin sulfate (Cellgro), 0.5 mg/ml, was added, and incubation continued for 15 minutes. The mixture was washed twice in PBS to remove gentamicin and free bacteria. Neutrophils were resuspended in HBSS with Ca2+ and Mg2+ and 0.5% BSA and returned to the incubator. Samples were taken at relevant time points, and neutrophils were washed once in ice-cold PBS and lysed with 100 μl LB broth plus 0.1% Triton X-100. Plate counts were made by serial dilution on LB agar.
NADPH oxidase activity was determined using Fluostar Omega (BMG Technologies). Neutrophils were isolated from bone marrow of healthy mice as described and resuspended at 1 × 106 cells/ml in HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 0.1% human serum albumin. 200 μl of cell suspension with lucigenin (Sigma-Aldrich) at a final concentration of 100 μM was stimulated with opsonized zymosan (OpZ) (Molecular Probes, Invitrogen) at a ratio of 2:1 OpZ to neutrophils or with 10 ng/ml PMA (Sigma-Aldrich). Zymosan particles were sonicated for 5 minutes, opsonized by incubation with pooled human serum for 30 minutes at 37°C, centrifuged, and resuspended in HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 0.1% human serum albumin. Chemiluminescence readings were taken every minute for 60 minutes and reported as relative luminescence units per minute.
2 × 105 neutrophils resuspended in DMEM containing 10% FCS or not were cultured in tissue culture wells previously coated with BSA (100 ng/ml BSA in PBS, 2 hours at 37°C). Cell death was determined by the propidium iodide uptake assay of DNA content as previously described (25). In brief, cells were resuspended in 0.5 ml PBS and then fixed by adding 1.5 ml ice-cold 70% ethanol and incubating for 1 hour at 4°C. Fixed cells were washed once with 10 ml PBS containing 10% FCS, resuspended in the same buffer containing 40 μg/ml RNase A, and 50 μg/ml propidium iodide and incubated for 30 minutes in the dark. Samples were acquired by FACS, and cell death was determined based on hypodiploid DNA content.
Murine bone marrow neutrophils were isolated as previously described and stained with calcein-AM (Invitrogen, C3099) at a final concentration of 5 μg/ml. Cells were incubated at 37°C for 30 minutes, washed, and stored in HBSS, without Ca2+ and Mg2+. Cells were diluted to 106/ml in HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 1% human serum albumin prior to addition to the plate. ZAS was prepared by incubating normal mouse serum with sonicated zymosan (Molecular Probes, Invitrogen) particles at 37°C for 30 minutes (61). The zymosan particles were removed by centrifugation, and the resulting serum was diluted to 10% in HBSS, with Ca2+ and Mg2+, 0.1% glucose, 1% human serum albumin. KC (R&D Systems, 1395-KC) was diluted to 50 nM and 5 nM in HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 1% human serum albumin. DPBS without Ca2+ and Mg2+ was used to determine random migration. The TX migration plate (Neuroprobe, 101-3) consisted of a 3-μm pore filter in a 96-well format. 30 μl of stimulant or PBS was placed in the wells below the filter. The filter was snapped in place, and 25,000 neutrophils in 25 μl droplets were placed in hydrophobic rings on the filter, which allowed contact with the stimulant below. Additionally, a standard curve of stained cells was prepared from both cell preparations, and 25 μl placed in separate wells below the filter, with no fluid on the filter top. The prepared plate was incubated for 3 hours at 37°C/5% CO2. Following incubation, the cells were removed from the top of the filter by gently wiping with tissues and rinsing with HBSS, without Ca2+ and Mg2+. The plate was read on a Tecan Safire2 at 485 nm (excitation) and 530 nm (emission). The number of migrated cells was calculated from the standard curves, based on relative fluorescence units, using Excel. Graphs were made using GraphPad Prism.
SVEC4-10 murine lymph node endothelial cells (ATCC CRL-2181) were grown in DMEM (Gibco, Invitrogen) supplemented with 10% FBS (Hyclone) and l-glutamine (Gibco). MBECs, a gift from Beverly Davidson (University of Iowa, Iowa City, Iowa, USA), were grown in DMEM/F12 (Gibco), supplemented with 10% FBS and 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco). Endothelial cells were cultured on gelatin-coated, 3.0-μm-pore-size Transwell filters (Costar, 3472) for 3 days at 37°C/5% CO2 with 1 ml of medium in the bottom chamber. Confluence was assessed by H&E staining. Freshly isolated neutrophils were diluted to 5 × 106/ml in HBSS, with Ca2+ and Mg2+, 0.1% glucose, and 1% human serum albumin. 100 μl of neutrophil suspension was added to the Transwell, and 1 ml of 10% ZAS/medium was placed in the chamber below. Neutrophils migrated in the apical-to-basolateral direction for 16 hours, and then were collected from the lower chamber and counted using a hemocytometer.
Murine pulmonary endothelial cells were isolated as previously described (26). Mice were anesthetized with ketamine/xylazine. The lungs were perfused with 10 ml of 2-mM EDTA in PBS, followed by 5 ml 0.25% trypsin/2 mM EDTA via the right ventricle. The lungs were removed and incubated at 37°C for 30 minutes in PBS. The lungs were minced with a sterile blade, suspended in 2 ml DMEM plus 10% FBS, and washed 10 times to elute cells from the lung. Cells, but not debris, were placed in a 15-ml conical tube and centrifuged 5 minutes at 250 g. Cells were resuspended in 1.5 ml EGB-2 medium, 2% FBS, and supplements (EGM-2 Bullet Kit, Lonza CC3162) and plated in two 6-well plates precoated with 0.2% gelatin/PBS. Medium was changed at 24 hours and every other day until use.
Mouse tracheal AECs were cultured on 3-μm-pore-size Transwell filter (Costar, 3470) based on a previous report (27). Two weeks after seeding, the differentiated AECs established transepithelial resistance >500 Ω•cm2. To investigate the transepithelial migration ability of neutrophils, freshly isolated bone marrow derived neutrophils pooled from 3 mice per group (200,000/well) were loaded on the basolateral side of AECs by inverting the Transwells. KC (50 nM) was added at the apical side of culture medium. After 16 hours of co-incubation, the cells in apical medium were collected and cytospinned on a glass microscope slide. Ly6G immunostaining was performed to identify neutrophils. Ly6G+ cells were quantified by counting 3 random fields under a ×20 objective lens.
Filters for TEM were processed using conventional methods. Briefly, samples were fixed in 2.5% gluteraldehyde in 0.1 M cacodylate buffer, postfixed in 1% osmium tetroxide, and en bloc stained in 2.5% aqueous uranyl acetate. The filters were then dehydrated in a graded series of ethanol washes, transitioned to Eponate-12 resin (Ted Pella Inc.), and cured overnight at 65°C. Ultrathin sections were post-stained with uranyl acetate and lead citrate and imaged in a JEOL JEM 1230 transmission electron microscope equipped with a Gatan 2k × 2k digital camera.
Unless otherwise noted, all data were evaluated using GraphPad Prism software. Error bars represent SEM. Statistical significance was determined by Mann-Whitney nonparametric t test. For cytokine assays, geometric mean fluorescence (MF) was analyzed using a 5-parameter logistic equation. Based on expected/observed concentration ratios, interpolated values greater than 10,000 pg/ml (highest concentration used for standard curve) were all arbitrarily set to 12,000 pg/ml, while values calculated to be less than 16 pg/ml were all set to 0. In the case of G-CSF, the upper limit was set to 240 ng/ml and the lower limit to 300 pg/ml. A statistical software package (SAS) was used to assess differences in concentrations of each cytokine between WT (C57BL/6) mice and TREM-1/3–deficient mice using the nonparametric Wilcoxon test. P values less than 0.05 were considered significant. For single lung suspension analysis, statistical analyses were performed using InStat software (GraphPad Software). Groups were compared using 2-way ANOVA with Bonferroni post-test analyses to determine whether there was a statistical significance of at least α = 0.05.
All protocols involving mice were approved by the University of Iowa Institutional Animal Care and Use Committee and were carried out in accordance with institutional guidelines and regulations. TREM-1/3–deficient animals were generated at Washington University in St. Louis with the approval of the Washington University Animal Care and Use Committee and were carried out in accordance with institutional guidelines and regulations.
We thank Bill Nauseef, Beverly Davidson, and Kristina Thiel for critical reading and editing of the manuscript. We also thank Mark Schultz for assistance with the MBECs. This work was supported by grants from the American Heart Association to J. Klesney-Tait (0675028N) and from the NIH to M. Colonna (GM077279-04). All imaging was carried out by the Central Microscopy Research Facilities. The data presented herein were obtained at the Flow Cytometry Facility, which is a Carver College of Medicine Core Research Facilities/Holden Comprehensive Cancer Center Core Laboratory at the University of Iowa. The Facility is funded through user fees and the financial support of the Carver College of Medicine, Holden Comprehensive Cancer Center, and Iowa City Veteran’s Administration Medical Center.
Conflict of interest: The authors have declared that no conflict of interests exists.
Citation for this article: J Clin Invest. 2013; 123(1):138–149. doi:10.1172/JCI64181.