Role of natural killer cells and surfactant protein D in dendritic cell lymph node homing: Effects of ozone exposure

Abstract

The role of natural killer (NK) cells, surfactant protein D (SP-D), IFN-γ and the effect of ozone (O₃) inhalation was studied on recirculation of pulmonary dendritic cells (DC) to the mediastinal lymph nodes. Both O₃ exposure and lack of SP-D reduced NK cell IFN-γ and lung tissue CCL21 mRNA expression and impaired DC homing to the mediastinal lymph nodes. Notably, addition of recombinant SP-D to naïve mononuclear cells stimulated IFN-γ release in vitro. Because NKp46, a glycosylated membrane receptor, was necessary for dose-dependent SP-D binding to NK cells in vitro and DC migration in vivo, we speculate that SP-D may constitutively stimulate IFN-γ production by NK cells possibly via NKp46. This mechanism could then initiate the IFN-γ/IL-12 feedback circuit, a key amplifier of DC lymph node homing. Inhibition of this process during an acute inflammatory response causes DC retention in the peripheral lung tissue and contributes to injury.

INTRODUCTION

Dendritic cell (DC) lymph node homing is required to achieve optimal host defense, inflammatory resolution and immune homeostasis in the lung. To reach the draining mediastinal lymph nodes DC are driven by CCR7, activated by its ligands CCL19 or CCL21 (1). We previously demonstrated that NK-cell derived IFN-γ was necessary for CCL21 expression and DC lymph node homing in influenza A virus infected mice (2). Pulmonary NK cells are capable of releasing large quantities of IFN-γ both in homeostatic and inflammatory conditions (3). While the trigger for this NK cell function is not known, lung-specific factors were suggested to be important. SP-D, an epithelial pulmonary collectin with prominent immune regulatory properties, is a likely local candidate to regulate NK cells. SP-D expression increases during airway inflammation and plays an important role in inflammatory resolution. SP-D^−/− mice spontaneously develop an abnormal pulmonary immune phenotype (4) and show impaired pulmonary DC lymph node homing (5).

We chose to study O₃-induced airway inflammation because in this model there is no specific antigen uptake involved. Therefore, following O₃ inhalation DC migration should be primarily modulated by tissue changes. O₃ mainly affects the distal air spaces, inducing epithelial damage with release of pro-inflammatory mediators and cellular infiltration of the lung of both humans and mice (6) within a few hours of exposure.

Based upon the proximity of lung residential DC, NK cells and SP-D in the distal air spaces, we hypothesized that NK cell-SP-D interactions activate IFN-γ release at the peripheral ends of lymphatic vessels that facilitates DC lymph node homing. We further hypothesized that this process is altered upon O₃ exposure.

MATERIALS AND METHODS

Mice and model of O3 exposure

Seven-to ten week old male C57BL/6 mice (Jackson Laboratory, ME) SP-D^−/− mice (a gift from Drs. Samuel Hawgood UCSF and Francis Poulain, UC Davis) and NKp46^−/− mice (a gift from Dr. O. Mandelboim, Hebrew University and Dr. Wayne Yokoyama, Washington University) were exposed to 3 parts per million (ppm) O₃ or filtered air for 2 h and sacrificed at different time points to obtain BAL, lung tissue and mediastinal lymph nodes (6). All procedures were approved by the IACUC of the University of Pennsylvania.

Tissue processing and analysis, RNA isolation, real-time qPCR and flow cytometry were performed as described before (2, 6, 7).

Pulmonary dendritic cell migration assay

Mice were anesthetized (ketamine/xylazine, i.p., 100/20 mg/kg; Butler, OH; Akorn, IL) and given 50 μl of CFSE (8 mM; Fluka, MO; i.t.), then exposed to O₃ or filtered air 6 h later. Lymph nodes were harvested and processed for FACS analysis (2).

Ex vivo SP-D binding to NK cells

Splenocytes were isolated from naïve wild type (WT) mice and cultured with or without rSP-D (1μg/ml, Sino Biological Inc. China) for 48 h. Released IFN-γ was assayed by ELISA.

Pulmonary lymphocytes were enriched from naïve SP-D^−/− mice and NKp46^−/− mice. NKp46 expression was determined using FACS gated on NK cells (NK1.1⁺CD3⁻).

Cells were incubated with rSP-D (0 [ctrl], 2 or 20μg/ml) for 2 h at 4°C. Surface binding of SP-D on NKp46⁺ cells (from SP-D^−/− lungs) and on NK1.1⁺CD3⁻ cells (from NKp46^−/− lungs) was assessed by FACS.

Data analysis

Student’s t-test with Welch’s correction (unpaired, one-tailed) was performed (Prism 6 software, GraphPad Inc., CA) with Bonferroni’s correction for multiple comparisons unless otherwise indicated. Data are expressed as mean±SEM; p<0.05 was considered statistically significant.

RESULTS AND DISCUSSION

O3 induced pulmonary inflammation was associated with impaired DC lymph node homing

Neutrophilic granulocyte infiltration to the airways together with IL-6 and KC release was accompanied by accumulation of activated, TNF-α⁺ DC in the bronchoalveolar lavage in O₃ exposed mice 12 h after exposure (Figure 1A–C). According to dogma, DC lymph node homing is significantly induced during the inflammatory airway response (1). However, distinct from allergen challenge (5), we saw no increase in DC numbers in draining mediastinal lymph nodes harvested 12 h after O₃ exposure. In fact, the proportion of dendritic cells (CD11c^highMHCII^high), especially the CD103⁺ population (denoting mucosal dendritic cells that migrated from the distal air spaces) was decreased in the posterior mediastinal lymph nodes of the O₃ exposed group (Figure 1D, E). To confirm that such decrease reflected impaired DC migration, we labeled airway cells by instilling CFSE prior to O₃ exposure of mice and found that the CFSE⁺ migratory DC proportion was significantly reduced in the lymph nodes (Figure 1E left panel). This change did not reflect an influx of unlabeled DC from the circulation because circulatory DC retain an immature phenotype with low expression of CCR7 that is required for thoracic lymph node entry through the high endothelial venules. Indeed, the total numbers of the CD11b⁻CD103⁺ DC in the lymph node 12 h post O₃ was also significantly decreased (1E right panel). Thus, O₃ induced airway inflammation was associated with increased presence of TNF-α⁺ DC in the air spaces and decreased mucosal DC homing to the mediastinal lymph nodes suggesting that impaired regulation of DC migration could amplify airway inflammation.

Figure 1. O₃ induced neutrophilic airway inflammation was associated with suppression of DC lymph node homing, inhibition of CCL21 mRNA and reduced IFN-γ production by NK cells.

C56BL/6 mice exposed to O₃ or filtered air (for 2 h) were sacrificed 12 h later or as indicated. (A): BAL differential cell counts of macrophages (MP), eosinophils (EP), neutrophils (NP) and lymphocytes (LC) (DiffKwik stained cytospin preps). (B): BAL KC and IL-6 measured by ELISA. Representative FACS plots of (C): TNF-α⁺ DC% in CD11c⁺ BAL cells (n=12, *p<0.05; Student’s t test), (D): lymph node DC (left panels) and CD11b⁺CD103⁻ and CD11b⁻CD103⁺ subsets. (E): CFSE was instilled (i.t.) 6 h before O₃ exposure. % DC of all cells or % CFSE⁺ of total DC. (F–G): Lung mRNA was quantified by a qPCR microarray. (H): Intracellular IFN-γ in pulmonary NK cells. Representative FACS plot from three independent experiments. (I): Number of IFN-γ positive lung NK cells. (A, B): Mean±SEM of n=8–10. (E, I): Mean±SEM of n=6–8 mice pooled from 2–3 independent experiments. *p<0.05; Student’s t test with Welch’s correction; Filtered air vs. O₃. (F–G): data points (n=6–8) were obtained from three independent experiments; *p< 0.05; **p<0.01 vs. 0 h (ANOVA with Bonferroni correction).

A dynamic shift in DC numbers and phenotype including reduced CCR7expression was found upon chronic O₃ exposure in a different study (8). CCR7 activation by the specific ligands CCL19/21 is essential for lymph node homing (1). While the mechanisms that regulate CCL19/21 expression are unclear, NK cell derived IFN-γ was implied (2). Here we found that O₃-suppressed DC lymph node homing was associated with inhibition of mRNA for both CCL21 and IFN-γ (Figure 1F–G). As naïve T lymphocytes are incapable of producing IFN-γ (9), the likely source of IFN-γ in the normal lung is the NK cell. Notably, not only the level of NK cell IFN-γ expression but also the total number of IFN-γ⁺ NK cells were reduced in the O₃ treated mice (Figure 1H–I) implying NK cell-derived IFN-γ in the underlying pathways of O₃-inhibited DC lymph node homing. Our data were corroborated by another study in which chronic O₃ exposure suppressed IFN-γ production by NK T cells (10). NK cell IFN-γ expression was also dampened upon co-culture with O₃ exposed epithelial cells (11), suggesting that epithelial-derived mediators may play a regulatory role in this process.

Surfactant protein D (SP-D) promoted IFN-γ release and DC migration to lymph nodes

SP-D is an immune regulatory product of the distal epithelium with the capability of inducing IFN-γ in immune cells (7). Naïve splenocytes cultured with murine recombinant SP-D significantly increased IFN-γ release (Figure 2A). This effect was specific and selective because SP-A did not change IFN-γ expression and other cytokines (IL-4 or IL-5) were not affected by SP-D (data not shown). Compared with wild type mice, SP-D^−/− mice had reduced pulmonary IFN-γ and CCL21 mRNA expression (Figure 2B). Thus, while NK cells were activated in the lung of SP-D^−/− mice, their capacity to produce IFN-γ was diminished.

Figure 2. Presence of SP-D promoted IFN-γ release and DC migration to lymph nodes.

(A): IFN-γ (ELISA) from naïve wild type (WT) splenocytes cultured with rSP-D (1μg/ml) for 48 h. (B): IFN-γ and CCL-21 mRNA in the lung from naïve WT and SP-D^−/− mice (qPCR, normalized to 0hr) (C–F): WT and SP-D^−/− mice were exposed to O₃ or filtered air for 2 h. BAL was obtained 12 h later. (C): BAL neutrophils were determined by total and differential cell counts using DiffKwik stained cytospin preparations. (D): Representative data of TNF-α⁺CD11c⁺ BAL cells (n=7–8). (E): % CD11c⁺MHC-II⁺ CFSE⁺ DC in WT and SP-D^−/− mediastinal lymph nodes 12 h post O₃. (CFSE instilled [i.t.] 6 h prior O₃). (F): IFN-γ mRNA in the lung from WT and SP-D^−/− mice 12 h post O₃ (qPCR). (B, D, F): mean±SEM of n=6 (3 repeat experiments); *p< 0.05 Student’s t test with Welch’s correction. rSP-D vs. medium (A); WT vs. SP-D^−/− (B,D). (C, E–F): FACS plots representative of 2 independent experiments.

Compared to WT mice, O₃ induced neutrophilia and DC numbers (with a markedly elevated proportion of activated, TNF-α⁺ cells) were enhanced in the BAL of SP-D^−/− mice (Figures 2C, D). After CFSE treatment, the CFSE⁺ DC count was reduced in the mediastinal lymph nodes of SP-D^−/− mice (Figure 2E). Pulmonary IFN-γ expression also showed a trend of reduction post O₃ (Figure 2F). Previous studies showed that a differential ability of mice to develop inflammation was inversely associated with the amount of SP-D in the airways: Genetically low SP-D producer BALB/c or SP-D^−/− mice were highly susceptible to O₃-induced airway inflammation with a prolonged and heightened neutrophilia (6). In our study, O₃-induced neutrophilia, TNF-α⁺ DC accumulation in the airways and the diminished CFSE⁺ DC migration to the lymph nodes were all amplified in SP-D^−/− mice. These support that SP-D is necessary to maintain pulmonary DC lymph node homing. Our data also raised the possibility that SP-D plays a role through stimulation of IFN-γ by NK cells.

SP-D bound to NK cells in vitro through NKp46

Whether SP-D can directly bind NK cells was not previously known. We hypothesized that one of the membrane receptors, NKp46, may mediate specific SP-D binding and its effects on NK cells. We chose NKp46 over the other natural cytotoxicity receptors (NKp30 and 44) because it is evolutionarily conserved (the only member of the NCR family found in mice), constitutively expressed and capable of inducing IFN-γ synthesis and release by NK cells (through activation of PI3K and PLC-γ) (12). SP-D can bind cell surface glycoconjugates (4) and NKp46 is a highly glycosylated receptor (13).

SP-D^−/− pulmonary NK cells expressing NKp46, but not NK cells from NKp46^−/− mice (Figure 3A–C), bound SP-D dose dependently. This suggested that NKp46 could indeed mediate SP-D induced IFN-γ production by NK cells. This was supported by our data on DC lymph node homing: While the total percentages of DC remained similar between wild type and NKp46^−/− mice, upon CFSE labeling, DC migration from the lungs to the posterior mediastinal lymph nodes was diminished in the absence of NKp46, indicating a migration defect (Figure 3D). This defect was accompanied by a significant reduction in lung IFN-γ expression (Figure 3E).

Figure 3. NKp46 was necessary for SP-D binding to NK cells and for DC lymph node homing.

(A): Surface binding of SP-D on pulmonary NKp46⁺NK1.1⁺CD3⁻ cells from naïve SP-D^−/− mice incubated with rSP-D ex vivo. FMO: Fluorescence Minus One: cells were stained with all antibodies except for anti SP-D. (B): NKp46 expression on pulmonary NK1.1⁺CD3⁻ cells from naïve WT and NKp46^−/− mice. (C): Surface binding of SP-D on pulmonary NK1.1⁺CD3⁻ cells from NKp46^−/− mice incubated with rSP-D ex vivo. (D): CFSE (i.t.) expressing DC in lymph nodes in WT and NKp46^−/− mice 12 h post O₃. % DC of all cells or % CFSE⁺ of total DC (E): IFN-γ mRNA in the lung from WT and NKp46^−/− mice 12 h post O₃ (qPCR). (A–C), 2 independent experiment. (D, E), Mean±SEM of n=6–10 mice from 2 independent experiments, *p<0.05; **p<0.01 Student’s t test with Welch’s correction. WT vs. NKp46^−/−.

Interactions of SP-D with NK cells through NKp46 may play an important role in homeostatic DC lymph node homing and maintenance of an inflammation free pulmonary environment. It could also provide the initial activation of T cell independent IFN-γ release. Our study suggests that O₃ inhalation suppresses the SP-D/NKp46/IFN-γ and CCL21 mediated DC recirculation to the mediastinal lymph nodes. This effect may contribute to acute lung injury.

Abbreviations

BAL

Bronchoalveolar lavage

DC

Dendritic cell(s)

EP

eosinophil

LC

lymphocyte

MP

macrophage

NK

Natural Killer

NP

neutrophil

rSP-D

recombinant surfactant protein D

WT

wild type

i.t

intratracheal