Recruitment of neutrophils from blood vessels to sites of infection represents one of the most important elements of innate immunity. Movement of neutrophils across blood vessel walls to the site of infection first requires that the migrating cells firmly attach to the endothelial wall. Generally, neutrophil extravasation is mediated at least in part by two classes of adhesion molecules, β2 integrins and selectins. However, in the case of streptococcal pneumonia, recent studies have revealed that a significant proportion of neutrophil diapedesis is not mediated by the β2 integrin/selectin paradigm. Galectin-3 is a β-galactoside-binding lectin implicated in inflammatory responses as well as in cell adhesion. Using an in vivo streptococcal pneumonia mouse model, we found that accumulation of galectin-3 in the alveolar space of streptococcus-infected lungs correlates closely with the onset of neutrophil extravasation. Furthermore, immunohistological analysis of infected lung tissue revealed the presence of galectin-3 in the lung tissue areas composed of epithelial and endothelial cell layers as well as of interstitial spaces. In vitro, galectin-3 was able to promote neutrophil adhesion to endothelial cells. Promotion of neutrophil adhesion by galectin-3 appeared to result from direct cross-linking of neutrophils to the endothelium and was dependent on galectin-3 oligomerization. Together, these results suggest that galectin-3 acts as an adhesion molecule that can mediate neutrophil adhesion to endothelial cells. However, accumulation of galectin-3 in lung was not observed during neutrophil emigration into alveoli induced by Escherichia coli infection, where the majority of neutrophil emigration is known to be β2 integrin dependent. Thus, based on our results, we propose that galectin-3 plays a role in β2 integrin-independent neutrophil extravasation, which occurs during alveolar infection with Streptococcus pneumoniae.
The bacterial pathogen Streptococcus pneumoniae is responsible for the majority of community-acquired cases of pneumonia (1). Infection of the lung with S. pneumoniae induces a variety of inflammatory responses, including recruitment of leukocytes to infected lesions (2, 3, 4). Among leukocytes, neutrophils are the first to be recruited to the lesions, where they phagocytose pneumococci and synthesize a variety of products, including active oxygen metabolites and defensins (5). While these products kill the invading pathogens, they can also severely damage the lung tissue itself (5).
In streptococcal pneumonia, neutrophils transmigrate to the inflammatory site by passing first across the capillary venule, then through the interstitial matrix, and finally across the epithelium to the alveoli (3). The smaller diameter of the capillary venule compared with that of neutrophils reduces the normally rapid flow of neutrophils through the bloodstream, resulting in “sequestered” neutrophils (3). It has been proposed that the sequestration of neutrophils in capillaries replaces the role of selectins, adhesion molecules for leukocytes that facilitate neutrophil diapedesis in the case of acute inflammation of the skin or peritoneal cavity. It has recently been shown that selectins do not, in fact, participate in neutrophil diapedesis to the S. pneumoniae-infected lung (6).
To initiate transendothelial migration, the sequestered neutrophils are required to adhere tightly to the vascular endothelium (a process known as flattening) to withstand the shear stress imparted by blood flow (3). In the case of skin inflammation, peritonitis, and Escherichia coli-induced alveolar inflammation, such tight adhesion is thought to be mediated by β2 integrins (3). In contrast, in the case of streptococcal pneumonia, neutrophil extravasation to the infected lung is observed even in β2 integrin knockout mice (7, 8) or in rabbits where β2 integrins have been neutralized by passive immunization with neutralizing Abs against β2 integrins (9, 10). These data suggest that a significant proportion of tight adhesion of neutrophils to the endothelium during streptococcal pneumonia is mediated by a non-β2 integrin pathway. However, the molecules responsible for this adhesion have not yet been identified (11).
Galectin-3 (previously known as Mac-2 Ag, CBP30, or CBP35) is an ∼30-kDa mammalian lectin composed of a C-terminal carbohydrate recognition domain (CRD)3 and a N-terminal domain containing multiple repeats of a sequence rich in glycine, proline, and tyrosine (reviewed in Refs. 12, 13, 14). This lectin binds β-galactoside-containing glycoconjugates, particularly those with a polylactosamine structure (15, 16). Galectin-3 is found in the cytoplasm of various cells, including peritoneal macrophages and alveolar macrophages (17, 18, 19, 20). However, peritoneal inflammatory macrophage secretes galectin-3 directly from the cytoplasm without compromising the membrane integrity (19, 20). The mechanism for this secretion is not clearly understood (21), but the fact that it is not observed in peritoneal resident macrophages or in monocytes suggests that galectin-3 secretion is regulated in a differentiation-dependent manner (19, 22). It has been suggested that both proinflammatory factor IL-1β and fibroblast growth factor are secreted by a similar mechanism (21). As the majority of galectin-3 ligands are cell surface glycoconjugates, release of galectin-3 from cells is the rate-limiting step for its ability to act as a lectin.
Upon binding to glycoconjugate ligands at the cell surface, galectin-3 molecules are able to oligomerize through their N-terminal domains (23, 24) and cause cross-linking of surface glycoproteins. It has been suggested that such cross-linking triggers signal transduction cascades involved in several innate immune responses: 1) the triggering of an oxidative burst in monocytes and neutrophils (25, 26, 27); 2) the augmentation of LPS-induced release of IL-1 in monocytes (28); 3) the initiation of degranulation in mast cells (29); 4) the selective down-regulation of IL-5 expression (30); and 5) chemoattraction of monocytes and macrophages (31). Another consequence of galectin-3-mediated cross-linking is cell adhesion. It has been suggested that galectin-3 can mediate the adhesion of lymphoma cells to lung microvascular endothelial cells (32) or the adhesion of neutrophils to laminin (33), although the implication of such adhesion in the context of leukocyte emigration has not been explored.
In this study we demonstrate by using a murine pneumonia model system that the extracellular release of galectin-3 occurs at the onset of neutrophil diapedesis to alveolar spaces infected by S. pneumoniae. Our results suggest that this release is from alveolar resident macrophages or pulmonary parenchyma cells but not from extravasated neutrophils. In addition, we show in vitro that galectin-3 mediates neutrophil adhesion to the endothelium through direct cellular cross-linking. These results suggest that galectin-3 acts as a novel type of adhesion molecule involved in streptococcal pneumonia in which a significant proportion of extravasation of neutrophils to the infected lung occurs in a manner independent of β2 integrins.
Materials and Methods
Human endothelial cell lines ECV304, HL-60, and hybridoma M3/188.8.131.52 (Mac-2, anti-galectin-3 Ab) were obtained from American Tissue Culture Collection (Manassas, VA). HUVECs were obtained from Cell Applications (San Diego, CA).
A rat mAb against galectin-3 was purified from the cultured medium (serum-free hybridoma medium (Life Technologies, Burlington, Ontario, Canada)) of the hybridoma M3/184.108.40.206 by a protein G-Sepharose column (Amersham Pharmacia Biotech, Piscataway, NJ).
Mouse streptococcal pneumonia model
A murine model of streptococcal pneumonia was used as described previously (4, 34). Briefly, lightly anesthetized female CD1 mice (18–20 g) received an inoculum of 107 CFU of S. pneumoniae serotype 3 in 50 μl of PBS applied at the tip of the nose and involuntarily inhaled. Infected animals were sacrificed by decerebration (under anesthesia) at different time points after infection over a 3-day period. In some experiments, lightly anesthetized female CD1 mice received an inoculum of 107 CFU of E. coli instead of S. pneumoniae to induce the extravasation of neutrophils to the alveoli.
In some experiments, the extravasation of neutrophils in S. pneumoniae-infected alveoli of female mice of strains C57BL/6 and BALB/c (instead of CD1) was also examined 6 h postinfection. The other S. pneumoniae strain, Harvard strain (kindly provided by Dr. C. M. Doerschuk, Case Western Reserve University, Cleveland, OH, 108 CFU/mice (7)), which was used in the papers suggesting β2 integrin-independent neutrophil extravasation, was also used in some experiments to induce neutrophil extravasation (6 h postinfection) to the infected alveoli of CD1, C57BL/6, and BALB/c mice.
Bronchoalveolar lavage (BAL) fluid was collected by gently washing the alveolar space twice with 1 ml of PBS through the trachea. The BAL fluid was centrifuged at 550 × g for 10 min to separate the cells from the supernatants. The cell-free BAL supernatants (BAL sup) were used for the estimation of galectin-3 and TNF-α. The cells were resuspended in PBS for multiple analyses, including a total cell count by hemacytometer and differentiation of cell populations by Diff-Quick (Baxter, Quebec, Canada) staining of cytospin preparations.
Analysis of galectin-3 in cell-free supernatants of BAL fluids or exudated leukocytes of BAL fluids
Cell-free supernatants of BAL fluids (50 μl) were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane (Schleicher & Schuell, Keene, NH) using a Tris-glycine buffer system. After incubation with 10% skimmed milk in TBST (20 mM Tris-HCl (pH 7.5),150 mM NaCl, 0.2% Tween 20), membranes were incubated with anti-galectin-3 Ab followed by anti-rat IgG-peroxidase (Amersham Pharmacia Biotech). Ab complexes bound onto membranes were detected by exposing to Blue XB-1 film (NEN Life Science, Boston, MA) after incubating with chemiluminescence substrate for peroxidase (NEN Life Science).
Exudated leukocytes (cell fractions) of BAL fluids were homogenated with 500 μl of buffer containing 1% Triton X-100, 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, and protease inhibitors mixture (buffer A; Sigma, Oakville, Ontario, Canada). The homogenates were then centrifuged at 35,000 × g for 15 min to obtain the cell extracts. Ten microliters of cell extracts were used for the estimation of galectin-3.
Mouse resident alveolar macrophages were purified from the BAL fluid of noninfected mice. BAL fluid was prepared as described above. After spinning down of cells and removal of contaminating RBCs by hypotonic lysis, the leukocytes in BAL fluid were resuspended in RPMI 1640 supplemented with 10% FCS (HyClone Laboratories, Logan, UT) and incubated at 37°C for 1 h on a tissue culture dish. Unbound cells were removed by washing three times with PBS. More than 95% of adherent cells showed an appearance characteristic of macrophages when examined by light microscopy.
Exudated neutrophils were purified from leukocytes that emigrated to air pouches on the dorsum of CD1 mice as described by Tessier et al. (35, 36). Briefly, an air pouch was raised on the dorsum by s.c. injection of 2 ml of sterile air. On day 6 after two injections of air 3 days apart, inflammation was induced by injecting 1 ml of 10 μg/ml LPS in PBS into the air pouch (35, 36). Six hours after the induction, emigrated leukocytes were obtained by washing the air pouches with PBS supplemented with 1 mM EDTA (2 ml, three times). Leukocytes were resuspended in 10% FCS-MEM and incubated at 37°C for 2 h on tissue culture dishes and nonadherent cells were collected. More than 95% of the nonadherent cells showed an appearance characteristic of neutrophils when Diff-Quick-stained and examined by light microscopy.
Alveolar macrophages or exudated neutrophils were homogenized with buffer A as described above, and the levels of galectin-3 in those cells were estimated by Western blotting with anti-galectin-3 Ab.
Secretion of galectin-3 from alveolar macrophages
The membrane fractions of S. pneumoniae were prepared as follows. S. pneumoniae was washed with PBS three times and resuspended in ice-cold buffer containing 20 mM Tris-HCl (pH 7.5). After freezing and thawing twice, insoluble membrane fractions were separated by centrifugation at 5,000 × g for 10 min and the membrane fractions were further washed with PBS twice. Mouse resident alveolar macrophages were purified as described above and were incubated overnight in fresh 10% FCS-RPMI 1640. Alveolar macrophages (1 × 105 cells) were incubated in the presence or absence of LPS (10 μg/ml) or the membrane fraction of S. pneumoniae (serotype 3) at macrophage:S. pneumoniae cell ratio of 1:50 for 2 h. After incubation, culture media were removed and centrifuged to obtain cell-free supernatants. Aliquots (40 μl) of the supernatants were used for the analysis of galectin-3 secretion.
Immunofluorescence analysis of BAL neutrophils
Exudated neutrophils in BAL fluid of mice infected with S. pneumoniae (4 h postinfection) were purified by brief panning of BAL fluids in a petri dish to remove macrophages. Nonbound neutrophils were then incubated with PBS supplemented with 0.1% sodium azide and 0.1% BSA (PAB) together with either anti-galectin-3 (Mac-2) Ab or control rat IgG followed by anti-rat IgG Ab labeled with Alexa 448 (Molecular Probes, Eugene, OR) as described previously (19, 20). FACS analysis was conducted by FACSCalibur flow cytometer (BD Biosciences, San Jose, CA) and neutrophil populations were selected based on the size and the granularity of cells at the time of data collection.
Fresh left lungs were perfused and fixed overnight at 4°C in 0.1 M phosphate-buffered 4% paraformaldehyde. Tissues were then dehydrated and embedded in paraffin. After blocking with normal rabbit serum (Vector Laboratories, Burlingame, CA), 5 μm of tissue sections were first incubated with anti-galectin-3 Ab (10 μg/ml) or normal rat IgG for 2 h at 37°C. Bound anti-galectin-3 Ab was visualized by biotinylated anti-rat IgG Ab followed by VECTASTAIN Elite ABC reagent following the manufacturer’s instructions (Vector Laboratories). A 0.5% toluidine blue O solution was used as counterstaining in some tissue sections. Specificity of labeling was tested by 1) omitting the primary Ab in the procedure or 2) replacing the primary Ab by normal rat IgG.
Galectin-3 and its CRD
Recombinant human galectin-3 was purified as described previously, with modification (15). Briefly, E. coli JM109 was transformed with an expression plasmid of human galectin-3 (kindly provided by Drs. J. Hirabayashi and K. Kasai, Teikyo University, Tokyo, Japan). Two liters of this E. coli (JM-HG29-c1.2) overnight bacteria culture medium were incubated with 1 mM isopropyl β-d-thiogalactoside for 3 h at 37°C to induce galectin-3 production and E. coli was pelleted by centrifugation. After sonicating bacteria in buffer B (20 mM Tris-HCl (pH 7.5), 0.15 M NaCl) containing 5 mM EDTA, 1 mM DTT, and protease inhibitor mixture (Sigma), cell homogenates were centrifuged to obtain a soluble fraction. This soluble fraction (∼50 ml) was then applied to 5 ml of asialofetuin-agarose (4 mg asialofetuin/ml gel), which was prepared with AminoLink Plus Coupling Gel (Pierce, Rockford, IL) following the manufacturer’s instructions. After extensive washing of the column (∼20 bed volume) with buffer B, galectin-3 was eluted with buffer B containing 100 mM lactose. The eluate was dialyzed against PBS containing 1 mM EDTA and then against PBS to remove lactose. Typically, 3–7 mg of galectin-3 was purified from 2-liter cultures of E. coli. The purity of galectin-3 was determined by Coomassie blue staining and silver staining of a SDS-polyacrylamide gel. Purified galectin-3 was filter-sterilized, kept at 4°C, and used within 1 mo.
The CRD of galectin-3 was prepared as described by Massa et al. (24). Briefly, galectin-3 was digested with collagenase VII (Sigma) in 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 2 mM CaCl2 for 37°C for 4 h. The digestion was stopped by adding EDTA (final concentration, 5 mM), and the CRD of galectin-3 was purified with asialofetuin agarose affinity column chromatography as described above.
In vitro neutrophil adhesion assay
Human neutrophils, which do not secrete galectin-3 (26), were purified from blood of healthy volunteers. Briefly, after incubating heparinized blood with 3% dextran, the leukocyte- and erythrocyte-rich fractions were subjected to Ficoll-Hypaque separation (Amersham Pharmacia Biotech) to purify neutrophils. When primed neutrophils were required, neutrophils were treated in vitro with fMLP (27, 37). When differentiated HL-60 cells were required, HL-60 cells were incubated overnight in the presence of all trans retinoic acid (1 μM).
Human endothelial cells ECV304 or HUVECs (Cell Applications) were plated (104 cells/well) to 96-well plates and grown to confluence in 10% FCS-M199 (Life Technologies) or in endothelial cell growth medium (Cell Applications), respectively. The endothelial monolayers were washed gently with serum-free M199 medium twice and 25 μl of fresh serum-free M199 medium was added to each well. Human neutrophils or HL-60 were labeled with calcein-acetoxymethyl (Molecular Probes) by following the manufacturer’s instructions. Labeled neutrophils (10 μl, 105 cells) together with PBS (25 μl) containing various amounts of galectin-3 were added to the wells and plates were incubated for the indicated time at 37°C. At the end of incubation, the plates were immersed in cold PBS and inverted. Excess liquid and nonadherent cells were removed by blotting onto paper towels (repeated three times) (38). Bound neutrophils were lysed by adding 1% Triton X-100-PBS and the associated fluorescence was measured using a fluorometric plate reader (PerSeptive Biosystems, Foster City, CA). When the adhesion assays were performed in the presence of lactose, the sodium chloride concentration in the PBS was adjusted to maintain appropriate osmolarity, i.e., 317 mosmol/L. In the case when endothelial cells were exposed to LPS, confluent endothelial monolayers were preincubated in the 10% FCS-M199 containing 2 μg/ml LPS for 2 h at 37°C. After preincubation, monolayers were washed and then the adhesion assays were conducted as described above in the presence of LPS (2 μg/ml).
Immunofluorescence analysis of human blood neutrophils and human endothelial cells ECV304
Human blood neutrophils were prepared as described above. Monolayers of ECV304 were trypsinized briefly to obtain single cell suspensions. Neutrophils or ECV304 cells (1 × 106 cells) were first incubated with galectin-3 (1 μM) for 10 min and unbound galectin-3 was removed by washing twice with PAB. To analyze the level of galectin-3 that bound to cells, cells were then incubated with anti-galectin-3 Ab or control rat IgG followed by anti-rat IgG Ab labeled with Alexa 448 (Molecular Probes). FACS analysis was conducted as described above.
Accumulation of soluble galectin-3 in BAL fluids of mice infected with S. pneumoniae
To study a role of galectin-3 in the recruitment of neutrophils into the lung tissue during infection with S. pneumoniae, a mouse model was used (4). Mice (CD1) were inoculated with S. pneumoniae (serotype 3; 1 × 107 CFU) intranasally, resulting in induction of typical pneumonia symptoms, including recruitment of leukocytes into alveoli and lung interstitium.
BAL fluids were collected from infected mice and centrifuged to separate the leukocytes that had emigrated to alveolar spaces, from supernatants (BAL sup) containing factors and proteins secreted into the alveolar spaces (4, 34). Galectin-3 content in the BAL sup was estimated by Western blotting with anti-galectin-3 mAb (Fig. 1⇓A). A very low level of galectin-3 was detected in the BAL sup from untreated mice and from mice infected for 2 h. However, 4 h after infection, the amount of galectin-3 in the BAL sup was significantly increased by an average of ∼5-fold compared with those from uninfected mice, reaching a peak (32-fold of those from uninfected mice) 12 h postinfection (Fig. 1⇓B). The level of galectin-3 in the BAL sup remained the same level up to 48 h and then gradually decreased by 72 h when 30% of mice began to die due to pneumonia (data not shown). Thus, these data indicate that galectin-3 is actively released into the alveolar space 4–48 h after infection with S. pneumoniae.
Correlation of the kinetics of accumulation of galectin-3 in alveoli with the kinetics of neutrophil emigration to alveoli during S. pneumoniae infection
The numbers of neutrophils in BAL fluids were increased during S. pneumoniae-induced inflammation (Fig. 2⇓) as previously reported (4, 34). In BAL fluids from mice treated with PBS (Fig. 2⇓) or from untreated mice (data not shown), negligible number of neutrophils (∼3 × 104 cells/ml) were present. The number of neutrophils in BAL fluids was significantly increased 4 h after S. pneumoniae inoculation (Fig. 2⇓), reaching a peak around 12 h, and was maintained at this level up to the point of death 96 h after infection. In contrast, a small but significant number of resident alveolar macrophages (∼3 × 105 cells/ml BAL fluid) was observed in BAL fluids from uninfected mice. The number of macrophages in BAL fluid was not altered significantly during the first 24 h after infection (Fig. 2⇓). Thus, the accumulation of galectin-3 in BAL sup (Fig. 1⇑) of infected animals correlated closely to the kinetics of neutrophil extravasation into alveolar spaces.
Endogenous galectin-3 expression in alveolar leukocytes during early streptococcal infection
The levels of endogenous galectin-3 in total leukocytes that were recovered from infected alveoli were next analyzed. Equal proportions of total lysates of cell fractions isolated from BAL fluids were applied to SDS-polyacrylamide gels and the amounts of endogenous galectin-3 were estimated by Western blotting. The levels of galectin-3 associated to BAL leukocytes from infected mice remained similar to those from uninfected mice (Fig. 3⇓A), despite the fact that 26- to 33-fold more neutrophils were present in those BAL fluids from infected alveoli compared with uninfected alveoli (Fig. 2⇑).
We next analyzed the level of endogenous galectin-3 in exudated neutrophils and alveolar resident macrophages. Neutrophils were purified from leukocytes that had emigrated to inflamed s.c. dorsal air pouch, and alveolar resident macrophages were prepared from BAL fluids. As Fig. 3⇑B shows, no significant level of galectin-3 was detected in lysates of the mouse neutrophil fraction (2 × 105 cells). By contrast, galectin-3 was readily detected in the extracts from 10 times lower number of alveolar macrophages (2 × 104 cells) (Fig. 3⇑B, lane 2). This result suggests that, unlike human neutrophils that express galectin-3 intracellularly (Ref. 39 and data not shown), murine neutrophils do not express endogenous galectin-3. Together, these data suggest that increased level of galectin-3 in BAL sup obtained from infected mice (Fig. 1⇑) is not due to the accumulation of neutrophils and thus galectin-3 is released from other cells, including alveolar macrophages.
Alteration of tissue distribution of galectin-3 during S. pneumoniae infection
Immunocytochemical analysis was next performed on sections of formalin-fixed lungs from mice infected with S. pneumoniae for 12 h to determine whether the distribution of galectin-3 in the lung was altered during the early stage of streptococcal pneumonia (Fig. 4⇓). In noninfected lungs, positive staining with anti-galectin-3 Ab was scattered (Fig. 4⇓A). The distribution and the shape of the labeled cells suggest that some resident alveolar macrophages were galectin-3 positive, consistent with a previous report suggesting that galectin-3 is localized to alveolar macrophages (40). No significant galectin-3 positive staining was observed in uninfected alveolar wall (Fig. 4⇓A). In striking contrast to the noninfected lung, galectin-3-positive staining in S. pneumoniae-infected lung was not limited to alveolar macrophages in S. pneumoniae-infected lung (Fig. 4⇓, B and D). A greater number of alveolar parenchyma cells particularly located in the vicinity of vascular regions were labeled (Fig. 4⇓, B and D). Dispersed interstitial macrophages, alveolar epithelial cell layer, and vascular endothelium were most likely among the galectin-3-positive cells. At some places, diffuse but distinctive colored deposits were also observed in the alveolar interstitium located in between the alveolar epithelium and the vascular endothelium (Fig. 4⇓B). In contrast, negative staining of infected lung tissue sections was displayed when anti-galectin-3 Ab was replaced with normal rat IgG (Fig. 4⇓C). Thus, the redistribution of galectin-3 we observed at 12 h postinfection occurred within the window of active recruitment of neutrophils that occurs during development of the pneumonia.
Secretion of galectin-3 from alveolar macrophages
We have previously reported that peritoneal inflammatory, but not resident and quiescent, peritoneal macrophages release galectin-3 extracellularly: about one-half of newly synthesized galectin-3 is found to be secreted from the activated macrophages within 3 h (19, 20), suggesting that the machinery involved in galectin-3 secretion is regulated in a manner dependent on the differentiation/activation stages of macrophages. Thus, we next investigated whether alveolar macrophages secrete galectin-3, contributing to the accumulation of galectin-3 in BAL fluids during streptococcal pneumonia.
Resident alveolar macrophages were obtained from untreated mice. Fresh medium was added to the cells and the cells were incubated for an additional 3 h in the presence of LPS or S. pneumoniae membrane fraction. Galectin-3 released into the medium was analyzed by Western blotting. As shown in Fig. 5⇓A, a small amount of galectin-3 was secreted from these control macrophages or macrophages incubated in the presence of LPS. In contrast, the release was significantly augmented when macrophages were incubated with the membrane fraction of S. pneumoniae (Fig. 5⇓A), suggesting that S. pneumoniae stimulates the release of galectin-3 from alveolar macrophages.
Binding of galectin-3 to the surface of neutrophils which emigrated to infected alveoli
As shown in Fig. 3⇑B, only a negligible amount of endogenous galectin-3 was found in mouse inflammatory neutrophils. However, our results show that galectin-3 accumulated in S. pneumoniae-infected alveoli, into which neutrophils emigrated. Thus, if mouse neutrophils express galectin-3 ligands on their surface, galectin-3 will bind to the surface of those emigrated neutrophils. Interestingly, Feuk-Lagerstedt et al. (41) have suggested that neutrophils express galectin-3 ligands such as CD66. Thus, we next studied whether galectin-3 that was released to the infected alveoli was associated with neutrophils that emigrated to the alveoli. Neutrophils obtained from S. pneumoniae-infected alveoli were treated with anti-galectin-3 Ab, followed by anti-rat IgG-Alexa 448, and the surface expression of galectin-3 was analyzed by flow cytometry. As shown in Fig. 5⇑B, a low but significant level of galectin-3 was found on the neutrophil surface. This result suggests that galectin-3 released into infected alveoli during infection binds to the cell surface of neutrophils, which emigrated to the alveoli.
Lack of accumulation of galectin-3 in alveoli infected with E. coli
Extravasation of neutrophils into the alveolar space can be induced by other bacterial pathogens. Thus, we investigated whether galectin-3 was released into the alveolar space during E. coli infection. Mice (BALB/c, CD1, or C57BL/6) were intranasally inoculated either with E. coli or S. pneumoniae (Harvard strain) as described above. As shown in Fig. 6⇓B, inoculation of BALB/c mice with E. coli induced a robust emigration of neutrophils into alveoli (about twice that induced by S. pneumoniae) and the release of a proinflammatory factor, TNF-α. In contrast, the amount of galectin-3 released into alveolar spaces infected with E. coli remained the same as that of the noninfected controls and substantially less than that found in S. pneumoniae infection (Fig. 6⇓A). We also found levels of galectin-3 in S. pneumoniae (Harvard strain)-infected alveoli of CD1 and C57BL/6 mice or in S. pneumoniae (serotype 3)-infected alveoli of CD1 mice comparable to those found in S. pneumoniae-infected alveoli of BALB/c mice (data not shown). We did not observe any significant increase of galectin-3 in E. coli-infected alveoli of CD1 and C57BL/6 mice (data not shown). Thus, this specific accumulation of galectin-3 induced by S. pneumoniae was observed when different mouse strains and S. pneumoniae strains were used. Together, these data demonstrate that accumulation of galectin-3 is not induced in E. coli infection and thus is specific to S. pneumoniae infection.
The ability of galectin-3 to mediate adhesion of neutrophils to endothelial cell layer in vitro
To investigate whether galectin-3 participates as an adhesion molecule in neutrophil diapedesis, its ability to mediate the adhesion of neutrophils to endothelial cells was analyzed in vitro. Human neutrophils were labeled with a fluorescent dye, calcein-acetoxymethyl, and added to a human endothelial cell line ECV304 monolayer in the presence or absence of galectin-3. As shown in Fig. 7⇓A, adhesion of neutrophils onto the endothelial monolayer was promoted 5-fold in the presence of exogenously added galectin-3 at concentrations >0.66 μM. The adhesion of neutrophils reached to plateau within 10 min (Fig. 7⇓B) in the presence of 0.66 μM galectin-3. When the human promyelocytic cell line, HL-60 cells (undifferentiated or differentiated with retinoic acid), which are suggested to express surface galectin-3 ligands different from those of neutrophils (41), were used instead of neutrophils, no obvious promotion of HL-60 binding to endothelial cells by galectin-3 was found (data not shown). Thus, those data demonstrated that galectin-3 can promote neutrophil adhesion to endothelial cells. Preincubation of endothelial cells or neutrophils with LPS or prepriming of neutrophils with low concentration of fMLP did not cause further increase of galectin-3-mediated neutrophil adhesion (data not shown).
We next studied whether ECV 304 endothelial cells as well as human neutrophils express galectin-3 ligands on their cell surface (Fig. 7⇑, C and D). ECV304 cells or neutrophils were first incubated with galectin-3. Then, cells were washed to remove unbound galectin-3 and the levels of cell surface-bound galectin-3 were analyzed by flow cytometry to study the expression of galectin-3 ligands on the cell surfaces. As shown in Fig. 7⇑, C and D, extracellularly added galectin-3 bound to the surface of endothelial cells and neutrophils, respectively, suggesting that both cells express galectin-3 ligands on the surface. Thus, together, the data suggest that peripheral blood neutrophils as well as endothelial cells express galectin-3 ligands on their surfaces to support galectin-3-dependent adhesion and that the adhesion can be achieved without any inflammatory priming of cells.
To study whether this adhesion resulted from the adhesion activity of galectin-3 itself rather than the mechanism in which galectin-3 is involved in the activation of some other adhesion molecules, the adhesion assay was performed at 4°C (Fig. 8⇓A). The presence of galectin-3 also increased the adhesion of neutrophils to the endothelial cells when the adhesion assay was performed at 4°C (nearly 75% of that at 37°C, Fig. 8⇓A), suggesting that galectin-3 directly mediates neutrophil adhesion.
Galectin-3-mediated adhesion was efficiently inhibited in the presence of lactose (100 mM), which can compete the binding of galectin-3 to their ligands (Fig. 8⇑B), but not by mannose (100 mM), which does not antagonize galectin-3 (data not shown), suggesting that lectin activity of galectin-3 mediates the adhesion of neutrophils to endothelial cells. Galectin-3 has only one CRD located at C-terminal and it is suggested that oligomerization through non-lectin N-terminal domain is essential to exert cross-linking activity of galectin-3 (23, 24). To investigate whether the N-terminal domain of galectin-3 is involved in neutrophil adhesion, truncated galectin-3 lacking N-terminal tandem repeating domain (galectin-3-CRD), instead of full-length galectin-3, was added in the neutrophil adhesion assay. The truncated galectin-3 failed to support neutrophil adhesion (Fig. 8⇑C), suggesting that the full-length lectin mediates neutrophil adhesion to endothelial cells through oligomerization of its N-terminal multiple repeating domain.
We also tested whether galectin-3 can also support human neutrophil adhesion to freshly isolated primary HUVEC. As shown in Fig. 8⇑D, in the presence of galectin-3 (above 1 μM), human neutrophils adhered to HUVEC layer to the same extent as observed with ECV304, suggesting that HUVEC also expresses galectin-3 ligands with which galectin-3 can mediate adhesion of neutrophils to endothelium. Interestingly, pretreatment of endothelial cells with LPS followed by the presence of LPS during the adhesion assay did not augment the number of adherent neutrophils in the presence of galectin-3, suggesting that HUVEC cells express galectin-3 ligand before LPS activation (Fig. 8⇑D).
This study provides evidences suggesting that galectin-3 plays a role as an adhesion molecule in the process of extravasation of neutrophils in a mouse model of streptococcal pneumonia, where conventional neutrophil adhesion molecules, β2 integrins, or selectins do not actively participate to form tight adhesion of neutrophils to the vessels (42). Thus, we found that the content of galectin-3 remarkably increased in the alveolar lavage obtained during the progression of murine streptococcal pneumonia. Those kinetics closely parallel the infiltration of neutrophils into the alveolar spaces. Immunohistochemistry analysis showed that galectin-3, whose presence was scattered in uninfected lungs, was distributed in the vicinity of vascular and alveolar regions of infected lungs. Furthermore, the accumulation of galectin-3 in alveoli was not observed when neutrophil extravasation was induced by alveolar E. coli infection, which is known to induce β2 integrin-dependent neutrophil extravasation in lungs. In addition, our data suggest that galectin-3 is released from the macrophages exposed to S. pneumoniae membrane fraction, and that in vitro galectin-3 supports neutrophil adhesion to an endothelial cell layer by cross-linking. Thus, we propose that galectin-3 acts as an adhesion molecule which mediates the adhesion of neutrophils to the endothelium during streptococcal pneumonia, in which the extravasation of neutrophils is suggested to be predominantly β2 integrin independent (3, 42).
While the majority of classical adhesion molecules are anchored to the cell surface, soluble galectin-3 has been proposed to act as an adhesion molecule for cell-cell and cell-cell matrices interaction in other systems by cross-linking cells and matrices (12, 13, 14); e.g., galectin-3 promotes the adhesion of neutrophils to laminin in vitro (33). Recently, kinetics of galectin-3 binding to laminin was analyzed by using surface plasmon resonance, with which the molecular interactions can be studied under the dynamic reaction equilibrium (rather than the stable equilibrium) in real time. This study suggests that the interaction of galectin-3 with laminin is closer to that of β2 integrin (high-affinity status) to its ligand, ICAM-1 (43, 44, 45), than that of other adhesion molecules, selectins with their glycoconjugate ligands (46, 47). For example, the Kd of galectin-3 is 1 μM, while Kd of a β2 integrin, LFA-1, is 0.5 μM. In contrast, selectins have lower binding affinity (Kd = 100–320 μM), consistent with the fact that selectins are involved in transient adhesion rather than the formation of tight adhesion (46, 47). As adhesion of neutrophils mediated by β2 integrins (high-affinity status) is known to be sufficient to form tight adhesion (3), the kinetics studies suggest that galectin-3, which has comparable kinetics parameter to β2 integrin, is able to mediate tight cell adhesion.
Relatively high concentrations of galectin-3 (>0.7 μM) are necessary to demonstrate adhesion activity in vitro. Similar concentrations of galectin-3 are also required to augment oxidative burst or to induce chemotaxis of monocytes and macrophages (25, 26, 27, 31). It has been suggested that this concentration is necessary for the efficient oligomerization of the lectin and its conversion into an effector capable of participating in multivalent interactions (23, 24). In some cells, the concentration of galectin-3 in cytoplasm is suggested to be relatively high (>4.5 μM) (24). In fact, as shown in Fig. 5⇑B, galectin-3 was found on the surface of alveoli-emigrated neutrophils, suggesting that the levels of galectin-3 in infected lesions in lungs reach to the concentration that is required for the oligomerization of galectin-3. In BAL fluids obtained from mice infected with S. pneumoniae, the concentration of galectin-3 is at least 60 nM, assuming that an airway surface liquid volume of mouse is ∼0.2 ml. Thus, when correcting for the large dilution effects in lavage fluids and the probability that the lectin is deposited at localized pulmonary sites, it seems very likely that functionally significant concentrations of galectin-3 would be reached during streptococcal infection.
In lung infection, Doerschuk and colleagues (3, 42) have demonstrated elegantly that emigration of neutrophils in lungs can proceed by β2 integrin-independent as well as the classical β2 integrin-dependent pathways. Their studies indicate that alveolar infection with E. coli induces the β2 integrin-dependent neutrophil emigration, while infection with S. pneumoniae preferentially induces the β2 integrin-independent one (7, 8, 9, 10). In regard to the latter, molecular mechanism of the specific induction of β2 integrin-independent pathway and the participating adhesion molecules have remained elusive (11). Interestingly, the accumulation of galectin-3 in infected alveoli was evident only in lung infected with S. pneumoniae but not with E. coli, despite the fact that both pathogens induced neutrophil emigration to the infected alveoli. Thus, the data suggest that galectin-3 is implicated in β2 integrin-independent neutrophil extravasation induced by S. pneumoniae pulmonary infection.
β2 integrin-independent neutrophil extravasation is also observed in some peritonitis models and, interestingly, the presence of inflammatory macrophages in the peritoneal cavity appears to be prerequisite for the induction of this β2 integrin-independent pathway (48, 49). Thus, it is demonstrated that inflammatory macrophages can elicit β2 integrin-independent neutrophil emigration when transferred to naive peritoneal cavities before the induction of acute peritonitis (3, 48). Winn et al. (49) have suggested that neutrophil emigration into inflamed peritoneal cavity at the late stage of inflammation (∼24 h) but not the early stage (∼6 h) is independent of β2 integrin. As inflammatory activated macrophages emigrated into peritoneal cavity in the late stage of inflammation (24 h) but not the early stage (50), those observations indicate that inflammatory macrophages but not peritoneal resident macrophages play a role in β2 integrin-independent neutrophil diapedesis. Interestingly, such inflammatory peritoneal macrophages, but not resident peritoneal macrophages, actively secrete galectin-3 (19). We demonstrate in this study that the onset of β2 integrin-independent neutrophil emigration is closely associated with alveolar accumulation of galectin-3, which mediates human neutrophil adhesion to human endothelium in vitro. Together, those data suggest that galectin-3 secreted from inflammatory macrophages acts as an adhesion molecule involved in β2 integrin-independent neutrophil extravasation.
Recently, galectin-3null mutant mice were established (51, 52). Those mice are viable and do not show overt abnormality under conventional animal housing condition (51, 52). However, it is suggested that in galectin-3null mice, neutrophil emigration into peritoneal cavity is reduced at the late stage of inflammation (∼1–4 days after induction) while neutrophil emigration is unaffected at the acute stage (∼4–6 h after induction) (51, 52). Characterization of galectin-3null mice has not been extensively conducted yet. Therefore, it is not clear whether the reduced rate of peritoneal neutrophil emigration found in those mice is strictly related to the lack of galectin-3 or results from secondary systemic phenotypic alterations, which have now been evoked in some adhesion molecule knockout mice (7, 8, 53). However, considering the reports mentioned above (48, 49), which suggest that inflammatory macrophages are prerequisite for the induction of β2 integrin-independent diapedesis of neutrophils, lack of galectin-3 release from inflammatory macrophage in galectin-3null mice could affect this process.
In addition to the role of galectin-3 as an adhesion molecule, there is a report that suggests that galectin-3 is a chemoattractant molecule for monocytes and macrophages, as demonstrated in vitro using Boyden chambers and in vivo using a dorsal air pouch model (31). In our pneumococcal pneumonia mouse model, macrophages/monocytes are actively recruited to the lung only 48 h after infection (34), while release and accumulation of galectin-3 in BAL fluids occurred 4 h postinfection. Thus, in this model it seems unlikely that galectin-3 served as a prime chemoattractant for monocytes/macrophages. Our recent studies suggest that conventional chemokines for monocytes such as macrophage-inflammatory protein-1, monocyte chemoattractant protein-1, and RANTES do not seem to participate in monocyte/macrophage recruitment in pneumococcal pneumonia as actively as in other inflammatory models, including the air pouch system (34, 36). Therefore, the recruitment of monocytes/macrophages in streptococcal pneumonia may be regulated in a different manner than into the air pouch and galectin-3 may not serve as efficiently to attract the cells in this pneumonia model.
In conclusion, our data suggest that galectin-3 acts as a novel adhesion molecule that mediates neutrophil adhesion to endothelium and is involved in β2 integrin-independent extravasation of neutrophils during alveolar S. pneumoniae-initiated inflammation. Recent works suggest that β1 integrins are implicated in the recruitment of neutrophils (54, 55, 56), although it has not been clarified yet whether β1 integrins play a role in the β2 integrin-independent neutrophil emigration observed in streptococcal pneumonia. However, galectin-3 and possibly various other types of non-β integrin adhesion proteins may play critical roles in streptococcal pneumonia. Finally, extravasation is known to prime/activate neutrophils (57), and in vitro galectin-3 can stimulate superoxide production through activation of NADPH-oxidase in exudated neutrophils (27). Thus, these recent studies provide increasing evidence that galectin-3 is a novel type of cytokine that can modulate the output of neutrophil-mediated innate immunity either as an adhesion molecule or as an immunomodulator.
We thank Drs. R. Colin Hughes, Yves Bergeron, Cristina L. Ward, and Masahiko S. Satoh for discussion and critical reading of the manuscript, Dr. Philippe Tessier for discussion, and Drs. K. Kasai and J. Hirabayashi for human galectin-3-containing expression plasmid.
↵1 This work was supported by Canadian Institutes of Health Research Grant MT-15498 (to S.S.). S.S. was supported by salary support for a new investigator from Fonds de la Recherche en Santé du Québec.
↵2 Address correspondence and reprint requests to Dr. Sachiko Sato, Glycobiology Laboratory, Center de Recherche en Infectiologie du Centre Hospitalier de l’Université Laval, 2705 Boulevard Laurier, Ste-Foy, Québec, Canada G1V 4G2. E-mail address:
↵3 Abbreviations used in this paper: CRD, carbohydrate recognition domain; BAL, bronchoalveolar lavage; BAL sup, BAL supernatant.
- Received June 8, 2001.
- Accepted December 17, 2001.
- Copyright © 2002 by The American Association of Immunologists