Abstract
Extravasation of leukocytes from peripheral blood is required for an effective inflammatory response at sites of tissue infection. Integrins help mediate extravasation and navigate the leukocyte to the infectious source. A novel role for integrins in regulating the effector response to a cell wall component of fungal pathogens is the subject of the current study. Although phagocytosis is useful for clearance of unicellular fungi, the immune response against large, noningestible hyphae is not well-understood. Fungal β-glucan, a pathogen-associated molecular pattern, activates production of superoxide anion in leukocytes without the need for phagocytosis. To model polymorphonuclear leukocyte (PMN) recognition of fungi under conditions in which phagocytosis cannot occur, β-glucan was covalently immobilized onto tissue culture plastic. Plasma membrane-associated respiratory burst was measured by reduction of ferricytochrome C. Results show that the human PMN oxidative burst response to immobilized β-glucan is suppressed by addition of β1 integrin ligands to the β-glucan matrix. Suppression was dose dependent and steric hindrance was ruled out. β1 integrin ligands did not affect respiratory burst to ingestible β-glucan-containing particles, phorbol esters or live yeast hyphae. Furthermore, in the absence of matrix, Ab activation of VLA3 or VLA5, but not other β1 integrins, also prevented β-glucan-induced respiratory burst. β1-induced suppression was blocked and burst response restored by treating neutrophils with either the cell-binding fragment of soluble human Fn, cyclic RGD peptide, or Ab specific to VLA3 or VLA5. Together these findings extend the functional role of β1 integrins to include modulating PMN respiratory burst to a pathogen-associated molecular pattern.
Polymorphonuclear leukocytes (PMNs)4 are the innate immune cells responsible for protection of the host against opportunistic fungal infections. Neutropenia is a primary risk factor for systemic candidiasis, a frequent hospital-acquired infection associated with 40% mortality and causing ∼10,000 deaths per year in the United States (1, 2, 3). β-glucans, a class of long-chain polymers of glucose in β-(1, 3) (1, 6) linkages, are a structural component of the fungal cell wall and a pathogen-associated molecular pattern (PAMP) that permits leukocytes to recognize and respond to fungi without need for opsonization (4). Although PAMPs such as β-glucan are a significant component of nonself recognition, cellular mechanisms that regulate the host response to these molecules are not well-understood.
Integrins are cell surface receptors that mediate the bidirectional transfer of signals across the plasma membrane. They are heterodimeric transmembrane proteins, each with a distinct α subunit noncovalently associated with a β subunit, and comprise an extensive superfamily of receptors that mediate cell-cell and cell-protein adhesive interactions (5). Human PMNs express a number of integrins including those of the β1 and β2 subfamilies (6). The β1 family predominantly binds extracellular matrix (ECM) proteins such as collagen, fibronectin (Fn), and laminin primarily through the recognition of two consensus binding motifs: Arg-Gly-Asp (RGD) and Leu-Asp-Val (LDV) (5). The β2 receptors are engaged by complement, ICAM-1, fibrinogen, factor X, as well as β-glucan (7).
Peripheral blood PMNs respond to infection within tissues by extravasation from the vasculature and chemotaxis through the stroma. These processes are regulated by multiple integrin-mediated interactions with cellular counterreceptors such as ICAM-1 and VCAM-1 and with components of the ECM (6). Integrin depletion or blockade impedes the ability of the host to mount a complete inflammatory response, as is observed in patients with leukocyte adhesion deficiency who lack β2 integrins and suffer from recurrent opportunistic infections (8).
The β2 integrin complement receptor 3 (CR3) (αMβ2; CD11b/CD18) has been shown to serve as a β-glucan receptor on human PMNs (9, 10, 11). Previous work from this laboratory found that CR3 activation by β-glucan up-regulates the directed migration of neutrophils toward a gradient of fMLP (12). A mechanism of cross-talk among members of the β1 and β2 integrin families was proposed to account for this finding, such that the activation of CR3 by β-glucan affected β1 integrin-dependent PMN chemotaxis. In this study, we considered whether there is reciprocal integrin cross-talk whereby occupancy of β1 integrins might regulate the response of PMNs to β-glucan.
The virulence of dimorphic fungi such as Candida albicans depends on their ability to switch between small, rounded blastoconidia and filamentous forms (13). Whereas host phagocytes can clear blastoconidia by ingestion, hyphal forms are too large to be phagocytosed. The primary cytotoxic response of neutrophils to hyphal fungi is the respiratory burst, i.e., the production of oxidative intermediates by the plasma membrane NADPH oxidase complex (14, 15). Work in this laboratory (16, 17) demonstrated that β-glucan is an accessible component of the hyphal cell wall that contributes significantly to the induction of a PMN respiratory burst response. In addition, we reported that β-glucan enhances PMN chemotaxis on ECM protein via specific β1 integrin regulation (12). We hypothesize here that β1 integrins regulate β-glucan-induced production of reactive oxygen intermediates by PMNs. Our current studies show that β1 integrins suppress the nonphagocytic respiratory burst to immobilized β-glucan. Suppression is lost upon PMN contact with hyphae, ensuring an effective oxidative response.
Materials and Methods
Reagents
Highly purified, soluble yeast β-glucan (PGG-Glucan; formerly Betafectin now ImPrime PGG) and endotoxin-free whole glucan particles (WGP) isolated from Saccharomyces cerevisiae were provided by Biothera. The β-glucan preparation contained <0.02% protein, <0.01% mannan, and 1% glucosamine. Zymosan was obtained from Sigma-Aldrich. Endotoxin-free BSA and dextran (∼80–120 kDa molecular mass) were purchased from Sigma-Aldrich and were suspended in Dulbecco’s PBS solution (dPBS). dPBS, HBSS, and DNase were obtained from Invitrogen Life Technologies. PMA was obtained from Sigma-Aldrich and kept as a stock of 20 μM in ethanol. Histopaque (Ficoll-Hypaque), fMLP, polymyxin B sulfate, ferricytochrome C were also purchased from Sigma-Aldrich. Rat tail collagen type I, fibronectin, vitronectin, collagen type IV, and poly-d-lysine were from (Fisher Scientific). Anti-human VLA1 (5E8D9) and VLA2 (A2-IIE10) were purchased from Upstate Biotechnology. Anti-VLA3 (P1B5), anti-VLA4 (P1H4), anti-VLA5 (P1D6), anti-VLA6 (NKI-GoH3), and the fibronectin cell-binding fragment were obtained from Chemicon International. The synthetic peptides cyclo-RGDfV and cyclo-RADfV were obtained from Peptides International. 5-(and-6) Chloromethyl-2′,7′-dichlorodihydrofluoroscein diacetate, acetyl ester (CM-H2DCFDA) was purchased from Molecular Probes. All reagents used contained <0.1 pg/ml endotoxin as determined by Limulus amebocyte lysate screening (BioWhittaker). When necessary, endotoxin removal was achieved using immobilized polymyxin B (Affinity Pak Detoxi-Gel; Pierce) followed by repeat Limulus testing.
Cell preparation
PMNs were isolated from blood donated by healthy human volunteers collected in EDTA-containing Vacutainer tubes (BD Biosciences). Histopaque was used for initial cell separation followed by sedimentation through 3% dextran. Contaminating erythrocytes were removed by hypotonic lysis, yielding a >95% pure neutrophil preparation of >90% viability by trypan dye exclusion. PMN were suspended in HBSS for adhesion assays or indicator-free HBSS for oxidative burst assays.
Preparation of microtiter plates: immobilization of polysaccharides, ECM proteins, and mAbs
For polysaccharide immobilization, varying concentrations of β-glucan or dextran were diluted into dPBS and plated in volumes of 100 μl onto a Costar 96-well Universal Binding polystyrene plate, manufactured with a photoactivatable cross-linking agent (Corning). The plate was incubated at 37°C in a humidified incubator with 5% (v/v) CO2 for 60 min to allow for adsorption. After incubation, the liquid was removed by flicking or aspiration and the plate was briefly irradiated in a UV Stratalinker 1800 (Stratagene). Photoactivation of the plate surface allows formation of a covalent linkage with the passively adsorbed carbohydrates. The method has been shown to have an approximate binding efficiency of 10–20% (18). The final content of β-glucan on the assay plate is similarly reduced. However, for consistency, the β-glucan concentrations reported in this study reflect the applied concentration. Plates were then blocked with low endotoxin BSA for 20–30 min in a 37°C incubator with 5% (v/v) CO2, at a concentration of 10 μg/ml. After blocking, wells were washed three times with dPBS before use.
In cases of complex coatings of polysaccharide and ECM components or Abs, the β-glucan or dextran solution was supplemented to the indicated concentration of ECM component or Ab. A total of 100 μl of this freshly prepared solution was then added per well with incubation, cross-linking, and blocking performed as described above.
Oxidative burst assay
Respiratory burst activity was determined by measuring the colorimetric change caused by superoxide anion (O⨪2) reduction of ferricytochrome C. Cells were plated at a concentration of 3 × 105 cells/well, in replicates of 3–6 wells, at a volume of 100 μl/well onto Costar Universal Binding plates coated with immobilized substrate prepared as described above. Where indicated, PMNs were assayed in the presence of fMLP, WGP, or zymosan as an additional stimulant added at time zero of the assay. WGP and zymosan were sonicated before use to produce single particle suspensions. Stimulant concentrations were from published reports and optimized for maximum generation of superoxide through preliminary work. A total of 20 nM PMA was used as a positive control for superoxide production (data not shown) while stimulants in the absence of immobilized polysaccharides in wells blocked with endotoxin-free BSA served as our negative controls. For Ab-blocking experiments, cells were preactivated with 1 nM fMLP for 10 min at room temperature to ensure active integrin conformation. Cells were then incubated with indicated mAbs for 30 min on ice before being plated. Finally, 100 μl/well of 100 μM ferricytochrome C was added to each well and absorbance was measured every 10 min for 60–90 min at dual wavelengths of 550 and 630 nm at 37°C using a Microplate BIO Kinetics Reader (BIO-TEK Instruments), running DeltaSoft3 software. Controls were done with superoxide dismutase (300 U/ml) which inhibited reduction of ferricytochrome C by O⨪2 (data not shown). Superoxide production was calculated for 60 min (unless otherwise indicated) with the following formula: (ΔE550 nm-630 nm) × 15.238 = nM/well; 15.238 is a predetermined absorbance constant and the final units are nanomoles O⨪2/3 × 105 cells/h.
To measure the generation of reactive oxygen species by PMN against live yeast, PMNs were suspended in HBSS (5 × 106/ml) containing 8 μM CM-H2DCFDA and equilibrated at room temperature in the dark for 30 min. CM-H2DCFDA is a cell-permeant indicator for the respiratory burst that becomes fluorescent upon oxidation. Cells were washed once in an excess volume of HBSS without cations and suspended to a concentration of 6 × 106/ml in HBSS with cations. C. albicans hyphae were grown on Costar 96-well high-binding plates (Fisher) coated with fibronectin or BSA to promote adhesion to the plate surface and minimize disturbance through multiple assay steps. Plates were prepared by incubating with 100 μl of 10 μg/ml fibronectin or BSA diluted in PBS for 1 h at 37°C, flicking to remove solution and washing once with PBS. Yeast hyphae were grown as described below. Before use in burst assay, filamentous yeast was washed by flooding wells with PBS and removing supernatant by aspiration. Hyphae were then preincubated with 50 μl given concentrations of fibronectin diluted into HBSS for 25 min before addition of 50 μl of PMNs. Under all conditions, 3 × 105 PMNs in HBSS (and fibronectin, where indicated) were added to each well for a total reaction volume of 100 μl/well and plates were incubated for an hour at 37°C. Fluorescence was measured using an FL800 Microplate Fluorescence Reader (BIO-TEK Instruments) at excitation/emission wavelength settings of 485/530 nm and a sensitivity setting of 75–80. Results of 4–6 replicate wells were calculated as the difference in measured fluorescence at 0 min subtracted from that at 60 min.
Yeast preparation
C. albicans was obtained from American Type Culture Collection (Wasson, strain 24433) and streaked onto Sabouraud agar plates (Difco) then allowed to form colonies overnight at 37°C. A single colony was picked, seeded into 10 ml of yeast extract-peptone-dextrose medium consisting of yeast extract, bactopeptone (both from Difco), and dextrose (Sigma-Aldrich) and grown overnight at 37°C with vigorous agitation (225 rpm) on a platform shaker. After culture in yeast extract-peptone-dextrose, blastoconidia were washed with PBS and counted on a hemacytometer under ×10 magnification, medium 199 supplemented with Earle’s balanced salt solution, l-glutamine, and 25 mM HEPES (BioWhittaker) was inoculated with 7 × 104 CFU/ml. A total of 100 μl/well of yeast suspension was distributed to prepared 96-well plates and filament formation was induced by incubation at 37°C overnight. The filamentous phenotype was confirmed by light microscopy.
ELISA
The binding of β-glucan to lactosylceramide or ECM protein was measured as previously described (19). Briefly, lactosylceramide (Matreya) was dissolved in ethanol at 1 mg/ml and used as a positive β-glucan binding control. Fibronectin, collagen type I, or BSA were dissolved in PBS at 1 mg/ml. Aliquots (20 μl) were applied in quadruplicate to the wells of a 96-well polystyrene plate (Costar) and air-dried. Plates were blocked by incubation with 300 μl/well of 1% gelatin (w/v) in PBS at 37°C for 1 h. Plates were rinsed with PBS and equilibrated at 37°C for 30 min before addition of 0.1 μg of β-glucan/well in 100 μl of PBS. Plates were incubated for 1 h at 37°C, rinsed with PBS to remove nonadsorbed β-glucan, then maintained at 37°C. In other experiments, β-glucan was covalently immobilized to Costar 96-well universal binding polystyrene plates in the presence or absence of ECM protein as described in the main Materials and Methods. These plates were then blocked with 1% gelatin (w/v) in PBS at 37°C for 1 h. For both methods, bound β-glucan was quantified by ELISA using a specific mAb in mouse ascites at a final concentration of 7 μg/ml (100 μl/well) (BFDIV; provided by Biothera). The generation of this Ab is described in Wakshull et al. (Ref. 20 ; http://patft.uspto.gov). Murine IgM (clone TEPC 183; Sigma-Aldrich) was used at the same concentration as a specificity control. Plates were incubated for 2 h at 37°C and washed three times with PBS containing 0.1% Tween 20 (polyoxyethylene sorbitan monolaurate; Sigma-Aldrich). Goat F(ab′)2 anti-mouse IgM conjugated to HRP (2 μg/ml, 100 μl/well; BioSource International) was added and incubated for 1 h at 37°C. The plates were washed three times and 3,3′, 5,5′-tetramethylbenzidine (Kirkegaard & Perry Laboratories) was added and incubated for 5 min at room temperature. Stop solution (Kirkegaard & Perry Laboratories) was added and the color development was read on a spectrophotometer at 450 nm with a reference wavelength of 630 nm on a Microplate BIO Kinetics Reader (Bio-Tek Instruments), using DeltaSoft3 software.
Data presentation
For superoxide production figures, values presented are mean ± SEM of results obtained from 3–10 independent experiments each with replicates of 3–6 wells. In superoxide production assays presented as a percentage of β-glucan-mediated burst, results represent mean percentage of O⨪2 production with respect to that elicited by 2.5 μg/well immobilized β-glucan ± SEM of 3–10 independent experiments, each with six replicate samples. Statistical analysis was performed using ANOVA-Newman-Keuls or the Student t test as appropriate. The null hypothesis was rejected if p ≤ 0.01.
Results
Fn suppresses the PMN respiratory burst response to β-glucan
To model the regulation of the respiratory burst under conditions where phagocytosis cannot take place, we covalently linked β-glucan, Fn, or other ECM proteins to a substratum and assayed superoxide (O⨪2) release of normal human PMNs as a measure of the plasma membrane respiratory burst. PMNs generated a robust response to immobilized β-glucan but not to dextran, a polysaccharide control (Fig. 1⇓A). The magnitude of the burst was reduced when β-glucan was immobilized with increasing amounts of Fn (Fig. 1⇓A). The respiratory burst of PMNs that were exposed to β-glucan-containing particles (zymosan or whole glucan particles) was not affected by the presence of Fn, suggesting that Fn regulates the respiratory burst to β-glucan when its presentation is restricted to the cell surface, but not when it is in ingestible form (Fig. 1⇓B). The oxidative response to PMA (17.6 ± 1.5 nM O⨪2/3 × 105 cells/h) was unaffected by the composition of the matrix (data not shown).
Fn inhibits PMN respiratory burst to immobilized β-glucan but not to zymosan or WGPs. O⨪2 production of PMNs as measured by ferricytochrome C reduction. A, Response to 2.5 μg β-glucan (▪) or dextran (□) immobilized with increasing concentrations of Fn on solid substrate. ∗, p < 0.01 vs β-glucan alone. Response to stimulation with 20 nM PMA was unaffected by presence or absence of Fn and/or β-glucan. B, PMNs were plated onto Fn (200 ng/well) and treated with β-glucan-containing particles, zymosan (100 μg/well), or whole glucan particles (100 μg/well). C, O⨪2 production by fMLP-activated PMNs treated with increasing concentrations of a soluble Fn central cell-binding peptide then stimulated with immobilized β-glucan (2.5 μg/well) and Fn (100 ng/well). All results represent mean percentage of O⨪2 production with respect to that elicited by 2.5 μg/well immobilized β-glucan. A total of 2.5 μg/well immobilized BSA (−) or dextran (Dex) were used as a control at each peptide concentration tested and values represent the mean for these pooled conditions. ∗, p < 0.01 vs β-glucan+Fn.
To demonstrate the receptor-mediated nature of the suppression by Fn, receptor activity was blocked by incubating cells with the 120-kDa cell-binding fragment of soluble human Fn before assay. Pretreatment restored as much as 75% of the PMN capacity for respiratory burst on β-glucan, despite the presence of immobilized Fn (Fig. 1⇑C). A technical detail that should be noted here is that to ensure active integrin conformation for efficient binding of soluble ligand, all blocking experiments were performed using PMNs primed with 1 nM fMLP (suboptimal for respiratory burst). Adhesion to Fn-coated surfaces, however, does not require preactivation to the high-affinity form (21).
The substantial respiratory burst to β-glucan seen when Fn receptors were blocked suggests that the immobilized Fn does not physically interfere with the ability of the cells to recognize the β-glucan. To further examine whether Fn can mask or sterically interfere with β-glucan, increasing concentrations of Fn were covalently linked with a fixed amount of β-glucan. ELISAs were performed using a β-glucan specific mAb (BFDIV), which is capable of detecting exposed carbohydrate (16). Detection of β-glucan was unaffected at all concentrations of Fn tested including those in excess of that needed for maximal suppression of the oxidative response to β-glucan (Fig. 2⇓A). ELISAs were also used to confirm that immobilized Fn and other ECM proteins are unable to directly bind or trap β-glucan (Fig. 2⇓B). Taken together, these findings provide serologic evidence for a lack of an interaction between β-glucan and Fn that might obviate cellular recognition of β-glucan.
Lack of interaction between β-glucan and Fn. A, β-glucan (2.5 μg/well) was cross-linked onto polystyrene 96-well plates with increasing amounts of Fn. β-glucan-specific Ab (BFDIV) was used to detect β-glucan and assessed by ELISA. B, Wells were precoated with lactosylceramide or indicated proteins before addition of β-glucan. After washing, adsorbed β-glucan was quantified by ELISA with BFDIV. Data in each figure represent mean ± SD of quadruplicate samples and are representative of three independent experiments; ND, not detectable.
Several ECM proteins are able to suppress PMN response to β-glucan
To investigate whether inhibition of the PMN oxidative response to β-glucan was Fn specific or a general property of ECM proteins, β-glucan was combined with varying concentrations of laminin, vitronectin, or collagens type I or IV. All matrix proteins tested had inhibitory effects similar to Fn on the PMN response to β-glucan (Figs. 3⇓, A and C–F). In contrast, poly-d-lysine and heparin control ligands had no effect (Fig. 3⇓A).
β1 integrin ECM ligands inhibit respiratory burst to β-glucan; Fn inhibition of β-glucan-stimulated burst is relieved by a soluble cyclic RGD peptide. O⨪2 production by human PMNs as measured by ferricytochrome C reduction. A, Response to 2.5 μg/well β-glucan immobilized with 200 ng/well ECM proteins: laminin, vitronectin, collagen I, or collagen IV. Poly-d-lysine and heparin were used as non-β1 integrin-binding control molecules. ∗, p < 0.01 vs β-glucan. B, Response of fMLP-activated PMNs pretreated with increasing concentrations of cyclic RGD peptide (▪) to block Fn binding or control peptide (□), then stimulated with 2.5 μg/well β-glucan immobilized with 100 ng/well fibronectin. Results in A and B represent mean percentage of O⨪2 production with respect to that elicited by 2.5 μg/well immobilized β-glucan. A total of 2.5 μg/well immobilized dextran (Dex) was used as a control at each peptide concentration tested and values represent the mean for these pooled conditions. ∗, p < 0.01 vs β-glucan+Fn. C–F, Superoxide production in response to mixtures of 2.5 μg/well immobilized β-glucan with increasing concentrations of various ECM proteins: C, laminin; D, vitronectin; E, collagen I; F, collagen IV. Samples denoted (−) contain only β-glucan; those denoted (#) contain no β-glucan. Data shown in C–F represent mean ± SD of triplicate samples and is representative of at least three individual experiments for each panel.
Cyclic RGD peptides serve as a mimetic to activate integrins when used at nanomolecular concentrations, but inhibit integrin binding at micromolecular concentrations by competition with RGD-binding sequences of ligand (22). Pretreatment of PMNs with the cyclic RGD peptide, cyclo-RGDfV, but not control peptide cyclo-RADfV, restored virtually 100% of the respiratory burst to β-glucan in the presence of Fn (Fig. 3⇑B), thus supporting an RGD-dependent mechanism of suppressing the respiratory burst to β-glucan, consistent with a β1 integrin-mediated effect. The reagent concentrations used in these experiments are in accordance with what other reports have shown to be effective to inhibit cell binding to matrix (23).
VLA3- or VLA5-dependent suppression of the respiratory burst to β-glucan
To identify specific β1 integrins capable of mediating suppression of the oxidative response to β-glucan, Abs against the α subunits of VLA-1 through VLA-6 were immobilized on to 96-well plates to induce cross-linking and activation of their cognate integrin on the PMN surface. Selectively activating VLA3 or VLA5, but not other β1 integrins inhibited β-glucan-induced O⨪2 production (Fig. 4⇓A) and did so in a dose-dependent fashion (Fig. 4⇓, B and C). To confirm the suppressive functions of VLA3 and VLA5, we pretreated fMLP-primed PMNs with specific blocking Abs before assay on β-glucan plus Fn. Ab blockade of either VLA5 or VLA3 was sufficient to prevent Fn binding and suppression of respiratory burst, while Ab blockade of VLA-4, known to be expressed on activated PMNs (24), was not (Fig. 4⇓D). Blocking both VLA5 and VLA3 together was not additive over either Ab alone, suggesting nonredundant roles for these integrins in mediating suppression of the response to β-glucan.
Ab activation of β1 integrins VLA3 or VLA5 mimics Fn inhibition of O⨪2 production by PMNs; Fn inhibition is relieved by integrin-blocking Abs. O⨪2 production by PMNs as measured by ferricytochrome C reduction in response to 2.5 μg/well β-glucan immobilized with 100 ng/well Fn. Response after (A) PMNs were treated with 400 ng/well mAb to β1 integrins VLA1–6, or 400 ng/well isotype control (IgG), and treatment with increasing concentrations of immobilized Ab to (B) VLA5 or (C) VLA3. D, fMLP-activated PMNs were treated with 2.5 μg/well soluble mAb to VLA3, VLA4, VLA5, a mixture of 2.5 μg each of VLA3 and VLA5, or isotype control, then tested for ability to respond to immobilized β-glucan+Fn. All results represent mean percentage of O⨪2 production with respect to that elicited by 2.5 μg/well immobilized β-glucan. A total of 2.5 μg/well immobilized BSA or dextran was used as a control for each Ab concentration tested and the baseline (−) represents the mean for these pooled conditions. ∗, p < 0.01 vs all other groups.
It should be noted that a survey of the literature indicates that the VLA3 Ab used in these experiments (mAb P1B5) is the best characterized blocking Ab specific for the α3 subunit. Specificity has been confirmed using a variety of approaches to rule out cross-reactivity with other β1 integrins and cell surface distribution of VLA3 was shown to be different from that of VLA5 (25, 26, 27).
Fn plus β-glucan down-regulates the PMN response to fMLP
Studies were performed to determine whether the suppression of the respiratory burst induced by a matrix of immobilized Fn plus β-glucan rendered cells refractory to other O⨪2-inducing stimuli. PMNs were placed in wells containing immobilized Fn plus β-glucan and simultaneously stimulated with fMLP. Although immobilized Fn alone and β-glucan alone both permitted a normal burst response to fMLP, the combination of Fn plus β-glucan decreased the respiratory burst to fMLP by 37% as compared with Fn alone (Fig. 5⇓).
Immobilized Fn + β-glucan inhibits respiratory burst to fMLP. PMNs were treated with (▪) or without (□) 1 μM fMLP upon placement into wells containing a matrix of 2.5 μg/well β-glucan immobilized with 100 ng/well Fn, or onto each alone. O⨪2 production was measured by ferricytochrome C reduction. Results represent mean percentage of O⨪2 production with respect to that elicited by 2.5 μg/well immobilized β-glucan. A total of 2.5 μg/well immobilized dextran or BSA was used as control for each condition tested and the baseline (−) represents the mean for these pooled conditions. ∗, p < 0.01 vs all other groups.
Fn does not inhibit PMN respiratory burst to yeast hyphae
This laboratory has recently reported that PMNs stimulated with the pathogenic yeast C. albicans release reactive oxygen species in a β-glucan-dependent fashion. To test whether Fn could affect the ability of PMNs to recognize the yeast, we determined the PMN respiratory burst response to the live filamentous form of C. albicans under several conditions. Cytochrome c reduction, as used in other experiments reported here, was not ideal to measure superoxide production of PMNs against live yeast due to interference with light detection of OD by the layer of Candida filaments (16). To quantify the production of reactive oxygen species in response to the filamentous yeast, we used CM-H2DCFDA, a cell-permeant indicator probe loaded into PMNs that becomes fluorescent when oxidized. This technique yielded results consistent with those obtained when cytochrome c reduction was used to measure the PMN response to immobilized purified β-glucan with or without Fn (data not shown).
Yeast assays were conducted with several approaches to make sure that outcomes were not simply due to experimental conditions. For yeast, conditions included: growth of C. albicans hyphal forms on microtiter wells coated with a range of concentrations of Fn or BSA (as a control); preincubating hyphae with soluble Fn or BSA before addition of PMNs; and allowing the soluble protein to remain for the duration of the assay or removing it before PMN addition. For each of the yeast culture conditions, PMNs were also subjected to several treatments. These included preincubation with Fn (from 0.1 to 10 μg/ml) or addition at the time of assay and pretreatment (or not) of PMNs with fMLP to assure activation of integrins.
For data shown in Fig. 6⇓, yeast hyphae were grown on Fn and soluble Fn remained in the assay medium. The figure illustrates the absence of any Fn effect on the respiratory burst and is representative of results obtained with all other conditions tested (data not shown). Regardless of the Fn concentration or the order in which assay components were assembled, we consistently observed that Fn did not alter the PMN burst response to unopsonized yeast filaments.
Fn does not affect the PMN respiratory burst response to C. albicans hyphae. Yeast hyphae were grown overnight on 1–2.5 μg/well immobilized Fn or BSA (as control). CM-H2DCFDA-loaded PMNs were added with 1 or 10 μg/ml soluble Fn and oxidant production was measured for 1 h. Fluorescence was normalized relative to that of PMN response to hyphae without Fn (0 μg/ml). Data are expressed as mean relative fluorescence ± the SEM from five experiments using PMNs from different donors and 4 to 6 wells of replicates for each condition.
Discussion
Pathogenic yeast and fungi elaborate β-glucan during the course of fungal growth such that β-glucan may be present in the interstitium surrounding a focus of infection (28, 29). Results from our laboratory have shown that a Fn matrix supplemented with fungal β-glucan permits enhanced chemotaxis of human PMNs (12). This effect was shown to be initiated by CR3 (αMβ2) recognition of β-glucan and could be functionally blocked with Abs against either VLA5, VLA3, or the VLA3-associated transmembrane-4 superfamily member CD151 (30). In addition, we demonstrated that yeast cell wall β-glucan is a significant element in the recognition of hyphae by PMNs, acting as a determinant for the induction of a respiratory burst (16, 17). In the current study, we show that the ability of human PMNs to produce superoxide against immobilized β-glucan is suppressed by Fn and selective activation of VLA3 or VLA5, but that Fn does not limit the respiratory burst to yeast hyphae. The emerging picture is that activation of β1 integrins may determine an inflammatory cell phenotype that is well-suited to eliminate the virulent, filamentous form of fungi by accelerating chemotaxis toward the foci of infection while suppressing the premature release of oxidants until the neutrophil establishes direct multifocal contact with hyphae.
In developing the assay system used in this study, the possibility that the addition of Fn to immobilized β-glucan could sterically mask β-glucan was ruled out; β-glucan was consistently detected by Ab, even after being combined with concentrations of Fn that were an order of magnitude in excess of that needed for maximal suppression of the respiratory burst (Fig. 2⇑A). In addition, there was no detectable binding of β-glucan to the ECM proteins Fn or collagen (Fig. 2⇑B), suggesting lack of physical complexing between β-glucan and these proteins. Experimental outcomes were unaffected whether matrix protein was added before, after, or at the time of β-glucan cross-linking (data not shown). When taken together, these data show that β-glucan and ECM are independently perceived by PMNs.
Specific clustering and activation of VLA3 or VLA5 was induced with immobilized mAbs to mimic ligand binding. Despite the absence of matrix protein, this strategy reproduced the Fn effect of suppressing respiratory burst to β-glucan (Fig. 4⇑, A–C). Furthermore, when cell surface VLA5 was blocked, PMN burst response to immobilized β-glucan plus Fn was restored (Fig. 4⇑D). These findings are consistent with the conclusion that activation of VLA5 has a primary role in moderating PMN response to a PAMP. Interestingly, the primary PMN receptor for β-glucan is also an integrin, CR3 (9, 12, 31), indicating that β1 to β2 integrin cross-talk may serve to explain our observations.
Contribution of VLA5 to regulating the response to a PAMP, while novel, is consistent with VLA5 being the dominant receptor for Fn. It was surprising to find, though, that VLA3 manipulation induced a similar response to immobilized Fn and β-glucan (Fig. 4⇑, A and C), because neither one represents a prototypic ligand for this integrin. Laminin-5 is the canonical ligand for VLA3 and binding is RGD independent (32), however, VLA3 has been shown to recognize RGD-containing ligands including Fn in some reports (27, 33). Although our results suggest behavior that is uncharacteristic for VLA3, they cannot be explained by Ab cross-reactivity. The specificity of the VLA3 Ab used in these studies has been extensively characterized and cross-reactivity with other β1 integrins ruled out (25, 27).
Although VLA3 and VLA5 function to suppress the respiratory burst response to immobilized β-glucan, it has not been determined whether suppression requires coincident activity of both integrins or can be induced by engagement of either one alone. Because the selective activation of either VLA3 or VLA5 was sufficient to inhibit β-glucan-induced O⨪2 production, blocking studies unexpectedly revealed that loss of either integrin alone, while preserving the function of the other, was sufficient to abrogate the suppression. These results suggest that the activity of both integrins may be required and nonredundant. Finding that inhibition of superoxide release is not restricted to any one β1 family member acquires physiologic significance when it is considered that an extravasated PMN will engage at least one, and more likely multiple, integrins regardless of the tissue into which it has migrated.
Additional experiments were conducted to determine whether the combination of immobilized Fn and β-glucan resulted in a generalized suppression of cells to generate an adhesion-dependent respiratory burst. It has been reported that PMNs generate an oxidative burst response to soluble inflammatory mediators under conditions where leukocyte integrins were ligated by matrix (34, 35). The present study shows that a response is permitted to proceed without impairment when PMNs are stimulated with the agonist fMLP on a matrix of Fn (Fig. 5⇑). In comparison, however, fMLP stimulation of cells on a matrix of Fn plus β-glucan showed greatly reduced O⨪2 release (Fig. 5⇑). Because the respiratory burst to the membrane-permeable agonist PMA was unaffected by the composition of the matrix (data not shown), the ability of the NADPH oxidase to assemble and properly function was not compromised. Therefore, it appears that Fn plus β-glucan renders PMNs refractory to further stimulation. Future experiments will determine whether this state can be overcome by greater amounts of β-glucan. If so, it would suggest that the ratio of ECM ligands to β-glucan ligands is one of the cues used by PMNs to determine when to initiate antimicrobial activity.
The coating of flat-bottom tissue-culture plates with matrix and β-glucan (as shown in Figs. 1–5⇑⇑⇑⇑⇑) allows PMNs to encounter both ligands on the same surface plane. In experiments to determine whether ECM affects the neutrophil response to hyphae, we considered that Candida filaments would protrude upward from the surface of the plate, allowing PMNs to contact the filaments before engaging matrix. As described in Results, we addressed this concern by expanding the experimental design to afford every possible opportunity for the PMNs to engage both fibronectin and the hyphae together as would occur in an infected tissue. Regardless of the technical variations used, all experiments consistently showed that integrin activation with Fn did not limit the PMN oxidative response to yeast as it did the response to β-glucan alone. Experiments are underway to determine a mechanistic explanation for the differential regulation of integrins on the PMN response to β-glucan. Our work suggests that integrins may prevent oxidative damage to healthy tissue by inhibiting the PMN response to β-glucan until coming into direct contact with fungal hyphae.
The engagement of multiple integrins in the β1 and β2 families is a necessary function for PMN extravasation and directed migration toward a site of infection. This report demonstrates a cross-regulatory relationship between endogenous integrin activation and the response to the fungal PAMP, β-glucan. Additional work will determine whether our observations are unique to β-glucan or if integrin regulation extends to innate immune reactions against other conserved microbial structures.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by grants from the National Institutes of Health (GM-066194 (to J.S.R.) and GM-42859 (to J.E.A.)); Grant Assistance in Areas of National Need from the U.S. Department of Education (to L.M.L. and X.M.O.); the American Heart Association (0535498T (to M.K.)); and allocations to the Department of Surgery by Rhode Island Hospital.
↵2 L.M.L. and X.M.O. contributed equally to the generation of this manuscript.
↵3 Address correspondence and reprint requests to Dr. Jonathan S. Reichner, Department of Surgery, Rhode Island Hospital, 593 Eddy Street, Providence, RI 02903. E-mail address: Reichner{at}Brown.edu
↵4 Abbreviations used in this paper: PMN, polymorphonuclear leukocyte; PAMP, pathogen-associated molecular pattern; ECM, extracellular matrix; Fn, fibronectin; CR3, complement receptor 3; O2⨪, superoxide anion; WGP, whole β-glucan particle; dPBS, Dulbecco’s PBS solution.
- Received June 20, 2006.
- Accepted March 21, 2007.
- Copyright © 2007 by The American Association of Immunologists