Lectin Site Ligation of CR3 Induces Conformational Changes and Signaling*

Background: CR3 is a β2 integrin that contains a lectin-like domain that binds the fungal pathogen-associated molecular pattern β-glucan. Results: Soluble β-glucan stabilizes an intermediate CR3 conformation that induces differential intracellular phosphorylation. Conclusion: CR3 is a signaling pattern recognition receptor for β-glucan. Significance: The CR3 receptor is a target for the design of novel immune modulators. Neutrophils provide an innate immune response to tissues infected with fungal pathogens such as Candida albicans. This response is tightly regulated in part through the interaction of integrins with extracellular matrix ligands that are distributed within infected tissues. The β2 integrin, CR3 (CD11b/CD18), is unique among integrins in containing a lectin-like domain that binds the fungal pathogen-associated molecular pattern β-glucan and serves as the dominant receptor for recognition of fungal pathogens by human granulocytes. β-Glucan, when isolated in soluble form, has been shown to be a safe and effective immune potentiator when administered therapeutically. Currently a pharmaceutical grade preparation of β-glucan is in several clinical trials with an anti-cancer indication. CR3 binding of extracellular matrix, carbohydrate, or both ligands simultaneously differentially regulates neutrophil function through a mechanism not clearly understood. Using FRET reporters, we interrogated the effects of soluble β-glucan on intracellular and extracellular CR3 structure. Although the canonical CR3 ligand fibrinogen induced full activation, β-glucan alone or in conjunction with fibrinogen stabilized an intermediate conformation with moderate headpiece extension and full cytoplasmic tail separation. A set of phosphopeptides differentially regulated by β-glucan in a CR3-dependent manner were identified using functional proteomics and found to be enriched for signaling molecules and proteins involved in transcriptional regulation, mRNA processing, and alternative splicing. These data confirm that CR3 is a signaling pattern recognition receptor for β-glucan and represent the first direct evidence of soluble β-glucan binding and affecting a signaling-competent intermediate CR3 conformation on living cells.

developed for immune priming with anti-infectious and anticancer indications that are being explored clinically (8 -12). Whether released from a fungal burden or administered therapeutically, s␤glu has been shown to influence the functional phenotype of neutrophils through changes in migration, adhesion, and respiratory burst (3,4,13).
How cells are able to discriminate the context of ␤-glucan presentation to provide the appropriate cellular response to direct contact with the intact pathogen (e.g. phagocytosis and production of reactive oxygen species) or to induce a primed state, inflammatory response, or changes in migration that are more suited to activation at a distance by soluble mediators is a topic of active research. Several distinct mammalian receptors for ␤-glucan have been identified in addition to CR3: Dectin-1, lactosylceraminde. and scavenger receptors (14 -17); however, it is still unclear how each receptor distinguishes its ligand. Goodridge et al. (18) have recently shown that although Dectin-1 can bind both soluble and particulate ␤-glucan, signaling is only activated by particulate ␤-glucans, positing a "phagocytic synapse" that allows discrimination of direct microbial contact from soluble mediators. These findings underscore a possible mechanistic niche for CR3 that shows distinct functional responses to both immobilized and soluble ␤-glucan (3, 4, 12, 13, 19 -21). Although functional evidence has been used to argue for CR3 as a receptor for ␤-glucan, there has been no direct evidence of ␤-glucan inducing a change in CR3 conformation at a physiochemical or structural level. In this study we investigate whether CR3 functions as a signaling receptor in response to s␤glu.
Integrins have long been thought to undergo conformational rearrangements upon activation speculated to correlate with increased ligand binding and initiation of intracellular signaling (22). The studies herein use FRET to investigate changes in the spatial separation of the ␣ and ␤ cytoplasmic domains and extracellular conformational changes of CR3 upon differential activation of the I-domain or the lectin-like site. The use of FRET is unique in that it allows highly sensitive changes in integrin activity to be measured in viable cells in real time. We have exploited this tool to investigate changes in conformation of both the intracellular and extracellular regions of CR3 in response to the I-domain ligand, fibrinogen (Fgn), and the lectin-like domain ligand, s␤glu. To extend our investigation we also employed a functional proteomics approach using isogenic K562 cells that differ only in their expression of CR3 to identify phosphopeptides that are modulated in response to s␤glu. These K562-derived cell lines allow a specific interrogation of the CR3-dependent processes and, unlike primary neutrophils, can be propagated in culture to facilitate an MS-coupled technique called stable isotope labeling with amino acids in cell culture (SILAC). The resultant phosphopeptide signature represents the first global demonstration of CR3-dependent signaling in response to s␤glu. This along with the associated conformational changes in intracellular and extracellular domains of CR3 may represent the ␤-glucan initiated primed state.
Neutrophil Preparation-With Rhode Island Hospital Institutional Review Board oversight and approval, neutrophils were isolated from healthy human volunteers into EDTA-containing Vacutainer tubes (BD Biosciences). Histopaque cell separation was followed by gravity sedimentation through 3% dextran. Hypotonic lysis to remove erythrocytes yielded a neutrophil purity of Ͼ95%. Neutrophils were suspended in Hanks' balanced salt solution (without Ca 2ϩ /Mg 2ϩ ) on ice until use. All reagents contained Ͻ0.1pg/ml endotoxin.
Cell Culture and Transfection-K562 cell line was from ATCC (Manassas, VA). K562-CR3 stable cell line was established as described (23). The K562 cell line stably expressing a FRET construct reporting the separation of the cytoplasmic tails of integrin LFA1 (designated in this work as K562:LFA1-FRETa) was established as described (24). K562 cells were cultured in RPMI 1640 Glutamax with 1% Normocin and 10% FBS. Puromycin or G418 were used where needed for selection.
Plasmid Construction-Generation of ␤ 2 -monomeric yellow fluorescent protein (mYFP) constructs was previously de-scribed (24). An ␣ M monomeric cyan fluorescent protein (mCFP) fusion construct with a linker sequence of APEPA-PRPTAAPEPAKRARDPPVAT was generated as described in the supplemental Experimental Procedures.
Generation of K562:CR3-FRETa Stable Cell Line-K562 cells co-transfected with ␣ M (25)mCFP, consisting of the ␣ M chain linked to a monomeric CFP by a 25-amino acid linker, and ␤ 2 (12)mYFP, consisting of the ␤ 2 chain linked to monomeric YFP by a 12-amino acid linker (described in supplemental Fig. S1A) were propagated under selective antibiotic pressure and underwent two rounds of selective enrichment by FACS. Putative clones were single-sorted into 96-well dishes, expanded, and characterized for FRET response and protein expression. From these, a stable line, K562:CR3-FRETa, was established and verified by Western blot (supplemental Fig. S1C), PCR, and sequence analysis. Both the reporter construct and stable line have competent receptor function (supplemental Figs. S1D and S2).
Cell Preparation and Activation for Cytoplasmic Tail FRET-2.5 ϫ 10 5 K562:CR3-FRETa cells or K562:LFA1-FRETa were washed and resuspended in a 0.75-1 ml of L-15 supplemented with 2 mg/ml D-glucose. Cells were treated in heatable glassbottom Delta T dishes or 15-ml tubes for analysis as 30 l drops on coverglass. Cells were treated at 37°C for 15 min in L-15 supplemented with 2 mg/ml D-glucose with or without 1 mM Mn 2ϩ . Treatment was in the presence of 1% human serum where indicated. I-domain ligand binding conditions were 25 g/ml FgnD or 50 g/ml full-length human Fgn. 20 nM phorbol 12-myristate 13-acetate or 50 g/ml s␤glu was supplemented where noted.
Microscopy and Cytoplasmic Tail FRET Analysis-Imaging was conducted on a Nikon Eclipse TE2000-U epifluorescence microscope (Tokyo, Japan) coupled to a CoolSNAP HQ CCD camera (Roper Scientific, Tucson, AZ) or iXon EM ϩ 897E backilluminated EMCCD camera (Andor, Belfast, UK) with a Bioptechs stage heater. Nikon CFP HQ, YFP HQ, and YFP BLEACH filter cubes without emission filters were used for CFP imaging, YFP imaging, and YFP bleaching. A xenon lamp illuminated cells through a 33-mm neutral density 4 (ND4) filter, and they were viewed through a 60ϫ oil immersion Plan APO objective lens. Exposure time was 500 ms for both CFP and YFP. Acceptor photobleaching was done for 3.5 min without an ND filter using the YFP BLEACH cube. For each field pre-and post-bleach images of CFP and YFP were acquired. Image acquisition, registration, background subtraction, and data analysis were performed using NIS Elements (Nikon) and Excel (Microsoft, Seattle, WA) computational software.
Cells with similar CFP and YFP intensities were selected. After registration and background subtraction, the CFP signal in the cell membrane was masked as a region of interest for analysis. FRET efficiency (E) was calculated as E ϭ 1 Ϫ (F CFP(Pre) /F CFP(Post) ) (25), where F CFP(Pre) and F CFP(Post) are the mean CFP emission intensity before and after YFP photobleaching (24).
FACS-based FRET Assay for Extracellular Extension of CR3-Experiments were performed as described (23) and expanded in supplemental Experimental Procedures. Briefly, neutrophils were suspended at 1 ϫ 10 7 cells/ml in BSSϩ buffer (0.1% BSA, 1 mM CaCl 2 , 1 mM MgCl 2 ) with 10% donor serum. Cells were stimulated with 10 nM bacterial n-formyl peptide (fMLP), 25 g/ml FgnD, 50 g/ml s␤glu, or combinations for 15 min at 37°C, and 100-l cell aliquots were incubated with FITC-IgG control (10 g/ml), FITC-ICRF44 (9.8 g/ml), or FITC-CBRM1/5 (15 g/ml) for 60 min with rotation at 4°C in foil. Where indicated, cells were incubated with 200 g/ml FITC-Fgn for 15 min at room temperature. Cells were washed 3 times with BSSϩ at 4°C and split into four 125-l aliquots in FACScompatible snap cap tubes on ice. Octadecylrhodamine B (ORB) was added to a final volume of 250 l and concentrations of 0, 75, 200, and 400 nM. Samples were incubated on ice for 20 min and analyzed by the Brown University Flow Cytometry and Sorting Facility on a BD FACSAria.
SILAC-SILAC is detailed in the supplemental Experimental Procedures. Briefly, 1 ϫ 10 7 K562 or K562-CR3 cells were collected after at least seven doublings in SILAC media, washed twice, and resuspended in 1 ml of L-15 supplemented with 2 mg/ml D-glucose and 10% human serum. 100 g/ml s␤glu was added to the treatment tubes. Cells were incubated at 37°C for 30 min, placed on ice, and washed twice with ice-cold dPBS, and aspirated pellets were frozen at Ϫ80°C. Eight peptide samples were generated by combing cell types, SILAC label, and treatment conditions. Phosphopeptides were TiO 2 -enriched, analyzed using automated nano-LC/ESI-MS, and processed using the High-throughput Autonomous Proteomic Pipeline platform. A list of phosphopeptides that changed significantly in K562-CR3 s␤glu treatment versus K562-CR3 control but were unchanged in K562 s␤glu treatment versus K562 control was identified (supplemental Table S1) as well as a list that changed significantly in K562-CR3 s␤glu treatment versus K562 s␤glu treatment but was unchanged in K562-CR3 control versus K562 control (supplemental Table S2).
Statistics-For SILAC experiments, q-values were calculated using R (R Foundation for Statistical Computing, Vienna, Austria), and significance was set at a q Ͻ 0.1. All other data were analyzed by Student's t test or analysis of variance with post hoc Newman-Keuls as appropriate using Excel and MATLAB computational software. Values of p Ͻ 0.01 were considered statistically significant.

Neutrophil CR3 Colocalizes with Exposed ␤-Glucan on Live
Yeast Hyphae-Previous data from our laboratory (3,20) have shown using both functional and antibody blocking studies that CR3 recognizes the fungal PAMP ␤-glucan on intact hyphae. An experiment was designed to demonstrate physical colocalization of CR3 with exposed ␤-glucan of live yeast hyphae in which neutrophils were added to established yeast hyphae. Cells were then fixed and stained for CR3 (red) and ␤-glucan (green). In areas of neutrophil-hyphae interaction, fluorescent microscopy reveals specific colocalization of CR3 to areas of exposed ␤-glucan (yellow), shown in Fig. 1B. These data support our model that CR3 acts as a pattern recognition receptor in the interaction of neutrophils with intact hyphae. It is known that s␤glu is elaborated as a breakdown product from the fungal cell surface and may be present in the blood stream and tissue surrounding the primary site of infection (7). How s␤glu primes peripheral blood leukocytes and signaling in response to s␤glu is poorly characterized. To investigate this, a series of experiments was designed to determine if CR3 functions as a signaling receptor in response to s␤glu.
Generation of K562 Cell Line Stably Expressing CR3-FRET Activation Reporter-Current mechanistic thought on the unique bidirectional signal transduction of integrins speculates that conformational rearrangements correlate with activation, increased ligand binding, and intracellular signaling (25). Kim et al. (24) provided compelling evidence for this hypothesis using a series of FRET reporters for the integrin LFA-1, suggesting that this bidirectional signal transduction is achieved through coupling the spatial separation of the ␣ and ␤ cytoplasmic domains with extracellular conformational change (24). The functional consequences and whether this manner of regulation is shared among integrins is a subject of current investigation. To that end we developed a FRET sensor to report the separation of CR3 cytoplasmic tails, allowing the measurement of integrin activation and ligand binding directly in living cells ( Figs. 2A and supplemental Fig. S1). From this construct a stable line, K562:CR3-FRETa, was established (Fig. 2). Both the reporter construct and stable line have competent receptor function shown by both adhesion (supplemental Fig. S1D) and the ability to agglutinate with either iC3b-coated whole glucan particles or unicellular C. albicans (supplemental Fig. S2).  The merged image indicates colocalization of CR3 to areas with immuno-accessible ␤-glucan in yellow (bottom right). Shown is a differential interference contrast (DIC) image (bottom left). FITC, Texas Red, and differential interference contrast images were acquired with a Nikon Eclipse TE2000-U epifluorescence microscope coupled to a CoolSNAP HQ CCD camera using NIS elements software. Nikon B-2E/C and Y-2E/C cubes were used for FITC and Texas Red imaging, respectively. Cells were visualized at 37°C in L-15 medium with 2 mg/ml glucose. A xenon lamp was used to illuminate the cells that were viewed through a 40ϫ Plan APO objective lens. FIGURE 2. The use of FRET to report integrin activation; development of a K562 cell line that stably expresses the CR3-FRET activation reporter; loss of FRET in activation reporter with native ligand. A, the integrin activation FRET reporter is shown. In the inactive, bent conformation (left), the cytoplasmic domains of the ␣ and ␤ subunits are closely apposed, allowing for efficient FRET between mCFP and mYFP fused to these C termini. In the active, extended conformation (right), the ␣ and ␤ subunits lose their interactions in the tailpiece leading to a separation greater than 100 Å between the mCFP and mYFP, thereby abolishing FRET. See also supplemental Figs. S1 and S2 for reporter construction and functional characterization. B, shown are bright field (BF), CFP, and YFP images of the K562 cell line stably expressing the CR3-FRET activation construct (K562:CR3-FRETa) acquired with a Nikon Eclipse TE2000-U epifluorescence microscope coupled to a CoolSNAP HQ CCD camera using NIS elements software. Nikon CFP HQ and YFP HQ cubes without emission filters were used for CFP and YFP imaging, respectively. Cells were visualized at 37°C in L-15 medium with 2 mg/ml glucose. A xenon lamp was used to illuminate the cells through a 33-mm ND4 filter, and the cells were viewed through a 60ϫ oil immersion Plan APO objective lens. Exposure time was 500 ms for both CFP and YFP. C, FRET efficiencies: basal FRET, non-activating treatment with 1 mM Mn 2ϩ FRET, activation with 1 mM Mn 2ϩ plus 25 g/ml FgnD or 20 nM phorbol 12-myristate 13-acetate (PMA). Each sample represents analysis of at least 25 cells in three separate experiments. *, p Ͻ 0.01.

␤-Glucan Increases Homotypic CR3 Interactions and Specifically Induces Separation of CR3 Cytoplasmic Tails-K562:
CR3-FRETa cells were treated with s␤glu plus or minus Mn 2ϩ in the presence or absence of FgnD. These cells formed large clumps in the presence of s␤glu and Mn 2ϩ , likely due to an increase in homotypic interactions, rendering FRET analysis difficult (Fig. 3A). K562:LFA1-FRETa cells, which express a FRET construct reporting the cytoplasmic tail separation of the ␤2 integrin LFA1, which lacks the lectin-like domain of CR3, were similarly treated but did not show homotypic aggregation (data not shown). As Mn 2ϩ facilitates activation but is not a requirement for integrin function, we performed our analysis in its absence. As shown in Fig. 3B, FRET is significantly reduced in the CR3-FRETa cells upon treatment with s␤glu. LFA1-FRETa cells lacking the lectin-like domain showed no separation of cytoplasmic tails in response to s␤glu treatment. Activation with FgnD or treatment with s␤glu and FgnD also significantly reduced FRET in the CR3-FRETa cells, but no additive effect was observed. This separation of cytoplasmic tails suggests a mechanism by which s␤glu may exert some of its immune priming effect. Experiments were conducted in the presence of human serum, but the observed cytoplasmic activation of CR3 was serum-independent for the treatments described here. Studies in the absence of serum or in the presence of heat-inactivated human serum showed no differences from the data reported in Fig. 3B (data not shown).
s␤glu Treatment of Human Neutrophils Reveals Activationspecific Epitope of CR3-Previously published functional experiments suggest that s␤glu does not fully activate cells. Although the CR3 FRET experiments shown above suggest that CR3 is activated by treatment with s␤glu, those experiments only reveal the intracellular conformation of the receptor. To investigate the extracellular conformation of CR3 after treatment with s␤glu, FACS experiments were performed on primary human neutrophils.
Cells were treated with either fMLP, FgnD, s␤glu, or a combination of FgnD and s␤glu in the presence of donor serum. Treated and control cells were labeled with FITC-ICRF44 (which binds to both inactive and extended conformations of CR3) (28), FITC-CBRM1/5 (which binds to an epitope revealed upon activation and extension), or FITC-Fgn (fibrinogen, a CR3 I-domain ligand, labeled with FITC) (Fig. 4A). fMLP treatment is known to enhance surface expression of CR3 as well as activating CR3, revealing the CBRM1/5-specific epitope (29). As shown in Fig. 4, B-D, with s␤glu treatment we also see significant enhancement of the surface expression of CR3 (126 Ϯ 18%), the exposure of the activation-specific epitope (138 Ϯ 18%), and binding of FITC-labeled Fgn (126 Ϯ 9%) but to a much lower extent to that seen with fMLP stimulation, which showed a significant increase of 206 Ϯ 37, 250 Ϯ 40, and 226 Ϯ 29%, respectively. Neutrophils treated with unlabeled FgnD had significantly lower levels of ICRF44, both alone (67 Ϯ 1%) or with s␤glu (72 Ϯ 3%), indicating CR3 internalization after binding. To specifically isolate the conformational dynamics of ligand bound CR3 on the cell surface, subsequent experiments used FITC-labeled fibrinogen (FITC-Fgn) as both a ligand and FRET reporter.
s␤glu Causes Small Magnitude of Extension of CR3 Extracellular Domain-Because it is unclear at what point in the extension of activated CR3 the CBRM1/5 epitope becomes accessible, this leaves at least two possible ways of interpreting these results. The data could indicate that s␤glu treatment causes a Bright field images were acquired for each condition with a Nikon Eclipse TE2000-U epifluorescence microscope coupled to a CoolSNAP HQ CCD camera using NIS Elements software. Cells were visualized at 37°C in L-15 medium with 2 mg/ml glucose. A xenon lamp was used to illuminate the cells through a 33-mm ND4 filter, and the cells were viewed through a 20ϫ Plan APO objective lens. See also supplemental Figs. S1 and S2 for reporter construction and functional characterization. B, shown is the graphic comparison of FRET efficiencies of K562:CR3-FRETa (gray bars) and K562:LFA1-FRETa (white bars) for each treatment, performed in the presence of human serum and in the absence of Mn 2ϩ . *, p Ͻ 0.01 versus untreated. This FACS-based assay reports the relative distance between a FITC-labeled mAb or a FITC-labeled protein and ORB, a membrane partitioning fluorophore. Neutrophils labeled with ORB and FITC-conjugated mAb or protein are assayed using flow cytometry. Transfer of energy from the FITC donor to the ORB acceptor causes a decrease in the fluorescence intensity measured in the FITC channel that we plot as the ratio of FITC mean fluorescence intensity in the absence of ORB acceptor (F D ) to that in the presence of ORB acceptor (F DA ) versus ORB mean fluorescence (Fig. 5A). The slope of these lines quantita-tively represents the extent of energy transfer from the donor to the acceptor. Conditions are compared by calculating a distance ratio (L 2 /L 1 ϭ (S 1 /S 2 ) 1/4 ) that represents the change in donor-acceptor separation, where L represents the distance of the donor fluorophore from the acceptor fluorophore in the cell membrane, and S is the slope as calculated above. Using FITC-ICRF44 and FITC-CBRM1/5, we see a significantly greater distance ratio for FITC-CBRM1/5 (1.46 Ϯ 0.11) than FITC-ICRF44 (1.25 Ϯ 0.04) with fMLP that is consistent with previous data (23) (Fig. 5E). See Fig. 5, B-D, for representative data plots used to determine slope to calculate distance ratios. FITC-IgG isotype control confirmed the specificity of signal (data not shown).

CR3 Is a Signaling Receptor for Soluble ␤-Glucan
As the active conformation of CR3 is thought to extend its headpiece a distance greater than 200 nm from the cell membrane where FRET would not be possible, the fact that greater distance ratios are obtained using FITC-CBRM1/5 versus FITC-ICRF44 suggests that upon activation there is a only a small fraction of total CR3 in the extended conformation and that a large, inactive fraction masks the decrease in FRET signal (Fig. 5B). With s␤glu treatment we saw much more modest, although significant distance ratios with 1.11 Ϯ 0.05 for FITC-ICRF44 in untreated neutrophils versus FITC-ICRF44 in s␤glutreated neutrophils and 1.24 Ϯ 0.08 versus FITC-CBRM1/5 in s␤glu-treated neutrophils (Fig. 5E). This suggests that s␤glu treatment results in less extension of the extracellular domain than seen for full activation with fMLP. We also used FITC-Fgn to investigate how dual ligation would influence distance ratios. FITC-Fgn-treated neutrophils alone versus FITC-Fgn-treated neutrophils in the presence of fMLP produced a distance ratio of 1.35 Ϯ 0.06, indicating significantly greater extension after fMLP stimulation. Interestingly, Plots represent a single blood donor, with duplicate samples for each condition. E, the slope of each line is combined with those of biological replicates to calculate distance ratios used to quantitatively compare relative distance between the FITC-labeled donor on the CR3 receptor and the ORB acceptor in the cell membrane between experimental conditions. *, p Ͻ 0.01 versus untreated or Fgn alone; **, p Ͻ 0.01 between treatment conditions. FITC-Fgn-treated neutrophils alone versus neutrophils treated with both FITC-Fgn and s␤glu had a distance ratio of 0.93 Ϯ 0.05, indicating a significant decrease in extension when CR3 is engaged by both Fgn and s␤glu (Fig. 5E). Regardless of treatment condition, ORB mean fluorescence was concentration-dependent and consistent within a given experiment, demonstrating that neutrophil activation does not alter incorporation of ORB into the membrane.
s␤glu Treatment Induces Phosphorylation of Signaling Molecules and Proteins Associated with Transcriptional Regulation, mRNA Processing and Alternative Splicing-In addition to examining the conformational changes in CR3 induced by s␤glu (Fig. 6), we pursued a functional proteomics approach to identify phosphopeptides that are present in differential abundance after s␤glu treatment. In these experiments K562 cell lines that were isogenic with the exception of CR3 expression were used. K563 and K562-CR3 cells with and without s␤glu treatment were combined into sample groups described below, tryptic peptides were generated and enriched for phosphopeptides using TiO2 column, and the identity and abundance of the resultant peptides was inter-rogated using an automated nano-LC/ESI-MS analysis. To report phosphopeptide ratios, all ratios were corrected for source protein variation, and ratios of three technical replicates from an individual biological replicate were weighted on intensity and averaged. Q-values were calculated, and significance was set at a q Ͻ 0.1. The final ratio of each peptide was calculated by averaging four biological replicates.
Changes in phosphorylation level can be affected by many influences. To identify those changes specifically related to CR3 and ␤-glucan treatment, we identified 317 peptides representing 111 proteins and 402 unique phosphorylation sites (356 Ser(P), 43 Thr(P), and 3 Try(P)) that changed significantly in K562-CR3-treated with s␤glu versus K562-CR3 control. From this set, a list of 53 unique phosphopeptides representing 39 proteins that changed significantly in K562-CR3 treated with s␤glu versus K562-CR3 control but remained unchanged in K562 treated with s␤glu versus K562 control was selected and is shown in supplemental Table S1. Functional annotation was manually curated from the Gene Ontology bioprocesses, Kyoto Encyclopedia of Genes and Genomes, and available characterization in the literature. To complement these data, we addi-

CR3 Is a Signaling Receptor for Soluble ␤-Glucan
tionally identified 346 peptides representing 110 proteins and 441 unique phosphorylation sites (380 Ser(P), 54 Thr(P), and 7 Tyr(P)) that changed significantly in K562-CR3 treated with s␤glu versus K562 treated with s␤glu. From this set a list of 131 unique phosphopeptides representing 94 proteins was generated that changed significantly in K562-CR3 treated with s␤glu versus K562 treated with s␤glu but remained unchanged in K562-CR3 control versus K562 control and is shown in supplemental Table S2. Many of the phosphorylation sites that show differential regulation are non-canonical, and therefore, antibody availability for purposes of confirmation was limiting. Western blot data generated using the commercially available phosphospecific antibody for the MAP kinase ERK1/2 (Thr-185/Tyr-187) shows a 50% reduction in phosphorylation in K562-CR3 cells upon treatment with soluble s␤glu but remains unchanged in K562 cells, supporting our SILAC data (supplemental Fig. S3). These phosphopeptide sets, schematized in Fig.  7, are differentially regulated in response to s␤glu in a CR3-dependent manner and identify CR3 as a signaling receptor for s␤glu.

DISCUSSION
When isolated in soluble form, ␤-glucan has been shown to be an effective immune potentiator and is the subject of several clinical trials for both anti-infectious and anti-cancer indications (8 -12). Therapeutically, s␤glu has shown an ideal clinical profile in supporting host defense without stimulating the production of cytokines, which when overproduced can result in significant morbidity and mortality (31)(32)(33). The ability to achieve cellular immune stimulation in the absence of cytokine production is a unique consequence of glucan binding to CR3, but the mechanism is poorly understood. Characterizing the physical changes in integrin conformation upon s␤glu binding and understanding the signal that is transduced in response to this binding provides insight into the mechanism of s␤glu as a therapeutic while also facilitating the design of targeted agonists.
FRET was used to investigate intracellular and extracellular conformational changes of the ␤ 2 integrin, CR3, in response to s␤glu and the native I-domain ligand, Fgn. This study repre- FIGURE 7. Set of phosphopeptides differentially regulated in response to s␤glu in a CR3-dependent manner. Schematic of the phosphopeptides found using SILAC in K562 and K562:CR3 cells to be differentially regulated in response to s␤glu in a CR3-dependent manner, organized by their likely cellular functions. See also supplemental Tables S1 and S2 and Fig. S3.
sents the first use of FRET to study PAMP recognition in live cells.
K562 cells expressing our intracellular FRET activation reporter, K562:CR3-FRETa, showed significant loss of FRET upon FgnD binding, supporting that, as with LFA-1, CR3 activation is correlated with a spatial separation of the cytoplasmic domains ( Figs. 2 and 3, and supplemental Fig. S1). Significantly reduced FRET was also shown upon treatment with s␤glu (Fig.  3). K562:LFA1-FRETa cells reported no loss of FRET in response to s␤glu, demonstrating this response is specific to CR3. Furthermore, although treatment with FgnD alone or in combination with s␤glu also significantly reduced FRET in the CR3-FRETa construct, no additive affect was observed. These studies were carried out without Mn 2ϩ to compensate for homotypic clumping, which likely explains the less robust loss of FRET observed in comparison to Fgn treatments in the presence of Mn 2ϩ . The results suggest a mechanism by which ␤-glucan may exhibit some of its priming effect. s␤glu is not known to directly activate neutrophils through CR3 but rather is thought to initiate a primed state that leads to a faster or an enhanced response when encountering a second stimulus. This FRET readout reports only the average separation distance between the cytoplasmic tails with "active" and "inactive," each likely covering a range of integrin conformations (Fig. 6). The CR3 activation reporter cannot distinguish an integrin in its inactive bent conformation from the intermediate activation state believed to be stabilized by Mn 2ϩ . Additionally, the FRET efficiency of each cell is actually an average of all labeled CR3 molecules on the membrane. A lower FRET efficiency means that more receptors have a separation of cytoplasmic tails but does not reflect the activation state of every single receptor. Intracellular FRET reporter data, although giving valuable insight, must be coupled with other means of investigation to give a clear picture of integrin dynamics.
To further investigate the conformational state induced by s␤glu, the FRET reporting system developed by Lefort et al. (23) was used to examine the extension of the extracellular domain of CR3 in response to s␤glu. Treatment with s␤glu caused a much more modest extracellular extension than would be expected, given the degree of cytoplasmic tail separation reported. Additionally, dual ligation with Fgn and s␤glu showed a smaller distance ratio than that of Fgn treatment alone, suggesting that dual ligation signals may be transduced via a distinct conformational stabilization to that of Fgn or s␤glu alone (Figs. 5 and 6). The extracellular extension data for dual ligation is restricted to reporting the conformation of surface receptors. As shown in Fig. 4, CR3 ligation by Fgn with and without s␤glu leads to receptor internalization (Fig. 4). To isolate the conformation of dual ligated receptor on the surface, our ORB studies used FITC-Fgn as both a ligand and FRET reporter. We speculate that dual ligation stabilizes a subset of CR3 molecules on the cell surface, whereas receptors that have transitioned to the fully extended state are rapidly internalized and cleared. It is also plausible that dual ligation may slow conversion kinetics to the fully activated state.
Ligand binding is not a constitutive property of integrins. Three distinct integrin conformations have been characterized by electron microscopy (34). In the low affinity conformation, the legs and cytoplasmic domains are in close apposition to each other and bent at the knee in an inverted V shape, allowing the head to be tucked in over the legs (Fig. 6) (25). The high affinity binding state requires a switchblade-like extension in which the integrin transitions to a fully extended conformation with an open headpiece and separation of the ␣ and ␤ subunits at their cytoplasmic and transmembrane leg domains (Fig. 6) (24,34). This unbending event is linked to the outward swing of the hybrid domain and a pulldown of the C-terminal helix of the ␤ I-like domain, which alters the geometry of the metal ion-dependent adhesion site to induce the active high affinity conformation. I-domain-containing integrins additionally require the metal ion-dependent adhesion site loops and Mg 2ϩ coordination of the activated ␤ I-like domain to interact with the C-terminal ␣ I-domain ␣-helix and pull downward. This action converts the ␣ I-domain through a downward displacement of one or two turns of the helix into intermediate and high affinity forms. These global conformational changes are coupled to specific inter-and intradomain rearrangements that stabilize the high affinity states of the ␤ I-like and ␣ I domains in the head region to permit binding of extrinsic ligands (35). In addition to these stabilized conformational states it is thought that a range of intermediates in integrin conformation are possibly based on the inherent flexibility of hinges and joints in the protein structure (36) (Fig. 6). Because of this structural flexibility, integrins in the resting conformation are thought to exhibit "breathing," which is a small degree of unbending.
In addition to the canonical "outside-in" signaling whereby ligand binding activates an intracellular signaling pathway, integrins can also transduce "inside-out" signals. In this case, signals received by cell surface receptors (e.g. chemokine receptors) initiate intracellular signals that impinge on the cytoplasmic domains, causing a dramatic increase of ligand affinity in the extracellular domain and possibly passing through intermediate affinity conformations (34) (Fig. 6).
The previously described integrin breathing coupled with a relaxation of leg restraints, which may result from the binding of adaptor molecules, such as talin, to the cytoplasmic tail regions in response to intracellular signaling, may allow further unbending and small outward movement of the hybrid domain sufficient to expose the epitopes of activation-specific or stimulatory antibodies (37). Fig. 6 schematizes some of these defined and hypothetical intermediate integrin conformations paired with their predicted state FRET-based readout. Our observed separation of cytoplasmic tails combined with only a modest increase in distance ratio after s␤glu treatment lends support to the existence of this hypothesized intermediate conformation. Perhaps binding by s␤glu occurs in small amounts, facilitated by the breathing of the integrin, which then transduces that signal via an inside-out mechanism to allow separation of cytoplasmic tails and a relaxation of the leg regions. This hypothesized conformation may structurally define the "primed state" induced by s␤glu.
In addition to changes in conformation induced by s␤glu, we were also interested in identifying CR3-dependent signaling responses to s␤glu. The ␤-glucan-dependent signaling events are not well characterized, and study results are further complicated, as different sources of ␤-glucan may mediate their effects through different signaling pathways (38 -41). Proteintyrosine kinase inhibitors genistein or herbimycin A were shown to significantly block CR3-dependent cytotoxicity induced by s␤glu (13). Li et al. (11) demonstrated that the tyrosine kinase Syk was phosphorylated by mimicking dual ligation with macrophage-processed ␤-glucan and anti-I domain antibodies and additionally that phospho-Syk could be immunoprecipitated with CR3. Dual ligation also activated PI3K, which was shown using specific inhibitors to be downstream of Syk phosphorylation (11). In our studies, using K562 cell lines that were isogenic with the exception of CR3 expression for SILACcoupled MS experiments allowed the generation of a s␤glu phosphopeptide map that is CR3-dependent. This technique is very robust and avoids some of the run variation and derivitization artifacts seen in other proteomic techniques. Phosphopeptides whose ratios varied in a CR3-dependent and s␤glumediated way include previously reported integrin-associated proteins, such as Vimentin, phosphatidylinositol 3-dependent protein kinase-1, STAT5B, Rho GTPase-activating protein 12, keritins 8 and 18, N-myc downstream regulated gene family member 3, scaffold attachment factor B-like transcription modulator, myosin heavy chain 10, sialophorin, and components of EGF signaling (42)(43)(44)(45)(46)(47)(48). Also represented were peptides known to be involved in the cellular response to other PAMPs, including ERK2 and TRAF (TNF receptor-associated factor)type zinc finger domain-containing protein 1, the former of which was confirmed by Western blot analysis (26, 30) (supplemental Fig. S3). Additionally, the overall phosphopeptide profile was heavily weighted toward signaling molecules and proteins involved in transcriptional regulation, mRNA processing, and alternative splicing. This suggests that s␤glu may exert its priming effects by altering the transcriptional state of the cell.
This global phosphoproteome is the first direct demonstration of CR3-dependent signal transduction in response to unprocessed s␤glu in human cells and demonstrates that CR3 is a functional pattern recognition receptor for fungal ␤-glucan. The phosphopeptide signature and conformational intermediate described here may define the s␤glu-primed state in human neutrophils. This work underscores the potential therapeutic value of CR3 as a target for the development of carbohydratebased immune modulators.