Novel collectin/C1q receptor mediates mast cell activation and innate immunity

The α₂β₁ integrin serves as a receptor for a number of matrix and nonmatrix ligands, including collagens, laminins, decorin, E-cadherin, matrix metalloproteinase-1 (MMP-1), endorepellin, and several viruses.^1,2  Multiple lines of evidence have established that the inserted, or I, domain of the α₂ integrin subunit mediates binding of the α₂β₁ integrin to its ligands. Indeed, recombinant α₂ integrin I domain specifically binds all known ligands of the α₂β₁ integrin.

Previous studies using inhibitory monoclonal antibodies directed against the α₂β₁ integrin suggested a number of roles for this integrin in different models of inflammation.^3-5  We recently reported that α₂β₁ integrin–deficient mice exhibit markedly diminished inflammatory responses to Listeria monocytogenes, a Gram-positive bacterium, and zymosan, a fungal polysaccharide.⁶  We also reported that the α₂β₁ integrin is expressed at high levels on all peritoneal mast cells (PMCs).⁶  PMCs have been shown to be required for the induction of the inflammatory response to infection within the peritoneal cavity.^7,8  Our results suggested that in vivo PMC activation in response to certain microorganisms requires the α₂β₁ integrin.

At the time of our previous report, the ligand within the peritoneal cavity for the α₂β₁ integrin on PMCs was unknown. We hypothesized that the α₂β₁ integrin interacted with members of the microbial pattern recognition molecule family of the innate immune system. Collectins, including mannose-binding lectin (MBL), surfactant protein A (SP-A) and D (SP-D), ficolins, and the C1q complement protein all contain collagen-like sequences and are known to coat the surfaces of microbes.⁹  Opsonization and binding to the surface of microbes orient the collagen stalk of the collectin family for binding to receptors on the cell surface.¹⁰ 

Here we report that C1q, MBL, and SP-A all serve as ligands for the α₂β₁ integrin and bind specifically to the α₂ I domain. We demonstrate that these ligands mediate PMC adhesion, and we show a requirement for the α₂β₁ integrin during models of in vitro PMC activation. The α₂β₁ integrin-collectin interaction establishes a potential role for this integrin in a number of disease processes involving the innate immune system.

α₂ integrin subunit–deficient mice and wild-type littermate controls on a C57BL/6 × 129/Sv background were used at 6 to 20 weeks of age.¹¹  C1q-deficient mice, originally generated by Walport and Botto,¹²  were obtained on a pure C57BL/6 background from Drs Michael S. Diamond (Washington University School of Medicine, St Louis, MO.) and Gregory L. Stahl (Harvard Medical School, Boston, MA.).^13,14  Mice were maintained under pathogen-free conditions in either Washington University School of Medicine (St Louis, MO) or Vanderbilt University School of Medicine (Nashville, TN) mouse facilities. Within individual experiments, mice were appropriately age and sex matched.

Static adhesion assays were performed in 96-well plates (Immulon 2HB; Thermo Labsystems, Franklin, MA) as follows.^15,16  Wells were coated with bovine serum albumin (BSA) (5 μg/mL; Sigma-Aldrich, St Louis, MO), type 1 collagen (25 μg/mL rat tail; BD Biosciences, San Diego, CA), human C1q (25 μg/mL; purchased from Sigma-Aldrich, Calbiochem, San Diego, CA), or prepared as described in Kolb et al¹⁷  with minor modifications), human MBL (10 μg/mL; US Biological, Swampscott, MA), human SP-A (25 μg/mL; purified as described briefly here and in Wright et al¹⁸ ), a matrix of Listeria monocytogenes, anti-Listeria antibody, and serum, or a matrix of BSA, anti-BSA, and serum. The biologic activity of commercial C1q was similar to purified C1q in ligand-binding assays. The Listeria or BSA matrix was formed by allowing Listeria (strain EGD, 1 × 10⁸ organisms/mL in 0.1 M carbonate buffer, pH 8.5) or BSA (5 μg/mL in PBS) to adhere to wells of a 96-well plate overnight. Unattached Listeria or BSA was removed, and polyclonal rabbit anti-Listeria antibody (serotypes 1 and 4, used at 1:200 dilution in PBS; Difco, Detroit, MI) or rabbit polyclonal anti-BSA antibody (IgG fraction, used at 1:1000 dilution in PBS; Invitrogen Life Technologies, Carlsbad, CA) was added and incubated at 37°C for 1 hour before washing 3 times. Fresh mouse serum at the indicated dilution in PBS was then added to the wells and incubated at 37°C for 1 hour. Serum from wild-type or C1q^-/- mice, human serum (Sigma-Aldrich), or human serum depleted of the C2 complement component (Sigma-Aldrich) was used for this step. All wells were blocked with BSA (0.1% in PBS) for 1 hour before the addition of cells. PMCs isolated from resident peritoneal exudates using Percoll gradient centrifugation to approximately 85% purity, ⁶  NMuMg-1 cells, NMuMg-3 cells, K562 cells, or K562 transfectants expressing the human α₂β₁ integrin¹⁵  were allowed to adhere for 1 hour at 37°C in the presence of 2 mM MgCl₂ ± 40 nmol PDB or of 2 mM EDTA. Nonadherent cells were removed by washing, and the remaining adherent cells were quantitated, as previously described.¹⁶ 

SP-A was prepared by the butanol extraction method, as described previously.¹⁸  Briefly, 1 to 2 mL alveolar proteinosis was extracted with 25 mL 1-butanol and dried over nitrogen overnight. The dried protein was suspended in HEPES buffer with 0.15 M NaCl and 20 mM n-octyl-β-D-glucoside, then centrifuged at 17 000g; this process was repeated once. The final pellet was suspended in 5 mM HEPES buffer with 1 mM EDTA (pH 7.5), incubated with polymyxin B–agarose beads (Sigma-Aldrich), and dialyzed for 48 hours. The lipopolysaccharide content in the SP-A preparations was monitored after polymyxin treatment by the Limulus lysate assay (Associates of Cape Cod, Falmouth, MA) and contained less than 0.05 endotoxin U/mL.

For in vitro mast cell activation by Listeria, purified PMCs (5 × 10⁴ cells/well) were incubated with a washed suspension of Listeria, rabbit anti-Listeria antibody, and serum. The suspension was formed by incubation of Listeria (1 × 10⁷ organisms) with anti-Listeria antibody (1:200 dilution in 0.3 mL PBS) overnight at 4°C, followed by washing and incubation with 50% serum for 1 hour at 37°C. The PMC-Listeria suspension was centrifuged for 15 minutes and then incubated for 1 hour at 37°C. Supernatants were analyzed by ELISA for IL-6 (BD Biosciences).

Cloning and expression of the human α₂ integrin subunit I domain has been described elsewhere.¹⁹  Briefly, cDNA encoding the α₂ I domain was amplified by PCR using the full-length α₂ integrin cDNA (a gift of Dr Martin E. Hemler). The product of the PCR reaction encodes Ser124-Met349 of the published α₂ integrin sequence.²⁰  PCR primers were designed such that a BglII site was introduced at the 5′ end, and a stop codon followed by an XhoI site was introduced at the 3′ end. The product was subcloned into the glutathione S-transferase (GST) fusion vector pGEX-5X-1 (Amersham Biosciences, Piscataway, NJ), and the GST-I domain fusion protein was expressed, purified, and characterized as previously described.²¹  The QuikChange mutagenesis method (Stratagene, La Jolla, CA) was used to create α₂ integrin I domain cDNAs containing the point mutations D151A, D254A, and E318A. The sequence of each of the mutants was verified using the BigDye terminator cycle sequencing method (Applied Biosystems, Foster City, CA).

Wells of a 96-well microtiter plate (Immulon 2 HB) were coated overnight at 4°C with 0.1 mL of 30 μg/mL rat tail type 1 collagen (Upstate Cell Signaling Solutions, Lake Placid, NY), 20 μg/mL human C1q (Sigma-Aldrich), 20 μg/mL human MBL, or 10 μg/mL human SP-A. Wells were washed twice with TBS (50 mM Tris-HCl, 150 mM NaCl, pH 7.4) and then blocked for 1 hour at room temperature with 0.15 mL of 300 μg/mL BSA in TBS (for type 1 collagen, SP-A, MBL, or C1q (Figure S1; see the Supplemental Figure link at the top of the online article, at the Blood website), or heat-treated undiluted porcine serum (for C1q in Figure 3) (Invitrogen Life Technologies). Blocking of C1q with BSA alone led to an increased background, as evidenced by increased binding of GST-alone (Figure S1), likely because the secondary antibodies used for detection in the ELISA bound directly to C1q. Thus, heat-treated porcine serum was used as a blocking agent for C1q. The porcine serum was heat treated to degrade the complement proteins. Heat-treated porcine serum did not influence α2 integrin I domain binding to BSA or type 1 collagen (data not shown). Binding of GST alone to the different ligands served as the negative control in all experiments.

Purified recombinant α₂ integrin subunit I domain proteins were diluted to the desired concentration in wash buffer (TBS containing 0.05% Tween-20, 30 μg/mL BSA, and 2 mM MgCl₂, 2 mM MnCl₂, 2 mM MgCl₂₊ 2 mM CaCl_2, 2 mM MnCl₂₊ 2 mM CaCl_2, or 1 mM EDTA). The wells were washed once with 0.15 mL appropriate wash buffer, and 0.1 mL recombinant I domain was added and allowed to interact with the type 1 collagen, C1q, SP-A, or MBL for 1.5 hours at room temperature. Wells were washed 3 times with 0.15 mL wash buffer, and 0.1 mL of a 1:8000 dilution of anti-GST antibody (Amersham Biosciences) in wash buffer was added for 1 hour at room temperature. Wells were washed 3 times, and 0.1 mL of a 1:20 000 dilution of pig anti–goat IgG horseradish peroxidase (Roche Applied Science, Indianapolis, IN) in wash buffer was added per well and incubated for 1 hour at room temperature. Wells were washed 3 times, and 0.1 mL tetramethylbenzidine dihydrochloride (Sigma-Aldrich), prepared according to the manufacturer's instructions, was added per well. After 10 to 15 minutes of substrate conversion, reactions were stopped with 0.025 mL 4 N H₂SO₄, and optical density was read at 450 nm using an Emax microplate reader (Molecular Devices, Sunnyvale, CA). All data are presented as means of triplicate determinations.

To understand the role of the α₂β₁ integrin in PMC activation in vivo, we hypothesized that the α₂β₁ integrin mediated an adhesive interaction between PMCs and Listeria organisms or Listeria-containing immune complexes. Indeed, purified PMCs from wild-type mice adhered only to the substrate formed from serum-treated Listeria-containing complexes (Figure 1A). Adhesion did not occur in the absence of Listeria, anti-Listeria antibody, or serum. Adhesion of wild-type and α₂-null PMCs to plate-bound immune complexes formed between BSA and anti-BSA antibody was also performed. Adhesion was α₂β₁ integrin–dependent and required serum. Wild-type PMCs adhered to type 1 collagen and fibronectin. Adhesion to the multicomponent substrate or to type-1 collagen was divalent cation–dependent and was inhibited by EDTA, as expected for α₂β₁ integrin–dependent adhesion (data not shown). PMCs from α₂β₁-null mice failed to adhere to either the Listeria-containing complexes or type 1 collagen, but they did adhere to fibronectin and to PMCs from wild-type controls. Wild-type PMCs adhered maximally with 100% serum (Figure 1B and data not shown). This serum component was heat labile at 56°C, suggesting a role for complement in the process (Figure 1C).

We hypothesized that the α₂β₁ integrin may interact with members of the collectin family of proteins or with C1q. To address the role of C1q or complement components downstream of C1q in forming an adhesive substrate, we evaluated the adhesion of wild-type PMCs to an immune complex formed from Listeria, anti-Listeria antibody, and serum from wild-type or C1q-deficient mice, human serum, or human serum depleted of C2 (Figure 1D). Wild-type mice serum, human serum, and human serum depleted of C2 all resulted in the formation of an adhesive matrix, suggesting that components of complement downstream of the C1q complex are not required for this interaction. Importantly, mouse serum lacking C1q failed to form an adhesive matrix for α₂β₁ integrin adhesion (Figure 1D).

To determine whether C1q alone served as a ligand for the α₂β₁ integrin, we compared the ability of C1q to support the adhesion of wild-type or α₂β₁-deficient PMCs. Wild-type PMCs adhered to purified C1q, type 1 collagen, and fibronectin but not to BSA, in a divalent-cation dependent manner (Figure 2A and data not shown). Adhesion was inhibited by EDTA (data not shown). α₂β₁–deficient PMCs failed to adhere to either C1q or type 1 collagen but did adhere to fibronectin. Integrin activation by phorbol dibutyrate (PDB) slightly enhanced α₂β₁ integrin–mediated adhesion of wild-type PMCs to C1q but had no effect on α₂β₁-null PMCs. These data strongly suggest that the α₂β₁ integrin expressed on PMCs mediates adhesion to C1q.

To determine whether the α₂β₁ integrin–C1q interaction was cell-type or species restricted, we used a series of well-characterized α₂β₁ integrin–expressing and –nonexpressing cell lines.¹  We compared adhesion of the NMuMg-1 cell line, a murine mammary epithelial cell line that expresses high levels of endogenous murine α₂β₁ integrin, with that of the NMuMg-3 cell line, a spontaneous subclone with undetectable levels of this integrin. NMuMg-1 cells adhered to C1q and type 1 collagen in a divalent cation–dependent manner, whereas NMuMg-3 cells failed to adhere to either C1q or collagen (Figure 2B). The K562 cell line, a human hematopoietic line that lacks expression of the α₂β₁ integrin, failed to adhere to C1q or collagen, whereas K562 transfectants expressing high levels of the human α₂β₁ integrin (K562-X2C2) adhered to C1q and collagen (Figure 2B). All cell lines adhered to fibronectin.

The α₂ subunit of the α₂β₁ integrin contains an inserted (I) domain shown to be a critical determinant for ligand recognition and binding. We hypothesized that C1q would bind to isolated α₂ integrin I domain. These experiments were conducted using a previously developed solid-phase binding assay based on an ELISA format (see “Materials and methods”). Blocking with BSA alone led to an increased background, as evidenced by increased GST binding (Figure S1), likely because the secondary antibodies used for detection in the ELISA bound directly to C1q. Thus, in Figure 3, heat-treated porcine serum was used as a blocking agent for C1q. As shown previously, the α₂ integrin I domain bound to type 1 collagen in a concentration-dependent manner (Figure 3A). The α₂ integrin I domain also bound to C1q (Figure 3B; Figure S1) in a concentration-dependent and a cation-dependent manner. Binding of the I domain to C1q was readily apparent when either serum or BSA was used to block the substrate.

Ligands of the α₂ integrin differ in their dependence on divalent cations for binding.²²  Divalent cations bind to the α₂ integrin I domain at the metal ion–dependent adhesion site (MIDAS) motif.²³  We examined the binding of recombinant α₂ integrin I domain MIDAS motif (D151A and D254A) mutants to C1q. We have previously shown that these mutations abrogate binding to collagen.²⁴  Although these mutations did not abrogate binding of recombinant α₂ integrin I domain to C1q, the binding was significantly reduced (Figure 3C). These experiments establish that the binding of C1q to the α₂β₁ integrin is mediated by the I domain in a divalent cation- and a MIDAS motif-dependent process.

Crystal structures suggest that the α₂ integrin I domain can adopt 2 conformations: a closed, lower affinity conformation and an open, higher affinity conformation.^25,26  As shown in Figure 4, PDB treatment slightly enhanced α₂β₁ integrin–dependent adhesion to C1q, suggesting that the activated conformation of α₂β₁ integrin would bind with greater apparent affinity to C1q. Within the α₂ integrin I domain, amino acid E318 forms a salt bridge with amino acid R288 that is important for maintaining the α₂ integrin I domain in the closed conformation. Disruption of this salt bridge by mutation of E318 to alanine promotes the transition to the open, higher affinity conformation and enhances α₂ integrin I domain binding to low-affinity ligands such as collagen IV and laminin I (T.P.S. and S.A.S., unpublished data, July 2004). Therefore, we examined the binding of the E318A mutant to C1q. As anticipated from the enhanced cell attachment observed in the presence of PDB, the E318A mutant bound to C1q with higher apparent affinity than it did to the wild-type α₂ integrin I domain (Figure 3D).

These data define C1q as a novel ligand for the α₂β₁ integrin and demonstrate specificity for the interaction of C1q and the α₂β₁ integrin. C1q shares many structural features with members of the collectin family. We therefore evaluated the ability of other collectin family members to serve as ligands for the α₂β₁ integrin. PMCs adhered to both SP-A and MBL in an α₂β₁ integrin–dependent manner (Figure 4A-B). Adhesion to SP-A and MBL was enhanced by integrin activation through PDB based on the concentration of SP-A or MBL and on Mg²⁺ and inhibited by EDTA (data not shown).

In contrast to I domain binding to C1q, wild-type α₂ integrin I domain failed to bind to SP-A in the presence of Mg²⁺ to any greater extent than in the presence of EDTA (Figure 5A). However, the gain-of-function E318A α₂ integrin I domain activation mutant bound SP-A robustly in a divalent cation–dependent manner (Figure 5A), suggesting that integrin activation is required for SP-A binding. Mg²⁺ effectively supported the binding of the activated, but not the wild-type, α₂ integrin I domain to SP-A. Mn²⁺, a divalent cation known to stabilize the open, higher affinity conformation of the α₂ integrin I domain, effectively supported the binding of both the wild-type α₂ integrin I domain and the E318A activation α₂ integrin I domain mutant to SP-A (Figure 5B).

In contrast to dependence on the activation of the α₂ integrin I domain for binding to SP-A, wild-type α₂ integrin I domain bound robustly to MBL. Optimal binding of the α₂ integrin I domain to MBL required the presence of Mg²⁺ and Ca²⁺. Surprisingly, the gain-of-function E318A activation mutant bound to MBL to a lesser extent than did the wild-type α₂ integrin I domain (Figure 5C).

The findings presented in this manuscript identify C1q and the collectin family of proteins as novel adhesive ligands for the α₂β₁ integrin. Mast cells adhere to complement-containing immune complexes in a C1q- and an α₂β₁ integrin–dependent manner. In vivo, α₂β₁ integrin expression by PMCs is required for the early innate immune response that involves mast cell activation and IL-6 secretion.⁶  We hypothesized that ligation of the α₂β₁ integrin by members of the collectin family and C1q contribute to mast cell activation and cytokine secretion. To test this hypothesis, we measured the secretion of IL-6 by PMCs after 1-hour stimulation with serum-treated Listeria-containing immune complexes (Figure 6A). Wild-type PMCs secreted a maximal amount of IL-6 in the presence of these complexes, whereas α₂-null PMCs failed to secrete detectable levels of IL-6. Although type 1 collagen serves as an adhesive ligand for the α₂β₁ integrin on PMCs, type 1 collagen could not substitute as a ligand to stimulate mast cell activation. Similar to the results with purified type 1 collagen, purified C1q did not stimulate IL-6 secretion (data not shown). Unfractionated peritoneal exudate cells (consisting predominantly of macrophages, lymphocytes, and only approximately 2% PMCs) from wild-type and α₂-null mice secreted equivalent amounts of IL-6 in response to Listeria-containing immune complexes. These data support the hypothesis that α₂β₁ integrin immune complex interaction provides an essential signal for cytokine secretion by mast cells in response to certain stimuli. In contrast, the α₂β₁ integrin is not required for IL-6 secretion in response to immune complexes by other cell types. Resident peritoneal macrophages, the likely source of the IL-6 secretion by the unfractionated cells, lack surface expression of the α₂β₁ integrin, as determined by flow cytometry (data not shown).

To define the requirement for C1q, we evaluated the secretion of IL-6 by wild-type and α₂-null PMCs after 1-hour stimulation with Listeria-containing immune complexes using serum from either wild-type or C1q-deficient mice (Figure 6B). Although serum from wild-type mice provides a signal for mast cell activation, C1q-deficient serum failed to stimulate cytokine secretion.

We have identified the α₂β₁ integrin as a novel receptor for C1q and collectin family members. The ability of mature connective tissue mast cells expressing the α₂β₁ integrin to interact with immune complexes that directly stimulate mast cell activation and cytokine secretion suggests many previously unexpected roles for the α₂β₁ integrin in the innate immune response and in modulating autoimmune and allergic disorders. The requirement for both C1q and the α₂β₁ integrin for mast cell cytokine secretion in response to Listeria explains why α₂β₁ integrin–deficient mice demonstrate a diminished inflammatory response to Listeria infection and zymosan stimulation in vivo.⁶  At the time of our previous report, the ligand for the α₂β₁ integrin during the PMC response to infection and the molecular mechanisms by which the α₂β₁ integrin stimulated mast cell activation were unknown. We hypothesized that the α₂β₁ integrin interacted with certain microbial pattern recognition molecules of the innate immune system. Collectins, ficolins, and the C1q complement protein all contain collagen-like sequences and are known to coat the surfaces of microbes.⁹  We now demonstrate that C1q, SP-A, and MBL all serve as ligands for the α₂β₁ integrin. It is likely that the collagen-like sequences within each of these proteins are recognized by the α₂β₁ integrin; investigations to test this idea are under way.

Previous studies suggested the possibility that a platelet collagen receptor, such as the α₂β₁ integrin, may be involved in C1q binding.²⁷  Human platelets adhered to C1q in a Mg²⁺-dependent manner. Monomeric C1q inhibited platelet aggregation stimulated by immune complexes or collagen. However, Peerschke and Ghebrehiwet²⁷  described a distinct C1qR on platelets that cross-reacted with collagen in a cation-independent manner. The possibility that the α₂β₁ integrin served as a receptor to mediate platelet adhesion to C1q was addressed by antibody inhibition experiments using the α₂β₁ integrin–specific antibody 6F1. In these experiments, platelet adhesion to collagen was inhibited 69% to 71% by 6F1 when platelets were allowed to adhere to collagen at 5 μg/mL or 1 mg/mL. However, 6F1 inhibited platelet adhesion to low-density C1q by 31% and to high-density C1q by 17%. The authors concluded that C1q was not a ligand for the α₂β₁ integrin. An alternative conclusion suggests that there are several platelet receptors for C1q. At low densities of C1q, the high-affinity α₂β₁ integrin receptor serves an important role in adhesion. In contrast, at high densities of C1q, other lower affinity C1q receptors are sufficient to mediate binding.

Not only have we defined a family of novel ligands for the α₂β₁ integrin, we have demonstrated in vivo and in vitro α₂β₁ integrin–dependent mast cell activation and cytokine secretion.⁶  Ongoing studies have defined an important role for mast cells in a variety of autoimmune disorders, including rheumatoid arthritis and asthma. We present 2 alternative models by which α₂β₁ integrin–ligand interactions may stimulate mast cell activation and cytokine secretion (Figure 7). Ligation of the α₂β₁ integrin alone appears insufficient to activate cytokine secretion because mast cell adhesion to collagen or C1q alone by the α₂β₁ integrin fails to support cytokine secretion. These experiments suggest that one or more additional signals emanating from an additional receptor are required to activate mast cell cytokine secretion in response to immune complexes.

The first model suggests that the α₂β₁ integrin serves as a coreceptor that must be ligated simultaneously with a second costimulatory receptor to elicit mast cell activation. In this model, the α₂β₁ integrin would provide a necessary costimulatory function required for mast cell activation in a manner reminiscent of the role proposed for the α₂β₁ integrin during platelet adhesion to collagen. In vivo and in vitro studies of platelet adhesion to collagen are consistent with a 2-step, 2-site model of collagen-induced platelet activation that involves glycoprotein VI (GPVI) and the constitutively associated Fc receptor common γ chain (FcRγ) and the α₂β₁ integrin.^28,29  Once GPVI binds to collagen, the FcRγ chain is phosphorylated, and the α₂β₁ integrin and the GPVI receptor cooperatively mediate platelet activation.^28,30,31 

If mast cell activation through the α₂β₁ integrin is similar to the platelet-collagen interaction, then perhaps the FcRγ chain is also involved during interaction between the mast cell and the immune complexes. Other potential receptors on mast cells for which the α₂β₁ integrin may serve as a coreceptor include toll-like receptors.³²  In a similar manner, dectin 1, a C-type lectin receptor for β-glucan–containing particles, collaborates with the toll-like receptors TLR2 to TLR6 to augment the inflammatory response to complex particles, including the yeast cell wall.^33,34 

The second model to explain the role of the α₂β₁ integrin in mast cell activation is indirect activation of a separate pathway. For example, binding of the α₂β₁ integrin to an immune complex containing C1q may directly activate the complement cascade, resulting in the deposition of C3b or iC3b and the generation of complement byproducts such as C3a or C5a. Binding of complement components to their cognate receptors on mast cells (complement receptor 1 [CR1], CR3, CR4, C3aR, or C5aR) would then stimulate mast cell activation, as has previously been shown.^32,35,36  Experiments to determine which mechanism(s) are operational are ongoing.

The α₂ integrin I domain, composed of an inserted sequence of 191 amino acids, mediates the binding of physiologic ligands to the α₂β₁ integrin. Our data indicate that the α₂ integrin I domain binds C1q, SP-A, and MBL in a divalent cation–dependent manner. To verify that the observed divalent cation dependence of α₂β₁ integrin–mediated cell adhesion to these ligands was a function of divalent cation binding to the α₂ integrin I domain, we tested the binding of 2 α₂ integrin I domain MIDAS motif mutants, D151A and D254A, to C1q. Both mutations markedly inhibited α₂ integrin I domain binding to C1q, confirming that the cation dependence of this interaction required I domain cation binding.

Stimulation of cells with PDB enhanced the attachment of PMCs to C1q, SP-A, and MBL. These findings suggested that integrin activation augmented binding to these ligands. This hypothesis was verified in experiments in which the E318A α₂ integrin I domain mutation, a change known to facilitate transition to the open, higher affinity conformation, enhanced binding to these ligands. Inflammatory signals, including cytokine stimulation, may serve to activate the α₂β₁ integrin on cells in vivo, permitting higher affinity binding to C1q and SP-A. Interestingly, the E318A α₂ integrin I domain mutant did not show the expected increase in affinity for MBL. However, activation of wild-type, full-length α₂β₁ integrin on the surfaces of cells by phorbol showed an increase in binding for all 3 ligands, including MBL. These results suggest that MBL, but not C1q or SP-A, directly interacts with the E318 residue in the α₂ integrin I domain, which is near the ligand-binding site. Studies are under way to evaluate this hypothesis. These results suggest that different ligands of the α₂β₁ integrin may use different side chains near the binding site to stabilize the receptor-ligand interaction. These differences may have important effects on the affinity of the receptor for different ligands and the ability of different ligands to signal through the receptor.

The collectin family of proteins recognizes pathogen-associated molecular patterns on microorganisms. Their role in the host innate immune system is to enhance and promote phagocytosis and inflammation at the site of microbial invasion. In addition, both C1q and SP-D are important in the phagocytosis of apoptotic cells.^37,38  Several different cell surface receptors for C1q and other collectin family members have been reported, including the C1q receptor for phagocytosis enhancement (C1qRp), CR1, calreticulin (CRT), and binding protein for the globular head of C1q (gC1qbp).^39-52  The precise role of each receptor remains an area of active investigation.

In conclusion, we now define the α₂β₁ integrin as a novel receptor for C1q and several collectin family members, and we demonstrate not only the structural requirements for C1q and collectin binding but also the biologic relevance of the α₂β₁ integrin/C1q interaction in mast cell activation. The α₂β₁ integrin is a well-characterized receptor with a large extracellular domain that binds to a variety of ligands and a transmembrane domain and a cytoplasmic domain that activate a number of signaling pathways. The α₂β₁ integrin was originally described as a type 1 collagen receptor on platelets, and C1q and collectins have a collagen like sequence with a Gly-X-Y repeat that forms a triple helix, providing an obvious motif for binding to the α₂β₁ integrin. Furthermore, we show that mast cell binding to immune complexes is dependent on C1q and the α₂β₁ integrin and that this interaction results in a biologically relevant response, cytokine secretion.

We thank Laura Wells for expert animal care, Drs Emil R. Unanue and Hector Molina for advice in many helpful discussions, Drs Michael Diamond, Gregory L. Stahl, and Marina Botto for providing the C1q^-/- mice, and Jean McClure for secretarial assistance.