The binding of soluble proteins to target surfaces is vital in triggering the immune response. However, structural insight into such processes is still lacking. Mannan-binding lectin (MBL) is a classic example of a pattern recognition molecule with important roles in innate immunity against microbial infections. By small angle x-ray scattering analysis we show that the large MBL complex in solution is folded into a ramified structure with a striking rotational symmetry and a structure permissive of elongation by unbending. Nevertheless, the structure in solution is found to be very stable. However, when the MBL molecule interacts with surface-immobilized ligands, the stable MBL structure is broken into a stretched state with separation of the ligand-binding domains as shown by high resolution atomic force microscopy. These studies provide a snapshot of the single molecule mechanics of MBL and the first direct evidence that the transition from the soluble state to surface-bound protein involves large conformational changes in the quaternary structure, thus highlighting the role of surface topography in immune recognition.

Mechanisms of immune protection against infection and disease are divided into the two major categories of adaptive and innate immunity. Adaptive immunity involves the somatic selection of effector cells that specifically recognize Ags, which differ from “self” molecules of the body. By contrast, the innate immune system depends on molecular mechanisms evolutionary selected to recognize conserved motifs or “patterns” on microbial surfaces (1). Mannan-binding lectin (MBL),4 also referred to as mannan-binding protein or mannose-binding protein, plays a significant role in the body’s immune response to infection as underscored by clinical studies as well as experimental evidence from animal models. Binding by MBL to microbial surfaces activates the complement system, which leads to the deposition of a cascade of plasma proteins that facilitates uptake by leukocytes or lysis of the microbe. Recent studies also suggest a role for MBL in the clearance of apoptotic cells and the elimination of tumor cells (2, 3).

The structure of MBL is complex; the primary structure of the MBL polypeptide contains an N-terminal segment of 21 residues with three cysteines followed by seven collagen-like Gly-X-Y repeats and, interrupted by Gly-Gln, a second collagen-like region with 12 Gly-X-Y repeats. This segment is joined to a short 34-residue neck region and the carbohydrate recognition domain (CRD) of 93 residues at the C terminus, bringing the molecular mass (Mr) of the monomer to 25,340 Da (4). In the folded molecule the collagen regions form a stalk of three monomers that are further organized in higher oligomers through a complex pattern of interchain disulfide bridge formation (5), with the size of the complexes ranging from three to six stalks (5, 6, 7). Structural characterization of MBL has been conducted with electron microscopy or with x-ray crystallography on fragments of the ligand recognition domains. However, the three-dimensional structure of the MBL complex is currently not known.

Small angle x-ray scattering (SAXS) (8, 9) is a technique used for gaining insights into macromolecular structure in solution. By contrast to crystallography and electron microscopy, SAXS may bring the experimental setup during image collection close to physiological conditions with regard to such important parameters as ionic strength and other additives that influence protein solubility. In addition, recent progress in the analysis of SAXS spectra allows for the reconstruction of macromolecular structures with a spatial resolution of 10 Å (10). Advances in making recombinant human MBL similar to natural MBL (11) has provided a novel source of material, bringing a characterization of this molecule into the realm of SAXS analysis.

The binding to target surfaces of MBL triggers an immune response. However, structural insight into the processes that accompanies immune protein binding to surfaces is still lacking. Atomic force microscopy (AFM) is the technique of choice for investigating conformational changes at the single molecule level and has provided a range of new opportunities for manipulating and visualizing large biomolecules in environments compatible with preserving their native structure. The diversity of chemistries that can be used in these experiments is permitting AFM imaging of molecules on ligand-coated surfaces, especially when these ligands are small chemical compounds such as monosaccharides.

In this study we report the three-dimensional structure of MBL as determined by SAXS. Surprisingly, we find that the highly ramified structure is stable in solution but with features that are permissive of conformational changes. Such changes were induced by applying MBL to ligand-coated surfaces as shown here by AFM images of single MBL molecules.

Recombinant human MBL was prepared and isolated from tissue culture medium with a purification method selective for MBLs with a natural size distribution as described elsewhere (11). For the AFM, size fractionation of the MBL for the recovery of complexes with nine monomers was conducted as described previously (7).

MBL in 140 mM NaCl and 10 mM Tris-HCl (pH 7.5) was concentrated by use of a centrifugal filter unit with a cutoff at 10,000 Da (Centricon; Millipore). The MBL concentration was determined by a time-resolved immunofluorometric assay routinely used for quantifying the absolute MBL concentration. Immediately following concentration, the MBL was used in SAXS experiments with a laboratory-based instrument at the University of Aarhus, Aarhus, Denmark (12) at concentrations of 4.25 mg/ml and 2.12 mg/ml. The data were converted to absolute scale by using the scattering from a pure water sample as the primary standard. The intensity was displayed as a function of the modulus of the scattering vector, q = (4 π/λ) sin(θ), where λ of 1.54 Å is the x-ray wavelength and 2θ is the angle between the incident and the scattered x-rays. The measured data were corrected for instrumental smearing effects using an indirect Fourier transformation (13, 14) and a resolution function estimated from the geometry of the SAXS instrument (15). The desmeared data were calculated in the same q points as the original data and, subsequently, noise of the same magnitude as that of the original data was added to the desmeared data to allow for the application of a conventional weighted least squares method in the further analysis. Several modeling approaches were used for the determination of protein structure in solution from the SAXS data. The program packages DAMMIN (16), GASBOR (17), SASREF (10), MASSHA (18), DAMAVER (19, 20), and CRYSOL (21) were applied to evaluate the structural content of the SAXS data. MOLMOL (22) was used for the display and analysis of structures.

Modified mica surfaces were prepared by the immersion of freshly cleaved mica surfaces in a solution of 0.05 mg/ml spermine (S4264; Sigma-Aldrich) for 10 min, followed by a rinse in deionized water and drying under a steam of N2, which creates stable and flat surfaces (23). Subsequently, 2 μl of a solution with 7 μg/ml unfractionated MBL in a buffer containing 150 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1.5 mM MgCl2, and 10 mM HEPES (pH 7.4) was deposited onto the modified mica. After incubation at room temperature for 5 min, 10 μl of the HEPES buffer was placed onto the mica for 10 min and the surfaces were briefly rinsed in deionized water by gentle aspiration and dried with N2 before AFM imaging. Surfaces modified with 2-amino-2-deoxy-d-mannose (M4670; Sigma-Aldrich) or 2-amino-2-deoxy-d-galactopyranose (G0500; Sigma-Aldrich) were made by immersing the mica surfaces for 10 min in a solution of either hexosamine dissolved at 0.5 M in water followed by the application of 7 μg/ml size-sorted MBL. The molecules were either rinsed and dried as for the analysis on spermine-coated surfaces or analyzed directly in liquid by tapping mode AFM.

All images were acquired using a commercial AFM with a NanoScope IIIa controller (Veeco). For the imaging of MBL in liquid, all topographies were recorded using the tapping mode AFM in liquid. The samples were mounted in a liquid cell with an O-ring seal and immediately filled with stock buffers as described above. Commercial silicon nitride cantilevers (V-shape, catalog no. OMCL-TR400PSAHW; Olympus) with a length of 120 μm, a normal spring constant of 0.1 N/m, and integrated sharpened tips were used. The cantilevers were oscillated at a resonance frequency of ∼8 kHz. The drive voltage to the cantilever was ∼400–800 mV. Initial engagement of the probe was performed by setting the scan size at 0 to minimize sample deformation or probe contamination. The topographic images were obtained by with scan rate of 1–2 Hz. All images were captured as 512 × 512-pixel photographs. Several images were obtained from separate locations across the mica surfaces to ensure reproducibility. For the imaging of dried samples, AFM imaging was performed in tapping mode by using a high aspect ratio, rectangular STING probe (DP14, MikroMasch); the probe had a resonance frequency at 180 kHz, a spring constant at 5 N/m, and tip radii of 5–10 nm. During imaging, a minimal loading force of 100 pN was applied at scan frequencies of 1–2 Hz using optimized feedback parameters. For each sample, surface images were obtained from three separate locations. The images were flattened using the NanoScope software (Veeco), excluding the particles from the flattened area followed by automated analysis using the SPIP (version 4.2) scanning probe image processor software (Image Metrology) to yield size histograms of the imaged particles.

To determine the three-dimensional structure of MBL in solution, we performed SAXS measurements on recombinant MBL (11). The recombinant MBL has previously been shown to form oligomeric structures very similar to the oligomers found in human plasma and provides a homogenous source of MBL compared with plasma-derived MBL that requires extensive fractionation to obtain purity. The SAXS intensity distribution as a function of the scattering vector modulus q (Fig. 1 A) is the average of the x-ray interference patterns of the randomly oriented individual MBL structures (8). A few integral parameters are easily derived from these data; the radius of gyration and maximum dimensions of the complexes were 11.0 ± 0.1 nm and ∼32 nm, respectively. The mass of the MBL complexes was determined by the forward scattering of the SAXS data to Mr = 266,000 ± 14,000 Da. Recently developed analysis methods were used for the three-dimensional structure reconstruction of small globular proteins from the SAXS intensity (17), whereas the determination of the solution structure of large, highly ramified proteins such as MBL is novel.

FIGURE 1.

Structure of MBL in solution determined by SAXS. A, Plot of the intensity I as a function of the modulus of the scattering vector q. Experimental data (indicated with open circles) were fitted with the dummy atom model (solid line) and the dummy residue model (hatched line). The atomic resolution model was tested with either the neck/CRD cluster in the conformation determined by x-ray crystallography (rigid body model 1; dotted line) or with an opening angle of 20° between the 3-fold rotational symmetry axis and the coiled neck regions (rigid body model 2; hatched and dotted line). For each model the goodness of fit is indicated by the corresponding χ value. B, Averaged ab initio dummy atom model generated with the DAMMIN (dummy atoms) algorithm (16 ). Starting from a spherical search space with a diameter equal to the maximum dimension of the MBL complexes (32 nm) and filled with closely packed spheres, the algorithm applied a simulated annealing optimization to find the set of spheres with the lowest discrepancy between the calculated and measured scattering data. As three-dimensional reconstructions from SAXS data cannot yield a unique solution (20 ), we obtained a better description of the particle shape by generating 10 independent models and applying a filtering method using the DAMAVER algorithm (1920 ). The models were aligned and compared with a repacked structure that shows the most probable structure. C, Dummy residue model of the MBL oligomers generated by the algorithm GASBOR (17 ) from the number of amino acid residues in the MBL monomer (228) and by imposing a C3 rotational symmetry constraint with 3 × 228 as the number of dummy residues. The shown structure was selected as the most representative from pairwise comparisons between 10 independent models. D, Atomic resolution model of MBL. By use of rigid body refinement in a simulated annealing optimization protocol, the program SASREF (10 ) searched the three-dimensional configurations of six connected domains of trimeric collagen segments (25 ) terminated by the neck/CRD cluster (24 ). For simplicity, each stalk region was modeled with six identical domains, not taking into account the interruption of the collagen region at Gly-43 or the 21 residues in the N-terminal noncollagenous region of MBL.

FIGURE 1.

Structure of MBL in solution determined by SAXS. A, Plot of the intensity I as a function of the modulus of the scattering vector q. Experimental data (indicated with open circles) were fitted with the dummy atom model (solid line) and the dummy residue model (hatched line). The atomic resolution model was tested with either the neck/CRD cluster in the conformation determined by x-ray crystallography (rigid body model 1; dotted line) or with an opening angle of 20° between the 3-fold rotational symmetry axis and the coiled neck regions (rigid body model 2; hatched and dotted line). For each model the goodness of fit is indicated by the corresponding χ value. B, Averaged ab initio dummy atom model generated with the DAMMIN (dummy atoms) algorithm (16 ). Starting from a spherical search space with a diameter equal to the maximum dimension of the MBL complexes (32 nm) and filled with closely packed spheres, the algorithm applied a simulated annealing optimization to find the set of spheres with the lowest discrepancy between the calculated and measured scattering data. As three-dimensional reconstructions from SAXS data cannot yield a unique solution (20 ), we obtained a better description of the particle shape by generating 10 independent models and applying a filtering method using the DAMAVER algorithm (1920 ). The models were aligned and compared with a repacked structure that shows the most probable structure. C, Dummy residue model of the MBL oligomers generated by the algorithm GASBOR (17 ) from the number of amino acid residues in the MBL monomer (228) and by imposing a C3 rotational symmetry constraint with 3 × 228 as the number of dummy residues. The shown structure was selected as the most representative from pairwise comparisons between 10 independent models. D, Atomic resolution model of MBL. By use of rigid body refinement in a simulated annealing optimization protocol, the program SASREF (10 ) searched the three-dimensional configurations of six connected domains of trimeric collagen segments (25 ) terminated by the neck/CRD cluster (24 ). For simplicity, each stalk region was modeled with six identical domains, not taking into account the interruption of the collagen region at Gly-43 or the 21 residues in the N-terminal noncollagenous region of MBL.

Close modal

Reconstruction of the MBL structure with “dummy atoms” through the simplest ab initio method showed that, in solution, the large MBL complexes have a striking 3-fold rotational symmetry axis with the stalks joined in a common hub (Fig. 1 B). The SAXS-determined Mr, taken together with the presence of three stalks with collagen regions that form trimers, indicates that the determined structure contains nine MBL monomers. Our data thus suggest that oligomers of this size are the dominating MBL species on a molar scale.

With the determination of the stoichiometry of the MBL quaternary structure and its symmetry, a more accurate structural modeling could be performed. In this approach the protein backbone is represented by a chain of connected spherical “dummy residues.” Using dummy residues, a search for the three-dimensional arrangement of the chain that gives the best fit to the experimental data was performed (17) (Fig. 1,C). As the resulting model gave an excellent fit with the experimental scattering data (Fig. 1,A), we were able to further detail the structural reconstruction and make an atomic model of MBL by use of crystallographic data for the structure of a trimer of neck regions with CRDs (24) and a collagen triple helix segment (25) (Fig. 1,D). Both the neck/CRD cluster and the collagen fragments have 3-fold symmetry. The model was built with the neck/CRD cluster and six 18-residue pieces of collagen structure to fill in the stalk in a 3-fold symmetry assembly. Although the fit was acceptable for values of the scattering modulus q between 0.03 Å−1 and 0.15 Å−1, the experimental data for lower values of q did not fit well with this approach (Fig. 1,A). A comparison between the dummy residue model (Fig. 1,C) and the crystal structure of the neck/CRD cluster showed less compactness of the dummy residue structure than expected from the crystallographic structure (Fig. 2,B). We decided to compare the SAXS-determined envelope with a trimeric neck/CRD cluster where the helical necks were tipped to form an angle with the symmetry axis (Fig. 2,A). The best result was obtained for a 20° angle between the axis of symmetry and the helical necks (Fig. 2,C). With the improvement of the χ value from 6.46 to 5.42, this model agreed significantly better with the dummy residue model as compared with the unperturbed crystal structure (Figs. 1,A and 2, B and C).

FIGURE 2.

Consistency between the trimeric neck/CRD structure determined by x-ray crystallography and the MBL structure determined by SAXS. A, Variations in the angle between the axis of symmetry of the trimeric neck/CRD and the neck regions applied for fitting the atomic resolution structure with the dummy residue SAXS model. B, Overlap between the structure of trimeric neck/CRD determined by x-ray crystallography (indicated in solid gray) and dummy residue model (shown in transparent red). C, Overlap between the crystallographic structure, modeled with an angle of 20° between the axis of rotational symmetry (shown in solid gray), and the neck regions and dummy residue model (shown in transparent red).

FIGURE 2.

Consistency between the trimeric neck/CRD structure determined by x-ray crystallography and the MBL structure determined by SAXS. A, Variations in the angle between the axis of symmetry of the trimeric neck/CRD and the neck regions applied for fitting the atomic resolution structure with the dummy residue SAXS model. B, Overlap between the structure of trimeric neck/CRD determined by x-ray crystallography (indicated in solid gray) and dummy residue model (shown in transparent red). C, Overlap between the crystallographic structure, modeled with an angle of 20° between the axis of rotational symmetry (shown in solid gray), and the neck regions and dummy residue model (shown in transparent red).

Close modal

To gain insights into the dimensions of MBL at the single molecule level, we applied samples of the MBL also used in SAXS for imaging by use of AFM. The MBL molecules were immobilized through nonspecific electrostatic attachment onto the positively charged spermine-coated surfaces (Fig. 3,A) through a recently developed procedure (23). The oligomers showed a characteristic bagel-like appearance, with the globular CRDs forming a circular ring. We were not able to visualize the hub of joined stalks in the center consistent with a firm attachment of the molecules to the substrate. Nevertheless, under these conditions the molecular dimensions were similar to those for the molecule in solution; the average cross-sectional diameter of 37 nm (Fig. 3,I) agrees well with the dimensions expected from the SAXS study. The height of the molecules were found to be on average 1.46 nm (Fig. 3,H). As expected from earlier reports where the size distribution of MBL oligomers were analyzed by use of electron microscopy (6), we also observed some size heterogeneity due to the variation in the number of stalks in the MBL oligomers (Fig. 3, BI).

FIGURE 3.

AFM images of MBL on a spermine surface. A, MBL particles on a 4-Å-thick spermine-coated mica surface. The scale bar indicates a distance of 500 nm. B–G, Close-up views of MBL molecules on the spermine-coated surface. For each particle the diameter is indicated by two black arrows and the distance. The spermine-coated surface had a root mean square roughness value of <0.2 nm based on 2 × 2-μm images. The frequency of particle heights (H) or particle widths (I) on the mannosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution, determined from >200 particles, the peak value is indicated ± SEM.

FIGURE 3.

AFM images of MBL on a spermine surface. A, MBL particles on a 4-Å-thick spermine-coated mica surface. The scale bar indicates a distance of 500 nm. B–G, Close-up views of MBL molecules on the spermine-coated surface. For each particle the diameter is indicated by two black arrows and the distance. The spermine-coated surface had a root mean square roughness value of <0.2 nm based on 2 × 2-μm images. The frequency of particle heights (H) or particle widths (I) on the mannosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution, determined from >200 particles, the peak value is indicated ± SEM.

Close modal

MBL binds mannose through the 3- and 4-hydroxyl groups in the equatorial plane of the hexose ring, whereas galactose with an alternate stereochemistry of the hydroxyl groups is not a ligand (26, 27). We therefore designed a ligand-modified mica surface by applying the positively charged mannosamine and, to generate a control nonligand surface, galactosamine to the negatively charged mica surfaces. To reduce the source of structural variability originating from the size heterogeneity of the MBL oligomers, we applied purified MBL oligomers containing nine monomers (7) to the surfaces coated with either mannosamine or galactosamine.

The MBL molecules were applied to mannosamine-coated surface and imaged in the liquid with buffer conditions identical with the experimental conditions for SAXS (Fig. 4,A). In this case the molecular dimensions were clearly larger than expected from the SAXS analysis or from the molecular dimensions obtained on the spermine-coated surface. On average, the molecules had a cross-sectional diameter of ∼64 nm and a height of 4 nm; a significant portion of the molecules had even larger dimensions with a diameter of 80 nm (Fig. 4, E and F). By contrast to the spermine-coated surface, the imaging did not resolve the molecules sufficiently to distinguish molecular substructure (Fig. 4, B–D). Imaging of the MBL oligomers in liquid on the galactosamine-coated surface was not possible because no molecules were bound to the surface (data not shown).

FIGURE 4.

AFM images of MBL in liquid on a mannosamine-coated surface. A), Overview of particle distribution. The scale bar indicates a distance of 500 nm. B–D, Close-up views of MBL particles on mannosamine-coated mica. The scale bars indicate a distance of 50 nm. The frequency of particle heights (E) or particle widths (F) on the mannosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution the peak value is indicated ± SEM.

FIGURE 4.

AFM images of MBL in liquid on a mannosamine-coated surface. A), Overview of particle distribution. The scale bar indicates a distance of 500 nm. B–D, Close-up views of MBL particles on mannosamine-coated mica. The scale bars indicate a distance of 50 nm. The frequency of particle heights (E) or particle widths (F) on the mannosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution the peak value is indicated ± SEM.

Close modal

To provide a more rigorous control on the influence of MBL binding to ligand surfaces vs spontaneous adsorption to nonligand surfaces, we also analyzed the dimensions of MBL in the dry state following application of the molecule to mannosamine or galactosamine surfaces (Figs. 5 and 6, respectively). The retention of MBL molecules was significantly higher on the mannosamine than on galactosamine surface, confirming the ligand properties of the mannosamine surface (Figs. 5,A and 6,A). On the mannosamine-coated surface the MBL molecules showed a homogeneous distribution and a high coverage (Fig. 5,A). On the nonligand galactosamine-covered surface the particles had variable dimensions (Fig. 6, A, D, and E) ranging from 19.4 to 63.6 nm in width with a peak value of ∼41.5 nm (Fig. 6, D and E), i.e., close to the dimensions for the soluble MBL molecules, and the individual MBL molecules revealed no molecular substructure (Fig. 6, B and C). By contrast, the dimensions of the MBL molecules adsorbed on a mannosamine ligand-modified surface were smaller, with the peak in width and height at 27.1 and 0.58 nm, respectively (Fig. 5, G and H). A thorough inspection and analysis of the MBL molecules on the mannosamine surface revealed three nodules (Fig. 5, B–F), in some cases with a visible hub at a point equidistant to all CRD clusters (Fig. 5, B and D–F). This finding reflects that in the course of the ligand binding the MBL molecule was broken up into a molecular structure with greatly stretched stalks compared with the dimensions in solution or on the nonligand surface. Estimates of the lengths of the invisible stalks measured as the distance from the hub to the CRD clusters reveals dimensions from 40 to 60 nm (Fig. 5, B–F).

FIGURE 5.

AFM analysis of MBL oligomers with nine monomers on mica surfaces coated with mannosamine. A, Overview of particle distribution. The scale bar indicates a distance of 500 nm. B—F, Close-up views of MBL particles on mannosamine-coated mica. The scale bars indicate a distance of 50 nm. In D–F a small hub is clearly visible in addition to CRD clusters. For the oligomers the stalk lengths, defined as the distance from the hub to the center of the CRD clusters, and the angles between the stalks are indicated. The frequency of particle heights (G) or particle widths (H) on the mannosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution the peak value is indicated ± SEM.

FIGURE 5.

AFM analysis of MBL oligomers with nine monomers on mica surfaces coated with mannosamine. A, Overview of particle distribution. The scale bar indicates a distance of 500 nm. B—F, Close-up views of MBL particles on mannosamine-coated mica. The scale bars indicate a distance of 50 nm. In D–F a small hub is clearly visible in addition to CRD clusters. For the oligomers the stalk lengths, defined as the distance from the hub to the center of the CRD clusters, and the angles between the stalks are indicated. The frequency of particle heights (G) or particle widths (H) on the mannosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution the peak value is indicated ± SEM.

Close modal
FIGURE 6.

AFM analysis of size-sorted MBL oligomers with nine monomers on mica surfaces coated with galactosamine. A, Overview of particle distribution. The scale bar indicates a distance of 500 nm. B and C, Close-up views of MBL particles on galactosamine-coated mica. The scale bars indicate a distance of 50 nm. The positions of the hub and stalks were approximated from the shape of the particles. Stalk lengths and the angles between them are indicated. Frequencies of particle heights (E) or particle widths (D) on the galactosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution the peak value is indicated ± SEM.

FIGURE 6.

AFM analysis of size-sorted MBL oligomers with nine monomers on mica surfaces coated with galactosamine. A, Overview of particle distribution. The scale bar indicates a distance of 500 nm. B and C, Close-up views of MBL particles on galactosamine-coated mica. The scale bars indicate a distance of 50 nm. The positions of the hub and stalks were approximated from the shape of the particles. Stalk lengths and the angles between them are indicated. Frequencies of particle heights (E) or particle widths (D) on the galactosamine-coated surface are shown together with a Gaussian fit to the distribution. For each distribution the peak value is indicated ± SEM.

Close modal

The principle of immune recognition through the binding of soluble molecules to target surfaces is fundamental and appears as widely dispersed in both innate and adaptive immunity (28). The binding of immunoglobulins or MBL to the surfaces of pathogens initiates an immune response through receptors on leukocytes or through activation of the complement system, which plays a critical role both in the immediate clearance of intruding microbes as well as in regulating the responses of cellular defense mechanisms. Although the transition from a soluble to a surface-immobilized state is critical to the proinflammatory function of these large plasma proteins, structural insight into this process has been limited. In the present study we report that ligand surface binding by MBL has pronounced structural consequences, with conformational changes on a scale of several nanometers.

The structure of MBL in solution as determined by SAXS is highly ramified yet with a very stable separation of the stalks. Our analysis by SAXS on unfractionated MBL suggests that the predominant molecular species contains a total of nine monomers and we were consequently able to establish the structure and molecular dimensions of this oligomer, which spans 32 nm. AFM images of MBL on a spermine-coated surface confirmed this molecular size with an average cross-sectional diameter of 37 nm of the examined particles. The stable structure of MBL in solution led us to investigate the consequences of ligand surface binding to the molecular architecture. We applied MBL oligomers to mica surfaces coated with mannosamine and conducted the AFM imaging in calcium-containing buffer. In this case the dimensions of the oligomers increased to 64 nm on average with a significant fraction of the oligomers having a cross-sectional diameter at 80 nm, which is close to 3-fold larger than the oligomers in solution. The binding of the oligomers to the surface-immobilized mannosamine was confirmed from comparison with imaging of surfaces coated with galactosamine, which is not a ligand for MBL. Under these conditions the AFM probe did not detect any MBL on the surface. We also analyzed the size distribution of oligomers when these were imaged in the dry state on either mannosamine or galactosamine surfaces. Under these conditions the MBL oligomers were stretched into a highly elongated state on the mannosamine surface with the molecule obtaining a cross-sectional diameter in the order of 100 nm. By contrast, the MBL oligomers maintained dimensions similar to the solution structure when imaged on galactosamine. These findings confirm the observations from the imaging in liquid state that the oligomers are greatly stretched upon binding to the surface. We note, however, that the dimensions of the oligomers in the dry state significantly exceed the dimensions obtained in liquid state to a level that suggests some deterioration in the molecular integrity of the oligomers. Collagen fibers are highly hydrated under physiological conditions, and Raspanti et al. (29) found that the procedure of dehydration may alter the appearance of collagen fibers in AFM. Furthermore, the evaporation of water during sample preparation alters the ionic conditions in a way that is not well specified and potentially affects the molecular structure of the proteins. The quantitative discrepancy between the cross-sectional diameters in liquid and dry mode AFM imaging is likely to derive from these factors.

Based on our determination of the solution structure of MBL, it seems reasonable to identify the structural source of the stretching observed on binding ligand-coated surfaces as originating, at least in part, from a simple unbending of the stalk curvature. However, we cannot exclude the possibility that other contributions to structural deformation, such as the unfolding of flexible secondary structure elements, may also be involved. An interesting observation is that the heights of the MBL molecules on the mannosamine-coated surfaces differ between the analysis done in the liquid state and the dry state. The hydrated molecule is on average almost 7-fold higher than the dehydrated molecule. Consequently, in the hydrated state at least a part of the molecule, presumably the common hub of the collagenous stalks, is not in direct contact with the ligand-coated surface. The cavity formed between the ligand-coated surface and MBL suggests that the conformational changes are not a random process dispersing the MBL stalk region on the surface but involve a regulated fixation of the structure such that only parts of the molecule, most likely the ligand binding domains, form contacts. In any case, the MBL solution structure appears as finely tuned to accommodate the conformational changes we here describe as a consequence of surface binding.

At a more detailed level of ligand recognition we considered the structure of the trimer of ligand recognition sites at the termini of each stalk. Each CRD contains one carbohydrate binding site. Previously, the crystal structures of both human and rat neck/CRD fragments have clearly indicated that the preferred complex is that of a trimer with a fixed distance between the ligand binding sites at 5 nm (24, 30). This distance and hence the stability of the trimer of CRDs has been suggested to play a crucial role in the preference for binding microbial rather than mammalian glycans (24, 30, 31). Our study, which is the first to address the three-dimensional orientation of the trimer of CRDs relative to the collagen stalks in the context of complete MBL oligomers, shows that the SAXS-determined envelope fits poorly with the anticipations from the structures determined by crystallography. In consequence, we compared the envelope in this region with a trimer altered by tipping the neck regions to form a 20° angle with the axis of symmetry. The wider span of this structure overlaps significantly better with the SAXS envelope. In interpreting this observation it should be taken into account that the SAXS structure represents the average structure of the MBL oligomers in solution. Evidence from proteolytic digestion and the primary structure of the molecule suggests that the trimer of neck/CRDs is attached through flexible loop regions to the rigid collagenous stalks (24, 32). Given the evidence that trimers form a stable unit, it seems likely that the broadened SAXS envelope is explained by a variation in the orientation of the neck/CRD trimer relative to the stalk. The 20° angle between the neck/CRDs and the axis of symmetry thus represents the outer boundary for the space that the trimers may occupy. Our study suggests a segmented nature of the MBL oligomers where the core structure of the oligomers, consisting of the stalks, is relatively stable in solution, whereas the orientation of CRD cluster relative to the stalk region is flexible. With a point of flexibility in the connection of the neck/CRD trimers to the stalks, the neck/CRD trimers may adapt the binding interface between the CRDs and ligand glycans. This binding process, which ultimately causes a fixation of the orientation of the CRD cluster, leads to the change of the structure of the stalks; although AFM imaging as conducted in the present study does not provide sufficient molecular detail to clarify the role of the CRD clusters, we suggest that the stretching of the MBL oligomers is driven by the anchoring of the CRD clusters to ligand glycans and that the flexible attachment of the clusters plays a significant role in this process.

An important aspect in host immunity is the features that distinguish the body’s own surfaces from that of intruding microbes. The function of MBL as a pattern recognition molecule is tightly linked with the high affinity for certain carbohydrate species enriched on microbial surfaces compared with the mammalian cell surface. We now show that ligand binding involves a complex structural transition that is clearly a consequence of the ligands being distributed on a surface. The capability of the target surface topography to accommodate large conformational changes in bound protein adds a novel macrostructural component to our understanding of immune recognition at the molecular level and may constitute a significant regulatory mechanism.

We thank Annette G. Hansen for excellent technical assistance and Dr. Gregers R. Andersen for advice.

Steffen Thiel has a financial interest in NatImmune A/S, a biotechnology company exploring the possibilities of therapy with recombinant MBL.

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 study was supported by grants from The LEO Pharma Foundation, Danish Medical Research Council, Danish Natural Science Research Foundation, and Danish National Research Foundation.

4

Abbreviations used in this paper: MBL, mannan-binding lectin; AFM, atomic force microscopy; CRD, carbohydrate recognition domain; SAXS, small angle x-ray spectroscopy.

1
Janeway, C. A., Jr, R. Medzhitov.
2002
. Innate immune recognition.
Annu. Rev. Immunol.
20
:
197
-216.
2
Holmskov, U., S. Thiel, J. C. Jensenius.
2003
. Collectins and ficolins: humoral lectins of the innate immune defense.
Annu. Rev. Immunol.
21
:
547
-578.
3
Takahashi, K., W. E. Ip, I. C. Michelow, R. A. Ezekowitz.
2006
. The mannose-binding lectin: a prototypic pattern recognition molecule.
Curr. Opin. Immunol.
18
:
16
-23.
4
Teillet, F., B. Dublet, J. P. Andrieu, C. Gaboriaud, G. J. Arlaud, N. M. Thielens.
2005
. The two major oligomeric forms of human mannan-binding lectin: chemical characterization, carbohydrate-binding properties, and interaction with MBL-associated serine proteases.
J. Immunol.
174
:
2870
-2877.
5
Jensen, P. H., D. Weilguny, F. Matthiesen, K. A. McGuire, L. Shi, P. Hojrup.
2005
. Characterization of the oligomer structure of recombinant human mannan-binding lectin.
J. Biol. Chem.
280
:
11043
-11051.
6
Lu, J. H., S. Thiel, H. Wiedemann, R. Timpl, K. B. Reid.
1990
. Binding of the pentamer/hexamer forms of mannan-binding protein to zymosan activates the proenzyme C1r2C1s2 complex, of the classical pathway of complement, without involvement of C1q.
J. Immunol.
144
:
2287
-2294.
7
Dahl, M. R., S. Thiel, M. Matsushita, T. Fujita, A. C. Willis, T. Christensen, T. Vorup-Jensen, J. C. Jensenius.
2001
. MASP-3 and its association with distinct complexes of the mannan-binding lectin complement activation pathway.
Immunity
15
:
127
-135.
8
Ehrenberg, W., A. Franks.
1952
. Small-angle x-ray scattering.
Nature
170
:
1076
-1077.
9
Koide, S., X. Huang, K. Link, A. Koide, Z. Bu, D. M. Engelman.
2000
. Design of single-layer β-sheets without a hydrophobic core.
Nature
403
:
456
-460.
10
Petoukhov, M. V., D. I. Svergun.
2005
. Global rigid body modeling of macromolecular complexes against small-angle scattering data.
Biophys. J.
89
:
1237
-1250.
11
Vorup-Jensen, T., E. S. Sorensen, U. B. Jensen, W. Schwaeble, T. Kawasaki, Y. Ma, K. Uemura, N. Wakamiya, Y. Suzuki, T. G. Jensen, et al
2001
. Recombinant expression of human mannan-binding lectin.
Int. Immunopharmacol.
1
:
677
-687.
12
Pedersen, J. S..
2004
. A flux- and background-optimized version of the NanoSTAR small-angle X-ray scattering camera for solution scattering.
J. Appl. Crystallogr.
37
:
369
-380.
13
Glatter, O..
1977
. New method for evaluation of small-angle scattering data.
J. Appl. Crystallogr.
10
:
415
-421.
14
Pedersen, J. S., S. Hansen, R. Bauer.
1994
. The aggregation behavior of zinc-free insulin studied by small-angle neutron scattering.
Eur. Biophys J.
22
:
379
-389.
15
Pedersen, J. S., D. Posselt, K. Mortensen.
1990
. Treatment of the resolution function for small-angle scattering.
J. Appl. Crystallogr.
23
:
321
-333.
16
Svergun, D. I..
1999
. Restoring low resolution structure of biological macromolecules from solution scattering using simulated annealing.
Biophys J.
76
:
2879
-2886.
17
Svergun, D. I., M. V. Petoukhov, M. H. Koch.
2001
. Determination of domain structure of proteins from X-ray solution scattering.
Biophys J.
80
:
2946
-2953.
18
Konarev, P. V., M. V. Petoukhov, D. I. Svergun.
2001
. MASSHA—a graphic system for rigid body modelling of macromolecular complexes against solution scattering data.
J. Appl. Crystallogr.
34
:
527
-532.
19
Kozin, M. B., D. I. Svergun.
2001
. Automated matching of high- and low-resolution structural models.
J. Appl. Crystallogr.
34
:
33
-41.
20
Volkov, V. V., D. I. Svergun.
2003
. Uniqueness of ab initio shape determination in small-angle scattering.
J. Appl. Crystallogr.
36
:
860
-864.
21
Svergun, D. I., C. Barberato, M. H. J. Koch.
1995
. CRYSOL—a program to evalate X-ray solution scattering of biological macromolecules from atomic coordinates.
J. Appl. Crystallogr.
28
:
768
-773.
22
Koradi, R., M. Billeter, K. Wuthrich.
1996
. MOLMOL: a program for display and analysis of macromolecular structures.
J. Mol. Graphics
14
:
51
-55.
23
Andersen, E. S., S. A. Contera, B. Knudsen, C. K. Damgaard, F. Besenbacher, J. Kjems.
2004
. Role of the trans-activation response element in dimerization of HIV-1 RNA.
J. Biol. Chem.
279
:
22243
-22249.
24
Weis, W. I., K. Drickamer.
1994
. Trimeric structure of a C-type mannose-binding protein.
Structure
2
:
1227
-1240.
25
Kramer, R. Z., J. Bella, P. Mayville, B. Brodsky, H. M. Berman.
1999
. Sequence dependent conformational variations of collagen triple-helical structure.
Nat. Struct. Biol.
6
:
454
-457.
26
Drickamer, K..
1992
. Engineering galactose-binding activity into a C-type mannose-binding protein.
Nature
360
:
183
-186.
27
Weis, W. I., K. Drickamer, W. A. Hendrickson.
1992
. Structure of a C-type mannose-binding protein complexed with an oligosaccharide.
Nature
360
:
127
-134.
28
Fujita, T..
2002
. Evolution of the lectin-complement pathway and its role in innate immunity.
Nat. Rev. Immunol.
2
:
346
-353.
29
Raspanti, M., A. Alessandrini, P. Gobbi, A. Ruggeri.
1996
. Collagen fibril surface: TMAFM, FEG-SEM and freeze-etching observations.
Microsc. Res. Tech.
35
:
87
-93.
30
Sheriff, S., C. Y. Y. Chang, R. A. B. Ezekowitz.
1994
. Human mannose-binding protein carbohydrate recognition domain trimerizes through a triple α-helical coiled-coil.
Struct. Biol.
1
:
789
-794.
31
Hoffmann, J. A., F. C. Kafatos, C. A. Janeway, R. A. Ezekowitz.
1999
. Phylogenetic perspectives in innate immunity.
Science
284
:
1313
-1318.
32
Quesenberry, M. S., K. Drickamer.
1991
. Determination of the minimum carbohydrate-recognition domain in two C-type animal lectins.
Glycobiology
1
:
615
-621.