The crystal structure of iC3b-CR3 αI reveals a modular recognition of the main opsonin iC3b by the CR3 integrin receptor

Complement activation on cell surfaces leads to the massive deposition of C3b, iC3b, and C3dg, the main complement opsonins. Recognition of iC3b by complement receptor type 3 (CR3) fosters pathogen opsonophagocytosis by macrophages and the stimulation of adaptive immunity by complement-opsonized antigens. Here, we present the crystallographic structure of the complex between human iC3b and the von Willebrand A inserted domain of the α chain of CR3 (αI). The crystal contains two composite interfaces for CR3 αI, encompassing distinct sets of contiguous macroglobulin (MG) domains on the C3c moiety, MG1-MG2 and MG6-MG7 domains. These composite binding sites define two iC3b-CR3 αI complexes characterized by specific rearrangements of the two semi-independent modules, C3c moiety and TED domain. Furthermore, we show the structure of iC3b in a physiologically-relevant extended conformation. Based on previously available data and novel insights reported herein, we propose an integrative model that reconciles conflicting facts about iC3b structure and function and explains the molecular basis for iC3b selective recognition by CR3 on opsonized surfaces.


Supplementary Methods
Small-angle X-ray diffraction. Synchrotron SAXS measurements (I(s) vs. s, where s = 4p sinq /l, 2q is the scattering angle, and l = 0.992 Å) were performed at the BM29 BioSAXS beamline at the European Synchrotron Radiation Facility (Grenoble, France) 1,2 in continuous-flow batch mode at 4 ºC. Supplementary Table 2 summarizes the SAXS data and provides additional information. SAXS intensity data were recorded from 30 µL samples at several different concentrations across the ranges 1.4−5.5 mg/mL (iC3b-CR3 aI complex) and 0.17−0.7 mg/mL (CR3 aI domain) and from a matching solvent blank. The individual 2D-data frames were radially averaged to produce unsubtracted 1D I(s) vs. s scattering profiles. For the final data reduction process statistical checks were performed to discard frames affected by radiation damage or systematic scaling errors. Data were averaged, buffer subtracted, and merged to produce the final SAXS profile for each species. The ATSAS 2.8 software package 3 and BioXTAS RAW 4 were used to extract structural information and perform ab initio shape restoration. Firstly, the number of Shannon channels and maximum usable s was estimated with SHANUM 5 . The extrapolated forwarding scattering at zero concentration I(0) and the radius of gyration Rg were evaluated using the Guinier approximation (ln I(s) vs. s 2 ) and from the real-space pair-distance distribution function (P(r) vs. r) calculated with GNOM 6 . From the P(r) profile the maximum particle dimension (Dmax) could be evaluated. Two separate concentration-independent methods were used to estimate the molecular mass of the iC3b-CR3 aI complex and the free αI domain: the volume of correlation (VC) 7 and the empirical correction to the Porod volume (Vp) 8 using the ATSAS dattools DATMOW, DATVC, and DATPOROD. DATCLASS classified the SAXS derived shapes as compact and potentially unique 9 . Ab initio shape restoration was performed using dummy beads from 50 independent runs of DAMMIF 10 , which were averaged with DAMAVER, clustered with DAMCLUST 8 , and further refined with DAMMIN 11 to create the final ab initio shape. In each case, 20-40 individual dummy bead (DAMMIF) models were selected that fitted the data as judged by the reduced c 2 test and the correlation map (CorMap) p-value; values of c 2 = 1 and p > 0.01 indicate the absence of any systematic discrepancy between the data and the fitted model 12 .
Individual models were aligned with SUPCOMB with a threshold on the normalized spatial discrepancy (NSD) of 0.5, i.e. NSD < 1 for similar aligned models 13 . The final average model of the iC3b-CR3 αI domain have been produced at an estimated resolution of ~42 ± 3 Å with SASRES 14 .
Molecular dynamics. All molecular dynamics (MD) simulations were carried out with GROMACS v. 2020.3 15 . Molecular systems consisting in iC3b U -CR3 aI or iC3b D -CR3 aI plus the CUB g motif (residues 913-954) and the intact connections with the MG7 domain or the TED domain (in total, residues 907-971) were created by manual editing of coordinate files with Coot 16 and the stereochemistry of the edited model was verified with MolProbity 17 . For the protein, ligands (glycan moieties), and ions (solvent counterions to neutralize the system's net charge), the OPLS/AA force field 18,19 was used, and TIP4P 20 was used for the water.
All systems were first subjected to minimization, followed by 1-ns equilibration of the NVT and NPT ensembles. All production runs (1 µs each) were carried out in the NPT ensemble at 300 K and 1 bar. Temperature was controlled by Nosé−Hoover 21,22 (coupling constant tt = 0.1 ps) and pressure by Parrinello−Rahman 23,24 (tp = 2.0 ps) schemes. To avoid harsh density oscillations, the first 5 ns of the NPT equilibration run were performed with Berendsen weak coupling 25 for temperature (tt = 0.1 ps) and pressure (tp = 2.0 ps). Periodic boundary conditions were applied in three-dimensional space, and electrostatic forces were calculated with the Particle Mesh Ewald (PME) method 26,27 using a real-space cutoff of 1.2 nm and an FFT grid density of 6.25 nm −1 .
Lennard-Jones interactions were truncated at 1.2 nm. Covalent bond lengths in the protein and ligands were constrained to their reference values with P-LINCS 28 . SETTLE was used to constrain the water geometry 29 . Equations of motion were integrated using the leapfrog scheme with a time step of 2 fs. During the first 1 ns of this run, all protein heavy atoms were harmonically restrained to their initial positions. The latter trajectories were analyzed, and the structures were clustered using the GROMOS algorithm 30 . Multiple related clusters were detected for iC3b U -CR3 aI or iC3b D -CR3 aI, whose central structure allowed the calculation of root-mean-square fluctuations (RMSF, in Å) per residue during the trajectory ( Supplementary Fig. 10a,c). A surface representation of each representative structure confirms that relatively modest displacement of the CUB g around its equilibrium position in either complex ( Supplementary Fig. 10b,d). Although the segment does not adopt stable secondary structures, we refer to the CUB g in these models as "stable" in the sense that it retains its overall shape and characteristics for the duration of the MD runs. During the simulation, we detected the formation of a few contact points between the CUB g -containing segment and other iC3b or CR3 aI surfaces, which appear to provide some additional stabilization energy to the complex.
Data availability. Source data are provided with this paper. Tables   Supplementary Table 1 1.60 The structure was determined from a single crystal. * Values in parenthesis are for the highestresolution shell.

Supplementary Table 2. Interactions observed between the TED domain of iC3b and CR3
aI. Interactions and other properties of the interface(s) were calculated with PISA 31 . The interface area is 465.6 Å 2 , the solvation energy is −1.2 kcal/mol, the total binding energy is −5.6 kcal/mol, and the hydrophobic p-value is 0.5. The number of hydrogen bonds is 10, and there are no salt bridges or disulfide bridges. Distances are shown in Å.

Supplementary Table 3. Interactions observed between the C3c moiety of iC3b U and CR3
aI or the TED domain. Interactions were calculated with PISA 31 . This interface is a composite interface made up of residues from the iC3b b chain with a minor contribution from the C3c a' fragment 1. Concerning the C3c moiety-CR3 aI domain interface, the interface area is 873.5 Å 2 , the solvation energy is 2.4 kcal/mol, the total binding energy is −3.0 kcal/mol, and the    31 . This interface is a composite interface made up of residues from the iC3b b chain and residues from the iC3b a' fragment 1 chain. For the C3c moiety-CR3 aI domain interaction, the interface area is 836.4 Å 2 , the solvation energy is −3.5 kcal/mol solvation energy, the total binding energy is −7.0 kcal/mol, the