FcγRIIB-I232T polymorphic change allosterically suppresses ligand binding

FcγRIIB binding to its ligand suppresses immune cell activation. A single-nucleotide polymorphic (SNP) change, I232T, in the transmembrane (TM) domain of FcγRIIB loses its suppressive function, which is clinically associated with systemic lupus erythematosus (SLE). Previously, we reported that I232T tilts FcγRIIB’s TM domain. In this study, combining with molecular dynamics simulations and single-cell FRET assay, we further reveal that such tilting by I232T unexpectedly bends the FcγRIIB’s ectodomain toward plasma membrane to allosterically impede FcγRIIB’s ligand association. I232T substitution reduces in situ two-dimensional binding affinities and association rates of FcγRIIB to interact with its ligands, IgG1, IgG2 and IgG3 by three to four folds. This allosteric regulation by an SNP provides an intrinsic molecular mechanism for the functional loss of FcγRIIB-I232T in SLE patients.


Introduction 36
Disorders or hyper activation of immune components could lead to autoimmune diseases. Malfunction of an immune receptor, FcγRIIB, is generally regarded as 38 destructive for immune system [1][2][3]

Results and Discussion
In this report, we firstly performed systemic examination over the association of 56 FcγRIIB-I232T with clinical manifestations of SLE. We enrolled 711 unrelated Chinese patients with SLE and complete clinical documents into this study (Table 1). 688 58 unrelated healthy Chinese volunteers with matched gender and age were then enrolled as controls. We confirmed a strong positive association of the homozygous FcγRIIB-60 I232T polymorphism with SLE (χ 2 = 7.224, p = 0.008, odds ratio with 95% confidence interval (CI) = 1.927) (  and abnormal elevation of PLCγ2 activation, proliferation and calcium mobilization 11 . FcγRIIB-I232T B cells lose the ability to inhibit the oligomerization of B cell receptors 92 6 (BCRs) upon co-ligation between BCR and FcγRIIB 12 . Recent live-cell imaging studies showed that B cells expressing FcγRIIB-I232T fail to inhibit the spatial-94 temporal co-localization of BCR and CD19 within the B cell immunological synapses 13 . Human primary B cells from SLE patients with homozygous FcγRIIB-I232T 96 mutation revealed hyper-activation of PI3K 13 . Thus, FcγRIIB-I232T is very likely the first example that a naturally occurring SNP within the TM domain of a single-pass 98 transmembrane receptor can cripple its function in principle and is significantly relevant in diseases. 100 These signaling events are usually triggered or followed by ligand engagement of FcγRIIB, while this function is disrupted by a single amino acid change from Ile to Thr 102 in the TM domain. Two early biochemical studies proposed a model of reduced affinity between FcγRIIB-I232T and lipid rafts to explain the functional relevance and effect of 104 this mutation [10][11] . Another model suggested that I232T mutation enforces the inclination of the TM domain and thereby reduces the lateral mobility and inhibitory 106 functions of FcγRIIB. However, both models assumed that FcγRIIB-I232T and FcγRIIB-WT (I232) have an equal capability to perceive and bind to the ligand, the IgG 108 Fc portion within the antibody antigen immune complexes. This important but experimentally un-proved pre-requisition in both models is based on the argument that 110 linker extends more in the I232T form than that in the WT, and the length between S218 and P221 peaks at 11 Å for I232T, 3 Å longer than the 8 Å peak position for I232 130 system ( Figure 1C). This length elongation further results in a different conformation of residue P217, the main chain dihedral angle of P217 in I232 system displays two 132 populations at 141°±23° and -50°±12°, respectively, but shifts to 4°±45° and -75°±12° in the T232 form ( Figure 1D and Figure 1-figure supplement 2). These effects 134 8 propagate and lead to striking effect on the extracellular domains of FcγRIIB. We found that the extracellular domain of the T232 form adopts significant different conformation 136 than that of the I232 form. The ectodomain of I232 maintains more straight conformation, whereas that of T232 bends down towards the lipid bilayer ( Figure 1E). 138 Statistical analyses show that the ectodomain inclination angle of the T232 form distributes across 30~60° with a sharper single-peak at 40° ( Figure 1E). In contrast, the 140 angle of the I232 form distributes more flatter with a most favorable probability ranging from 50° to 70° ( Figure 1E). The distance of C1 domain is much closer to the membrane 142 for the T232 form than the I232 form ( Figure 1E). These results suggest that the T232 morphism (or I232T mutation) may reduce the antibody recognition ability of FcγRIIB 144 via two aspects. First, although the Fc binding site is not buried, the orientation and membrane binding of T232 may sterically prevent the accessibility of the Fc portion of 146 IgG, as significant clashes between docked Fc and the membrane are observed ( Figure   1-figure supplement 1B). Second, T232 is more rigid (or less flexibility, Figure 1E) 148 such that the chance to associate with the ligand is decreased (thus the ligand association rate may be significantly reduced).  Ectodomain orientation changes of a receptor can significantly affect its in-situ 194 binding affinity with its ligands 17 . We therefore predict that titling FcγRIIB ectodomain towards plasma membrane by I232T polymorphism may attenuate its ligand binding 196 affinity, especially the association rate. To test this hypothesis, we applied wellestablished single-cell biomechanical apparatus with adhesion frequency assay 18 to 198 directly and quantitatively measure in-situ two-dimensional (2D) binding kinetics of the WT or I232T FcγRIIB binding with its ligands ( Figure 3A). It revealed that the in-200 12 situ 2D effective binding affinity of FcγRIIB232I with MERS virus S protein human IgG1 antibody (anti-S) is about three times higher than that of FcγRIIB232T binding 202 with same antibody (A c K a =3.03±0.15×10 -7 and 0.80±0.04×10 -7 µm 4 , respectively), whereas that of human IgG4 is hardly measured as its binding is too weak and beyond 204 the detection limit (10 -8 µm 4 ) 18 of this assay ( Figure 3B and 3D), which is consistent with previous reported FcγRIIB/IgG4 binding affinity is far less than IgG1 1 . Moreover, 206 off-rates of the WT and the I232T form binding with human IgG1 are similar (7.75±1.42 and 7.62±1.41 s -1 , respectively) ( Figure 3B and 3F), while the 2D effective 208 on-rate of the I232T binding with human IgG1 is three times slower than that of the WT binding with same ligand (Figure 3B and 3E). These kinetics data strongly support our 210 prediction that I232T polymorphism tilts FcγRIIB ectodomain more recumbent toward the plasma membrane so that its ligand binding domain is harder to be accessed, which 212 reduces FcγRIIB/IgG1 binding on-rate. The conclusion is also confirmed by FcγRIIB binding with another human IgG1 (HIV1 gp120 human IgG1, anti-gp120) (Figures 3C 214 to 3F). That is, the 2D effective affinity of FcγRIIB binding with human IgG1 and onrate both are three times higher than those of 232T's (A c K a =7.74±0.24×10 -7 and  reported [10][11] . The DNA sequencing was done by BGI (Beijing). The Pearson chi-square tests were performed for the comparison of differences between cases and controls at 256 genotype model (recessive model CC vs. TT+TC). The odds ratios (OR), 95% confidence intervals (CI) and p value for recessive model analysis were calculated using 258 logistic regression, adjusting for age and sex. In statistical analyses, p value of less than 0.05 was considered statistically significant. 260

Molecular Dynamics Simulations
Structure model of the full human FcγRIIB system (residues A46-I310) was built 262 by fusing the crystal structure of the ectodomain (PDB code 2FCB, residues A46-Q215) 16 to the transmembrane (TM) helix (residues M222-R248) model obtained in previous 264 study 16 , the stalk (residues A216-P221) and cytoplasmic regions (residues K249-I310) are randomly placed. An asymmetric lipid bilayer with the membrane lateral area of 266 100×100 Å 2 was generated with Membrane Builder in CHARMM-GUI 19  The resulted systems were subjected to productive simulations for 200 ns with 2 fs timestep without any constrains, and the snapshots of the last 80 ns (sampled at 10 286 ps intervals) were used for detailed analyses including the probability distributions of hydrogen bonds, tilting angles of the TM helix, inclination angles of ectodomain, the 288 distance between Ig-like C2-type 1 domain and lipid bilayer. The tilting angle of TM helix is defined as the angle between TM helix and membrane plane, similar as that 290 used in previous study 16 . The inclination angle of ectodomain is defined as the angle between the membrane plane and the vector linking NT-terminal of TM helix (M222-292 I224) and linker region of Ig-like C2-type 1 and 2 domain (S130-W132). The distance between Ig-like C2-type 1 domain and lipid bilayer is defined as the length between 294 center of mass (COM) of this domain and the heavy atoms of phospholipid head in the normal direction of bilayer. 296 All simulations were performed with NAMD2 software 21 using CHARMM36m force field with the CMAP correction 22 . The simulations were performed in NPT 298 ensemble (1 atm, 310K) using a Langevin thermostat and Nosé-Hoover Langevin piston method 23 , respectively. 12Å cutoff with 10 to 12 Å smooth switching was used 300 for the calculation of the van der Waals interactions. The electrostatic interactions were computed using the particle mesh Eward method under periodic boundary conditions. 302 The system preparations and illustrations were conducted using VMD.

RBC preparation
Streptavidin-coated red blood cells (RBCs) preparation have been described 324 previously 18 . IgG was biotinylated by EZ-Link Sulfo-NHS-LC-Biotin kits (Thermo Fisher Scientific). Different amounts of biotinylated IgG was linked into RBCs through 326 SA-biotin interaction at RT for 30 min, respectively. IgG-coated RBCs were obtained for micropipette adhesion frequency assay to measure 2D binding kinetics of 328 FcγRIIB/IgG. All above experimental processes were followed by the institutional ethical review board of Zhejiang University. 330

2D binding kinetics measurements
The micropipette adhesion frequency assay was applied to measure FcγRIIB/IgG 332 2D in-situ binding kinetics. The detail experimental progress was previously described 18 . In brief, biotinylated human antibodies (IgG1 or IgG4) were coated on red 334 blood cell (RBC) with streptavidin-biotin association. Two opposing micropipettes aspirated the RBC and FcγRIIB A20II1.6 B cell (232I or 232T) respectively to operate 336 contact-retraction cycles manipulation. Through these 50 contact-retraction cycles, the binding frequency was acquired with definite contact area and a series of setting contact 338 time (0.1, 0.2, 0.5, 1 and 2 s). 3~4 cell pairs were tested for each setting contact time.
And these data were fitted by probabilistic kinetic model. In order to accurately 340 calculate 2D binding affinity and on-rate, these two surface molecular densities  Conformational difference of I212-S220 regions after aligning residues S218 to S220. The association between FcγRIIB ectodomain and membrane in I232T polymorphism is mediated by multiple residues. The residues important to the association can be obtained by comparing the contact ratios per residue for WT (232I, green) or the FcγRIIB-232T (232T, blue).
Regions with greater contact ration differences are highlighted in the insets. Of them, N168, K170, S191 and D193 may play more essential roles.