Antigen-Induced Oligomerization of the B Cell Receptor Is an Early Target of FcγRIIB Inhibition

Abstract

The FcγRIIB is a potent inhibitory coreceptor that blocks BCR signaling in response to immune complexes and, as such, plays a decisive role in regulating Ab responses. The recent application of high-resolution live cell imaging to B cell studies is providing new molecular details of the earliest events in the initiation BCR signaling that follow within seconds of Ag binding. In this study, we report that when colligated to the BCR through immune complexes, the FcγRIIB colocalizes with the BCR in microscopic clusters and blocks the earliest events that initiate BCR signaling, including the oligomerization of the BCR within these clusters, the active recruitment of BCRs to these clusters, and the resulting spreading and contraction response. Fluorescence resonance energy transfer analyses indicate that blocking these early events may not require molecular proximity of the cytoplasmic domains of the BCR and FcγRIIB, but relies on the rapid and sustained association of FcγRIIB with raft lipids in the membrane. These results may provide novel early targets for therapies aimed at regulating the FcγRIIB to control Ab responses in autoimmune disease.

B lymphocytes are activated by the binding of Ag to the clonally distributed BCRs (1, 2). The FcγRIIB is a potent inhibitor of B cell activation and plays a central role in regulating Ab responses (3–5). The FcγRIIB, the only classic FcR expressed by B cells, has been shown to regulate both the magnitude and persistence of Ab responses through effects on both B cells and plasma cells (4, 6). Deficiencies in FcγRIIB result in susceptibility to autoimmune diseases and in certain genetic backgrounds, severe autoimmune disease and death (7). The FcγRIIB blocks BCR signaling cascades when colligated to the BCR through the binding of Ag-containing immune complexes (ICs) (4). Ag binding to the BCR induces the phosphorylation of the ITAMs within the cytoplasmic domains of the Igα and Igβ-chains of BCR complex by the Src family kinase Lyn (2). Colligating the BCR and FcγRIIB leads to the phosphorylation of the ITIM in the cytoplasmic domain of the FcγRIIB, resulting in the recruitment of the SH2-domain containing inositol-5-phosphatase (SHIP) that hydrolyses the PI3K product PtdIns(3,4,5)P₃ (PIP3) to PtdIm(4,5)P₂ blocking the recruitment of PH domain-containing proteins and inhibiting BCR signaling (8).

Our current understanding of the mechanisms that underlie the inhibition of B cell activation by the FcγRIIB is based almost exclusively on biochemical analyses of events that occur down-stream of Lyn’s phosphorylation of the BCR in response to soluble ICs. Advances in high-resolution live cell imaging technologies and their application to the study of B cell activation are providing a new view of the earliest events in the initiation of BCR signaling preceding Lyn’s phosphorylation of the BCR that follow within seconds of membrane-associated Ag binding (9, 10). Batista et al. (11) first showed that B cells were strongly activated by Ags presented on APCs, and in the process the B cells formed immune synapses (ISs) similar to those described for T cells. Most recently, the initiation of B cell activation has been explored in vitro following the B cell’s encounter with Ag attached to planar lipid bilayers (12, 13), mimicking the presentation of Ag to B cells on the surfaces of APCs, as current evidence suggests occurs in vivo (14–17). Using fluid lipid bilayers containing Ag, and allowing higher resolution imaging, B cells were shown to first touch Ag-containing lipid bilayers in a few contact points in which signaling active BCR microclusters formed and triggered actin-dependent spreading (13). The spreading response significantly increased the area of contact of the B cells with the Ag-containing bilayers, promoting the formation of additional new BCR microclusters in the leading edge of the cells. The newly formed BCR microclusters moved along actin fibers from the periphery to the center of the contact area, where they accumulated to form a central IS. The B cell then contracted and drove all the BCR microclusters to coalesce into the central IS. The amount of Ag that the B cells engaged was dependent on the degree of spreading, which in turn was amplified by BCR-Ag engagement. As a consequence, this process provided a mechanism for B cells to discriminate the amount and affinity of Ags. These remarkable studies provided evidence that Ag recognition and B cell activation were far more dynamic processes than previously understood.

Focusing on the initial formation of BCR microclusters, before establishing a central IS and using single molecule imaging techniques, we recently showed that the binding of B cells to Ag in a planar fluid lipid bilayer resulted in the assembly of immobile, signaling active BCR oligomers (18). A mutational analysis of the BCR provided evidence that BCR oligomerization involved a novel mechanism in which Ag binding induced a conformational change in the Cμ4 domain of the Fc portion of the BCR’s mIgM, revealing an oligomerization interface resulting in BCR microclustering (18). These events were BCR intrinsic and did not require a signaling competent BCR (18). We coined this the conformation-induced oligomerization model for BCR microclustering and signaling (9). Additional recent studies showed that newly formed BCR microclusters perturbed the local lipid environment, leading to their transient association with a lipid raft probe (19). A subsequent, more stable association of the BCR microclusters with membrane-tethered Lyn was observed, and this association correlated both temporally and spatially with the transition of the cytoplasmic domains of the microclustered BCRs from a closed to an open signaling active form (19). We postulated that the condensing of raft lipids around the BCR microclusters was necessary for the stabilization of BCR oligomers and may induce the conformational change in the cytoplasmic domains (9). We speculate these early events might be targets of FcγRIIB inhibition based on our earlier study showing that, when colligated to the BCR, the FcγRIIB destabilized BCR and membrane raft lipids interaction and blocked the formation of an IS (20).

In this study, we provide evidence that the colligation of the FcγRIIB with the BCR by membrane-associated ICs blocks the early dynamic events in B cell activation that precede the BCR’s activation by Lyn. We show that when colligated to the FcγRIIB, the Ag-engaged BCRs fail to form immobile BCR oligomers, fail to undergo a transition from a closed to an open signaling active form, and fail to trigger cell spreading. These effects of the FcγRIIB on the BCR may not require molecular proximity of the BCR and FcγRIIB, but rather appeared to rely on the FcγRIIB’s ability to perturb the local lipid environment. FcγRIIBs with loss of function mutations in either the cytoplasmic domain’s ITIM motifs or the transmembrane domain and that block the association of the FcγRIIB with raft lipids were unable to block these early events. These results indicate that the FcγRIIB interrupts BCR signaling at an earlier step than was evident from biochemical studies of B cells responding to soluble ICs. The novel mechanism described in this study likely complements the well-documented ability of the FcγRIIB to block the downstream BCR signaling through the lipid phosphatase SHIP, and may provide new targets to moderate the function of the FcγRIIB in autoimmunity.

Materials and Methods

Mice, cells, Abs, and Ags

Primary B cells were isolated from spleens of IgH^B1-8/B1-8 Igκ ^−/− transgenic mice (provided by M. Shlomchik, Yale University, New Haven, CT) by negative selection using MACS sorting as described previously (18). Isolated B cells were cultured overnight with CpG and LPS (Calbiochem, San Diego, CA) and experimented next day. The J558L B cells stably expressing B1-8-γ-cyan fluorescent protein (CFP) and Igα-yellow fluorescent protein (YFP) (γCαY) were established and characterized as previously described (21). Daudi B cell line stably expressing Lyn16-CFP-YFP fusion protein and CH27 B cell line stably expressing the lipid raft probe Lyn16-CFP, the nonlipid raft probe CFP-Ger, or the full length LynFL-CFP were established and characterized as previously reported (19). ST486, a human B cell line and A20II1.6, a mouse B cell line, both negative for endogenous FcγRIIB expression, were purchased from the American Type Culture Collection (Manassas, VA). PT67 cell line, a virus packaging cell line, was purchased from BD Clontech (Palo Alto, CA).

Cy3- and Cy5-conjugated Fab goat Abs specific for mouse IgM and IgG were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). Cy3-conjugated Fab rat mAb specific for mouse IgM (clone no. II/41) was purchased from Rockland (Gilbertsville, PA). Rabbit IgG Abs specific for BSA (rabbit anti-BSA) were purchased from Bethyl Laboratories (Montgomery, TX) and F(ab′)₂ rabbit anti-BSA were prepared as described (20). PE-conjugated and biotin-conjugated rat mAb specific for mouse FcγRIIB (clone no. 2.4G2) and APC-conjugated mouse mAb specific for human FcγRIIB (clone no. FLI8.26) were purchased from BD Pharmingen (San Diego, CA). Biotin-conjugated mouse mAb specific for human FcγRIIB (clone no. AT10) was purchased from AbD Serotec (Raleigh, NC). Biotin-conjugated F(ab′)₂ goat Abs specific for mouse IgG and IgM, biotin-conjugated F(ab′)₂ rabbit Abs specific for human IgM, and streptavidin were purchased from Jackson ImmunoResearch Laboratories. Rat mAb specific for mouse FcγRIIB (clone no. 190907) was purchased from R&D Systems (Minneapolis, MN) and Fabs mAb 190907 were prepared using an immobilized papain kit (Pierce, Rockford, IL) following the manufacturer’s protocol. Conjugations of Abs with Alexa514, 568, or 647 were performed using Alexa Fluor mAb labeling kits (Molecular Probes, Eugene, OR) following manufacturer’s protocols. BSA conjugated 1:14 with 4-hydroxy-5-iodo-3-nitrophenyl acetyl (NIP) (NIP₁₄-BSA) and BSA conjugated 1:16 with phosphorylcholine (PC₁₆-BSA) were purchased from Biosearch Technologies (Novato, CA). NIP₁₄-BSA and PC₁₆-BSA were conjugated to a Cys-containing peptide terminated with a His₁₂ tag (ASTGTASACTSGASSTGSH₁₂) using succinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate (Pierce) following manufacturer’s protocols. ICs were formed by mixing 10 nM His₁₂ tagged NIP₁₄-BSA or PC₁₆-BSA with 20 nM rabbit anti-BSA (for IgG-IC) or F (ab′)₂ rabbit anti-BSA [for F(ab′)₂-IC]. Recombinant ICAM-1 with a His₁₂ tag was a gift of J. Huppa (Stanford University, Palo Alto, CA). The mouse ICAM-1/huFc chimera protein with a His₁₂ tag was purchased from R&D Systems. Conjugation of His₁₂-tagged NIP₁₄-BSA and PC₁₆-BSA to Cy5 and His₁₂-tagged ICAM-1 to AlexaFluor488 were performed following manufacturer’s protocols (19).

Plasmid constructs and transfections

Plasmids expressing Lyn16-YFP or Lyn16-CFP and containing monomeric CFP or yellow fluorescent protein (YFP) were constructed as described (19). The pMSCV series plasmids were purchased from Clontech. Plasmids of pCDNA3.1 and pCDNA6 series were purchased from Invitrogen (Carlsbad, CA). Plasmids of pECFP-N1 or pEYFP-N1 were purchased from BD Clontech, and the GFP gene in these two plasmids was modified using a QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) to reduce the dimerization of GFP as suggested (22). LynFL-CFP plasmid expressing full length Lyn kinase and CFP-Ger plasmid expressing the geranylgenylated CFP were constructed as described (19, 23).

Plasmids of pBLUEScript containing mouse FcγRIIB1 and pMSCVpuro plasmids containing human FcγRIIB WT or I232T loss of function mutant were kindly provided by Z. I. Honda (Tokyo University, Tokyo, Japan) (24). Using these plasmids as templates, the genes encoding mouse or human FcγRIIB were amplified by PCR and inserted into pECFP-N1, pEYFP-N1, or pCDNA6/Bsd via the restriction enzyme sites of SacII and HindIII. To acquire constructs of FcγRIIB fused with monomeric YFP in pMSCV plasmid, the released fragments containing genes of FcγRIIB and YFP from plasmid pEYFP-N1-FcγRIIB by BglII and NotI double digestion were subcloned into the multiple cloning site of pMSCV plasmid with a NotI restriction enzyme site placed into its multiple cloning site (23). In the case of mouse FcγRIIB, the BglII restriction site AGATCT carried by mouse FcγRIIB itself was synonymously mutated to AGACCT using a QuikChange Mutagenesis Kit (Stratagene). The Y309F mutation in mouse FcγRIIB ITIM motif, and the CD314 mutant deleting the C-terminal 16aa residues in mouse FcγRIIB was made using a QuikChange mutagenesis kit (Stratagene). A J558L B cell line expressing the fluorescence resonance energy transfer (FRET) pair B1-8-γ-CFP and γCαY was generated and maintained in the laboratory as described (21). Mouse FcγRIIB WT or Y309F and human FcγRIIB WT or I232T were transfected into γCαY B cells in pCDNA6 by electroporation. Cell lines stably expressing the plasmid were acquired by blasticidin selection and cell sorting. PT67 packaging cells expressing murine stem cell virus (MSCV) containing FcγRIIB-YFP were prepared as described (18, 19). Briefly, the PT67 cell line was transfected with a pMSCV plasmid containing mouse FcγRIIB-YFP construct by spin infection and the virus in the supernatant was used to infect a CH27 B cell line stably expressing Lyn16-CFP. Stable cell lines expressing Lyn16-CFP and FcγRIIB-YFP were acquired by puromycin selection and cell sorting. Transient transfection was performed using Amaxa transfection kits and the transfected B cells were imaged after overnight culture.

Preparation of planar lipid bilayers containing ICAM-1 and ICs

Planar fluid lipid bilayers were prepared as previously detailed (25–27). Briefly, Ni-NTA-containing lipid bilayers were prepared using the 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) lipid and 1,2-dioleoyl-sn-glycero-3-[N(5-amino-1-carboxypentyl) iminodiacetic acid]-succinyl (nickel salt) (DOGS-Ni-NTA; Avanti Polar Lipids, Alabaster, AL) with 90% DOPC and 10% DOGS-Ni-NTA. ICs were formed by mixing 10 nM His₁₂ tagged NIP₁₄-BSA or PC₁₆-BSA with 20 nM rabbit BSA-specific (for IgG-IC) or F(ab′)₂ rabbit BSA-specific [for F(ab′)₂-IC]. F(ab′)₂-IC or IgG-IC was further mixed with 10 nM His₁₂ tagged ICAM-1, and then incubated on Ni-NTA–containing lipid bilayers for 20 min for binding. After washing, the F(ab′)₂-IC or IgG-IC containing lipid bilayers were ready to be used in total internal reflection fluorescence (TIRF) imaging. ICAM-1, IgG-IC and F(ab′)₂-IC were mobile on planar lipid bilayer determined as described previously (19, 20).

Biotin-containing planar lipid bilayers were prepared with 99% DOPC and 1% 1,2-Dioleoyl-sn-Glycero-3-phosphoethanolamine-cap-biotin (DOPE-cap-biotin) (Avanti Polar Lipids) to which biotinylated ICAM-1 and Abs were attached through streptavidin as previously reported (19). Briefly, 50 nM streptavidin was incubated with biotin-containing lipid bilayer for 10 min. After washing, 10 nM biotinylated ICAM-1 and 10 nM biotinylated F(ab′)₂ Abs specific for either mouse or human Igand/or10 nM biotinylated rat Abs specific for either mouse or human FcγRIIB were bound to the planar lipid bilayer.

Live cell imaging by total internal reflection fluorescence microscopy and image processing

TIRF images were acquired using an Olympus IX-81 microscope (Melville, NY) equipped with a TIRF port, Cascade II 512 × 512 electron-multiplying CCD camera (Roper Scientific, Tucson, AZ), Olympus 100 × 1.45 N.A. or Zeiss 100 × 1.4 objective lens (Carl Zeiss, Salem, MA). A 442-nm diode pump solid state laser, a 488-nm/514-nm argon gas laser and a 568-nm/647-nm red krypton/argon gas laser were equipped and used as indicated. All images were acquired at 37°C on a heating stage. The acquisition was controlled by Metamorph (Molecular Devices, Sunnyvale, CA) and the exposure time was 100 ms for 512 × 512 pixels image, unless specially indicated. The acquired images were analyzed and processed with Image Pro Plus (Media Cybernetics, Silver Spring, MD), Image J (National Institutes of Health) or Matlab (The MathWorks, Natick, MA) software as indicated. Before analysis, images were split, aligned, background subtracted, and corrected for spectral bleed-through using Image Pro Plus or Matlab software depending on needs. Characterization of the B cell clustering was the protocol reported by Batista et al. (12). Briefly, the B1-8 primary B cells were labeled with 200 nM Fab IgM-specific Cy3-conjugated rat mAb (clone II/41). The labeled cells were washed twice and then loaded to the lipid bilayer containing F(ab′)₂-IC or IgG-IC. TIRF images were acquired every 2 s immediately after the loading of the B cells to the chamber. Cy3 was excited by 514 nm laser, and Cy3 fluorescence was collected through a 550/40 ET BP emission filter. The spreading area, the mean fluorescence intensity, and number of the BCR microclusters were measured using Image J software. The number of BCR microclusters per cell at maximal spreading was manually counted after background filtering the TIRF image by Matlab software. The threshold efficiency used for the filtering was best estimated by comparing the mean fluorescence intensity (FI) of BCR microclusters with the mean FI of IgM-specific Cy3-conjugated monoclonal Fab fragment nonspecifically bound to the lipid bilayer, which was acquired with the same total internal reflection fluorescence microscopy (TIRFM) settings used for B cell image. Trajectories of individual microclusters were tracked automatically using a Matlab-supported code. The mean square displacement (MSD) for each BCR microcluster was calculated from positional coordinates as reported (28, 29). A detailed description of the automatic tracking is given below in Single particle tracking and analysis. To sort the movement of individual BCR microcluster into different categories (random, directed, or confined movement), the MSD plot of each BCR microcluster was fitted with the three functions to define random, directed, and confined movement, respectively as below:

$$F(t)=a+4\ast D\ast t;$$ (1)

$$F(t)=a+4\ast D\ast T+\upsilon 2\ast (T2);$$ (2)

$$F(t)=a+(L2/3\ast [1-exp(-12\ast D()\ast T/[L2])]).$$ (3)

For each fit, the square of correlation coefficient (R²) is acquired as a parameter to quantify the goodness of each fit. In this report, if the R² values of one certain BCR microcluster from all three fittings are <0.33, we will sort this specific BCR microcluster into the category of complex type of movement, suggesting that the movement of this BCR microcluster is not a simple random, directed, or confined movement. The percentage of complex type of movement is < 10% in any examined groups in this report. Otherwise, the SD of all three R² values from one certain BCR microcluster is calculated; if SD < 0.015, we will sort this BCR microcluster into the category of random type of movement; if SD ≥ 0.015, we will sort this BCR microcluster into the category of directed or confined type of movement, whichever R² value is greater.

Two-color TIRF images of BCRs and FcγRIIBs on the encounter of B1-8 primary B cells on planar lipid bilayers were captured every 2 s using multiple dimensional acquisition mode controlled by the Metamorph system. The B1-8 primary B cells were prelabeled with IgM-specific Cy5-conjugated mAb Fab and FcγRIIB-specific Alexa568-conjugated mAb Fab. Cy5 and Alexa568 were excited by 647 and 568 nm laser, respectively, and the switch of 647 and 568 nm lasers were completed by an acousto-optical tunable filter. Cy5 and Alexa568 fluorescence were collected by 665 LP and 605/40 BP emission filters, respectively, through a 488/568/647 dichroic wheel filter cube. Quantitative evaluation of the colocalization of green and red images in two-color TRIF images was performed based on the intensity correlation analysis as described (30), and Pearson’s correlation index was given as the output result using the WCIF plugin of Image J software.

FRET image acquisition and FRET efficiency calculation

The acquisition of CFP, FRET, and YFP images by TIRF microscopy (TIRFM) and the calculation of FRET efficiency were as described (18, 19, 21, 23). A 442-nm laser (Melles Griot, Carlsbad, CA) was used to excite CFP, and a 514-nm argon gas laser was used to excite YFP in time lapse TIRF FRET live cell imaging. CFP and FRET TIRF images were acquired simultaneously using a dual image splitter (MAGS Biosystems, Tuscon, AZ) equipped with a 505 dichroic beamsplitter and HQ485/30 (CFP) and HQ550/30 (FRET) emission filters (Chroma Technology, Rockingham, VT). YFP channel images were acquired through the same dual image splitter with a HQ540/30 emission filter. CFP FRET dual view and YFP images were sequentially acquired every 2 s with alternative switching 442 or 514-nm laser by an acousto-optical tunable filter. Images were captured into 16-bit grayscale 512 × 512 pixel images without binning and averaging by a Cascade II electron-multiplier CCD camera (Photometrics, Tuscon, AZ), controlled by Metamorph software (Molecular Devices, Downingtown, PA). FRET images obtained by TIRFM were analyzed by the sensitized acceptor emission method as described (21, 23, 31, 32). The FRET efficiency normalized with YFP intensity (Ea) was used in this report because Ea was a more reliable indicator for FRET efficiency, depending less on the donor stoichiometry as reported previously (21). The CFP FRET dual view images from 1.0-μm blue and green polystyrene beads (excitation at 430 nm and emission at 465 nm; Molecular Probes) served as a dual-view image alignment reference. Correction factors for donor (CFP) bleed-through (β) in the FRET channel and acceptor (YFP) crosstalk (γ) in the FRET channel were obtained from images of single CFP- or YFP-expressing CH27 B cells acquired using the same TIRFM settings as for the experimental cells. In our TIRFM settings, the bleed-through of YFP emission to the CFP channel (δ) from the YFP single-positive cell was negligible and thus was counted as zero in the calculation of FRET efficiency.

The calculation of Ea based FRET efficiency using sensitized donor (CFP), FRET, and acceptor (YFP) images were described previously (19, 20, 21, 23). CFP FRET dual-view and YFP images acquired in FRET TIRFM were aligned and split into individual CFP (D), FRET (F) and YFP (A) images using an Image Pro Plus software package (Media Cybernetics). The individual CFP, FRET, and YFP images were background-subtracted and smoothed by a Gauss filter using Image Pro Plus. FRET efficiency (Ea) was calculated by the following equation as described previously (18, 19, 21, 23):

$$\text{Ea}=(F-\beta D-\gamma A)/\gamma A{K}_{\text{A}}.$$

K_A is the ratio of the extinction coefficient of donor (εCFP) and acceptor (εYFP) when excited at donor excitation wavelength 442 nm. In our experiments, K_A was calculated as reported previously (18, 19, 21, 23):

$$\begin{array}{c}{K}_{\text{A}}=(F-\beta D-\gamma A)/(\gamma A{E}_{\text{bleaching}})\\ E_{\text{bleaching}}=(1-D_{\text{before}}/D_{\text{after}}).\end{array}$$

D_before and D_after were the CFP fluorescence intensities in the donor channel before and after complete photo-bleaching of acceptor YFP. To calculate the D_before and D_after, we used Daudi human B cells expressing Lyn16-CFP-YFP fusion protein, a construct with a 1:1 CFP and YFP ratio and fixed distance between CFP and YFP proteins producing positive FRET efficiency (18, 19, 21, 23). For the quantification of FRET efficiency at the single-cell level over time, the mean FI from the TIRF images of CFP (D), FRET (F) and YFP (A) were acquired by the autotracking function of the Image Pro Plus software. When tracking, the contact area of the cell as a whole was taken as the region of interest, and the 3 × 3-pixel autotracking mode was used in the software setup. The background FI in close proximity to BCR microclusters was used as the lower threshold in the segmentation setup of the autotracking mode. The ratio of donor and acceptor (CFP/YFP) was calculated based on the YFP and the corrected CFP FI, with the FRET loss compensated as reported in previous studies (18, 19, 21, 23). CFP/YFP ratio was calculated as:

$$\text{CFP}:\text{YFP}\text{ratio}=[D/A+(F-\beta D-\gamma A)/{AK}_{\text{D}}]\times (\text{CFP}:\text{YFP}\text{ratio}\text{control}).$$

K_D is the multiplied results of two ratios: ratio 1 × ratio 2. Ratio 1 is the quantum yields of acceptor (Q_YFP) divided by donor (Q_CFP). Ratio 2 is the constant F (C_F) divided by the constant D (C_D), where C_F is a constant defining the efficiency of detection acceptor FI with the TIRF FRET filter setup, and C_D is a constant defining the efficiency of donor FI with the TIRF CFP filter setup. In our experiments, the CFP:YFP ratio control and the K_D were calculated as reported previously (18, 19, 21, 23) from the positive FRET control Daudi B cells expressing a nonzero FRET positive control Lyn16-CFP-YFP fusion protein as mentioned above:

$$K_{\text{D}}=(F-\beta D-\gamma A)/({DE}_{\text{bleaching}})-(F-\beta D-\gamma A)/D.$$

E_bleaching was calculated from Daudi B cells expressing Lyn16-CFP-YFP fusion protein as described above. In this report, FRET efficiency (Ea) was given as the mean ± SD or ΔFRET ± SD as specified. Images shown in the figures and videos were converted to 8-bit grayscale or RGB24 (for Ea-based FRET efficiency images) from the original 16-bit grayscale TIRF images.

The acquisition of Cy3- and Cy5- paired TIRF FRET images and the calculation of Cy3-Cy5 FRET efficiency were as previously reported (18, 19, 21, 23), and were similar to the CFP and YFP paired TIRF FRET system. A 514-nm argon gas laser and a 647-nm red krypton/argon gas laser were used to excite Cy3 and Cy5, respectively. Cy3 and FRET channel images were acquired simultaneously using a dual image splitter (MAGS Biosystems) equipped with a 565 dichroic beamsplitter and HQ550/30 (Cy3) and HHQ-LP665 (FRET) emission filters (Chroma Technology, Rockingham, VT). Cy5 channel images were acquired using the same dual image splitter with the HHQ-LP665 emission filter. Calculation of Cy3 and Cy5 FRET efficiency were similar to the CFP and YFP FRET system as described above. For Cy3 and Cy5 imaging, CH27 B cells labeled with 100 nM Cy3- or Cy5-conjugated Fab Abs specific for mouse IgM served as single Cy3- or Cy5-positive control and B cells labeled with a mixture of 100 nM Cy3- and Cy5-conjugated Fab Abs specific for mouse IgM served as FRET-positive controls to provide positive FRET efficiency as reported previously (18, 19, 21, 23).

Single particle tracking and analysis

For single-molecule TIRF imaging, B1-8 primary B cells were labeled with an appropriate amount of Cy3-Fab anti-IgM to provide resolution of a single BCR molecule without the need for further extensive photobleaching as reported (18). Labeled B cells were washed twice before incubation with ICs containing lipid bilayers, and images were taken using TIRFM with a 514-nm laser at a power of 5 mW, as measured at the objective lens in epifluorescence mode. Residual laser light was blocked using a 514-nm notch filter (Semrock, Rochester, NY) and a 550/40 ET BP emission filter was used for collection of Cy3 fluorescence. When imaging, a cropped region of ~100 × 100 pixels from intact 512 × 512 pixels CCD was used, and the exposure time was set at 35 ms. Streamline acquisition mode was used to capture single BCR molecules for 300 frames over 10 s.

Analysis of single-molecule tracking was as described in our previous study (18). Acquired images were background subtracted. Single-molecule tracking and analysis was performed using Matlab (The Mathworks) code based on available tracking algorithms (28, 29, 33). The resulting trajectories were visually inspected and occasional errors in tracking were manually corrected. The positions of the diffraction-limited spots in the trajectories were refined using two-dimensional Gaussian fit. MSD and instant diffusion coefficients for each BCR molecule trajectories (D₀, based on time intervals of 35–140 ms) were calculated from positional coordinates as reported (28, 29) and plotted as cumulative probability distribution graphs.

Results

In response to membrane associated ICs, the BCR and FcγRIIB colocalize in microscopic clusters

Live cell TIRFM was used to record the real-time responses of primary B cells to BCR crosslinking alone or to BCR-FcγRIIB colligation. Primary B cells from the spleens of IgH^B1-8/B1-8 Igκ^−/− transgenic mice expressing the B1-8 BCR specific for the hapten of NIP (34) were pretreated with CpG and LPS according to a standard laboratory protocol that allows for transfection of the cells when applicable before imaging. Cells were incubated with Cy5-conjugated Fab IgM-specific mAb (Cy5-Fab anti-IgM) to image the BCR and with Alexa 568-conjugated Fab FcγRIIB-specific mAb (Alexa 568 Fab anti-FcR) to image the FcγRIIB. To colligate the BCR and FcγRIIB, ICs formed by HIS₁₂-tagged NIP₁₄-BSA and intact rabbit-BSA-specific IgG (IgG-IC) were attached to nickel-containing lipid bilayers. To crosslink the BCR alone, ICs formed by HIS₁₂-tagged NIP₁₄-BSA and F(ab′)₂ of rabbit-BSA–specific IgG [F(ab′)₂–IC] were used. HIS₁₂–tagged ICAM-1 was also attached to nickel-containing lipid bilayers as described (18). Fluorescence recovery after photobleaching in TIRFM showed that IgG-IC and F(ab′)₂-IC were similarly mobile in the planar lipid bilayers (Supplemental Fig. 1A). Flow cytometric analyses indicated that soluble IgG-IC and F(ab′)₂-IC bound similarly to the B1-8 primary B cells (Supplemental Fig. 1B) and that the Alexa 568-Fab anti-FcR used to label the FcγRIIB did not block the binding of IgG-ICs to the FcγRIIB on B1-8 primary B cells (Supplemental Fig. 1C). Measuring calcium fluxes by flow cytometry showed that IgG-ICs inhibited the calcium responses of B1-8 primary B cells compared with F(ab′)₂-IC (Supplemental Fig. 1D).

We determined the influence of the BCR and FcγRIIB colligation on the spatial and temporal distribution of the BCR and FcγRIIB microclusters on the plasma membrane of B1-8 primary B. In B cells placed on lipid bilayers containing ICAM-1 but no Ag, the BCR and FcγRIIB overlapped by ~40% as quantified by Pearson’s correlation index (Fig. 1A, Supplemental Video 1A). Because neither BCR nor FcγRIIB is engaged in this case, the 40% degree of colocalization presumably represents the overlap in expression of the two receptors imposed by the membrane topology. The extent of colocalization of the BCR and FcγRIIB was increased to 60% in B cells in which the BCR was crosslinked alone by F(ab′)₂-ICs (Fig. 1A, Supplemental Video 1B). In B cells in which the BCR and FcγRIIB were colligated by IgG-ICs, the colocalization was nearly complete (Fig. 1A, 1B, Supplemental Video 1C). In these cells, the FcγRIIB and BCR appeared to colocalize from the earliest time points of contact with the IgG-IC–containing lipid bilayer (TIRF image at 2 s) and continued to colocalize as new FcγRIIB and BCR microclusters formed (Fig. 1A, Supplemental Video 1C). We did not observe this enhanced early colocalization in B cells placed on lipid bilayers containing F(ab′)₂-ICs as a control (Fig. 1A).

FIGURE 1.

In response to planar lipid bilayers containing IgG-IC, the BCR and FcγRIIB form microscopic clusters in which the BCR and FcγRIIB cytoplasmic domains are not in close molecular proximity. A, Merged TIRF images of B1-8 primary B cells labeled with Cy5-Fab anti-IgM (red) and Alexa 568-Fab anti-FcR (green) within the first 36 s after encountering planar lipid bilayers containing ICAM-1 plus F(ab′)₂-IC or IgG-IC, or ICAM-1 alone are given at the indicated time points, and shown in Supplemental Video 1A–C. Scale bar is 1.5 μm. B, The extent of colocalization of the BCR and FcγRIIB was compared based on the Pearson’s correlation index as described (45). The data represent the mean and SD of 6–8 cells in three independent experiments. Mann-Whitney U tests were performed for statistical comparisons. C, CH27 B cells expressing the FRET pair Igα-CFP and FcγRIIB-YFP or (D) CH27 cells labeled with a 1:1 mixture of 100 nM Cy3-Fab anti-IgM (FRET donor) and Cy5-Fab anti-IgM (FRET acceptor), or (E) CH27 cells labeled with a 1:1 mixture of 100 nM Cy3-Fab anti-FcγRIIB (FRET donor) and Cy5-Fab anti-IgM (FRET acceptor) were imaged by TIRFM after placing on lipid bilayers containing F(ab′)₂-IC or IgG-IC. Data in the upper panel of C–E are means and SD of FRET efficiency from 8 (C, D) or 11 (E) B cells acquired on either F(ab′)₂-IC– or IgG-IC–containing lipid bilayers in three experiments. Data in the lower panel of C–E represent the plot and linear fitting of the maximal FRET efficiency versus corresponding FRET donor to FRET acceptor ratio from all the imaged cells in C–E.

The presence of the majority of BCRs and FcγRIIBs in microscopic clusters in response to IgG-IC suggested that the BCR and FcγRIIB might be in close molecular proximity within the microclusters. We used fluorescence resonance energy transfer (FRET) as a molecular ruler to measure the distance between the cytoplasmic domains of the BCRs and FcγRIIBs in the microscopic clusters. To do so, PC-specific CH27 B cells expressing the BCR’s Igα-chain containing the FRET donor CFP in its C terminus (Igα-CFP) and FcγRIIB containing the FRET acceptor YFP in its C terminus (FcγRIIB-YFP) were imaged on lipid bilayers containing F(ab′)₂-IC or IgG-IC formed using HIS₁₂-tagged PC₁₆-BSA and either F(ab′)₂ or whole IgG BSA-specific Abs. No significant FRET was observed in either case throughout the experiment (Fig. 1C, top) and there was no significant correlation between FRET efficiency and CFP:YFP ratio (Fig. 1C, bottom). A concern in interpreting the failure to observe FRET is that there may be a physical constraint on the particular donor and acceptor construct analyzed that precludes their coming into close molecular proximity. To address this possibility, we generated cell lines in which the CFP and YFP were interchanged, Igα-YFP and FcγRIIB-CFP. However, we did not observe significant FRET upon BCR and FcγRIIB colligation (data not shown). We also changed the method by which the BCR and FcγRIIB were colligated using biotinylated F(ab′)₂ IgM-and FcγRIIB-specific Abs tethered to a streptavidin bilayer, but still failed to observe FRET (data not shown). Thus, although collectively these results are consistent with the failure of the cytoplasmic domains of the colligated BCR and FcγRIIB to come into close molecular proximity, we cannot rule out the possibility that physical constraints prevented FRET.

We also used TIRF FRET to measure the distances between the ectodomains of BCRs. BCRs were incubated with 100 nM of Cy3-conjugated (FRET donor) and 100 nM of Cy5-conjugated (FRET acceptor) Fab of polyclonal Abs specific for IgM (Cy3- or Cy5-Fab anti-IgM). The initial FRET was significant because of the close proximity of Cy3- and Cy5-Fab anti-IgM bound to the same mIgM (Fig. 1D, top). However, the FRET increased ~50% from this initial level during the experiment as previously reported (21), and the increase in FRET efficiency was similar for B cells placed on F(ab′)₂-IC– as compared with IgG-IC–containing lipid bilayers (Fig. 1D, top). This observation indicates that the PC₁₆-BSA as the Ag in the ICs clustered the ectodomains of BCRs into molecular proximity to a similar degree. In this case, the FRET efficiency was positively correlated with the Cy3:Cy5 ratio, indicating that the FRET increase may be caused by the change of the concentration of the BCRs in the microclusters (Fig. 1D, bottom). To measure the distance between the extracellular domains of the BCR and FcγRIIB, the CH27 B cells were labeled with equal 100-nM Fabs of Cy3-anti–FcR and Cy5-anti–IgM. A small but significant increase in FRET, independent of the Cy3:Cy5 ratio, was observed in the case of BCR and FcγRIIB colligation by IgG-IC compared with BCR crosslinking alone by F(ab′)₂-IC (Fig. 1E).

These results indicate that although the BCR and FcγRIIB colocalized in microscopic clusters on IgG-IC–containing bilayers and the ectodomains were in contact, the cytoplasmic domains may not come into close molecular proximity or at least no closer than 10 nm, the maximal distance over which FRET can occur.

FcγRIIB blocks BCR microclustering induced B cell spreading

The effect of BCR-FcγRIIB colligation on the ability of B cells to spread on the lipid bilayers in response to Ag as described by Batista et al. (12, 13) was determined. To do so, B cells were labeled with Cy3-conjugated Fab IgM-specific mAb (clone II/41), and the cells were imaged on F(ab′)₂-IC– or IgG-IC–containing lipid bilayers. Time lapse TIRFM showed that the first contacts of B cells with the IgG-IC– and F(ab′)₂-IC–containing lipid bilayers were very similar (Fig. 2A, Supplemental Video 2A, 2B). In both cases, B cells first touched the planar lipid bilayers through small protrusions in which BCR microclusters formed and grew with time. However, starting at 8 s, the behaviors of the B cells on the IgG-IC– and F(ab′)₂-IC–containing lipid bilayers were dramatically different. For B cells in which the BCR was crosslinked alone on F(ab′)₂-IC–containing lipid bilayers, the initial BCR microclusters grew quickly and after a few seconds a spreading response was triggered, as indicated by the broadening of the visible contact area of these B cells in the TIRF image (Fig. 2A; Supplemental Videos 2A, 3A). As the B cells rapidly spread over the lipid bilayers, new BCR-Ag microclusters continued to form at the periphery of the spreading cells. Even after the B cells had spread maximally, additional new microclusters continued to form in the ruffling region of the membrane. The cells then slowly contracted moving the BCR-Ag microcluster toward the B cell’s center, forming a central IS (Fig. 2A; Supplemental Videos 2A, 3A). Five minutes after the B cells’ initial encounter with the F(ab′)₂-IC–containing lipid bilayers, plasma membrane ruffling and the formation of new BCR microclusters subsided. During this process the number of BCRs in the contact area increased as measured by the fluorescence intensity in the contact area. The observed B cell spreading and contraction response to F(ab′)₂-IC–containing planar lipid bilayer was similar to B cell responses to planar lipid bilayers containing Ag alone as first reported by Batista et al. (12, 13) and recently in our own study (18).

FIGURE 2.

The FcγRIIB blocks B cell Ag-induced spreading. A, Time lapse TIRF images of B1-8 primary B cells labeled with Cy3-Fab anti-IgM after encountering either F(ab′)₂-IC– or IgG-IC–containing planar lipid bilayers are given at the indicated time points (Supplemental Video 2A,2B). Scalebar is 1.5 μm. The area of the B cell contact with the bilayer and the mean FI within the area of contact are plotted over time in the lower panel. B, Comparisons of parameters of the spreading response of the B1-8 B cells placed on F(ab′)₂-IC– or IgG-IC–containing lipid bilayers. The data represent the mean and SD of 29 cells on F(ab′)₂-IC– or 33 cells on IgG-IC–containing lipid bilayers in three independent experiments. Mann-Whitney U tests were performed for statistical comparisons. C, MSDs of the BCR microcluster trajectories in B cells placed on F(ab′)₂-IC–or IgG-IC–containing lipid bilayers. D, Percentage of BCR microclusters showing directed, random, confined, or complex movements in B cells on F(ab′)₂-IC– or IgG-IC–containing lipid bilayer determined as detailed in Materials and Methods. E, Two-dimensional trajectories of four typical individual BCR microclusters of B cells placed on F(ab′)₂-IC–containing lipid bilayers (blue trajectories) or IgG-IC–containing lipid bilayers (red trajectories). F, FI, mean ± SE of all BCR microcluster trajectories with tracking steps ≥10 is plotted over time for B cells placed on F(ab′)₂-IC–containing lipid bilayers (314 trajectories) or on IgG-IC–containing lipid bilayers (208 trajectories).

In sharp contrast to the B cell responses to BCR crosslinking alone, the B cell spreading on the IgG-IC–containing lipid bilayer was greatly reduced and the time required to reach the maximal spreading was significantly increased (Fig. 2A; Supplemental Videos 2B, 3B). The total number of BCR-Ag microclusters formed at the point of maximal spreading and the FI of the contact area at the point of maximal spreading were significantly decreased for B cells on IgG-IC–containing compared with F(ab′)₂-IC–containing bilayers (Fig. 2A, 2B). The observed decreases were presumably a consequence of the diminished spreading of B cells in which the BCR and FcγRIIB were colligated. Although the contact areas of the final IS of the B cells on IgG-IC– versus F(ab′)₂-ICs were similar, the total FI of the IS was significantly lower in B cells in which the BCR and FcγRIIB were colligated versus the BCR was crosslinked alone (Fig. 2B). This observation suggests that the FcγRIIB limited the number of BCR complexes that accumulated into the final IS.

Although these data were acquired from LPS-CpG preincubated B1-8 primary B cells by standard laboratory protocol, we confirmed that FcγRIIB colligated to BCR significantly inhibited B cell spreading and accumulation of BCR-Ag into the IS, using naive B1-8 primary B cells that were freshly isolated from B1-8 IgH^B1-8/B1-8 Igκ^−/− transgenic mice without LPS-CpG preincubation (Supplemental Fig. 1E–H). Similarly, the colocalization of the BCR and FcγRIIB was significantly enhanced in naive B1-8 primary B cells upon BCR and FcγRIIB colligation (data not shown).

Focusing on the first 40 s upon the B1-8 primary B cells’ encountering the planar lipid bilayers containing ICs, individual BCR microclusters were tracked from their formation in the points of initial contact with the lipid bilayers (Fig. 2C–F; Supplemental Video 4A, 4B). The MSD of BCR microclusters was much greater for the B cells placed on F(ab′)₂-IC–containing lipid bilayers as compared with the B cells placed on IgG-IC–containing lipid bilayers (Fig. 2C). We also characterized the movement of BCR microclusters as complex, directed, confined, or random by fitting the MSD plots of each BCR cluster with three functions to define directed, confined, and random movement as detailed in Materials and Methods. If a BCR microcluster showed R² values < 0.33 for each function, its behavior was considered to be complex. BCR microclusters that formed in response to F(ab′)₂-IC showed more directed movement toward the B cell synapse as compared with BCR microclusters formed in response to IgG-IC (Fig. 2D, 2E). Measurements of the average FI for all the individual BCR microclusters over time showed that the BCR microclusters formed in response to IgG-IC increased more slowly and reached much lower levels of maximal FI as compared with the BCR microclusters that formed in response to F(ab′)₂-IC (Fig. 2F). These analyses suggest that the FcγRIIB interfered with the directed mobility of BCR microclusters and the ability of BCR microclusters to grow in FI.

Colligation of the BCR and FcγRIIB blocks conformational changes in the cytoplasmic domains of the BCR microclusters that accompany BCR signaling

Our previous studies using FRET technology to measure the distances between the cytoplasmic domains of the BCR’s Ig, Igα, and Igβ showed that the initiation of signaling from Ag-clustered BCRs was accompanied by a conformational change in the cytoplasmic domains of the BCRs from a closed to an open form and the simultaneous phosphorylation of the BCRs by Lyn (19, 21). We determined the effect of colligating the BCR and FcγRIIB on the transition of ICs-engaged BCR microclusters from a closed to an open form. To do so, we used the previously well-characterized J558L B cell line expressing endogenous Igβ-chains and a transfected NIP-specific B1-8-γH chain containing a FRET donor CFP in its C terminus (B1-8-γ-CFP) and an Igα-chain containing a FRET acceptor YFP in its C terminus (Igα-YFP) (21). These cells do not express endogenous FcγRIIBs and were thus transfected with a construct containing the mouse FcγRIIB (Supplemental Fig. 2A). We refer to these B cells as γCαY-mo FcγRIIB. When placed on F(ab′)₂-IC–containing lipid bilayers to crosslink the BCRs, the γCαY-mo FcγRIIB cells show a FRET pattern similar to that previously described for γCαY cells responding to Ag-containing bilayers (Fig. 3A; Supplemental Fig. 2C, Supplemental Video 5A) (19). The FRET efficiency increased rapidly, indicating the clustering of the cytoplasmic domains of the Ag-engaged BCRs into a closed conformation. FRET efficiency reached a peak at 30 s and then decreased, indicating that the cytoplasmic domains had moved apart or opened with the initiation of BCR signaling (Fig. 3A; Supplemental Fig. 2C, Supplemental Video 5A). As reported earlier (19), FRET was detected primarily in the periphery of the spreading cell where new BCR clusters formed (Fig. 3A, Supplemental Video 5A). In contrast, when placed on IgG-IC–containing lipid bilayers to colligate the BCR and FcγRIIB, the γCαY-mo FcγRIIB cells showed only a slight gradual increase in FRET efficiency with time (Fig. 3B; Supplemental Fig. 2C, Supplemental Video 5B). This result indicates that although the BCRs form microscopic clusters and the BCRs’ ectodomains FRET as shown above (Fig. 1D), the cytoplasmic domains of the clustered BCRs do not come into close molecular proximity and FRET when the BCR and FcγRIIB are colligated.

FIGURE 3.

Colligation of the BCR and FcγRIIB blocks conformational changes within the cytoplasmic domains of the BCR. γCαY-mo FcγRIIB WT were placed on (A) lipid bilayers containing F(ab′)₂-IC or on (B) lipid bilayers containing IgG-IC. FRET efficiencies were calculated by the sensitized acceptor emission method as described in Materials and Methods. Upper panels, the B1-8-γ-CFP, Igα-YFP, and FRET efficiency images (pseudo-colored, with blue indicating low and red indicating high FRET efficiency) are given at the indicated time points and in Supplemental Video 5A, 5B. Lower panels, the FRET efficiency is plotted over time. A statistical comparison of the changes in FRET efficiencies (ΔFRET) for all imaged cells in A (16 cells) and B (13 cells) is given in Supplemental Fig. 2C. C, γCαY-mo FcγRIIB WT cells pretreated with 50 μM Src-family kinase inhibitor PP2 were imaged on IgG-IC–containing lipid bilayers and the data displayed as in A and in Supplemental Video 6A. D, mo FcγRIIB Y309F cells were imaged on IgG-IC–containing lipid bilayers and the data displayed as in A and in Supplemental Video 6B. A statistical comparison for B (13 cells) and D (17 cells) is given in Supplemental Fig. 2D. Hu FcγRIIB WT (E) or hu FcγRIIB I232T (F) were imaged on IgG-IC–containing lipid bilayers and the data displayed as in A and in Supplemental Video 7A, 7B. A statistical comparison for E (12 cells), and F (13 cells) is given in Supplemental Fig. 2E.

The ability of FcγRIIB to block the BCR cytoplasmic domains coming into close molecular proximity was dependent on the function of Src-family kinases. Pretreating the γCαY-mo FcγRIIB cells with PP2 resulted in an increase in FRET efficiency when the FcγRIIB were colligated with the BCR on IgG-IC–containing bilayers (Fig. 3C, Supplemental Video 6A). However, the dynamic loss of FRET efficiency was not observed following BCR-FcγRIIB colligation in PP2 treated cells. Rather, the cytoplasmic domains of the BCRs remained in a stable closed conformation (Fig. 3C, Supplemental Video 6A). We previously reported a similar FRET pattern for PP2-treated cells in response to BCR crosslinking alone, indicating that to achieve an open conformation, an Src-family kinase activity is required (21). Because PP2 affects the activation of both BCR and FcγRIIB, to determine the effect of the FcγRIIB activity alone on the Ag-induced change in the BCRs’ cytoplasmic domains, we analyzed γCαY cells expressing a mouse FcγRIIB loss of function mutation in which tyrosine 309 within the ITIM was replaced by phenylalanine (γCαY-mo FcγRIIB Y309F; Supplemental Fig. 2A) (35). When placed on IgG-IC–containing lipid bilayers these cells showed a FRET pattern similar to that of BCR crosslinking alone (Fig. 3D; Supplemental Fig. 2D, Supplemental Video 6B). Thus, in the absence of the ability of FcγRIIB to be phosphorylated on tyrosine 309 of its ITIM motif, the cytoplasmic domains of the BCR, although colligated to the FcγRIIB, still clustered in close proximity in a closed conformation and subsequently opened with the initiation of BCR signaling. We also analyzed another FcγRIIB mutant, FcγRIIB-CT314, in which the C-terminal 16 residues were deleted. These residues have been reported to cooperate with the ITIM motif of FcγRIIB to enhance its association with SHIP and to mediate FcγRIIB-dependent SHIP phosphorylation (36). Compared with the ITIM motif mutant, FcγRIIB-Y309F (the FcγRIIB-CT314 mutant) was much less impaired in its ability to block the increase in BCR FRET (Supplemental Fig. 2D; p = 0.0003). Thus, the ability of the FcγRIIB to block BCR activation does not appear to be highly dependent on its 16 C-terminal residues.

Earlier biochemical studies demonstrated that the loss of function trans-membrane polymorphism of FcγRIIB associated with systemic lupus erythematosus (SLE) in which isoleucine at position 232 was replaced by threonine (hu FcγRIIB I232T) correlated with its inability to perturb and associate with detergent-insoluble sphingolipid- and cholesterol-rich lipid raft microdomain upon BCR and FcγRIIB colligation (24, 37). γCαY cells expressing the wild type human FcγRIIB (γCαY-hu FcγRIIB WT; Supplemental Fig. 2B) showed a slight gradual increase in FRET efficiency when the BCR and FcγRIIB were colligated on IgG-IC–containing lipid bilayers (Fig. 3E; Supplemental Fig. 2E, Supplemental Video 7A). In contrast, colligating the BCR and FcγRIIB in γCαY B cells expressing hu FcγRIIB I232T (γCαY-hu FcγRIIB I232T) yielded a FRET pattern similar to those obtained from BCR crosslinking alone (Fig. 3F; Supplemental Fig. 2B, 2E, Supplemental Video 7B).

Collectively, these observations indicate that the ability of FcγRIIB to block the early clustering of the cytoplasmic domains of the BCRs into close molecular proximity in a ‘closed’ form and the subsequently ‘opening’ is dependent on the ability of FcγRIIB to be phosphorylated in its ITIM motif by Src family kinase and to associate with raft lipids.

BCR-FcγRIIB colligation leads to rapid association of FcγRIIB with a lipid raft probe and with Lyn tyrosine kinase

We recently reported using TIRF FRET microscopy that Ag-driven BCR microclusters transiently associate with a lipid raft probe that leads to the more stable association with the raft lipid-tethered Lyn kinase (19). The observation that the loss of function hu FcγRIIB-I232T that cannot associate with raft lipids did not block BCR clustering, suggested that the FcγRIIB may interact with raft lipids to block early BCR clustering events. To directly measure the real-time association of the FcγRIIB with raft lipids, CH27 B cell lines expressing the lipid raft probe, Lyn16-CFP, which contains the first 16 aa residues of Lyn and is both myristoylated and palmitoylated and targeted to lipid raft microdomains (19, 22, 23, 31), were transfected with mouse FcγRIIB-YFP (Supplemental Fig. 3A). Time lapse TIRF imaging showed no significant FRET efficiency between the lipid raft probe, Lyn16-CFP, and FcγRIIB-YFP when the BCRs were cross-linked alone on F(ab′)₂-IC–containing bilayers (Fig. 4A, 4C; Supplemental Video 8A). In contrast, when the BCR and FcγRIIB were colligated on IgG-IC–containing bilayers, FRET efficiency was observed between the FcγRIIB and the lipid raft probe, and increased rapidly for the first 2 min before reaching a plateau that was sustained over the time course of the experiment (Fig. 4B, 4C; Supplemental Fig. 3B, Supplemental Video 8B). A plot of the maximal FRET efficiency versus the corresponding CFP:YFP ratio from all cells imaged did not show an obvious correlation (Supplemental Fig. 3C). In addition, FRET efficiency was specific and was not observed between FcγRIIB-YFP and a nonraft lipid control probe, CFP-Ger (19, 22, 23), following BCR- FcγRIIB colligation (Fig. 4C). The association of Ag-driven BCR clusters with the lipid raft probe in response to membrane-bound Ags was previously shown to be independent of the activity of Src-family kinases and occurred in PP2-treated cells. Similarly, the association of the FcγRIIB-YFP with the lipid raft probe Lyn16-CFP was not dependent on FcγRIIB signaling as FRET efficiency was observed following BCR-FcγRIIB colligation in PP2 pretreated cells (Fig. 4C). The stable interaction of the FcγRIIB with the lipid raft probe is in striking contrast to the highly transient association of Ag-induced BCR clusters with the lipid raft probe as reported previously (19).

FIGURE 4.

Colligation of the BCR and FcγRIIB leads to rapid and sustained association of the FcγRIIB with a lipid raft probe. CH27 B cells expressing Lyn16-CFP and FcγRIIB-YFP were placed on lipid bilayers containing either (A) F(ab′)₂-IC or (B) IgG-IC. Data are displayed as in Fig. 3A. C, The mean and SD of the maximal increase of FRET efficiency (ΔFRET) between 0 and 180 s are given from 8–14 cells in three independent experiments, for (left to right) cells expressing: Lyn16-CFP and FcγRIIB-YFP placed on F(ab′)₂-IC–containing lipid bilayers; or Lyn16-CFP and FcγRIIB-YFP; the nonraft lipid probe CFP-Ger and FcγRIIB-YFP or Lyn16-CFP and FcγRIIB-YFP in B cells pretreated with the Src-family kinase inhibitor PP2, each placed on IgG-IC–containing lipid bilayers. Mann-Whitney U tests were performed for statistical comparisons.

FRET was also measured between hu FcγRIIB-YFP and Lyn16-CFP expressed in the human FcγRIIB-deficient B cell line ST486 (24). Following colligation of the human FcγRIIB and the BCR on biotin-containing lipid bilayers in which biotinylated mouse mAb specific for human FcγRIIB (clone no. AT10) and biotinylated F(ab′)₂ rabbit Abs specific for human IgM were attached through streptavidin, a rise in FRET efficiency was observed that was sustained through the time course of the experiment (Supplemental Fig. 4A, Supplemental Video 9A). In contrast, a construct that contained the loss of function hu FcγRIIB I232T and YFP (hu FcγRIIB-I232T YFP) (25, 26), did not stably FRET with the lipid raft probe Lyn16-CFP as indicated in Supplemental Video 9B following the colligation with BCRs, and showed no increase in FRET efficiency throughout the time course (Supplemental Fig. 4B, Supplemental Video 9B). This result was consistent with earlier biochemical analyses demonstrating that the loss of function mutant hu FcγRIIB-I232T was not able to associate with detergent insoluble lipid rafts microdomains after the colligation with BCRs (24, 37).

The association of the BCR-colligated FcγRIIB with the lipid raft probe suggested that following colligation the FcγRIIB might also associate with the lipid-raft tethered full length Lyn kinase as we previously reported for BCR following Ag binding (19). To determine whether this was the case, the FcγRIIB-deficient mouse B cell line, A20II1.6, expressing full length Lyn kinse, LynFL-CFP, and FcγRIIB-YFP was imaged on lipid bilayers crosslinking BCR alone, no FRET change was observed throughout the time course (Supplemental Fig. 5A, 5D). In contrast, when imaging on lipid bilayer inducing BCR and FcγRIIB colligation, the cells showed a rapid rise in FRET efficiency that reached a peak between 30 and 50 s, and then declined to base line level (Supplemental Fig. 5B, 5D). Additionally the loss of function mutant at ITIM motif of mo FcγRIIB-Y309F YFP did not show FRET with LynFL-CFP following the colligation (Supplemental Fig. 5C, 5D). Taken together these results indicate that when colligated to the BCR, FcγRIIB blocks the very early events in BCR activation as it simultaneously stably associates with raft lipids.

FcγRIIB blocks the formation of submicroscopic immobile BCR oligomers

We recently provided evidence using single molecule imaging techniques that within microscopic BCR clusters signaling-active immobile BCR oligomers were formed (18). Formation of the immobile oligomers was a BCR intrinsic property and did not require a signaling competent BCR. In this study, we used the same single molecule imaging techniques to determine the effect of colligating the BCR and FcγRIIB on the formation of immobile BCR oligomers. Using single molecule TIRF imaging, we analyzed and compared the diffusion coefficients from >1000 individual BCR trajectories from B1-8 primary B cells following BCR crosslinking alone or BCR and FcγRIIB colligation induced on planar lipid bilayers. The data are given in cumulative probability distribution plots of the diffusion coefficients of single BCR molecules. The median diffusion coefficients can be calculated from the data and setting a cutoff diffusion coefficient of 0.01μm²/s as immobile, the fraction of immobile BCRs can be determined. In B cells placed on lipid bilayers that contained no Ag, the majority of BCRs were mobile with a median diffusion coefficient of 0.1 μm²/s (Fig. 5A, Supplemental Video 10A), and ~16% of the total BCRs were immobile (Fig. 5A) consistent with our previous report (18). The mobility of BCRs in B cells placed on F(ab′)₂-IC–containing lipid bilayers to crosslink the BCR alone were significantly slower with a median diffusion coefficient of 0.02μm²/s (Fig. 5A, Supplemental Video 10B) and the fraction of immobile BCR oligomers increased to 36% (Fig. 5A). The diffusion coefficient of BCRs in B cells placed on IgG-IC–containing lipid bilayers to colligate BCR and FcγRIIB were significantly faster as compared with BCR crosslinking alone with a median diffusion coefficient of 0.03 (Fig. 5A, Supplemental Video 10C) and the fraction of immobile BCR oligomers fell to 24% (Fig. 5A). Pretreating the B cells with PP2 did not significantly change the median diffusion coefficient or the fraction of BCRs that were immobilized (38%) in cells in which the BCR was crosslinked alone (Fig. 5B), consistent with our published results (18). Interestingly, PP2 treatment of B cells resulted in a slight change in the shape of the cumulative probability distribution plots suggesting that blocking Src kinase induced a small variation in the distribution of diffusion coefficients of individual BCRs. However, PP2 pre-treatment of the B cells placed on IgG-IC–containing lipid bilayers to colligate BCR and FcγRIIB resulted in the immobilization of BCRs to a similar degree (38%) as compared with BCR crosslinking alone. Thus, the ability of the FcγRIIB to block the formation of submicroscopic immobile BCR oligomers when colligated with BCRs was dependent on FcγRIIB signaling (Fig. 5B).

FIGURE 5.

FcγRIIB blocks the Ag-driven oligomerization of the BCRs. A, Cumulative probability distribution of the instant diffusion coefficient of individual BCR molecules of B1-8 primary B cells labeled with Cy3-Fab IgM-specific mAb and placed on lipid bilayers containing: ICAM-1 alone (2195 trajectories), ICAM-1 plus F(ab′)₂-IC (1720 trajectories), or IgG-IC (1299 trajectories). The acquisition and analysis of single-molecule TIRF images are described in detail in Materials and Methods, and TIRF videos showing the time lapse diffusion mobility of BCR molecules in these three cases are shown in Supplemental Video 10A–C, respectively. Instant diffusion coefficients for each of the BCR molecule trajectories are plotted in the cumulative probability distribution graphs. Data represent 1299–2195 BCR molecules from three independent experiments. Significant differences by Kolmogorov-Smirnov tests are indicated (***). The p value for the plots of F(ab′)₂-IC and IgG-IC was 2.85E-13. B, B1-8 primary B cells were pretreated with the Src-family kinase inhibitor PP2 before placing them on F(ab′)₂-IC–containing lipid bilayers (2408 trajactories), on IgG-IC–containing lipid bilayers (3014 trajectories) or on lipid bilayers containing ICAM-1 alone (1776 trajectories). Significant differences by Kolmogorov-Smirnov tests are indicated (***). At the bottom panel, the percentage of immobile and mobile fractions of BCR molecules at different experimental conditions was shown schematically with a cutoff diffusion coefficient of 0.01 μm²/s, as described in Materials and Methods.

Discussion

The advent of new live cell imaging techniques has provided both the temporal and spatial resolution over the time and length scales that are critical to decipher the earliest events in B cell activation. Although our understanding of these events is still incomplete, recent results from high resolution live cell imaging studies are providing the first glimpse into the process by which B cell signaling is initiated (9, 10, 38). Here we investigated the effect of colligating the FcγRIIB to the BCR through ICs on these early events, by imaging both the BCR and FcγRIIB during the first stages of B cell encounter of ICs incorporated into a planar lipid bilayer to mimic an APC surface. We conclude that the FcγRIIB blocks the earliest events in BCR activation including the formation of signaling active, immobile BCR oligomers and the transition of the cytoplasmic domains from a ‘closed’ to a signaling active ‘open’ conformation. The FcγRIIB accomplished this as it stably associated with raft lipids and transiently associated with Lyn, events that were necessary to exert its inhibitory function.

We observed that when colligated by ICs the FcγRIIB and the BCR colocalized in microscopic clusters. However, using FRET as a molecular ruler to measure the distances between BCRs and FcγRIIB ectodomains and cytoplasmic domains within clusters we observed that although the ectodomains appeared to interact weakly, the cytoplasmic domains were not in close molecular proximity and did not FRET. The inability of the colligated BCRs and FcγRIIBs to stably FRET may be, in part, a reflection of the size of IgG-IC formed by the BSA and BSA-specific Abs, which were estimated to be ~19–20 nm by light scattering and electron microscopy in early studies (39–43) and could potentially hold the BCRs and FcγRIIBs apart. It is also possible that the particular FRET pairs used are constrained in some way to prevent them from stably interacting. The observation that the BCRs and FcγRIIBs do not directly physically interact is important, because it rules out physical obstruction of BCR oligomerization as the major reason for FcγRIIB inhibition of B cell signaling.

We also observed that the spreading response of B cells over the Ag-containing bilayer triggered by signaling from BCRs in the initial contacts of the B cell with the bilayer was severely impaired by FcγRIIB colligation. Batista and colleagues earlier provided evidence that spreading was a mechanism by which B cells increased the amount of Ag engaged and increased the sensitivity of the B cell to the affinity of the Ag (13). By limiting the ability of the BCR to spread, the FcγRIIB indirectly reduced the amount of Ag engaged by the BCR and the magnitude of BCR signaling. We also observed that fluorescence intensity of BCR microclusters initiated by IgG-IC binding increased more slowly as they moved toward the center of the contact area and reached much smaller maximal intensity as compared with microclusters formed by F(ab′)₂-ICs. This result indicates that the FcγRIIB interfered with the ability of BCR microclusters to recruit new BCRs as they moved toward the center of the contact area, and thus significantly altered the quality of these microclusters.

Using a combination of FRET and TIRFM we interrogated the BCRs within the microscopic clusters that formed in response to ICs to determine if they were undergoing conformational changes in their cytoplasmic domains that we previously described accompanied the initiation of signaling (19, 21). We showed that when colligated to the FcγRIIB by ICs, the cytoplasmic domains of the BCRs do not come into close molecular proximity in microscopic clusters and fail to undergo a conformation change in their cytoplasmic domains from a closed to a signaling active open form. Our recent studies using single molecule tracking in TIRFM provided evidence that the initiation of BCR signaling also involves a conformational change in the Cμ4 domain of the BCR’s mIg that lead to the oligomerization of the BCRs into signaling active complexes (18). Using the same single molecule techniques we provide evidence here that when colligated to the FcγRIIB, BCRs showed impaired ability to form immobile, signaling-active BCR oligomers. These results provide evidence that the FcγRIIB is able to block the earliest events in BCR activation including BCR-intrinsic Ag-induced oligomerization.

If the FcγRIIB does not physically obstruct the BCRs from oligomerization, as indicated by the lack of FRET efficiency between the colligated BCRs and FcγRIIBs, how might the FcγRIIB exert its inhibitory effect on early events in BCR activation? One clue comes from the behavior of an FcγRIIB that carries a systemic lupus erythematosus-associated loss of function mutation in its trans-membrane domain, hu FcγRIIB I232T (24, 37). Hu FcγRIIB I232T was shown by biochemical techniques to be unable to inhibit BCR activation because the trans-membrane mutation excluded it from cholesterol-rich and sphingolipid-rich raft membrane microdomains. We recently showed that Ag-induced BCRs microcluster rapidly and transiently perturbed the local lipid environment resulting in the association of BCRs with a lipid raft probe (19, 20). This transient membrane perturbation was followed by a more stable association of the BCR microclusters with Lyn. The association of the BCR with raft lipids spatially and temporally correlated with the change in the cytoplasmic domains of the BCRs from closed to open forms. Based on these observations, we proposed that the perturbation of the membrane and the accumulation of raft lipids around the BCR oligomers as they form may exert a force on the transmembrane domains of the BCRs through the thickening and curvature of lipid raft membranes, opening the cytoplasmic domains (9). We previously showed that the colligation of the FcγRIIB with the BCR blocked the association of BCRs with raft lipids (20). In this study, we show that the FcγRIIB itself, when colligated to the BCR, becomes stably associated with raft lipids, and the loss of function hu FcγRIIB I232T that is unable to associate with raft lipids cannot block the Ag-driven early changes in the BCR. Collectively, these results suggest that the FcγRIIB, in stably associating with raft lipids, perturbs the membrane in such a way that it destabilizes the formation of BCR oligomers.

The results presented in this study point to a role of the FcγRIIB in blocking BCR-intrinsic events in the activation of the BCRs that precede and are independent of the phosphorylation of the BCR by Lyn and the subsequent recruitment of PI3K. However, the ability of the FcγRIIB to block these events required a functional FcγRIIB that contained an intact ITIM and presumably could recruit SHIP. If, in responding to ICs, the FcγRIIB blocks BCR activation at an early point that precedes BCR signaling, what role does the well-established recruitment of SHIP by the FcγRIIB play in blocking BCR activation? Cambier et al. recently showed that engagement of the BCR and FcγRIIB by ICs lead to the formation of what the authors referred to as mobile PIP3 scavenger complexes composed of SHIP and Dok-1 (3). The authors suggested that mobile PIP3 scavenger complexes localized to the membrane through the PH domain of Dok-1 would be free to hydrolyze the PIP3 products of other activating receptors (3). We speculate that FcγRIIB may serve to block the early BCR-intrinsic events in BCR activation while, at the same time, blocking signaling from other activating receptors and any BCRs that escape the FcγRIIB block in BCR oligomerization.

The observation of a novel mechanism by which the FcγRIIB blocks BCR activation reported in this study was made in B cell-engaging, membrane-associated ICs. Studying the B cell engagement of ICs on membranes in vitro would seem to be highly relevant to the in vivo situation given the number of studies that report B cells engaging Ag in lymph nodes on the surfaces of APCs (44–47). It is still too early to know how similar B cell activation to membrane-associated Ags is to Ags in solution. Important differences have already been observed between soluble and membrane associated Ags in the requirement for CD19 (12) and for multivalent Ags to physically crosslink the BCR (18). However, it is still unclear how the two activation modes relate to one another, as is the mechanism of FcγRIIB inhibition of B cell activation in the two modes of Ag engagement.

The results presented in this study provide evidence for a novel mechanism by which the FcγRIIB blocks BCR signaling, namely by preventing the earliest events in Ag-driven BCR activation, including the formation of signaling active immobile oligomers. Such a mechanism may complement the well-established ability of the FcγRIIB to block the downstream BCR signaling through the action of the lipid phosphatase SHIP on the generation of PIP3 by PI3K.

Abbreviations used in this paper

CFP

cyan fluorescent protein

DOPC

1,2-dioleoyl-sn-glycero-3-phosphocholine

Ea

fluorescence resonance energy transfer efficiency normalized with yellow fluorescent protein intensity

FI

fluorescence intensity

FRET

fluorescence resonance energy transfer

IC

immune complex

IS

immune synapse

MSCV

murine stem cell virus

MSD

mean square displacement

NIP

4-hydroxy-5-iodo-3-nitrophenyl acetyl

PC

phosphorylcholine

TIRF

total internal reflection fluorescence

TIRFM

TIRF microscopy

YFP

yellow fluorescent protein