Research ArticleB CELLS

Intrinsic properties of human germinal center B cells set antigen affinity thresholds

See allHide authors and affiliations

Science Immunology  30 Nov 2018:
Vol. 3, Issue 29, eaau6598
DOI: 10.1126/sciimmunol.aau6598

Optimal affinity

Germinal center (GC) B cells are essential to generating protective antibody responses and are selected through a process of affinity maturation. Kwak et al. now define intrinsic properties of human GC B cells that are critical to antigen affinity discrimination. They identified BCR-containing actin-rich pod-like structures that facilitated formation of highly stable synapses and antigen internalization but only when GC B cells engaged high-affinity antigens. These findings reveal the importance of these structures in setting thresholds for affinity selection and driving GC B cell responses.

Abstract

Protective antibody responses to vaccination or infection depend on affinity maturation, a process by which high-affinity germinal center (GC) B cells are selected on the basis of their ability to bind, gather, and present antigen to T follicular helper (Tfh) cells. Here, we show that human GC B cells have intrinsically higher-affinity thresholds for both B cell antigen receptor (BCR) signaling and antigen gathering as compared with naïve B cells and that these functions are mediated by distinct cellular structures and pathways that ultimately lead to antigen affinity– and Tfh cell–dependent differentiation to plasma cells. GC B cells bound antigen through highly dynamic, actin- and ezrin-rich pod-like structures that concentrated BCRs. The behavior of these structures was dictated by the intrinsic antigen affinity thresholds of GC B cells. Low-affinity antigens triggered continuous engagement and disengagement of membrane-associated antigens, whereas high-affinity antigens induced stable synapse formation. The pod-like structures also mediated affinity-dependent antigen internalization by unconventional pathways distinct from those of naïve B cells. Thus, intrinsic properties of human GC B cells set thresholds for affinity selection.

INTRODUCTION

A hallmark of immunological memory is the affinity maturation of antibody responses (1, 2). Underlying affinity maturation is the stochastic process of somatic hypermutation (SHM) and the subsequent competitive selection of B cells that have acquired affinity-enhancing mutations for the immunizing antigen or invading pathogen (25). The processes of SHM and affinity-based selection occur in spatially distinct, specialized microenvironments within B cell follicles of secondary lymphoid organs termed germinal center (GC) dark zones (DZs) and light zones (LZs), respectively (57). Naïve B cells first encounter antigens in B cell follicles on the surfaces of subcapsular sinus macrophages, dendritic cells (DCs), or follicular dendritic cells (FDCs) (8, 9), triggering B cell antigen receptor (BCR) signaling and antigen extraction, internalization, processing, and presentation of the antigen on major histocompatibility complex (MHC) class II molecules (4, 10). B cells then migrate to the border of the follicles with the T cell zone, where they present antigen to T cells that have been recently primed by antigen presented on DCs to differentiate into T follicular helper (Tfh) cells (11, 12). The resulting B cell–Tfh cell interaction has the potential to drive B cells to several fates, including differentiation to GC cells (1316). GC B cells first enter GC DZs, where they proliferate and undergo SHM before entering the GC LZs where antigen affinity–based selection occurs (1719). Current evidence indicates that GC LZ selection is a competitive process dependent, in large part, on the amount of antigen B cells that are able to gather, process, and present to Tfh cells (20). LZ GC B cells have several potential fates that depend on their interactions with Tfh cells including apoptosis, positive selections for reentry into the DZ for further proliferative expansion, and SHM or differentiation into memory B cells (MBCs) or long-lived plasma cells (PCs) (6).

Thus, there appear to be at least two key checkpoints in the process of affinity maturation: one for naïve B cells and one for LZ GC B cells. At these checkpoints, the affinity of B cells for antigen is tested by the ability of the BCR to differentially signal in response to and internalize, process, and present antigen to T cells. However, despite the central role of these checkpoints in affinity selection and maturation, we have a limited understanding of the potential of human naïve or GC B cells at these checkpoints to discriminate antigen affinity or of the cellular and molecular mechanisms that facilitate affinity discrimination by naïve and GC B cells. Here, we describe intrinsic properties of human GC B cells that set thresholds for affinity selection, providing a mechanistic framework for antigen affinity discrimination by GC B cells.

RESULTS

GC B cells engage antigen through BCRs concentrated in unconventional F-actin and ezrin-rich pod-like structures

To understand how human naïve and GC B cells engage membrane-associated antigens, we obtained differential interference contrast (DIC) and interference reflection microscopy (IRM) images of live B cells isolated from human tonsils and placed on antigen-containing membranes (Fig. 1A and movie S1). B cell subpopulations were purified from single-cell suspensions of human tonsil B cells obtained by negative selection and sorted on the basis of expression of CD10, immunoglobulin D (IgD), and CXCR4 into the four B cell subpopulations (fig. S1): naïve B cells (IgD+ CD10), MBCs (IgD CD10), LZ GC B cells (IgD CD10+ CXCR4Lo), and DZ GC B cells (IgD CD10+ CXCR4Hi). To further characterize these purified subpopulations, we quantified the cell surface expression of the BCR by flow cytometry using fluorescently conjugated antibodies specific for either the λ or κ light chains. Naïve B cells expressed the highest level of BCR followed by MBCs, LZ GC B cells, and DZ GC B cells (fig. S2A). The expression of IgM, IgG, and IgA isotypes was also lower on GC B cells (fig. S2B). As a surrogate membrane-associated antigen, we used streptavidin-conjugated F(ab′)2 light chain–specific antibodies (anti-λ/κ) attached to biotinylated lipids incorporated into fluid planar lipid bilayers (PLBs). When placed on antigen-containing PLB, naïve B cells appeared to form large flat, stable contact areas with the antigen-containing PLB and maintain active ruffling around the cell’s periphery, consistent with conventional immune synapses (Fig. 1A and movie S1). The interaction of the naïve B cell with the bilayer was quantified over time and shown as a kymograph (Fig. 1B). The behavior of the LZ GC B cells on the antigen bilayer was remarkably different. Rather than forming flat stable contact areas, the cells touched the bilayer through highly dynamic, large pod-like structures (Fig. 1A and movie S1). This behavior is displayed as a kymograph, showing the dynamic movement of the cell over time (Fig. 1B). In the absence of antigen, the LZ GC B cell surfaces were smooth and the cells made contact with the bilayer through a small, stable area, as evident in the IRM images (Fig. 1A and movie S1). We extended the IRM analyses to include a comparison of DZ GC B cells in addition to naïve and LZ GC B cells. Naïve B cells rapidly responded to the antigen-containing PLB, forming flat contact areas in uniform close proximity to the membrane that were stable over 20 min (Fig. 1C and movie S2). In contrast, both DZ and LZ GC B cells made small, transient, highly dynamic contacts with the antigen-containing PLB for several minutes before the contacts stabilized but never reached the same high-intensity, uniform contact areas observed for naïve B cells. Quantification of the IRM images by several methods, including the intensity of the image from the center of the contact area to the edge (fig. S2C), the percent change in intensity from frame to frame (fig. S2D), and the number of contact sites over time (fig. S2E), supported the conclusion that naïve B cells rapidly formed uniform contacts with the antigen-containing PLB, in contrast to GC B cells that contacted the PLB in highly dynamic small points over several minutes.

Fig. 1 GC B cells engage antigen through BCRs concentrated in F-actin and ezrin-rich pod-like structures.

(A) DIC, IRM, and merged images of immune synapses of live naïve B cells placed on antigen-containing PLB and live LZ GC B cells placed on antigen-containing PLB or PLB with no antigen. (B) Kymographs of DIC images of naïve and LZ GC B cell immune synapses on antigen-containing PLB. (C) Membrane movement in the immune synapses of live naïve, LZ GC, and DZ GC B cells with time on antigen-containing PLB imaged by IRM. (D) STED super-resolution images of F-actin formed in immune synapses of naïve B cells and GC B cells placed on PLB that contained antigen. (E) Colocalization of F-actin and BCR in immune synapses of naïve and LZ GC B cells imaged by confocal microscopy on PLB that contained antigen. (F) Left: Confocal microscopy images of immune synapses of naïve B cells and LZ GC B cells on antigen-coated PLB stained with Alexa Fluor 488 phalloidin for F-actin (green) and antibodies specific for ezrin (red). Quantification of the MFI of ezrin (top right panel) and colocalization of ezrin with F-actin (bottom right panel) in the immune synapse of confocal images. (G) Bottom and orthogonal views of F-actin (green) and BCR (red) in naïve and LZ GC B cells imaged by confocal microscopy on antigen-containing PLB. (H) Side and top views of naïve and LZ GC B cells imaged by scanning electron microscopy on PLB without or with antigen. (I) Colocalization of pod-like structures and antigens in immune synapse of LZ GC B cells imaged by TIRFM on antigen-containing PLB. Scale bars, 5 μm. **P ≤ 0.01 and ****P ≤ 0.0001 (unpaired t test). Data are representative of two experiments [mean and SEM (F)].

To determine a role of actin in antigen contact, we used super-resolution stimulated emission depletion (STED) microscopy to obtain nanoscale images of the organization of F-actin in naïve and LZ GC B cells placed on antigen-containing PLBs for 25 min, fixed, permeabilized, and stained with Alexa Fluor 488–conjugated phalloidin (Fig. 1D). The high-resolution images showed that naïve B cells formed conventional mesh-like F-actin structures in the contact area with weak actin signal intensity. In contrast, LZ GC B cells showed a high-intensity punctate pattern of F-actin–rich structures primarily in the cell’s periphery. Colocalization of the BCR and F-actin in the area of contact indicated that, in contrast to naïve B cells, for which BCR clusters were not directly associated with polymerized F-actin (Fig. 1E), BCRs were associated with actin-rich structures in LZ GC B cells (Fig. 1E).

We also determined the amount of ezrin concentrated in the contact area of the B cells with the antigen-containing PLB and the degree of colocalization of ezrin with F-actin by confocal microscopy. Ezrin is a cytoplasmic peripheral membrane protein that is enriched in actin-rich cell surface projections (2123) and regulates morphological and cytoskeletal changes in B cells (24). By flow cytometry, ezrin expression in GC B cells was about 10-fold higher than that in naïve B cells (fig. S2F). In naïve B cells, ezrin was primarily in the cell periphery where ezrin and F-actin were only partially colocalized (Fig. 1F). In contrast, in LZ GC B cells, ezrin was concentrated in a punctate pattern and was more highly colocalized with F-actin (Fig. 1F), suggesting that the GC B cell pod-like structures were both actin and ezrin rich.

Confocal z-stack three-dimensional (3D) images of cross sections of naïve and LZ GC B cells placed on antigen-containing PLBs revealed remarkable details of cellular architecture in the contact area that was consistent with the DIC, IRM, STED, and confocal images (Fig. 1G). Naïve B cells formed contact with the antigen-containing PLB in a single plane in close opposition to the PLB, resulting in a relatively uniform distribution of BCRs under a plane of F-actin. In contrast, LZ GC B cells contacted the antigen-containing PLB through actin-rich pod-like structures, the tips of which concentrated BCRs. The dimensions of these structures were about 0.2 to 0.5 μm in diameter by 1.5 μm in length, much larger than microvilli through which T cells have been recently demonstrated to search for antigen (25). These structures were also evident by scanning electron microscopy (Fig. 1H). By scanning electron microscopy, in the resting state, LZ GC B cells and naïve B cells showed distinct morphologies with LZ GC B cells having a corrugated membrane topology with long, thin protrusions, whereas naïve B cells had smoother membranes with short, thick bulges. More pronounced differences were observed between naïve B cells and LZ GC B cells that were activated on antigen-coated PLBs. Naïve B cells showed a flat area of contact with the bilayer visible by scanning electron microscopy in both top and side views, in contrast to LZ GC cells in which pod-like contacts were visible in the side view.

The observation that BCRs were concentrated in the area of contact of the pod-like structures with the antigen-containing PLB predicted that antigen gathered from the PLB would also be concentrated in the contact areas of the pod-like structures. LZ GC B cells were placed on PLB-containing fluorescently labeled anti-λ/κ for 20 min and imaged. The DIC and antigen merged image showed antigen concentrated at the ends of the pod-like structures (Fig. 1I).

Human GC B cells signal in response to and exert pulling forces on membrane-bound antigens through pod-like structures

BCRs were labeled with Alexa Fluor 488 Fab or Alexa Fluor 647 Fab anti-IgM for naïve B cells or Alexa Fluor 488 Fab or Alexa Fluor 647 Fab anti-IgG for GC B cells, and B cells were placed on antigen-containing PLBs for 7 or 25 min. Cells were fixed, permeabilized, and stained with fluorescently conjugated antibodies specific for the phosphorylated forms of kinases and phosphatases activated by the initiation of BCR signaling, including PI3K (phosphatidylinositol 3-kinase) and SHP-1 (Fig. 2A) and Syk, PLCγ2, Btk, Vav, and Cbl (fig. S3, A and B). Naïve B cells spread on the antigen-containing PLB over 25 min, and pPI3K and pSHP-1 accumulated with BCRs in the contact area, forming a central cSMAC structure (Fig. 2A). In contrast, GC B cells did not form centralized BCR-pPI3K and BCR–pSHP-1 clusters, but rather, pPI3K and pSHP-1 were concentrated with BCRs in a punctate pattern primarily in the cell periphery, consistent with BCR signaling through pod-like structures. Quantification of images showed that pPI3K and pSHP-1 were recruited earlier to naïve BCRs as compared with GC B cells but were more highly colocalized with the BCRs in GC B cells as compared with naïve B cells both in resting cells and at early and late phases of the response (Fig. 2B). Similarly, the colocalization of pSyk, pPLCγ2, pVav, pBtk, pCbl, and CD19 with the BCR was consistently higher in GC cells than in naïve B cells (fig. S3, B and C, and movie S3) but similar between IgM and IgG GC B cells (fig. S3D). Time-lapse live-cell total internal reflection fluorescence microscopy (TIRFM) imaging of the growth of individual BCR microclusters showed that microclusters grew rapidly on naïve B cells, whereas growth of GC B cell BCR microclusters was slower (fig. S3E).

Fig. 2 Human GC B cells signal in response to and exert pulling forces on membrane-bound antigens through pod-like structures.

(A) Immune synapses of naïve B cells and GC B cells imaged by TIRFM on PLB containing F(ab′)2 anti-λ/κ. (B) MFI of pPI3K and pSHP-1 and colocalization of BCR with pPI3K or pSHP-1 in the contact area of naïve and GC B cells placed on antigen (Ag)–containing PLBs for 7 or 25 min. (C) Top: Atto 647N FI (red) and Atto 550 FI (green) merged images. Bottom: Sensor opening ratio (the ratio of Atto 647N FI to Atto 550 FI). (D) Quantification of sensor opening for 9-pN sensor and control sensor. (E) Colocalization of sensor opening locations and contact sites to the membrane with time imaged by TIRFM and IRM. Scale bars, 5 μm. *P ≤ 0.05 and ****P ≤ 0.0001 (unpaired t test). Data are representative of two experiments [mean and SEM (B and D)].

We also compared the ability of LZ and DZ GC B cells and naïve B cells to exert a pulling force on membrane antigen. To do so, we used F(ab′)2 anti-κ/λ as a surrogate antigen conjugated to a 9-pN DNA-based force sensor attached to PLBs (fig. S4A). The force sensor contained a central DNA hairpin that unwinds when pulling forces are exerted by the BCR on the antigen, displacing an Atto 647N dye from a dark quencher at the base of the hairpin, resulting in increased Atto 647N fluorescence over the fluorescence of a force-insensitive dye Atto 550 attached to the double-stranded DNA at the top of the sensor. Opening is expressed as the ratio of the signal intensity of Atto 647N to that of Atto 550. Control sensors lack the hairpin and cannot be opened by force.

Time-lapse TIRFM imaging of the sensor’s Atto 550 fluorescence showed that naïve B cells rapidly spread over the PLB, capturing the sensor and accumulating it in the center of the contact area within 3 to 4 min, forming a flat immune synapse where the sensor persisted for several minutes (Fig. 2C and movie S4). Quantification of the sensor opening showed that naïve B cells were unable to exert a force sufficient to open the sensor (Fig. 2D and movie S4). LZ GC B cells showed a punctate pattern of clustered intensity, consistent with the sensor binding to BCR concentrated in the pod-like structures. We observed robust Atto 647N intensity that coincided with areas of Atto 550 intensity (Fig. 2C), indicating that LZ GC B cells were able to exert a pulling force on the antigen of at least 9 pN through the pod-like structures. Analysis of DZ GC B cells showed a similar result.

To confirm that the LZ GC B cell’s pulling force was exerted by the BCRs in the actin-rich pod-like structures, we simultaneously acquired images of the antigen force sensor opening and IRM images and merged the two. LZ GC B cells exerted force on the sensors beginning about 5 min after the cells were placed on the sensor-containing PLBs, and the opening of the sensors was restricted to the areas of contact of the LZ GC B cell’s pod-like structures with the membrane (Fig. 2E and movie S5).

Behavior of GC B cells in response to antigen is dictated by intrinsic affinity thresholds

We determined whether naïve and GC B cells were able to discern affinity for membrane-associated antigens. As surrogate antigen, we used κ-specific monoclonal antibodies (anti-κ mAbs) that were generated in mice to either rat κ chains and showed a low affinity for human κ chains [the equilibrium dissociation constant (KD) = 3.9 × 10−7], as determined by Biacore surface plasmon resonance analysis, or that were generated in mice to human κ chains and showed a high affinity for human κ chains (KD = 2.4 × 10−9). Streptavidin-conjugated anti-κ mAbs were bound to PLBs that contained biotinylated lipids. λ Light chain–expressing B cells were removed from purified naïve B cell and LZ GC B cell subpopulations by negative sorting. The κ light chain–expressing LZ GC and naïve B cells were placed on PLBs that contained low- or high-affinity anti-κ mAbs for 7 or 25 min. Cells were permeabilized and stained with antibodies specific for the phosphorylated forms of the BCR’s Igα chain, CD79a (pCD79A), and several components of the B cell signaling cascade, including Syk (pSyk), BLNK (pBLNK), and PLCγ2 (pPLCγ2). Cells were imaged by TIRFM, and the MFIs (mean fluorescence intensities) in the contact areas between the B cells and the antigen-containing PLBs were quantified.

The TIRFM images showed that naïve B cells formed immune synapses similarly on both the high- and low-affinity antigen-containing PLBs (Fig. 3A) and accumulated similar amounts of BCR in the area of contact between the B cell and the PLB in response to low- and high-affinity antigens at 7 min (Fig. 3B). At 25 min, the accumulation of the BCR was less for low- versus high-affinity antigens; nonetheless, the accumulation was significantly greater than that of resting naïve B cells (Fig. 3B). In contrast, LZ GC B cells did not form stable contacts with the low-affinity antigen-containing bilayers (Fig. 3A), nor did they accumulate BCRs in the contact area between the B cell and the PLBs to an amount that was significantly greater than that accumulated by resting LZ GC B cells placed on bilayers that did not contain antigen (Fig. 3B). When placed on the high-affinity antigen-containing PLBs, LZ GC B cells accumulated BCRs in a punctate pattern (Fig. 3A). The amount of LZ GC BCRs that accumulated in the interface between the B cell and the high-affinity antigen-containing PLB was significantly greater than that of resting LZ GC B cells not exposed to antigen, although less than that of naïve B cells (Fig. 3B).

Fig. 3 Human GC B cells have a high-affinity threshold for antigen.

Naïve and LZ GC B cells were placed on PLB that did not contain antigen (resting) or placed on PLB that contained either high- or low-affinity anti-κ mAbs. (A) TIRFM images were obtained from cells stained for the BCR and pCD79A. (B and C) Quantification of the BCR (B) and pCD79A (C) in the contact area of the B cells with the PLB in TIRFM images. (D to F) Quantification of TIRF images of cells stained with pSyk (D), pBLNK (E), and pPLCγ2 (F) in the contact area with the PLB of naïve and LZ GC B cells activated by high- and low-affinity antigen for 7 or 25 min. (G) DIC and IRM images of naïve and LZ GC B cells imaged by TIRFM on PLB containing either high- or low-affinity anti-κ mAbs. The still images were taken between 3 and 12 min of movie S6. (H) Kymographs of IRM images of naïve and LZ GC B cell immune synapse on low- or high-affinity antigen-containing PLB. Red arrowheads on low-affinity naïve B cell are extended membrane ruffles and on high-affinity LZ GC B cell are pod-like structures that established stable contacts with antigen-containing PLB. (I) Quantitative analysis of pod-like structure movement in DIC images (number of tracks: n = 347 for high affinity early, n = 887 for high affinity late, n = 241 for low affinity early, and n = 432 for low affinity late). Scale bars, 5 μm. ns > 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001 (unpaired t test). Data are representative of two experiments [mean and SEM (B to F)].

Quantification of the phospho-signaling components in the contact area of the B cells with the bilayers showed a similar pattern. Naïve B cells induced phosphorylation of CD79a, Syk, BLNK, and PLCγ when placed on either low- or high-affinity antigen-containing PLBs (Fig. 3, C to F). In contrast, LZ GC B cells were unable to initiate phosphorylation of signaling components in response to low-affinity antigen-containing PLB (Fig. 3, C to F) but responded to high-affinity antigen-containing PLB. Both the DIC and IRM images (Fig. 3G and movie S6) as well as quantitative analyses of both the IRM and DIC images displayed as kymographs (Fig. 3H) are also shown. High-affinity antigens immobilized naïve B cells in stable large contact areas with the PLB (Fig. 3G and movie S6), as quantified in the kymograph (Fig. 3H). In contrast, naïve B cells on low-affinity antigen PLB formed smaller contact areas and continued to extend membrane ruffles around the periphery, presumably continuing to search for antigen (Fig. 3G and movie S6), as indicated by red arrowheads in the kymograph (Fig. 3H). LZ GC B cells placed on high-affinity antigen-containing PLB vigorously explored the PLB through their peripheral pods (Fig. 3H and movie S6). However, with time, their pod formed stable contacts with the PLB, as indicated by the red arrowheads in the kymograph (Fig. 3H). Low-affinity LZ GC B cells failed to form stable contacts with the membranes but rather continued to make highly dynamic contacts with the PLBs (Fig. 3, G and H, and movie S6), even occasionally moving over the PLB (movie S7). Thus, the behavior of the B cells on the antigen-containing bilayers appears to be dictated by the B cells’ affinity for antigen.

To quantify the movement of pod-like structures in the DIC images, we identified individual pod-like structures on PLBs and measured their movement using a particle tracking algorithm (fig. S4B). During the early stage of contact, LZ GC B cells engaged with high-affinity antigen moved their pods more vigorously in a broader area (median diffusion, 2.5 × 10−3 μm2/s) relative to LZ GC B cells engaged with low-affinity antigen (median diffusion, 1.7 × 10−3 μm2/s) (Fig. 3I). Movement of pod-like structures markedly slowed during the late stage of contact for LZ GC B cells engaged with high-affinity antigen (median diffusion, 1.0 × 10−3 μm2/s) relative to earlier time points and to LZ GC B cells engaged with low-affinity antigen (median diffusion, 1.3 × 10−3 μm2/s) (Fig. 3I), indicating that as more BCRs engage with high-affinity antigen, pod movement stabilizes.

Extraction and trafficking of membrane-associated antigen distinguishes GC B cells

We tested the ability of GC B cells and naïve B cells to extract antigen from pliant plasma membrane sheets (PMSs) presenting either high- or low-affinity anti-κ mAbs. Total purified human tonsil B cells were incubated on PMS coated with fluorescently labeled high- or low-affinity anti-κ for 2 hours, and the B cells were harvested and analyzed by flow cytometry to identify B cells that had acquired fluorescent antigen and, within this population, to identify naïve B cells and GC B cells (Fig. 4A). Naïve B cells acquired similar amounts of antigen from high- and low-affinity PMS. In contrast, significantly fewer GC B cells acquired antigen from the low-affinity PMS as compared with the high-affinity PMS (Fig. 4A).

Fig. 4 Extraction and trafficking of membrane-associated antigen distinguishes GC B cells.

(A) Antigen extraction by naïve and GC B cells placed for 2 hours on PMS that contained either high- or low-affinity anti-κ mAbs. Antigen-positive B cell populations were quantified by flow cytometry. (B) Transport of extracted antigens in naïve and LZ GC B cells placed on antigen-containing PMS. F-actin (green) and antigen (red). (C) Distance of MTOC in naïve and LZ GC B cells to antigen-containing PMS. F-actin (green), γ-tubulin (cyan), and antigen (red). (D) Localization and quantification of SNX9 or SNX18 in naïve and LZ GC B cells 45 min after activation on antigen-containing PMS imaged by confocal microscopy. F-actin (green), SNX9 and SNX18 (cyan), and antigen (red). MFI of SNX9 and SNX18 per each z-stack from the cell bottom to the top (n = 25 per group). (E) Localization of SNX9 or SNX18 with MTOC in naïve and LZ GC B cells 45 min after activation on antigen-containing PMS. α-Tubulin (green), SNX9 and SNX18 (cyan), and antigen (red). (F) Localization pattern of LAMP-1 in naïve and LZ GC B cells (left) 30 min after activation on antigen-containing PMS. 3D colocalization of antigen with LAMP-1 in naïve and LZ GC B cells (right). LAMP-1 (cyan) and antigen (red). (G) Trafficking of extracted antigens to acidic intracellular compartments in naïve and LZ GC B cells with time on antigen-containing PMS. (H) pHrodo FI of naïve B cells and LZ GC B cells (bottom left). Antigen and pHrodo FIs for naïve and LZ GC B cells (top left). Percentage of pHrodo-positive naïve B cells (top right) and LZ GC B cells (bottom right) for B cells that have acquired similar amount of antigen. ns > 0.05, **P ≤ 0.01, and ****P ≤ 0.0001 (unpaired t test). Data are representative of two (B to F) or three experiments (A, G, and H) [mean and SEM (A, C, D, and F)].

We characterized the patterns of trafficking and internalization of membrane-associated antigen captured by BCRs of LZ GC B cells and naïve B cells placed on PLB that contained Atto 633–labeled antigen, fixed, permeabilized, and stained with Alexa Fluor 488–conjugated phalloidin. On naïve B cells, antigen-bound BCRs accumulated in the plane of the contact area with the PMS in the center of the synapse where they were internalized (Fig. 4B and movie S8). No cell surface–associated antigen was detected outside of the contact area, as shown in the 3D surface reconstruction (Fig. 4B). In contrast, LZ GC B cells engaged antigens via pod-like structures and trafficked antigen away from the contact area along the sides of the cells to distal sites for internalization (Fig. 4B and movie S8). To determine the polarity of the cells, we located the microtubule-organizing centers (MTOCs) using γ-tubulin–specific antibodies in naïve and LZ GC B cells after activation on antigen-containing PMS. In most of the activated naïve B cells, MTOCs were polarized and located proximal to PMS (Fig. 4C). In contrast, MTOCs in activated LZ GC B cells were not polarized toward the PMS and appeared to be dispersed large distances from the PMS (Fig. 4C).

BCR-mediated antigen internalization in B cells is primarily by clathrin-mediated endocytosis (CME) (26). Sorting nextin 9 (SNX9) and SNX18 are key components of the endocytosis machinery and modulators of CME (2729). By flow cytometry, the expression of both SNX9 and SNX18 was four- to sixfold higher in naïve B cells as compared with LZ GC B cells (fig. S5, A and B). By confocal microscopy in naïve B cells, both SNX9 and SNX18 were concentrated with antigen in the immune synapse (Fig. 4D), polarized toward the antigen-containing PMS as shown, and quantified in longitudinal sections (Fig. 4D and fig. S5, C to E). In marked contrast, in LZ GC B cells, SNX9 and SNX18 showed no pattern of accumulation but rather were dispersed throughout the cell (Fig. 4D). In addition, 3D reconstructions of images of α-tubulin, SNX9 or SNX18, and antigen showed the polarization of α-tubulin and SNX9 and SNX18 toward the contact area of naïve B cells with the antigen-containing PLB and antigen internalization from the contact site (Fig. 4E and movie S9). In contrast, α-tubulin and SNX9 and SNX18 were not polarized in LZ GC B cells, and the antigen was trafficked along the outside of the cell to distal sites (Fig. 4E and movie S9).

It has been shown, primarily in naïve mouse B cells, that extracted antigens are internalized and trafficked to specialized LAMP-1+ MHC class II–containing late endosomes (MIIC) for antigen processing (30). In activated naïve B cells, LAMP-1+ vesicles were polarized toward the immune synapse and antigens gathered by BCRs were highly colocalized with LAMP-1+ vesicles (Fig. 4F and movie S10). LAMP-1+ vesicles in LZ GC B cells were less polarized, and less antigen was colocalized with LAMP-1+ vesicles.

We next followed the trafficking of antigen internalized by naïve B cells and LZ GC B cells with time. Unsorted purified tonsil B cells were incubated on PMS containing DyLight 650–conjugated F(ab′)2 anti-λ/κ and the pH-sensitive dye, pHrodo, the FI of which increases with increasing pH. Cells were removed from the PMS and analyzed by flow cytometry to identify naïve B cells and LZ GC B cells and to quantify the DyLight 650 and pHrodo FI. In controls in which the PMS contained DyLight 650–conjugated bovine serum albumin and pHrodo, no DyLight 650 or pHrodo FI was detected in naïve and LZ GC B cells (Fig. 4G). Between 30 and 120 min on the PMS, the amount of antigen associated with naïve and LZ GC B cells increased, as did the pHrodo FI; however, the temporal patterns of antigen uptake and pHrodo FI differed between naïve and LZ GC B cells (Fig. 4G). Overlapping the data in Fig. 4G, it is apparent that naïve B cells extracted more antigen with time as compared with LZ GC B cells (fig. S5F) and trafficked a larger portion of the acquired antigen to acidic compartments as indicated by increased pHrodo FI. Comparing naïve B cells with LZ GC B cells that acquired equivalent amounts of antigen at 120 min, it was clear that naïve B cells trafficked a larger portion of the antigen to acidic compartments: More than 80% of naïve B cells were pHrodo positive as compared with 20% of LZ GC B cells, and the pHrodo FI was higher for naïve B cells as compared with LZ GC B cells (Fig. 4H).

Human LZ GC B cells express IRF-4 in response to high-affinity antigen in combination with Tfh cell help

We characterized the response of naïve and LZ GC B cells to high- versus low-affinity membrane-bound antigens in the presence or absence of CD40-specific mAb, interleukin-4 (IL-4), and IL-21 to mimic Tfh cell help. We quantified the expression of IRF-4, a gene required for PC differentiation, and found that LZ GC B cells increased transcription of IRF-4 modestly in response to high- but not low-affinity antigens (Fig. 5A). However, the largest increase in IRF-4 expression was in response to a combination of high-affinity antigen and Tfh cell help. In contrast to LZ GC B cells, naïve B cells expressed IRF-4 equivalently in response to both high- and low-affinity antigens (Fig. 5B). Tfh cell help independently induced IRF-4 expression and augmented the IRF-4 expression to both low- and high-affinity antigens. Together, these data provide evidence that the high-affinity threshold for activation of LZ GC B cells as compared with naïve B cells has the functional outcome of maximally triggering high-affinity LZ GC B cells to differentiate to PCs in the presence of Tfh cell help.

Fig. 5 Human GC B cells express IRF-4 in response to high-affinity antigen in combination with Tfh cell help.

(A and B) Relative mRNA expression level of IRF-4 in LZ GC (A) and naïve B (B) cells after activation with antigens on PLB and/or Tfh-like stimuli composed of anti-CD40 mAb, IL-21, and IL-4. ns > 0.05 and *P ≤ 0.05 (paired t test) [mean and SEM (A and B)].

The differential expression of surface markers may further contribute to affinity discrimination by human GC B cells

We carried out an analysis of more than 330 B cell surface markers on resting and antigen-activated naïve B cells, DZ and LZ GC B cells, and MBCs. The data are expressed as a heat map comparing LZ and DZ GC B cells and MBCs with naïve B cells (Fig. 6A and fig. S6). LZ GC and DZ GC B cells showed nearly identical cell surface expression profiles with the exception of a few previously reported markers including CD83, CD86, and CXCR4. In contrast, the cell surface expression of markers on MBC more closely resembled that of naïve B cells. Many environmental sensing co-receptors involved in the regulation of BCR signaling were down-regulated in GC B cells including FcεRII, FcγRII, CR1, SIGLEC-2, PECAM-1, and CR2, suggesting that the response of GC B cells to antigen may be minimally influenced by environmental signals.

Fig. 6 The differential expression of key surface markers may further contribute to affinity discrimination by human GC B cells.

(A) Heat map: Comparisons of differential surface molecule expression. (B and C) Surface levels of CD29, CD49d, and VLA-4 active epitope in naïve and LZ GC B cells. (D) Binding ability of naïve and LZ GC B cells to VCAM-1–coated beads. ****P ≤ 0.0001 (unpaired t test). Data are from one of five individuals (B) or from representative of three experiments (D) [mean and SEM (C and D)].

Integrins were also differentially expressed by GC B cells as compared with naïve B cells and, given their importance in B cell activation, we quantified the expression levels of a variety of human integrins and found a substantial difference between naïve B cells and LZ GC B cells in their expression of integrins α4, αM, β1, and β7 (Fig. 6A and fig. S7, A and B). We further characterized VLA-4 expression as it plays a key role in stabilizing the initial interaction of B cells with antigens associated with the surface of antigen-presenting cells (APCs) that express its ligand VCAM-1 (31). VLA-4/VCAM-1 interactions also facilitate B cell activation after BCR antigen binding (31). As compared with naïve B cells, the surface expression of the VLA-4 chains CD29 (β1) and CD49d (α4) was greatly reduced on LZ GC B cells (Fig. 6, B and C), as was VLA-4 in the high-affinity state (Fig. 6, B and C) (31).

To determine whether the decreased expression of VLA-4 resulted in a reduced ability of GC B cells to adhere to VCAM-1 on surfaces, we incubated unsorted purified tonsillar B cells with an excess of VCAM-1–coated magnetic agarose beads at 37°C for 20 min and analyzed the composition of bead-bound and bead-unbound B cells by flow cytometry. The vast majority of B cells that bound to the VCAM-1 beads were naïve B cells (Fig. 6D). GC B cells were nearly exclusively recovered in the unbound fraction, indicating that the low expression of VLA-4 on GC B cells had the functional consequence of limiting the ability of GC B cells to engage VCAM-1 surfaces. Together, these results suggest that GC B cells may be highly dependent on BCR-antigen interactions in the absence of strong integrin-mediated cell-cell contact with APCs.

DISCUSSION

We determined that, as compared with human naïve B cells, human GC B cells have a higher intrinsic affinity threshold for antigen. We observed that independently of other extrinsic factors, such as competition among B cells for antigen, the intrinsic affinity of GC B cells for an antigen dictated each subsequent step in B cell activation from the magnitude of BCR signaling to the receptivity of BCR-stimulated GC B cells to Tfh cell signals that drive IRF-4 expression and PC differentiation. We provided evidence that BCRs on LZ GC B cells are concentrated in distinct, highly dynamic, ezrin- and actin-rich pod-like structures through which the BCRs engage antigen, signal, exert pulling forces, and extract antigen from membranes. In contrast, the BCRs on naïve B cells function in flat, stable membrane contacts, with antigen-containing surfaces displaying the well-described features of immune synapses and cSMACs. The role of these pod-like structures in establishing high-affinity thresholds for GC B cells is, at present, a matter of speculation. These structures could serve to isolate small numbers of GC B cells, reducing the B cell avidity for antigen, forcing B cells to function through only monomeric affinity-dependent interactions as described earlier (32) or isolate BCRs from the trans-activity of BCR signaling molecules such as has been described for Syk (33). On a larger size scale, the dynamic nature of the GC pod-like structures may underlie the rapid scanning behavior of LZ GC B cells over FDC observed in vivo in mice (9). We observed that LZ GC B cells engaging high-affinity antigen moved their pods more vigorously in broader areas as compared with LZ GC B cells engaging low-affinity antigen, suggesting that LZ GC B cells modulate their search for antigen in response to BCR affinity. Our results indicate that GC B cells isolated from tonsils have already established a high-affinity threshold for antigen, allowing only cells that meet or exceed this threshold to be selected. Because the threshold appears to be an intrinsic feature of GC B cells, increases in antibody affinity with time could reflect a resetting of the threshold, for example, in LZ GC B cells that reenter the GC DZ.

We also observed marked differences in the internalization and the intracellular trafficking of antigen to acidic compartments for processing. For example, we observed that naïve B cells acquired more antigens with time as compared with GC B cells and immediately delivered a greater portion of their acquired antigen to acidic compartments for processing. In contrast, only a small portion of GC B cells that acquired a similar amount of antigen were able to traffic antigen into acidic compartments. The high threshold for processing of internalized antigen may provide a mechanism by which GC B cells temporally or spatially regulate the presentation of antigen to Tfh cells during GC LZ competition.

In terms of the outcome of LZ GC B cell encounter with antigen, we demonstrated that LZ GC B cells optimally increased IRF-4 transcription in response to high-affinity antigen and Tfh-like stimuli. In mice, it has been shown that sustained high expression of IRF-4 promotes PC differentiation (3437). Our observation is consistent with the findings of Kräutler et al. (38) that provided evidence in mice for a two-signal mechanism in the differentiation of LZ GC B cells to PCs: the first provided by antigen and the second by Tfh cells. In contrast to LZ GC B cells, increases in IRF-4 expression in naïve B cells were relatively affinity independent and more highly dependent on Tfh cell help, suggesting that low-affinity B cells may capture sufficient antigen to activate Tfh cells to allow differentiation directly to short-lived PCs in vivo. The expression of CD40, IL4R, and IL21R in GC B cells was slightly higher than that in naïve B cells (fig. S6C), which could contribute to higher IRF-4 mRNA expression in GC B cells as compared with naïve B cells in response to Tfh-like stimuli alone. However, the induction of IRF-4 mRNA expression was antigen affinity dependent only in GC B cells.

We also provided evidence for a variety of additional differences between GC B cells and naïve B cells that together could further contribute to establishing antigen affinity thresholds. For example, we showed that, as compared with naïve B cells, human LZ GC B cells express very little VLA-4, which is unique to human GC B cells, as mouse GC B cells show comparable surface expression level of VLA-4 in comparison with naïve B cells (39). VLA-4 is an integrin that binds to its ligand, VCAM-1, expressed by FDCs that present antigen to LZ GC B cells and serves to lower the affinity threshold for BCR signaling. It is of interest that in contrast to VLA-4, the expression of which was decreased on LZ GC B cells as compared with naïve B cells, the levels of expression of LFA-1 were similar for LZ GC B cells and naïve B cells. LFA-1 is an integrin that both stabilizes B cell interactions with APCs and plays a critical role in B cell–T cell interactions (40). This suggests that once LZ GC B cells succeed in triggering BCR signaling in a mostly VLA-4–independent fashion, the GC B cells will not be compromised in subsequent LFA-1–dependent processes, including interactions with Tfh cells.

In the studies described here, we limited our analysis to a comparison of GC B cells and naïve B cells. It will be of interest in future studies to understand the mechanisms by which MBCs that have undergone affinity selection in GCs discriminate antigen affinity. Wang et al. (41) recently reported that, in mice, GC B cells expressing IgG BCRs generated greater traction forces on membrane-associated antigen during the initiation of activation than did naïve B cells, consistent with increased expression of motor proteins in MBCs. Preliminary results from our studies are consistent with this observation, showing that MBCs were able to exert pulling force on membrane-associated antigen that was sufficient to open 9-pN DNA force sensor, whereas naïve B cells were not. How this function of MBCs might contribute to affinity discrimination remains to be determined. Our studies here were limited to κ-specific mAbs serving as surrogate antigens that differed about 60-fold. It will be of interest to develop additional reagents that allow a more precise assessment of the degree of affinity discrimination of these two human B cell subsets.

Our studies presented here provided evidence for intrinsic properties of LZ GC B cells that may contribute to enhanced affinity discrimination. Further elucidating the mechanisms of affinity discrimination may provide new strategies for the development of vaccines that require the generation of affinity-matured antibody responses.

MATERIALS AND METHODS

Purification and sorting of human tonsil B lymphocytes

Tonsils were obtained from DC-CFAR Basic Science Core and Children’s National Medical Center (CNMC) from patients undergoing tonsillectomies. Use of these tonsils for this study was determined to be exempt from review by the NIH Institutional Review Board in accordance with the guidelines issued by the Office of Human Research Protections. B cells were isolated from tonsil single tonsil cell suspensions by negative selection using the EasySep Human B Cell Enrichment Kit (STEMCELL) and, when indicated, were sorted to obtain populations of naïve, MBC, and LZ and DZ GC B cells staining with fluorescently labeled anti-IgD, anti-CD10, anti-CD184, and anti-λ using a FACSAria II cell sorter (BD Biosciences).

Microscopy

PLBs containing either goat F(ab′)2 anti-human Igκ (SouthernBiotech) and anti-human Igλ (SouthernBiotech) or, for affinity measurements, anti-human Igκ or anti-rat Igκ as surrogate antigens were prepared as described (42). For fixed cell imaging, cells were stained with antibodies as indicated, placed on PLB for the times indicated, fixed and permeabilized in saponin, and stained with the appropriate antibodies to detect intracellular proteins. Images were acquired using an Olympus IX-81 TIRF microscope system and analyzed using MATLAB (MathWorks) software. Pearson’s correlation coefficients were calculated using MATLAB software. Time-lapse live-cell TIRF images were acquired at 6-s intervals for 20 min, as described for cells stained with antibodies as indicated. Time-lapse live-cell DIC TIRF image acquisition and analysis were performed as previously described (43). Time-lapse live-cell IRM TIRF images were acquired at 3-s intervals for 20 min, and MATLAB algorithms were developed to perform automated image analysis as detailed in Supplementary Methods. BCR cluster growth was measured in live-cell TIRF images using MATLAB, as described (43). For super-resolution images of F-actin structure in the immune synapse, cells were stained with phalloidin (Thermo Fisher Scientific) and images were acquired by STED microscopy (Leica). Scanning electron microscopy images were acquired from fixed samples coated with 5 nm of gold in an EMS 575-X sputter coater (Electron Microscopy Sciences, Hatfield, PA) and imaged with a Hitachi S-3400 N1 scanning electron microscope (Hitachi High Technologies America Inc., Pleasanton, CA). Confocal images were obtained using a Zeiss-880 confocal microscope, and stacked images were deconvolved with Huygens Software. The MFI and the Pearson’s correlation coefficients of fluorescent proteins were calculated from background-subtracted images using Zen software (Zeiss). Distance of fluorescent proteins from PMS and MFI per each z-stack were measured using ImageJ [National Institutes of Health (NIH)]. 3D images and movies and colocalizations were generated using Imaris (Bitplane).

Flow cytometry

To quantify cell surface marker expression, purified tonsillar B cells from five different individuals were stained with a near-infrared (IR) LIVE/DEAD marker (Thermo Fisher Scientific) and fluorescently labeled specific antibodies on ice, washed, and analyzed by flow cytometry using a BD LSR II instrument and the data were analyzed using FlowJo (FlowJo, LLC) and Prism (GraphPad) software. For quantification of intracellular proteins, cells were fixed and permeabilized using BD Cytofix/Cytoperm (BD Biosciences), incubated with fluorescently labeled specific antibodies at room temperature for 2 hours, and washed, and when indicated, cells were incubated with fluorescently labeled secondary antibodies, washed, and analyzed.

VCAM-1 binding assay

Purified tonsil B cells were incubated with nickel agarose magnetic beads (Sigma) coated with recombinant human VCAM-1 Fc chimera proteins (R&D Systems) or purified human IgG (MP Biomedicals) at 37°C for 30 min. Unbound cells were collected using a magnetic separation stand, and bound cells were eluted from the beads with Hanks’ balanced salt solution containing 150 mM imidazole (Sigma) elution buffer. Cells were analyzed by flow cytometry (BD LSR II), and flow cytometry data were analyzed with FlowJo (FlowJo, LLC) and Prism (GraphPad) software.

Detection and tracking of pod-like structures

Time-lapse live-cell DIC and IRM images were acquired at 3-s intervals for 20 min after placing cells on antigen-containing PLBs. Analyses of time-lapse acquisitions were performed with MATLAB and ImageJ. Variations in intensity in DIC image sequences were normalized using linear regression in MATLAB. Images were sharpened using an unsharp masking algorithm in MATLAB to enhance the contrast of the pod-like structures. Pixels were proportionately scaled down and binned into 8-bit unsigned integers to reduce the effects of small variations in pixel intensities. Masks were created using a method similar to topographic prominence in ImageJ by iteratively applying the maximum function to identify the local intensity maximum of pod-like structures as a single point. Points were then tracked over time, and their diffusion coefficients were calculated using MATLAB code developed to track single particles as described previously (32).

Measurement and analysis of pulling forces

Antigen-conjugated force and control sensors were produced, and measurement of sensor opening was conducted as described previously (44) using sensors containing goat F(ab′)2 anti-human Igκ and anti-human Igλ as surrogate antigens. For the movies, images were background subtracted and flatfield corrected in ImageJ (NIH), and value 1000 was added to Atto 550 images before ratio calculation to reduce background. For statistical analysis, images were acquired after 20 min of incubation of the cells in antigen-containing PLB, background subtracted, and flatfield corrected in ImageJ. Using the threshold function, the ratio of Atto 647N to Atto 550 fluorescence in antigen clusters of each cell was calculated using ImageJ, and statistical analysis of sensor opening from each B cell subset was calculated using Prism software (GraphPad).

Antigen extraction and trafficking

To measure antigen extraction, PMSs containing either fluorescently labeled anti-human Igκ (GeneTex) or anti-rat Igκ (Thermo Fisher Scientific) were prepared as described (44). Cells were harvested and stained with a near-IR LIVE/DEAD marker (Thermo Fisher Scientific), anti-IgD Fab (SouthernBiotech), and anti-CD10 (BD Biosciences), and the signal was quantified by flow cytometry (BD LSR II). To measure antigen trafficking, cells were placed on PMSs containing DyLight 550 and pHrodo avidin (Thermo Fisher Scientific)–conjugated goat F(ab′)2 anti-human Igκ (SouthernBiotech) and anti-human Igλ (SouthernBiotech), as described above. Cells were harvested and stained with a near-IR LIVE/DEAD marker (Thermo Fisher Scientific), anti-IgD Fab (SouthernBiotech), and anti-CD10 (BD Biosciences), and the signal was quantified by flow cytometry (BD LSR II).

Activation of naïve and LZ GC B cells and measurement of IRF-4 mRNA expression

B cells were placed on PLB containing anti-human Igκ or anti-rat Igκ antibodies and incubated at 37°C with 5% CO2 and 75% humidity for 1 hour. RPMI-10 culture media containing anti-CD40 antibody (2 μg/ml; R&D Systems), IL-4 (40 ng/ml; R&D Systems), and IL-21 (2 μg/ml; BioLegend) or RPMI-10 culture media alone was added to the cells, and cells were incubated at 37°C for an additional 2 hours and harvested. Cells were lysed, RNA was reverse transcribed, and complementary DNA was used to run TaqMan Gene Expression Assays using a TaqMan Gene Expression Cells-to-CT kit (Thermo Fisher Scientific) on CFX Connect Real-Time PCR Detection System (Bio-Rad). TaqMan Gene Expression Assays detecting IRF-4 (Hs00180031_m1; Thermo Fisher Scientific) and ACTB (Hs1060665_g1; Thermo Fisher Scientific) were used to measure levels of mRNA, and ACTB was used for normalization. Data were analyzed using Prism software (GraphPad).

Measurement and analysis of surface molecule expression

Purified tonsil B cells from three different individuals were incubated at 37°C for 2 hours with goat F(ab′)2 anti-human Igκ (10 μg/ml) (SouthernBiotech) and anti-human Igλ (10 μg/ml) (SouthernBiotech). Cells were then washed; stained with a LIVE/DEAD marker (Thermo Fisher Scientific), anti-CD19 (BioLegend), anti-IgD (Miltenyi), anti-CD10 (BD Biosciences), and anti-CD184 (BD Biosciences); and barcoded as described (45). Cells were washed, combined, and further stained for surface markers using a LEGENDScreen (BioLegend) human cell screening kit. Fluorescent signals were quantified by flow cytometry (BD LSR II). Flow cytometry data were analyzed with FlowJo v.10.1 and Microsoft Excel 2016.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/3/29/eaau6598/DC1

Methods

Fig. S1. Fluorescence-activated cell sorting gating strategy for naïve, memory, LZ GC, and DZ GC B cells.

Fig. S2. Quantification of dynamic movement of unique pod-like structures formed in the immune synapse of LZ GC B cell and expression level of BCR and ezrin.

Fig. S3. Antigen-induced BCR signaling in naïve B cells and GC B cells.

Fig. S4. Design of force sensor and detection of pod-like structures.

Fig. S5. Polarization of SNX9 and SNX18 in naïve and LZ GC B cells after activation on antigen-containing PMS.

Fig. S6. The differential expression of key surface markers.

Fig. S7. Surface expression of LFA-1 and specificity of VLA-4 binding.

Movie S1. Distinctive membrane dynamics in the immune synapse of live naïve and LZ GC B cells placed on antigen-containing PLB were visualized by DIC and IRM.

Movie S2. The dynamics of membrane movement in the immune synapse upon activation were visualized by IRM.

Movie S3. Time-lapse imaging of CD19 and BCR in the immune synapse of naïve B cells and LZ GC B cells.

Movie S4. Pulling forces in the immune synapse of live naïve and LZ GC B cells were monitored with time using DNA-based force sensors.

Movie S5. Colocalization of the cell contact sites and the locations of pulling force in LZ GC B cell.

Movie S6. The dynamics of membrane movement in the immune synapse of naïve and LZ GC B cell upon activation by high- or low-affinity antigen were visualized by DIC and IRM.

Movie S7. Motile LZ GC B cells were observed upon activation by low-affinity antigen.

Movie S8. Distinct antigen transport pattern shown in human LZ GC B cells on PMS presenting surrogate antigens as compared with naïve B cells.

Movie S9. Localization of MTOC with SNX9 or SNX18 after activation in naïve and LZ GC B cells.

Movie S10. Localization of LAMP-1 after activation in naïve and LZ GC B cells.

Table S1. Raw data file (Excel file).

REFERENCES AND NOTES

Funding: This work was supported by the Intramural Research Program of the NIH, National Institute of Allergy and Infectious Diseases. P.T. and K.M.S. were supported by the H2020 European Research Council (Consolidator grant 648228) and the Francis Crick Institute, which receives its core funding from Cancer Research UK, the UK Medical Research Council, and the Wellcome Trust. We would like to thank DC-CFAR Basic Science Core and Children’s National Medical Center (CNMC) for providing tonsil specimens. Author contributions: K.K. and S.K.P. conceived and designed the study. K.K., N.Q., H.S., A.S., P.H., and J.L. performed experiments. K.K., N.Q., H.S., A.S., J.M.-L., J.B., J.L., H.X., and C.X. analyzed data and performed statistical analysis. K.M.S. and P.T. prepared DNA force sensor. K.K. wrote the original manuscript. N.Q., H.S., and S.K.P. edited the original manuscript. S.K.P. led the study. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article