Imaging of Germinal Center Selection Events During Affinity Maturation

See allHide authors and affiliations

Science  26 Jan 2007:
Vol. 315, Issue 5811, pp. 528-531
DOI: 10.1126/science.1136736


The germinal center (GC) is an important site for the generation and selection of B cells bearing high-affinity antibodies, yet GC cell migration and interaction dynamics have not been directly observed. Using two-photon microscopy of mouse lymph nodes, we revealed that GC B cells are highly motile and extend long cell processes. They transited between GC dark and light zones and divided in both regions, although these B cells resided for only several hours in the light zone where antigen is displayed. GC B cells formed few stable contacts with GC T cells despite frequent encounters, and T cells were seen to carry dead B cell blebs. On the basis of these observations, we propose a model in which competition for T cell help plays a more dominant role in the selection of GC B cells than previously appreciated.

Germinal centers (GC) represent critical sites within organized lymphoid tissues in which B cell responses to antigen are amplified and refined in specificity. A classical model of GC function holds that B cells in the dark zone undergo rapid rounds of proliferation and somatic hypermutation of their antibody genes, followed by exit from the cell cycle and movement to the light zone, where the B cells undergo selection based on the affinity of their surface antibody for antigen (15). The selection process is thought to involve competition between GC B cells for capture of antigen in the form of immune complexes displayed on the processes of follicular dendritic cells (FDCs) (1, 3, 5, 6). However, recent experimental evidence and computer simulations have contradicted aspects of this classical model (3, 79), and the mechanism by which selection occurs in the GC has remained elusive. One of the challenges in studying GC selection has been a lack of knowledge about the dynamics of GC B cells in their complex physiological milieu.

We developed a system to study the dynamics of GC B cells within intact mouse lymph nodes (LNs) by two-photon microscopy [the system is described in detail in the supporting online material (SOM) text] (10). In this system, 1 to 2% of GC B cells expressed green fluorescent protein (GFP) (Fig. 1A and figs. S1 and S2) and by our measurements were able to undergo the normal processes of class switching, somatic hypermutation, and affinity maturation (figs. S3 and S4). The dynamics of GC B cell motility in explanted LNs were compared with two other B cell populations: (i) naïve B cells (representing B cells before antigen encounter) in the follicular mantle that surrounds the GC and (ii) plasma cells (representing post-selected B cells that secrete antibodies) in the LN medulla (SOM text). GC B cells had highly dynamic shapes, extending dendritic processes resembling pseudopods as they moved (Fig. 1B and movie S1), whereas naïve follicular mantle B cells and plasma cells exhibited a more round phenotype (Fig. 1B). Similar observations were made when GC and follicular mantle B cells were imaged by intravital microscopy of intact LNs (movie S2). Tracking analysis indicated that GC B cells were highly motile, similar to follicular mantle B cells, whereas plasma cells showed little motility (Fig. 1C; fig. S5, A to E; fig. S6; and movie S3). GC B cell motility was partially dependent on the chemokine CXCL13 (Fig. 1C and fig. S5) that is expressed by FDCs (SOM text) (11).

Fig. 1.

Dynamics and motility of GC B cells compared with follicular mantle (FM) B cells and plasma cells (PC). (A) An 18-μm maximum intensity z-projection from two-photon microscopy image stacks of a GC and FM in an intact LN. A time-lapse recording corresponding to the center of this region is shown in movie S1. (B) Representative time-lapse images from two-photon microscopy showing the morphology of a GC B cell, FM B cell, and PC. The FM B cells in this experiment were naïve GFP+ cells that were also labeled with CMTMR (10), and only the GFP channel is shown in the images. Scale bars, 10 μm. (C) Superimposed 15-min tracks of 40 randomly selected cells of each indicated type in the xy plane, setting the starting coordinates to the origin. Units are in micrometers. WT, wild type. Each color represents one cell's path. (D) (Left) Maximum-intensity projection of FM (red) and GC (green) B cells. (Right) Tracks of FM (red) and GC (green) B cells. The gridlines are separated by 20 μm.

On average, GC B cells formed separate clusters from naïve B cells in the follicular mantle (Fig. 1, A and D), giving the impression of distinct regions typically seen in static images, such as those obtained by immunohistochemistry (for an example, see fig. S2C). However, examination of the boundary between these regions revealed that the tracks of GC B cells and follicular mantle B cells were overlapping (Fig. 1D and movie S4), indicating that a physical boundary does not demarcate the GC and follicular mantle. Instead, the majority of cells turned after crossing from one region to the other, suggesting that the GC and follicular mantle are segregated by the responses of cells to attractive (or repulsive) cues.

The behavior of cells in GC dark and light zones was examined by labeling light zone FDCs in vivo with immune complexes containing the fluorescent protein phycoerythrin (PE) (Fig. 2A and movie S5). B cells in dark and light zones were similarly motile (fig. S7). In 1 hour, most GC B cells appeared to stay within the dark or light zones, although a measurable proportion (5 to 8%) traveled along relatively straight paths from one zone to the other, covering substantial distances (Fig. 2, B and C, and movies S6 and S7). These observations extend previous conclusions based on fixed tissue analysis (1) that GCB cells migrate from the dark zone to the light zone and provide direct evidence that GC B cells return from the light zone to the dark zone. Within the light zone, FDC processes appeared to undulate, perhaps due to displacement by migrating GC B cells (movie S5). However, there was little evidence that GC B cells pause on FDC processes, in contrast to observations of the T cell–DC interactions in the T cell area (12). GC B cells in the light zone followed straighter paths than those in the dark zone (fig. S7, B and C), which may relate to the high concentration of CXCL13 on the FDC processes in the light zone (11).

Fig. 2.

GC B cell movement within and between GC dark and light zones. (A) Immunofluorescence of cryostat sections showing extensive overlap of in vivo deposited PE immune complexes (PE IC) (red) with FDC-M2 antibody staining (green). The follicular mantle is shown with staining for immunoglobulin D (IgD) (blue). Scale bar, 100 μm. (B) Representative manual classification (10) of cell tracks into groups with respect to the PE+ light zone, corresponding to GC 1A in (C). The gridlines are separated by 20 μm. (C) Frequency of cells traveling between the light and dark zones in three GCs imaged 7 days after immunization. Imaging sessions of 1 hour each were subdivided into two segments (A and B) to facilitate analysis.

During time-lapse recordings, we occasionally observed cell division in both light zones (Fig. 3A) and dark zones (Fig. 3B), consistent with other studies that have questioned the classical model that proliferation only occurs in the dark zone (3, 8, 13). We further analyzed the relationship between GC B cell position and cell-cycle behavior by flow cytometry (SOM text), using surface abundance of the chemokine receptor CXCR4 (Fig. 3C) as a marker to delineate cells in dark (CXCR4hi) and light (CXCR4lo) zones (11). Analysis of DNA content (Fig. 3D) revealed that a similar proportion of GC B cells were in S phase in both dark and light zones (Fig. 3E). A subset of cells was tracked over one complete cell cycle by pulse labeling with 5-bromo-2′-deoxyuridine (BrdU), a thymidine analog that is incorporated into cells undergoing DNA synthesis (S phase) near the time of BrdU injection (14). A similar proportion of GC B cells in dark and light zones were labeled 30 min after BrdU injection; however, after 5 hours, at which time the cells had divided and returned to G1 phase, they accumulated in the dark zone but not in the light zone (Fig. 3, F and G; fig. S8; and fig. S9). These observations indicate that cells that divide in the light zone do not stay resident there—perhaps they return to the dark zone, exit the GC, or undergo apoptosis. From 5 to 12 hours after BrdU injection, the G1 phase cells that had accumulated in the dark zone began to appear in the light zone, suggesting that the light zone is replenished continuously by an influx of cells from the dark zone that recently completed a cell cycle (Fig. 3, F and G; fig. S8; and fig. S9). Cells were not seen reentering S phase until 12 hours after they were first labeled with BrdU in S phase (Fig. 3G and fig. S9, C and D), indicating that the average cell-cycle time is 12 hours or longer, rather than 6 to 7 hours as has been concluded in previous studies (1).

Fig. 3.

Cell-cycle analysis in dark and light zones. (A and B) Single-plane time-lapse images of dividing GC B cells in the light (A) and dark (B) zones. The image sequences begin when the cells have rounded up and stopped moving. (C) CXCR4 is expressed more highly on GC B cells in the dark zone than in the light zone. (Left) Representative immunohistochemistry of a cryostat section of an immunized mouse LN (antibodies used for staining, bottom right). (Right) Representative gating of CXCR4lo (light zone) and CXCR4hi (dark zone) GC B cells by flow cytometry. (D) Representative cell-cycle analysis of GC B cells by flow cytometric measurement of DNA content. DAPI, 4′,6-diamidino-2-phenylindole. (E) Frequency of cells that were in each stage of the cell cycle in the CXCR4lo and CXCR4hi GC B cell subsets. (F) Frequency of cells that were BrdU+ in the CXCR4lo or CXCR4hi GC B cell subsets after BrdU injection at 0 hours. (G) Frequency of cells that were in each stage of the cell cycle among BrdU+ GC B cells. Data points in (E) to (G) represent individual mice. Similar data were obtained in splenic GCs in other experiments. GC B cells in (C) to (G) were gated as in fig. S9A.

T cells are concentrated in the light zone (1), and T cell help is thought to be essential for GCs (3, 15). To understand the mode of B cell–T cell interaction in GCs, we visualized the interactions between GFP+ GC B cells and GC T cells expressing cyan fluorescent protein (CFP) (movie S8). Most contacts between GC B and T cells were of short duration (Fig. 4A), and only about 4% of GC B cell–T cell encounters led to the formation of stable conjugates, defined here as contacts lasting for more than 5 min (Fig. 4, B and C). In contrast, early in the immune response before GC formation, more than 50% of contacts between cognate antigen-specific B and T cells led to the formation of stable conjugates (Fig. 4C and movie S9) (16). Upon encounters leading to stable interactions with GC B cells, rapidly migrating GC T cells sharply decreased their motility to match B cell motility (Fig. 4D), and conjugates were led by B cells (movie S10), similar to the dynamics of stable B-T conjugates early in the immune response (16). The median velocity of GC T cells stably interacting with GFP+ B cells was below 10 μm/min (Fig. 4E). Only 32% of total GC T cells showed a median velocity below 10 μm/min (Fig. 4E), and the motility coefficient of GC T cells was greater than that of GC B cells (Fig. 4F and fig. S5E), strongly suggesting that the majority of T cells in GCs were not engaged in stable interactions with B cells. Our analyses suggest that each GC B cell encounters as many as 50 T cells per hour and that the majority of GC B cells are capable of binding antigen (fig. S3B). It thus seems likely that T cell help is limited, not only because there are fewer T cells than B cells, but also because of additional mechanisms suppressing stable B-T interactions in GCs.

Fig. 4.

Dynamics of GC T cell interactions with live GC B cells and dead B cell blebs. (A and B) Time-lapse images of a brief contact (A) and a stable conjugate [(B), arrowhead and dotted line] of a GC B cell (green) and T cell (blue). (C) Contact times between antigen-specific B and T cells at the indicated number of days after immunization. Black bar segments indicate conjugates that could not be tracked for their entire duration and therefore are underestimates. Data are from three independent experiments. (D) Magnitudes of the velocities of a GC B cell and T cell that formed a stable conjugate. (E) Median velocity of total GC T cells, GC T cells stably interacting with GFP+ GC B cells, and GC T cells stably interacting with GFP+ blebs. (F) Displacement of GC T cells plotted against the square root of time. Blue line, mean of four imaging data sets from four recipient mice (error bars indicate SD); red line, mean of seven T cells stably interacting with GC B cells; M, motility coefficient (10). (G) Image of a GC B cell undergoing cell death. Dotted line, 5-min track of migration preceding the first time point; arrowheads, fragments of the cell. (H) Image of a GC T cell (blue) picking up a GC B cell bleb (green). (I) Image of a GC T cell (blue) carrying a GC B cell bleb (green, indicated by arrowheads) over a path shown with a dotted line. Some follicular mantle cells (red) are also visible. (J) Contact time distribution between GC T cells and GFP+ GC B cell blebs from three independent experiments. Black bar segments indicate incomplete tracking as in (C). Scale bars, 10 μm.

During the selection process in GCs, many GC B cells die and are visible by histology as “tingible bodies” inside macrophages (1). The cell bodies of dying GFP+ GC B cells were observed to undergo fragmentation (Fig. 4G), but surprisingly, this occurred outside of macrophages (movie S11). Blebs of dead GFP+ GC B cells appeared to be taken up by multiple macrophages (movie S11), although some blebs moved rapidly away from the original location of cell death, as if carried by motile cells (movie S2). Indeed, some GFP+ B cell blebs were attached to and carried by rapidly migrating CFP+ T cells (Fig. 4, H and I, and movie S12). All T cells that carried GFP+ B cell blebs had a median velocity greater than 10 μm/min (Fig. 4E), suggesting that they were not undergoing stable interactions with living B cells. The GFP+ GC B cells represent only about 1 to 2% of GC B cells, and we observed about 0.5% of T cells carrying GFP+ blebs; by extrapolation, at least one quarter of the GC T cells may be associated with one or more blebs from dead GC B cells. A higher frequency of bleb–T cell interactions were stable compared with live B cell–T cell interactions (Fig. 4, C and J), suggesting that these dead B cell fragments may affect the availability of T cell help in GCs.

Our findings reveal that GC B cells are highly motile and exhibit a probing behavior as they travel over the antigen-bearing FDC network. The lack of GC B cell pausing suggests that the selection mechanism does not involve competition for adhesion to FDCs, whereas the rapid movement of B cells in close proximity to each other raises the possibility that high-affinity cells remove surface-bound antigen from lower-affinity cells. The observed migration of GC B cells from light to dark zones is consistent with GC B cells undergoing repeated rounds of mutation and selection within a given GC (17). Our estimate that GC B cells spend only several hours in the light zone suggests a limited amount of time to access helper T cells. Given that stable interactions of GC B cells with GC T cells were infrequent, it seems possible that T cell help is a limiting factor driving selection of higher-affinity B cell clones. In vitro studies have shown that T cells responding to antigen-presenting B cells can be sensitive to variations in the affinity of the B cell receptor across several orders of magnitude (18). We propose a selection model in which newly arising mutated GC B cells with higher affinity for antigen obtain and process greater amounts of antigen in a given period of time and then outcompete the surrounding B cells and B cell blebs for the attention of GC T cells.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S9


Movies S1 to S12

References and Notes

View Abstract

Navigate This Article