Dynamic signaling by T follicular helper cells during germinal center B cell selection

See allHide authors and affiliations

Science  29 Aug 2014:
Vol. 345, Issue 6200, pp. 1058-1062
DOI: 10.1126/science.1257861

T and B cells' intricate molecular dance

Generating high-affinity antibodies to fight infection is no easy task. To do so requires multiple steps, including T cells interacting with antibody-producing B cells in lymph nodes. These interactions select B cells expressing high-affinity antibodies for further proliferation, ensuring that the immune response generates high-affinity antibodies in large quantities. Shulman et al. use fluorescent live-cell imaging in mice to determine the molecular details of these interactions. They find that T cells engage B cells in short-lived mobile contacts during selection. These contacts cause T cells to flux calcium and produce proteins called cytokines, which probably drive B cells to proliferate and produce high-affinity antibodies.

Science, this issue p. 1058


T follicular helper (TFH) cells select high-affinity, antibody-producing B cells for clonal expansion in germinal centers (GCs), but the nature of their interaction is not well defined. Using intravital imaging, we found that selection is mediated by large but transient contacts between TFH and GC B cells presenting the highest levels of cognate peptide bound to major histocompatibility complex II. These interactions elicited transient and sustained increases in TFH intracellular free calcium (Ca2+) that were associated with TFH cell coexpression of the cytokines interleukin-4 and -21. However, increased intracellular Ca2+ did not arrest TFH cell migration. Instead, TFH cells remained motile and continually scanned the surface of many GC B cells, forming short-lived contacts that induced selection through further repeated transient elevations in intracellular Ca2+.

Germinal centers (GCs) are specialized microanatomical sites where B cells undergo clonal expansion, somatic hypermutation, and affinity maturation (13). Through iterative cycles of diversification and selection, the GC produces high-affinity memory B and plasma cells (24). Selection of high-affinity GC B cells requires their interaction with T follicular helper (TFH) cells, which must discern among B cell clones according to their surface density of peptide–major histocompatibility complex II (pMHCII) (5). GC B cells are then programmed by TFH cells to expand and hypermutate in direct proportion to the levels of cognate antigen presented (6). These events are controlled by TFH cell–derived signals, including membrane-bound inducible costimulator (ICOS) and CD40L and the cytokines interleukin-4 (IL-4) and IL-21 (7, 8), which are delivered in short-lived intercellular contacts (9).

To examine how the interactions between TFH cells and GC B cells control selection, we imaged cells expressing genetically encoded fluorescent proteins in vivo by means of two-photon laser-scanning microscopy (TPLSM). Ovalbumin (OVA)–specific, T cell receptor (TCR) transgenic OT-II T cells expressing DsRed were adoptively transferred into congenic mice before priming with OVA in alum. After 2 to 3 weeks, a 95:5 mixture of nonfluorescent Ly75−/− and GFP+ Ly75+/+ (Ly75 encodes the cell-surface receptor DEC205) B cells, both specific for NP (4-hydroxy-3-nitrophenylacetyl; B1-8hi) (10), was transferred before boosting with soluble OVA conjugated to NP (NP–OVA) (11, 12). To induce selection, we increased the levels of pMHCII on the surface of GC B cells 7 to 8 days later through injection of DEC205 antibody fused to the cognate antigen OVA (αDEC-OVA) (Fig. 1A). This chimeric antibody targets DEC205, an endocytic receptor that carries associated proteins into MHCII-processing compartments of GC B cells (5). As a result, targeted GC B cells are initially retained in the GC light zone (LZ) and thereafter proliferate in the dark zone (DZ) (5, 6). As a control, mice were injected with chimeric DEC205 antibody fused to an irrelevant antigen (Plasmodium falciparum circumsporozoite protein, αDEC-CS) (5). Popliteal lymph nodes were exposed after 4 to 10 hours, GCs were imaged by means of TPLSM (Fig. 1A), and the results were subjected to colocalization analysis (fig. S1).

Fig. 1 Dynamics of TFH and B cell interactions in the GC.

(A) Timeline of the experimental protocol. i.p., intraperitoneally; s.c., subcutaneously. (B) GCs containing a 5:95 mixture of B1-8hi GFP+ Ly75+/+ B cells (light blue), B1-8hi Ly75−/− B cells (nonfluorescent), and OT-II DsRed+ T cells (red) were imaged by means of TPLSM after subcutaneous injection of αDEC-OVA or αDEC-CS as control. T-B contacts (green) were detected by means of colocalization. Collapsed z-stacks of 40-μm depth (in 5-μm steps) are shown. (Bottom) T and B cell dynamics over time. Images correspond to movies S1 to S3. (C) Velocity analysis of OT-II T cells and B1-8hi B cells in GCs. (D) T-B contact duration as measured by the lifetime of the T-B colocalized areas. Percentages indicate events lasting >5 min, marked by a dashed red box. (E) T-B conjugate velocity was measured as the velocity of the T-B colocalized area. (F) Contact size analysis as measured by T-B colocalized area. Each data point represents a single cell, and red lines represent mean values. Data in (C) to (F) were pooled from two to four mice imaged in two to four independent experiments. *P < 0.0001; two-tailed Student’s t test.

Consistent with previous observations (9, 11, 1316), GC lymphocytes were highly motile (TFH cells, 9 μm/min and B cells, 6.6 μm/min) (Fig. 1, B and C, and movie S1). Under steady-state conditions, in which an unknown fraction of B cells are being positively selected, the majority of T-B contacts were short-lived (Fig. 1, B and D, and movie S1). Positive selection through αDEC-OVA injection was associated with a reduction in both GC B and TFH cell velocities, along with an increase in the duration of the T-B contacts (P < 0.0001) (Fig. 1, B to D, and movies S2 and S3). In particular, the fraction of conjugates lasting 5 min or longer increased from 2.7 to 20.7% of the total interactions (Fig. 1D), and these occasionally moved at the B cell velocity (4.16 μm/min on average) (Fig. 1E). Although most of the conjugates moved short distances, and it was not possible to determine which one of the partners drags the other (movie S2), in those cases that could be interpreted the conjugates were led by the B cell and rarely, if at all, by the T cell (movie S3). Thus, even under conditions of enforced selection, most T-B interactions resembled those found under physiologic conditions in that they remained transient, with T cells forming and breaking contacts with multiple B cells (Fig. 1D and movie S3). Volume analysis of the T-B colocalized area revealed that the average contact size of the stable T-B conjugates (>5 min) was enlarged threefold during positive selection compared with control (Fig. 1F). As expected, polyclonal follicular B cells within the mantle zone did not slow down or form contacts with TFH cells after αDEC-OVA injection (fig. S2 and movie S4). We conclude that positive selection through increased pMHCII on GC B cells is associated with longer but dynamic T-B contacts involving a larger surface area between the two interacting cells.

To examine whether positive selection interferes with the interactions between TFH cells and GC B cells presenting low levels of pMHCII, we directly imaged selected and nonselected B cells within the same GC. GCs containing OT-II DsRed+ T cells and a mixture of B1-8hi GFP+ Ly75−/−, CFP+ Ly75+/+, and nonfluorescent Ly75−/− B cells at a ~5:5:90 ratio were generated and imaged as in Fig. 1A. Positive selection of the B1-8hi CFP+ Ly75+/+ B cells was induced by injecting αDEC-OVA. When compared directly with Ly75−/− B cells presenting lower levels of pMHCII in the same GC, positively selected Ly75+/+ B cells interacted for a longer time with TFH cells (P < 0.0001) (Fig. 2, A and B, and movie S5) and formed a greater number of stable contacts (> 5 min) (Fig. 2C). However, selection of Ly75+/+ GC B cells did not alter the behavior of nonselected Ly75−/− cells, which showed no significant change in contact duration or in the total number of interactions (> 0.5 min) with TFH cells (Figs. 1D and 2, B and C, and movie S5). The majority of TFH cells interacting with selected cells concurrently formed transient contacts with nonselected B cells (Fig. 2A and movie S5). We conclude that the increase in contact duration of TFH cells with selected B cells does not substantially affect their interactions with nonselected B cells.

Fig. 2 TFH cell interactions with selected and nonselected GC B cells.

(A) GCs containing a ~5:5:90 mixture of B1-8hi CFP+ Ly75+/+ B cells (blue), B1-8hi GFP+ Ly75−/− (green), Ly75−/− nonfluorescent B cells, and OT-II DsRed+ T cells (red) were imaged by means of TPLSM after subcutaneous injection of αDEC-OVA. Contacts between OT-II T cells and either B1-8hi Ly75+/+ or Ly75−/− B cells are shown in yellow and gray, respectively. 40-μm deep, collapsed z-stacks (5-μm steps) are shown. Arrowheads indicate B1-8hi Ly75+/+ and Ly75−/− GC B cells interacting simultaneously with one OT-II T cell. (B) Quantitation of contact duration between OT-II T cells and B1-8hi Ly75+/+ or Ly75−/− B cells. Each data point represents a single cell, and red lines represent mean values. Percentages indicate events >5 min, marked by a dashed red box. (C) The number of B cell contacts (>0.5 min, left; >5 min, right) with OT-II T cells was normalized to the number of B cells in each group. Exp., experiment. Data in (B) and (C) were pooled from four independent experiments. *P < 0.0001; two-tailed Student’s t test.

Our experiments indicate that increased contact size and duration correlates with the amount of pMHCII presented by GC B cells and their subsequent clonal expansion in the DZ (6), yet how these interactions affect TFH T cell receptor (TCR) signaling is unknown. To examine the relationship between T-B contacts, TCR signaling, and B cell selection in the GC, we sought to measure intracellular Ca2+ levels in TFH cells. Traditional Ca2+ dyes cannot be used for this purpose because T cells divide extensively and dilute such tracers before reaching the GC (17). To circumvent this issue, we used mice expressing a genetically encoded Ca2+ indicator (GCaMP3), which changes its fluorescence intensity according to intracellular Ca2+ levels (18, 19). Lymphocytes from these mice showed an increase in intracellular fluorescence when stimulated with a Ca2+ ionophore or when stimulated through their antigen receptor (fig. S3 and movie S6).

Changes in Ca2+ fluorescence are best measured by comparison with a second dye that is insensitive to changes in intracellular Ca2+ levels. We therefore induced and imaged GCs, as described in Fig. 1A, using OT-II GCaMP3+ DsRed+ T cells and measured the ratio of GCaMP3:DsRed fluorescence intensity (fig. S4). To determine whether TFH cell Ca2+ content is associated with changes in cellular dynamics, velocities at successive time points were measured (instantaneous velocity) and correlated with intracellular Ca2+ content (17). Under steady-state conditions, spikes in Ca2+ content in GC TFH cells were rare (Fig. 3, A to C; fig. S5A; and movie S7). In contrast, αDEC-OVA injection increased the proportion of GC TFH cells with GCaMP3:DsRed ratios above 0.05 from 9 to 68% (Fig. 3, A to C, and movie S7) and the average ratio by 8.3-fold (fig. S5A). Single TFH cells that were actively engaged in B cell selection showed sustained increases in intracellular Ca2+ over time, which did not drop to control levels during the observation period (Fig. 3B). In addition, single TFH cells also displayed frequent Ca2+ spike transients (Fig. 3, B and D). Both sustained and transient increases in TFH intracellular Ca2+ levels were a result of TCR:pMHCII interactions (Fig. 3, B to D, and fig. S5A). We conclude that selection is associated with TCR:pMHCII–dependent increases in TFH cell Ca2+ content that were both transient and long-term.

Fig. 3 TFH cell Ca2+ signaling during B cell selection.

(A) GCs containing a 5:95 mixture of B1-8hi CFP+ Ly75+/+ B cells (light blue), B1-8hi Ly75−/− B cells (nonfluorescent), and OT-II DsRed+ GCaMP3+ T cells (red, Ca2+ low; yellow, Ca2+ high) were imaged by TPLSM in untreated mice or after subcutaneous injection of αDEC-OVA. (Bottom) Dynamic changes in GCaMP3 fluorescence in individual OT-II T cells over time. (B) Traces show changes in the GCaMP3/DsRed ratios over time for six single T cells in each condition. αI-Ab was injected intravenously after αDEC-OVA injection. (C) Scatter plots depict the GCaMP3/DsRed ratio versus instantaneous velocity as measured at successive 30-s intervals. Each dot represents a single cell at a single time point. Average velocity (V) and GCaMP3/DsRed ratios (R) of two to three experiments are indicated in red and green, respectively. (D) GCaMP3/DsRed ratio fluctuations in single cells (expressed as variance) under control and αDEC-OVA conditions. (E) GCaMP3/DsRed spikes of nine cells were synchronized. Traces of average GCaMP3/DsRed ratios and corresponding instantaneous velocities are shown. The synchronized spikes fall in the black rectangle. Error bars indicate SEM. (F) OT-II GCaMP3+ T cells and B1-8hi tdTomato+ Ly75+/+ B cells were imaged in GCs over time as in (A). Images correspond to movie S11. (G) Traces show GCaMP3 mean fluorescence intensities for five OT-II T cells in contact with B1-8hi tdTomato+ Ly75+/+ GC B cells. (H) Initiation of contacts were synchronized in 4 OT-II GCaMP3+ T cells, and the corresponding average of GCaMP3 fluorescence intensity was traced before and during the contact (au, arbitrary units). Data are representative of two or three mice imaged in two or three independent experiments. *P < 0.0001; two-tailed Student’s t test.

Increases in intracellular Ca2+ content in naïve T cells result in motility arrest and enhanced effector functions (17, 20, 21). By correlating intracellular Ca2+ content and instantaneous velocity in unperturbed GCs, we found that TFH cells moved at an average speed of 8.76 μm/min regardless of their Ca2+ content (Fig. 3C and fig. S5B). Consistently, in αDEC-OVA–targeted GCs the velocity of TFH cells with high- and low-Ca2+ content was decreased to an average of 6.24 μm/min, and few, if any, of the actively engaged TFH cells stopped moving or lost their morphological polarity while signaling (Fig. 3C, fig. S5B, and movies S7 to S9). However, there was no clear correlation between the level of Ca2+ increase and cell motility because both TFH cells with high- and low-Ca2+ content had the same average velocity (Fig. 3C, fig. S5B, and movie S7). Nevertheless, by synchronizing transient small Ca2+ peaks of several TFH cells, we found that these were precisely associated with a reduction in instantaneous velocity of ~2.5 μm/min (Fig. 3E, fig. S6, and movie S10).

Given these broad changes in Ca2+ content and the regulated T-B interactions, we sought to examine TFH intracellular Ca2+ levels during the formation of T-B contacts. To this end, we imaged selection in GCs containing a 95:5 mixture of B1-8hi nonfluorescent Ly75−/− B cells, tdTomato+ Ly75+/+ B cells, and OT-II GCaMP3+ T cells. Injection of αDEC-OVA induced the formation of T cell contacts with Ly75+/+ B cells that were associated with transient spikes in TFH intracellular Ca2+ content (Fig. 3, F and G, and movies S11 and S12). To determine whether these Ca2+ transients take place preferentially during T-B conjugate formation, we synchronized clearly isolated contact events and measured the average GCaMP3 fluorescence intensity in TFH cells before and during these interactions. Although the intracellular Ca2+ levels in TFH cells were high in αDEC-OVA–targeted GCs (Fig. 3B), an additional small increase at the onset of contact with selected B cells was detected (Fig. 3H and movie S10).

Changes in intracellular Ca2+ concentration control T cell effector functions, including the expression of IL-4 and IL-21 (20, 22), both of which are required for effective B cell immune responses (23). T effector cells can express either or both of these cytokines, and this multifunctionality is associated with enhanced immunity and vaccine responses (2428). To determine whether the long-term increase in TFH intracellular Ca2+ levels alters the quality of GC TFH cells, we examined T cells derived from IL4/IL21 double reporter mice expressing Il4-IRES-GFP (29) and Il21-IRES-Katushka (fig. S7). Polyclonal OVA-primed CD4+ T cells isolated from these mice were adoptively transferred into TCRβ-deficient mice along with B1-8hi Ly75+/+ and Ly75−/− B cells at a 5:95 ratio. Recipients were boosted with NP-OVA and injected 7 days later with either αDEC-CS or αDEC-OVA so as to induce selection. TFH cells were examined for cytokine expression 9 hours later by means of flow cytometry. αDEC-OVA injection resulted in a significant decrease in IL-4IL-21 TFH cells and a concomitant increase in IL-4+IL-21+ multifunctional TFH cells in the absence of cell division or a significant change in the number of single-cytokine-producing cells (Fig. 4, A and B, and fig. S8). Moreover, there was also a small but significant increase in the amount of IL-21, but not IL-4, produced by the double-positive cells, as measured by an increase in the mean fluorescence intensity of the reporters (Fig. 4, C and D). Thus, increased TFH Ca2+ signaling during B cell selection is associated with a rapid change in the quality of the GC TFH response, with increased development of multifunctional TFH cells producing Ca2+–dependent cytokines.

Fig. 4 Multifunctional TFH cells during B cell selection.

(A and B) CD4+ T cells derived from OVA-immunized Il4-IRES-GFP and Il21-IRES-Katushka double knock-in mice and a 5:95 ratio of B1-8hi Ly75+/+ and Ly75−/− B cells were transferred into TCRβ-deficient mice, before boosting with NP-OVA. After 7 days, mice were injected with αDEC-CS or αDEC-OVA, and lymph nodes were analyzed 9 hours later. The proportions of cytokine-expressing cells among TFH (CD8, B220, CD4+, CD44+, CD62L, CXCR5high, and PD-1high) subsets are indicated. (C and D) gMFI of IL-4 (C) or IL-21 (D) reporter expression. Each data point represents a single mouse, and lines represent mean values. Pooled data from three experiments each with three or four mice per condition. *P = 0.025; **P = 0.014; ***P = 0.02; two-tailed Student’s t test. NS, not significant.

Our results demonstrate that TFH cells respond to pMHCII on GC B cells during selection differently than the prolonged interactions of T cells with dendritic cells (DCs) in the T zone or with B cells at the T-B border (3036). The transient interactional dynamics of TFH cells in GCs allow them to continuously seek and find B cells presenting high levels of pMHCII and provide them with preferential help while permitting competitive opportunities for other GC B cells. This mode of B cell scanning allows TFH cells to interact with many cells presenting a range of pMHCII levels, rather than forming a prolonged contact with a single high-affinity B cell.

Our experiments show that the increased size and duration of contacts between GC TFH and selected B cells prolongs Ca2+ signaling and modifies the quality of the GC TFH response. These events induce coexpression of the Ca2+-dependent cytokines IL-4 and IL-21, which endow TFH cells with effector capabilities that facilitate high-affinity B cell selection.

Supplementary Materials

Materials and Methods

Figs. S1 to S8


Movies S1 to S12

  • * Present address: Immunology Institute, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA.

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank G. Victora for helpful discussions and suggestions, D. Bosque and T. Eisenreich for help with mouse colony management, A. Abadir for protein production, and K. Yao for technical help. The data reported in this manuscript are tabulated in the main paper and in the supplementary materials. Z.S. is a Human Frontiers of Science Program fellow (reference LT000340/2011-L). A.D.G. is supported by NIH Medical Scientist Training Program grant T32GM07739 to the Weill Cornell/Rockefeller/Sloan-Kettering Tri-Institutional MD-PhD Program. Support for the Rockefeller University multiphoton microscope was granted by the Empire State Stem Cell Fund through New York State Department of Health contract C023046. This work was supported by NIH grants AI037526-19, AI072529-06, and AI100663-02 to M.C.N. and NIH grants AR40072-24 and AR053495-08 (the content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH) and the Alliance for Lupus Research to J.E.C. M.C.N. and R.A.F. are HHMI investigators.
View Abstract

Navigate This Article