Report

T Follicular Helper Cell Dynamics in Germinal Centers

See allHide authors and affiliations

Science  09 Aug 2013:
Vol. 341, Issue 6146, pp. 673-677
DOI: 10.1126/science.1241680

Help Shared

Germinal centers are specialized structures within lymph nodes, where B cells undergo the changes required to produce high-affinity antibodies. This process relies on T follicular helper (Tfh) cells. The dynamic properties of Tfh cells and how they affect the selection of B cells, however, are not well understood. Using two-photon laser scanning microscopy of mouse lymph nodes, Shulman et al. (p. 673, published online 25 July) find that Tfh cells are not restricted to a single germinal center, but instead emigrate into neighboring germinal centers within the same lymph nodes. Furthermore, newly activated T cells can enter already established germinal centers and presumably influence ongoing B cell selection and differentiation. Such active movement may ensure maximal diversification of the B cell response and promote the production of high-affinity antibodies.

Abstract

T follicular helper (TFH) cells are a specialized subset of effector T cells that provide help to and thereby select high-affinity B cells in germinal centers (GCs). To examine the dynamic behavior of TFH cells in GCs in mice, we used two-photon microscopy in combination with a photoactivatable fluorescent reporter. Unlike GC B cells, which are clonally restricted, TFH cells distributed among all GCs in lymph nodes and continually emigrated into the follicle and neighboring GCs. Moreover, newly activated TFH cells invaded preexisting GCs, where they contributed to B cell selection and plasmablast differentiation. Our data suggest that the dynamic exchange of TFH cells between GCs ensures maximal diversification of T cell help and that their ability to enter ongoing GCs accommodates antigenic variation during the immune response.

T cells play a pivotal role in affinity maturation by selecting B cells to enter the germinal center (GC), regulating GC positive selection, and directing B cell differentiation to plasma cells and memory B cells (16). These events are orchestrated by a specialized population of GC-resident T follicular helper (GC-TFH) cells that develop in concert with GC B cells (713). TFH cells express the chemokine receptor CXCR5, the CD28 family members ICOS and PD-1, the transcription factor Bcl-6, and the cytokines interleukin-4 (IL-4) and IL-21; many of these are required for GC B cell survival and differentiation (7). Less is known about the dynamic properties of TFH cells and how they might affect GC B cell selection.

To document the kinetics of T cell expansion and localization during the GC reaction, we examined whole-mounted lymph nodes by means of two-photon laser scanning microscopy (TPLSM) (14, 15). In agreement with previous reports (16), large numbers of antigen-specific T cells were found in the T cell zone 3 days after immunization with 4-hydroxy-3-nitrophenylacetyl-ovalbumin (NP-OVA). T cells then spread throughout the lymph node, including B cell follicles and nascent GCs, by day 5 (Fig. 1A). T cells began to concentrate in GCs only after 8 to 11 days (Fig. 1A), coinciding with the peak of T cell help to GC B cells (5). Thus, unlike B cell clones, which are thought to expand in a confined microanatomical region to produce pauciclonal GCs (17, 18), responding T cells are initially evenly distributed throughout the entire lymph node and accumulate in GCs only after they have coalesced.

Fig. 1 Expansion kinetics and anatomical distribution of TFH cells during germinal center formation.

(A) Left: 3 × 104 GFP+ OVA-specific T cells (OT-II) and 1.5 × 106 tdTomato+ B1-8hi B cells (corresponding to 1.5 × 105 Igλ+, NP-specific B cells) were transferred into wild-type mice 1 day before subcutaneous immunization with NP-OVA in alum. Popliteal lymph nodes were explanted at 0, 3, 5, 8, and 11 days after immunization and subjected to whole–lymph node scanning by TPLSM. Collapsed Z-stacks (total volume 50 to 100 μm; 5-μm steps) are shown. Scale bars, 300 μm. Right: Quantitation of T cell density and density ratio (proportion of T cells within GC to T cells outside of GC) in two independent experiments. (B and C) Expression of TFH markers on PAGFP+ OT-II T cells photoactivated in different lymph node regions, as described in fig. S1. Each symbol represents one experiment. gMFI, geometric mean fluorescence intensity.

TFH and GC-TFH cells are commonly defined on the basis of functional properties and expression of cell surface markers (8, 19) rather than anatomical localization. To verify the correspondence between surface phenotype and microanatomical location, we used photoactivatable green fluorescent protein (PAGFP) to label cells within spatially restricted areas (3). Flow cytometric analysis of photoactivated OT-II T cells (fig. S1, A to C) showed that CXCR5 and PD-1 expression were highest among T cells physically inside the GC and lowest among T cells in the paracortex, whereas T cells in the follicle outside the GC showed intermediate levels of expression of these molecules (Fig. 1, B and C, and fig. S1D). In contrast, ICOS expression was comparable in all three locations (Fig. 1, B and C, and fig. S1D). Similar results were obtained by photoactivation of endogenous polyclonal T cells (fig. S2). Thus, although CXCR5 and PD-1 expression can be used to distinguish between GC-TFH and paracortical T cells, they cannot be used to definitively distinguish follicular TFH cells from either of these populations.

To determine whether GC-TFH cells, like their B cell counterparts, are clonally restricted within individual GCs (17, 18), we immunized mice that had received a mixture of OT-II T cells expressing one of three fluorescent proteins and visualized GCs 10 days after immunization (Fig. 2A). As suggested by our initial observations, the proportion of T cells of each color within individual GCs was constant across all GCs in the same lymph node, indicating that T cells are not clonally restricted within GCs (Fig. 2, A and B, and fig. S3A).

Fig. 2 GC-TFH cells in individual germinal centers are not clonally restricted.

(A) A total of 3 × 105 OT-II T cells expressing CFP, GFP, or DsRed (~105 of each color) and 3 × 106 to 5 × 106 nonfluorescent B1-8hi B cells (corresponding to 3 × 105 to 5 × 105 Igλ+, NP-specific cells) were transferred into wild-type recipients 1 day before subcutaneous immunization with NP-OVA in alum. Draining lymph nodes were explanted 10 days later and imaged by TPLSM. Follicular dendritic cell networks were labeled by injection of a mixture of NP-tdTomato and NP-YPet (orange). Central image shows a collapsed Z-stack, 50 μm deep (10-μm steps), of a whole lymph node containing multiple GCs, magnified in the outer panels. Scale bars, 300 μm (large panel), 50 μm (side panels). (B) Quantitation of data as in (A) across multiple lymph nodes from different mice. Each bar represents a single GC. Numbers on top of bars indicate the number of T cells counted in each GC. (C) T and B cells as described in (A) were transferred to H-2Aβ-GFP mice to allow visualization of early GCs, and lymph nodes were imaged 5 days after immunization with NP-OVA in alum. Image of one GC and quantitation of multiple GCs are shown. Scale bar, 100 μm. Each set of experiments was performed twice (additional images and quantitation in fig. S3).

This pattern of T cell distribution might result from initial colonization by multiple clones and/or from T cell exchange between GCs. To examine the first of these possibilities, we analyzed early GCs. GC-TFH cell distribution was homogeneous across GCs even at the earliest observable time point, which suggests that individual GCs are indeed colonized by several distinct T cell clones (Fig. 2C and fig. S3, B and C). We confirmed this by transferring a much smaller number of T cells (3 × 104 total T cells per mouse) at a 19:1 ratio of GFP:DsRed cells (fig. S4), as well as by reconstituting T cell receptor β–deficient mice with mixtures of genetically labeled polyclonal CD4+ T cells (fig. S5). In both cases, T cells were evenly distributed across GCs after immunization (figs. S4 and S5). Thus, we find no evidence that GC-TFH cells are clonally restricted to individual GCs; instead, GCs appear to be colonized by a heterogeneous population of T cells representing all responding T cell clones.

To determine whether GC-TFH cells can also exchange between GCs by migration, we imaged these cells by TPLSM. Short-term intravital movies showed that GC-TFH cells occasionally migrated out of GCs (movie S1). Whether these emigrants are in fact leaving the GC rather than making a transient foray into the follicle could not be determined because of constraints in the duration of conventional intravital imaging movies (20).

To observe GC-TFH cell migration for longer periods, we photoactivated PAGFP–OT-II TFH cells within single GCs (fig. S6A). Photoactivation did not alter the motility of PAGFP+ OT-II TFH cells relative to controls (movie S2), and, because GC-TFH cells were in G1 phase (fig. S6B), activated PAGFP was not diluted by cell division, allowing for long-term cell tracking. Immediately after photoactivation, PAGFP+ OT-II TFH cells were restricted to the targeted region (fig. S6C). In contrast, ~20 hours after activation, 32% of photoactivated GC-TFH cells were found outside the original GC (Fig. 3, A and B). Emigrating cells were found almost entirely within neighboring follicles and other GCs, and only very rarely (1 cell out of 487 analyzed) in the T cell zone (Fig. 3B, fig. S6D, and movie S3). Consistent with their near-absence from the T cell zone, we could detect only a very small number of photoactivated cells in the blood or pooled distal lymphoid organs of mice examined 36 hours after photoactivation (1 to 4 cells per mouse, seven mice in three independent experiments; fig. S6E). Transfer of PAGFP+ OT-II T cells before priming with OVA did not substantially alter GC size or the ratio of T cells to B cells in the GC (fig. S6, F and G). Thus, GC-TFH cell emigration to the follicle and neighboring GCs cannot be attributed to a nonphysiological density of TFH cells.

Fig. 3 GC-TFH cells exchange between neighboring germinal centers.

(A) GCs containing OT-II PAGFP+ TFH cells were generated as in fig. S6A, and the light zone of a single GC was photoactivated within a living mouse. Central image shows a collapsed Z-stack of the cortical half (depth 250 μm) of a whole–lymph node scan taken ~20 hours after photoactivation. Outer panels (i-vi) are high-magnification images of the regions indicated in the center panel; panel vii shows the original photoactivated GC. Scale bars, 300 μm (center panel), 20 μm (panels i to vi), 30 μm (panel vii). Green spots are placed over photoactivated OT-II T cells in low-magnification panels to improve visualization. (B) Quantitation of three experiments as in (A) (95, 184, and 208 photoactivated cells analyzed in three independent experiments). Each color represents one experiment. Average distance covered by photoactivated cells was 334 μm; the farthest 10% of cells traveled on average 790 μm. (C) Single GCs in popliteal lymph nodes containing PAGFP+ OT-II TFH cells were photoactivated in vivo as in (A), then explanted and dissected as shown in fig. S6. Flow cytometry plots show PD-1 and CXCR5 expression in emigrant GC-TFH cells (pooled data from two mice). Data are representative of five mice from three independent experiments.

To determine the phenotype of emigrant GC-TFH cells, we microdissected lymph nodes into two fragments, one of which contained the original photoactivated GC (Fig. 3C and fig. S7, A and B). Immediately after activation (0 hours), the lymph node fragment contralateral to the activated GC contained no photoactivated cells (fig. S7C). In contrast, after 24 hours, photoactivated cells corresponding to GC-TFH cells that had emigrated from the original GC could also be found in the nonphotoactivated half of the lymph node (Fig. 3C). In agreement with the lymph node scans (Fig. 3A), emigrant GC-TFH cells remained phenotypically similar to authentic GC-TFH cells, although expression of PD-1 and CXCR5 was slightly decreased (Fig. 3C and fig. S7, D and E). We conclude that GC-TFH cells emigrate and redistribute throughout the follicles and to neighboring GCs, but only rarely leave the follicle to enter circulation in the time frame analyzed.

The exchange of T cells between GCs suggested a hitherto unappreciated dynamic equilibrium between T cells in the follicles and in different GCs. We reasoned that if GCs are in fact open to TFH cell exchange, then newly activated T cells may also be able to join an ongoing GC reaction. To examine this possibility, we transferred GFP-expressing OT-II cells into wild-type recipients that were then primed with OVA to induce OT-II expansion. These mice then received B1-8hi cells expressing cyan fluorescent protein (CFP–B1-8hi cells) and were immunized with NP–chicken gamma globulin (CGG) in alum to produce GCs in which endogenous CGG-specific TFH cells provide help for CFP–B1-8hi cells. The immunized mice were then boosted subcutaneously with soluble NP-OVA to trigger invasion of these GCs by OT-II cells (fig. S8A). Before the boost, small numbers of GFP–OT-II cells were found scattered throughout the T cell zone, occasionally entering existing GCs for brief periods (Fig. 4A, fig. S8B, and movie S4). One day after immunization, GFP+ cells could be seen proliferating in foci of T and B cells at the follicle–T zone border (Fig. 4A and movie S5). By day 3 after boost, GFP–OT-II T cells were found throughout the lymph node follicles and within preexisting GCs (Fig. 4A and movie S6). All GCs evaluated contained invading GFP–OT-II T cells, indicating that these cells were indeed entering existing GCs rather than producing GCs de novo. Accordingly, control mice boosted with NP-OVA without preimmunization with NP-CGG had no GCs at this time point (fig. S8C). Over time, GFP–OT-II T cells began to accumulate progressively within GCs, so that by day 11 after boost, the newly activated T cells were mostly concentrated inside these structures (Fig. 4A, fig. S8B, and movie S6), interacting actively with GC B cells (Fig. 4B and movie S7). Therefore, ongoing GCs are open to invasion by newly activated TFH cells.

Fig. 4 Newly activated TFH cells can enter ongoing germinal center reactions.

(A) Kinetics of invasion of ongoing GCs by newly activated T cells as described in fig. S8A. Top: Montage showing whole–lymph node TPLSM scans. Collapsed Z-stacks are 90 μm deep (pre and day 1) or 40 μm deep (days 3 to 11). GCs containing CFP+ B1-8hi B cells (blue) and NP-tdTomato–coated follicular dendritic cells (red) are circled with dotted lines. Green spots are placed over GFP+ T cells to improve visualization. Background outside lymph nodes was deleted. Bottom: Higher-magnification images of GCs circled in yellow in the top panel. Scale bars, 0.5 mm (top), 100 μm (bottom). (B) Time series showing cognate T cell–B cell interactions in an invaded GC on day 11 after boost. Scale bar, 20 μm. (C and D) Recipient mice were treated with αDEC205-OVA or PBS 6 days after NP-OVA boost as outlined in fig. S8D. Flow cytometry plots show proportion of Ly75+/+ and Ly75–/– B1-8hi GC B cells (C) or percentage of plasmablasts among all single cells (D) 3 days after αDEC205-OVA treatment. (E) Flow cytometry plots comparing the proportion of GC cells (B220+Fas+CD38) in mice treated as outlined in fig. S8E or receiving PBS instead of the NP-OVA boost at day 0. The rightmost panels show means ± SEM of two or three experiments.

To determine whether invading TFH cells contribute to the GC reaction, we measured their ability to provide help to B cells selectively expressing high levels of cognate peptide–major histocompatibility complex (pMHC) by using an antibody to DEC-205 (encoded by Ly75) fused to OVA (αDEC205-OVA) to target antigen to GC B cells (3, 4). OVA-specific GC-TFH cells were induced to invade a CGG-specific GC containing a mixture of 85% Ly75−/− CD45.1/2 and 15% Ly75+/+ CD45.1/1 B1-8hi B cells, and cognate pMHC expression was artificially increased on the Ly75+/+ B cells by injection of αDEC205-OVA (fig. S8D). Whereas phosphate-buffered saline (PBS) or control antibody had no effect, injection of αDEC205-OVA resulted in selective and marked expansion of the Ly75+/+ but not Ly75−/− GC B cells, and an increase in plasmablast frequency by nearly an order of magnitude (Fig. 4, C and D). We conclude that T cells that invade ongoing GCs actively participate as helper T cells and positively select B cells expressing high levels of pMHC.

Boosting ongoing NP-CGG–specific GCs with NP-OVA (as detailed in Fig. 4A) resulted in a factor of 5 increase in the proportion of GC B cells 5 days after boost (from 0.9 to 5.1%), whereas the proportion of transferred to endogenous B cells remained constant (Fig. 4E). Thus, GC invasion by newly activated TFH cells can augment ongoing GC reactions.

Physical restriction of responding B cells to a single GC minimizes competition between B cells in different GCs, thereby preventing the immune response from converging on a single dominant clone (21). In contrast, free movement of TFH cells between GCs may be advantageous, in that it ensures diversified and robust support for B cell clonal expansion and affinity maturation (2224). Thus, diversity in the antibody response appears to be favored by restricting B cell clones to single GCs while exposing them to a multiplicity of different T cell clones that transit between different GCs.

In addition to polyclonal colonization and dynamic exchange between GCs, newly activated T cells can also enter ongoing GCs. This raises the possibility that T cells that are activated late in the immune response can join and potentially prolong the GC reaction. Such a feature may be an important advantage in responding to chronic infections and/or pathogens that diversify their antigenic epitopes during infection, such as HIV.

Supplementary Materials

www.sciencemag.org/cgi/content/full/science.1241680/DC1

Materials and Methods

Figs. S1 to S8

Table S1

Movies S1 to S7

References (2529)

References and Notes

  1. See supplementary materials on Science Online.
  2. Acknowledgments: We thank H. Ploegh, K. Strijbis, and H. Sive for mice and equipment use and E. Browne for helpful discussion. The data presented are tabulated in the main paper and in the supplementary materials. 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. Supported by NIH Medical Scientist Training Program grant T32GM07739 to the Weill Cornell/Rockefeller/Sloan-Kettering Tri-Institutional MD-PhD Program (A.D.G.), NIH grant 5DP5OD012146-02 and March of Dimes Foundation Basil O’Connor Award (G.D.V.), and NIH grants AI037526-19, AI072529-06, and AI100663-01 (M.C.N.). Z.S. is a Human Frontiers of Science fellow. G.P. is a Swiss National Science Foundation fellow. M.C.N. is an HHMI investigator.
View Abstract

Navigate This Article