Landscape of T Follicular Helper Cell Dynamics in Human Germinal Centers Emmanuel Donnadieu, Kerstin Bianca Reisinger, Sonja Scharf, Yvonne Michel, Julia Bein, Susanne Hansen, This information is current as Andreas G. Loth, Nadine Flinner, Sylvia Hartmann and of September 28, 2021. Martin-Leo Hansmann J Immunol published online 22 July 2020 http://www.jimmunol.org/content/early/2020/07/21/jimmun

ol.1901475 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2020/07/21/jimmunol.190147 Material 5.DCSupplemental http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 28, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2020 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published July 22, 2020, doi:10.4049/jimmunol.1901475 The Journal of Immunology

Landscape of T Follicular Helper Cell Dynamics in Human Germinal Centers

Emmanuel Donnadieu,*,1 Kerstin Bianca Reisinger,†,1 Sonja Scharf,† Yvonne Michel,† Julia Bein,†,‡ Susanne Hansen,† Andreas G. Loth,x Nadine Flinner,{ Sylvia Hartmann,†,‡ and Martin-Leo Hansmann†,‡,{

T follicular helper (Tfh) cells play a very important role in mounting a humoral response. Studies conducted in mouse models have revealed with good kinetic and spatial resolution the dynamics of these cells in germinal centers (GC) and their cross-talk with B cells upon an immune response. However, whether a similar migratory behavior is performed by human Tfh cells is unclear, as technology to track them in situ has been lacking. In this study, we combined traditional immunohistochemistry and real-time fluorescent imaging approaches on fresh human adenoid slices to provide static and dynamic information on Tfh cells. Our data indicate that GC light zones are composed of two distinct areas in terms of Tfh cell distribution and migration. In the outer GC Downloaded from light zones, Tfh cells migrate actively and with a high ability to form dynamic clusters showing intense and rapid reorganization. In these outer regions, Tfh cells demonstrate multiple interactions between each other. Conversely, in central regions of GC light zones, Tfh cells are much more static, forming long-lasting conjugates. These findings reveal for the first time, to our knowledge, the dynamic behavior whereby Tfh cells migrate in human GC and highlight the heterogeneity of GC for Tfh cell motility. The Journal of Immunology, 2020, 205: 000–000. http://www.jimmunol.org/

subset of CD4 T cells, T follicular helper (Tfh) cells a pre-Tfh phenotype with the expression of canonical markers specialize in helping B cells to produce Abs. Mature Tfh such as CXCR5, PD-1, ICOS, and Bcl6. These cells migrate to the A cells localize to follicles in secondary lymphoid organs, T–B junction of secondary lymphoid follicles where the first T–B where they provide help to B cells in germinal centers (GC). interaction takes place. Tfh then move in the GC, a microana- Various help mechanisms have been reported that include contact- tomical structure in which proliferating Ag-specific B cells un- dependent process, such as CD40 ligands, and soluble molecules, dergo Ig affinity maturation, class-switch recombination, and such as IL-21. These contact-dependent and soluble signals support differentiation into long-lived plasma cells and memory B cells. survival and differentiation of GC B cells [for a review, see (1)]. Time-lapse two-photon laser-scanning microscopy experiments by guest on September 28, 2021 Recent work has significantly expanded our understanding of performed in mouse models have clarified how Tfh move and how when Tfh cells are formed and how these cells contribute to they interact with B cells in GC (3–5). A striking feature that help B cells (2). Tfh cell differentiation is initiated when naive emerged in the real-time imaging studies was that GC Tfh cells CD4+ T cells in the T cell zone encounter activated Ag-presenting were highly motile and frequently engage with B cells (3). Two dendritic cells. Activated T cells primed by dendritic cells acquire very distinct T–B conjugates were noticed. Unlike the frequent formation of stable conjugates during initial T cell–B cell inter- action at the T–B junctions, GC Tfh cells interact briefly with *De´partement Immunologie, Inflammation, et Infection, Institut Cochin, INSERM, U1016, CNRS, UMR8104, Universite´ de Paris, F-75014 Paris, France; †Dr. Senckenberg GC B cells. Only a low proportion of GC B cells form stable Institute of Pathology, Goethe University, 60590 Frankfurt am Main, Germany; conjugates with GC Tfh (3, 4). It was shown that these B cells ‡Reference and Consultant Center for Lymph Node and Lymphoma Diagnostics, x engaging Tfh in stable conjugates were capable of collecting and 60590 Frankfurt, Germany; Department of Otolaryngology, Head and Neck Sur- gery, University Hospital, 60590 Frankfurt am Main, Germany; and {Frankfurt presenting more Ag to Tfh cells. These prolonged interactions Institute for Advanced Studies, 60438 Frankfurt, Germany were responsible for promoting the positive selection of B cell 1E.D. and K.B.R. contributed equally to this work. clones (6). ORCIDs: 0000-0002-4985-7254 (E.D.); 0000-0002-5439-3909 (K.B.R.); 0000-0002- Several elements have been shown to control Tfh cell migration 0300-7027 (S.S.); 0000-0003-3424-1091 (S.H.). and interaction with B cells. One can mention the costimulatory Received for publication December 17, 2019. Accepted for publication June 28, receptor ICOS, which regulates Tfh recruitment to follicles but also 2020. regulates cognate T–B interactions in the GC (4, 5). A role of the This work was supported by Deutsche Forschungsgemeinschaft (FOR1961 Control-T chemokine receptor CXCR5, which binds to CXCL13 produced consortium, Grants HA1284/7-2 and HA6145/2-1). by follicular dendritic cells (FDC), stromal cells that build the M.-L.H. and E.D. designed the study; K.B.R., Y.M., S.H., N.F., J.B., and E.D. performed research; A.G.L. provided human samples; S.S. and E.D. analyzed data; architecture of GC light zones, was also demonstrated. S.H. provided insights and advice; M.-L.H. and E.D. wrote the manuscript. These elegant studies, which described the dynamics of Tfh in Address correspondence and reprint requests to Dr. Emmanuel Donnadieu, De´partement intact GC, have all been performed in mouse models that did not Immunologie, Inflammation, et Infection, Institut Cochin, Universite´ de Paris, 22 rue entirely reflect the true situation in humans. For instance, most ´ Mechain, 75014 Paris, France. E-mail address: [email protected] studies relied on the use of a cognate Ag, which usually binds to The online version of this article contains supplemental material. high-affinity BCRs and TCRs. In addition, several important Abbreviations used in this article: 3D, three-dimensional; FDC, follicular dendritic questions remain to be addressed concerning the role of different cell; GC, germinal center; Tfh, T follicular helper. subsets of Tfh cells and whether Tfh cells communicate to cells Copyright Ó 2020 by The American Association of Immunologists, Inc. 0022-1767/20/$37.50 other than just B cells.

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1901475 2 IMAGING OF Tfh CELLS IN HUMAN LYMPHOID TISSUES

In human lymphoid tissues, our knowledge of Tfh cells is mainly and Alexa Fluor 405–anti-human fibronectin (clone HFN 7.1; Novus Bi- based on flow cytometry experiments or immunostained sections ologicals). All Abs were diluted in RPMI 1640 without Phenol red and m for different lymphoid subpopulations (7). These studies have used at a concentration of 10 g/ml. To concentrate the Abs on the tissue, a stainless steel ring was placed to the agarose surrounding the slice. In some highlighted some common features but also some notable differ- experiments, fresh slices were labeled for 15 min at 37˚C with DAPI and ences between mouse and human Tfh cells (8). In addition, these DRAQ5 to assess the nuclei of dead and live cells. are static evaluations that may not reflect the dynamic process T cells were imaged with a Leica SP8 confocal microscope (Leica ongoing during an immune response. Microsystems) equipped with a 37˚C thermostated chamber. The temper- ature was constant, and oxygen was controlled by self-written appropriate In this study, we tried to combine the advantage of cell differ- software. For dynamic imaging, adenoid slices were secured with a entiation by immunohistochemistry with a dynamic approach in stainless steel slice anchor (Warner Instruments) and perfused at a rate of human fresh lymphoid tissues that we have established in human 0.8 ml/min with a solution of RPMI 1640 without Phenol red, bubbled with tumors (9, 10) and more recently in human adenoids (11). Tissue 95% O2 and 5% CO2. Ten minutes later, images from a first microscopic 3 3 3 slices enabled us to monitor for the first time, to our knowledge, field were acquired with either a 10 ,25 ,or63 objective. For four- dimensional analysis of cell migration, stacks of 10–12 sections (z step = human Tfh cells in space and time. 5 mm) were acquired every 10–20 s for 20 min at depths up to 80 mm. Our results reveal the existence of two distinct navigational Regions were selected for imaging when B cell follicles (CD19 or CD35) modes of Tfh cells that depend on their localization. In outer GC and PD-1+ were simultaneously present in the same microscopic field. For light zones, Tfh migrate rapidly and exhibit a prominent ability to most of the adenoids included in the study, between two and four micro- scopic fields were selected for time-lapse experiments. briefly interact with each other. Conversely, in more central GC areas, Tfh cells are mostly static and engage in long-lived inter- Data analysis action with other Tfh cells. A three-dimensional (3D) image analysis was performed on x, y, and z Downloaded from planes using Imaris 9.2 (Bitplane AG). First, superficial planes from the Materials and Methods top of the slice to 15 mm in depth were removed to exclude T cells located Human samples near the cut surface. Cellular motility parameters were then calculated using Imaris. Tracks .10% of the total recording time were included in the Fresh human adenoids were obtained after routine adenectomy. This tissue analysis. 3D reconstruction of the sequential z series was performed using turns out to be better preserved than tonsils, which, in our experience, the surface tool of Imaris. include less damage, such as necrosis and fibrosis. This procedure was http://www.jimmunol.org/ approved by the local ethical committee of Frankfurt University Hospital Immunohistochemistry (number 376/16) and by declared consents. Formalin-fixed adenoid samples embedded in paraffin were cut into slices Dynamic imaging of Tfh in human adenoid slices (18–33 mm) using a microtome. Afterwards, we deparaffined the sections for 10 min in a xylene bath. For rehydrating, descending ethanol series Experiments were performed with fresh adenoid specimens obtained 30 min were used before incubating the slices twice in 100% ethanol at room to 1 h after adenectomy. Human adenoid slices were prepared as described temperature. Then, slices were incubated twice for 1 min in 96% ethanol; previously (11). Briefly, samples were embedded in 5% low-gelling- afterwards they were incubated with 70% ethanol for 1 min, followed by temperature agarose (Type VII-A; Sigma-Aldrich) prepared in PBS. aqua dest. for 5 min. Epitope retrieval was performed by pressure cooking Three hundred and fifty micrometer slices were cut with a vibratome the loaded microscopy slides for 90 s in EDTA (pH 80) before staining

(VT1000S; Leica Biosystems) in a bath of ice-cold PBS. Slices were sections were transferred into distilled water. Monoclonal mouse anti- by guest on September 28, 2021 transferred to 0.4-mm organotypic culture inserts (Millicell; Millipore) in human PD-1 (clone NAT 105; Abcam), anti-human FOXP3 (clone 35-mm petri dishes containing 1.1 ml of RPMI 1640 without Phenol red. 236A/E7; Abcam), anti-human CD35 (clone EP197; Bio SB), anti-human Live vibratome sections were stained for 15 min at 37˚C with the following CD4 (clone 4B12; Dako), and polyclonal rabbit anti-human active caspase Abs: Alexa Fluor 647–anti-human PD-1 (clone EH12.2H7; BioLegend), 3 (Diagnostic BioSystems) served as primary Abs. As a secondary Ab, we FITC–anti-human CD35 (clone E11; BioLegend), FITC–anti-human IgD used the VectaFluor Ready-to-Use DyLight 594 Ab Kit that included (clone IA6-2; BioLegend), Alexa Fluor 647–anti-human CD3 (clone normal horse serum (2.5%) for blocking and a goat anti-mouse Ig or goat UCHT1; BD Biosciences), Alexa Fluor 488–anti-human CD19 (clone HIB anti-rabbit amplifier Abs. Anti-goat IgG is the dye labeled Ig G. (cat. no. 19; BioLegend), Pacific Blue–anti-human CD20 (clone 2H7; BioLegend), DK-2594; Vector Laboratories, Burlingame, CA.). Nucleic acid staining

FIGURE 1. Tfh cell distribution in human GC. (A) CD35 and IgD immunofluorescent staining in one human adenoid paraffin section. (B) PD-1 and CD4 immunostaining in one human adenoid paraffin section. (C) PD-1 and CD20 immu- nofluorescent staining in one fresh human adenoid vibratome section. Dotted lines indicate the outer edge of the GC and the border between peripheral and central GC regions. (D) PD-1 and CD35 immunofluorescent staining in one human adenoid paraffin section. Dashed lines indicate the outer edge of the GC. Boxed areas one and two are shown at higher magnification (right). Scale bar, 20 mm. Results are repre- sentative of at least three human samples. MZ, mantle zone. The Journal of Immunology 3 Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 2. Tfh cells exhibit a heterogeneous distribution in human GC. (A) PD-1 immunostaining in one human adenoid paraffin section. Right, Zoom showing the presence of single cells and clusters of cells. (B) Same image as (A) but after automated analysis showing small (left) and large (right) objects in yellow. Objects were identified based on PD-1 staining. Right, Histogram showing the size distribution of all objects in the image shown in (B). Results are representative of similar analysis performed on seven different GCs from four adenoids. was performed with DAPI (D9542; Sigma-Aldrich, St. Louis, MO). A distributed. In particular, focal PD-1+ cell clusters became evident, Vector TrueVIEW Quenching Kit (Vector Laboratories, Burlingame, CA) especially in the outer regions of GC (Figs. 1D, 2A, Supplemental was used to minimize autofluorescence. Protocols were performed as rec- Fig. 1). We then performed an automated image analysis based ommended by the manufacturer. on the size of objects positive for PD-1. Our data revealed sev- Statistical analysis eral populations that can be classified into large objects, likely An unpaired student t test was performed to determine statistical signi- composed of clusters of several Tfh lymphocytes, and into small ficance. All statistical analyses were performed with Prism 6 (GraphPad). objects, likely composed of single lymphocyte (Fig. 2B). Inter- The following p values were used to indicate significance: *p , 0.05, estingly, Tfh cell clusters were preferentially distributed at the , , **p 0.01, and ***p 0.001. GC periphery, whereas single Tfh cells were in the center. Results However, with this magnification one cannot figure out how many cells constitute these clusters and whether these cells move In human lymphoid tissues, GC Tfh cells exhibit a or are static. We thus decided to combine confocal microscopy nonuniform distribution and a preparation of fresh tissue slices to characterize in detail We used conventional immunohistochemistry and immunofluo- the dynamics of Tfh cells in human adenoids. rescence staining on paraffin samples to assess the distribution of Tfh cells in human adenoids. As shown in Fig. 1A, several fol- Tracking of T cells in human lymphoid tissue slices using licular compartments can be distinguished, including a mantle confocal microscopy zone positive for IgD and a light zone positive for CD35. CD35 is Over the last several years, we have established and used an ap- a specific marker of FDC that preferentially populate GC light proach based on vibratome slices made from fresh human tissues, zones. Tfh cells, characterized as being positive for CD4 and PD-1 including adenoids (9–11). This technique enables the monitoring on consecutive sections, were found concentrated in the outer zone of immune cells and assessment of their motile behaviors in a (Fig. 1B). This peripheral zone was also found denser for CD20- preserved tissue environment. The tracked cells can be either positive B cells (Fig. 1C). Looking in more detail with high mag- immune cells (e.g., T cells) that have been previously purified, nification, we confirmed that Tfh cells were not homogenously loaded with a fluorescent dye, and plated onto fresh slices, or they 4 IMAGING OF Tfh CELLS IN HUMAN LYMPHOID TISSUES Downloaded from http://www.jimmunol.org/

FIGURE 3. T cell tracking in human fresh adenoid slices. Tracks of FIGURE 4. B cell behavior in human fresh adenoid slices. (A) Snap- individual endogenous CD3 T cells (green) in relation to CD19 B cells shots of a time lapse with CD20 B cells (blue) in one human adenoid slice. (red) and fibronectin (blue) in one human adenoid slice. Tracks during the PD-1 Tfh cells are superimposed in the bottom picture. Circles in dashed whole recording (20 min) are in light gray. See also Supplemental Video 1. lines indicate clusters of B cells with low motions. White stars indicate Results are representative of three independent experiments. fast-moving B cells. See also Supplemental Video 2. (B) Snapshots of a time lapse with one PD-1–positive cells engaging with a CD20-positive B cell (white dot). Results shown are representative of more than five inde- can be endogenous cells fluorescently labeled with directly cou- pendent experiments. by guest on September 28, 2021 pled Abs. In this study, we used confocal microscopes to monitor movement patterns and speeds of fluorescent resident T cells and, especially, Tfh cells. composed of clusters of several B lymphocytes that remain at- To this end, fresh human adenoids were rapidly (within 30–60 tached during the 15-min recording. These clusters were not im- min) obtained after routine adenectomy. Slices of 350 mm were motile but exhibited a slow motion. In contrast, individual B cells made using a vibratome and subsequently labeled for 15 min with that migrate rapidly were also in evidence. Tfh cells were found directly coupled fluorescent Abs. Fig. 3 and Supplemental Video 1 interacting transiently with many B cells (Supplemental Video 2). illustrate a low-magnification overview of the distribution and Fig. 4B illustrates a dynamic conjugate formed between a Tfh cell dynamics of T lymphocytes labeled with anti-CD3 Abs. The and a B cell. Rapidly after its contact with the B cell, the Tfh cell follicle was identified with anti-CD19 Abs. Anti-fibronectin Abs showed marked shape changes characterized by cell protrusions were used to define structural determinants. The T zone enriched that surrounded the B cell body. in CD3 lymphocytes presents a typical reticular network made of Tfh cells in GC light zones of human lymphoid tissues exhibit thin and spaced fibers of fibronectin. Our data indicate that many two different migratory patterns of the T cells in the T zone migrate rather actively (mean velocity of 6 mm/min) as already described in mouse models (12, 13). We We decided thereafter to analyze Tfh cell velocity in GC light zones also noticed the presence of motile CD3 T cells, likely Tfh cells, using the tracking software Imaris. As shown in Fig. 5 and in the GC, and we decided to concentrate our efforts on these cells Supplemental Video 3, two different patterns of Tfh cell motility using higher-magnification confocal microscopy. were noticed. In the outer GC light zones, Tfh cells migrate rapidly, with a mean velocity of 5.8 mm/min. Conversely, in Tfh and B cell migration in GC light zones of human central regions Tfh cells were more static, with a mean velocity of lymphoid tissues 0.8 mm/min. Supplemental Fig. 3 and Supplemental Video 4 To define Tfh cell dynamics we used anti–PD-1 Abs. We checked confirm this differential pattern of Tfh cell migration in another that the stained cells were also positive for CD4 and not for GC. Overall, we observed such different motility behavior in more FOXP3, a marker of T follicular regulatory cells, and therefore than 20 different GC from 12 different human biopsies. Because can be considered as bona fide Tfh cells (Supplemental Fig. 2A, Tfh cells in central regions were nearly static, we performed ex- 2B). GC light zones were identified using CD35 Abs. We first periments to rule out potential artifacts created by our experi- assessed the behavior of Tfh cells in relation to B cells labeled mental procedure. Such a motility difference was not due to the with CD20 Abs. As reported in mouse models (5), B cells in GC Ab used to stain Tfh cells, as a similar observation was made light zones do not migrate as rapidly as Tfh cells (Fig. 4A, when an anti-CD3 Ab was employed to label T cells (Fig. 5C, Supplemental Video 2). B cells usually formed a dense network Supplemental Video 5). In these conditions, T cells migrate, once The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/

FIGURE 5. Tfh cells migrate more actively at the periphery of human GC light zones. (A) Tracks of individual endogenous PD-1 T cells in one human by guest on September 28, 2021 adenoid. Tracks are color coded according to PD-1 cell displacement length. See also Supplemental Video 3. Student t test. ***p , 0.001. (B) Mean velocity of PD-1–positive cells analyzed from Supplemental Video 3. Results are representative of similar analysis performed on 20 different GCs from 12 adenoids. (C) Tracks of individual endogenous CD3 T cells in one human adenoid. Tracks are color coded according to CD3 cell displacement length. See also Supplemental Video 5. again, more actively in outer GC light zones as compared with cells with the formation of highly dynamic clusters, as shown in more central areas. A high death rate is usually a feature of GC. the Supplemental Video 7. Fig. 6A and Supplemental Video 8 We thus wondered whether cells undergoing apoptosis, either in a illustrate the typical behavior of one Tfh cell localized in the physiological manner or due to the cutting procedure, might be outer follicle that moves fast before forming a transient conjugate more numerous in central regions and thus might contribute to the with another PD-1–positive cell and then leaving this cell to regain low motility of Tfh cells. To this end, we measured cell apoptosis motility. One can notice that the formation of a conjugate is as- by immunohistochemistry on paraffin-fixed adenoid samples by sociated with a sharp decrease in the cell speed. The T cell ve- using an anti-active caspase-3 Ab. Our results confirm the presence locity before the contact was around 10 mm/min, dropped to 1–2 of cells undergoing apoptosis in GC regions (Supplemental Fig. mm/min during the interaction phase, and resumed to 10 mm/min 4A). However, such cells were distributed homogenously within when the two PD-1–positive cells came off. the whole GC and were not particularly enriched in central GC By contrast, the pattern of Tfh cells in the central area of the GC regions. In addition, cells that undergo apoptosis were not labeled light zone was strikingly different. Although clusters of PD-1 cells with anti–PD-1 Abs. Because the specificity of the signal coming were also observed in these regions, conjugates were much more from the anti-active caspase-3 Ab was not optimal, we decided to stable, as shown in the Fig. 6B and Supplemental Video 9. In this confirm this result by assessing live and dead cells in fresh adenoid example, two Tfh cells formed a conjugate that lasted the whole slices. Two fluorescent dyes, DRAQ5 and DAPI, that enabled us to recording (20 min). We then analyzed in both regions the average distinguish live cells from dead cells were used. Results revealed contact time between one Tfh cell and another. In central regions, that dead cells, which accounted for ∼20% of all cells, were not the average contact time between two Tfh cells was 11.9 min particularly enriched in central GC areas (Supplemental Fig. 4B– (Fig. 7A). At the GC periphery, the conjugates were more tran- D). As illustrated in the Supplemental Video 6, many live cells sient, with an average contact time of 2.7 min. With a fast mi- labeled with DRAQ5 were motile during the 15-min recording. gration and an ability to form brief contact with each other, Tfh cells in the GC periphery usually interact with many different Tfh cells in GC light zones interact with each other PD-1–positive cells during the recording. Fig. 7B represents the Tfh cells were not only faster in the GC light zone periphery but also number of contacts and their duration for 16 different cells lo- exhibited a prominent capacity to interact with other PD-1–positive calized in outer GC light zones. Although most Tfh cells interact 6 IMAGING OF Tfh CELLS IN HUMAN LYMPHOID TISSUES Downloaded from http://www.jimmunol.org/

FIGURE 6. Tfh cells engage in either short or long-lasting homotypic conjugates. (A) Snapshots of a time lapse with one PD-1–positive cell engaging transiently with another PD-1–positive cell. The trajectory of the first PD-1 cell is shown in green. See also Supplemental Video 8. Right, The instantaneous by guest on September 28, 2021 velocity plotted against time for the tracked cell. Note the drop in T cell velocity during the cell–cell interaction. (B) Snapshots of a time lapse with one PD-1–positive cell stably engaging with another PD-1–positive cell. See also Supplemental Video 9. Right, The instantaneous velocity plotted against time for the tracked cell. Results shown are representative of more than 10 independent experiments. with two or three different cells, some cells, such as number 18, we add another layer of information by providing evidence that come into contact with nine other PD-1–positive cells during a Tfh cells also communicate between them by forming either brief 20-min recording. Fig. 7C and Supplemental Video 10 show, with or prolonged conjugates, depending on their locations in GCs. high resolution, Tfh cell clusters that break up and rebuild. One The immunology field has mainly focused on studying T cell can notice extensive surface contacts between Tfh cells when interaction with dendritic cells, B cells, or malignant cells. Few they are in a cluster, suggesting potential cell–cell communica- studies have assessed T cell–T cell interactions. Of note, Krummel’s tions. Such extensive surface contacts from homotypic conju- (14) laboratory has reported in mouse models the occurrence of gates were also clearly visible in 3D projection images (Fig. 7D, functional homotypic interactions between activated CD8 T cells. Supplemental Video 11). Interestingly, such conjugates were shown to play a role in the Overall these results reveal two different anatomical GC light differentiation of activated CD8 T cells (15). The implication of zone regions in which human Tfh behave very differently. They such interactions for Tfh cell functions and underlying molecular also demonstrate the prominent ability of Tfh to interact with mechanisms is not known for the moment. From our movies, we each other. noticed that the size of the T cell–T cell contact areas was important (see Fig. 6C for example). This might suggest that intracellular Discussion signals can possibly be transmitted, modulating Tfh cell fate and To our knowledge, this is the first study of the dynamic behavior of maintenance of an appropriate helper state. Tfh cells in human tissue. For this study, we have set up an ex vivo Our results demonstrate two different landscapes in GC light imaging system for the in situ labeling and real-time monitoring zones in which Tfh behave very differently. Tfh are dynamic at the of Tfh cells in human lymphoid tissues. Using this approach, we periphery and more static in the center. For the moment, the reasons demonstrated two distinct modes of Tfh migration. In outer GCs of these distinct navigational modes are unknown. They probably light zones, Tfh move fast, with a remarkable ability to form brief reflect two different light zone territories with the periphery and multiple homotypic interactions. Conversely, Tfh cells in more having a marked Tfh cell stimulatory migration ability. Of in- central GC areas are static, forming long-lasting conjugates with terest, immunohistological studies from Ian MacLennan and his each other. colleagues (16) have revealed the presence of five different fol- Mouse studies have revealed with great detail the spatiotemporal licular compartments in human tonsils. Apart from the follicular interactions between Tfh cells and B cells in GCs (2). In this study, mantle and the dark zone, these studies provided the evidence that The Journal of Immunology 7 Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 7. Analysis of Tfh cell homotypic interaction in human GC. (A) Mean contact time between two PD-1–positive cells in GCs. Results are from seven GCs with more than 50 cells analyzed. (B) Contact time of 16 Tfh cells with other Tfh cells in one GC. Each dot represents a cell–cell contact. The red mark indicates the mean contact time of every PD-1–positive cell analyzed. (C) Snapshots of a time lapse with PD-1–positive cells (red) in a cluster that progressively breaks apart. See also Supplemental Video 10. (D) Confocal images displaying homotypic conjugates of PD-1–positive cells. Left, A two-dimensional maximum intensity projection image. Right, A 3D reconstruction image. See also Supplemental Video 11. light zones are subdivided into three areas. The basal light zone molecules can be important too. Notably, Tfh cells express high was found adjacent to the dark zone, whereas the apical light zone levels of PD-1, a protein that has been shown to control T cell was close to the follicular mantle. Between the apical light activation and also control migration (20). Remarkably, a recent zone and the follicular mantle was a narrow outer zone where imaging study performed in mice evidenced an essential role of activated CD4 T CD40L cells accumulated (17). It is likely that PD-1 in controlling Tfh cell positioning and function in GCs (21). the outer zone described in these studies corresponds to the pe- Experiments with blocking anti–PD-1 Abs should be informative ripheral regions enriched in fast-migrating Tfh cells that we in deciphering the role of PD-1 in human Tfh displacement ob- characterized in human adenoids. The same group has also shown served in our settings. In tumors, one of us has shown that CD8 that the division of light zones into three regions was a feature of T cell migration was negatively controlled by collagen fibers and chronically inflamed lymphoid tissues such as palatine tonsils macrophages (9, 10). Whether macrophages, which are abun- (18). In contrast, human lymph nodes, the site of an acute immune dantly present in lymphoid organs, trap Tfh cells in the center of response, were composed of an homogenous light zone devoid of GCs remains an open question that can be addressed in our model. an outer region. The exploration of these different zones, their Apart from external determinants, one cannot rule out the relationship and regulation during an immune response, as well as possibility of two different subsets of Tfh showing different mi- their respective function will be an important topic for the future. gration behaviors and distributed differently in human GCs. It has Recent imaging studies have extended our knowledge of external long been known that GC T cells are heterogeneous in phenotype factors that control the motility of T cells and their ability to interact and function, comprising different helper and regulatory subsets with other cells (19). One can mention soluble factors such as (22). In particular, Tfh cells share important plastic features chemokines. We can thus speculate that CXCL13 produced by the with other CD4+ helper T cells; so it possible that the distinct outer zones of FDC is more concentrated at the periphery than at motile behaviors we evidenced reflect two Tfh populations the center of the light zone, explaining our results. Accessory endowed with distinct modes of migration. Of note, a recent study 8 IMAGING OF Tfh CELLS IN HUMAN LYMPHOID TISSUES has highlighted the different migration pattern of Th1 and Th2 References cells in the mouse dermis, indicating important consequences 1. Vinuesa, C. G., M. A. Linterman, D. Yu, and I. C. MacLennan. 2016. Follicular for the function of these cells. In this study, Th1 cells migrate helper T cells. Annu. Rev. Immunol. 34: 335–368. 2. Qi, H. 2016. T follicular helper cells in space-time. Nat. Rev. Immunol. 16: locally in a chemokine-dependent manner, whereas Th2 cells 612–625. show a broader migration mode independent of chemokines 3. Allen, C. D., T. Okada, H. L. Tang, and J. G. Cyster. 2007. Imaging of germinal but dependent on the integrin a b3 (23). Knowing that Tfh center selection events during affinity maturation. Science 315: 528–531. v 4. Liu, D., H. Xu, C. Shih, Z. Wan, X. Ma, W. Ma, D. Luo, and H. Qi. 2015. T-B- cells subsets are linked to their Th cell counterpart, we can cell entanglement and ICOSL-driven feed-forward regulation of germinal centre hypothesize that distinct motility programs are initiated during reaction. Nature 517: 214–218. 5. Xu, H., X. Li, D. Liu, J. Li, X. Zhang, X. Chen, S. Hou, L. Peng, C. Xu, W. Liu, Tfh differentiation. et al. 2013. Follicular T-helper cell recruitment governed by bystander B cells From a technological perspective, the successful ex vivo staining and ICOS-driven motility. Nature 496: 523–527. not only with anti–PD-1 but also with many other directly coupled 6. Victora, G. D., T. A. Schwickert, D. R. Fooksman, A. O. Kamphorst, M. Meyer- Hermann, M. L. Dustin, and M. C. Nussenzweig. 2010. Germinal center dy- Abs indicates that the current system may be used to study the namics revealed by multiphoton microscopy with a photoactivatable fluorescent behavior of other cells in normal but also pathological conditions reporter. Cell 143: 592–605. and relate cell navigation modes with a patient outcome. 7. Schmitt, N., and H. Ueno. 2013. Human T follicular helper cells: development and subsets. Adv. Exp. Med. Biol. 785: 87–94. The tissue slice assay that we established and used in this study 8. Ueno, H., J. Banchereau, and C. G. Vinuesa. 2015. Pathophysiology of T fol- presents a number of advantages but it has some limitations that licular helper cells in humans and mice. Nat. Immunol. 16: 142–152. 9. Peranzoni, E., J. Lemoine, L. Vimeux, V. Feuillet, S. Barrin, C. Kantari-Mimoun, need to be taken into account and that could have amplified or N. Bercovici, M. Gue´rin, J. Biton, H. Ouakrim, et al. 2018. Macrophages impede minimized the real situation. As a matter of fact, we cannot rule CD8 T cells from reaching tumor cells and limit the efficacy of anti-PD-1 out that the long-lasting homotypic Tfh cell conjugates that we treatment. Proc. Natl. Acad. Sci. USA 115: E4041–E4050. Downloaded from 10. Salmon, H., K. Franciszkiewicz, D. Damotte, M. C. Dieu-Nosjean, P. Validire, found in our study are due to experimental conditions. The A. Trautmann, F. Mami-Chouaib, and E. Donnadieu. 2012. Matrix architecture cutting procedure will damage the tissue, which may affect the defines the preferential localization and migration of T cells into the stroma of behavior of Tfh cells, especially near the cut surface. By using human lung tumors. J. Clin. Invest. 122: 899–910. 11. Donnadieu, E., Y. Michel, and M. L. Hansmann. 2019. Live imaging of resident fluorescent dyes specific for live and dead cells, we estimated that T-cell migration in human lymphoid tissue slices using confocal microscopy. our slice preparation contains ∼20% of dead cells. Although Methods Mol. Biol. 1930: 75–82. 12. Asperti-Boursin, F., E. Real, G. Bismuth, A. Trautmann, and E. Donnadieu. these dead cells were not particularly enriched in GC central 2007. CCR7 ligands control basal T cell motility within lymph node slices in a http://www.jimmunol.org/ areas, it is possible that an excess of cell mortality in certain phosphoinositide 3-kinase-independent manner. J. Exp. Med. 204: 1167–1179. follicular regions alters the motility of Tfh cells. In addition, a 13. Miller, M. J., S. H. Wei, I. Parker, and M. D. Cahalan. 2002. Two-photon im- aging of lymphocyte motility and antigen response in intact lymph node. Science number of soluble factors (e.g., chemokines) instrumental in 296: 1869–1873. controlling the migration of Tfh in the center could have been 14. Sabatos, C. A., J. Doh, S. Chakravarti, R. S. Friedman, P. G. Pandurangi, lost during the cutting procedure. One way to bypass these A. J. Tooley, and M. F. Krummel. 2008. A synaptic basis for paracrine interleukin-2 signaling during homotypic T cell interaction. Immunity 29: 238–248. problems is by monitoring cells located at several tens of mi- 15. Ge´rard, A., O. Khan, P. Beemiller, E. Oswald, J. Hu, M. Matloubian, and crons from the surface in presumably healthier regions of the M. F. Krummel. 2013. Secondary T cell-T cell synaptic interactions drive the differentiation of protective CD8+ T cells. Nat. Immunol. 14: 356–363. tissue. We have used a confocal microscope for our study, but it 16. Hardie, D. L., G. D. Johnson, M. Khan, and I. C. M. MacLennan. 1993. is likely that two-photon microscopes will increase the spatial Quantitative analysis of molecules which distinguish functional compartments by guest on September 28, 2021 resolution in depth. within germinal centers. Eur. J. Immunol. 23: 997–1004. 17. Casamayor-Palleja, M., M. Khan, and I. C. M. MacLennan. 1995. A subset of In conclusion, our results suggest that Tfh cells interact not only CD4+ memory T cells contains preformed CD40 ligand that is rapidly but with B cells but also with each other. An important question for the transiently expressed on their surface after activation through the T cell receptor future is what do they say to each other and how do these talks complex. J. Exp. Med. 181: 1293–1301. 18. Brachtel, E. F., M. Washiyama, G. D. Johnson, K. Tenner-Racz, P. Racz, and affect Tfh cell fate and differentiation. We believe that the tissue I. C. MacLennan. 1996. Differences in the germinal centres of palatine tonsils slice assay, combined with dynamic imaging microscopy, will and lymph nodes. Scand. J. Immunol. 43: 239–247. 19. Krummel, M. F., F. Bartumeus, and A. Ge´rard. 2016. T cell migration, search enable us to provide important insights into such novel mode of strategies and mechanisms. Nat. Rev. Immunol. 16: 193–201. cell–cell communication. 20. Brunner-Weinzierl, M. C., and C. E. Rudd. 2018. CTLA-4 and PD-1 control of T-cell motility and migration: implications for tumor immunotherapy. Front. Immunol. 9: 2737. Acknowledgments 21. Shi, J., S. Hou, Q. Fang, X. Liu, X. Liu, and H. Qi. 2018. PD-1 controls follicular We thank Dr. Fransesco Pampaloni and Heinz Schewe (Goethe Universita¨t T helper cell positioning and function. Immunity 49:264–274.e4. Frankfurt am Main) for help in setting up the confocal microscope for our 22. Song, W., and J. Craft. 2019. T follicular helper cell heterogeneity: time, space, and function. Immunol. Rev. 288: 85–96. first imaging experiments. 23. Gaylo-Moynihan, A., H. Prizant, M. Popovic, N. R. J. Fernandes, C. S. Anderson, K. K. Chiou, H. Bell, D. C. Schrock, J. Schumacher, T. Capece, et al. 2019. Programming of distinct chemokine-dependent and -independent search Disclosures strategies for Th1 and Th2 cells optimizes function at inflamed sites. Immunity The authors have no financial conflicts of interest. 51:298–309.e6.