B Cell Speed and B-FDC Contacts in Germinal Centers Determine Plasma Cell Output via Swiprosin-1/EFhd2.

Article B Cell Speed and B-FDC Contacts in Germinal Centers Determine Plasma Cell Output via Swiprosin-1/EFhd2 Graphical Abstract Highlights dEFhd2 limits migration of GC B cells and hapten-specific immunity dFast GC B cells are selected earlier and prone to differentiate dEFhd2 fosters contacts of GC B cells with follicular dendritic cells dContact of fast GC B cells with follicular dendritic cells is decisive Authors Dorothea Reimer, Michael Meyer-Hermann, Asylkhan Rakhymzhan, ..., Anja E. Hauser, Raluca A. Niesner, Dirk Mielenz Correspondence [email protected] In Brief Reimer et al. show that EFhd2 restricts migration of germinal center B cells and hapten-specific immunity. Selection of speedy germinal center B cells occurs earlier and they are more likely to differentiate. EFhd2 supports contact of germinal center B cells with follicular dendritic cells, enabling successful competition of rapidly migrating germinal center B cells. Reimer et al., 2020, Cell Reports 32, 108030 August 11, 2020 ª2020 The Authors. https://doi.org/10.1016/j.celrep.2020.108030 ll Article B Cell Speed and B-FDC Contacts in Germinal Centers Determine Plasma Cell Output via Swiprosin-1/EFhd2 Dorothea Reimer, 1 Michael Meyer-Hermann, 2 Asylkhan Rakhymzhan, 3 Tobit Steinmetz, 1 Philipp Tripal, 4 Jana Thomas, 1 Martin Boettcher, 5 Dimitrios Mougiakakos, 5 Sebastian R. Schulz, 1 Sophia Urbanczyk, 1 Anja E. Hauser, 3,6 Raluca A. Niesner, 3,7,8 and Dirk Mielenz 1,8,9, * 1 Division of Molecular Immunology, Universita ¨tsklinikum Erlangen, Nikolaus-Fiebiger-Zentrum, FAU Erlangen-N€ urnberg, Erlangen, Germany 2 Department of Systems Immunology and Braunschweig, Integrated Centre of Systems Biology, Helmholtz Center for Infection Research, Braunschweig, Germany 3 Deutsches Rheumaforschungszentrum (DRFZ), Berlin, Germany 4 Optical Imaging Center (OICE), Universita ¨tsklinikum Erlangen, FAU Erlangen-N€ urnberg, Erlangen, Germany 5 Department of Internal Medicine V, Universita ¨tsklinikum Erlangen, FAU Erlangen-N€ urnberg, Erlangen, Germany 6 Charite ´– University Medicine, Berlin, Germany 7 Dynamic and Functional In Vivo Imaging, Veterinary Medicine, Freie Universita ¨t, Berlin, Germany 8 These authors contributed equally 9 Lead Contact *Correspondence: dirk.mielen[email protected] https://doi.org/10.1016/j.celrep.2020.108030 SUMMARY Plasma cells secreting affinity-matured antibodies develop in germinal centers (GCs), where B cells migrate persistently and directionally over defined periods of time. How modes of GC B cell migration influence plasma cell development remained unclear. Through genetic deletion of the F-actin bundling protein Swiprosin-1/EF-hand domain family member 2 (EFhd2) and by two-photon microscopy, we show that EFhd2 restrains B cell speed in GCs and hapten-specific plasma cell output. Modeling the GC reaction reveals that increasing GC B cell speed promotes plasma cell generation. Lack of EFhd2 also reduces contacts of GC B cells with follicular dendritic cells in vivo. Computational modeling uncovers that both GC output and antibody affinity depend quantitatively on contacts of GC B cells with follicular dendritic cells when B cells migrate more persistently. Collectively, our data explain how GC B cells integrate speed and persistence of cell migration with B cell receptor affinity. INTRODUCTION The development of memory B cells and long-lived plasma cells (PCs) secreting affinity-matured antibodies (Abs) (Berek et al., 1991) depends on the germinal center (GC) reaction. GCs are transient structures that form in secondary lymphatic organs around the follicular dendritic cell (FDC) network 4 days after T-cell-dependent (TD) antigen challenge (Gatto and Brink, 2010). Mature GCs are highly ordered structures consisting of histologically distinct regions, the dark zone (DZ) and the light zone (LZ), the latter of which is centered on chemokine (C-X-C motif) ligand 13 (Cxcl13)-expressing FDCs (Victora and Nussenzweig, 2012). Typically, GCs formed in response to alum precipitated haptenated model antigens last 21–28 days and then resolve as antigen becomes limiting and Ab feedback ensues (Gatto and Brink, 2010). Long-lasting or even chronic GCs develop after immunization with particulate antigen, such as sheep red blood cells (SRBCs) (Dogan et al., 2009), in response to certain infections as a consequence of hyperimmunization or naturally in the Peyer’s patches. Immunoglobulin (Ig) gene pedigrees derived from GC B cells (Berek et al., 1991) suggested a Darwinian mechanism of Ab affinity maturation. These observed Ig gene mutation rates are reconciled with DZ/LZ architecture of GCs and cell-cycle duration in the cyclic-reentry model of the GC reaction (Kepler and Perelson, 1993). Accordingly, somatic hypermutation (SHM) occurs in the DZ and clonal selection occurs in the LZ, predicting a requirement for interzonal B cell migration. In vivo multiphoton microscopy corroborated this prediction, with C-X-C motif chemokine receptor 4 (Cxcr4)-expressing centroblasts migrating from the DZ to the LZ and Cxcr5-expressing centrocytes migrating from the LZ to the DZ (Hauser et al., 2007). GC B cells migrate in a directed manner (O’Connor et al., 2011), whereby the net flow appears to be from the DZ to the LZ (Beltman et al., 2011), with only few B cells migrating back into the DZ (Victora et al., 2010). Tagging GC B cells with photoactivatable GFP in vivo has revealed that those B cells with superior antigen presentation ability are preferentially selected by T follicular helper (Tfh) cells, resulting in their migration back into the DZ and increased PC generation (Victora et al., 2010). There is evidence that B cell/T Cell Reports 32, 108030, August 11, 2020 ª2020 The Authors. 1 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). ll OPEN ACCESS (legend on next page) 2Cell Reports 32, 108030, August 11, 2020 Article ll OPEN ACCESS cell interactions determine the number of divisions of a B cell in the DZ (Meyer-Hermann et al., 2012). B cell clones carrying selected B cell receptor (BCR) with comparably high affinity are predominantly found in the PC compartment, while memory B cells harbor less affine BCR genes and arise earlier in the GC reaction (Weisel et al., 2016). As PCs carry Ig genes encoding higher affine BCRs and later Abs, a role for BCR signaling in addition to BCR-mediated endocytosis in dictating GC fate toward the PC program is not too speculative. Indeed, initiation of the PC program requires BCR contact with antigen contained in immune complexes bound on FDCs, followed by completion of the PC program by signals derived from Tfh cells (Kra ¨utler et al., 2017) and specialized stromal cells localized at the T cell/B cell border (Zhang et al., 2018b). In line, B cells directly test BCR affinity on antigen presented by FDCs through force application via myosin II motors in a specialized GC synapse (Natkanski et al., 2013;Nowosad et al., 2016). Only BCR signaling, together with CD40 stimulation, enables centrocytes to upregulate Myc (Luo et al., 2018), highlighting a role for BCR signaling and T cell help in the selection of GC cells. Whereas transcription factors such as Irf4 (Klein et al., 2006), Myc (Calado et al., 2012), or Rel (Heise et al., 2014) that direct the cell fate in GCs have been well characterized, proteins linking the BCR or CD40 to GC output are less established. Those proteins might be found among those controlling B cell migration or B cell/FDC interaction in GCs (Tolar, 2017), because a main task of B cells having undergone and survived SHM in the DZ (Mayer et al., 2017)is to find antigen bound on FDCs as well as a cognate Tfh cell. In line, many reports emphasize the importance of cell migration patterns for GC dynamics, particularly for their PC output (reviewed in Victora and Nussenzweig, 2012). However, it remains elusive how different cell migration patterns directly impact the GC reaction and, consequently, PC maturation and which molecules within B cells drive their motility behavior. Swiprosin-1/EFhand domain family member 2 (EFhd2; also known as Swiprosin1, and not to be confused with Swiprosin-2/EFhd1; D€ utting et al., 2011)isa30-kDa Ca 2+ - and F-actin-binding protein (Hagen et al., 2012;Huh et al., 2013;Kwon et al., 2013) with a C-terminal coiled-coil domain and functional SH3-binding sites at the N terminus (Kroczek et al., 2010). Null mutation of EFhd2 (EFhd2KO) revealed that EFhd2 negatively controls GC expansion in response to secondary Nippostrongylus brasiliensis infection (Brachs et al., 2014). EFhd2 also controls development of IgG1 and IgE PCs as well as IgM and IgE serum Abs in response to infection with N. brasiliensis (Brachs et al., 2014) in a B-cellintrinsic manner. How EFhd2 controls GC-derived PC generation and whether EFhd2 controls Ab quality was unknown. EFhd2 supports lamellipodia formation, lung cancer, and melanoma metastasis as well as macrophage migration through regulation of actin dynamics by Rho GTPases (Fan et al., 2017;Huh et al., 2015;Tu et al., 2018;Zhang et al., 2018a). Based thereupon, we hypothesized that EFhd2 controls the GC reaction and PC development via control of actin dynamics and cell migration. Hence, we analyzed B cell dynamics of EFhd2 wild-type (WT) and knockout (KO) B cells in vitro and in vivo using two-photon microscopy, together with in silico modeling and a detailed kinetic analysis of the GC and PC response against nitrophenol (NP)- keyhole limpet hemocyanin (KLH). We found that EFhd2 limits B cell migration and that enhanced B cell migration through lack of EFhd2 leads to increased PC output. However, a mathematically predicted increase in BCR affinity was counterbalanced by reduced interactions of EFhd2KO B cells with FDCs. Using this iterative approach, we defined the general concept (independently of EFhd2KO) that cell migration alone cannot elicit stronger GC reactions or Ab affinity, and B cell/FDC interactions are required to support the accelerating effect of B cell migration. RESULTS EFhd2KO B Cells Migrate Faster in Germinal Centers To test whether EFhd2 affects B cell migration and PC generation, we isolated and labeled EFhd2WT and EFhd2KO B cells (Brachs et al., 2014) from SRBC-immunized mice and transferred them back into SRBC-immunized WT recipients (Figure 1A). Migration of fluorescently labeled B cells in GCs and through the FDC network was assessed in living spleen by multiphoton microscopy (Heinig et al., 2014;Rakhymzhan et al., 2017) (Video S1;Figure 1B). Hereby, we assessed several parameters: (1) velocity, (2) speed, (3) distance, (4) displacement, and (5) directedness (Figure 1C–1G; see Figure S1 for definitions and explanations). EFhd2KO B cells migrated approximately two times faster in GCs in competition with WT B cells, with both mean velocity and speed being elevated (Figure 1C, D). Displacement of EFhd2KO B cells and covered distance were proportionally enhanced (Figure 1E, F), revealing unaltered directedness (displacement/distance; Masuzzo et al., 2016)(Figure 1G), although EFhd2 has been proposed to determine the Figure 1. EFhd2 Deficiency Enhances B Cell Motility in GCs (A) WT and EFhd2KO donor as well as WT recipient mice were immunized intraperitoneally (i.p.) with SRBCs. At day 7, splenic B cells from donor mice were isolated and fluorescently labeled. WT B cells, Hoechst, EFhd2KO B cells, CMTPX, and vice versa, mixed in a 1:1 ratio, were transferred intravenously (i.v.) into WT recipient mice. One day after transfer, the FDC network was labeled via i.v. application of anti-CD21/32-Atto590 Fab 2 -Ab, and migration of WT and EFhd2KO B cells within splenic GC was tracked. (B) Representative image of two adjacent GCs, identified via the FDC networks (turquoise, anti-CD21/32-Atto590), containing WT B cells (blue, Hoechst) and EFhd2KO B cells (red, CMTPX). Scale bar, 100 mm. (C–G) Plots on the left side display WT and EFhd2KO B cells tracked within one representative recipient mouse (one dot = one cell). Plots on the right side show mean values for WT and EFhd2KO B cells in five independent recipient mice (one dot = one mouse). (C and D) Analysis of cell movement by cell velocity (C) and speed (D). (E and F) Analysis of the actually covered paths quantified by displacement (E) and distance (F). (G) The directedness of cell migration was calculated as the ratio of distance to displacement. Data are represented as mean values; statistics: N = 2, n = 5; significance was analyzed using Mann-Whitney test for single cell comparison (****p < 0.0001, left panels in C–G) and paired t test for comparing mean values of recipient mice (nonsignificant [n.s.], p > 0.05; *p %0.05; ***p %0.001, right panels in C–G). Cell Reports 32, 108030, August 11, 2020 3 Article ll OPEN ACCESS direction of cell protrusions (Huh et al., 2015). Importantly, the speed of WT B cells was exactly as reported previously (Allen et al., 2007;Hauser et al., 2007;Schwickert et al., 2007). Migration, activation, and attachment of B cells is dictated by regulated expression of chemokine receptors Cxcr4 and Cxcr5, integrins, and their ligands, such as the lymphocytefunction-associated antigen 1 (LFA-1)/intercellular adhesion molecule 1 (ICAM-1) combination, as well as by B cell activation (Carrasco et al., 2004;Liu et al., 2016; reviewed in Victora and Nussenzweig, 2012). The faster migration of EFhd2KO B cells in vivo was not due to altered expression of Cxcr4, Cxcr5, LFA-1, very late antigen 4 (VLA-4), ICAM-1, or ICAM-2 (Figure S2). We next wished to determine which activating stimulus triggers EFhd2KO B cells. Since GC B cells have received a plethora of signals that cannot be dissected, we measured chemokinesis of SiR-actin-labeled B cells on recombinant ICAM1Fc fusion protein (Liu et al., 2016)(Videos S2 and S3;Figure 2A). Anti-BCR-activated, but not anti-CD40/interleukin-4 (IL-4)-activated, EFhd2KO B cells migrated faster on ICAM-1 in vitro in response to Cxcl13 (triggering Cxcr5) (Figure 2B), but not to Cxcl12 (triggering Cxcr4) (Figure 2C). Intriguingly, BCR activation in combination with anti-CD40/IL-4 stimulation arrested B cell migration (Figure 2B) in both WT and EFhd2KO B cells. Taken together, EFhd2 limits B cell migration in GCs and in vitro.In particular, BCR activation appears to modulate EFhd2-restricted B cell migration. Computational Modeling of GC Dynamics Based on Experimentally Determined Parameters of B Cell Migration To test whether more rapid cell migration modulates GC responses we supplied an established mathematical model of the GC reaction (Binder and Meyer-Hermann, 2016;Figge et al., 2008;Meyer-Hermann, 2014;Meyer-Hermann et al., 2009,2012) with the B cell migration data (shown in Figure 1). As EFhd2KO B cells migrated approximately two times faster (compare Figure 1), we simulated GCs with normal B cell migration rates (7.5 mmin/min/1.5 min persistence time, black lines) and two times faster B cells (15 mmin/min/1.5 min persistence time, pink lines). Faster migration may also be the result of increased directional persistence (Maiuri et al., 2015). We therefore also modeled B cells with two times and four times more persistence (7.5 mmin/min/3.0 min persistence and 7.5 mmin/ min/6.0 min persistence, green and blue lines) and with twotimes-faster-migrating B cells with a 2-fold increased persistence (7.5 mmin/min/3.0 min persistence, red line). Elevating cell migration or persistence in silico was sufficient to elicit a dose-dependent increase in the total number of GC and output cells (PCs and memory B cells) (Figures 3A and 3B). Importantly, the effect of speed was more dominant than the effect of persistence (Figure 3B). Predicted affinity maturation of GC B cells was unaltered by speed or persistence, whereas doubled speed, but not doubled persistence, predicted a small (15%) increase in affinity of GC output cells (Figure 3C). By plotting selection probability versus affinity, it became evident that faster B cells are likely to be selected early (days 3–5), but not later (days 5–9) (Figure 3D). Taken together, increasing cell migration and persistence in simulated GCs led to the following expectations that were evaluated in our model system, EFhd2KO mice (experimentators were blinded to the results of the simulation): (1) increased GC B cell numbers, (2) enhanced PC output, and (3) a moderate increment of the mean affinity of GC-derived BCRs. EFhd2 Limits PC Differentiation Doubling B cell speed in GCs predicted increased GC B cell numbers (Figure 3A). However, we did not register higher numbers of GC B cells or hapten-specific GC B cells at any Figure 2. BCR Activation Enhances Migration of EFhd2KO B Cells on ICAM-1 and CXCL13 (A) Naive splenic B cells from WT and EFhd2KO mice were isolated and activated with anti-IgM Ab, anti-CD40 Ab + IL-4, or anti-IgM Ab + anti-CD40 Ab + IL-4 for 48 h. (B and C) B cell migration on ICAM-1 in the presence of CXCL13 (B) or CXCL12 (C) was analyzed by spinning disc confocal microscopy. Datasets are shown for individual cells (one dot = one B cell) activated as described above. Statistics: N = 2, n = 2–3; significance was analyzed using Mann-Whitney test (n.s. > 0.05; **p %0.01). 4Cell Reports 32, 108030, August 11, 2020 Article ll OPEN ACCESS time point between days 7 and 21 upon NP-KLH immunization of EFhd2KO mice (Figures 4A–4C). To test nevertheless whether EFhd2 influences PC generation from GCs, we crossed EFhd2KO mice with Blimp1:GFP reporter mice (Kallies et al., 2004), immunized mice with NP-KLH, and analyzed Blimp1 expression (GFP fluorescence) in CD38 low CD95 + B cells at day 21. We found a clearly Blimp1-positive cell fraction in CD19 + CD38 low CD95 GL7  cells (Figure 4D). Intriguingly, we detected a higher frequency of Blimp1-positive CD38 low CD95 + GL7  cells in EFhd2KO mice (Figure 4E), while total cell numbers were hardly elevated (Figure 4F). We explain this by reduced frequencies and numbers of CD19 + CD38 low B cells in the Blimp1:GFP;EFhd2KO mice (Figures 4G and 4H). In summary, Blimp1-positive late GC B cells are increased in relation to the Figure 3. Enhanced B Cell Speed and Persistence Increase GC-Derived PC Output and Affinity (A and B) In silico simulation of GC modulated by speed and persistence B cell migration. Predicted results are shown for the total number of GC B cells, their distribution in dark zone (DZ) and light zone (LZ) (A), and the generation of output cells (B). (C) Affinity maturation was simulated for GC B cells as well as for the output cells. (D) The selection probability in dependence of B cell affinity was analyzed for different time points during the GC course. Data are shown as mean (solid lines) of 50 simulations ±SD (gray shades). Cell Reports 32, 108030, August 11, 2020 5 Article ll OPEN ACCESS Figure 4. Normal Hapten-Carrier-Induced GC B Cells but Increased GC-Derived PCs in EFhd2-Deficient Mice (A) WT and EFhd2KO mice were immunized i.p. with 100 mgNP 29 -KLH in alum and analyzed after 7, 14, and 21 days. Antigen-specific and total GC B cell and PC populations from spleen were determined via fluorescence-activated cell sorting (FACS). Merged FACS dot plots of GC B cells are displayed for day 21, pregated on singlets and viable lymphocytes. (B and C) Frequencies of total GC B cells defined as CD19 + CD38 lo CD95 + GL7 + at days 7, 14, and 21 post-immunization. Respective antigen-specific GC B cells, identified by binding to NP 28 -PE, were quantified as frequency of NP + cells of total CD19 + CD38 lo CD95 + GL7 + B cells. Data are represented as mean values. Statistics: N = 2, n = 4–8; values passed the test for Gaussian distribution (Shapiro-Wilk normality test), and significance was therefore analyzed using unpaired t test (n.s., p > 0.05). (D) Blimp1:GFP mice and Blimp1:GFP 3EFhd2KO mice were immunized i.p. with 100 mgNP 29 -KLH in alum. Spleens were analyzed for Blimp1:GFP expression of CD19 + CD38 lo CD95 + GL7 lo B cells by FACS. Merged FACS dot plots of GC-derived PCs are displayed for day 21, pre-gated on singlets and viable lymphocytes. (legend continued on next page) 6Cell Reports 32, 108030, August 11, 2020 Article ll OPEN ACCESS frequencies and numbers of CD19 + CD38 low B cells. These data point to incremented PC generation within EFhd2KO GCs as a consequence of cell migration, which is in agreement with the predictions (Figure 3). If actin-dependent cytoskeletal reorganization can in fact influence PC generation from GCs, then the expression of actin modulating proteins, including EFhd2, should be altered in (1) GCs and (2) the fraction of GC B cells that is predestined to become PCs (fraction 1 in Ise et al., 2018; Cxcr4 low , CD86 + , Bcl6 low , Irf4 + , and CD69 high ). We therefore analyzed a pre-existing RNA sequencing (RNA-seq) dataset from fraction 1 B cells as well as other GC subsets (Ise et al., 2018; fraction 2: Cxcr4 low , CD86 + , Bcl6 high , Irf4  , and CD69 int ; fraction 3: Cxcr4 low , CD86 + , Bcl6 high , and Irf4  , CD69 low ; fraction 4: Cxcr4 low , CD86 + , Bcl6 low , Irf4  , and CD69 low ; and plasmablasts). These re-analyses showed that genes controlling the B cell actin cytoskeleton, including EFhd2, are sharply regulated at the transition of GC B cells and PCs and strongly enriched in fraction 1 (Figure S3), supporting the idea of an actin-dependent checkpoint in GCs controlling PC development. EFhd2 Limits Antigen-Specific PC Output from the GC We next tracked hapten-specific GC B cells, PCs, and serum Abs over 21 days after NP-KLH immunization (Figure 5). Flow cytometry revealed increased NP-specific PC frequencies in the spleen and bone marrow of EFhd2KO mice at day 21 (Figures 5C and 5F). In addition, we observed increased NP-specific serum IgG (of all isotypes; not depicted here), but not IgM Abs (Figures 5G and 5H), in immunized EFhd2KO mice. These results are in agreement with GC simulation and the increase in Blimp1-expressing B cells at day 21 in immunized Blimp1:GFP;EFhd2KO mice (Figure 4A). The increased NP-specific Abs of EFhd2KO mice were dominantly of lower affinity (Figure 5H). The ratio of high- (anti NP(4)-BSA) versus low-affine (anti NP(29)-BSA) Abs that was predicted to be slightly increased was surprisingly not altered. If affinity maturation was severely altered, then NP binding to GC B cells as determined by flow cytometry would be altered as well, which was not the case. Considering that measurement of affinity maturation by ELISA is widely used, these experiments strongly suggest that affinity maturation is not altered in EFhd2KO B cells. Nevertheless, the observed increase in NP-specific Abs appeared to be specific, as we did not observe increased formation of auto-Abs, at least not anti-double stranded DNA Abs (Figure S4). Furthermore, the T-cell-independent type 2 antigen NP-Ficoll did not elicit an increased response in EFhd2KO mice (Figure S4). Thus, several lines of evidence reveal that EFhd2 limits PC generation from GCs, with cell migration limited by EFhd2 likely playing a role therein (Figure 3B). The increase in PCs was not due to increased B cell/T cell interactions as determined by conventional flow as well as Amnis imaging cytometry (Figures S5 and S6)or increased Tfh cell numbers (Figure S7). An in silico prediction that was not confirmed in EFhd2KO mice (Figure 4A) was the increase in GC B cells induced by increasing B cell speed or persistence in silico (Figure 3A) and an increase in Ab affinity. This raised the possibility that cell migration was not the only parameter altered by EFhd2 deficiency in GC B cells. Since the actin cytoskeleton is also crucial for the B cell/FDC synapse (Tolar, 2017), we next assessed how EFhd2KO B cells interact with FDCs. EFhd2KO B Cells Have Less Contact with FDCs To test whether EFhd2KO B cells interact differently with FDCs in vivo, we tracked EFhd2KO and WT B cells within the FDC network (Figure 6). The colocalization volume of EFhd2KO B cells with FDCs was reduced (Figure 6B), but the duration of interactions was similar (Figure 6C). However, the B cell/FDC contacts appeared to be more confined on the FDC surface, since the displacement of EFhd2KO B cell contacts was reduced (Figure 6D). We conclude that EFhd2KO GC B cells interact differently with FDCs than WT GC B cells. These data could be explained by altered actin cytoskeleton dynamics. The Actin Cytoskeleton of EFhd2KO B Cells Is Highly Dynamic Activated B cells move on ICAM-1 in the presence of Cxcl13 in vitro by exhibiting frequent, spontaneous dilation and shrinking events at the sides of the leading membrane edge, which is predictive of turning versus directional persistence (Liu et al., 2016). These conditions are at least partially found on and within the FDC network. The faster migration of BCR-activated EFhd2KO B cells in response to Cxcl13 (Figure 2B) coincided with increased actin dynamics (Figure 6E). We observed specifically that the Uropod (trailing edge; Figure 6F) is stable in WT B cells but highly dynamic in EFhd2KO B cells (see static images in Figure 6E and Videos S2 [WT] and S3 [EFhd2KO]), showing many protrusion-contractile events in EFhd2KO B cells (Figure 6G). We also observed an increased pool of F-actin at the trailing edge of EFhd2KO B cells (Figure 6H). A Decrease of B-FDC Contact Probability Opposes the GC Increase Induced by High Speed and Persistence If there was a quantitative or qualitative difference in the B cell/ FDC synapses of EFhd2KO B cells, then positive selection could be impaired, reducing the frequency of recycling GC B cells. In this case, the increased GC output predicted to occur when GC B cells migrate faster could be counterbalanced. To reconcile the computer simulation with our experimental data, we changed the likelihood of B cell/FDC interaction as a surrogate (E) Frequencies of GC-derived PCs defined as Blimp1 + CD19 + CD38 low CD95 + GL7 low at day 21 post-immunization. Data are represented as mean values. Statistics: N = 2, n = 5; significance was analyzed using Mann-Whitney test (**p %0.01). (F) Absolute numbers of GC-derived PCs defined as Blimp1 + CD19 + CD38 low CD95 + GL7 low at day 21 post-immunization. Data are represented as mean values. Statistics: N = 2, n = 5; significance was analyzed using Mann-Whitney test. (G) Frequencies of GC B cells defined as CD19 + CD38 low at day 21 post-immunization. Data are represented as mean values. Statistics: N = 2, n = 5; significance was analyzed using Mann-Whitney test. (H) Absolute numbers of GC B cells defined as CD19 + CD38 low at day 21 post-immunization. Data are represented as mean values. Statistics: N = 2, n = 5; significance was analyzed using Mann-Whitney test. Cell Reports 32, 108030, August 11, 2020 7 Article ll OPEN ACCESS measure for the reduced B cell/FDC contacts we observed (Figure 6B). Decreasing B cell/FDC interaction probability by half indeed completely normalized the elevated GC response elicited by doubled B cell speed (Figure 7A) but did not affect quantitative GC output (Figure 7B), thereby approaching closely the experimental data obtained with EFhd2KO mice immunized with NP-KLH. In silico, the reduced antigen uptake led to less intense Tfh cell signals, which induced less division in the DZ and are reflected in a reduced DZ/LZ ratio (Gitlin et al., 2014; Meyer-Hermann, 2014)(Figure 7A). Considering a reduced B cell/FDC interaction probability in the case of simulated doubled B cell persistence instead of doubled B cell speed reduced the increased GC response below the standard setting (Figure 7A), reduced the DZ/LZ ratio even more, and also reduced the increase of GC output (Figure 7B). The higher affinity of GC output cells induced by doubling speed was hardly affected by in silico Figure 5. EFhd2 Deficiency Increases Antigen-Specific PC Output without Increasing Ab Affinity WT and EFhd2KO mice were immunized i.p. with 100 mgNP 29 -KLH in alum, and spleen and bone marrow were analyzed after 7, 14, and 21 days for antigen-specific and total plasmablast/PC populations via FACS. NP-binding immunoglobulins were detected by serum-ELISA. Merged FACS dot plots are displayed for day 21 for the spleen (A) and bone marrow (BM) (D), pre-gated on singlets and viable lymphocytes. Frequencies of total plasmablasts/PCs defined as CD138 + TACI + in spleen (B) and bone marrow (E). Antigen-specific plasmablasts/PCs, identified by binding to NP 28 - PE, were quantified as the frequency of NP + cells of total CD138 + TACI + cells for spleen (C) and bone marrow (F). Sera from day 21 were analyzed for highand low-affine NP-specific IgM (G) and IgG (H). Data are represented as mean values. Statistics: N = 2, n = 4–8; Shapiro-Wilk and unpaired t test (n.s. > 0.05; *p %0.05; **p %0.01). reduction of B cell/FDC interaction but was decreased in the case of doubled persistence (Figure 7C). We conclude that speed and persistence of B cell migration in GCs, along with the propensity to access antigen on FDCs or find supporting stromal cells, are decisive for quantitative and qualitative GC output. How can these theoretical data in turn be validated? In EFhd2KO mice, we did observe increased antigen-specific PC output without an increase in affinity (Figure 5). These data point to a functional consequence of the observed reduced B cell/FDC contacts, which could be reduced positive selection, explaining the phenomenon of increased PC output without increased Ab affinity. Our experiments were performed under noncompetitive conditions, where even B cells without increased BCR affinity may benefit from enhanced migration. To create a competitive situation, we adoptively transferred Ly5.2 EFhd2KO or Ly5.2 WT B cells into Ly5.1 hosts and then immunized the mice and tracked GC B cells. We observed that EFhd2KO B cells are outcompeted at day 21, but not yet at day 14 (Figure S8). A similar finding was also observed in Blimp1:GFP;EFhd2KO mice (Figure 4G). We propose that B cell/FDC contacts are important for GC maintenance when B cells show increased motility under competitive conditions. DISCUSSION B cells in GCs that have undergone and survived SHM (Mayer et al., 2017) constantly move to find selective antigen bound on FDCs, cognate Tfh cells, and stromal cell support (Ise et al., 2018;Kra ¨utler et al., 2017;Zhang et al., 2018b), equipping 8Cell Reports 32, 108030, August 11, 2020 Article ll OPEN ACCESS STAR+METHODS KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Anti CD16/32, clone 93 Biolegend # 101302, RRID:AB_312801 Anti TACI/CD267 (APC, clone eBio8F10-3) eBioscience Cat# 17-5942-82, RRID:AB_842758 Anti Ly5.2/CD45.2 (PerCPCy5.5, clone 104) eBioscience Cat# 35-0454-82, RRID:AB_469725 Anti CXCR4/CD184 (PE, clone 2b11) eBioscience Cat# 12-9991-82, RRID:AB_891391 Anti CD19 (BV421/ APCFire750, clone 6D5) Biolegend Cat# 115558, RRID:AB_2572120 Anti CD38 (PerCPCy5.5, clone 90) Biolegend Cat# 102722, RRID:AB_2563333 Anti CD138 (PECy7, clone 281-2) Biolegend Cat# 142513, RRID:AB_2562197 Anti Ly5.1/CD45.1 (PB or PECy7, clone A20) Biolegend Cat# 110730, RRID:AB_1134168 Anti CXCR5/CD185 (Biotin or PEDazzle, clone L138D7) Biolegend Cat# 145509, RRID:AB_2562125 Anti VLA-4/CD49d-CD29 (PerCPCy5.5, clone R1-2) Biolegend Cat# 103619, RRID:AB_2563701 Anti ICAM-2/CD102 (AF488, clone 3C4) Biolegend Cat# 105609, RRID:AB_2264501 Anti LFA-1/CD11a (AF488, clone M17/4) Biolegend Cat# 101111, RRID:AB_493432 Anti GL-7 (AF647, clone GL-7) BD Biosciences Cat# 561529 Anti CD95 (PECy7, clone Jo2), BD Biosciences Cat# 554254, RRID:AB_395326 Anti ICAM-1/CD54 (APC, clone 3E2) BD Biosciences Cat# 561605 Anti CD4 (FITC, clone gk1.5) BD Biosciences Cat# 561828 Anti CD21/35 (clone 8C12) BD Biosciences Cat# 558768, RRID:AB_397114 Goat anti mouse IgM-HRP Southern Biotech Cat# 1021-05 Goat anti mouse IgG-HRP Southern Biotech Cat# 1030-05 Goat anti-IgM AffiniPure F(ab’) 2 mchain specific Jackson ImmunoResearch Cat# 115-006-020, RRID:AB_2338469 Anti CD40, clone FGK45, Rolink et al., 1996 RRID:AB_2490239 Chemicals, Peptides, and Recombinant Proteins 4-Hydroxy-3-nitrophenylacetyl (Phycoerythrin) Biosearch Technologies. Cat# N-5070 4-Hydroxy-3-nitrophenylacetyl (keyhole limpet hemocyanin) Biosearch Technologies. Cat# N-5060 4-Hydroxy-3-nitrophenylacetyl (bovine serum albumin) Biosearch Technologies. Cat# N-5050 Hoechst 33342 Sigma Cat# B2261 CellTracker Red CMTPX Thermo Fisher Cat# C34552 Imject Alum Thermo Fisher Cat# 77161 Poly-L-Lysine Sigma Cat# P8920 Calf thymus DNA Sigma Cat# D1501 Interleukin-4, premium grade Miltenyi Biotech Cat# 130-097-761 Cxcl12, carrier free Biolegend Cat# 581204 Cxcl13, carrier free Biolegend Cat# 583904 Fatty acid free BSA Sigma Cat# A8806 Protein A Sigma Cat# P6031 Protein A Sepharose Thermo Scientific Cat# 101041 SiR-actin Spirochrome Cat# SC001 Streptavidin (Cy5) Jackson ImmunoResearch Cat# 016-170-084 Critical Commercial Assays EasySep B cell isolation kit Stem Cell Technologies Cat#19854 Experimental Models: Cell Lines 293 cells ATCC Cat# CRL-1573, RRID:CVCL_0045 (Continued on next page) Cell Reports 32, 108030, August 11, 2020 e1 Article ll OPEN ACCESS RESOURCE AVAILABILITY Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Dirk Mielenz ([email protected]) Materials Availability All unique reagents generated in this study are available from the Lead Contact. The plasmid pFuse-rIgG–Fc2-ICAM1 generated in this study has been deposited to Addgene. The accession number for pFuse-rIgG–Fc2-ICAM1 Addgene: 156462. Data and Code Availability This study did not generate new resource datasets or code. EXPERIMENTAL MODEL AND SUBJECT DETAILS All experimental procedures were done in agreement with animal protocols approved by the Government of Lower Franconia, Bavaria, Germany. Both female and male mice were used in the experiments. Mice were maintained on a 12-h light/dark cycle with free access to food and water according to governmental rules. The Blimp1:GFP (B6/Prdm1 tm1Nutt ) and EFhd2KO (Efhd2 tm1(KOMP)Vlcg ) mice used in this study were described previously (Brachs et al., 2014;Kallies et al., 2004). All mice were in C57BL/6 backgrounds and between 8-12 weeks old. Sex matched littermates or age and sex matched animals were used as controls. METHOD DETAILS Two-photon laser-scanning microscopy All imaging experiments of ex vivo spleen slices were performed using a specialized multi-photon laser-scanning microscope based on a commercial scan head (TriMScope II, LaVision BioTec, Bielefeld, Germany) previously described (Rakhymzhan et al., 2017). The detection of the fluorescence signals was accomplished with photomultiplier tubes in the ranges 460 ±30 nm (Hoechst), 525 ±25 nm (autofluorescence), 593 ±20 nm (CMTPX) and 655 ±20 nm (CD21/35-Fab-Alexa 647). The excitation of Hoechst, CMTPX and Alexa Continued REAGENT or RESOURCE SOURCE IDENTIFIER Experimental Models: Organisms/Strains Ly5.1 (B6.SJL-Ptprc a Pepc b /BoyJ) mice Jackson Laboratories Cat# JAX:002014, RRID:IMSR_JAX:002014 EFhd2KO (Efhd2 tm1(KOMP)Vlcg ) mice Brachs et al., 2014 MGI # 5707812, RRID:MGI:5707812 Blimp1:GFP (B6/Prdm1 tm1Nutt ) mice Kallies et al., 2004 MGI# 3510704 Oligonucleotides ICAM-1Fcfwd tgtcacgaattcgcaggtatccatccatcccagagaag Invitrogen; this study N/A ICAM-1Fcrev atgcagatctgttattttgagagtggtacagt Invitrogen; this study N/A Recombinant DNA pFUSE-rIgG-Fc2 Invivogen Cat# pfuse-rfc1 pFUSE-rIgG-Fc2-ICAM1 Addgene; this study Addgene ID 156462 Software and Algorithms GraphPad Prism Graphpad https://www.graphpad.com/scientificsoftware/prism/ Imaris Oxford Instruments https://imaris.oxinst.com/ Volocity Quorum technologies https://quorumtechnologies.com/volocity R Pheatmaps Kolde, 2015 https://cran.r-project.org/web/packages/ pheatmap/index.html Shortest path algorithm Rakhymzhan et al., 2017 N/A Kaluza Beckman Coulter https://www.beckman.de/flow-cytometry/ software/kaluza Other mslide 8-well polymer chambered coverslips Ibidi Bioscience Cat# 80826 Sheep red blood cells Fiebig Na ¨hrstofftechnik Cat# 31100100 e2 Cell Reports 32, 108030, August 11, 2020 Article ll OPEN ACCESS 647 was performed at 850 nm of Ti:Sa laser and 1100 nm of OPO. In all experiments, we focused the laser beams with an 20x objective lens (Apochromat water-immersion, NA = 1.0, WD = 2 mm, Zeiss, Jena, Germany) and maximally used an average laser power of 8 mW to avoid photodamage. The pulse width of the Ti:Sa beam amounted to 160 fs, that of the OPO beam to 150 fs, under the objective lens. Analysis of two-photon imaging data All time-lapse 3D fluorescence data acquired in spleen slices by two-photon microscopy were loaded either in Imaris (BitPlane) or Volocity (Quorum technologies), post-processed with a Gaussian-filter to reduce signal noise and contrasted for best image quality. The objects in each spectral channel of the 4D data were segmented using a standardized automatic pipeline based on the watershed algorithm. The pipeline relies on signal discrimination using the measure of local signal-to-background ratio followed by object merging or separation according to the expected average cell volume. The segmentation results were subsequently inspected by the researchers to exclude non-sense segmentation. After this step, we determined cell numbers and volume. For object tracking over time we used the shortest-path-algorithm, which was previously demonstrated to reliably perform for the used fluorescent cell concentration per imaged volume unit (Rakhymzhan et al., 2017). Also the tracking results were subsequently inspected to exclude artifacts. The results of this step were for each cell: 1) time period the cell could be monitored (time span), 2) position (x,y,z) to each time point. Based on this information the 3) track length (distance), 4) displacement length (displacement), 5) immediate velocity to each time point, 6) average velocity for the whole time span (velocity), 7) displacement rate averaged for the whole time span (displacement rate) and 8) directedness as the ratio between displacement length and track length were determined. Cell-cell contacts were quantified by measuring the colocalization of the corresponding spectral channels, chosen to have a minimum cross-talk. Colocalization is defined as overlap of two spectral signals measured in each pixel. Discrimination between signal and background was performed relying on local signal-to-background ratio. The same segmentation pipeline and tracking algorithm as described for cells were used to identify and follow cell-cell contact objects over time. Analysis of gene expression data Heatmaps of normalized expression values derived from publicly available gene expression data based on RNA Seq (Ise et al., 2018) were generated using the R heatmaps package (Kolde, 2015). Mice Constitutively EFhd2 deficient mice (EFhd2KO) on a C57BL/6 background were described previously (Brachs et al., 2014). Blimp1:GFP reporter mice were kindly provided by Stephen Nutt (Kallies et al., 2004) and crossed to EFhd2KO mice. Ly5.1 (B6.SJL-Ptprc a Pepc b /BoyJ) mice were bought from The Jackson Laboratory. Mice were used at 8–12 weeks of age, wild-type (WT) littermates or age and sex matched animals from the same facility served as controls. Mice were kept under pathogen free conditions in the IVC of the Franz-Penzoldt-Center (Erlangen, Germany), experiments were performed according to ethical guidelines for animal experimentation under control of the government of Lower Franconia, Bavaria, Germany. T-dependent immunizations Mice were injected with 100mgNP 29 -KLH (Biosearch Technologies) in Alum (ThermoScientific) in a 1:2 ratio (200ml total volume) or with 2x10 9 sheep red blood cells (SRBCs, Fiebig Na ¨hrstofftechnik) in 200ml PBS intraperitoneally (i.p.). Isolation of primary murine cells from spleen and bone marrow Spleen was transferred in cold R10 medium (RPMI 1640, 10% FCS, 2mM glutamate, 1mM sodium pyruvate, 50 U/ml penicillin G, 50 mg/ml streptomycin, 50 mMb-mercaptoethanol) and gently passed through a 70mm cell strainer (BD) using the plunger of a 5ml syringe (BD). Femur and tibia were flushed with cold R10 medium using a 27G cannula (BD). Cell suspensions were pelleted by centrifugation at 300xg for 5min at 4C. Erythrocytes were lysed upon resuspension in red blood cell-lysis buffer (150mM NH 4 Cl, 10mM KHCO 3 , 100mM EDTA) for 5min at room temperature. The reaction was stopped by adding cold R10 medium before centrifugation at 300xg for 5min at 4C. The final cell suspensions were kept in cold R10 medium after filtration through 30mm mesh filter (Sysmex). Detection of surface antigens by flow cytometry 2x10 6 - 4x10 6 cells were pelleted in FACS tubes (Micronic) at 300xg for 5min at 4C and resuspended in 50ml of unlabeled anti-CD16/ 32 Ab (10 mg/ml in FACS-buffer (PBS, 2%FCS, 0.05% sodium azide)) for 15min on ice. Cells were washed once with FACS-buffer by centrifugation at 300xg for 5min at 4C, resuspended in 50ml FACS-buffer containing the respective fluorochrome-coupled Abs and incubated for 20min on ice in the dark. Cells were washed twice with FACS-buffer by centrifugation at 300xg for 5min at 4C. Data were acquired using a Gallios flow cytometer (Beckman Coulter). Analyses were performed using Kaluza version 1.3 and 2.1 (Beckman Coulter). Abs and other reagents are described in the key resources table. Enzyme-linked immunosorbent assay (ELISA) Serum samples from NP-KLH immunized mice (see above) were analyzed in duplicates serially diluted on 96-well flat-bottom microtiter plates (Greiner bio-one) coated with 1mg/ml NP 4 -BSA or NP 20 -BSA conjugates (Biosearch Technologies) in 50ml/well coating Cell Reports 32, 108030, August 11, 2020 e3 Article ll OPEN ACCESS buffer (15mM Na 2 CO 3 , 35mM NaHCO 3 ) overnight at 4C. For anti double-stranded DNA ELISA, plates were coated with Poly-L-Lysin (10 mg/ml) in 10mM Tris/ 0.1mM EDTA / pH8.0 overnight at 4C, washed 3 times with Tris/EDTA buffer and coated with dsDNA (10 mg/ml) from calf thymus (Sigma) in Tris/EDTA. Plates were washed 3x with PBS, 0.05% Tween20 and blocked for 1h at room temperature with 275ml/well of PBS, 2%FCS. Binding of Abs from pre-diluted serum samples was allowed for 1h at room temperature. Captured NP-specific Abs were detected with goat anti-mouse IgM and IgG specific horseradish peroxidase (HRP)- coupled Abs (1:1000, Southern Biotech) and the ELISA was developed using o-phenoldiamindihydrochloride (OPD) substrate (20mM Na 2 HPO 4 , 7mM citric acid, 0.001% OPD, 0.035% H 2 O 2 ) and acid stop (0.5M H 2 SO 4 ). Optical density (oD) was measured at 490nm on a SpectraMax 190 Microplate Reader (Molecular Devices). Plates were normalized using ELISA IgM and IgG standards as internal reference. Purification of murine B lymphocytes from spleen B cells were enriched from splenic cell suspensions using the EasySep Mouse B cell isolation negative selection kit (EasySep #19854, StemCell Technologies, Inc.) according to the manufacturer’s instructions. In short, spleen cells were resuspended in MACS buffer (PBS, 2% FCS, 2mM EDTA), surface blocked with rat serum and immunomagnetically enriched for untouched naive B cells. Purity of isolated B cells was verified by surface stain for CD19. Usually, an enrichment of > 95% was achieved. In vitro cultivation of primary murine B cells Splenic B cells were cultured with a starting concentration of 1x10 6 cells/ ml in R10 medium (RPMI1640, 10% fetal calf serum (FCS), 2mM glutamate, 1mM sodium pyruvate, 50 U/ml penicillin G, 50 mg/ml streptomycin, 50 mMb-mercaptoethanol) for 48h at 37C and 5% CO 2, supplemented with 10mg/ml anti-IgM F(ab’) 2 Ab (AffiniPure F(ab’) 2 fragment goat anti-mouse IgM, mchain specific, Jackson ImmunoResearch), 10mg/ml rat anti-CD40 Ab (clone FGK45, (Rolink et al., 1996) and 0.1U/ml IL-4 (mouse IL-4, premium-grade, Miltenyi Biotec). In vitro migration assay of activated B cells Migration assays were performed as described in Liu et al. (2016). ICAM1-Fc fusion protein was obtained by cloning a PCR fragment (fwd primer: tgtcacgaattcgcaggtatccatccatcccagagaag, rev: atgcagatctgttattttgagagtggtacagt) obtained using murine lung cDNA as a template. The PCR product was cloned into pFUSE-rIgG-Fc2 (Invivogen) via BglII/EcoRI restriction sites, replacing the endogenous signal peptide with the vector encoded IL-2 signal sequence and the membrane domain with a rabbit IgG Fc2 part. ICAM-1Fc was purified from the supernatant of transfected 293 cells by Protein A Sepharose (Thermo Sientific). 8-well polymer chambered coverslips (Ibidi) were coated with 10mg/ml Protein A (Sigma Aldrich) and 5mg/ml recombinant mouse CXCL13 or CXCL12 (carrier-free, Biolegend) in 200ml PBS/well for 2h at room temperature. Chambers were rinsed once with 10mg/ml fatty acid-free BSA (Sigma Aldrich) in PBS and coated with 10mg/ml ICAM-1 Fc in PBS for 2h at room temperature. After additional washing chambers were blocked with 10mg/ml fatty acid free BSA in PBS over night at 4C. The coverslips were finally washed with pre-warmed migration medium (phenol red-free RPMI1640, 2mM glutamate, 1mM sodium pyruvate, 50 U/ml penicillin G, 50 mg/ml streptomycin, 50 mMb-mercaptoethanol, 10mg/ml fatty acid-free BSA) and kept humidified at 37C until use. B cells were washed by centrifugation at 300xg for 5min at 4C and seeded in pre-warmed phenol red-free migration medium at a concentration of 5x10 5 cells/ ml in 200ml/well. F-Actin of living cells was labeled using SiR-actin (1:1000, Spirochrome). Spinning disc confocal microscopy Live cell migration was analyzed using the Zeiss Spinning Disc Axio Observer Z1, with a LD LCI Plan-Apochromat 25x/0.8 Imm Korr DIC M27 or Plan-Apochromat 63x/1.40 Oil M27 objective. All experiments were performed at 37C and 5%CO 2 . Time series were acquired using an Evolve camera and Zeiss Zen software. Fluorescence (638nm laser wavelength) and DIC was recorded. Images were processed with the Fiji software using manual tracking macro to track cell migration. The freeware Chemotaxis and Migration Tool (Ibidi) was used for further quantification of cell velocity. Adoptive B cell transfer Immunization of donor (WT and EFhd2KO) and recipient mice with SRBC and the isolation of splenic B cells from the donors at d7 post antigen challenge was performed as described above. 20x10 6 B cells per genotype were fluorescently labeled with Hoechst and CMTPX (both Thermo Scientific) or vice versa. WT and EFhd2KO B cells were mixed in a 1:1 ratio and transferred i.v. into recipient mice. 24h post transfer and 4h before imaging, follicular dendritic cells (FDCs) of recipient mice were in vivo labeled via i.v. application of anti-CD21/35-Atto590 F(ab’)-Ab or CD21/35-Alexa647 F(ab’)-Ab (coupled in-house at DRFZ, Berlin). For multi-photon microscopy of B cells in splenic germinal centers, spleen was isolated, transferred into pre-warmed R10 medium and cut into slices (1-2mm thickness). Migration of cells was tracked for 30-50min. B-T-conjugates were quantified as described previously (Lee et al., 2017;Reinhardt et al., 2009). Briefly, 15x10 6 splenic B cells from WT and EFhd2KO mice, isolated as described above, were transferred i.v. in a total volume of 75ml PBS into congenically marked Ly5.1 recipient mice. Recipients were challenged with 100mg NP-KLH in alum (described above) 24h later and analyzed at d14 and d21 post antigen application. e4 Cell Reports 32, 108030, August 11, 2020 Article ll OPEN ACCESS Imaging Cytometry Samples were prepared as described above (Detection of surface antigens by flow cytometry) and subjected to analysis with an AmnisFlowSight (Luminex Corp., Austin, TX). Fluorescence spillover was compensated using single-stained cells. Samples were focused based on the signal of CD45.1 (Channel06, PE/Cy7). Debris was gated out from the focused cells based on area and aspect ratio of the bright field (BF). For each sample 3000 events in the CD19 + /CD4 + double positive gate were recorded. Data were analyzed with IDEAS v6.2 (Luminex Corp.). In silico GC model In silico germinal center modeling was as previously described (LEDA model) (Meyer-Hermann, 2014;Meyer-Hermann et al., 2012; Meyer-Hermann et al., 2018). The motility properties were introduced in Figge et al. (2008) and fine-tuned in Binder and Meyer-Hermann (2016). The multiple B-T-interactions have been introduced in detail in Meyer-Hermann et al. (2018). QUANTIFICATION AND STATISTICAL ANALYSIS Values were assessed for Gaussian distribution using the Shapiro-Wilk normality test. Mann-Whitney test was used for non-Gaussian distributed datasets. Datasets revealing Gaussian-like distribution were assessed by Student’s t test. Differences between the analyzed groups were considered to be statistically significant with p values < 0.05. Data were analyzed using Prism (GraphPad). Cell Reports 32, 108030, August 11, 2020 e5 Article ll OPEN ACCESS