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

A icle B Cell Speed and B-FDC Con ac s in Ge minal Cen e s De e mine Plasma Cell Ou pu ia Swip osin-1/EFhd2 G aphical Abs ac Highligh s dEFhd2 limi s mig a ion o GC B cells and hap en-speci ic immuni y dFas GC B cells a e selec ed ea lie and p one o di e en ia e dEFhd2 os e s con ac s o GC B cells wi h ollicula dend i ic cells dCon ac o as GC B cells wi h ollicula dend i ic cells is decisi e Au ho s Do o hea Reime , Michael Meye -He mann, Asylkhan Rakhymzhan, ..., Anja E. Hause , Raluca A. Niesne , Di k Mielenz Co espondence di k.mielenz@ au.de In B ie Reime e al. show ha EFhd2 es ic s mig a ion o ge minal cen e B cells and hap en-speci ic immuni y. Selec ion o speedy ge minal cen e B cells occu s ea lie and hey a e mo e likely o di e en ia e. EFhd2 suppo s con ac o ge minal cen e B cells wi h ollicula dend i ic cells, enabling success ul compe i ion o apidly mig a ing ge minal cen e B cells. Reime e al., 2020, Cell Repo s 32, 108030 Augus 11, 2020 ª2020 The Au ho s. h ps://doi.o g/10.1016/j.cel ep.2020.108030 ll A icle B Cell Speed and B-FDC Con ac s in Ge minal Cen e s De e mine Plasma Cell Ou pu ia Swip osin-1/EFhd2 Do o hea Reime , 1 Michael Meye -He mann, 2 Asylkhan Rakhymzhan, 3 Tobi S einme z, 1 Philipp T ipal, 4 Jana Thomas, 1 Ma in Boe che , 5 Dimi ios Mougiakakos, 5 Sebas ian R. Schulz, 1 Sophia U banczyk, 1 Anja E. Hause , 3,6 Raluca A. Niesne , 3,7,8 and Di k Mielenz 1,8,9, * 1 Di ision o Molecula Immunology, Uni e si a ¨ sklinikum E langen, Nikolaus-Fiebige -Zen um, FAU E langen-N€ u nbe g, E langen, Ge many 2 Depa men o Sys ems Immunology and B aunschweig, In eg a ed Cen e o Sys ems Biology, Helmhol z Cen e o In ec ion Resea ch, B aunschweig, Ge many 3 Deu sches Rheuma o schungszen um (DRFZ), Be lin, Ge many 4 Op ical Imaging Cen e (OICE), Uni e si a ¨ sklinikum E langen, FAU E langen-N€ u nbe g, E langen, Ge many 5 Depa men o In e nal Medicine V, Uni e si a ¨ sklinikum E langen, FAU E langen-N€ u nbe g, E langen, Ge many 6 Cha i e ´– Uni e si y Medicine, Be lin, Ge many 7 Dynamic and Func ional In Vi o Imaging, Ve e ina y Medicine, F eie Uni e si a ¨ , Be lin, Ge many 8 These au ho s con ibu ed equally 9 Lead Con ac *Co espondence: di k.mielen[email p o ec ed] h ps://doi.o g/10.1016/j.cel ep.2020.108030 SUMMARY Plasma cells sec e ing a ini y-ma u ed an ibodies de elop in ge minal cen e s (GCs), whe e B cells mig a e pe sis en ly and di ec ionally o e de ined pe iods o ime. How modes o GC B cell mig a ion in luence plasma cell de elopmen emained unclea . Th ough gene ic dele ion o he F-ac in bundling p o ein Swip o- sin-1/EF-hand domain amily membe 2 (EFhd2) and by wo-pho on mic oscopy, we show ha EFhd2 e- s ains B cell speed in GCs and hap en-speci ic plasma cell ou pu . Modeling he GC eac ion e eals ha inc easing GC B cell speed p omo es plasma cell gene a ion. Lack o EFhd2 also educes con ac s o GC B cells wi h ollicula dend i ic cells in i o. Compu a ional modeling unco e s ha bo h GC ou pu and an i- body a ini y depend quan i a i ely on con ac s o GC B cells wi h ollicula dend i ic cells when B cells mig a e mo e pe sis en ly. Collec i ely, ou da a explain how GC B cells in eg a e speed and pe sis ence o cell mig a ion wi h B cell ecep o a ini y. INTRODUCTION The de elopmen o memo y B cells and long-li ed plasma cells (PCs) sec e ing a ini y-ma u ed an ibodies (Abs) (Be ek e al., 1991) depends on he ge minal cen e (GC) eac ion. GCs a e ansien s uc u es ha o m in seconda y lympha ic o gans a ound he ollicula dend i ic cell (FDC) ne wo k 4 days a e T-cell-dependen (TD) an igen challenge (Ga o and B ink, 2010). Ma u e GCs a e highly o de ed s uc u es consis ing o his ologically dis inc egions, he da k zone (DZ) and he ligh zone (LZ), he la e o which is cen e ed on chemokine (C-X-C mo i ) ligand 13 (Cxcl13)-exp essing FDCs (Vic o a and Nus- senzweig, 2012). Typically, GCs o med in esponse o alum p ecipi a ed hap ena ed model an igens las 21–28 days and hen esol e as an igen becomes limi ing and Ab eedback en- sues (Ga o and B ink, 2010). Long-las ing o e en ch onic GCs de elop a e immuniza ion wi h pa icula e an igen, such as sheep ed blood cells (SRBCs) (Dogan e al., 2009), in esponse o ce ain in ec ions as a consequence o hype immu- niza ion o na u ally in he Peye ’s pa ches. Immunoglobulin (Ig) gene pedig ees de i ed om GC B cells (Be ek e al., 1991) sug- ges ed a Da winian mechanism o Ab a ini y ma u a ion. These obse ed Ig gene mu a ion a es a e econciled wi h DZ/LZ a - chi ec u e o GCs and cell-cycle du a ion in he cyclic- een y model o he GC eac ion (Keple and Pe elson, 1993). Acco d- ingly, soma ic hype mu a ion (SHM) occu s in he DZ and clonal selec ion occu s in he LZ, p edic ing a equi emen o in e zonal B cell mig a ion. In i o mul ipho on mic oscopy co obo a ed his p edic ion, wi h C-X-C mo i chemokine ecep o 4 (Cxc 4)-exp essing cen oblas s mig a ing om he DZ o he LZ and Cxc 5-exp essing cen ocy es mig a ing om he LZ o he DZ (Hause e al., 2007). GC B cells mig a e in a di ec ed manne (O’Conno e al., 2011), whe eby he ne low appea s o be om he DZ o he LZ (Bel man e al., 2011), wi h only ew B cells mig a ing back in o he DZ (Vic o a e al., 2010). Tagging GC B cells wi h pho oac i a able GFP in i o has e ealed ha hose B cells wi h supe io an igen p esen a ion abili y a e p e e en ially selec ed by T ollicula helpe (T h) cells, esul ing in hei mig a ion back in o he DZ and inc eased PC gene a ion (Vic o a e al., 2010). The e is e idence ha B cell/T Cell Repo s 32, 108030, Augus 11, 2020 ª2020 The Au ho s. 1 This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). ll OPEN ACCESS (legend on nex page) 2Cell Repo s 32, 108030, Augus 11, 2020 A icle ll OPEN ACCESS cell in e ac ions de e mine he numbe o di isions o a B cell in he DZ (Meye -He mann e al., 2012). B cell clones ca ying selec ed B cell ecep o (BCR) wi h compa ably high a ini y a e p edominan ly ound in he PC compa men , while memo y B cells ha bo less a ine BCR genes and a ise ea lie in he GC eac ion (Weisel e al., 2016). As PCs ca y Ig genes encoding highe a ine BCRs and la e Abs, a ole o BCR signaling in addi- ion o BCR-media ed endocy osis in dic a ing GC a e owa d he PC p og am is no oo specula i e. Indeed, ini ia ion o he PC p og am equi es BCR con ac wi h an igen con ained in im- mune complexes bound on FDCs, ollowed by comple ion o he PC p og am by signals de i ed om T h cells (K a ¨u le e al., 2017) and specialized s omal cells localized a he T cell/B cell bo de (Zhang e al., 2018b). In line, B cells di ec ly es BCR a - ini y on an igen p esen ed by FDCs h ough o ce applica ion ia myosin II mo o s in a specialized GC synapse (Na kanski e al., 2013;Nowosad e al., 2016). Only BCR signaling, oge he wi h CD40 s imula ion, enables cen ocy es o up egula e Myc (Luo e al., 2018), highligh ing a ole o BCR signaling and T cell help in he selec ion o GC cells. Whe eas ansc ip ion ac o s such as I 4 (Klein e al., 2006), Myc (Calado e al., 2012), o Rel (Heise e al., 2014) ha di ec he cell a e in GCs ha e been well cha ac e ized, p o eins linking he BCR o CD40 o GC ou pu a e less es ablished. Those p o eins migh be ound among hose con olling B cell mig a ion o B cell/FDC in e ac- ion in GCs (Tola , 2017), because a main ask o B cells ha ing unde gone and su i ed SHM in he DZ (Maye e al., 2017)is o ind an igen bound on FDCs as well as a cogna e T h cell. In line, many epo s emphasize he impo ance o cell mig a ion pa e ns o GC dynamics, pa icula ly o hei PC ou pu ( e- iewed in Vic o a and Nussenzweig, 2012). Howe e , i emains elusi e how di e en cell mig a ion pa e ns di ec ly impac he GC eac ion and, consequen ly, PC ma u a ion and which mole- cules wi hin B cells d i e hei mo ili y beha io . Swip osin-1/EF- hand domain amily membe 2 (EFhd2; also known as Swip osin- 1, and no o be con used wi h Swip osin-2/EFhd1; D€ u ing e al., 2011)isa30-kDa Ca 2+ - and F-ac in-binding p o ein (Hagen e al., 2012;Huh e al., 2013;Kwon e al., 2013) wi h a C- e minal coiled-coil domain and unc ional SH3-binding si es a he N e - minus (K oczek e al., 2010). Null mu a ion o EFhd2 (EFhd2KO) e ealed ha EFhd2 nega i ely con ols GC expansion in esponse o seconda y Nippos ongylus b asiliensis in ec ion (B achs e al., 2014). EFhd2 also con ols de elopmen o IgG1 and IgE PCs as well as IgM and IgE se um Abs in esponse o in ec ion wi h N. b asiliensis (B achs e al., 2014) in a B-cell- in insic manne . How EFhd2 con ols GC-de i ed PC gene a ion and whe he EFhd2 con ols Ab quali y was unknown. EFhd2 suppo s lamellipodia o ma ion, lung cance , and melanoma me as asis as well as mac ophage mig a ion h ough egula ion o ac in dynamics by Rho GTPases (Fan e al., 2017;Huh e al., 2015;Tu e al., 2018;Zhang e al., 2018a). Based he eupon, we hypo hesized ha EFhd2 con ols he GC eac ion and PC de el- opmen ia con ol o ac in dynamics and cell mig a ion. Hence, we analyzed B cell dynamics o EFhd2 wild- ype (WT) and knockou (KO) B cells in i o and in i o using wo-pho on mic o- scopy, oge he wi h in silico modeling and a de ailed kine ic analysis o he GC and PC esponse agains ni ophenol (NP)- keyhole limpe hemocyanin (KLH). We ound ha EFhd2 limi s B cell mig a ion and ha enhanced B cell mig a ion h ough lack o EFhd2 leads o inc eased PC ou pu . Howe e , a ma he- ma ically p edic ed inc ease in BCR a ini y was coun e bal- anced by educed in e ac ions o EFhd2KO B cells wi h FDCs. Using his i e a i e app oach, we de ined he gene al concep (independen ly o EFhd2KO) ha cell mig a ion alone canno elici s onge GC eac ions o Ab a ini y, and B cell/FDC in e ac- ions a e equi ed o suppo he accele a ing e ec o B cell mig a ion. RESULTS EFhd2KO B Cells Mig a e Fas e in Ge minal Cen e s To es whe he EFhd2 a ec s B cell mig a ion and PC gene a- ion, we isola ed and labeled EFhd2WT and EFhd2KO B cells (B achs e al., 2014) om SRBC-immunized mice and ans- e ed hem back in o SRBC-immunized WT ecipien s (Fig- u e 1A). Mig a ion o luo escen ly labeled B cells in GCs and h ough he FDC ne wo k was assessed in li ing spleen by mul i- pho on mic oscopy (Heinig e al., 2014;Rakhymzhan e al., 2017) (Video S1;Figu e 1B). He eby, we assessed se e al pa ame e s: (1) eloci y, (2) speed, (3) dis ance, (4) displacemen , and (5) di ec edness (Figu e 1C–1G; see Figu e S1 o de ini ions and explana ions). EFhd2KO B cells mig a ed app oxima ely wo imes as e in GCs in compe i ion wi h WT B cells, wi h bo h mean eloci y and speed being ele a ed (Figu e 1C, D). Displacemen o EFhd2KO B cells and co e ed dis ance we e p opo ionally enhanced (Figu e 1E, F), e ealing unal e ed di ec edness (displacemen /dis ance; Masuzzo e al., 2016)(Fig- u e 1G), al hough EFhd2 has been p oposed o de e mine he Figu e 1. EFhd2 De iciency Enhances B Cell Mo ili y in GCs (A) WT and EFhd2KO dono as well as WT ecipien mice we e immunized in ape i oneally (i.p.) wi h SRBCs. A day 7, splenic B cells om dono mice we e isola ed and luo escen ly labeled. WT B cells, Hoechs , EFhd2KO B cells, CMTPX, and ice e sa, mixed in a 1:1 a io, we e ans e ed in a enously (i. .) in o WT ecipien mice. One day a e ans e , he FDC ne wo k was labeled ia i. . applica ion o an i-CD21/32-A o590 Fab 2 -Ab, and mig a ion o WT and EFhd2KO B cells wi hin splenic GC was acked. (B) Rep esen a i e image o wo adjacen GCs, iden i ied ia he FDC ne wo ks ( u quoise, an i-CD21/32-A o590), con aining WT B cells (blue, Hoechs ) and EFhd2KO B cells ( ed, CMTPX). Scale ba , 100 mm. (C–G) Plo s on he le side display WT and EFhd2KO B cells acked wi hin one ep esen a i e ecipien mouse (one do = one cell). Plo s on he igh side show mean alues o WT and EFhd2KO B cells in i e independen ecipien mice (one do = one mouse). (C and D) Analysis o cell mo emen by cell eloci y (C) and speed (D). (E and F) Analysis o he ac ually co e ed pa hs quan i ied by displacemen (E) and dis ance (F). (G) The di ec edness o cell mig a ion was calcula ed as he a io o dis ance o displacemen . Da a a e ep esen ed as mean alues; s a is ics: N = 2, n = 5; signi icance was analyzed using Mann-Whi ney es o single cell compa ison (****p < 0.0001, le panels in C–G) and pai ed es o compa ing mean alues o ecipien mice (nonsigni ican [n.s.], p > 0.05; *p %0.05; ***p %0.001, igh panels in C–G). Cell Repo s 32, 108030, Augus 11, 2020 3 A icle ll OPEN ACCESS di ec ion o cell p o usions (Huh e al., 2015). Impo an ly, he speed o WT B cells was exac ly as epo ed p e iously (Allen e al., 2007;Hause e al., 2007;Schwicke e al., 2007). Mig a ion, ac i a ion, and a achmen o B cells is dic a ed by egula ed exp ession o chemokine ecep o s Cxc 4 and Cxc 5, in eg ins, and hei ligands, such as he lymphocy e- unc ion-associa ed an igen 1 (LFA-1)/in e cellula adhesion molecule 1 (ICAM-1) combina ion, as well as by B cell ac i a ion (Ca asco e al., 2004;Liu e al., 2016; e iewed in Vic o a and Nussenzweig, 2012). The as e mig a ion o EFhd2KO B cells in i o was no due o al e ed exp ession o Cxc 4, Cxc 5, LFA-1, e y la e an igen 4 (VLA-4), ICAM-1, o ICAM-2 (Fig- u e S2). We nex wished o de e mine which ac i a ing s imulus igge s EFhd2KO B cells. Since GC B cells ha e ecei ed a ple ho a o signals ha canno be dissec ed, we measu ed che- mokinesis o SiR-ac in-labeled B cells on ecombinan ICAM- 1Fc usion p o ein (Liu e al., 2016)(Videos S2 and S3;Figu e 2A). An i-BCR-ac i a ed, bu no an i-CD40/in e leukin-4 (IL-4)-ac i- a ed, EFhd2KO B cells mig a ed as e on ICAM-1 in i o in esponse o Cxcl13 ( igge ing Cxc 5) (Figu e 2B), bu no o Cxcl12 ( igge ing Cxc 4) (Figu e 2C). In iguingly, BCR ac i a ion in combina ion wi h an i-CD40/IL-4 s imula ion a es ed B cell mig a ion (Figu e 2B) in bo h WT and EFhd2KO B cells. Taken oge he , EFhd2 limi s B cell mig a ion in GCs and in i o.In pa icula , BCR ac i a ion appea s o modula e EFhd2- es ic ed B cell mig a ion. Compu a ional Modeling o GC Dynamics Based on Expe imen ally De e mined Pa ame e s o B Cell Mig a ion To es whe he mo e apid cell mig a ion modula es GC e- sponses we supplied an es ablished ma hema ical model o he GC eac ion (Binde and Meye -He mann, 2016;Figge e al., 2008;Meye -He mann, 2014;Meye -He mann e al., 2009,2012) wi h he B cell mig a ion da a (shown in Figu e 1). As EFhd2KO B cells mig a ed app oxima ely wo imes as e (compa e Figu e 1), we simula ed GCs wi h no mal B cell mig a- ion a es (7.5 mmin/min/1.5 min pe sis ence ime, black lines) and wo imes as e B cells (15 mmin/min/1.5 min pe sis ence ime, pink lines). Fas e mig a ion may also be he esul o inc eased di ec ional pe sis ence (Maiu i e al., 2015). We he e- o e also modeled B cells wi h wo imes and ou imes mo e pe sis ence (7.5 mmin/min/3.0 min pe sis ence and 7.5 mmin/ min/6.0 min pe sis ence, g een and blue lines) and wi h wo- imes- as e -mig a ing B cells wi h a 2- old inc eased pe sis- ence (7.5 mmin/min/3.0 min pe sis ence, ed line). Ele a ing cell mig a ion o pe sis ence in silico was su icien o elici a dose-dependen inc ease in he o al numbe o GC and ou pu cells (PCs and memo y B cells) (Figu es 3A and 3B). Impo an ly, he e ec o speed was mo e dominan han he e ec o pe sis- ence (Figu e 3B). P edic ed a ini y ma u a ion o GC B cells was unal e ed by speed o pe sis ence, whe eas doubled speed, bu no doubled pe sis ence, p edic ed a small (15%) inc ease in a ini y o GC ou pu cells (Figu e 3C). By plo ing selec ion p ob- abili y e sus a ini y, i became e iden ha as e B cells a e likely o be selec ed ea ly (days 3–5), bu no la e (days 5–9) (Fig- u e 3D). Taken oge he , inc easing cell mig a ion and pe sis- ence in simula ed GCs led o he ollowing expec a ions ha we e e alua ed in ou model sys em, EFhd2KO mice (expe i- men a o s we e blinded o he esul s o he simula ion): (1) inc eased GC B cell numbe s, (2) enhanced PC ou pu , and (3) a mode a e inc emen o he mean a ini y o GC-de i ed BCRs. EFhd2 Limi s PC Di e en ia ion Doubling B cell speed in GCs p edic ed inc eased GC B cell numbe s (Figu e 3A). Howe e , we did no egis e highe numbe s o GC B cells o hap en-speci ic GC B cells a any Figu e 2. BCR Ac i a ion Enhances Mig a ion o EFhd2KO B Cells on ICAM-1 and CXCL13 (A) Nai e splenic B cells om WT and EFhd2KO mice we e isola ed and ac i a ed wi h an i-IgM Ab, an i-CD40 Ab + IL-4, o an i-IgM Ab + an i-CD40 Ab + IL-4 o 48 h. (B and C) B cell mig a ion on ICAM-1 in he p esence o CXCL13 (B) o CXCL12 (C) was analyzed by spinning disc con ocal mic oscopy. Da ase s a e shown o indi idual cells (one do = one B cell) ac i a ed as desc ibed abo e. S a- is ics: N = 2, n = 2–3; signi icance was analyzed using Mann-Whi ney es (n.s. > 0.05; **p %0.01). 4Cell Repo s 32, 108030, Augus 11, 2020 A icle ll OPEN ACCESS ime poin be ween days 7 and 21 upon NP-KLH immuniza ion o EFhd2KO mice (Figu es 4A–4C). To es ne e heless whe he EFhd2 in luences PC gene a ion om GCs, we c ossed EFhd2KO mice wi h Blimp1:GFP epo e mice (Kallies e al., 2004), immunized mice wi h NP-KLH, and analyzed Blimp1 exp ession (GFP luo escence) in CD38 low CD95 + B cells a day 21. We ound a clea ly Blimp1-posi i e cell ac ion in CD19 + CD38 low CD95 GL7  cells (Figu e 4D). In iguingly, we de- ec ed a highe equency o Blimp1-posi i e CD38 low CD95 + GL7  cells in EFhd2KO mice (Figu e 4E), while o al cell numbe s we e ha dly ele a ed (Figu e 4F). We explain his by educed e- quencies and numbe s o CD19 + CD38 low B cells in he Blimp1:GFP;EFhd2KO mice (Figu es 4G and 4H). In summa y, Blimp1-posi i e la e GC B cells a e inc eased in ela ion o he Figu e 3. Enhanced B Cell Speed and Pe sis ence Inc ease GC-De i ed PC Ou pu and A ini y (A and B) In silico simula ion o GC modula ed by speed and pe sis ence B cell mig a ion. P edic ed esul s a e shown o he o al numbe o GC B cells, hei dis ibu ion in da k zone (DZ) and ligh zone (LZ) (A), and he gene a ion o ou pu cells (B). (C) A ini y ma u a ion was simula ed o GC B cells as well as o he ou pu cells. (D) The selec ion p obabili y in dependence o B cell a ini y was analyzed o di e en ime poin s du ing he GC cou se. Da a a e shown as mean (solid lines) o 50 simula ions ±SD (g ay shades). Cell Repo s 32, 108030, Augus 11, 2020 5 A icle ll OPEN ACCESS Figu e 4. No mal Hap en-Ca ie -Induced GC B Cells bu Inc eased GC-De i ed PCs in EFhd2-De icien Mice (A) WT and EFhd2KO mice we e immunized i.p. wi h 100 mgNP 29 -KLH in alum and analyzed a e 7, 14, and 21 days. An igen-speci ic and o al GC B cell and PC popula ions om spleen we e de e mined ia luo escence-ac i a ed cell so ing (FACS). Me ged FACS do plo s o GC B cells a e displayed o day 21, p e- ga ed on single s and iable lymphocy es. (B and C) F equencies o o al GC B cells de ined as CD19 + CD38 lo CD95 + GL7 + a days 7, 14, and 21 pos -immuniza ion. Respec i e an igen-speci ic GC B cells, iden i ied by binding o NP 28 -PE, we e quan i ied as equency o NP + cells o o al CD19 + CD38 lo CD95 + GL7 + B cells. Da a a e ep esen ed as mean alues. S a is ics: N = 2, n = 4–8; alues passed he es o Gaussian dis ibu ion (Shapi o-Wilk no mali y es ), and signi icance was he e o e analyzed using unpai ed es (n.s., p > 0.05). (D) Blimp1:GFP mice and Blimp1:GFP 3EFhd2KO mice we e immunized i.p. wi h 100 mgNP 29 -KLH in alum. Spleens we e analyzed o Blimp1:GFP exp ession o CD19 + CD38 lo CD95 + GL7 lo B cells by FACS. Me ged FACS do plo s o GC-de i ed PCs a e displayed o day 21, p e-ga ed on single s and iable lymphocy es. (legend con inued on nex page) 6Cell Repo s 32, 108030, Augus 11, 2020 A icle ll OPEN ACCESS equencies and numbe s o CD19 + CD38 low B cells. These da a poin o inc emen ed PC gene a ion wi hin EFhd2KO GCs as a consequence o cell mig a ion, which is in ag eemen wi h he p edic ions (Figu e 3). I ac in-dependen cy oskele al eo gani- za ion can in ac in luence PC gene a ion om GCs, hen he exp ession o ac in modula ing p o eins, including EFhd2, should be al e ed in (1) GCs and (2) he ac ion o GC B cells ha is p edes ined o become PCs ( ac ion 1 in Ise e al., 2018; Cxc 4 low , CD86 + , Bcl6 low , I 4 + , and CD69 high ). We he e- o e analyzed a p e-exis ing RNA sequencing (RNA-seq) da ase om ac ion 1 B cells as well as o he GC subse s (Ise e al., 2018; ac ion 2: Cxc 4 low , CD86 + , Bcl6 high , I 4  , and CD69 in ; ac ion 3: Cxc 4 low , CD86 + , Bcl6 high , and I 4  , CD69 low ; ac ion 4: Cxc 4 low , CD86 + , Bcl6 low , I 4  , and CD69 low ; and plasma- blas s). These e-analyses showed ha genes con olling he B cell ac in cy oskele on, including EFhd2, a e sha ply egula ed a he ansi ion o GC B cells and PCs and s ongly en iched in ac ion 1 (Figu e S3), suppo ing he idea o an ac in-depen- den checkpoin in GCs con olling PC de elopmen . EFhd2 Limi s An igen-Speci ic PC Ou pu om he GC We nex acked hap en-speci ic GC B cells, PCs, and se um Abs o e 21 days a e NP-KLH immuniza ion (Figu e 5). Flow cy- ome y e ealed inc eased NP-speci ic PC equencies in he spleen and bone ma ow o EFhd2KO mice a day 21 (Figu es 5C and 5F). In addi ion, we obse ed inc eased NP-speci ic se um IgG (o all iso ypes; no depic ed he e), bu no IgM Abs (Figu es 5G and 5H), in immunized EFhd2KO mice. These esul s a e in ag eemen wi h GC simula ion and he inc ease in Blimp1-exp essing B cells a day 21 in immunized Blimp1:GFP;EFhd2KO mice (Figu e 4A). The inc eased NP-spe- ci ic Abs o EFhd2KO mice we e dominan ly o lowe a ini y (Fig- u e 5H). The a io o high- (an i NP(4)-BSA) e sus low-a ine (an i NP(29)-BSA) Abs ha was p edic ed o be sligh ly inc eased was su p isingly no al e ed. I a ini y ma u a ion was se e ely al e ed, hen NP binding o GC B cells as de e mined by low cy- ome y would be al e ed as well, which was no he case. Conside ing ha measu emen o a ini y ma u a ion by ELISA is widely used, hese expe imen s s ongly sugges ha a ini y ma u a ion is no al e ed in EFhd2KO B cells. Ne e heless, he obse ed inc ease in NP-speci ic Abs appea ed o be speci ic, as we did no obse e inc eased o ma ion o au o-Abs, a leas no an i-double s anded DNA Abs (Figu e S4). Fu he mo e, he T-cell-independen ype 2 an igen NP-Ficoll did no elici an inc eased esponse in EFhd2KO mice (Figu e S4). Thus, se e al lines o e idence e eal ha EFhd2 limi s PC gene a ion om GCs, wi h cell mig a ion limi ed by EFhd2 likely playing a ole he ein (Figu e 3B). The inc ease in PCs was no due o inc eased B cell/T cell in e ac ions as de e mined by con en ional low as well as Amnis imaging cy ome y (Figu es S5 and S6)o inc eased T h cell numbe s (Figu e S7). An in silico p edic ion ha was no con i med in EFhd2KO mice (Figu e 4A) was he in- c ease in GC B cells induced by inc easing B cell speed o pe sis ence in silico (Figu e 3A) and an inc ease in Ab a ini y. This aised he possibili y ha cell mig a ion was no he only pa ame e al e ed by EFhd2 de iciency in GC B cells. Since he ac in cy oskele on is also c ucial o he B cell/FDC synapse (Tola , 2017), we nex assessed how EFhd2KO B cells in e ac wi h FDCs. EFhd2KO B Cells Ha e Less Con ac wi h FDCs To es whe he EFhd2KO B cells in e ac di e en ly wi h FDCs in i o, we acked EFhd2KO and WT B cells wi hin he FDC ne wo k (Figu e 6). The colocaliza ion olume o EFhd2KO B cells wi h FDCs was educed (Figu e 6B), bu he du a ion o in- e ac ions was simila (Figu e 6C). Howe e , he B cell/FDC con- ac s appea ed o be mo e con ined on he FDC su ace, since he displacemen o EFhd2KO B cell con ac s was educed (Fig- u e 6D). We conclude ha EFhd2KO GC B cells in e ac di e - en ly wi h FDCs han WT GC B cells. These da a could be ex- plained by al e ed ac in cy oskele on dynamics. The Ac in Cy oskele on o EFhd2KO B Cells Is Highly Dynamic Ac i a ed B cells mo e on ICAM-1 in he p esence o Cxcl13 in i o by exhibi ing equen , spon aneous dila ion and sh ink- ing e en s a he sides o he leading memb ane edge, which is p edic i e o u ning e sus di ec ional pe sis ence (Liu e al., 2016). These condi ions a e a leas pa ially ound on and wi hin he FDC ne wo k. The as e mig a ion o BCR-ac i a ed EFhd2KO B cells in esponse o Cxcl13 (Figu e 2B) coincided wi h inc eased ac in dynamics (Figu e 6E). We obse ed speci - ically ha he U opod ( ailing edge; Figu e 6F) is s able in WT B cells bu highly dynamic in EFhd2KO B cells (see s a ic images in Figu e 6E and Videos S2 [WT] and S3 [EFhd2KO]), showing many p o usion-con ac ile e en s in EFhd2KO B cells (Figu e 6G). We also obse ed an inc eased pool o F-ac in a he ailing edge o EFhd2KO B cells (Figu e 6H). A Dec ease o B-FDC Con ac P obabili y Opposes he GC Inc ease Induced by High Speed and Pe sis ence I he e was a quan i a i e o quali a i e di e ence in he B cell/ FDC synapses o EFhd2KO B cells, hen posi i e selec ion could be impai ed, educing he equency o ecycling GC B cells. In his case, he inc eased GC ou pu p edic ed o occu when GC B cells mig a e as e could be coun e balanced. To econ- cile he compu e simula ion wi h ou expe imen al da a, we changed he likelihood o B cell/FDC in e ac ion as a su oga e (E) F equencies o GC-de i ed PCs de ined as Blimp1 + CD19 + CD38 low CD95 + GL7 low a day 21 pos -immuniza ion. Da a a e ep esen ed as mean alues. S a- is ics: N = 2, n = 5; signi icance was analyzed using Mann-Whi ney es (**p %0.01). (F) Absolu e numbe s o GC-de i ed PCs de ined as Blimp1 + CD19 + CD38 low CD95 + GL7 low a day 21 pos -immuniza ion. Da a a e ep esen ed as mean alues. S a is ics: N = 2, n = 5; signi icance was analyzed using Mann-Whi ney es . (G) F equencies o GC B cells de ined as CD19 + CD38 low a day 21 pos -immuniza ion. Da a a e ep esen ed as mean alues. S a is ics: N = 2, n = 5; signi icance was analyzed using Mann-Whi ney es . (H) Absolu e numbe s o GC B cells de ined as CD19 + CD38 low a day 21 pos -immuniza ion. Da a a e ep esen ed as mean alues. S a is ics: N = 2, n = 5; signi icance was analyzed using Mann-Whi ney es . Cell Repo s 32, 108030, Augus 11, 2020 7 A icle ll OPEN ACCESS measu e o he educed B cell/FDC con ac s we obse ed (Fig- u e 6B). Dec easing B cell/FDC in e ac ion p obabili y by hal indeed comple ely no malized he ele a ed GC esponse elici ed by doubled B cell speed (Figu e 7A) bu did no a ec quan i a- i e GC ou pu (Figu e 7B), he eby app oaching closely he expe imen al da a ob ained wi h EFhd2KO mice immunized wi h NP-KLH. In silico, he educed an igen up ake led o less in ense T h cell signals, which induced less di ision in he DZ and a e e lec ed in a educed DZ/LZ a io (Gi lin e al., 2014; Meye -He mann, 2014)(Figu e 7A). Conside ing a educed B cell/FDC in e ac ion p obabili y in he case o simula ed doubled B cell pe sis ence ins ead o doubled B cell speed educed he inc eased GC esponse below he s anda d se ing (Figu e 7A), educed he DZ/LZ a io e en mo e, and also educed he in- c ease o GC ou pu (Figu e 7B). The highe a ini y o GC ou pu cells induced by doubling speed was ha dly a ec ed by in silico Figu e 5. EFhd2 De iciency Inc eases An i- gen-Speci ic PC Ou pu wi hou Inc easing Ab A ini y WT and EFhd2KO mice we e immunized i.p. wi h 100 mgNP 29 -KLH in alum, and spleen and bone ma ow we e analyzed a e 7, 14, and 21 days o an igen-speci ic and o al plasmablas /PC pop- ula ions ia FACS. NP-binding immunoglobulins we e de ec ed by se um-ELISA. Me ged FACS do plo s a e displayed o day 21 o he spleen (A) and bone ma ow (BM) (D), p e-ga ed on single s and iable lymphocy es. F equencies o o al plasmablas s/PCs de ined as CD138 + TACI + in spleen (B) and bone ma ow (E). An igen-speci ic plasmablas s/PCs, iden i ied by binding o NP 28 - PE, we e quan i ied as he equency o NP + cells o o al CD138 + TACI + cells o spleen (C) and bone ma ow (F). Se a om day 21 we e analyzed o high- and low-a ine NP-speci ic IgM (G) and IgG (H). Da a a e ep esen ed as mean alues. S a is- ics: N = 2, n = 4–8; Shapi o-Wilk and unpai ed es (n.s. > 0.05; *p %0.05; **p %0.01). educ ion o B cell/FDC in e ac ion bu was dec eased in he case o doubled pe sis ence (Figu e 7C). We conclude ha speed and pe sis ence o B cell mig a ion in GCs, along wi h he p open- si y o access an igen on FDCs o ind suppo ing s omal cells, a e decisi e o quan i a i e and quali a i e GC ou pu . How can hese heo e ical da a in u n be alida ed? In EFhd2KO mice, we did obse e inc eased an igen-speci ic PC ou pu wi hou an inc ease in a ini y (Fig- u e 5). These da a poin o a unc ional consequence o he obse ed educed B cell/FDC con ac s, which could be educed posi i e selec ion, explaining he phenomenon o inc eased PC ou pu wi hou inc eased Ab a ini y. Ou expe i- men s we e pe o med unde noncom- pe i i e condi ions, whe e e en B cells wi hou inc eased BCR a ini y may bene i om enhanced mig a ion. To c ea e a compe i i e si ua ion, we adop i ely ans- e ed Ly5.2 EFhd2KO o Ly5.2 WT B cells in o Ly5.1 hos s and hen immunized he mice and acked GC B cells. We obse ed ha EFhd2KO B cells a e ou compe ed a day 21, bu no ye a day 14 (Figu e S8). A simila inding was also obse ed in Blimp1:GFP;EFhd2KO mice (Figu e 4G). We p opose ha B cell/FDC con ac s a e impo an o GC main enance when B cells show inc eased mo ili y unde compe i i e condi ions. DISCUSSION B cells in GCs ha ha e unde gone and su i ed SHM (Maye e al., 2017) cons an ly mo e o ind selec i e an igen bound on FDCs, cogna e T h cells, and s omal cell suppo (Ise e al., 2018;K a ¨u le e al., 2017;Zhang e al., 2018b), equipping 8Cell Repo s 32, 108030, Augus 11, 2020 A icle ll OPEN ACCESS STAR+METHODS KEY RESOURCES TABLE REAGENT o RESOURCE SOURCE IDENTIFIER An ibodies An i CD16/32, clone 93 Biolegend # 101302, RRID:AB_312801 An i TACI/CD267 (APC, clone eBio8F10-3) eBioscience Ca # 17-5942-82, RRID:AB_842758 An i Ly5.2/CD45.2 (Pe CPCy5.5, clone 104) eBioscience Ca # 35-0454-82, RRID:AB_469725 An i CXCR4/CD184 (PE, clone 2b11) eBioscience Ca # 12-9991-82, RRID:AB_891391 An i CD19 (BV421/ APCFi e750, clone 6D5) Biolegend Ca # 115558, RRID:AB_2572120 An i CD38 (Pe CPCy5.5, clone 90) Biolegend Ca # 102722, RRID:AB_2563333 An i CD138 (PECy7, clone 281-2) Biolegend Ca # 142513, RRID:AB_2562197 An i Ly5.1/CD45.1 (PB o PECy7, clone A20) Biolegend Ca # 110730, RRID:AB_1134168 An i CXCR5/CD185 (Bio in o PEDazzle, clone L138D7) Biolegend Ca # 145509, RRID:AB_2562125 An i VLA-4/CD49d-CD29 (Pe CPCy5.5, clone R1-2) Biolegend Ca # 103619, RRID:AB_2563701 An i ICAM-2/CD102 (AF488, clone 3C4) Biolegend Ca # 105609, RRID:AB_2264501 An i LFA-1/CD11a (AF488, clone M17/4) Biolegend Ca # 101111, RRID:AB_493432 An i GL-7 (AF647, clone GL-7) BD Biosciences Ca # 561529 An i CD95 (PECy7, clone Jo2), BD Biosciences Ca # 554254, RRID:AB_395326 An i ICAM-1/CD54 (APC, clone 3E2) BD Biosciences Ca # 561605 An i CD4 (FITC, clone gk1.5) BD Biosciences Ca # 561828 An i CD21/35 (clone 8C12) BD Biosciences Ca # 558768, RRID:AB_397114 Goa an i mouse IgM-HRP Sou he n Bio ech Ca # 1021-05 Goa an i mouse IgG-HRP Sou he n Bio ech Ca # 1030-05 Goa an i-IgM A iniPu e F(ab’) 2 mchain speci ic Jackson ImmunoResea ch Ca # 115-006-020, RRID:AB_2338469 An i CD40, clone FGK45, Rolink e al., 1996 RRID:AB_2490239 Chemicals, Pep ides, and Recombinan P o eins 4-Hyd oxy-3-ni ophenylace yl (Phycoe y h in) Biosea ch Technologies. Ca # N-5070 4-Hyd oxy-3-ni ophenylace yl (keyhole limpe hemocyanin) Biosea ch Technologies. Ca # N-5060 4-Hyd oxy-3-ni ophenylace yl (bo ine se um albumin) Biosea ch Technologies. Ca # N-5050 Hoechs 33342 Sigma Ca # B2261 CellT acke Red CMTPX The mo Fishe Ca # C34552 Imjec Alum The mo Fishe Ca # 77161 Poly-L-Lysine Sigma Ca # P8920 Cal hymus DNA Sigma Ca # D1501 In e leukin-4, p emium g ade Mil enyi Bio ech Ca # 130-097-761 Cxcl12, ca ie ee Biolegend Ca # 581204 Cxcl13, ca ie ee Biolegend Ca # 583904 Fa y acid ee BSA Sigma Ca # A8806 P o ein A Sigma Ca # P6031 P o ein A Sepha ose The mo Scien i ic Ca # 101041 SiR-ac in Spi och ome Ca # SC001 S ep a idin (Cy5) Jackson ImmunoResea ch Ca # 016-170-084 C i ical Comme cial Assays EasySep B cell isola ion ki S em Cell Technologies Ca #19854 Expe imen al Models: Cell Lines 293 cells ATCC Ca # CRL-1573, RRID:CVCL_0045 (Con inued on nex page) Cell Repo s 32, 108030, Augus 11, 2020 e1 A icle ll OPEN ACCESS RESOURCE AVAILABILITY Lead Con ac Fu he in o ma ion and eques s o esou ces and eagen s should be di ec ed o and will be ul illed by he Lead Con ac Di k Mielenz ([email p o ec ed]) Ma e ials A ailabili y All unique eagen s gene a ed in his s udy a e a ailable om he Lead Con ac . The plasmid pFuse- IgG–Fc2-ICAM1 gene a ed in his s udy has been deposi ed o Addgene. The accession numbe o pFuse- IgG–Fc2-ICAM1 Addgene: 156462. Da a and Code A ailabili y This s udy did no gene a e new esou ce da ase s o code. EXPERIMENTAL MODEL AND SUBJECT DETAILS All expe imen al p ocedu es we e done in ag eemen wi h animal p o ocols app o ed by he Go e nmen o Lowe F anconia, Ba- a ia, Ge many. Bo h emale and male mice we e used in he expe imen s. Mice we e main ained on a 12-h ligh /da k cycle wi h ee access o ood and wa e acco ding o go e nmen al ules. The Blimp1:GFP (B6/P dm1 m1Nu ) and EFhd2KO (E hd2 m1(KOMP)Vlcg ) mice used in his s udy we e desc ibed p e iously (B achs e al., 2014;Kallies e al., 2004). All mice we e in C57BL/6 backg ounds and be ween 8-12 weeks old. Sex ma ched li e ma es o age and sex ma ched animals we e used as con ols. METHOD DETAILS Two-pho on lase -scanning mic oscopy All imaging expe imen s o ex i o spleen slices we e pe o med using a specialized mul i-pho on lase -scanning mic oscope based on a comme cial scan head (T iMScope II, LaVision BioTec, Biele eld, Ge many) p e iously desc ibed (Rakhymzhan e al., 2017). The de ec ion o he luo escence signals was accomplished wi h pho omul iplie ubes in he anges 460 ±30 nm (Hoechs ), 525 ±25 nm (au o luo escence), 593 ±20 nm (CMTPX) and 655 ±20 nm (CD21/35-Fab-Alexa 647). The exci a ion o Hoechs , CMTPX and Alexa Con inued REAGENT o RESOURCE SOURCE IDENTIFIER Expe imen al Models: O ganisms/S ains Ly5.1 (B6.SJL-P p c a Pepc b /BoyJ) mice Jackson Labo a o ies Ca # JAX:002014, RRID:IMSR_JAX:002014 EFhd2KO (E hd2 m1(KOMP)Vlcg ) mice B achs e al., 2014 MGI # 5707812, RRID:MGI:5707812 Blimp1:GFP (B6/P dm1 m1Nu ) mice Kallies e al., 2004 MGI# 3510704 Oligonucleo ides ICAM-1Fc wd g cacgaa cgcagg a cca cca cccagagaag In i ogen; his s udy N/A ICAM-1Fc e a gcaga c g a gagag gg acag In i ogen; his s udy N/A Recombinan DNA pFUSE- IgG-Fc2 In i ogen Ca # p use- c1 pFUSE- IgG-Fc2-ICAM1 Addgene; his s udy Addgene ID 156462 So wa e and Algo i hms G aphPad P ism G aphpad h ps://www.g aphpad.com/scien i ic- so wa e/p ism/ Ima is Ox o d Ins umen s h ps://ima is.oxins .com/ Voloci y Quo um echnologies h ps://quo um echnologies.com/ oloci y R Phea maps Kolde, 2015 h ps://c an. -p ojec .o g/web/packages/ phea map/index.h ml Sho es pa h algo i hm Rakhymzhan e al., 2017 N/A Kaluza Beckman Coul e h ps://www.beckman.de/ low-cy ome y/ so wa e/kaluza O he mslide 8-well polyme chambe ed co e slips Ibidi Bioscience Ca # 80826 Sheep ed blood cells Fiebig Na ¨h s o echnik Ca # 31100100 e2 Cell Repo s 32, 108030, Augus 11, 2020 A icle ll OPEN ACCESS 647 was pe o med a 850 nm o Ti:Sa lase and 1100 nm o OPO. In all expe imen s, we ocused he lase beams wi h an 20x objec- i e lens (Apoch oma wa e -imme sion, NA = 1.0, WD = 2 mm, Zeiss, Jena, Ge many) and maximally used an a e age lase powe o 8 mW o a oid pho odamage. The pulse wid h o he Ti:Sa beam amoun ed o 160 s, ha o he OPO beam o 150 s, unde he objec- i e lens. Analysis o wo-pho on imaging da a All ime-lapse 3D luo escence da a acqui ed in spleen slices by wo-pho on mic oscopy we e loaded ei he in Ima is (Bi Plane) o Voloci y (Quo um echnologies), pos -p ocessed wi h a Gaussian- il e o educe signal noise and con as ed o bes image quali y. The objec s in each spec al channel o he 4D da a we e segmen ed using a s anda dized au oma ic pipeline based on he wa e - shed algo i hm. The pipeline elies on signal disc imina ion using he measu e o local signal- o-backg ound a io ollowed by objec me ging o sepa a ion acco ding o he expec ed a e age cell olume. The segmen a ion esul s we e subsequen ly inspec ed by he esea che s o exclude non-sense segmen a ion. A e his s ep, we de e mined cell numbe s and olume. Fo objec acking o e ime we used he sho es -pa h-algo i hm, which was p e iously demons a ed o eliably pe o m o he used luo escen cell con- cen a ion pe imaged olume uni (Rakhymzhan e al., 2017). Also he acking esul s we e subsequen ly inspec ed o exclude a - i ac s. The esul s o his s ep we e o each cell: 1) ime pe iod he cell could be moni o ed ( ime span), 2) posi ion (x,y,z) o each ime poin . Based on his in o ma ion he 3) ack leng h (dis ance), 4) displacemen leng h (displacemen ), 5) immedia e eloci y o each ime poin , 6) a e age eloci y o he whole ime span ( eloci y), 7) displacemen a e a e aged o he whole ime span (displacemen a e) and 8) di ec edness as he a io be ween displacemen leng h and ack leng h we e de e mined. Cell-cell con ac s we e quan- i ied by measu ing he colocaliza ion o he co esponding spec al channels, chosen o ha e a minimum c oss- alk. Colocaliza ion is de ined as o e lap o wo spec al signals measu ed in each pixel. Disc imina ion be ween signal and backg ound was pe o med elying on local signal- o-backg ound a io. The same segmen a ion pipeline and acking algo i hm as desc ibed o cells we e used o iden i y and ollow cell-cell con ac objec s o e ime. Analysis o gene exp ession da a Hea maps o no malized exp ession alues de i ed om publicly a ailable gene exp ession da a based on RNA Seq (Ise e al., 2018) we e gene a ed using he R hea maps package (Kolde, 2015). Mice Cons i u i ely EFhd2 de icien mice (EFhd2KO) on a C57BL/6 backg ound we e desc ibed p e iously (B achs e al., 2014). Blimp1:GFP epo e mice we e kindly p o ided by S ephen Nu (Kallies e al., 2004) and c ossed o EFhd2KO mice. Ly5.1 (B6.SJL-P p c a Pepc b /BoyJ) mice we e bough om The Jackson Labo a o y. Mice we e used a 8–12 weeks o age, wild- ype (WT) li e ma es o age and sex ma ched animals om he same acili y se ed as con ols. Mice we e kep unde pa hogen ee con- di ions in he IVC o he F anz-Penzold -Cen e (E langen, Ge many), expe imen s we e pe o med acco ding o e hical guidelines o animal expe imen a ion unde con ol o he go e nmen o Lowe F anconia, Ba a ia, Ge many. T-dependen immuniza ions Mice we e injec ed wi h 100mgNP 29 -KLH (Biosea ch Technologies) in Alum (The moScien i ic) in a 1:2 a io (200ml o al olume) o wi h 2x10 9 sheep ed blood cells (SRBCs, Fiebig Na ¨h s o echnik) in 200ml PBS in ape i oneally (i.p.). Isola ion o p ima y mu ine cells om spleen and bone ma ow Spleen was ans e ed in cold R10 medium (RPMI 1640, 10% FCS, 2mM glu ama e, 1mM sodium py u a e, 50 U/ml penicillin G, 50 mg/ml s ep omycin, 50 mMb-me cap oe hanol) and gen ly passed h ough a 70mm cell s aine (BD) using he plunge o a 5ml sy inge (BD). Femu and ibia we e lushed wi h cold R10 medium using a 27G cannula (BD). Cell suspensions we e pelle ed by cen i- uga ion a 300xg o 5min a 4C. E y h ocy es we e lysed upon esuspension in ed blood cell-lysis bu e (150mM NH 4 Cl, 10mM KHCO 3 , 100mM EDTA) o 5min a oom empe a u e. The eac ion was s opped by adding cold R10 medium be o e cen i uga ion a 300xg o 5min a 4C. The inal cell suspensions we e kep in cold R10 medium a e il a ion h ough 30mm mesh il e (Sysmex). De ec ion o su ace an igens by low cy ome y 2x10 6 - 4x10 6 cells we e pelle ed in FACS ubes (Mic onic) a 300xg o 5min a 4C and esuspended in 50ml o unlabeled an i-CD16/ 32 Ab (10 mg/ml in FACS-bu e (PBS, 2%FCS, 0.05% sodium azide)) o 15min on ice. Cells we e washed once wi h FACS-bu e by cen i uga ion a 300xg o 5min a 4C, esuspended in 50ml FACS-bu e con aining he espec i e luo och ome-coupled Abs and incuba ed o 20min on ice in he da k. Cells we e washed wice wi h FACS-bu e by cen i uga ion a 300xg o 5min a 4C. Da a we e acqui ed using a Gallios low cy ome e (Beckman Coul e ). Analyses we e pe o med using Kaluza e sion 1.3 and 2.1 (Beck- man Coul e ). Abs and o he eagen s a e desc ibed in he key esou ces able. Enzyme-linked immunoso ben assay (ELISA) Se um samples om NP-KLH immunized mice (see abo e) we e analyzed in duplica es se ially dilu ed on 96-well la -bo om mic o- i e pla es (G eine bio-one) coa ed wi h 1mg/ml NP 4 -BSA o NP 20 -BSA conjuga es (Biosea ch Technologies) in 50ml/well coa ing Cell Repo s 32, 108030, Augus 11, 2020 e3 A icle ll OPEN ACCESS bu e (15mM Na 2 CO 3 , 35mM NaHCO 3 ) o e nigh a 4C. Fo an i double-s anded DNA ELISA, pla es we e coa ed wi h Poly-L-Lysin (10 mg/ml) in 10mM T is/ 0.1mM EDTA / pH8.0 o e nigh a 4C, washed 3 imes wi h T is/EDTA bu e and coa ed wi h dsDNA (10 mg/ml) om cal hymus (Sigma) in T is/EDTA. Pla es we e washed 3x wi h PBS, 0.05% Tween20 and blocked o 1h a oom em- pe a u e wi h 275ml/well o PBS, 2%FCS. Binding o Abs om p e-dilu ed se um samples was allowed o 1h a oom empe a u e. Cap u ed NP-speci ic Abs we e de ec ed wi h goa an i-mouse IgM and IgG speci ic ho se adish pe oxidase (HRP)- coupled Abs (1:1000, Sou he n Bio ech) and he ELISA was de eloped using o-phenoldiamindihyd ochlo ide (OPD) subs a e (20mM Na 2 HPO 4 , 7mM ci ic acid, 0.001% OPD, 0.035% H 2 O 2 ) and acid s op (0.5M H 2 SO 4 ). Op ical densi y (oD) was measu ed a 490nm on a Spec aMax 190 Mic opla e Reade (Molecula De ices). Pla es we e no malized using ELISA IgM and IgG s anda ds as in e nal e e ence. Pu i ica ion o mu ine B lymphocy es om spleen B cells we e en iched om splenic cell suspensions using he EasySep Mouse B cell isola ion nega i e selec ion ki (EasySep #19854, S emCell Technologies, Inc.) acco ding o he manu ac u e ’s ins uc ions. In sho , spleen cells we e esuspended in MACS bu e (PBS, 2% FCS, 2mM EDTA), su ace blocked wi h a se um and immunomagne ically en iched o un ouched nai e B cells. Pu i y o isola ed B cells was e i ied by su ace s ain o CD19. Usually, an en ichmen o > 95% was achie ed. In i o cul i a ion o p ima y mu ine B cells Splenic B cells we e cul u ed wi h a s a ing concen a ion o 1x10 6 cells/ ml in R10 medium (RPMI1640, 10% e al cal se um (FCS), 2mM glu ama e, 1mM sodium py u a e, 50 U/ml penicillin G, 50 mg/ml s ep omycin, 50 mMb-me cap oe hanol) o 48h a 37C and 5% CO 2, supplemen ed wi h 10mg/ml an i-IgM F(ab’) 2 Ab (A iniPu e F(ab’) 2 agmen goa an i-mouse IgM, mchain speci ic, Jackson ImmunoResea ch), 10mg/ml a an i-CD40 Ab (clone FGK45, (Rolink e al., 1996) and 0.1U/ml IL-4 (mouse IL-4, p emium-g ade, Mil enyi Bio ec). In i o mig a ion assay o ac i a ed B cells Mig a ion assays we e pe o med as desc ibed in Liu e al. (2016). ICAM1-Fc usion p o ein was ob ained by cloning a PCR agmen ( wd p ime : g cacgaa cgcagg a cca cca cccagagaag, e : a gcaga c g a gagag gg acag ) ob ained using mu ine lung cDNA as a empla e. The PCR p oduc was cloned in o pFUSE- IgG-Fc2 (In i ogen) ia BglII/EcoRI es ic ion si es, e- placing he endogenous signal pep ide wi h he ec o encoded IL-2 signal sequence and he memb ane domain wi h a abbi IgG Fc2 pa . ICAM-1Fc was pu i ied om he supe na an o ans ec ed 293 cells by P o ein A Sepha ose (The mo Sien i ic). 8-well polyme chambe ed co e slips (Ibidi) we e coa ed wi h 10mg/ml P o ein A (Sigma Ald ich) and 5mg/ml ecombinan mouse CXCL13 o CXCL12 (ca ie - ee, Biolegend) in 200ml PBS/well o 2h a oom empe a u e. Chambe s we e insed once wi h 10mg/ml a y acid- ee BSA (Sigma Ald ich) in PBS and coa ed wi h 10mg/ml ICAM-1 Fc in PBS o 2h a oom empe a u e. A e addi ional washing chambe s we e blocked wi h 10mg/ml a y acid ee BSA in PBS o e nigh a 4C. The co e slips we e inally washed wi h p e-wa med mig a ion medium (phenol ed- ee RPMI1640, 2mM glu ama e, 1mM so- dium py u a e, 50 U/ml penicillin G, 50 mg/ml s ep omycin, 50 mMb-me cap oe hanol, 10mg/ml a y acid- ee BSA) and kep humidi ied a 37C un il use. B cells we e washed by cen i uga ion a 300xg o 5min a 4C and seeded in p e-wa med phenol ed- ee mig a ion medium a a concen a ion o 5x10 5 cells/ ml in 200ml/well. F-Ac in o li ing cells was labeled using SiR-ac in (1:1000, Spi och ome). Spinning disc con ocal mic oscopy Li e cell mig a ion was analyzed using he Zeiss Spinning Disc Axio Obse e Z1, wi h a LD LCI Plan-Apoch oma 25x/0.8 Imm Ko DIC M27 o Plan-Apoch oma 63x/1.40 Oil M27 objec i e. All expe imen s we e pe o med a 37C and 5%CO 2 . Time se ies we e acqui ed using an E ol e came a and Zeiss Zen so wa e. Fluo escence (638nm lase wa eleng h) and DIC was eco ded. Images we e p ocessed wi h he Fiji so wa e using manual acking mac o o ack cell mig a ion. The eewa e Chemo axis and Mig a ion Tool (Ibidi) was used o u he quan i ica ion o cell eloci y. Adop i e B cell ans e Immuniza ion o dono (WT and EFhd2KO) and ecipien mice wi h SRBC and he isola ion o splenic B cells om he dono s a d7 pos an igen challenge was pe o med as desc ibed abo e. 20x10 6 B cells pe geno ype we e luo escen ly labeled wi h Hoechs and CMTPX (bo h The mo Scien i ic) o ice e sa. WT and EFhd2KO B cells we e mixed in a 1:1 a io and ans e ed i. . in o ecipien mice. 24h pos ans e and 4h be o e imaging, ollicula dend i ic cells (FDCs) o ecipien mice we e in i o labeled ia i. . applica ion o an i-CD21/35-A o590 F(ab’)-Ab o CD21/35-Alexa647 F(ab’)-Ab (coupled in-house a DRFZ, Be lin). Fo mul i-pho on mic oscopy o B cells in splenic ge minal cen e s, spleen was isola ed, ans e ed in o p e-wa med R10 medium and cu in o slices (1-2mm hick- ness). Mig a ion o cells was acked o 30-50min. B-T-conjuga es we e quan i ied as desc ibed p e iously (Lee e al., 2017;Rein- ha d e al., 2009). B ie ly, 15x10 6 splenic B cells om WT and EFhd2KO mice, isola ed as desc ibed abo e, we e ans e ed i. . in a o al olume o 75ml PBS in o congenically ma ked Ly5.1 ecipien mice. Recipien s we e challenged wi h 100mg NP-KLH in alum (desc ibed abo e) 24h la e and analyzed a d14 and d21 pos an igen applica ion. e4 Cell Repo s 32, 108030, Augus 11, 2020 A icle ll OPEN ACCESS Imaging Cy ome y Samples we e p epa ed as desc ibed abo e (De ec ion o su ace an igens by low cy ome y) and subjec ed o analysis wi h an Am- nisFlowSigh (Luminex Co p., Aus in, TX). Fluo escence spillo e was compensa ed using single-s ained cells. Samples we e ocused based on he signal o CD45.1 (Channel06, PE/Cy7). Deb is was ga ed ou om he ocused cells based on a ea and aspec a io o he b igh ield (BF). Fo each sample 3000 e en s in he CD19 + /CD4 + double posi i e ga e we e eco ded. Da a we e analyzed wi h IDEAS 6.2 (Luminex Co p.). In silico GC model In silico ge minal cen e modeling was as p e iously desc ibed (LEDA model) (Meye -He mann, 2014;Meye -He mann e al., 2012; Meye -He mann e al., 2018). The mo ili y p ope ies we e in oduced in Figge e al. (2008) and ine- uned in Binde and Meye -He - mann (2016). The mul iple B-T-in e ac ions ha e been in oduced in de ail in Meye -He mann e al. (2018). QUANTIFICATION AND STATISTICAL ANALYSIS Values we e assessed o Gaussian dis ibu ion using he Shapi o-Wilk no mali y es . Mann-Whi ney es was used o non-Gaussian dis ibu ed da ase s. Da ase s e ealing Gaussian-like dis ibu ion we e assessed by S uden ’s es . Di e ences be ween he analyzed g oups we e conside ed o be s a is ically signi ican wi h p alues < 0.05. Da a we e analyzed using P ism (G aphPad). Cell Repo s 32, 108030, Augus 11, 2020 e5 A icle ll OPEN ACCESS