© 2015 Nature America, Inc. All rights reserved. of Bcl-6 expression. In addition, the transcription factor Batf has has Batf factor transcription the addition, In expression. Bcl-6 of the initiation of the T initiation Received Received 6 May; accepted 15 June; published online 27 July 2015; Yuzhang Wu and IL-21–STAT3 (refs. (refs. IL-21–STAT3 and IL-6–STAT1, including factors, IL-12–STAT4 of family transcription B cell follicles and initiate GC reactions GC initiate and follicles cell B into migrate they Subsequently, cells. B activated engage they where expression but repressed Blimp1 expression. TCF-1-null T TCF-1-null expression. Blimp1 but repressed expression Mechanistically, to TCF-1 bound directly the (GC) B cells in secondary lymphoid tissues lymphoid in B secondary (GC) cells cells abundant chemokine receptor CXCR5 to enable homing to B cell cell B to follicles homing enable to CXCR5 receptor chemokine abundant (T into memory B cells and long-lived plasma cells plasma long-lived and cells B memory into antigen B receptors cells high-affinity bearing for differentiation final GC of selection the and reaction GC a of establishment the facilitate X.Z. X.Z. ( Bcl-6 expression drives the differentiation of fully functional T functional fully of differentiation the drives expression Bcl-6 4 1 ICOS receptor costimulatory the and CD40) receptor costimulatory the for ligand (the CD40L expressing and IL-4 and (IL-21) 21 interleukin cytokines the secreting by help Induction of the transcriptional repressor Bcl-6 in repressor CD4 of Induction the transcriptional cells plasma B of and cells long-lived by memory the production inducing Most antimicrobial vaccines licensed for human use elicit their effects T inhibits Lifan Xu differentiation of T The transcription factor TCF-1 initiates the nature immunology nature University, Yangzhou, China. CD4 activated in regulated is expression Bcl-6 how understand to critical (T transcriptional repressor Bcl-6, identified as a ‘master regulator’ regulator’ ‘master a as T of identified Bcl-6, repressor transcriptional of of T as Thus, TCF-1 functions an hub of axis to upstream important and the initiate the secure Bcl-6–Blimp1 differentiation CXCR5 immunization or infection after 3 or 2 day at CXCR5 with Haoqiang Wang These These authors contributed equally to this work. Institute of Immunology, Third Military Medical University, Chongqing, China. FH FH T FH FH 1 FH 1 [email protected] cells), which for immunity.are cells), B essential cell–mediated by (encoded Blimp1 factor In the transcription contrast, + cells), a unique subset of CD4 subset a unique cells), , , both of which are generally differentiated from germinal center 4 T cells. Several axes of cytokines and members of the STAT the of members and cytokines of axes Several cells. T . Given the central role of Bcl-6 in T in Bcl-6 of role central the Given . differentiation is a multistage, multifactorial process multifactorial multistage, a is differentiation cells acute during viral infection. differentiation + 4– Bcl-6 6 FH , where they engage cognate B cells and provide essential essential provide and cells B cognate engage they where , 1 , 4 differentiation by antagonizing Bcl-6. Here Bcl-6. we by TCF-1 factor for was found that both antagonizing essential the transcription differentiation , , Yi Cao + T 1 , FH 5 FH -like cells accumulate at the T cell–B cell border, cell cell–B T at the accumulate cells -like 1 differentiation and the effector function of differentiated T and of the function differentiated effector differentiation , , Tingting Zhao 9– 1

1 3 , 4 15– 1 Jiangsu Co-innovation Jiangsu Center Co-innovation for Prevention and Control of Important Animal Infectious Diseases and Zoonoses, Yangzhou, China. aDV , emerges on early T early on emerges , , , Zhunyi Xie ) ) or Y.W. ( 1 A 7 ), have been linked to the induction induction the to linked been have ), NCE ONLINE PUBLIC ONLINE NCE + [email protected] helper T cells, are specialized to are specialized T cells, helper 7 , 8 . 1 1 5 3 1 , , Zhonglei Fan These These authors jointly directed this work. should Correspondence be addressed to L.Y. ( , during which reinforced reinforced which during , , 2 4 . Follicular helper T cells helper . Follicular , , Qizhao Huang FH FH differentiation, it is is it differentiation, FH Bcl6 3 -like cells together together cells -like . T . A TION FH promoter and promoter cells during acute viral infection cells express express cells

doi:10.1038/ni.322 ). + 2 1 T T helper of cells is for follicular cells critical the differentiation , 2 3 . Next, Next, . 8 , , Aijian Qin FH . The The . 1 , , Qiang Bai cells upregulated genes associated with non-T genes associated cells upregulated FH 2 Ministry Ministry of Education Key Laboratory for Avian Preventive Medicine, Yangzhou

Prdm1 been reported to bind to the the to bind to reported been with those of CD8 of those with of Bcl-6 expression in activated CD4 activated in expression Bcl-6 of standing, whether and how other factors are involved in the regulation promotes commitment to the T helper type 2 (T 2 type helper T the to commitment promotes promoter and strongly represses represses strongly and promoter The transcription factor Blimp1 (encoded by production interferon- repressing GATA-3while factor transcription cytokine receptor chain cytokine IL-2R the of expression of upregulation to leads which expression, h frain f CD8 of formation the cl development for cell critical T is and pathway signaling Wnt canonical the of effector tion of various T cell responses, its tion of role T in various responses, T cell T helper type 1 (T 1 type helper T Bcl-6 inhibits its own transcription expression ypoyi coimnnii vrs (LCMV) virus choriomeningitis lymphocytic with infection acute during mRNA TCF-1 of expression high have of effects the tory T by inducing expression of the transcription factor Eomes factor of the transcription by expression inducing been determined. been TCF-1 has been shown to favor the formation of memory CD8 immune responses is also emerging. In a a In emerging. also is responses immune The transcription factor TCF-1 (encoded by TCF-1 factor (encoded The transcription The transcriptional signatures of T of signatures transcriptional The 9 2 5 , 3 1 ′ , , Jianqiang Ye , Xia , YangXia regulatory regions, which promoted Bcl-6 which promoted regions, regulatory 2 2 4 3 . Despite these profound effects of regula on TCF-1 the effects profound these . Despite . In addition, TCF-1 reportedly reduces the inflamma the reduces reportedly TCF-1 addition, In . FH cells during acute viral infection. cells acute during viral infection. + H H memory precursor cells precursor memory 1) differentiation by inducing expression of the the of expression inducing by differentiation 1) 17 subset 17 of subset helper T cells by IL-17A dampening 1 19 , , Ran He , 2 + 2 0 memory cells mainly by inducing Eomes Eomes inducing by mainly cells memory 1 . The role of TCF-1 in T cell–mediated cell–mediated T in TCF-1 of role The . 2 . . However, Eomes and both IL-2R , 3 , , Xinyuan Zhou Bcl6 1 locus and activate its expression its activate and locus Bcl6 , , Yaxing Hao 14 . Despite such advances in under + transcription T cells has remained unclear. remained has cells T FH cells overlap considerably considerably overlap cells [email protected] Listeria FH FH Prdm1 1 2 differentiation has not differentiation . Indeed, both subsets subsets both Indeed, . cell lineages. cell lineages. Tcf7 1 1 , H 2 5 1 . TCF-1 regulates regulates TCF-1 . 2) fate but inhibits inhibits but fate 2) , Lilin , YeLilin , TICLES E L C I RT A infection model, model, infection ) ) binds to the ) ) is a downstream 1

4 . Furthermore, Furthermore, .

21 Prdm1

, γ β 2 2 (IFN-

seem to seem + . . TCF-1 1

T cells ,

5 & ), ), Bcl6 ) 1 γ 8 β  - - - ) .

© 2015 Nature America, Inc. All rights reserved. or at least ten cells ( cells ten least at or percent percent cells in each. * expression in T in expression molecules CXCR5 and Bcl-6 and was negatively correlated with with correlated T the negatively of was expression and Bcl-6 and CXCR5 molecules Scale Scale bar, 5 T in TCF-1 for enrichment nuclear substantial observed we infection, cells adopting the T the adopting cells tion. Thus, our findings suggest that TCF-1 promotes T promotes TCF-1 that suggest findings our Thus, tion. increased and Bcl-6 with complex a formed TCF-1 In addition, expression. Blimp1 but repressed expression Bcl-6 vated bound the directly to TCF-1 that determined we study, this In pathway. enhanced in cells committed to the T the to committed cells in enhanced the cells, expression and of nuclear TCF-1 localization was selectively cells (Tim3 T high TCF-1 expression in the T the in expression TCF-1 high SMARTA cells (right). ( (right). SMARTA cells naive in MFI the to normalization by followed (Isotype), antibody control a isotype-matched with obtained results of MFI the of subtraction by calculated cells cells and T 66–77 (gp66) presented by I-A presented (gp66) 66–77 acids amino glycoprotein of epitope LCMV for specific receptor gen naive SMARTA cells (which have transgenic transferred expressionwe of a differentiation, T cell bifurcated anti this during TCF-1 of sion given adoptive transfer of SMARTA cells (CD45.1 SMARTAof cells transfer adoptive given acute infection with LCMV. with ( infection acute T Higher TCF-1 expression in T RESULTS axis. Bcl-6–Blimp1 the of upstream mechanisms regulatory multiple through entiation Figure 1 Figure be unnecessary for T S E L C I RT A  at day 2 and 8 after infection of the host mice with LCMV, and in naive (CD44 LCMV, with naive in mice and host the of infection 8 after 2 and day at differentiate into differentiate T Upon acute viral naive infection, virus-specific CD4 strated strated that during the of bifurcated differentiation T ( infection after ( B granzyme We also observed the higher expression of expression WeTCF-1 in the higher T observed also (Tim3 virus-specific T virus-specific merely represents a nonfunctional marker of marker T the a nonfunctional represents merely positively correlated with expression of the key T key the of expression with correlated positively tenfold tenfold lower in T At day 2 after infection, we observed that TCF-1 expression was nearly of strain the with LCMV.Armstrong infected we which subsequently a

H FH Events (% of max) FH Next we investigated the correlation between TCF-1 expression and 100 1 cells 1 atcells day 8 ( after infection 20 40 60 80 cells, but this was minimal in T in minimal was this but cells, differentiation. At day 2 after infection, TCF-1 expression was was expression TCF-1 infection, after 2 day At differentiation. 0 lo TCF-1 CXCR5 Bcl6

0 Selective enhancement of TCF-1 expression in expression TCF-1 of enhancement Selective Isotype FH Day 2 lo promoter and and promoter µ cells from mice as in 10 CXCR5 m. m. ( + Fig. 1 Fig. H 3 ) from the spleen of wild-type (CD45.2 wild-type of spleen the ) from FH 1 cells (Tim3 1 cells 10 Supplementary Fig. 1 Fig. Supplementary cells relative to that in T in that to relative cells c 4 ) ) Flow cytometry of SMARTA cells in the spleen of mice as in H Naive 10 H b 1 1 cells (Tim3 c H ) per group (error bars ( bars (error group ) per + H 1 1 cells or T 5 FH ). We observed a similar phenomenon at day 8 day at phenomenon similar a Weobserved ). P b ) ) relative to its expression in naive cells ( 1 fate. 1 1 cell–associated molecules CD25, Tim3 and and Tim3 CD25, molecules cell–associated 1 < 0.001 (unpaired two-tailed ) Confocal microscopy (left) of the localization of TCF-1 to the nucleus (stained with the DNA-binding dye DAPI) in sorted T sorted in DAPI) dye DNA-binding the with (stained nucleus the to TCF-1 of localization the of (left) microscopy ) Confocal differentiation a Day 8 Prdm1 T hi ) Flow cytometry of TCF-1 TCF-1 of cytometry ) Flow H CXCR5 1 b FH FH ) into wild-type C57BL/6J recipients, recipients, C57BL/6J ) into wild-type hi a at day 2 with LCMV,after infection and in naive SMARTA of and intensity TCF-1 cells in (control), (right). nuclei FH Fig. Fig. 1 CXCR5 cells 5 cells than in naive CD4 T − FH ′ cell subset serves a function or or function a serves subset cell ) and T ) and H regulatory regions, which acti which regions, regulatory 1 cells ( cells 1 25 FH 27 ). Together these data demon data Together ). these a H , 2 ). Furthermore, at). Furthermore, day 2 after ,

1 cells after after 1 cells TCF-1 expression (fold) 2 fate and was diminished in in diminished was and fate 6 − 0.5 1.0 1.5 2.0 2.5 8 a . . Thus, it is unclear whether ) ) but twofold higher in T FH . . To investigate the expres 0 , b cells cells ), s.e.m.). ), T N Fig. 1 Fig. + ), analyzed analyzed ), Day 2 * * H T 1 + t

* -test). -test). Data are of representative two ( FH ) mice ) mice FH b +

FH

T cells generally ). H -differentiation -differentiation -differentiation -differentiation Bcl6 FH 1 1 cells and T T N cells than in cells c

Day 8 transcrip * FH * + TCF-1 H T 1 Fig. Fig. 1 T cells 10 10 10 * differ 2 3 4 FH CXCR5 37.2 47.4 43.0 10 0 b a FH FH ). ). ------DAPI

3 10 4 tetramer-positive SLAM tetramer-positive influenced the programming of T of programming the influenced ( proliferation ( cells T regulatory follicular of differentiation enhanced to due not probably was mice CD4 tetramer-positive the of 35% approximately whereas mice, from cells T ( cells T and PD-1 ( PD-1 and (data (data not in The shown). the reduction T immunoglobulin G (IgG), relative to that of their control counterparts of number GC B as well as cells, a much lower of titer LCMV-specific Role of TCF-1 in T in TCF-1 of Role in in also much lower in in lower much also tion of T of tion differentia the influence then and loads antigen affect may which observed a similar phenotype for bulk activated Foxp3 for activated bulk phenotype a similar observed out out control mice ( mice control assess the role of cell-autonomous TCF-1 in regulating T in TCF-1 regulating role of the cell-autonomous assess lower expression of Bcl-6, ICOS and CXCR5 in in CXCR5 and ICOS Bcl-6, of expression lower tiation, we generated chimeras by reconstituting irradiated wild-type wild-type by we irradiated chimeras tiation, generated reconstituting lox abundance abundance of T T other for recapitulated was T for essential is TCF-1 of abundance greater a whether To investigate Cre) to generate mice with conditional deletion of of deletion conditional with mice generate to Cre) cell–specific T the from recombinase Cre of the T the in in CD4 tetramer–positive of gp66 number and frequency a similar observed ( T cells in the control mice were T were mice control the in cells T expansion of virus-specific virus-specific of expansion lo 10.7 a 10 Supplementary Fig. 2a Fig. Supplementary Naive CD62L FH at day 2 after infection with LCMV. Numbers in quadrants indicate 5 In Tcf7 Tcf7 P-flanked P-flanked TCF-1 Cd4 differentiation during differentiation acute we viral infection, crossed mice with + 10 10 10 Tcf7 FH T cells in in cells T 2 3 4 −/− −/− hi cell population through a feedback loop. To more precisely precisely To more loop. feedback a through population cell -Cre) ( -Cre) Bcl-6 41.3 43.8 6.93 ) SMARTA cells (N) (left), and summary of TCF-1 expression, expression, TCF-1 of summary and (left), (N) ) SMARTA cells 10 −/− FH Tcf7 mice ( mice might augment the effect of TCF-1 deficiency on on deficiency TCF-1 of effect the augment might mice aDV 2 cells Supplementary Fig. 2c Fig. Supplementary mice, CD8 b 10 Tcf7 fl/fl Tcf7 , Fig. 2 Fig. c Supplementary Fig. 2f Fig. Supplementary 3 A Supplementary Fig. 2b Fig. Supplementary FH ) ) or three ( Supplementary Fig. 2d Supplementary NCE ONLINE PUBLIC ONLINE NCE Cd4 8.03 2 Tcf7 10 alleles ( alleles −/− 1 cells, T . In addition, the lower abundance of GC B cells cells B GC of abundance lower the addition, In . 4 Supplementary Fig. 2e Fig. Supplementary H c Tcf7 FH 1 Ce ie cle ‘ called mice; -Cre mice had largely abrogated TCF-1 expression expression TCF-1 abrogated largely had mice 10 10 10 −/− ). 2 3 4 differentiation + Tcf7 T cells are also deficient in TCF-1 expression, mice and control mice ( mice control and mice CD25 −/− 2.02 13.6 32.4 lo a 10 0 ). At day 8 after infection with LCMV, we LCMV,we with infection after 8 day At ). Tcf7 ) ) independent experiments with four mice ( CXCR5 mice than in control mice ( mice control in than mice −/− Tcf7 FH 3 fl/fl 10 mice had a much lower and frequency cell markers, including ICOS, Bcl-6 Bcl-6 ICOS, including markers, cell 4 −/− ) to mice with transgenic expression expression ) to transgenic with mice 52.0 10 + 5 FH FH CD4 T T ). The number of T of number The ). FH A differentiation, as indicated by indicated as differentiation, cells ( cells 10 10 10 FH TION , 2 3 4 g ), indicative of normal clonal clonal normal of indicative ), ). ). Consistent with their lower cells were present in in present were cells Tim3 ). Indeed, TCF-1 deficiency deficiency TCF-1 Indeed, ). + 4.5 8.11 26.7 FH Tcf7 10 0 T cells. However, very few few However,very cells. T

cell population in population cell Fig. 2 Fig. ) or altered apoptosis or or apoptosis altered or ) nature immunology nature −/− 3 10

Intensity of nuclear TCF-1 Cd4 4 ie hr) CD4 here). mice’ 100 150 200 Tcf7 50.7 10 50 a Tcf7 0 5 ). This phenotype phenotype This ). promoter ( promoter −/− 10 10 10 fl/fl − 2 3 4 Tcf7 10 mice than in in than mice CD4 Fig. 2 Fig. Granzyme B * FH 1.16 16.4 43.0 2 mice with mice FH 10 * n CD4 in cells was was cells 3 differen 10 + * Tcf7 Tcf7 T cells 4 b 10 ). We ).

Cd4 T T Naive H 39.4 5 FH H a 1 1 10 −/− −/− , c 6 + + + ) - - - -

© 2015 Nature America, Inc. All rights reserved. per group (error bars ( bars (error group per paired ( paired and quantification of the MFI of CXCR5, ICOS and Bcl-6 on T on Bcl-6 and ICOS CXCR5, of MFI the of quantification and T tetramer-positive on CXCR5 T cells was inherently associated with aberrant development of development with aberrant CD4 T associated was cells inherently for green fluorescent protein (GFP for fluorescent green positive as (marked cells SMARTA retrovirus-transduced of ability host mice with LCMV. the This system enabled us to infected directly compareand the recipients wild-type into cells the transferred then to down SMARTAknock TCF-1 acutely in expression activated cells, RNA (shRNA) hairpin short and system retroviral a used we cells, T T cells, but CD4 (CD45.1 ( virus-specific T virus-specific virus. At day 8 after infection, the frequency of T ( origin cells of wild-type or or wild-type of cells indicate percent SLAM percent indicate areas outlined to adjacent Numbers (right). results those of summary and (left), infection LCMV 8 after day at (Ctrl) mice control frequency in in frequency cells (marked as (marked cells GFP marrow cells from among the cells outlined above (bottom). ( (bottom). above outlined cells the among T CD4 in the spleen of of spleen the in same mouse. ( mouse. same (CD45.1 2 Figure nature immunology nature T impaired the that sibility frequency of SLAM of frequency Fig. 2h Fig. or not transduced (GFP transduced not or uh oe TF1 xrsin hn i ter GFP their did than expression TCF-1 lower much tion, GFP tion, mouse after infection ( infection after mouse in SMARTA cells expressing in SMARTA expressing cells tetramer-positive T tetramer-positive tetramer-positive CD4 tetramer-positive percent indicate areas outlined to adjacent Numbers origin. day 8 after infection. Numbers adjacent to outlined areas indicate percent SLAM percent indicate areas outlined to adjacent Numbers infection. 8 after day wild-type (WT) and and (WT) wild-type tetramer–positive CD4 tetramer–positive ( infection. viral acute to response in h oiis n hi coa epnin f ermrpstv CD4 tetramer-positive of expansion clonal their in origins the between an we difference appreciable did not observe infection, after also also significantly lower in TCF-1-deficient T and of tetramer-positive CD4 tetramer-positive of and Bcl-6 ( Bcl-6 Fig. 2k Supplementary a a Supplementary Fig. 2j Fig. Supplementary FH ) Flow cytometry of gp66 gp66 of cytometry ) Flow SLAM TCF-1 regulates thymocyte development thymocyte regulates TCF-1 Next we infected infected we Next 10 10 10 10 + cells ( cells 0 2 3 4 5 T cells in the spleen of chimeras generated with a mixture of a mixture with generated chimeras of spleen the in T cells ), assessed at day 8 after infection of the host with LCMV, with host the of infection 8 after day at assessed ), CXCR5 Fig. 2 Fig. e 10 0

Fig. 2d Fig. , + TCF-1 is necessary for for necessary is TCF-1 + g ) ) recipient mice with a mixture of congenitally marked bone + ) donor mice (60%) ( (60%) mice donor ) ) two-tailed ) two-tailed Fig. 2 Fig. SMARTA cells expressing SMARTA expressing cells d 3 Ctrl 34.9 g 10 (top left), and MFI of CXCR5, ICOS and Bcl-6 on T on Bcl-6 and ICOS CXCR5, of MFI and left), (top f ). FH ) Flow cytometry of SMARTA cells transduced (GFP transduced SMARTAof cells cytometry ) Flow 4 Tcf7 , 10 + e differentiation differentiation e T cell populations of ). The expression of CXCR5, ICOS and Bcl-6 was was Bcl-6 and ICOS CXCR5, of expression The ). 5 FH ). Tcf7 Tcf7 Tcf7 −/− cells as in in as cells lo Tcf7 a + + t − T cells (top) and SLAM and (top) T cells lo – mice and and mice − T cells T cells CXCR5 -test). Data are representative of three ( three of representative are Data -test). −/− ) to differentiate ) into to T differentiate c ), then transferred into recipients subsequently infected with LCMV ( LCMV with infected subsequently recipients into transferred then ), CXCR5 Tcf7 −/− −/− Supplementary Fig. 2i Fig. Supplementary ), s.e.m.). ), ), ), with lower expression of CXCR5, ICOS and

). Notably,). T −/− bone marrow cells ( cells marrow bone origin; lines connect data points for the the for points data connect lines origin; 3.48

(CD45.2 –/– aDV + bone marrow chimeras with influenza A influenza with chimeras marrow bone FH T cells of wild-type or or wild-type of T cells FH + Tcf7 cells as in in as cells + T a

A T

. . ( differentiation of TCF-1-null CD4 TCF-1-null of differentiation NCE ONLINE PUBLIC ONLINE NCE FH Supplementary Fig. 2h Fig. Supplementary FH c

-specific shRNA cells ( shRNA cells -specific cells. ( cells. c ) Expression of Bcl-6, ICOS and and ICOS Bcl-6, of ) Expression + + 2 cells than did those of wild-type wild-type of those did than cells lo + ) ) donor mice (40%) and wild-type )) and non-transduced SMARTA )) and non-transduced Bcl-6 MFI (×10 ) SLAM CXCR5 10 15 FH 0 5 Tcf7 Tcf7 cells (%) 10 20 30 40 e differentiation was impaired impaired was differentiation a 0 ) Summary of T of ) Summary b . . ( *** ) Quantification of total total of ) Quantification −/− -specific shRNA displayed shRNA displayed -specific d *** lo Supplementary Supplementary Tcf7 Ctrl Tcf7 Ctrl 20 ) Flow cytometry of cytometry ) Flow FH CXCR5 origin exhibited a lower FH , 2 FH cells than in wild-type ). At day 8 after infec after 8 At day ). –/– –/– 9 2 cells within the same within cells . . To out pos the rule ICOS MFI (×10 ) Tcf7 cells was diminished 0 2 4 6 + T A *** −/− TION b FH − 5 FH

counterparts counterparts T cells (×10 )

FH cells cells

10 Fig. 2f Fig. cell cell 0 2 4 6 8 FH ). At day 8 day At ).

a

cells (presented as in in as (presented cells 3 – CXCR5 MFI (×10 ) c

FH Tcf7 Ctrl *

4 5 6 7 + ,

f

) with retrovirus expressing empty vector (EV) or or (EV) vector empty expressing retrovirus ) with

, , shRNA TCF-1-

g g

–/–

) or two ( two ) or and * EV f + + + - -

d SLAM SLAM gp66 10 10 10 10 10 10 10 10 10 10 10 lar frequency lar and frequency number SLAM of virus-specific control, we administered tamoxifen to i to tamoxifen administered we control, T cells T ( cells to infection after 7 day to induce deletion of 4 day from administration by tamoxifen followed LCMV, with mice these infected next We here). mice’ variant fused to Cre (ERT2-Cre) with with (ERT2-Cre) Cre receptor to estrogen fused variant tamoxifen-sensitive a expressing mice crossed (GFP Tamoxifen treatment led to the efficient deletion of TCF-1 in CD4 in TCF-1 of deletion efficient the to led treatment Tamoxifen indicated that the importance of TCF-1 for T for TCF-1 of importance the that indicated among TCF-1-null CD4 TCF-1 expression before and infection a observed reduction of nearly showed defective T LCMV. GFP LCMV. and assessed and the early assessed T infection. viral acute another by shared also was but infection LCMV to unique not in tamoxifen-treated and vehicle-treated i and vehicle-treated in tamoxifen-treated Tcf7 Temporal dependence of T of dependence Temporal idtp CD4 wild-type T TCF-1-deficient in lower was CXCR5 and ICOS of Bcl-6, expression of T of To TCF-1 is for whether investigate important the early commitment for priming early T early priming for d 0 0 2 3 4 5 0 3 4 5 2 3 4 5 FH , o eemn te oe f C- i lt T late in TCF-1 of role the determine To e CXCR5 lo CXCR5 CD44 ) independent experiments with at least three ( three least at with experiments ) independent 0 0 cells than in cells ( wild-type CXCR5 fl/fl FH 0 10 − GFP ) ) control cells ( cells, we knocked down TCF-1 expression in SMARTA cells SMARTAcells in expression TCF-1 down knocked we cells, ERT2-Cre mice for inducible deletion of 2 40.7 10 10 WT 10 Fig. Fig. 3 e 19.2 3 3 43.6 ). * ). 42.3 3 10 10 – 10 + 4 4 4 T P 10 10 10 + FH < 0.05, ** < 0.05, b 5 5 5 Supplementary Fig. 2i Fig. Supplementary SMARTA cells in which TCF-1 was knocked down down knocked was TCF-1 which in cells SMARTA ). ). Notably, at a day we simi 8 observed after infection, cells. ( cells. + cls ( cells T Tcf7 FH GFP Tcf FH 15.8 Fig. 3 differentiation relative to that of non-transduced 18.0 7 differentiation. 19.4 41.6 g + fl/fl –/– ) Summary of T of ) Summary + P FH T cells compared with their frequency among < 0.01 and *** and < 0.01 during the late phase of T differentiation at differentiation day 2 with after infection a upeetr Fg 3a Fig. Supplementary ). ). This suggested that TCF-1 was essential FH ICOS SLAMloCXCR5+ g

differentiation on TCF-1 on differentiation 2 e Supplementary Fig. 3b Supplementary MFI ( 10 ) lo + × cells (%) SLAM CXCR5 10 20 30 40 50 60 10

), assessed in the host spleen at spleen host the in assessed ), 2 4 6 8 Tcf7 ICOS MFI (×10 ) cells (%) 10 20 30 40 50 FH 0 2 4 6 8 0 VTCF-1- EV EV -specific shRNA (TCF-1-shRNA) (TCF-1-shRNA) shRNA -specific cell frequency in in frequency cell GFP GFP P Tcf7 < 0.001 (unpaired ( (unpaired < 0.001 Tcf7 shRNA TCF-1- Tcf7 shRNA – + *** Tcf7 WT ** −/− ** ** −/− FH Tcf7 FH fl/fl –/– a mice for 4 d to ablate ablate to d 4 for mice – differentiation, we we differentiation, differentiation was was differentiation mice ( mice c lo ie o generate to mice ) or four ( four ) or fl/fl TICLES E L C I RT A CXCR5 FH Bcl-6 CXCR5 MFI ). Moreover, the the Moreover, ). MFI (×102) (×102) (called ‘i differentiation. 40 45 50 55 60 Bcl-6 MFI 2 1 2 4 3 CXCR5 MFI (×10 ) 100 Fig. 3 Fig. f ). ). These data 50 60 70 80 90 10 11 (top left), left), (top 6 7 8 9 + VTCF-1- EV EV d T – a g – c FH Tcf7 ) mice ) mice c shRNA TCF-1- ). ). As a *** shRNA ) ) or * cells ** *

−/−  + -

© 2015 Nature America, Inc. All rights reserved. experiments with at least seven ( seven least at with experiments B cells as in in as B cells least six ( six least t given no cells (−) or i or (−) cells no given i ( quantification staining with peanut agglutinin (PNA), in mice without cell transfer, cell without (PNA), in mice agglutinin peanut with staining by terized high expression of the receptor cell surface Fas and (CD95) PNA virus-specific T virus-specific of differentiation late the for dispensable was and priming early for mainly responsible was TCF-1 that notion the supported data these hypothesis, we sorted fully differentiated T differentiated fully sorted we hypothesis, effector function of T of function effector transferred these T these transferred Tim3 percent indicate (left) areas outlined to adjacent Numbers left. at results the of summary right, (left); infection 2 after day at spleen host the in ( LCMV with infected subsequently transduced (GFP transduced spleen of i of spleen cells cells ( ( cells treatment efficiently diminished the TCF-1 expression in Tamoxifen i vehicle. or tamoxifen of administration the by followed (CD45.1 mice T of (right) in as mice of spleen the in adoptive transfer of T of transfer adoptive 1 d given later, and, LCMV with CD4 Figure 4 Figure T of abundance the in ninefold T 3 Figure S E L C I RT A  and frequency (middle) and number (right) of T of (right) number and (middle) frequency and (left), infection 8 after day at LCMV, with analyzed infection 4 d before adjacent to outlined areas (left) indicate percent SLAM percent indicate (left) areas outlined to adjacent relative to their abundance in vehicle-treated mice ( after infection. ( infection. after 8 day at analyzed 7, day 4 to day from (Veh) (Tam) vehicle or tamoxifen recipient mice (CD45.1 mice recipient NS, not significant; * significant; not NS, of the host with tamoxifen or vehicle, and analysis at day 5 after cell transfer. ( transfer. cell 5 after day at analysis and vehicle, or tamoxifen with host the of T of differentia function late effector the during modulate might tion expression TCF-1 that hypothesized We from i from (CD45.2 ( specific shRNA, or not transduced (GFP transduced not or shRNA, specific percent SLAM percent Role of TCF-1 in the effector function of T of function effector the in TCF-1 of Role wild-type mice (CD45.2 mice wild-type Tcf7 -test). Data are representative of two independent experiments with at with experiments independent two of representative are Data -test). a FH

a ,

b 3

TCF-1 MFI (×10 ) ( infection. viral acute in cells ) TCF-1 expression ( expression ) TCF-1 + hi 10 15 lo −/− cells in the spleen of i of spleen the in cells 0 5 FAS Tcf7 Fig. 4 CXCR5 Fig. 4 Fig. T

+ TCF-1 is critical for the the for critical is TCF-1 of differentiation late not but early for essential is TCF-1 FH a ) at day 7 after infection with LCMV (equal number of each genotype), followed by treatment treatment by followed genotype), each of number (equal LCMV with infection 7 after day ) at hi −/− Tcf7 ) or three ( three ) or FH * GC B cells gated on the B220 the on gated B cells GC cells in the spleen of spleen the in cells b and wild-type donor mice mice donor wild-type and c cells. Numbers adjacent to outlined areas (left) indicate indicate (left) areas outlined to adjacent Numbers cells. a + ). ). At day 5 after transfer, we found few GC B cells, charac ( lo ) but minimally altered the number of transferred T transferred of number the altered minimally but ) T Veh Tam −/− + b d CXCR5 c FH ) at day 1 after infection of the recipients with LCMV,with recipients of the ) at infection day 1 after ) of transferred transferred ) of ) and plasma cells as in in as cells plasma ) and + FH mice infected with LCMV, followed by administration of administration by LCMV, with followed infected mice ) Flow cytometry of gp66 tetramer–positive CD4 tetramer–positive gp66 of cytometry ) Flow ) with retrovirus expressing empty vector or or vector empty expressing retrovirus ) with cells. ( cells. cells. b FH Tcf7 FH b FH P – T cells + FH ( (unpaired < 0.001 cells. cells. cells sorted sorted cells T cells (the same number of each) into recipient recipient into each) of number same (the cells + d 10 10 10 10 a ) infected ) infected ) mice per group (error bars ( bars (error group per ) mice ) and ) and −/− FH 0 2 4 6 b + b ) Expression of TCF-1 in CD4 in TCF-1 of ) Expression cells. ( cells. and wild-type T wild-type and Tcf7 ) at day 7 after infection with LCMV and and LCMV with infection after 7 day at ) (left), and frequency (middle) and number number and (middle) frequency and (left), NS

−/− a

, a b Supplementary Fig. 2i Fig. Supplementary

mice treated with tamoxifen or vehicle vehicle or tamoxifen with treated mice d ) Flow cytometry of SMARTAof cells cytometry ) Flow FH Veh Tam ) or three ( three ) or

) Flow cytometry of tetramer-positive tetramer-positive of cytometry ) Flow cells in tamoxifen-treated mice mice tamoxifen-treated in cells

− ), then transferred into recipients recipients into transferred then ), e + ( e c CD19 b f CD138 FAS FH ). * ). – 10 10 10 10 10 10 10 10 0 0 d cells as in in as cells c 3 4 5 2 2 3 4 5 ) or paired ( paired ) or FH – P FH PNA B220 + f < 0.001 (unpaired two-tailed two-tailed (unpaired < 0.001 0 0 ) mice per group (error bars ( bars (error group per ) mice cells. Numbers Numbers cells. population ( population 3.55 FH cells from i from cells 10 10 Veh Veh FH 3 3 cells 10 b 10 – lo 5.10 Fig. Fig. 3 cells. To test our our Totest cells. 4 4 10 d 10 + CXCR5 iTcf7 ), assessed assessed ), iTcf7 T cells from the the from T cells ), s.e.m.). ), a 5 5 , a b ) two-tailed ) two-tailed , analyzed at day 5 after cell transfer. Numbers adjacent to outlined areas indicate percent percent indicate areas outlined to adjacent Numbers transfer. cell 5 after day at , analyzed –/– –/– d Tcf7 1.05 Tcf7 c ). ). Together + ), or CD138 or ), Tcf7 Tam T Tam + -

T cells T cells FH −/− 2.66 −/− cells. cells. T or or FH FH

- -

3.36 hi B220 Veh Veh whereas those that received exogenous T exogenous received that those whereas tive transfer of donor T Enhanced T Enhanced differentiation. of on of T early priming the its effect was independent which cells, than cells plasma that of control mice (that i received and number of frequency PNA lower significantly a exhibited tamoxifen with treated were and cells would be sufficient to improve expression TCF-1 T enhanced whether investigate to sought we Next TCF-1 modulated the effector function of already differentiated T differentiated of already function effector the modulated TCF-1 and were treated with vehicle) ( vehicle) with treated were and GC GC B cell ( population transferred T transferred of function effector the evaluate to us enabled which system, this in a c shRNA TCF-1- , t , b -test). NS, not significant. Data are representative of two independent independent two of representative are Data significant. not NS, -test). 6.08 e , ) Flow cytometry of cells in the spleen of recipient mice as in in as mice recipient of spleen the in cells of cytometry ) Flow d d c lo a EV WT , WT plasma cells ( cells plasma f ), s.e.m.). ), SLAM SLAM Tim3 3.16 10 10 10 10 10 10 10 10 10 10 10 aDV 0 0 0 2 3 4 5 2 3 4 5 Tam 3 4 5 Tam CXCR5 CXCR5 CXCR5 FH 0 FH 0 0 5.63 A GFP differentiation via TCF-1 overexpression TCF-1 via differentiation 10 cells cells Veh 10 e Tam Veh 10 NCE ONLINE PUBLIC ONLINE NCE 3 42.9 41.2 3 3 10 10 4713.4 34.7 24.6 29.4 10 e – 4 ). ( ). 4 4 10 10 0.57 10 in in vivo 5 Fig. Fig. 4 5 5 d

FH , – – f ) Frequency (left) and number (right) of GC of (right) number and (left) ) Frequency cells accelerated the formation of GC B cells 0.90 GFP Tam . Of note, mice that received i . that received Of note, mice c 31.9 4.97 ). These ). results These that suggested the adop + hi FH iTcf7 Fig. 4c Fig. iTcf7 FAS d f differentiation. We used a retroviral Plasma cells (%) + – Tam GC B cells (%) a + – Tam T–– – WT T–– – WT –/– 0 1 2 3 4 0 2 4 6 8 –/– A hi lo + lo + Tim3loCXCR5+

TION SLAM CXCR5 SLAM CXCR5 GC B and cells CD138

– cells (%) cells (%) cells (%) + + f * 10 20 30 40 50 * 20 40 60 10 20 30 40 50 ). These data indicated that that indicated data These ). + + FH 0 0 0

–+ – – NS –+ – – NS cells exhibited a distinct distinct a exhibited cells ++ GFP GFP ++ nature immunology nature NS VTCF-1- EV * – – – – – – + – shRNA

iTcf7 TFH cells TFH cells * iTcf7 10 10 10 10 10 10 10 6 6 Tcf7 Tam a + – Tam Plasma cells ( 10 ) WT GC B cells ( 10 ) × × 4 5 6 0 2 4 6 T–– – WT –/– 0 1 2 3 4 0 1 2 3 –/– −/− b Tcf7 NS

* 3 + + – – – TCF-1 MFI ( 10 ) + × T * * 0 2 4 6 8 a hi + + , −/− FH b B220 –+ – – –+ – – NS NS Tam Veh Tam Veh Tam Veh ++ ++ cells T * – – – – – – FH FH FH lo -

© 2015 Nature America, Inc. All rights reserved. from three mice per group ( group per mice three from replicates two with experiments independent mice per group ( group per mice three from pooled replicates biological two with expression in wild-type T wild-type in expression two-tailed two-tailed of selected genes in in genes selected of (bottom) in T in (bottom) downregulated or (top) upregulated signatures with wild-type and and wild-type with TCF-1 isoform TCF-1 sion of the p33 (truncated) or p45 (full-length) T expression in expression and and in genes Gata3, Il4 Gata3, cells cells relative to their expression in wild-type T ** more ‘concentrated’ in the wild-type or Tcf7 or wild-type the in ‘concentrated’ more signatures gene downregulated or upregulated as (defined genes’ ‘enriched indicate rectangles LCMV; red with infection 8 after day at TCF-1-deficient samples, and expression of CXCR5 mRNA was was mRNA CXCR5 T in of wild-type upregulated considerably expression and LCMV samples, with TCF-1-deficient infection after 8 day ( at mice TCF-1-deficient and in non-T in differentiation of SLAM of differentiation the increased substantially isoform either of expression ectopic that observed LCMV, we mice. At day 8 after infection of the hosts with recipient wild-type into cells the transferred T ( T regulates TCF-1 which by mechanisms molecular the elucidate To rndcin ytm o civ overexpres achieve to system transduction T TCF-1-deficient 5 Figure nature immunology nature ( regulated and 513 genes that were upregulated in TCF-1-deficient T codes accession infection ( infection each after 1 day at mice of recipient into population number equal an infection transferred after and 8 day p33 on mice overexpress donor from to transduced or vector Transcriptional profiles of TCF-1-deficient T TCF-1-deficient of profiles Transcriptional SMARTA T SMARTA cells B helping for tial poten greater held TCF-1 overexpressing accession codes codes accession 1 Table ferentiation of functional T dif functional of the ferentiation increase to sufficient was TCF-1 of expression heightened and plasma cells in mice that received p33-overexpressing T p33-overexpressing received that mice in cells plasma and B GC of number and frequency greater a observed we transfer, after downregulated in T in downregulated Supplementary Supplementary Fig. 4a Supplementary Fig. 4d Fig. Supplementary Supplementary Fig. 5a Fig. Supplementary FH FH H P We further investigated whether T whether investigated We further By microarray analysis, we identified 569 genes that were down were that genes 569 identified we analysis, microarray By 1 cell samples ( samples cell 1 population). ( population). < 0.01 and *** and < 0.01 differentiation, we sorted T we sorted differentiation, Tcf7 1 2 , for gene-set–enrichment analysis (GSEA) with our data. data. our with (GSEA) analysis gene-set–enrichment for ,

FH ). Then, we selected genes from published data sets (GEO (GEO sets data published from genes selected we Then, ). −/− a Distinct transcriptional profiles of profiles transcriptional Distinct and 4 genes of interest ( interest of 4 genes and and and cells, from published data (GEO (GEO data published from cells, t T -test). Data are from one experiment experiment one from are Data -test). Supplementary Fig. 4c Fig. Supplementary FH FH FH Tcf7 Icos cells relative to their expression expression their to relative cells cells. ( cells. cells transduced with empty empty with transduced cells 3 a GSE2137 0 , b FH GSE2137 b in SMARTA cells. We SMARTAthen cells. in ) Heat map of 51 selected selected 51 of map ) Heat Tcf7 (red)) expressed in wild-type wild-type in expressed (red)) ) or are representative of two two of representative are ) or −/− P Supplementary Fig. 5b Fig. Supplementary cells. ( cells. b < 0.001 (unpaired (unpaired < 0.001 , normalized to their their to , normalized FH T c −/− ) Quantitative RT-PCR ) Quantitative FH cells relative to their expression in non-T in expression their to relative cells FH

T , lo 9 b cells. * cells. ). Together these results demonstrated that that demonstrated results these Together ). n vivo in a c cell samples, as well as wild-type or as as wild-type well samples, cell ). TCF-1 mRNA was barely detectable in in detectable barely was mRNA TCF-1 ). FH 9 aDV and and ). CXCR5 ) GSEA of gene gene of ) GSEA ; error bars, s.e.m.). bars, ; error and and FH cells sorted sorted cells A cells. GSE2138 NCE ONLINE PUBLIC ONLINE NCE FH GSE2138 P . We sorted sorted We . Rorc < 0.05, < 0.05, + cells and cells T T ). At 5 d 5 At ). , FH FH

cells cells 1 cells cells FH −/− ), ), 1 ). ) that are upregulated and and upregulated are that )

cell samples relative to its relative samples cell - - FH H 1 cells from wild-type 1 from wild-type cells cells ( FH c a

cells Il21 expression Bcl6 expression Enrichment score Enrichment score A –0.8 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1 TION 0.1 0.2 0.3 0.4 0.5 0.6 0.8 0.9 0.7 Supplementary (fold) (fold) 0.5 1.0 1.5 0.5 1.0 1.5 0 0 0 0 Tcf7 Tcf7

T T *** *** FH FH –/– –/– *** *** WT FH Tcf7 T T cells cells H H 1 1 Tcf7 −/− FH FH - -

Il4 expression Icos expression –/– Downregulated in T Downregulated in

(fold) (fold) T Upregulated in 0.5 1.0 1.5 0.5 1.0 1.5 T those results by quantitative RT-PCR, we noted that expression of of T the expression that noted we RT-PCR, quantitative by results those and and from the GSEA results and another 4 genes of interest ( T ‘concentration’ in wild-type represented’ that of wild-type T represented’ that of wild-type in T in T-bet) and genes cell–relevant cells cells compared with that T of wild-type expression was higher in TCF-1-deficient T in TCF-1-deficient was higher expression T not, while the gene set linked to the T the to linked set gene the while not, ( the gene set associated with the T the with associated set gene the h gn-xrsin atr o TF1dfcet T TCF-1-deficient of pattern gene-expression the T This This analysis revealed that T wild-type Il4 are essential for priming T priming for essential are receptors IL-6R the signal-transducing Fig. 5 Fig. 0 0 FH FH FH was T lower in TCF-1-deficient cells than in wild-type T wild-type in than cells cells ( cells cells than in wild-type T wild-type in than cells T T Icos *** FH * FH FH FH *** ** c cells) was ‘concentrated’ in in ‘concentrated’ was cells) cell–associated genes genes cell–associated ). Similarly, the expression of of expression the Similarly, ). ) and noted a distinct gene-expression pattern in pattern ) gene-expression and a noted distinct T T H H Gzmb Fig. 5 Fig. 1 1 FH FH

Il6ra expression Prdm1 expression (which encodes granzyme B) was higher in TCF-1-null c

). We also observed that the expression of of expression the that observed We ). also (fold) (fold) 0.5 1.0 1.5 0 0 2 4 6 8 Tbx21 T *** T *** FH FH WT WT *** *** (which encodes the transcription factor factor transcription the encodes (which T T FH H H FH 1 1 FH FH Bcl6 differentiation b FH cells, whereas expression of the T the of expression whereas cells, cells ( cells cells ( cells

FH Il6st expression Tbx21 expression cells. We cells. 51 genes assessed further FH and and Tcf7 lineage, but lineage, (fold) (fold) 0.5 1.0 1.5 cells than in wild-type T than cells in wild-type Tcf7 FH 0 0 1 2 3 α Fig. 5 Fig. Il6ra FH –/– Icos Fig. 5 Fig. and gp130, respectively, that cells showed enrichment for H T T *** *** −/− cells ( FH FH 1 lineage (downregulated (downregulated lineage 1 *** *** was lower in TCF-1-null TCF-1-null in lower was FH and and T a T T ), which suggested that suggested ), which c cells than in wild-type than in cells wild-type 15 FH H H ). In addition, addition, In ). 1 1 Fig. Fig. 5 , Tcf7 1 Il6st cells relative to its its to relative cells 6 , was much lower lower much was , TICLES E L C I RT A Ascl2 expression Gzmb expression Expression (fold)

FH –0.28 −/− WT , which encode encode which , b Rorc , (fold) (fold) –0.19 0.5 1.0 1.5 ). Confirming ). Confirming el ‘under- cells

10 20 30 40 –0.09 T 0 0 0 Tcf7 FH 0.19 T T *** Gata3 , ** 0.09 FH FH cells did cells *** Il21 *** 0.28 −/− FH Prdm1 Il6st Slamf6 Cd200 Tox Cd22 Sostdc1 Icos Plagl1 Ccr7 Il6ra Tox2 Farsb Nsg2 P2rx7 H2–Ob Cxcr5 Id3 Tcf7 Pou2af1 Bcl6 Il4 Cebpa Il21 Gata3 Il12rb2 Bhlhe40 Rorc Foxp3 Slamf7 Tbx21 E2f8 Ccl4 Gzmb Ifngr1 Ccr5 Ccr2 Lrrk1 Ccna2 Ggt1 Id2 Cxcr6 Prdm1 Csf1 Ccl3 Slamf1 Prf1 Klrg1 Gzma Fasl Cd86 Aff3 Ascl2 Acsl3 Aurkb Arsb T T cells H T and and H 1 1 H Il4 FH  1

© 2015 Nature America, Inc. All rights reserved. (Mut; black font) font) black (Mut; mutated or (WT; font) red a wild-type Thy-1.1 expression on GFP on expression Thy-1.1 (P- cassette a PGK-EGFP and Thy-1.1, encoding sequence (SIN), repeats terminal long the in mutations self-inactivating Prdm1 or tion −900) (negative control) in SLAM in control) (negative −900) Prdm1 ( position number. ( number. position indicate above numbers orientation; and site start transcription indicate arrows EGFP); and 1) kinase phosphoglycerate encoding gene the 5 Bcl6 encodes Foxp3, for regulatory T cells) T regulatory for Foxp3, encodes mice (CD45.2 Ascl2 relative to their expression in wild-type T wild-type in expression their to relative cell–associated signatures were upregulated ( cells), ( PCR. quantitative by followed (IgG), antibody control isotype-matched or (anti-TCF-1) TCF-1 to antibody with ChIP by LCMV,with analyzed infection 8 after day at mice wild-type from deep sequencing (ChIP-Seq) ofthymocytes (ChIP-Seq) sequencing deep Using published data for chromatin immunoprecipitation followed by We next investigated whether TCF-1 directly regulates of regulation Direct lineages. other of programs differentiation T cell population (left) and MFI of Thy-1.1, normalized to GFP expression (right), in mice as in Figure Figure 6 T TCF-1-deficient in S E L C I RT A  aa niae ta TF1 ntae te T the initiated TCF-1 that indicated data (CD45.1 TCF-1-deficient T T in TCF-1-deficient T in TCF-1-deficient of the host with LCMV. Numbers adjacent to outlined areas indicate percent Thy-1.1 region as in Bcl6 constructs containing a wild-type or mutated (Lef1 wild-type or mutated ‘strengthening’ T ‘strengthening’ ( of Moreover, TCF-1-null T Moreover,TCF-1-null Using ChIP followed by quantitative PCR, we observed enrichment enrichment observed we PCR, quantitative by followed ChIP Using 5 the in three in consensus TCF-1-binding sequences the Genome Browser website (data not shown). We screened two conservedBcl6 locus and that at showed and enrichment the two peaks peak a single identified confirmed that genes associated with T with associated genes that confirmed 5c Fig. t-test). Data are representative of three ( expression of genes encoding other factors not associated with the T the with associated not factors other encoding genes of expression a Supplementary Fig. 6 Fig. Supplementary b FH FH ′

FH ) ) Binding of TCF-1 to conserved motifs in the regulatory region of of region regulatory , Bcl6 Prdm1 c cells in the cells spleen of wild-type recipient SLAM and cells ) Retroviral reporter constructs containing containing constructs reporter ) Retroviral cell lineage, including including lineage, cell 3 promoter as in promoter region ( 5′-Reg) (positive control) or reporter 1 , which encodes a transcription factor critical for for critical factor transcription a encodes which , , than did wild-type T wild-type did than , promoter (P- promoter without TCF-1-binding motifs ( motifs TCF-1-binding without Rorc and ). Quantitative RT-PCR analysis of an additional set of genes genes of set additional an of analysis RT-PCR Quantitative ).

+ TCF-1 TCF-1 regulates the transcription 5 ) transduced with a reporter containing ′ c (f), assessed at day 8 after infection (which encodes ROR encodes (which regulatory region (5 region regulatory Prdm1 + ) given transfer of SMARTA cells ′ regulatory region of of region regulatory Prdm1 locus, respectively, through the use of the UCSC Bcl6 d via direct binding. FH FH , hi b (d) or Prdm1 f Lef1 ) Flow cytometry analyzing analyzing cytometry ) Flow CXCR5 cell–associated signatures while suppressing the the suppressing while signatures cell–associated Bcl6 cells than in wild-type T promoter (P- promoter ) and to a region in a region to ) and Prdm1 FH FH Bcl6 Pgk1 5′ regulatory region cells ( cells cells than in wild-type T wild-type in than cells ) ) ( −500) and FH + Prdm1 SLAM − c T ( cells exhibited much lower expression of expression lower much exhibited cells and and Gata3 ), as well as well as ), (promoter of (promoter FH Prdm1 H ′ 1 cells sorted sorted 1 cells Supplementary Fig. 5d Fig. Supplementary -Reg) and and -Reg) cells ( cells lo lo 5′ regulatory CXCR5 CXCR5 γ Bcl6 Prdm1 t, t, for T (which encodes GATA-3,T for encodes (which −24500) −24500)

Prdm1

a) or two ( Prdm1 Fig. 5 Fig. ) ) ( 3 +

+ b 2

FH

) by TCF-1 by , was significantly higher in in higher significantly was , H

FH 17 cells) and 17 cells) cells were downregulated downregulated were cells

c ( Bcl6 3 ). Of particular note, the the note, particular Of ). Supplementary Supplementary Fig. 5e 3 cells. Collectively, these these Collectively, cells. FH Supplementary Fig. 6 Fig. Supplementary FH d–g) experiments with four replicates ( and CD8 and cells ( cell fate by directly directly by fate cell promoter region and Mut WT f d b a

CD4 CD4 Binding (%) FH 10 10 10 10 10 10 0.4 0.8 1.2 0 0 Supplementary Bcl6 3 4 5 3 4 5 cells ( cells 0 ), whereas T ), whereas SIN SIN Thy-1.1 Thy-1.1 + Foxp3 0 10 0 Cxcr5 T cells T and 10 –650 ** WT WT –500 Bcl6 3.74 6.53 ** P-Bcl6 P-Bcl6 –400, AGACA –400, CAAAG 3 3 Fig. 5 Fig. –335, TGTCT –335, CTTTG 10 10 Lef1 5′-Reg NS (which (which induc Prdm1 4 4 10 10 Exon 1 2 0 5 5 +318 we , Thy-1.1 Thy-1.1 H H c –24500 Prdm1 ** ). ). ). ). ), ), ** 1 2 - .

Mut1 NS Mut 14.0 26.5 P-Pgk1 P-Pgk1 transduced with the vector encoding the mutated mutated the encoding vector the with transduced counterparts their of that than Thy-1.1 of (MFI) intensity rescence exhibited higher frequency of Thy-1.1 frequency higher exhibited wild-type the encoding vector the with transduced retroviral system retroviral ( LCMV atwith from mice day 8 infection after sion in GFP expres Thy-1.1 much sites higher exhibited mutated TCF-1-binding encoding vector the that observed LCMV.with infected With this sequently - sub we which mice, recipient wild-type into cells those transferred duced SMARTA and vector(s) retroviral cells with the corresponding 5 this system. Notably,of system. that GFP this populations we observed Prdm1 5 included a included ( transduction retroviral of indicator an as served which cassette, PGK-EGFP a as well as activity, transcriptional of reporter cloned the fused and motifs TCF-1-binding mutated or wild-type containing +318) IgG to each locus in sorted T sorted in locus each to IgG of region the to TCF-1 of binding the for consistent with a published report published a with consistent mutated TCF-1-binding sites as a positive control positive a as sites TCF-1-binding mutated mutated and wild-type of the constructs We similar prepared Prdm1 5′-Reg ′ ′ regulatory region, at day 8 after infection ( infection after 8 day at region, regulatory regulatory regions (positions −24790 to −23864) fused to the the to fused −23864) to −24790 (positions regions regulatory T T T T Prdm1 Subsequently, we performed reporter assays using a self-inactivating –900 H H FH FH + 1 +IgG 1 +anti-TCF-1 cells. ( +IgG +anti-TCF-1 EGFP EGFP promoter (positions −757 to +180) ( +180) to −757 (positions promoter a) or four mice ( Mut2 WT 17.0 24.9 Prdm1 aDV Lef1 e + d (e) or SIN SIN ,g) Frequency of Thy-1.1 T P-Bcl6 FH A 5 NCE ONLINE PUBLIC ONLINE NCE Mut1-2 Mut2 relative to the binding of isotype-matched control control isotype-matched of binding the to relative ′ cells than did the vector encoding the wild-type 3 Mut1 WT c Bcl6 regulatory region reporter containing wild-type or wild-type containing reporter region regulatory 4 f (g). * . We cloned the We. the cloned Mut1-2 Mut 4.99 25.7 promoters to sequence encoding Thy-1.1 as a as Thy-1.1 encoding sequence to promoters d–g) per group (error bars ( SIN SIN SIN SIN P < 0.01 and **

g e –24790 –24609, TGTTGTCT –24609, CTTTGTTG –24609, TGTTGTCT + + –24609, CTTTGTTG 5′-Reg 5′-Reg 5′-Reg 5′-Reg FH –24520, AGACA –24520, CAAAG –24520, AGACA Thy-1.1 cells (%) Thy-1.1 cells (%) –24520, CAAAG 10 15 20 25 10 20 30 cells, but not in T in not but cells, 0 5 0 Bcl6 –24076, AGACA –24076, AGACA –24076, CAAAG –24076, CAAAG 5′-Reg Prdm1 5′-Reg + 3 Lef1 Lef1 Bcl6 cells in the GFP

–23864 5 ** A promoter and the 5 the and promoter and suggested the reliability of of reliability the suggested and * TION

** –757 P-Prdm1 P-Prdm1 P-Prdm1 P-Prdm1 + ** cells and a higher mean fluo and mean a cells higher 5 P < 0.001 (unpaired two-tailed promoter (positions −650 to to −650 (positions promoter P-Bcl6 ** ′ regulatory region with the the with region regulatory in in vivo

Exon 1 Exon 1 Exon 1 Exon 1 6 Fig. nature immunology nature +180 Thy-1.1 Thy-1.1 Thy-1.1 Thy-1.1 a,eg), s.e.m.). Fig. 6d Fig. 2 Thy-1.1 MFI (×102) 6 Fig. Thy-1.1 MFI (×10 ) we system, reporter 0 1 2 3 4 0 2 4 6 c + 3 H Prdm1 5′-Reg 5′-Reg ). In addition, we we addition, In ). (transduced) T 5 P-Pgk1 P-Pgk1 P-Pgk1 P-Pgk1 Lef1 1 cells, obtained obtained cells, 1 . We then trans We . then a ** Bcl6 Bcl6 ** , ). e ** ** ), which was was which ), P-Bcl6 ′ EGFP EGFP EGFP EGFP regulatory regulatory ** promoter promoter promoter promoter + T Fig. 6 Fig. FH Mut1-2 Mut2 Mut1 WT Mut WT Prdm1 SIN SIN SIN SIN cells cells Lef1 FH b

). ). - - -

© 2015 Nature America, Inc. All rights reserved. TLE3 and TLE4. ( TLE4. and TLE3 TLE2, TLE1, to antibody Anti-TLE14, anti-Myc. with analysis immunoblot by followed vector, empty and/or Bcl-6 Myc-tagged TLE3, untagged p33, Flag-tagged encoding plasmids of combinations various overexpressing cells 293T representative of three independent experiments ( experiments independent three of representative during T during stabilized resultant and Wnt signaling activated the Given ( anti-Myc. or anti-Bcl-6 with analysis immunoblot by followed vector, empty and/or of form stabilized a Myc-tagged Bcl-6, untagged p45, Flag-tagged encoding plasmids of blot) (above combinations various overexpressing cells 293T of lysates of ih yae o atvtd CD4 activated of lysates with a result we further confirmed by endogenous co-immunoprecipitation sion of ICAT (a 9-kilodalton protein that inhibits this interaction locus exhibited an elevated frequency of Thy-1.1 of frequency elevated an exhibited locus that that stabilized and Bcl-6 p45, Weoverexpressed β (throughout). ( (throughout). immunoprecipitation without lysates of margin) right (antibodies, analysis immunoblot Input, anti-Myc; with analysis immunoblot by followed (EV), vector empty or (Flag-p45) p45 or p33) (Flag- p33 Flag-tagged and (Myc–Bcl-6) Bcl-6 Myc-tagged overexpressing cells 293T of lysates cls ehne frain f h p3Bl6 ope ( complex p33–Bcl-6 the of formation enhanced cells) T expression via direct binding to the the to binding direct via expression day 3 after infection with LCMV ( LCMV with infection after 3 day ICAT, at SMARTAoverexpressing of assessed cells transfer adoptive e t a euto i T in reduction a to led assay of 293T human embryonic kidney cells co-transfected with with co-transfected plasmidscells expressing kidney Bcl-6 andembryonic the human p33 or 293T p45 of isoform assay of TCF-1 ( binding also observed of TCF-1 to Bcl-6 in a co-immunoprecipitation found that interference found of that interference the populations with mutated TCF-1-binding sequences in the the in sequences TCF-1-binding mutated with populations ( enhances and protein 7 Figure ( nature immunology nature these results demonstrated that TCF-1 promoted demonstrated results these wild-type the the with encoding transduced vector counterparts their of that to relative Thy-1.1 heterodimer with p45 (ref. (ref. p45 with heterodimer stabilized pathway, Wnt the of activation Upon T cells stimulated with anti-CD3 and anti-CD28 and infected with retrovirus expressing a mutated a mutated expressing retrovirus with infected and anti-CD28 and anti-CD3 with stimulated T cells functioned as a transcription factor and not as a co-activator of of co-activator a as not T and regulate to Bcl-6 factor transcription a as functioned ner a in family TLE the of corepressors transcriptional with a complex forms constitutively p33 p45, to trast areas indicate percent Thy-1.1 percent indicate areas outlined to adjacent Numbers + Bcl-6-ov). (p33-ov Bcl-6 and p33 both or (Bcl-6-ov) Bcl-6 (p33-ov), p33 overexpressing cells or (−) cells transduced (P- factors to regulatetranscription of gene set expressiona with ascomplexes a co-activatorprotein-protein forms TCF-1 or co-repressor Bcl-6 and TCF-1 between interaction Protein-protein repressed c a i. 6d Fig. -catenin is involved in the interaction between p45 and Bcl-6. Bcl-6. and p45 between interaction the in involved is -catenin ) Immunoprecipitation (as in in (as ) Immunoprecipitation ) Immunoprecipitation (IP), with anti-Flag, of anti-Flag, with (IP), ) Immunoprecipitation Prdm1 1 9 . . Notably, TLE3 (a dominant member of the TLE family in CD4 β ctnn irpe te 4–c- itrcin ( interaction p45–Bcl-6 the disrupted -catenin

, TCF-1 physically interacts with Bcl-6 Bcl-6 with interacts physically TCF-1 ) (negative control), as well as vectors for the overexpression of p33 and Bcl-6, followed by 2 d of incubation with IL-2, gated on non- on gated IL-2, with incubation 2 d of by followed Bcl-6, and p33 of overexpression the for vectors as well as control), ) (negative e FH Prdm1 ), which indicated that TCF-1 positively regulated regulated positively TCF-1 that indicated which ), differentiation b ) Immunoprecipitation (as in in (as ) Immunoprecipitation d β expression via direct DNA-binding activity. DNA-binding direct via expression ) Flow cytometry of GFP of cytometry ) Flow -catenin (Myc– -catenin FH Bcl6 differentiation. In support of that proposal, we proposal, of that In support differentiation.

transcription. transcription. FH aDV 3 differentiation in recipient mice given given mice recipient in differentiation 1 + a , these data suggested that p45 mainly mainly p45 that suggested data these , cells. ( cells. 1 β ) of lysates of lysates ) of A 9 -catenin–p45 complex by overexpres -catenin–p45 NCE ONLINE PUBLIC ONLINE NCE β ). We further investigated whether whether investigated further We ). + -cat) -cat) Prdm1 cls ( cells T

Supplementary Fig. 8 Fig. Supplementary e ) MFI of Thy-1.1 in cells as in in as cells in Thy-1.1 of ) MFI

β + ou ( locus ctnnidpnet man -catenin-independent Bcl6 CD4 a ) upeetr Fg 7 Fig. Supplementary + β ou. GFP locus.

-catenin and observed observed and -catenin a

– i. 6f Fig. c IP: anti-Flag EV Flag-p33 Flag-p45 Myc–Bcl-6 d a ) or two independent experiments with three replicates ( replicates three with experiments independent two ) or Bcl6 Input P-Bcl6 Mut β + cells and MFI of of MFI and cells A -catenin forms a forms -catenin P- TION Prdm1 expression but expression , g ). Together Together ). +

T ). In con In ). β CD4 7 Fig. Fig. 7 Fig. -catenin -catenin Fig. 7 + + 10 10 10 10 – – FH Prdm1 0 2 3 4 5 3 Thy-1.1 + + – – 6 36 cell cell 010 Bcl6 . We d , b 3 c a + + – – , normalized to GFP expression. * expression. GFP to , normalized 2 7 ). ). 10 ). ). ), ), ), + - - - ) 1.81 – 1.15 Anti-Myc Anti-Flag Anti-Myc 3

10 4 10 (refs. observed a comparable MFI for Thy-1.1 in Bcl-6-transduced CD4 Bcl-6-transduced in Thy-1.1 for MFI comparable a observed mutated to its lack of binding sites for either TCF-1 (refs. (refs. TCF-1 either for sites binding owing of control, lack its negative to a as +180) to −1119 (positions promoter motifs. In addition, we included a reporter containing the the containing reporter a included we TCF-1-binding addition, In inactivated motifs. with construct mutated a selected we the wild-type wild-type the with transduced cells of that than Thy-1.1 of expression lower much SMRATA the with mutated transduced cells T cells and non-transduced CD4 non-transduced and cells T the transcription of of transcription the favored potentially complex p33–Bcl-6 the that suggested data these on effect no to little exerted Thy-1.1 ( expressing expressing binding to the promoter region promoter the to binding of Bcl-6 targets. Because Bcl-6 represses its own transcription through transcription the modulated interactions such whether investigated Bcl-6. with complex a of formation the through quency of Thy-1.1 of quency fre lower a to led TCF-1 of LCMV, with knockdown infection after activity of the the of activity TCF-1 decreased expression would whether the affect transcriptional and TCF-1 Thy-1.1 driven by the by the driven Thy-1.1 hc idctd ht 3 pooe T promoted p33 that indicated which ( T regulate to Bcl-6 of upstream acted TCF-1 that confirm Tofurther axis Bcl-6–Blimp1 the of upstream acts TCF-1 Thy-1.1 of frequency highest the exhibited p33 and Bcl-6 both of expression Fig. 8a Fig. 5 FH Given the physical interaction between p33 and Bcl-6, we further further we Bcl-6, and p33 between interaction physical the Given differentiation, we combined shRNA-mediated knockdown of of knockdown shRNA-mediated combined we differentiation, p33-ov Input IP: anti-Flag EV Myc– Bcl-6 Flag-p45 b , 38 2.97 1.37 + , 3 Bcl6 b cells in the GFP the in cells β Fig. Fig. 7d ). 9). After transduction of the Thy-1.1 reporter containing the -cat ). Furthermore, regardless of TCF-1-knockdown status, status, TCF-1-knockdown of regardless Furthermore, ). Tcf7 Bcl6 promoter and Bcl-6 and/or p33 expression plasmids, we Bcl6 Bcl6 Bcl-6-ov -specific shRNA than among ‘TCF-1-intact’ T shRNA among than ‘TCF-1-intact’ -specific , promoter–Thy-1.1 reporter systems to investigate investigate to systems reporter promoter–Thy-1.1 + + + – e 1.82 1.47 ). ). As‘double’ expected, expression of p33 and Bcl-6 promoter ( promoter + promoter. As expected, in cells analyzed at day 3 day at analyzed cells in expected, promoter.As Bcl6 + + + – cells and lower MFI of Thy-1.1 among T among Thy-1.1 of MFI lower and cells Bcl6 + + + Bcl6 reporter (P- reporter + + – –– during T during P + Bcl-6-ov p33-ov+ promoter. To avoid occupancy by TCF-1, by TCF-1, promoter. To occupancy avoid < 0.05 (paired two-tailed two-tailed (paired < 0.05 population as well as the highest MFI for for MFI highest the as well as population Anti-Flag Anti-Myc Anti-Bcl-6 Anti-Myc Anti-Bcl-6 11.3 1.65 d Prdm1 Fig. 8a Fig. , e ; error bars ( bars ; error 1 + 4 FH T cells, whereas cells with ectopic ectopic with cells whereas cells, T Bcl6 , we used a reporter construct of of construct reporter a used we , Input c IP: anti-Flag EV Tle3 Myc-Bcl-6 Flag-p33 promoter ( promoter

2 differentiation. e Thy-1.1 MFI (×10 ) , b 0 1 2 3 4 Mut) or or Mut) ). FH P-Bcl6 Mut differentiation, probably probably differentiation, e ), s.e.m.). ), Bcl6 * * * Prdm1 + + + – Fig. 7d Fig. promoter exhibited promoter exhibited + + + – TICLES E L C I RT A t P-Prdm1 -test). Data are are Data -test). 20 promoter promoter + + – – p33-ov +Bcl-6-ov Bcl-6-ov p33-ov – Anti-Flag Anti-TLE1–4 Anti-Myc Anti-Myc , , 3 e 3 ). Together ). ) or Bcl-6 Bcl-6 or ) FH FH Prdm1 cells cells cells cells  + -

© 2015 Nature America, Inc. All rights reserved. with with three ( NS, of Data not are experiments two significant. representative independent ( eage emerges as early as 2 d after infection, but how the T sion sion are more prone into T to differentiate MRA el (CD45.1 co-transduced cells we SMARTA this, confirm To TCF-1. of knockdown by ated oe oal, Thy-1.1 notably, More virus-specific CD4 virus-specific 1.1 regulating T regulating data in axis these Bcl-6–Blimp1 of Together regulator upstream an as TCF-1 TCF-1. established of knockdown by mediated was that virus-activated CD4 virus-activated f h AFCE fml o tasrpin factors transcription of family ATF-CREB the of of formation members or c-Jun factor the transcription the with through heterodimers ATF3 effect auto-repressive the reversing by expression gene own its activate to mechanism similar a uses ATF3 Bcl-6 forms a complex with p33. Indeed, the repressor transcriptional cells, cells, whereas only ~15% of the GFP fate is determined remains unknown. In this study, we showed showed we study, this T in In diverged expression TCF-1 that unknown. remains determined is fate counteract Bcl-6-mediated auto-repression, which is similar to to similar is which auto-repression, to Bcl-6-mediated factor transcription counteract another recruit might complex p33–Bcl-6 due to shRNA-mediated knockdown had diminished early T early diminished had knockdown shRNA-mediated to due shRNA (marked as GFP overexpression of Bcl-6 completely ‘rescued’ the T the ‘rescued’ completely Bcl-6 of overexpression the Thy-1.1 the ( (CD45.1 the T the to commitment early dictated expression TCF-1 Thus, ferentiation. decreased in T decreased depend on its binding partners binding its on depend nario. First, the transcriptional sce repression or this activation of in Bcl-6 might involved be might mechanisms possible Two feedback. promoted promoted Bcl6 infection. viral overexpression would ‘rescue’ the defects in T 8 Figure S E L C I RT A  (CD45.2 after infection, being increased in early T early in increased being infection, after cells. ( cells. During acute viral infection, commitment to the T DISCUSSION after infection. Numbers above outlined areas indicate percent Thy-1.1 percent areas indicate above outlined Numbers after infection. by of followed infection the host mice with LCMV and at analysis day 3 or wild-type mutated shRNA (with human specific CD4 as and (hCD4) (left margin) a reporter) transfer of transfer SMARTA cells (CD45.1 of wild-type of mice (CD45.2 wild-type comparable to that of Thy-1.1 of that to comparable T at percent to results areas indicate adjacent (left) left. Numbers outlined by of followed plots), infection the host mice with LCMV; of summary right, (above TCF-1-shRNA and/or Bcl-6 (Bcl-6-ov) overexpressing retrovirus to differentiate into T Consistent with published data published with Consistent c a FH

) Flow cytometry ) of Flow cytometry SMARTA cells in mice (CD45.2 wild-type ) Flow cytometry ) of Flow cytometry hCD4 Of note, in addition to directly regulating the transcription of of transcription the regulating directly to addition in note, Of We reasoned that if TCF-1 acts upstream of Bcl-6 pathway, Bcl-6 Bcl-6 pathway, Bcl-6 of upstream acts TCF-1 if that Wereasoned + cells. * cells. ). We then transferred these cells into wild-type recipient mice mice recipient wild-type into cells these transferred Wethen ). , the p33 isoform of TCF-1 formed a complex with Bcl-6 and and Bcl-6 with complex a formed TCF-1 of isoform p33 the , FH b ) ) MFI of in Thy-1.1 cells as in

lineage versus commitment to the T the to commitment versus lineage + TCF-1 TCF-1 acts on T + ) transduced with retrovirus expressing empty or vector expressing ) with retrovirus transduced ), which we subsequently infected with LCMV ( LCMV with infected subsequently we which ), P a Bcl6 , < ( 0.001 (unpaired + b FH SMARTA cells differentiated into SLAM SMARTA differentiated cells ) ) or six ( H differentiation. transcription by reversing Bcl-6-mediated negative negative Bcl-6-mediated reversing by transcription 1 cells. We reasoned that during early viral infection, We1 infection, cells. viral early that during reasoned + Bcl6 T cells unable to their TCF-1 increase expression + c FH T cells with higher TCF-1 expression are poised ) ) mice per group bars (error ( FH cells, whereas those with lower TCF-1 expres promoter reporter (above plots) (as plots) (above reporter in promoter + + + + ) ) and virus expressing Bcl-6 (marked as Thy- ) given adoptive transfer ) of transfer given SMARTAadoptive cells differentiation upstream of upstream Bcl-6 signaling. differentiation + GFP GFP ) with retrovirus encoding encoding retrovirus with ) + + b SLAM + el ehbtd T exhibited cells + 9 ) ) or ( paired cells ( cells ) not transduced ) (−) not with or transduced transduced and our data reported above, 70% of and our data reported 8 ; the activation might occur when when occur might activation the ; a lo , normalized , to GFP normalized expression. + cells T became CXCR5 H Fig. 8 Fig. 1 cells and T and cells 1 c H ) two-tailed ) two-tailed FH + 1 cells. 1 In of cells. support that, c T H FH ), which suggested that that suggested which ), cells but substantially substantially but cells FH 1 lineage during acute acute during lineage 1 differentiation medi b cells in the spleen H , c 1 lineage or T FH ), ), s.e.m.). FH FH FH differentiation differentiation 4 cell deficiency deficiency cell 0 t lo cells at 2 days days 2 at cells H cells ( -test). -test). . Second, the the Second, . CXCR5 1-versus-T Tcf7 + ) ) given Fig. Fig. 6 Tcf1- -specific -specific Fig. 8 Fig. Fig. Fig. 8

FH FH + b T dif + ), ), lin

c c FH FH ). ). ). ). - - - - -

ated T ated ently or jointly to increase TCF-1 expression in early T early in expression TCF-1 increase to jointly or ently independ work mechanisms proposed these whether determine to fashion through the STAT3 signaling axis signaling STAT3 the through fashion of favor upregulation TCF-1 the in immediate expression a paracrine secreted by dendritic cells, such as cells, by IL-6 and dendritic secreted IL-12 (refs. possibilities exist for achieving this regulation. First, certain cytokines plex might be more important for the effector function of T of function effector the for important more be might complex. plex p33–Bcl-6 the by not com but However, that the p33–Bcl-6 we phase, at speculate the effector alone, TCF-1 by driven was was probably transmitted from naive CD4 naive from transmitted probably was tion of T of tion differentia the to regulate mechanisms of regulatory layers multiple upstream of the the of upstream of transcription Tcf7 the promote might receptors of family Notch the sensory neurons sensory of development the during Tlx3 factor transcription the and DRG11 of complex the through by Brn3a DRG11 factor transcription factor recruited newly transcription the of auto-repression of reversal early early commitment to the T of the kb upstream −1.3 own expression by binding to the TCF-1-binding sequence at position Although Although these these in complex p33–Bcl-6 the of absence the to due probably responses, the first wave of Bcl-6 expression for priming of the early T early the of priming for expression Bcl-6 of wave context, first the this In induction. new required expression Bcl-6 whereas of Bcl-6 (data not shown), shown), not (data Bcl-6 of TCF-1 expression in early T early in expression TCF-1 T FH Although the extrinsic and intrinsic factors involved in increasing in increasing involved factors and intrinsic extrinsic the Although Our study revealed that TCF-1 primed early commitment of the the of commitment early primed TCF-1 that revealed study Our fate and also modulated the effector function of fully differenti of fully function effector the modulated also and fate through conserved putative binding sites at a position −31 kb kb −31 position a at sites binding putative conserved through EV c shRNA TCF-1- a Tcf7 FH FH cells. TCF-1 expression preceded Bcl-6 expression during during expression Bcl-6 preceded expression TCF-1 cells. −/− Tcf7 aDV cells during acute viral infection. viral acute during cells SLAM CD4 T 10 10 10 10 10 10 10 0 0 −/− A 3 4 5 3 4 5 2 FH NCE ONLINE PUBLIC ONLINE NCE 4 Thy-1.1 CXCR5 Tcf7 cells. and wild-type T wild-type and 0 1 TCF-1-shRNA 0 . Together these data established that TCF-1 used used TCF-1 that established data these Together . +Bcl-6-ov 10 10 locus WT 21.5 10.7 – 3 73.9 3 26.9 10 10 Tcf7 4 4 FH P-Bcl6 10 10 4 Tcf7 FH 5 3 5 promoter fate the because main of fraction TCF-1 . Third, TCF-1 might also regulate its its regulate also might TCF-1 Third, . cells remain to be elucidated, several several elucidated, be to remain cells −/− TCF-1-shRNA Bcl-6-ov FH cells were less ‘helpful’ to B cell cell B to ‘helpful’ less were cells A Mut 0.87 0.64 TION cells expressed a similar level level similar a expressed cells 72.3 14.7 4 3 . Further studies are needed are needed studies . Further

+ 4 T cells upon activation, activation, upon cells T 2 nature immunology nature . Second, signaling via via signaling Second, . b 2 SLAMloCXC5+ cells (%) Thy-1.1 MFI (×10 ) Bcl-6-ov Bcl-6-ov TCF-1-shRNA + TCF-1-shRNA – 0 1 2 3 4 5 20 40 60 80 0 P-Bcl6 Mut P-Bcl6 WT EV * FH 15 * * * -like cells. -like , 1 NS -shRNA TCF-1 7 FH ), ), might * FH cells. cells. fate fate - - - -

© 2015 Nature America, Inc. All rights reserved. entiated T entiated induced T induced of the TCF-1 pathway to improve the quantity and of quality vaccine- responses. These findings provide important insights into modulation 12. 11. 10. 9. 8. 7. 6. 5. 4. 3. 2. 1. reprints/index.htm at online available is information permissions and Reprints The authors declare no competing interests.financial Z.X.; and X.Z., L.Y. and Y.W. the study. supervised L.Y. the designed study, analyzed the data and wrote the paper with L.X., Y.C. and performed the experiments; Q.B. analyzed the microarray data by GSEA; X.Z. and L.X., Y.C., Z.X., Q.H., Q.B., X.Y., Y.H.,R.H., H.W., T.Z., Z.F., A.Q. and J.Y. of Mouse Consortium) for permission to use the mouse with conditional knockout We thank the Institut Clinique de la Souris (part of the International Knockout online version of the pape Note: Any Supplementary Information and Source Data files are available in the codes. Accession the in vers available are references associated any and Methods M T TCF-1 initiated infection, nature immunology nature Institutions (A.Q.). and the Priority Academic Program Development of Jiangsu Higher Education 81471624 to L.Y. and 31000631 to Z.X.), the TalentChina1000 Plan Program (L.Y.) and X.Z.), the National Natural FoundationScience of China (31470870 to X.Z., National Basic Research Program of China (973 Program, to 2013CB531500, L.Y. center of Third Medical Military University for cell Supportedsorting. by the for performing microarray experiments and data analysis; and the core facility R. University)Ahmed (Emory for SMARTA mice and retroviral vectors; CapitalBio C AUTHO A c OM Tcf7 ethods In conclusion, we have demonstrated that in response to acute viral

kno T cells with B cell helper function. helper cell B with cells T Choi,Y.S. D. Yu, R.I. Nurieva, R.J. Johnston, disease. in roles and function, differentiation, cell helper follicular T S. Crotty, helper follicular T of Pathophysiology C.G. Vinuesa, & J. Banchereau, H., Ueno, C.H. Kim, P.Schaerli, D. Breitfeld, (TFH). cells T CD4 helper Follicular Crotty,S. centers. Germinal M.C. Nussenzweig, & G.D. Victora, Offit, P.A. W.A.& Orenstein, S.A., Plotkin, , 1545–1552 (2000). 1545–1552 192, production. immunoglobulin support and follicles, cell B to localize of T follicular helper cell differentiation. cell helper follicular T of (2011). lineage commitment. lineage n gria cne-oaie sbe o CXCR5 of subset center-localized germinal a in Science Immunity mice. and humans in cells 1373–1381 (2001). 1373–1381 and have the capacity to form memory.formcapacitytheto have and 429–457 (2012). 429–457 ion of the pape the of ion P ; ; H.H. Xue (University of Iowa) for E TI wl R R C h tasrpinl erso Bl6 iet T olclr epr cell helper follicular T directs Bcl-6 repressor transcriptional The al. et NG , 1001–1005 (2009). 1001–1005 325, FH FH edgmen ON etal. , 529–542 (2014). 529–542 41, Subspecialization of CXCR5 of Subspecialization al. et FI cells to perform their effector function in helping B in helping cell function effector their to perform cells cells. CXC chemokine receptor 5 expression defines follicular homing follicular defines expression 5 receptor chemokine CXC al. et Follicular B helper T cells express CXC chemokine receptor 5, receptor chemokine CXC express cells T helper B Follicular al. et TRIBUTI c6 eits h dvlpet f fliua hle cells. helper follicular T of development the mediates Bcl6 al. et N Bcl6 expressing follicular helper CD4 Tcells are fate committed early l Bcl6 and Blimp-1 are reciprocal and antagonistic regulators antagonistic and reciprocal are Blimp-1 and Bcl6 al. et . A GEO: microarray data, data, microarray GEO: N r CIAL CIAL I t . r Immunity s . ONS

Nat. Immunol. Nat. N aDV T FH E , 457–468 (2009). 457–468 31, R A differentiation and ‘warranted’ differ and ‘warranted’ differentiation ES NCE ONLINE PUBLIC ONLINE NCE J. Exp. Med. Exp. J. T J. Immunol.J. S Tcf7 Science , 142–152 (2015). 142–152 16, 6th edn. (Elsevier, 2013). (Elsevier, edn. 6th Vaccines fl/fl + T cells: B helper activity is focused is activity helper B cells: T mice and retroviral vectors; Annu. Rev. Immunol. Rev. Annu. GSE

, 1553–1562 (2000). 1553–1562 192, , 1006–1010 (2009). 1006–1010 325, +

cells. T 190 6569 , 4014–4026(2013)., nu Rv Immunol. Rev. Annu. http://www.nature 3 A . TION . x. Med. Exp. J.

, 621–663 29,

J. Exp. Med. Exp. J. online online

.com/ 193,

30,

-

25. 24. 43. 42. 41. 40. 39. 38. 37. 36. 35. 34. 33. 32. 31. 30. 29. 28. 27. 26. 23. 22. 21. 20. 19. 18. 17. 16. 15. 14. 13.

ensures CD4 ensures ky euaoy lmn o te ee noig PU.1. encoding gene the (2006). of element regulatory key a Yang, Y., Xu, J., Niu, Y., Bromberg, J.S. & Ding, Y. T-bet and eomesodermin play eomesodermin and T-betY. Ding, & J.S. Bromberg, Y., Niu, J., Xu, Y.,Yang, regulates negatively factor-1 cell T J.M. Sen, & A. Ghosh, A., Sharma, Q., Yu, ee, B.N. Weber, L. Durant, I. Regadas, the and 3 factor transcription Activating P.N. Anderson, & G. Raivich, D., Hunt, K. Hatzi, G.D. Barish, K. Satoh, safeguard Lef1 and Tcf1 output, to inception From H.H. Xue, & F.C. Steinke, Yu, S. F.Rosenbauer, Li, L. Yamane,J., W.E.Zhu, Paul, & populations. H. cell T CD4 effector of Differentiation X. Liu, The W. Held, & H. F.,Clevers, V.,Beermann, Ioannidis, S. Verbeek, J.S. Hale, H.D. Marshall, potentCrotty,STAT5 Y.S.,S.a& Choi,Yang,J.A.R.J.,isJohnston,J.A., Diamond, u Q. Yu, functional and precursors memory of generation edge: Cutting H.H. Xue, & X. Zhou, X. Zhou, F.C.Steinke, signaling Wnt the by responses cell T mature of Regulation D.M. Zhao, & H.H. Xue, W.Ise, S. Nakayamada, D. Eto, STAT1edge: required Cutting Crotty,is S. & C. Lao, Yang,Y.S.,J.A., D., Choi, Eto, helper follicular in Bcl6 of regulation Dynamic S. Crotty, & J.A. Yang,Y.S., Choi, H. Xu, expression. epr ad hle 1cl lnae ae eeae atr ct vrl infection. viral acute after generated are lineages Immunity 1-cell helper T and helper- and memory Th1 CD4 roles in T cell development and malignancy.developmentand cell T in roles n rdnaty nue pia fliua hle C4 cl (f) differentiation. ONE PLoS (Tfh) cell T CD4 helper follicular optimal induce redundantly and T-helper-celldevelopment. T cell pathogenicity and homeostasis. and pathogenicity cell T (2004). 8017–8021 101, signaling. Wnt of regulator negative a (ICAT), factor phosphorylation. differentiation. (2011). T cell factor 1. factor cell T xrsin f L1 fml o ctkns n poet mc fo experimental from mice protects and cytokines encephalomyelitis. of autoimmune family IL-17 of expression CD8 memory pathway. differentiation. cell helper follicular early Immunol. J. for induction Bcl6 IL-6-mediated for cells. (Tfh) T CD4 nervous system. nervous cells. immune innate and (2014). cells T of development Immunol. Rev. Annu. differentiation.cell TFH ofregulator negative Th17. versus (2008). Th1 8700–8710 181, to differentiation cell T directing in roles critical factor-1. transition. ICOS-driven motility. ICOS-driven differentiation. T cell fate and interact with Runx3 to silence to Runx3 with interact and fate cell T innate immune response. immune innate , 646–656 (2014). 646–656 15, mechanisms. wth eobnto i bt B el ad cells. T (2011). and cells B both in recombination switch (2009). transcription factor GATA-3 and repressing interferon- et al. A far downstream enhancer for murine Bcl11b controls its T-cell specific et al. Follicular T-helper cell recruitment governed by bystander B cells and cells B bystander by governed recruitment T-helpercell Follicular al. et cl fco 1 ntae te hle tp 2 ae y nuig the inducing by fate 2 type helper T the initiates 1 factor cell T al. et The transcription factor BATF controls the global regulators of class- of regulators global the BATFcontrols factor transcription The al. et IL-21 and IL-6 are critical for different aspects of B cell immunity cell B of aspects different for critical are IL-6 and IL-21 al. et rncito fco aheesue ooou 2 ntae follicular initiates 2 homologue achaete-scute factor Transcription al. et Differentiation and persistence of memory CD8 memory of persistence and Differentiation al. et J. Immunol. J. Ann. NY Acad. Sci. Acad. NY Ann. , 805–817 (2013). 805–817 38, Anteriorization of neural fate by inhibitor of inhibitor by fate neural of Anteriorization al. et C6 rhsrts f cl dfeetain i mlil distinct multiple via differentiation cell Tfh orchestrates BCL6 al. et Immunity The TCF-1 and LEF-1 transcription factors have cooperative and opposing itnt eoy CD4 memory Distinct al. et , e17739 (2011). e17739 6, ies tres f h tasrpin atr TT cnrbt to contribute STAT3 factor transcription the of targets Diverse al. et Blood + , 3049–3053 (2013). 3049–3053 190, + n M-o-otiig -el atr eurd o thymocyte for required factor T-cell HMG-box-containing An al. et Bcl-6 and NF- and Bcl-6 al. et ul oe f l3 s ouao o Prl tasrpin and transcription Prrxl1 of modulator as Tlx3 of role Dual al. et TCF-1 and LEF-1 act upstream of Th-POK to promote the CD4 the promote to Th-POK of upstream act LEF-1 and TCF-1 al. et CD8 J. Exp. Med. Exp. J. T cells depends on T cell factor-1 and lymphoid enhancer-binding lymphoid and factor-1 cell T on depends cells T A critical role for TCF-1 in T-lineage specification and specification T-lineage in TCF-1 for role critical A al. et Differential expression of Ly6C and T-bet distinguish effector T-betdistinguish and Ly6C of expression Differential al. et Nature Nature Lymphoid cell growth and transformation are suppressed by suppressed are transformation and growth cell Lymphoid al. et Immunity Front. Mol. Neurosci. Mol. Front. Biochim. Biophys. Acta Biophys. Biochim. al T1 el ifrnito i mre b a f cell-like Tfh a by marked is differentiation cell Th1 Early al. et + , 902–911 (2013). 902–911 122, thymocyte survival. thymocyte Curr. Opin. Immunol. Opin. Curr. , 919–931 (2011). 919–931 35,

+ Nature , 445–489 (2010). 445–489 28, , 2722–2726 (2012). 2722–2726 189, cell properties during viral infection. , 70–74 (1995). 70–74 374, , 63–68 (2011). 63–68 476, Genes Dev. Genes , 229–240 (2010). 229–240 33,

Nature , 539–553 (2015). 539–553 212,

, 523–527 (2013). 523–527 496, , 16–33 (2012). 16–33 1247, J. Immunol. J. B cistromes mediate opposing regulation of the of regulation opposing mediate cistromes κB 507

, 2760–2765 (2010). 2760–2765 24, , 513–518 (2014). 513–518 , Immunity , 7 (2012). 7 5, + Nat. Immunol. Nat.

cls ih omtet o follicular T to commitment with cells T

, 366–372 (2013). 366–372 25, , 1121–1131 (2014). 1121–1131 1839, Immunity , 3946–3952 (2011). 3946–3952 186, J. Exp. Med. Exp. J. in CD8 in Cd4 , 605–615 (2010). 605–615 32, γ. a. Immunol. Nat.

Nat. Immunol. rc Nt. cd Si USA Sci. Acad. Natl. Proc. muo. Res. Immunol. 37 , 691–697 (2001). 691–697 2, -catenin–TCF-1 pathway β-catenin–TCF-1 , 813–826 (2012).813–826 , a. Genet. Nat.

+ 209 Immunity T cells. T TICLES E L C I RT A -catenin and T cell T and β-catenin + , 243–250 (2012).243–250 , T cells depend on depend cells T Nat. Immunol. Nat. 536–543 12, 35, 633–646 10, 992–999 . Immunol. J. 45–55 59, 27–37 38,  +

© 2015 Nature America, Inc. All rights reserved. staining has been described been has staining inserted the multiple cloning site into the original construct by replacing replacing by construct original the the into site cloning multiple the inserted Laboratory. Wild-type mice and and mice Wild-type Laboratory. intranasally intranasally with 0.1 LD of this strain were used to establish acute in infection mice. Mice were infected Cytoperm Fixation/Permeabilization Kit (554714, BD Biosciences). T Biosciences). BD (554714, Kit Fixation/Permeabilization Cytoperm Cytofix/ a with performed were Ki67 and B granzyme for Staining (wt/vol). 2 Table in listed are staining cytometry flow for used reagents and antibodies The (Emory). Health of Institutes National US the of facility core LCMV epitope of was amino 66–77 by acids provided glycoprotein a tetramer Animals, infectious agents and tamoxifen treatment. tamoxifen and agents infectious Animals, METHODS ONLINE nature immunology nature provided by H.H. Xue (University of Iowa) of (University Xue H.H. by provided Star). Star). Major histocompatibility complex class II (I-A FACSCanto II (BD and Biosciences) were analyzed with FlowJo software (Tree antibodies. and cytometry Flow mouse. per day per mg 1 of at dose a mice 8-week-old to 6- into neally intraperito of injected was concentration Sigma-Aldrich) (S5007; a oil sunflower in at mg/ml 10 Sigma-Aldrich) (T5648; Tamoxifen Rico/8/H1N1). provided by R. Ahmed (Emory University), and University), 2 by × provided (Emory R. 10 Ahmed was strain) (Armstrong virus LCMV University. Medical Military Third the of Committees Use and Care Animal Institutional the of guidelines the with accordance in performed were experiments mouse All University. Medical Military Third the of regulations biosafety institutional with accordance in of with LCMV A Mice or infected reconstitution. were virus influenza housed omization or ‘blinding’. Bone marrow chimeras were after infected 8–10 weeks were infected at 6–10 weeks of age, and both sexes were included without rand performed by of for with cells the incubation tetramer 1 performed h at 37 °C. For was complex staining class II Major tetramer histocompatibility eBioscience). (00-5523; Set with the Foxp3/Transcription Buffer Factor Staining performed by transfection of 293T cells with the retroviral vectors along with plasmid plasmid pCL with along vectors retroviral the with cells packaged 293T of were transfection by Retroviruses sequencing. DNA by verified were sequences in presented are mutations All construct. the into insertion 5 mutated or wild-type to promoter the fused and 5 mutated and wild-type the cloned we Similarly, tions −650 to +318) and these promoters inserted into construct. the modified SMARTA(CD45.1 Clinique de la Souris (part of the International Knockout Mouse Consortium). n ad 5B/J (CD45.2 C57BL/6J and in, of expression transgenic with Mice (LCMV glycoprotein amino acids 61–77) into SMARTA mice. After 18 h, h, 18 After mice. SMARTA into 61–77) acids amino glycoprotein (LCMV originally constructed based on the pQCXIP backbone (Clontech) backbone on the pQCXIP based constructed originally was vector reporter retroviral self-inactivated The pMKO.1-hCD4. and GFP Tcf7 sequence was subcloned into the MIGR2 The vector. MIGR4. Theand shRNAMIT sequenceMIGR1, targetingvectors the into cloned and fied The (hCD4). CD4 human truncated ing in the constructs MIGR1 and pMKO.1-GFP, respectively, with sequence encod and pMKO.1-hCD4 were obtained by replacement of GFP encoding sequence (MSCV-IRES-hCD4) constructs MIGR4 The University). (Emory Ahmed R. MIT pMKO.1-GFP, constructs The (MSCV-IRES-Thy-1.1) and MIGR2 (MSCV-IRES-hCD2) were obtained from Iowa). of obtained (University were Xue Tle3 H.H. untagged from and ICAT Flag-tagged TCF-1, untagged or transduction. stabilized Myc-tagged expressing constructs retroviral and constructs Retroviral (552598; Kit Flow BrdU a with stained instructions. manufacturer’s the to according was Biosciences) BD cells T in BrdU killed. were BrdU in (5-bromodeoxyuridine) 0.5 ml PBS) 3 intraperitoneally h before mice mg (1.5 BrdU given were BrdU, mice analog thymidine the of incorporation HpaI (positions (positions −1119 to +180) and the andwild-type mutated XhoI (AGAAGCCAGTCATCAAGAAAC) was cloned into vectors pMKO.1- vectors into cloned was (AGAAGCCAGTCATCAAGAAAC) eco -Thy-1.1- . . SMARTA were cells activated . Surface staining was performed in PBS containing 2% BSA or FBS FBS or BSA 2% containing PBS in performed was staining Surface . -Thy-1.1- EcoRI BamHI + ) mice were provided by R. Ahmed (Emory University). University). (Emory Ahmed R. by provided were mice ) - EcoRV 50 Prdm1 (50% lethal dose) influenza A virus (strain A/Puerto ′ sequence with with sequence -regulatory region (positions −24790 to −23864) −23864) to −24790 (positions region -regulatory 9 - + . Staining of Bcl-6, TCF-1, T-bet and Foxp3 was was Foxp3 and T-bet TCF-1, Bcl-6, of Staining . BamHI n CD45.1 and promoter (positions −757 to +180) and the the and +180) to −757 (positions promoter Flow cytometry data were acquired with a with acquired were data cytometry Flow Tcf7 Cd4 . Then, we cloned the the cloned we Then, . in in vivo −/− -Cre, mice with the ERT2-Cre knock- ERT2-Cre the with mice -Cre, mice and their control littermates littermates control their and mice Bcl6 XhoI + 2 by injection of by 200 injection ) mice were from the Jackson Jackson the from were mice ) 0 coding sequences were ampli were sequences coding MIGR1 (MSCV-IRES-GFP) (MSCV-IRES-GFP) MIGR1 with permission from Institut Institut from permission with - BstZ17I ′ regulatory regions before before regions regulatory b ) ) tetramer specific for the 5 β plaque-forming units plaque-forming - -catenin, Flag-tagged Flag-tagged -catenin, Bcl6 MluI

Tcf7 Prdm1 promoter (posi Supplementary Supplementary Figure 6b Figure - fl/fl NruI µ Tcf7 g g of peptide mice were were mice 3 promoter promoter 4 - . . We first HindIII coding coding in in vivo FH , c . All All . cell cell ------

ChIP. primers for ‘test genes’ ( genes’ ‘test for primers Vazyme) on a TouchCFX96 and the System appropriate(Bio-Rad) Real-Time the for analyzed of expression genes with AceQ various was qPCR SYBR Green Master Mix (Q111, cDNA resulting The Scientific). Thermo (K1632; Kit Synthesis cDNA Strand First Minus H RevertAid a with reverse-transcribed and extracted was Total RNA Technologies). Life (10296; reagent LS TRIzol T mice mice (CD45.1 Institute) was used for analysis for used was Institute) (Broad software GSEA analysis. microarray for CapitalBio to submitted was and Technologies) (Life protocol reagent TRIzol the to according extracted was Total RNA University. Medical Military Third of center facility core the at Biosciences) (BD a sorter FACSAria with cell II Technologies) Life (10296; 2 Table Supplementary in identified (all anti-CXCR5 and anti-CD25 anti-GITR, anti-SLAM, CD44, markers (Lin lineage for positive of cells depletion to subjected were LCMV with infection Microarray and bioinformatics analysis. bioinformatics and Microarray LCMV. with infection before reconstitution for weeks 8–10 Quantitative RT-PCR. Quantitative of 550 rads (CD45.1 each) wild-type bone marrow were transferred intravenously into lethally irradiated (two doses evr. ufcs f h Lin the of Surfaces Beaver). from BioLegend) coupled to Mag500 BeaverBeads Streptavidin all Matrix (22302; (PK136); anti-NK1.1 and (TER-119) anti-TER119 (N418), (RB6-8C5), anti-CD11c anti-Gr-1 (M1/70), anti-CD11b (RA3-6B2), anti-B220 (53-6.7), Tcf7 chimeras. marrow bone of Generation (Zeiss). software Examiner Image LSM with processed ined with a Zeiss LSM 510 confocal fluorescence microscope. The images were Kit (P-7481; Technologies) Antifade Life ProLong with and slides were exam on mounted were Coverslips nucleus. the define to used was Sigma-Aldrich) D9542; (4,6-diamidino-2-phenylindole; DAPI Technology). Signaling Cell and were with (C46C7; anti-TCF-1 stained were cells and fixed permeablized, units (day 2 or 3) or were infected intraperitoneally with 2 × 10 × 2 with intraperitoneally infected were or 3) or 2 (day units as described as Immunofluorescence staining. formaldehyde in medium. Chromatin fragments were prepared as described followed by immunoprecipitation with anti-TCF-1 (C63D9; Cell Signaling Signaling Cell (C63D9; anti-TCF-1 with immunoprecipitation by followed (CD45.2 wild-type naive into transferred adoptively were SMARTA cells) transduced wild-type mice wild-type (CD45.2 forming units (day 8 or later) LCMV Armstrong strain the following day. following the T of evaluation strain For Armstrong LCMV later) or 8 (day units forming cells, total splenocyte samples from wild-type and and wild-type from samples splenocyte total cells, CD45.1 naive later) or 8 day at analysis (for conjugated goat anti–mouse IgG (1036-05; SouthernBiotech) as described as SouthernBiotech) IgG (1036-05; (HRP)- anti–mouse goat conjugated peroxidase horseradish antibody secondary the and lysates LCMV assay. immunosorbent Enzyme-linked transfer. cell after 5 day at analysis by followed vehicle, or tamoxifen with treated subsequently were chimera, 5 × 10 × 5 chimera, slips (354085; BD Biosciences) on a 24-well plate (2 × 10 (2 plate on a 24-well Biosciences) BD (354085; slips Adoptive transfer. Adoptive LCMV.mice, with hosts the of recipient by infection followed into transferred were cells SMARTA transduced the Then, and 20 ng/ml Sigma-Aldrich) of Miltenyi (H9268; Biotec). IL-2 (130-098-221; tion (800 activated SMARTA cells were ‘spin-infected’ for 90 min at 37 °C by centrifuga H 1 cells from wild-type and and wild-type from cells 1 −/− Sorted T Sorted (CD45.2 + g ) mice, which were infected intravenously with 1 × 10 ) with freshly harvested retrovirus supernatants, 8 supernatants, retrovirus ) harvested with freshly 4 + 5 cells) through the use of biotin-conjugated antibodies (anti-CD8 + . Cells were sorted and then were transferred to 12-mm cover 12-mm to transferred were then and sorted were Cells . ) ) 1 d after of infection the host mice with LCMV, then the hosts H 6 + cells of a 4:6 mixture of of mixture 4:6 a of cells 1 or T or 1 ) mice and C57BL/6J wild-type (CD45.1 wild-type C57BL/6J and mice ) A total of 2 × 10 × 2 of total A FH cell function, 1 × 10 × 1 function, cell FH For comparison of gene expression in T in expression gene of comparison For + Supplementary Table 3 Supplementary ), followed by direct sorting into TRIzol LS reagent reagent LS TRIzol into sorting direct by followed ), ) ) were adoptively transferred into recipientwild-type cells underwent crosslinking for 10 min with 1% 1% with min 10 for crosslinking underwent cells − Tcf7 el wr te sand ih anti-CD4anti- with stained then were cells 4 Immunofluorescence staining was performed 6 . −/− + ) ) recipients. Recipient mice were allowed mice, the cells were sorted directly into into directly sorted were cells the mice, 6 (for analysis at day 2 or 3) or 2 × 10 × 2 or 3) or 2 day at analysis (for LCMV-specific IgG LCMV-specific was with titrated Tcf7 Bone marrow was collected from from collected was marrow Bone For isolation of T of isolation For 5 sorted T sorted + −/− SMARTA cells (or retrovirus- (or cells SMARTA bone marrow and C57BL/6J C57BL/6J and marrow bone ). Tcf7 FH −/− cells from i from cells 3 cells per well). The The well). per cells doi:10.1038/ni.3229 mice at day 8 after after 8 day at mice + µ ) mice. For each each For mice. ) 6 FH plaque-forming g/ml polybrene polybrene g/ml cells and T and cells FH Tcf7 cells and and cells 5 plaque- −/− or H 4 2 4 1 1 4 - - - . ,

© 2015 Nature America, Inc. All rights reserved. Technology) or rabbit IgG (A7016; Beyotime) coupled to Dynabeads Protein G doi:10.1038/ni.3229 with primers ( primers with performed was PCR Quantitative water. in eluted was and Qiagen) (28104; kit purification PCR a with purified was DNA Technologies). Life (10004D; muorcptto ad muolt analysis. immunoblot and Immunoprecipitation ChIP. each to added chromatin of to amount the of samples the for normalization and used was precipitation immuno before chromatin sheared of aliquot an was DNA input The sites. then were incubated overnight with 2 2 with overnight incubated were then Transfection TranIT-293 and with were 24 and, after extracted h, lysates the cell Mirus) 2705, (MIR Reagent cells 293T into transfected were constructs anti-Bcl-6 (D-8, Santa Cruz Biotechnology) or rabbit IgG (A7016, Beyotime) Beyotime) or IgG rabbit (A7016, Biotechnology) Cruz Santa (D-8, proteins, anti-Bcl-6 Cell endogenous (4681; of TLE4 immunoprecipitation and For TLE3 Technology). TLE2, Signaling TLE1, to antibody or Cruz Biotechnology) Santa (N-3; anti-Bcl6 Sigma-Aldrich), (M2; HRP–anti-Flag Abmart), immunopre the (19C2; times, HRP–anti-Myc with Life immunoblot by analyzed three were samples cipitated (10004D; washed G were Protein samples Dynabeads After with Technologies). incubation of h 2 by followed Supplementary Table 3 Supplementary ) flanking the putative TCF-1-binding TCF-1-binding putative the flanking ) µ g of anti-Flag (M2; Sigma-Aldrich), Sigma-Aldrich), (M2; anti-Flag of g The MIGR1 retroviral retroviral MIGR1 The - -

46. 45. 44. chimera experiments, a paired two-tailed two-tailed paired a experiments, chimera of for used calculation (GraphPad). An unpaired two-tailed two-tailed unpaired An (GraphPad). analysis. Statistical analysis Statistical was conducted with Prism 6.0 software for used were Biosciences) BD (K112-91; analysis. immunoblot anti-Bcl-6 and Technology) Signaling Cell (C46C7; anti-TCF-1 and immunoprecipitation, for used was an unpaired one-tailed one-tailed unpaired an of calculation for used was val P values. Subramanian, A. Subramanian, vaccination mucosal Efficient X. Zhu, & D.C. Y.,Roopenian, Bai, R., Zeng, L., Ye, M.A. Rasheed, mediated by the neonatal Fc receptor. Fc neonatal the by mediated virus-specific long-lived plasma cells. plasma long-lived virus-specific for interpreting genome-wide expression profiles. expression genome-wide interpreting for 15545–15550 (2005). 15545–15550 nelui-1 s ciia ctkn fr h gnrto of generation the for cytokine critical a is Interleukin-21 al. et Gene set enrichment analysis: a knowledge-based approach knowledge-based a analysis: enrichment set Gene al. et P t- values. For retroviral transduction and bone marrow Fortransduction values. retroviral test with 95% confidence interval for calculation of calculation for interval confidence 95% with test P values. For microarray analysis, we used used we analysis, microarray For values. t- J. Virol. J. Nat. Biotechnol. Nat. test with 95% confidence interval was was interval confidence 95% with test t- test with 95% confidence inter confidence 95% with test , 7737–7746 (2013). 7737–7746 87, Proc. Natl. Acad. Sci. USA Sci. Acad. Natl. Proc. nature immunology nature , 158–163 (2011). 158–163 29,

102, -

Supplementary Figure 1

TCF-1 expression positively correlates with signatures associated with the TFH lineage.

Flow cytometry analysis of TCF-1 expression and other markers related to TFH and TH1 differentiation in SMARTA cells of spleens from SMARTA-chimera mice at day 8 after LCMV infection. The numbers in quadrants indicate the percentage of cells in each. The data are representative of three independent experiments.

Nature Immunology: doi:10.1038/ni.3229

Nature Immunology: doi:10.1038/ni.3229 Supplementary Figure 2

TCF-1 is essential for TFH differentiation after acute viral infection.

(a) Expression of TCF-1 in CD4+ T cells in spleens from Tcf7–/– and control mice. (b) Flow cytometry of CD4+ T cells in spleens from Tcf7–/– and control mice at day 8 after LCMV infection (left) and quantification of GP66-tetramer-positive cells (right). The numbers above the outlined areas indicate the percentages of tetramer-positive cells (left). (c) Flow cytometry of tetramer-positive CD4+ T cells –/– in spleens from Tcf7 and control mice as in b. The numbers above the outlined areas indicate the percentages of TFH cells. (d) Flow cytometry of Foxp3– CD4+ T cells in spleens from Tcf7–/– and control mice (top). The numbers above the outlined areas indicate the hi + percentages of CD44 CXCR5 TFH cells. The graphs show the frequency (below, left) and number (below, right) of TFH cells. (e) Flow cytometry of Foxp3+ CD4+ T cells in spleens from Tcf7–/– and control mice (top). The numbers above the outlined areas indicate the + + percentages of Foxp3 CXCR5 TFR cells. The graphs show the frequency (below, left) and number (below, right) of TFR cells. (f) Flow –/– cytometry of TFH cells in spleens from Tcf7 and control mice as in d. The numbers in quadrants indicate the percentage of cells in –/– each. (g) Quantification of BrdU incorporation (left) and Ki-67 expression (right) in TFH cells from spleens of Tcf7 and control mice as in d. (h) Setup of BM chimera experiments. Donor BM cells from Tcf7–/– (CD45.2) and WT (CD45.1) mice were mixed (4:6), and transferred to irradiated WT (CD45.1) mice. After reconstitution, the recipient mice were infected with LCMV. (i) Setup of retrovirus- mediated TCF-1 knockdown in SMARTA cells. Retrovirus-transduced SMARTA cells (CD45.1) were adoptively transferred into WT mice (CD45.2), followed by LCMV infection. (j) Flow cytometry analysis of TCF-1 expression in transduced (GFP+) and non-transduced (GFP–) SMARTA cells at day 8 after infection. (k) Flow cytometry of GFP– and GFP+ SMARTA cells at day 8 after infection. The numbers adjacent to the outlined areas indicate the percentage of corresponding population. NS, not significant; *P < 0.05 and ***P < 0.001 (unpaired two-tailed t-test). The data are representative of two (c–g,j,k) or three (a,b) independent experiments with at least three mice per group (error bars (a,b,d,e,g), SEM).

Nature Immunology: doi:10.1038/ni.3229

Supplementary Figure 3

TCF-1 is required for TFH differentiation after infection with influenza A virus.

(a) Flow cytometry of Foxp3– CD4+ T cells in mediastinal lymph nodes of BM chimeras as in Supplementary Fig. 2h at day 8 after + influenza A virus infection, showing the CXCR5 TFH population (left) and a summary of the results (right). The numbers adjacent to the outlined areas indicate the percentages of TFH cells. (b) Quantification of MFIs of Bcl-6, ICOS and CXCR5 on TFH cells as in a. *P < 0.05 and **P < 0.01 (paired two-tailed t-test). The data are representative of two independent experiments with four mice per group (error bars, SEM).

Nature Immunology: doi:10.1038/ni.3229

Supplementary Figure 4

Ectopic expression of TCF-1 enhances TFH differentiation.

(a) Flow cytometry analysis of GFP– (non-transduced) and GFP+ (overexpressing p33 or p45) SMARTA cells in spleens at day 8 after lo + LCMV infection. The numbers adjacent to the outlined areas indicate the percentages of SLAM CXCR5 TFH cells. (b) Summary of TFH frequency and quantification of MFIs of CXCR5, ICOS and Bcl-6 on TFH cells expressing p33 or p45. (c) Experimental setup. Retroviruses overexpressing p33 were introduced into CD45.1+ SMARTA cells, and the cells were transferred into WT recipients + (CD45.2), which were subsequently infected with LCMV. At day 8 after infection, transduced GFP SMARTA TFH cells were sorted and transferred into day 1-infected recipients, followed by analysis at day 5 after cell transfer. (d) The graphs show a summary of the frequency and number of PNAhiFAShi GC B cells (top) gated on the B220+CD19+ population and CD138hiB220lo plasma cells (below) in spleens at day 5 after cell transfer as in c. *P < 0.05, **P < 0.01 and ***P < 0.001 (unpaired (d) or paired (b) two-tailed t-test).The presented data are representative of two (d) or three (a,b) independent experiments with at least three (a,b) or four (d) mice per group (error bar (d), SEM).

Nature Immunology: doi:10.1038/ni.3229

Supplementary Figure 5

Altered gene expression in TCF-1-deficient TFH cells.

–/– (a) Sorting strategy. (b–e) WT and Tcf7 TFH and TH1 cells were sorted on day 8 after LCMV infection, followed by mRNA extraction and RT-qPCR analysis. NS, not significant; *P < 0.05, **P < 0.01 and ***P < 0.001 (unpaired two-tailed t-test).The data are representative of two (b–e) independent experiments with two replicates from three mice per group (error bars (b–e), SEM).

Nature Immunology: doi:10.1038/ni.3229

Supplementary Figure 6

Alignment of the putative TCF-1-binding sites in the Bcl6 promoter and Prdm1 5′ regulatory regions.

The conserved TCF-1-binding motifs “CAAAG” (or “CTTTG” on the reverse strand) are highlighted in red, and their locations relative to the transcriptional starting site (TSS) of Bcl6 or Prdm1 are marked.

Nature Immunology: doi:10.1038/ni.3229

Supplementary Figure 7

TCF-1 physically interacts with Bcl-6 protein.

Immunoprecipitation of lysates of activated CD4+ T cells from infected mice at day 8 after infection with a Bcl-6 antibody or control IgG and immunoblot analysis with Bcl-6 and TCF-1 antibodies. The presented data are representative of two independent experiments.

Nature Immunology: doi:10.1038/ni.3229

Supplementary Figure 8

ICAT inhibits TFH differentiation.

Flow cytometry analysis of GFP– (non-transduced) and GFP+ (overexpressing ICAT) SMARTA cells in spleens at day 3 after LCMV lo + infection. The numbers adjacent to the outlined areas indicate the percentages of Tim3 CXCR5 TFH cells (left). The graph shows a summary of results (right). ***P < 0.001 (paired two-tailed t-test).The presented data are representative of three independent experiments with at least five mice per group.

Nature Immunology: doi:10.1038/ni.3229 –/– Supplementary Table 1. Differential gene expression between Tcf7 and WT TFH cells. –/– Fold Changes (Tcf7 /WT) Gene Symbol Gene Name 21.4 Fgl2 fibrinogen-like protein 2 15.6 AA467197 expressed sequence AA467197 13.8 Ccl4 chemokine (C-C motif) ligand 4 13.0 Chi3l3 chitinase 3-like 3 12.3 Ccr9 chemokine (C-C motif) receptor 9 10.5 Actn2 actinin alpha 2 10.4 Ccl1 chemokine (C-C motif) ligand 1 10.0 Ccr5 chemokine (C-C motif) receptor 5 8.8 Il10 interleukin 10 8.3 Gzmb granzyme B 7.0 1500009L16Rik RIKEN cDNA 1500009L16 gene 6.7 Ccr2 chemokine (C-C motif) receptor 2 5.6 Adam8 a disintegrin and metallopeptidase domain 8 5.5 Depdc1a DEP domain containing 1a 5.4 Nuf2 NUF2, NDC80 kinetochore complex component, homolog (S. cerevisiae) 5.3 Itga1 integrin alpha 1 5.2 Iglv1 immunoglobulin lambda variable 1 5.1 Igkv15-103 Immunoglobulin kappa chain variable 15-103 5.0 Cish cytokine inducible SH2-containing protein 4.8 Ifng interferon gamma 4.8 Cdc6 cell division cycle 6 4.7 Cdkn1a cyclin-dependent kinase inhibitor 1A (P21) 4.7 Tmem163 transmembrane protein 163 4.6 Gstt1 glutathione S-transferase, theta 1 4.5 Esco2 establishment of cohesion 1 homolog 2 (S. cerevisiae) 4.5 Coro2a coronin, actin binding protein 2A 4.5 Car1 carbonic anhydrase 1 4.5 Bub1 budding uninhibited by benzimidazoles 1 homolog (S. cerevisiae) 4.5 Dapk2 death-associated protein kinase 2 4.4 Cxcr6 chemokine (C-X-C motif) receptor 6 4.4 Aurkb aurora kinase B 4.3 Brca1 breast cancer 1 4.3 Lrr1 leucine rich repeat protein 1 4.3 Rrm2 ribonucleotide reductase M2 4.2 Ttk Ttk protein kinase 4.2 E2f8 E2F transcription factor 8 4.2 E2f7 E2F transcription factor 7 4.2 Rad51ap1 RAD51 associated protein 1 4.1 Cep55 centrosomal protein 55 4.1 Clspn claspin homolog (Xenopus laevis) 4.0 Ggt1 gamma-glutamyltransferase 1 4.0 4933413G19Rik RIKEN cDNA 4933413G19 gene 4.0 Xcl1 chemokine (C motif) ligand 1 4.0 Hemgn hemogen 4.0 Cd24a CD24a antigen 3.9 Cenpi centromere protein I 3.9 Mt2 metallothionein 2 3.9 Sytl2 synaptotagmin-like 2 3.9 Ciita class II transactivator 3.9 Ccna2 cyclin A2 3.9 Bhlhe40 basic helix-loop-helix family, member e40 3.9 Spc25 SPC25, NDC80 kinetochore complex component, homolog (S. cerevisiae) 3.9 Lamc1 laminin, gamma 1 3.9 6230424C14Rik RIKEN cDNA 6230424C14 gene 3.8 Birc5 baculoviral IAP repeat-containing 5 3.8 Plac8 placenta-specific 8 3.8 Gem GTP binding protein (gene overexpressed in skeletal muscle) 3.8 Dscc1 defective in sister chromatid cohesion 1 homolog (S. cerevisiae) 3.8 Ncf1 neutrophil cytosolic factor 1 3.8 Acoxl acyl-Coenzyme A oxidase-like 3.8 Fignl1 fidgetin-like 1 3.7 Exo1 exonuclease 1 3.7 Fasl Fas ligand (TNF superfamily, member 6) 3.7 C330027C09Rik RIKEN cDNA C330027C09 gene 3.7 Trip13 thyroid hormone receptor interactor 13 3.7 Ckap2 cytoskeleton associated protein 2 3.6 Klf1 Kruppel-like factor 1 (erythroid) 3.6 Serpinb9b serine (or cysteine) peptidase inhibitor, clade B, member 9b 3.6 Spag5 sperm associated antigen 5 3.6 Car2 carbonic anhydrase 2 3.6 Shcbp1 Shc SH2-domain binding protein 1 3.6 Chst11 carbohydrate sulfotransferase 11 3.6 2810417H13Rik RIKEN cDNA 2810417H13 gene 3.5 Prf1 perforin 1 (pore forming protein) 3.5 Cdc45 cell division cycle 45 3.5 Rad51 RAD51 homolog (S. cerevisiae) 3.5 Prdm1 PR domain containing 1, with ZNF domain 3.5 A630038E17Rik RIKEN cDNA A630038E17 gene 3.5 Nusap1 nucleolar and spindle associated protein 1 3.5 Klrb1f killer cell lectin-like receptor subfamily B member 1F 3.4 Tnfsf10 tumor necrosis factor (ligand) superfamily, member 10 3.4 Zwilch Zwilch, kinetochore associated, homolog (Drosophila) 3.4 Chek1 checkpoint kinase 1 3.4 Mcm10 minichromosome maintenance deficient 10 (S. cerevisiae) 3.4 Lztfl1 leucine zipper transcription factor-like 1 3.4 Depdc1b DEP domain containing 1B 3.4 Cenpw centromere protein W 3.3 Cenpk centromere protein K

Nature Immunology: doi:10.1038/ni.3229 3.3 Dennd5a DENN/MADD domain containing 5A 3.3 Cdca3 cell division cycle associated 3 3.3 Cpa3 carboxypeptidase A3, mast cell 3.3 Cdkn3 cyclin-dependent kinase inhibitor 3 3.3 Tacc3 transforming, acidic coiled-coil containing protein 3 3.3 Hmmr hyaluronan mediated motility receptor (RHAMM) 3.3 Ska1 spindle and kinetochore associated complex subunit 1 3.3 Ncaph non-SMC condensin I complex, subunit H 3.3 Ncapg2 non-SMC condensin II complex, subunit G2 3.2 Klrg1 killer cell lectin-like receptor subfamily G, member 1 3.2 Prc1 protein regulator of cytokinesis 1 3.2 Ccdc99 coiled-coil domain containing 99 3.2 Pbk PDZ binding kinase 3.2 Mis18bp1 MIS18 binding protein 1 3.2 Stil Scl/Tal1 interrupting locus 3.2 LOC641050 uncharacterized protein LOC641050 3.2 Tpx2 TPX2, microtubule-associated protein homolog (Xenopus laevis) 3.2 Tktl1 transketolase-like 1 3.2 Gins1 GINS complex subunit 1 (Psf1 homolog) 3.2 Kif18a kinesin family member 18A 3.2 Dtl denticleless homolog (Drosophila) 3.2 Ccl6 chemokine (C-C motif) ligand 6 3.2 Anln anillin, actin binding protein 3.2 Klra17 killer cell lectin-like receptor, subfamily A, member 17 3.2 Casc5 cancer susceptibility candidate 5 3.1 Chsy1 chondroitin sulfate synthase 1 3.1 Kif18b kinesin family member 18B 3.1 Cdk1 cyclin-dependent kinase 1 3.1 Ccnb2 cyclin B2 3.1 Asf1b ASF1 anti-silencing function 1 homolog B (S. cerevisiae) 3.1 Rad54l RAD54 like (S. cerevisiae) 3.1 Sgol2 shugoshin-like 2 (S. pombe) 3.1 Spc24 SPC24, NDC80 kinetochore complex component, homolog (S. cerevisiae) 3.1 Kif20a kinesin family member 20A 3.1 Kif2c kinesin family member 2C 3.0 Nedd4 neural precursor cell expressed, developmentally down-regulated 4 3.0 Cenpp centromere protein P 3.0 Cenpe centromere protein E 3.0 Mki67 antigen identified by monoclonal antibody Ki 67 3.0 Rgs18 regulator of G-protein signaling 18 3.0 Arsb arylsulfatase B 3.0 Uhrf1 ubiquitin-like, containing PHD and RING finger domains, 1 3.0 Slc4a1 solute carrier family 4 (anion exchanger), member 1 3.0 Kif11 kinesin family member 11 3.0 Csf1 colony stimulating factor 1 (macrophage) 3.0 Ncapg non-SMC condensin I complex, subunit G 3.0 Perp PERP, TP53 apoptosis effector 3.0 Sgol1 shugoshin-like 1 (S. pombe) 3.0 Cdca5 cell division cycle associated 5 3.0 Nek2 NIMA (never in mitosis gene a)-related expressed kinase 2 3.0 Cdca8 cell division cycle associated 8 3.0 2210011C24Rik RIKEN cDNA 2210011C24 gene 2.9 Fbxo5 F-box protein 5 2.9 Tmem176a transmembrane protein 176A 2.9 H2-Ab1 histocompatibility 2, class II antigen A, beta 1 2.9 Mcpt8 mast cell protease 8 2.9 Lmo2 LIM domain only 2 2.9 Glrx glutaredoxin 2.9 Cdc25c cell division cycle 25C 2.9 Cldn13 claudin 13 2.9 Mad2l1 MAD2 mitotic arrest deficient-like 1 2.9 Cit citron 2.9 Irak3 interleukin-1 receptor-associated kinase 3 2.9 Cenph centromere protein H 2.9 Foxp3 forkhead box P3 2.9 Kif4 kinesin family member 4 2.8 Gm11428 predicted gene 11428 2.8 Id2 inhibitor of DNA binding 2 2.8 Fam64a family with sequence similarity 64, member A 2.8 Ccnd1 cyclin D1 2.8 Tal1 T cell acute lymphocytic leukemia 1 2.8 Ccnf cyclin F 2.8 Ptgr1 prostaglandin reductase 1 2.8 Mybl2 myeloblastosis oncogene-like 2 2.8 Clic4 chloride intracellular channel 4 (mitochondrial) 2.8 Tk1 thymidine kinase 1 2.8 F2r coagulation factor II (thrombin) receptor 2.8 Rcn1 reticulocalbin 1 2.8 Ahsp alpha hemoglobin stabilizing protein 2.8 Padi2 peptidyl arginine deiminase, type II 2.8 Csrp3 cysteine and glycine-rich protein 3 2.8 Gata2 GATA binding protein 2 2.8 Top2a topoisomerase (DNA) II alpha 2.8 F13a1 coagulation factor XIII, A1 subunit 2.8 Ect2 ect2 oncogene 2.7 Atp8b4 ATPase, class I, type 8B, member 4 2.7 Fam54a family with sequence similarity 54, member A 2.7 Anxa1 annexin A1 2.7 Tcf19 transcription factor 19 2.7 Lgals3 lectin, galactose binding, soluble 3 2.7 Ndc80 NDC80 homolog, kinetochore complex component (S. cerevisiae)

Nature Immunology: doi:10.1038/ni.3229 2.7 4930547N16Rik RIKEN cDNA 4930547N16 gene 2.7 Igj immunoglobulin joining chain 2.7 H2-Aa histocompatibility 2, class II antigen A, alpha 2.6 Csf2rb colony stimulating factor 2 receptor, beta, low-affinity (granulocyte-macrophage) 2.6 Gm3579 predicted gene 3579 2.6 Gen1 Gen homolog 1, endonuclease (Drosophila) 2.6 Slc43a3 solute carrier family 43, member 3 2.6 Ms4a2 membrane-spanning 4-domains, subfamily A, member 2 2.6 Kctd12 potassium channel tetramerisation domain containing 12 2.6 Pmepa1 prostate transmembrane protein, androgen induced 1 2.6 Arnt2 aryl hydrocarbon receptor nuclear translocator 2 2.6 Beta-s hemoglobin, beta adult s chain 2.6 Gata1 GATA binding protein 1 2.6 Gmnn geminin 2.6 Ccnb1 cyclin B1 2.6 Sema4a sema domain, immunoglobulin domain (Ig), transmembrane domain (TM) and short cytoplasmic domain, (semaphorin) 4A 2.6 Plk1 polo-like kinase 1 2.6 4930579G24Rik RIKEN cDNA 4930579G24 gene 2.6 Napsa napsin A aspartic peptidase 2.5 4632434I11Rik RIKEN cDNA 4632434I11 gene 2.5 Ckap2l cytoskeleton associated protein 2-like 2.5 D17H6S56E-5 DNA segment, Chr 17, human D6S56E 5 2.5 Smc2 structural maintenance of chromosomes 2 2.5 Cks1b CDC28 protein kinase 1b 2.5 Cdc20 cell division cycle 20 2.5 Arhgap11a Rho GTPase activating protein 11A 2.5 Ppp3ca protein phosphatase 3, catalytic subunit, alpha isoform 2.5 Snx9 sorting nexin 9 2.5 Tipin timeless interacting protein 2.5 Plekhf1 pleckstrin homology domain containing, family F (with FYVE domain) member 1 2.5 AI661384 expressed sequence AI661384 2.5 Spry2 sprouty homolog 2 (Drosophila) 2.5 Kif23 kinesin family member 23 2.5 Cd7 CD7 antigen 2.5 Plek pleckstrin 2.5 Hist1h2ag histone cluster 1, H2ag 2.5 Atad5 ATPase family, AAA domain containing 5 2.5 Tmprss13 transmembrane protease, serine 13 2.4 Tm6sf1 transmembrane 6 superfamily member 1 2.4 Gas2l3 growth arrest-specific 2 like 3 2.4 Prr11 proline rich 11 2.4 Prim1 DNA primase, p49 subunit 2.4 Serpinb9 serine (or cysteine) peptidase inhibitor, clade B, member 9 2.4 Tspan33 tetraspanin 33 2.4 Mcm3 minichromosome maintenance deficient 3 (S. cerevisiae) 2.4 Tgfbi transforming growth factor, beta induced 2.4 BC028528 cDNA sequence BC028528 2.4 Stmn1 stathmin 1 2.4 Nr4a2 nuclear receptor subfamily 4, group A, member 2 2.4 Gstm5 glutathione S-transferase, mu 5 2.4 Tmcc2 transmembrane and coiled-coil domains 2 2.4 Ermap erythroblast membrane-associated protein 2.4 Rfc4 replication factor C (activator 1) 4 2.4 Orc6 origin recognition complex, subunit 6 2.4 Mcm5 minichromosome maintenance deficient 5, cell division cycle 46 (S. cerevisiae) 2.3 Klrk1 killer cell lectin-like receptor subfamily K, member 1 2.3 Aqp1 aquaporin 1 2.3 Melk maternal embryonic leucine zipper kinase 2.3 Apitd1 apoptosis-inducing, TAF9-like domain 1 2.3 Tyms thymidylate synthase 2.3 Fabp5 fatty acid binding protein 5, epidermal 2.3 Apobr apolipoprotein B receptor 2.3 Ercc6l excision repair cross-complementing rodent repair deficiency complementation group 6 - like 2.3 Kntc1 kinetochore associated 1 2.3 Gda guanine deaminase 2.3 Alox5ap arachidonate 5-lipoxygenase activating protein 2.3 Rrm1 ribonucleotide reductase M1 2.3 Obfc2a oligonucleotide/oligosaccharide-binding fold containing 2A 2.2 Ehd1 EH-domain containing 1 2.2 Plk4 polo-like kinase 4 2.2 Il1rl1 interleukin 1 receptor-like 1 2.2 1700025G04Rik RIKEN cDNA 1700025G04 gene 2.2 Mt1 metallothionein 1 2.2 Tmem176b transmembrane protein 176B 2.2 Rfc3 replication factor C (activator 1) 3 2.2 Fam72a family with sequence similarity 72, member A 2.2 Kif20b kinesin family member 20B 2.2 Acpl2 acid phosphatase-like 2 2.2 Socs2 suppressor of cytokine signaling 2 2.2 Cpd carboxypeptidase D 2.2 Slc2a3 solute carrier family 2 (facilitated glucose transporter), member 3 2.2 Furin furin (paired basic amino acid cleaving enzyme) 2.2 Slamf1 signaling lymphocytic activation molecule family member 1 2.2 6330503K22Rik RIKEN cDNA 6330503K22 gene 2.2 Slamf7 SLAM family member 7 2.2 Ccl9 chemokine (C-C motif) ligand 9 2.2 Itgb3 integrin beta 3 2.2 Hdc histidine decarboxylase 2.2 Ccnd2 cyclin D2 2.2 Dna2 DNA replication helicase 2 homolog (yeast) 2.2 Lair1 leukocyte-associated Ig-like receptor 1

Nature Immunology: doi:10.1038/ni.3229 2.2 Klrc1 killer cell lectin-like receptor subfamily C, member 1 2.2 Pygl liver glycogen phosphorylase 2.2 Slc16a6 solute carrier family 16 (monocarboxylic acid transporters), member 6 2.2 Pif1 PIF1 5'-to-3' DNA helicase homolog (S. cerevisiae) 2.2 Scpep1 serine carboxypeptidase 1 2.2 Lig1 ligase I, DNA, ATP-dependent 2.1 Eif4e3 eukaryotic translation initiation factor 4E member 3 2.1 Lmnb1 lamin B1 2.1 Acsbg1 acyl-CoA synthetase bubblegum family member 1 2.1 Errfi1 ERBB receptor feedback inhibitor 1 2.1 Vcl vinculin 2.1 Kcnk5 potassium channel, subfamily K, member 5 2.1 Poc1a POC1 centriolar protein homolog A (Chlamydomonas) 2.1 Ifngr1 interferon gamma receptor 1 2.1 Fen1 flap structure specific endonuclease 1 2.1 Ly6c1 lymphocyte antigen 6 complex, locus C1 2.1 Gzmk granzyme K 2.1 Ccdc34 coiled-coil domain containing 34 2.1 Kif22 kinesin family member 22 2.1 Itga2b integrin alpha 2b 2.1 Cenpn centromere protein N 2.1 Cables1 CDK5 and Abl enzyme substrate 1 2.1 Ighg3 Immunoglobulin heavy constant gamma 3 2.1 Esm1 endothelial cell-specific molecule 1 2.1 Gpc1 glypican 1 2.1 Gfi1b growth factor independent 1B 2.1 Sfpi1 SFFV proviral integration 1 2.1 Zfp367 zinc finger protein 367 2.1 Fam82a1 family with sequence similarity 82, member A1 2.1 Syce2 synaptonemal complex central element protein 2 2.1 BC005685 cDNA sequence BC005685 2.1 Cmklr1 chemokine-like receptor 1 2.1 Psrc1 proline/serine-rich coiled-coil 1 2.1 Racgap1 Rac GTPase-activating protein 1 2.1 Incenp inner centromere protein 2.1 Miat myocardial infarction associated transcript (non-protein coding) 2.1 Snai3 snail homolog 3 (Drosophila) 2.1 Psmc3ip proteasome (prosome, macropain) 26S subunit, ATPase 3, interacting protein 2.1 Nmral1 NmrA-like family domain containing 1 2.1 E330016A19Rik RIKEN cDNA E330016A19 gene 2.1 Sass6 spindle assembly 6 homolog (C. elegans) 2.1 2810408I11Rik RIKEN cDNA 2810408I11 gene 2.0 Gins2 GINS complex subunit 2 (Psf2 homolog) 2.0 Nkg7 natural killer cell group 7 sequence 2.0 Pkig protein kinase inhibitor, gamma 2.0 Hist2h3c1 histone cluster 2, H3c1 2.0 Gpr68 G protein-coupled receptor 68 2.0 Rap1gap2 RAP1 GTPase activating protein 2 2.0 Rras2 related RAS viral (r-ras) oncogene homolog 2 2.0 Eme1 essential meiotic endonuclease 1 homolog 1 (S. pombe) 2.0 Tmem107 transmembrane protein 107 2.0 Cdca2 cell division cycle associated 2 2.0 Serpinb6b serine (or cysteine) peptidase inhibitor, clade B, member 6b 2.0 Chn2 chimerin (chimaerin) 2 2.0 Dusp5 dual specificity phosphatase 5 2.0 Mcm6 minichromosome maintenance deficient 6 (MIS5 homolog, S. pombe) (S. cerevisiae) 2.0 Cdkn2c cyclin-dependent kinase inhibitor 2C (p18, inhibits CDK4) 2.0 Serpini1 serine (or cysteine) peptidase inhibitor, clade I, member 1 2.0 Chaf1b chromatin assembly factor 1, subunit B (p60) 2.0 Casp7 caspase 7 2.0 Hmbs hydroxymethylbilane synthase 2.0 Zcchc11 zinc finger, CCHC domain containing 11 2.0 Hba-a1 hemoglobin alpha, adult chain 1 2.0 Cdt1 chromatin licensing and DNA replication factor 1 2.0 Ctsw cathepsin W 2.0 Fanca Fanconi anemia, complementation group A 2.0 Mxd3 Max dimerization protein 3 2.0 Baiap3 BAI1-associated protein 3 2.0 BC030867 cDNA sequence BC030867 2.0 Vim vimentin 2.0 Fkbp2 FK506 binding protein 2 2.0 Plp2 proteolipid protein 2 2.0 Nrm nurim (nuclear envelope membrane protein) 2.0 Il12rb1 interleukin 12 receptor, beta 1 1.9 Laptm4b lysosomal-associated protein transmembrane 4B 1.9 Kcnk6 potassium inwardly-rectifying channel, subfamily K, member 6 1.9 Csrp1 cysteine and glycine-rich protein 1 1.9 Ffar2 free fatty acid receptor 2 1.9 Tmem48 transmembrane protein 48 1.9 Gsg2 germ cell-specific gene 2 1.9 Mcm7 minichromosome maintenance deficient 7 (S. cerevisiae) 1.9 Mxra8 matrix-remodelling associated 8 1.9 C3ar1 complement component 3a receptor 1 1.9 2010002N04Rik RIKEN cDNA 2010002N04 gene 1.9 Plcg2 phospholipase C, gamma 2 1.9 Rfc5 replication factor C (activator 1) 5 1.9 D2Ertd750e DNA segment, Chr 2, ERATO Doi 750, expressed 1.9 Asb2 ankyrin repeat and SOCS box-containing 2 1.9 Brca2 breast cancer 2 1.9 Hmgn2 high mobility group nucleosomal binding domain 2 1.9 Mcm4 minichromosome maintenance deficient 4 homolog (S. cerevisiae)

Nature Immunology: doi:10.1038/ni.3229 1.9 Hist1h2bc histone cluster 1, H2bc 1.9 Cks2 CDC28 protein kinase regulatory subunit 2 1.9 Hip1 huntingtin interacting protein 1 1.8 Timeless timeless homolog (Drosophila) 1.8 Kpna2 karyopherin (importin) alpha 2 1.8 Eef2k eukaryotic elongation factor-2 kinase 1.8 2010111I01Rik RIKEN cDNA 2010111I01 gene 1.8 Ugcg UDP-glucose ceramide glucosyltransferase 1.8 Fgr Gardner-Rasheed feline sarcoma viral (Fgr) oncogene homolog 1.8 Glipr2 GLI pathogenesis-related 2 1.8 F2rl2 coagulation factor II (thrombin) receptor-like 2 1.8 Dnmt1 DNA methyltransferase (cytosine-5) 1 1.8 Sepn1 selenoprotein N, 1 1.8 Mgat4a mannoside acetylglucosaminyltransferase 4, isoenzyme A 1.8 Dennd4a DENN/MADD domain containing 4A 1.8 Serpinb1a serine (or cysteine) peptidase inhibitor, clade B, member 1a 1.8 Tbx21 T-box 21 1.8 H2afx H2A histone family, member X 1.8 Rom1 rod outer segment membrane protein 1 1.8 Alox5 arachidonate 5-lipoxygenase 1.8 Dsn1 DSN1, MIND kinetochore complex component, homolog (S. cerevisiae) 1.8 Pafah2 platelet-activating factor acetylhydrolase 2 1.8 Dut deoxyuridine triphosphatase 1.8 Hat1 histone aminotransferase 1 1.8 Ccne1 cyclin E1 1.8 Nhsl2 NHS-like 2 1.8 E2f1 E2F transcription factor 1 1.8 Runx2 runt related transcription factor 2 1.8 Ncapd2 non-SMC condensin I complex, subunit D2 1.8 Prim2 DNA primase, p58 subunit 1.8 Cenpl centromere protein L 1.8 Selenbp1 selenium binding protein 1 1.8 Vopp1 vesicular, overexpressed in cancer, prosurvival protein 1 1.8 Mlf1ip myeloid leukemia factor 1 interacting protein 1.7 4831426I19Rik RIKEN cDNA 4831426I19 gene 1.7 Fes feline sarcoma oncogene 1.7 Ncapd3 non-SMC condensin II complex, subunit D3 1.7 Gpx7 glutathione peroxidase 7 1.7 Slbp stem-loop binding protein 1.7 L1cam L1 cell adhesion molecule 1.7 Zyx zyxin 1.7 Cdc7 cell division cycle 7 (S. cerevisiae) 1.7 Trex1 three prime repair exonuclease 1 1.7 Dpy19l1 dpy-19-like 1 (C. elegans) 1.7 Pdss1 prenyl (solanesyl) diphosphate synthase, subunit 1 1.7 Cenpa centromere protein A 1.7 Pak1 p21 protein (Cdc42/Rac)-activated kinase 1 1.7 Aim1 absent in melanoma 1 1.7 Smpdl3b sphingomyelin phosphodiesterase, acid-like 3B 1.7 Xdh xanthine dehydrogenase 1.7 Rap2a RAS related protein 2a 1.7 Atad2 ATPase family, AAA domain containing 2 1.7 Blm Bloom syndrome, RecQ helicase-like 1.7 Rnaseh2b ribonuclease H2, subunit B 1.7 Rrbp1 ribosome binding protein 1 1.7 Tuba3a tubulin, alpha 3A 1.7 Tnfrsf9 tumor necrosis factor receptor superfamily, member 9 1.7 Zbtb8a zinc finger and BTB domain containing 8a 1.7 Serinc3 serine incorporator 3 1.7 2610020H08Rik RIKEN cDNA 2610020H08 gene 1.7 Tpst1 protein-tyrosine sulfotransferase 1 1.7 Cbx5 chromobox 5 1.7 Samhd1 SAM domain and HD domain, 1 1.7 Fhl2 four and a half LIM domains 2 1.7 H2afz H2A histone family, member Z 1.7 Pmf1 polyamine-modulated factor 1 1.7 Lrrk1 leucine-rich repeat kinase 1 1.7 Cd226 CD226 antigen 1.7 Ptgir prostaglandin I receptor (IP) 1.7 Rhd Rh blood group, D antigen 1.7 Fcgr2b Fc receptor, IgG, low affinity Iib 1.7 Tfrc transferrin receptor 1.7 Insl6 insulin-like 6 1.7 2610318N02Rik RIKEN cDNA 2610318N02 gene 1.7 Tpm4 tropomyosin 4 1.7 Plekhg3 pleckstrin homology domain containing, family G (with RhoGef domain) member 3 1.7 Rad21 RAD21 homolog (S. pombe) 1.7 Hmgb3 high mobility group box 3 1.7 Cd72 CD72 antigen 1.7 Ung uracil DNA glycosylase 1.7 Pola1 polymerase (DNA directed), alpha 1 1.7 Hmgb2 high mobility group box 2 1.7 Gtse1 G two S phase expressed protein 1 1.7 AU015263 expressed sequence AU015263 1.7 Wdhd1 WD repeat and HMG-box DNA binding protein 1 1.6 Sept11 septin 11 1.6 Cd48 CD48 antigen 1.6 Gm14005 predicted gene 14005 1.6 Plxnd1 plexin D1 1.6 Lyl1 lymphoblastomic leukemia 1 1.6 Mms22l MMS22-like, DNA repair protein

Nature Immunology: doi:10.1038/ni.3229 1.6 Dlgap5 discs, large (Drosophila) homolog-associated protein 5 1.6 H2-Oa histocompatibility 2, O region alpha locus 1.6 Arl6ip1 ADP-ribosylation factor-like 6 interacting protein 1 1.6 Topbp1 topoisomerase (DNA) II binding protein 1 1.6 Fam174b family with sequence similarity 174, member B 1.6 Ccl5 chemokine (C-C motif) ligand 5 1.6 Cchcr1 coiled-coil alpha-helical rod protein 1 1.6 N4bp1 NEDD4 binding protein 1 1.6 Chek2 checkpoint kinase 2 1.6 Lrrc40 leucine rich repeat containing 40 1.6 Evi5 ecotropic viral integration site 5 1.6 Nr4a3 nuclear receptor subfamily 4, group A, member 3 1.6 Pold1 polymerase (DNA directed), delta 1, catalytic subunit 1.6 Wdr67 WD repeat domain 67 1.6 Mcm2 minichromosome maintenance deficient 2 mitotin (S. cerevisiae) 1.6 E2f2 E2F transcription factor 2 1.6 Nrn1 neuritin 1 1.6 Xrcc2 X-ray repair complementing defective repair in Chinese hamster cells 2 1.6 Tmpo thymopoietin 1.6 Abhd15 abhydrolase domain containing 15 1.6 Hirip3 HIRA interacting protein 3 1.6 Ankrd32 ankyrin repeat domain 32 1.6 Pqlc3 PQ loop repeat containing 1.6 Rbbp8 retinoblastoma binding protein 8 1.6 Bik BCL2-interacting killer 1.6 Cdk2 cyclin-dependent kinase 2 1.6 Ier3 immediate early response 3 1.6 Ncald neurocalcin delta 1.6 Tuba1c tubulin, alpha 1C 1.6 Kifc1 kinesin family member C1 1.6 Casp3 caspase 3 1.6 Fanci Fanconi anemia, complementation group I 1.6 Gadd45g growth arrest and DNA-damage-inducible 45 gamma 1.6 Ppfia4 protein tyrosine phosphatase, receptor type, f polypeptide (PTPRF), interacting protein (liprin), alpha 4 1.6 Alyref Aly/REF export factor 1.6 Paqr4 progestin and adipoQ receptor family member IV 1.5 Muc20 mucin 20 1.5 Acot10 acyl-CoA thioesterase 10 1.5 Cd244 CD244 natural killer cell receptor 2B4 1.5 Gemin6 gem (nuclear organelle) associated protein 6 1.5 Rassf4 Ras association (RalGDS/AF-6) domain family member 4 1.5 Ankrd49 ankyrin repeat domain 49 1.5 Cyp4f18 cytochrome P450, family 4, subfamily f, polypeptide 18 1.5 Ust uronyl-2-sulfotransferase 1.5 Pak4 p21 protein (Cdc42/Rac)-activated kinase 4 1.5 Tesk2 testis-specific kinase 2 1.5 Pcna proliferating cell nuclear antigen 1.5 A130009E19Rik RIKEN cDNA A130009E19 gene 1.5 Trp53i11 transformation related protein 53 inducible protein 11 1.5 Rrad Ras-related associated with diabetes 1.5 2310061C15Rik RIKEN cDNA 2310061C15 gene 1.5 Siva1 SIVA1, apoptosis-inducing factor 1.5 H2-Ea-ps histocompatibility 2, class II antigen E alpha, pseudogene 1.5 4930524J08Rik RIKEN cDNA 4930524J08 gene 1.5 Cd47 CD47 antigen (Rh-related antigen, integrin-associated signal transducer) 1.5 Hells helicase, lymphoid specific 1.5 Jmjd5 jumonji domain containing 5 1.5 Chst15 carbohydrate (N-acetylgalactosamine 4-sulfate 6-O) sulfotransferase 15 1.5 St14 suppression of tumorigenicity 14 (colon carcinoma) 1.5 Slfn9 schlafen 9 1.5 Cpox coproporphyrinogen oxidase -13.2 Sostdc1 sclerostin domain containing 1 -11.0 Nsg2 neuron specific gene family member 2 -9.2 Itga7 integrin alpha 7 -9.2 Fcrl1 Fc receptor-like 1 -8.9 Ces2d-ps carboxylesterase 2D, pseudogene -8.9 Fam101b family with sequence similarity 101, member B -7.8 Id3 inhibitor of DNA binding 3 -7.3 Plagl1 pleiomorphic adenoma gene-like 1 -6.7 Ccr7 chemokine (C-C motif) receptor 7 -5.9 Snn stannin -5.7 Gdpd3 glycerophosphodiester phosphodiesterase domain containing 3 -5.5 Padi4 peptidyl arginine deiminase, type IV -5.3 Ascl2 achaete-scute complex homolog 2 (Drosophila) -5.2 Egln3 EGL nine homolog 3 (C. elegans) -5.1 H2-Ob histocompatibility 2, O region beta locus -4.9 Tox2 TOX high mobility group box family member 2 -4.7 Pou2af1 POU domain, class 2, associating factor 1 -4.6 2610019F03Rik RIKEN cDNA 2610019F03 gene -4.5 Acpp acid phosphatase, prostate -4.3 Bmp7 bone morphogenetic protein 7 -4.3 Ptrh1 peptidyl-tRNA hydrolase 1 homolog (S. cerevisiae) -4.3 Cd22 CD22 antigen -4.2 F2rl1 coagulation factor II (thrombin) receptor-like 1 -4.0 Gpm6b glycoprotein m6b -4.0 Asap1 ArfGAP with SH3 domain, ankyrin repeat and PH domain1 -3.9 Aff3 AF4/FMR2 family, member 3 -3.9 St6gal1 beta galactoside alpha 2,6 sialyltransferase 1 -3.9 Rpl39l ribosomal protein L39-like -3.9 Coro2b coronin, actin binding protein, 2B -3.9 Dapl1 death associated protein-like 1 -3.7 Ext1 exostoses (multiple) 1

Nature Immunology: doi:10.1038/ni.3229 -3.7 Il6st interleukin 6 signal transducer -3.6 Folr4 folate receptor 4 (delta) -3.5 Ephx1 epoxide hydrolase 1, microsomal -3.5 Angptl2 angiopoietin-like 2 -3.4 Slc5a3 solute carrier family 5 (inositol transporters), member 3 -3.4 Prickle1 prickle homolog 1 (Drosophila) -3.3 Arxes1 adipocyte-related X-chromosome expressed sequence 1 -3.3 Galnt9 UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase 9 -3.3 Matk megakaryocyte-associated tyrosine kinase -3.3 Cxcr5 chemokine (C-X-C motif) receptor 5 -3.2 Vipr1 vasoactive intestinal peptide receptor 1 -3.2 Ispd isoprenoid synthase domain containing -3.2 Plekha8 pleckstrin homology domain containing, family A (phosphoinositide binding specific) member 8 -3.2 Cd200 CD200 antigen -3.1 P2rx7 purinergic receptor P2X, ligand-gated ion channel, 7 -3.1 BC017612 cDNA sequence BC017612 -3.0 Axin2 axin2 -3.0 Il6ra interleukin 6 receptor, alpha -3.0 Ar androgen receptor -2.9 2310022B05Rik RIKEN cDNA 2310022B05 gene -2.9 Adk adenosine kinase -2.9 Pfn2 profilin 2 -2.9 Sox4 SRY-box containing gene 4 -2.9 Fam43a family with sequence similarity 43, member A -2.8 Zfp318 zinc finger protein 318 -2.8 Fam46c family with sequence similarity 46, member C -2.8 Ccdc126 coiled-coil domain containing 126 -2.8 Il4 interleukin 4 -2.8 Tnfrsf13b tumor necrosis factor receptor superfamily, member 13b -2.8 Dennd2d DENN/MADD domain containing 2D -2.8 Tm7sf2 transmembrane 7 superfamily member 2 -2.8 Prkcz protein kinase C, zeta -2.8 Parp3 poly (ADP-ribose) polymerase family, member 3 -2.7 D8Ertd82e DNA segment, Chr 8, ERATO Doi 82, expressed -2.7 Rnf167 ring finger protein 167 -2.7 Arhgap5 Rho GTPase activating protein 5 -2.6 Ikbke inhibitor of kappaB kinase epsilon -2.6 Slamf6 SLAM family member 6 -2.6 Armcx2 armadillo repeat containing, X-linked 2 -2.6 Btla B and T lymphocyte associated -2.6 Rapgef5 Rap guanine nucleotide exchange factor (GEF) 5 -2.6 D4Wsu53e DNA segment, Chr 4, Wayne State University 53, expressed -2.5 1110034G24Rik RIKEN cDNA 1110034G24 gene -2.5 Rnf19a ring finger protein 19A -2.5 Gpr183 G protein-coupled receptor 183 -2.5 Cnga1 cyclic nucleotide gated channel alpha 1 -2.5 Bphl biphenyl hydrolase-like (serine hydrolase, breast epithelial mucin-associated antigen) -2.5 Gramd4 GRAM domain containing 4 -2.5 Tspan32 tetraspanin 32 -2.4 Bcl6 B cell leukemia/lymphoma 6 -2.4 Rnf128 ring finger protein 128 -2.4 Pdlim1 PDZ and LIM domain 1 (elfin) -2.4 Psen2 presenilin 2 -2.4 Macrod1 MACRO domain containing 1 -2.4 Ddit4 DNA-damage-inducible transcript 4 -2.4 Msrb2 methionine sulfoxide reductase B2 -2.4 Pgap1 post-GPI attachment to proteins 1 -2.4 Tespa1 thymocyte expressed, positive selection associated 1 -2.4 Cxcr4 chemokine (C-X-C motif) receptor 4 -2.4 Tnfsf8 tumor necrosis factor (ligand) superfamily, member 8 -2.3 2010004M13Rik RIKEN cDNA 2010004M13 gene -2.3 BC021614 cDNA sequence BC021614 -2.3 Abhd8 abhydrolase domain containing 8 -2.3 Setd4 SET domain containing 4 -2.3 Fam134b family with sequence similarity 134, member B -2.3 Bckdhb branched chain ketoacid dehydrogenase E1, beta polypeptide -2.3 Dsel dermatan sulfate epimerase-like -2.3 Jmy junction-mediating and regulatory protein -2.3 Tnfsf11 tumor necrosis factor (ligand) superfamily, member 11 -2.3 Cdk5r1 cyclin-dependent kinase 5, regulatory subunit 1 (p35) -2.3 Dnahc8 dynein, axonemal, heavy chain 8 -2.3 6330403K07Rik RIKEN cDNA 6330403K07 gene -2.2 Fam69b family with sequence similarity 69, member B -2.2 1700019D03Rik RIKEN cDNA 1700019D03 gene -2.2 Polg polymerase (DNA directed), gamma -2.2 Pacsin1 protein kinase C and casein kinase substrate in neurons 1 -2.2 1700021C14Rik RIKEN cDNA 1700021C14 gene -2.2 Cbr1 carbonyl reductase 1 -2.2 Mgarp mitochondria localized glutamic acid rich protein -2.2 Enc1 ectodermal-neural cortex 1 -2.2 Ccdc91 coiled-coil domain containing 91 -2.2 Dennd4c DENN/MADD domain containing 4C -2.2 E230008J23Rik RIKEN cDNA E230008J23 gene -2.2 C78339 expressed sequence C78339 -2.2 Cd9 CD9 antigen -2.2 Arhgef4 Rho guanine nucleotide exchange factor (GEF) 4 -2.2 Adck3 aarF domain containing kinase 3 -2.2 Trib2 tribbles homolog 2 (Drosophila) -2.2 Dguok deoxyguanosine kinase -2.2 2310001H17Rik RIKEN cDNA 2310001H17 gene -2.2 Inpp5f inositol polyphosphate-5-phosphatase F -2.2 Dus2l dihydrouridine synthase 2-like (SMM1, S. cerevisiae) -2.2 Kdm5b lysine (K)-specific demethylase 5B -2.1 2610035D17Rik RIKEN cDNA 2610035D17 gene -2.1 Hdac10 histone deacetylase 10

Nature Immunology: doi:10.1038/ni.3229 -2.1 Smyd4 SET and MYND domain containing 4 -2.1 Cd320 CD320 antigen -2.1 Decr2 2-4-dienoyl-Coenzyme A reductase 2, peroxisomal -2.1 Metap1d methionyl aminopeptidase type 1D (mitochondrial) -2.1 Zbtb32 zinc finger and BTB domain containing 32 -2.1 Mgst2 microsomal glutathione S-transferase 2 -2.1 6330509M05Rik RIKEN cDNA 6330509M05 gene -2.1 Immp2l IMP2 inner mitochondrial membrane peptidase-like (S. cerevisiae) -2.1 Thumpd2 THUMP domain containing 2 -2.1 Ccdc28b coiled coil domain containing 28B -2.1 Fgfr1op Fgfr1 oncogene partner -2.1 Ldlrap1 low density lipoprotein receptor adaptor protein 1 -2.1 Klc3 kinesin light chain 3 -2.1 Bbs2 Bardet-Biedl syndrome 2 (human) -2.1 Rassf2 Ras association (RalGDS/AF-6) domain family member 2 -2.1 Gpr155 G protein-coupled receptor 155 -2.0 E230032D23Rik RIKEN cDNA E230032D23 gene -2.0 Pura purine rich element binding protein A -2.0 Tesc tescalcin -2.0 St3gal1 ST3 beta-galactoside alpha-2,3-sialyltransferase 1 -2.0 Nudt14 nudix (nucleoside diphosphate linked moiety X)-type motif 14 -2.0 Smad5 SMAD family member 5 -2.0 Zkscan14 zinc finger with KRAB and SCAN domains 14 -2.0 Fbxo32 F-box protein 32 -2.0 Pptc7 PTC7 protein phosphatase homolog (S. cerevisiae) -2.0 Ankrd28 ankyrin repeat domain 28 -2.0 Mrps6 mitochondrial ribosomal protein S6 -2.0 Lpp LIM domain containing preferred translocation partner in lipoma -2.0 Zhx2 zinc fingers and homeoboxes 2 -2.0 Ryk receptor-like tyrosine kinase -2.0 Slc26a11 solute carrier family 26, member 11 -2.0 Tcp11l2 t-complex 11 (mouse) like 2 -2.0 A2ld1 AIG2-like domain 1 -2.0 Ccs copper chaperone for superoxide dismutase -2.0 Mccc2 methylcrotonoyl-Coenzyme A carboxylase 2 (beta) -2.0 Bag3 BCL2-associated athanogene 3 -2.0 Iffo2 intermediate filament family orphan 2 -2.0 Tnfsf14 tumor necrosis factor (ligand) superfamily, member 14 -2.0 Faah fatty acid amide hydrolase -2.0 Mboat1 membrane bound O-acyltransferase domain containing 1 -2.0 Clybl citrate lyase beta like -2.0 Hsdl1 hydroxysteroid dehydrogenase like 1 -2.0 Trim8 tripartite motif-containing 8 -2.0 Gstt2 glutathione S-transferase, theta 2 -2.0 Krt10 keratin 10 -2.0 Cbx7 chromobox 7 -2.0 Isyna1 myo-inositol 1-phosphate synthase A1 -2.0 Slc41a1 solute carrier family 41, member 1 -2.0 Taf4b TAF4B RNA polymerase II, TATA box binding protein (TBP)-associated factor -2.0 Traf5 TNF receptor-associated factor 5 -2.0 Tnfrsf25 tumor necrosis factor receptor superfamily, member 25 -2.0 Bzw2 basic leucine zipper and W2 domains 2 -1.9 Trbv13-2 T cell receptor beta, variable 13-2 -1.9 Lnx2 ligand of numb-protein X 2 -1.9 Tnfrsf22 tumor necrosis factor receptor superfamily, member 22 -1.9 0610011F06Rik RIKEN cDNA 0610011F06 gene -1.9 Mkl2 MKL/myocardin-like 2 -1.9 2310058N22Rik RIKEN cDNA 2310058N22 gene -1.9 Usp33 ubiquitin specific peptidase 33 -1.9 Nbn nibrin -1.9 Kcnn4 potassium intermediate/small conductance calcium-activated channel, subfamily N, member 4 -1.9 Fam82b family with sequence similarity 82, member B -1.9 Top1mt DNA topoisomerase 1, mitochondrial -1.9 B230354K17Rik RIKEN cDNA B230354K17 gene -1.9 Tspan5 tetraspanin 5 -1.9 Ppp4r2 protein phosphatase 4, regulatory subunit 2 -1.9 Hivep2 human immunodeficiency virus type I enhancer binding protein 2 -1.9 Tmco6 transmembrane and coiled-coil domains 6 -1.9 Cenpv centromere protein V -1.9 Hook1 hook homolog 1 (Drosophila) -1.9 Rab37 RAB37, member of RAS oncogene family -1.9 Pex26 peroxisomal biogenesis factor 26 -1.9 Usp6nl USP6 N-terminal like -1.9 Zfp827 zinc finger protein 827 -1.9 Fam71b family with sequence similarity 71, member B -1.9 Grhpr glyoxylate reductase/hydroxypyruvate reductase -1.9 Ascc1 activating signal cointegrator 1 complex subunit 1 -1.9 Fbxo4 F-box protein 4 -1.9 Sgsh N-sulfoglucosamine sulfohydrolase (sulfamidase) -1.9 Dap3 death associated protein 3 -1.9 Mxd4 Max dimerization protein 4 -1.9 Bpgm 2,3-bisphosphoglycerate mutase -1.9 Dleu2 deleted in lymphocytic leukemia, 2 -1.9 Prmt2 protein arginine N-methyltransferase 2 -1.9 Cdc14b CDC14 cell division cycle 14B -1.9 Tmem62 transmembrane protein 62 -1.9 Thtpa thiamine triphosphatase -1.9 Fam118a family with sequence similarity 118, member A -1.9 Dusp12 dual specificity phosphatase 12 -1.9 Wbscr27 Williams Beuren syndrome chromosome region 27 (human) -1.9 Lta lymphotoxin A -1.9 U2af1l4 U2 small nuclear RNA auxiliary factor 1-like 4 -1.9 Atp9a ATPase, class II, type 9A -1.8 Thada thyroid adenoma associated -1.8 Dgka diacylglycerol kinase, alpha

Nature Immunology: doi:10.1038/ni.3229 -1.8 Nrip1 nuclear receptor interacting protein 1 -1.8 Tagap1 T cell activation GTPase activating protein 1 -1.8 Zfp566 zinc finger protein 566 -1.8 Psmg1 proteasome (prosome, macropain) assembly chaperone 1 -1.8 Mcee methylmalonyl CoA epimerase -1.8 Taf1c TATA box binding protein (Tbp)-associated factor, RNA polymerase I, C -1.8 Zfp623 zinc finger protein 623 -1.8 Grtp1 GH regulated TBC protein 1 -1.8 Pdk1 pyruvate dehydrogenase kinase, isoenzyme 1 -1.8 Camsap2 calmodulin regulated spectrin-associated protein family, member 2 -1.8 Grcc10 gene rich cluster, C10 gene -1.8 Rere arginine glutamic acid dipeptide (RE) repeats -1.8 Zfp560 zinc finger protein 560 -1.8 Zfp157 zinc finger protein 157 -1.8 Kbtbd11 kelch repeat and BTB (POZ) domain containing 11 -1.8 Wdr26 WD repeat domain 26 -1.8 Armc3 armadillo repeat containing 3 -1.8 Dtd1 D-tyrosyl-tRNA deacylase 1 homolog (S. cerevisiae) -1.8 5830428M24Rik RIKEN cDNA 5830428M24 gene -1.8 Lancl1 LanC (bacterial lantibiotic synthetase component C)-like 1 -1.8 Ldhb lactate dehydrogenase B -1.8 Slc39a11 solute carrier family 39 (metal ion transporter), member 11 -1.8 Chic1 cysteine-rich hydrophobic domain 1 -1.8 Dbp D site albumin promoter binding protein -1.8 Zmat3 zinc finger matrin type 3 -1.8 Zfp770 zinc finger protein 770 -1.8 Amz2 archaelysin family metallopeptidase 2 -1.8 Pik3ip1 phosphoinositide-3-kinase interacting protein 1 -1.8 Ankmy2 ankyrin repeat and MYND domain containing 2 -1.8 Cd4 CD4 antigen -1.8 A930005H10Rik RIKEN cDNA A930005H10 gene -1.8 Tk2 thymidine kinase 2, mitochondrial -1.8 Il16 interleukin 16 -1.8 Elk4 ELK4, member of ETS oncogene family -1.8 B930025B16Rik RIKEN cDNA B930025B16 gene -1.8 Mccc1 methylcrotonoyl-Coenzyme A carboxylase 1 (alpha) -1.8 Egr1 early growth response 1 -1.8 Fbxw8 F-box and WD-40 domain protein 8 -1.8 Gm13012 predicted gene 13012 -1.8 Rcl1 RNA terminal phosphate cyclase-like 1 -1.8 Plxnc1 plexin C1 -1.8 H2-Ke6 H2-K region expressed gene 6 -1.8 Ptpn11 protein tyrosine phosphatase, non-receptor type 11 -1.8 Abcd3 ATP-binding cassette, sub-family D (ALD), member 3 -1.8 Sidt1 SID1 transmembrane family, member 1 -1.8 Prkd2 protein kinase D2 -1.8 Sit1 suppression inducing transmembrane adaptor 1 -1.8 Lrsam1 leucine rich repeat and sterile alpha motif containing 1 -1.8 Pank1 pantothenate kinase 1 -1.8 9030425P06Rik RIKEN cDNA 9030425P06 gene -1.8 Tsen54 tRNA splicing endonuclease 54 homolog (S. cerevisiae) -1.8 Orai2 ORAI calcium release-activated calcium modulator 2 -1.8 Mettl8 methyltransferase like 8 -1.8 Ccm2 cerebral cavernous malformation 2 -1.7 D18Ertd653e DNA segment, Chr 18, ERATO Doi 653, expressed -1.7 B3gnt5 UDP-GlcNAc:betaGal beta-1,3-N-acetylglucosaminyltransferase 5 -1.7 Zfp790 zinc finger protein 790 -1.7 Rph3al rabphilin 3A-like (without C2 domains) -1.7 Ppie peptidylprolyl isomerase E (cyclophilin E) -1.7 Add3 adducin 3 (gamma) -1.7 Pccb propionyl Coenzyme A carboxylase, beta polypeptide -1.7 Nipsnap1 4-nitrophenylphosphatase domain and non-neuronal SNAP25-like protein homolog 1 (C. elegans) -1.7 Akap8l A kinase (PRKA) anchor protein 8-like -1.7 Slc9a9 solute carrier family 9 (sodium/hydrogen exchanger), member 9 -1.7 Utrn utrophin -1.7 Gemin8 gem (nuclear organelle) associated protein 8 -1.7 6820431F20Rik RIKEN cDNA 6820431F20 gene -1.7 Zfp874b zinc finger protein 874b -1.7 Scd2 stearoyl-Coenzyme A desaturase 2 -1.7 2010003O02Rik RIKEN cDNA 2010003O02 gene -1.7 Slc25a33 solute carrier family 25, member 33 -1.7 Klc4 kinesin light chain 4 -1.7 Skap1 src family associated phosphoprotein 1 -1.7 Kdm1a lysine (K)-specific demethylase 1A -1.7 Acyp1 acylphosphatase 1, erythrocyte (common) type -1.7 Isoc2b isochorismatase domain containing 2b -1.7 Tcf7 transcription factor 7, T cell specific -1.7 Mlh3 mutL homolog 3 (E coli) -1.7 Sco2 SCO cytochrome oxidase deficient homolog 2 (yeast) -1.7 2210013O21Rik RIKEN cDNA 2210013O21 gene -1.7 Prmt3 protein arginine N-methyltransferase 3 -1.7 Pdlim7 PDZ and LIM domain 7 -1.7 Fam73a family with sequence similarity 73, member A -1.7 Eid2 EP300 interacting inhibitor of differentiation 2 -1.7 Josd2 Josephin domain containing 2 -1.7 Cnp 2',3'-cyclic nucleotide 3' phosphodiesterase -1.7 D3Ertd254e DNA segment, Chr 3, ERATO Doi 254, expressed -1.7 Eefsec eukaryotic elongation factor, selenocysteine-tRNA-specific -1.7 Ptplad1 protein tyrosine phosphatase-like A domain containing 1 -1.7 Npepl1 aminopeptidase-like 1 -1.7 Tiam1 T cell lymphoma invasion and metastasis 1 -1.7 Spint2 serine protease inhibitor, Kunitz type 2 -1.7 Tmem55a transmembrane protein 55A -1.7 Rapgef6 Rap guanine nucleotide exchange factor (GEF) 6 -1.7 Exoc6b exocyst complex component 6B

Nature Immunology: doi:10.1038/ni.3229 -1.7 Icos inducible T cell co-stimulator -1.7 Zfp260 zinc finger protein 260 -1.7 Alkbh2 alkB, alkylation repair homolog 2 (E. coli) -1.7 Gm17491 predicted gene, 17491 -1.7 Spr sepiapterin reductase -1.7 Insr insulin receptor -1.7 Sept1 septin 1 -1.7 Tmem42 transmembrane protein 42 -1.7 9430023L20Rik RIKEN cDNA 9430023L20 gene -1.7 Zfp82 zinc finger protein 82 -1.7 Itpr2 inositol 1,4,5-triphosphate receptor 2 -1.7 Pebp1 phosphatidylethanolamine binding protein 1 -1.7 Ivd isovaleryl coenzyme A dehydrogenase -1.7 Stx6 syntaxin 6 -1.7 Evl Ena-vasodilator stimulated phosphoprotein -1.7 Ash1l ash1 (absent, small, or homeotic)-like (Drosophila) -1.7 Wdr74 WD repeat domain 74 -1.7 Ralgps2 Ral GEF with PH domain and SH3 binding motif 2 -1.7 1300018J18Rik RIKEN cDNA 1300018J18 gene -1.7 Pnpla7 patatin-like phospholipase domain containing 7 -1.7 Znf512b zinc finger protein 512B -1.7 Sh3kbp1 SH3-domain kinase binding protein 1 -1.7 5730408K05Rik RIKEN cDNA 5730408K05 gene -1.7 Btg2 B cell translocation gene 2, anti-proliferative -1.7 Atg14 autophagy related 14 -1.7 Mfge8 milk fat globule-EGF factor 8 protein -1.7 Kdm3a lysine (K)-specific demethylase 3A -1.7 Fam162a family with sequence similarity 162, member A -1.7 Polr3e polymerase (RNA) III (DNA directed) polypeptide E -1.7 Abi2 abl-interactor 2 -1.7 Tox thymocyte selection-associated high mobility group box -1.7 Plk1s1 polo-like kinase 1 substrate 1 -1.7 Nt5e 5' nucleotidase, ecto -1.7 Slc11a2 solute carrier family 11 (proton-coupled divalent metal ion transporters), member 2 -1.7 Numb numb gene homolog (Drosophila) -1.7 Diexf digestive organ expansion factor homolog (zebrafish) -1.7 Akr7a5 aldo-keto reductase family 7, member A5 (aflatoxin aldehyde reductase) -1.7 Ypel3 yippee-like 3 (Drosophila) -1.7 Tatdn2 TatD DNase domain containing 2 -1.7 Fbxo33 F-box protein 33 -1.7 Drap1 Dr1 associated protein 1 (negative cofactor 2 alpha) -1.7 Bdh1 3-hydroxybutyrate dehydrogenase, type 1 -1.6 1810063B07Rik RIKEN cDNA 1810063B07 gene -1.6 Otud1 OTU domain containing 1 -1.6 5430417L22Rik RIKEN cDNA 5430417L22 gene -1.6 Tacc2 transforming, acidic coiled-coil containing protein 2 -1.6 Tmem38b transmembrane protein 38B -1.6 Sertad3 SERTA domain containing 3 -1.6 Gtpbp5 GTP binding protein 5 -1.6 Ehd3 EH-domain containing 3 -1.6 Hddc2 HD domain containing 2 -1.6 Acot2 acyl-CoA thioesterase 2 -1.6 Crlf3 cytokine receptor-like factor 3 -1.6 Trappc9 trafficking protein particle complex 9 -1.6 Sec31a Sec31 homolog A (S. cerevisiae) -1.6 Ankrd46 ankyrin repeat domain 46 -1.6 Sqstm1 sequestosome 1 -1.6 Ppp1r14b protein phosphatase 1, regulatory (inhibitor) subunit 14B -1.6 A830021K08Rik RIKEN cDNA A830021K08 gene -1.6 Lancl3 LanC lantibiotic synthetase component C-like 3 (bacterial) -1.6 Echdc1 enoyl Coenzyme A hydratase domain containing 1 -1.6 Ttll4 tubulin tyrosine ligase-like family, member 4 -1.6 Sh3glb2 SH3-domain GRB2-like endophilin B2 -1.6 Rsu1 Ras suppressor protein 1 -1.6 Btbd6 BTB (POZ) domain containing 6 -1.6 Pvt1 plasmacytoma variant translocation 1 -1.6 Ppm1f protein phosphatase 1F (PP2C domain containing) -1.6 Utp20 UTP20, small subunit (SSU) processome component, homolog (yeast) -1.6 Hint2 histidine triad nucleotide binding protein 2 -1.6 Ehhadh enoyl-Coenzyme A, hydratase/3-hydroxyacyl Coenzyme A dehydrogenase -1.6 Mettl22 methyltransferase like 22 -1.6 Vamp1 vesicle-associated membrane protein 1 -1.6 Rpa1 replication protein A1 -1.6 Akap9 A kinase (PRKA) anchor protein (yotiao) 9 -1.6 Setx senataxin -1.6 Rnpepl1 arginyl aminopeptidase (aminopeptidase B)-like 1 -1.6 B3galt4 UDP-Gal:betaGlcNAc beta 1,3-galactosyltransferase, polypeptide 4 -1.6 Ccdc28a coiled-coil domain containing 28A -1.6 Calcrl calcitonin receptor-like -1.6 1110002L01Rik RIKEN cDNA 1110002L01 gene -1.6 Zfand1 zinc finger, AN1-type domain 1 -1.6 Clk1 CDC-like kinase 1 -1.6 D11Wsu99e DNA segment, Chr 11, Wayne State University 99, expressed -1.6 Xrcc6bp1 XRCC6 binding protein 1 -1.6 Rcbtb1 regulator of chromosome condensation (RCC1) and BTB (POZ) domain containing protein 1 -1.6 Sgms1 sphingomyelin synthase 1 -1.6 Icam2 intercellular adhesion molecule 2 -1.6 Mtap7 microtubule-associated protein 7 -1.6 Ablim1 actin-binding LIM protein 1 -1.6 Deaf1 deformed epidermal autoregulatory factor 1 (Drosophila) -1.6 Plin3 perilipin 3 -1.6 Wdfy2 WD repeat and FYVE domain containing 2 -1.6 Csnk1g2 casein kinase 1, gamma 2 -1.6 Hibadh 3-hydroxyisobutyrate dehydrogenase -1.6 Mpst mercaptopyruvate sulfurtransferase

Nature Immunology: doi:10.1038/ni.3229 -1.6 Zfp932 zinc finger protein 932 -1.6 1190007I07Rik RIKEN cDNA 1190007I07 gene -1.6 Pop5 processing of precursor 5, ribonuclease P/MRP family (S. cerevisiae) -1.6 Raf1 v-raf-leukemia viral oncogene 1 -1.6 Dhodh dihydroorotate dehydrogenase -1.6 Snap47 synaptosomal-associated protein, 47 -1.6 Rbak RB-associated KRAB repressor -1.6 Rccd1 RCC1 domain containing 1 -1.6 Lyrm4 LYR motif containing 4 -1.6 Gpbp1l1 GC-rich promoter binding protein 1-like 1 -1.6 Cluap1 clusterin associated protein 1 -1.6 1110018J18Rik RIKEN cDNA 1110018J18 gene -1.6 Mbtps1 membrane-bound transcription factor peptidase, site 1 -1.6 Ankrd23 ankyrin repeat domain 23 -1.6 Tmx4 thioredoxin-related transmembrane protein 4 -1.6 BB283564 expressed sequence BB283564 -1.6 Akt3 thymoma viral proto-oncogene 3 -1.6 Ddx41 DEAD (Asp-Glu-Ala-Asp) box polypeptide 41 -1.6 Tmem11 transmembrane protein 11 -1.6 Gm15417 predicted gene 15417 -1.6 Capns1 calpain, small subunit 1 -1.6 2310045N01Rik RIKEN cDNA 2310045N01 gene -1.6 Mzt2 mitotic spindle organizing protein 2 -1.6 Jmjd6 jumonji domain containing 6 -1.6 Ogt O-linked N-acetylglucosamine (GlcNAc) transferase (UDP-N-acetylglucosamine:polypeptide-N-acetylglucosaminyl transferase) -1.6 Itga6 integrin alpha 6 -1.6 Phpt1 phosphohistidine phosphatase 1 -1.6 Prkcq protein kinase C, theta -1.6 Glod4 glyoxalase domain containing 4 -1.6 Fam189b family with sequence similarity 189, member B -1.6 Klhl5 kelch-like 5 (Drosophila) -1.6 Bivm basic, immunoglobulin-like variable motif containing -1.6 Nars2 asparaginyl-tRNA synthetase 2 (mitochondrial)(putative) -1.6 Rab3ip RAB3A interacting protein -1.6 B3gnt1 UDP-GlcNAc:betaGal beta-1,3-N-acetylglucosaminyltransferase 1 -1.6 Nek3 NIMA (never in mitosis gene a)-related expressed kinase 3 -1.6 Fto fat mass and obesity associated -1.6 Orc5 origin recognition complex, subunit 5 -1.6 BC018507 cDNA sequence BC018507 -1.6 2310030N02Rik RIKEN cDNA 2310030N02 gene -1.6 Abhd6 abhydrolase domain containing 6 -1.6 Slc36a4 solute carrier family 36 (proton/amino acid symporter), member 4 -1.6 2310016M24Rik RIKEN cDNA 2310016M24 gene -1.6 Tgif1 TGFB-induced factor homeobox 1 -1.6 2810422O20Rik RIKEN cDNA 2810422O20 gene -1.6 Wdyhv1 WDYHV motif containing 1 -1.6 Pigy phosphatidylinositol glycan anchor biosynthesis, class Y -1.6 Rdh5 retinol dehydrogenase 5 -1.6 9430091E24Rik RIKEN cDNA 9430091E24 gene -1.6 Trim32 tripartite motif-containing 32 -1.6 Hpcal1 hippocalcin-like 1 -1.6 Gtf2i general transcription factor II I -1.6 Ovca2 candidate tumor suppressor in ovarian cancer 2 -1.6 Pgs1 phosphatidylglycerophosphate synthase 1 -1.6 Hadha hydroxyacyl-Coenzyme A dehydrogenase/3-ketoacyl-Coenzyme A thiolase/enoyl-Coenzyme A hydratase (trifunctional protein), alpha -1.6 Acp5 acid phosphatase 5, tartrate resistant -1.6 Frat2 frequently rearranged in advanced T cell lymphomas 2 -1.6 Zfp954 zinc finger protein 954 -1.6 Mfng MFNG O-fucosylpeptide 3-beta-N-acetylglucosaminyltransferase -1.6 Cd5 CD5 antigen -1.6 Lsg1 large subunit GTPase 1 homolog (S. cerevisiae) -1.6 Lipa lysosomal acid lipase A -1.6 1810020D17Rik RIKEN cDNA 1810020D17 gene -1.6 Zfp7 zinc finger protein 7 -1.6 Txndc15 thioredoxin domain containing 15 -1.6 Eci1 enoyl-Coenzyme A delta isomerase 1 -1.6 1110037F02Rik RIKEN cDNA 1110037F02 gene -1.6 Tle3 transducin-like enhancer of split 3, homolog of Drosophila E(spl) -1.6 Snx20 sorting nexin 20 -1.6 Paqr7 progestin and adipoQ receptor family member VII -1.6 Arhgef3 Rho guanine nucleotide exchange factor (GEF) 3 -1.6 Fam69a family with sequence similarity 69, member A -1.6 Egr2 early growth response 2 -1.6 Mib2 mindbomb homolog 2 (Drosophila) -1.6 Ecsit ECSIT homolog (Drosophila) -1.6 Cdk2ap2 CDK2-associated protein 2 -1.6 Zfp12 zinc finger protein 12 -1.6 Narf nuclear prelamin A recognition factor -1.6 Asb3 ankyrin repeat and SOCS box-containing 3 -1.6 Bri3bp Bri3 binding protein -1.6 Gnl1 guanine nucleotide binding protein-like 1 -1.6 D830012I24Rik RIKEN cDNA D830012I24 gene -1.6 2310009A05Rik RIKEN cDNA 2310009A05 gene -1.6 Dnajb2 DnaJ (Hsp40) homolog, subfamily B, member 2 -1.5 Ass1 argininosuccinate synthetase 1 -1.5 Narg2 NMDA receptor-regulated gene 2 -1.5 Tmem18 transmembrane protein 18 -1.5 Calcoco1 calcium binding and coiled coil domain 1 -1.5 Zxdb zinc finger, X-linked, duplicated B -1.5 Pi4kb phosphatidylinositol 4-kinase, catalytic, beta polypeptide -1.5 Use1 unconventional SNARE in the ER 1 homolog (S. cerevisiae) -1.5 Uap1l1 UDP-N-acteylglucosamine pyrophosphorylase 1-like 1 -1.5 Fam169b family with sequence similarity 169, member B -1.5 AI462493 expressed sequence AI462493 -1.5 Gpd1l glycerol-3-phosphate dehydrogenase 1-like

Nature Immunology: doi:10.1038/ni.3229 -1.5 Gnpat glyceronephosphate O-acyltransferase -1.5 Cnot6l CCR4-NOT transcription complex, subunit 6-like -1.5 Bckdha branched chain ketoacid dehydrogenase E1, alpha polypeptide -1.5 Ddb2 damage specific DNA binding protein 2 -1.5 Trpv2 transient receptor potential cation channel, subfamily V, member 2 -1.5 BC056474 cDNA sequence BC056474 -1.5 Grap GRB2-related adaptor protein -1.5 Cxxc1 CXXC finger 1 (PHD domain) -1.5 1700029I01Rik RIKEN cDNA 1700029I01 gene -1.5 Rspry1 ring finger and SPRY domain containing 1 -1.5 Fam55c family with sequence similarity 55, member C -1.5 Ap1m2 adaptor protein complex AP-1, mu 2 subunit -1.5 Dcxr dicarbonyl L-xylulose reductase -1.5 Fbxl17 F-box and leucine-rich repeat protein 17 -1.5 Pecam1 platelet/endothelial cell adhesion molecule 1 -1.5 Zfp84 zinc finger protein 84 -1.5 Ppcs phosphopantothenoylcysteine synthetase -1.5 Tnfaip8l2 tumor necrosis factor, alpha-induced protein 8-like 2 -1.5 Stk38 serine/threonine kinase 38 -1.5 Kansl2 KAT8 regulatory NSL complex subunit 2 -1.5 Mrpl24 mitochondrial ribosomal protein L24 -1.5 Arl13b ADP-ribosylation factor-like 13B -1.5 Slc35c1 solute carrier family 35, member C1 -1.5 P4ha1 procollagen-proline, 2-oxoglutarate 4-dioxygenase (proline 4-hydroxylase), alpha 1 polypeptide -1.5 Ccdc85b coiled-coil domain containing 85B -1.5 Tbxa2r thromboxane A2 receptor -1.5 Lmf1 lipase maturation factor 1 -1.5 Fam109a family with sequence similarity 109, member A -1.5 Neu1 neuraminidase 1 -1.5 BB182297 expressed sequence BB182297 -1.5 Nit2 nitrilase family, member 2 -1.5 Eif3k eukaryotic translation initiation factor 3, subunit K -1.5 Fam100a family with sequence similarity 100, member A -1.5 Smpdl3a sphingomyelin phosphodiesterase, acid-like 3A -1.5 Atic 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/IMP cyclohydrolase -1.5 Clec2i C-type lectin domain family 2, member i -1.5 Nlk nemo like kinase -1.5 Adamts6 a disintegrin-like and metallopeptidase (reprolysin type) with thrombospondin type 1 motif, 6 -1.5 Acsf3 acyl-CoA synthetase family member 3 -1.5 Prmt5 protein arginine N-methyltransferase 5 -1.5 Jakmip1 janus kinase and microtubule interacting protein 1 -1.5 2610002J02Rik RIKEN cDNA 2610002J02 gene -1.5 N4bp2 NEDD4 binding protein 2 -1.5 Kdelc1 KDEL (Lys-Asp-Glu-Leu) containing 1 -1.5 Smpd2 sphingomyelin phosphodiesterase 2, neutral -1.5 AA408213 expressed sequence AA408213 -1.5 Mum1 melanoma associated antigen (mutated) 1 -1.5 Maz MYC-associated zinc finger protein (purine-binding transcription factor) -1.5 Sh2b3 SH2B adaptor protein 3 -1.5 Ppm1m protein phosphatase 1M -1.5 Galnt11 UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase 11 -1.5 Dnaja4 DnaJ (Hsp40) homolog, subfamily A, member 4 -1.5 Bbs4 Bardet-Biedl syndrome 4 (human) -1.5 2210021J22Rik RIKEN cDNA 2210021J22 gene -1.5 Atp9b ATPase, class II, type 9B -1.5 Endog endonuclease G -1.5 1110012L19Rik RIKEN cDNA 1110012L19 gene -1.5 D10Jhu81e DNA segment, Chr 10, Johns Hopkins University 81 expressed -1.5 Nap1l4 nucleosome assembly protein 1-like 4 -1.5 Cc2d1b coiled-coil and C2 domain containing 1B -1.5 Gatsl3 GATS protein-like 3 -1.5 Vgll4 vestigial like 4 (Drosophila) -1.5 Rabac1 Rab acceptor 1 (prenylated) –/– Data represent the average of 2 indiependent samples of WT and Tcf7 TFH cells analyzed by microarray. –/– Positive fold changes indicate upregualted genes in Tcf7 TFH cells and are shown in bold. –/– Negative fold changes indicate downregulated genes in Tcf7 TFH cells. All fold changes are statistically significant (P < 0.05, unpaired one-tailed t -test). Genes highlighted in yellow are validated by RT-qPCR.

Nature Immunology: doi:10.1038/ni.3229 Supplementary Table 2. Antibodies and reagents used in flow cytometry. Antibody target/Reagent Clone/Cat. No. Dilution Provider CD4 RM4-5 1:100 Biolegend CD8 53-6.7 1:100 BD Biosciences CD19 6D5 1:100 Biolegend CD25 PC61.5 1:100 Biolegend CD44 IM7 1:100 eBioscience FAS JO2 1:100 BD Biosciences SLAM TC15-12F12.2 1:100 Biolegend ICOS C398.4A 1:100 Biolegend CD45.1 A20 1:100 Biolegend CD45.2 104 1:100 Biolegend PD-1 RMP1-30 1:100 eBioscience Thy-1.1 OX-7 1:100 Biolegend Granzyme B GB11 1:20 Life Technologies T-bet 4B10 1:100 Biolegend Foxp3 FJK-16s 1:100 eBioscience GITR DTA-1 1:100 eBioscience Bcl-6 K112-91 1:20 BD Biosciences Tim3 215008 1:20 R&D Systems PNA FL-1071 1:500 Vector Labs CD138 281-2 1:100 BD Biosciences Annexin V Kit 88-8102 1:50 eBioscience B220 RA3-6B2 1:100 eBioscience Streptavidin 25-4317-82 1:500 eBioscience Live/Dead Kit L10119 1:100 Life Technologies CD62L MEL-14 1:100 eBioscience Biotin Goat anti-Rat IgG 112-065-143 1:1000 Jackson Immunoresearch Ki-67 Kit B56 1:2 BD Biosciences Fc-blocker 2.4G2 1:100 BD Biosciences BrdU 3D4 1:50 BD Biosciences CXCR5 2G8 1:100 BD Biosciences TCF-1 C46C7 1:100 Cell Signalling Technology hCD2 RPA-2.10 1:20 Biolegend hCD4 OCT4 1:20 Biolegend

Nature Immunology: doi:10.1038/ni.3229 Supplementary Table 3. Primers used in quantitative PCR. For gene expression analysis Gene symbol 5' primer 3' primer Actb 5'-aatcgtgcgtgacatcaaag 5'-ggattccatacccaagaagg Bcl6 5'-agacgcacagtgacaaacca 5'-agtgtgggtcttcaggttgg Icos 5'-tgccgtgtctttgtcttctg 5'-cttcccttggtcttggtgag Prdm1 5'-agtgcaatgtctgtgccaag 5'-ttgagattgcttgtgctgct Tbx21 5'-caatgtgacccagatgatcg 5'-gcgttctggtaggcagtcac Gzmb 5'-cctcctgctactgctgacct 5'-taaggccatgtagggtcgag Il21 5'-cgcctcctgattagacttcg 5'-aaaacaggcaaaagctgcat Il4 5'-ggcattttgaacgaggtcac 5'-aaatatgcgaagcaccttgg Il6ra 5'-gcaagaatcctcgtccatgt 5'-gtggaggagaggtcgtcttg Il6st 5'-catgctttcaggctttcctc 5'-tcccactggcacagcatatt Ascl2 5'-tactcgtcggaggaaagcag 5'-acccagggatgcagcttag Tcf7 5'-caatctgctcatgccctacc 5'-cttgcttctggctgatgtcc Cxcr5 5'-catgggctccatcacataca 5'-ggcatgaataccgccttaaa Gata3 5'-cttatcaagcccaagcgaag 5'-cattagcgttcctcctccag RORgt 5'-ggacagggagccaagttctca 5'-cacaggtgataaccccgtagtgg Foxp3 5'-agaagctgggagctatgcag 5'-tactggtggctacgatgcag Id3 5'-atctcccgatccagacagc 5'-gagagagggtcccagagtcc Pou2af1 5'-gacatgtacgtgcagcctgt 5'-cgggtgtagcagtgcttctt Slamf6 5'-ccctggaatgcagtatggtt 5'-gctctgggaggactctggat Cd200 5'-cctggggaatgtgattgact 5'-ctcctgatttcggtgacgtt Sostdc1 5'-gaggcaggcatttcagtagc 5'-gtatttggtggaccgcagtt Cd22 5'-gagcaagctcaccttccaac 5'-cacctctgtgggattgacct Tox2 5'-aatgagccacagaagccagt 5'-ccttcttggcagcttcagtc Nsg2 5'-ggcttgactctgctctgacc 5'-cagccagctgcaaatgatta Plagl1 5'-gagcagaggaaggagcagaa 5'-atatggggagttgggggtag Tox 5'-gaggatgcctccaagatcaa 5'-gcctgggtatcacgaaagaa Ccl4 5'-agccagctgtggtattcctg 5'-gaggaggcctctcctgaagt E2f8 5'-agtagaattccgggcagctt 5'-ccagatcttccacgtggtct Id2 5'-catcagcatcctgtccttgc 5'-gtgttctcctggtgaaatgg Ggt1 5'-ttctctccgaagagcgtagc 5'-ggtggcgtagaactcagagc Ccr2 5'-attctccacaccctgtttcg 5'-gattcctggaaggtggtcaa Aurkb 5'-gaagaagagccgtttcatcg 5'-ggatgttgggatgtttcagg Bhlhe40 5'-gaccggattaacgagtgcat 5'-tcaatgctttcacgtgcttc Slamf7 5'-gcagatccccaaagtggtaa 5'-aggacggagtgctgtcatct Fasl 5'-gcagaaggaactggcagaac 5'-ttaaatgggccacactcctc Slamf1 5'-tcgagtccatggatgcaata 5'-ggctggcagtgatttgattt For TCF-1 occupancy in genomic locations Genome location 5' primer 3' primer Bcl6 –0.5K 5'-gggtctggggctaattcttc 5'-tagctggaaggagctgtggt Prdm1 –24.5K 5'-atgtttcggaggaggttgtg 5'-aagggtgggaccagattctt Prdm1 –0.9K 5'-tcagccccagtctgtcttct 5'-ggcaaaagaccattgaagga

Nature Immunology: doi:10.1038/ni.3229 Biosensors and Bioelectronics 94 (2017) 305–311

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

journal homepage: www.elsevier.com/locate/bios

A dynamic sandwich assay on magnetic beads for selective detection of MARK single-nucleotide mutations at room temperature

Junxiu Wanga, Guoliang Xiongb, Liang Mac, Shihui Wanga, Xu Zhoua, Lei Wanga, Lehui Xiaod, ⁎ ⁎ Xin Sua, , Changyuan Yua, a Beijing Key Laboratory of Bioprocess, College of Life Science and Technology, Beijing University of Chemical Technology, Beijing 100029, China b Department of Nephrology, Shenzhen Affiliated Hospital, Guangzhou University of Chinese Medicine, Shenzhen, Guangdong 518033, China c Clinical Laboratory, China-Japan Friendship Hospital, Beijing 100029, China d College of Chemistry, Nankai University, Tianjin 300071, China

ARTICLE INFO ABSTRACT

Keywords: Single-nucleotide mutation (SNM) has proven to be associated with a variety of human diseases. Development SNM of reliable methods for the detection of SNM is crucial for molecular diagnosis and personalized medicine. The Transient DNA binding sandwich assays are widely used tools for detecting nucleic acid biomarkers due to their low cost and rapid Magnetic beads signaling. However, the poor hybridization specificity of signal probe at room temperature hampers the Cancer biomarker detection discrimination of mutant and wild type. Here, we demonstrate a dynamic sandwich assay on magnetic beads for SNM detection based on the transient binding between signal probe and target. By taking the advantage of mismatch sensitive thermodynamics of transient DNA binding, the dynamic sandwich assay exhibits high discrimination factor for mutant with a broad range of salt concentration at room temperature. The beads used in this assay serve as a tool for separation, and might be helpful to enhance SNM selectivity. Flexible design of signal probe and facile magnetic separation allow multiple-mode downstream analysis including colorimetric detection and isothermal amplification. With this method, BRAF mutations in the genomic DNA extracted from cancer cell lines were tested, allowing sensitive detection of SNM at very low abundances (0.1–0.5% mutant/ wild type).

1. Introduction probe and signaling probe eliminates false-positive effectively. Super- sandwich architecture developed by Xia and co-workers further Sequencing technology has advanced the understanding of the role amplifies signal and push the detection limit down (Jiang et al., of genomic diversity in human diseases. The detection of small change 2012; Liu et al., 2013). However, the hybridization based probes used in genome, particularly single-nucleotide mutation (SNM) is of great in sandwich assay are not sufficiently sensitive to single-nucleotide significance because SNM has been prominent biomarkers for a variety variation because single mismatch in DNA duplex only produces a of diseases (Diehl et al., 2008; Kathiresan and Srivastava, 2012; Van small change in hybridization thermodynamics at room temperature. Cauwenberghe et al., 2016). Various techniques have demonstrated the Recently, various strategies have been developed to improve the capability for SNM detection, such as next-generation sequencing specificity of hybridization based probes, such as probe-pair strategies (Schmitt et al., 2012), allele specific amplification (Khripin et al., (Wang and Zhang, 2015), dynamic DNA nanotechnology (Hwang et al., 2000), simulation-guided probes (Wang and Zhang, 2015), nuclease 2016) and nanopore techniques (Ang and Yung, 2012), whereas the assisted methods (Xiao et al., 2012), electrochemical sensors (Das high cost and complicated design of these approaches hamper further et al., 2015), and microarray (LaFramboise, 2009). Sandwich assay has applications. Hence, low-cost and simple probe design which can be been widely used in the fields of biomarker detection including coupled with sandwich assay is urgently needed for high-confidence proteins, nucleic acids and small molecules (Shen et al., 2014). The SNM sensing. advances in nanotechnology and bio-conjugate chemistry enable multi- Short DNA oligonucleotides (diffusing strands) can transiently bind ple signaling mode in sandwich assay providing high sensitivity (Ding with complementary “docking” strands immobilized on interface at et al., 2016; Pallela et al., 2016). Sandwich assay is an excellent room temperature (Johnson-Buck et al., 2013; Jungmann et al., 2010) platform for nucleic acid sensing. The simultaneous binding of capture The binding events can be detected by total internal reflection excita-

⁎ Corresponding authors. E-mail addresses: [email protected] (X. Su), [email protected] (C. Yu). http://dx.doi.org/10.1016/j.bios.2017.03.023 Received 19 January 2017; Received in revised form 7 March 2017; Accepted 11 March 2017 Available online 12 March 2017 0956-5663/ © 2017 Elsevier B.V. All rights reserved. J. Wang et al. Biosensors and Bioelectronics 94 (2017) 305–311 tion by using fluorescently labeled diffusing strand. The transient at room temperature. Next, upon magnetic separation the supernatant binding of diffusing strand with its target possesses unique kinetic and beads were collected, respectively. The beads were rinsed by the signature, which can be easily distinguished from nonspecific binding Tris buffer for three times, and then stored in 1×PBS at 4 °C. The free events. Based on this technique, DNA-point accumulation imaging in DNA in the supernatant was measured to calculate the conjugation nanoscale topography (DNA-PAINT) and quantitative-PAINT yield. (qPAINT) have been developed for subcellular super-resolution ima- ging (Jungmann et al., 2014) and biomolecules counting inside cells (Jungmann et al., 2016). In light of DNA-PAINT, Walter and co- 2.3. SNM detection by DSA using fluorescent probe workers reported a kinetic fingerprinting approach for the amplifica- tion-free detection of nucleic acids at single-molecule level, which 1 mg/mL capture probe coated beads and 50 nM target single fi fl allows direct quanti cation of miRNA biomarker in clinical bio uids strand DNA were incubated in a NaCl contained Tris buffer (10 mM (Johnson-Buck et al., 2015). However, the high cost of single molecule Tris-HCl, pH 7.5) for 10 min at room temperature followed by adding apparatus and the complexity of statistical model hampers its applica- short fluorescent DNA probe to reach a final concentration of 50 nM. tions in point-of-care diagnosis. Meanwhile, Magnetic beads or nano- After 10 min, the first magnetic separation was performed; and the particles provide powerful platform for biomolecule sensing because supernatant and beads were collected, respectively. The fluorescence the superparamagnetic property makes facile separation processes (Ex: 490 nm, Em: 510 nm) in the supernatant was measured on a (Rocha-Santos, 2014). Detection methods for nucleic acids have been multilabel reader (EnVision, PerkinElmer, UK). The beads were developed by using magnetic particles (Chen et al., 2012; Sloane et al., dissolved in ion-free water and incubated for 10 min. The second 2015). magnetic separation was performed and supernatant was collected. Here, we describe a simple and low cost method for highly selective Prior to measuring the fluorescence of this supernatant, Tris buffer detection of SNM at room temperature by constructing dynamic (final concentration 10 mM, pH 7.5) was added to maintain high sandwich assay (DSA) on magnetic beads. Unlike common sandwich quantum yield of dye at neutral pH. To ensure high reproducibility, assay for nucleic acids, the signal probe used in DSA is short, allowing magnetic separation rack was used for each separation and the transient binding with its target at room temperature. Owing to the centrifugate tube was shacked frequently during the time of incubation. differential hybridization thermodynamics of short fully complemen- tary and single mismatched DNA duplex, highly selective SNM detec- tion is readily achieved. The magnetic beads employed in DSA serve as 2.4. Visualized detection of SNM by using G-rich probe a platform for separation and signaling and might be helpful to enhance the single-nucleotide selectivity due to the particle surface G-rich probe was used instead of the short fluorescent probes. 4 μL effect. Flexible design of signal probes and facile magnetic separation of DMSO containing 62.5 μM hemin was added to the supernatant of provide multiple-mode downstream analysis (i.e., colorimetric detec- the second magnetic separation to a final volume of 50 μL. The tion and isothermal amplification). The high discrimination capability supernatant was incubated at 37 °C for 1 h to facilitate the formation of DSA enables the detection of cancer-related SNM from cell lines at of G-quadruplex-hemin complex. The supernatant was then combined low abundance (0.1–0.5% mutant/wild type). with 143 μL HEPES buffer (50 mM HEPES, 20 mM KCl, 400 mM NaCl, 2% DMSO, 0.1% Triton X-100, pH 5.2). 2 μL of 4 mM ABTS and 2. Materials and methods 5 μL of 4 mM H2O2 were added to launch the catalytic oxidation. The color of the solution was observed after 10 min. 2.1. Materials

Oligonucleotides were synthesized and purified by HPLC (Sangon, Shanghai, China) and their sequences are listed in Table S1. Amine 2.5. EXPAR as the downstream assay for DSA modified magnetic beads (d=2 µm, 10 mg/mL) were purchased from Beaver Nanotech Co. (Shanghai, China). Nt.BstNBI nicking enzyme The EXPAR probe which can serve as the primer for EXPAR was fl and Vent (exo-) DNA polymerase and their corresponding buffers were used instead of the uorescent probe. The EXPAR reaction was carried μ ff ， obtained from New England Biolabs (MA, USA). Hemin, glutaralde- out in 20 L EXPAR bu er (22.5 mM Tris-HCl 5 mM (NH4)2SO4, μ hyde, 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) and 25 mM NaCl, 5 mM KCl, 3.5 mM MgCl2, 0.1% Triton X-100, 50 g/mL μ 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) were BSA, pH 8.3) containing 20 nM template, 0.25 mM dNTPs, 0.05 U/ L μ purchased from Sigma (St. Louis, MO, USA). SYBR Green I was Vent (exo-) DNA polymerase, 0.3 U/ L Nt.BstNBI, 1×SYBR Green, and purchased from Thermo Fisher (MA, USA). dNTPs were from released EXPAR probe at 55 °C. Fluorescence was recorded in the Tiangen Biotech Co. (Beijing, China). DNase/RNase free deionized SYBR/FAM channel of real-time PCR cycler (Mastercycler realplex, water from Tiangen Co. was used in all experiments. All of other Eppendorf, Germany) with a time interval of 8 s. chemicals are analytical grade without further purification unless mentioned. 2.6. Detection of BRAF mutations at different frequencies 2.2. Preparation of biotinylated single strand DNA coated magnetic beads Genomic DNAs were extracted from A375 and HEK-293T cell lines by QIAamp DNA Micro Kit (QIAGEN, Germany). PCR was performed Amine-modified magnetic beads were dissolved in 15% glutaralde- in 1×ThermoPol Buffer containing 500 nM primers (reverse primer is hyde solution and incubated for 1 h in 1.7 mL centrifugate tube. Upon 5′-phosphorylated), 0.2 mM dNTPs, 0.025 U/μL Taq DNA polymerase magnetic separation, the beads were collected and rinsed by 1×PBS for and 0.02 ng/μL genomic DNA. The PCR amplification was carried out three times. Then the beads were mixed with 0.25 mg/mL streptavidin on GeneAmp 9700 thermal cycler (ABI, USA) for 30 cycles (94 °C 30 s, in 1×PBS for at room temperature for 1 h, and the mixture was stand at 55 °C 30 s, 72 °C 30 s). Next, lambda exonuclease (final concentration, 4 °C overnight. Excess streptavidin was flushed out by three volumes of 0.1 U/μL) was added to digest the 5′-phosphorylated strands which 1×PBS. 2 mg/mL streptavidin modified magnetic beads and 1.6 μM was extended by the 5′-phosphorylated primer for 20 min at 37 °C biotinylated single strand DNA (capture probe) were incubated in Tris followed by inactivation of enzyme at 85 °C for 10 min. The PCR buffer (10 mM Tris-HCl, 1 M NaCl, 0.1% Tween-20, pH 7.5) for 30 min products were diluted to appropriate concentrations for DSA.

306 J. Wang et al. Biosensors and Bioelectronics 94 (2017) 305–311

Fig. 1. (A) Illustration of the preparation of biotinylated capture probe coated magnetic beads. The conjugation yield is approximately 50%. (B) Schematic representation of DSA. The mutant and wild type are captured on magnetic beads through stable hybridization between capture probe and targets. The addition of short fluorescent DNA probe leads transient binding between probe and targets. The first magnetic separation is carried out to separate the beads and diffusing strand. The beads are re-suspended in ion-free water followed by the second magnetic separation. The fluorescence intensity of mutant is significantly higher than that of wild type in the second supernatant. (C) The workflow and the duration time of each step.

3. Results and discussion the second supernatant (-salt). This suggests the probe preferentially hybridizes with mutant rather than wild type. We used discrimination 3.1. The principle of dynamic sandwich assay for SNM detection factor (DF) to evaluate the single-nucleotide discrimination capability

of DSA. DF is defined as the ratio of FFmb− and FFwb− where Fm, Fw, and The principle of the DSA for SNM detection is illustrated in Fig. 1. Fb represent the fluorescence intensity of the second supernatant for The streptavidin is conjugated on amine-modified Fe3O4 beads by mutant, wild type, and background (no target), respectively. As using glutaraldehyde as the cross-linker. The biotinylated capture expected, DSA exhibits remarkable DFs for SNM and 2-nt mutation probe is further immobilized on streptavidin coated magnetic beads by using 10-nt probe in the presence of 400 mM NaCl (Fig. 2B). In through the interaction of biotin and streptavidin (Fig. 1A). The SEM of contrast, stable sandwich assay (SSA) which uses long fluorescent streptavidin coated beads is shown in Fig. S1. The capture probe is probe (20-nt) does not distinguish mutant and wild type efficiently designed fully-complementary with the common region of wild type because the probe has close binding affinities with mutant and wild and mutant. Target single strand DNAs can be captured on the surface type (Fig. 2B). The dynamic range of DSA based on fluorescent probe is of magnetic beads through their stable hybridization with capture 0.1–50 nM and the detection limit is 0.1 nM with a signal/noise ratio probe (20-bp). The short fluorescent probe (9–12-nt) serving as signal of 3 (Fig. S2). probe is fully complementary with mutant but forms single mismatch with wild type. Upon the addition of signal probe, the transient binding 3.2. Optimization of the assay conditions for high DF between target and signal probe occurs in a salt-contained buffer at room temperature. The dynamic sandwich configuration is therefore The thermodynamics and kinetics of DNA hybridization, particu- constructed on the beads surface where the diffusing probe and bound larly transient hybridization depends on salinities. NaCl is always used probe reach an equilibrium. Due to the significant difference on the for tuning the stability of DNA duplex. In the accepted models of the hybridization thermodynamics of mutant and wild type, more probes effects of cation on hybridization thermodynamic, the effects of Na+ are preferentially bind to mutant rather than wild type. Upon the first assumed to be purely entropic (Tan and Chen, 2006). The hybridization magnetic separation, the beads are re-suspended in a salt-free solution. of fluorescent probe and target in DSA causes change in the total In the absence of salt, the stability of both target-signal probe duplexes number of paired bases leading change of entropy. Thus, the DF can be is weak resulting in the release of bound signal probes. Because Fe O 3 4 affected by NaCl concentration. Probe length is another critical factor magnetic particles have strong quenching effect on the fluorescence of on hybridization thermodynamics. The hybridization stability can be organic dyes (Ma et al., 2006; Yu et al., 2013), we carried out second greatly increased in the regime of transient binding by increasing the magnetic separation to measure the fluorescence of released probes. length of a duplex by 1 bp through the formation of an additional two The fluorescence intensity of the supernatant for mutant is substan- to three hydrogen bonds and two base-stacking contacts (Dupuis et al., tially higher than that of wild type since more signal probes are 2013). Probe length and NaCl concentration in the buffer for the first released from mutant target (Fig. 1B). For clearance, the workflow magnetic separation were therefore optimized. The fluorescence in- and the duration time of each step are presented in Fig. 1C. tensity of the supernatant upon the second separation is shown in Fig. To demonstrate the feasibility of our method, we used a 50-nt S3, and the DFs are calculated (Fig. S4). Compared with other probes single-stranded DNA containing an A > G mutation as a proof of (Fig. S4A, C and D), 10-nt probe shows relatively higher DF in a wide concept. A 10-nt fluorescent signal probe which is complementary range of NaCl concentration. The mutant and wild type are well with mutant and forms single mismatch with wild type was used. As distinguished in the presence of 150–400 mM NaCl (Fig. S4B). The shown in Fig. 2A, for the assay of mutant, less diffusing signal probes discrimination capability is relatively weak with low or high concentra- are found in the first supernatant (+salt), however, more are found in tion of NaCl where the binding affinities of mutant and wild type are

307 J. Wang et al. Biosensors and Bioelectronics 94 (2017) 305–311

Fig. 2. (A) Fluorescence intensities of mutant, wild type and background (no targets) in supernatants after the first (red) and the second (blue) magnetic separation. (B) Comparison of discrimination factor for SNM and 2-nt mutation between dynamic sandwich assay and stable sandwich assay (SSA). The signal probes for DSA and SSA are 10-nt and 20-nt, respectively. The concentrations of capture probe, targets, and signal probe are 1 μM, 50 nM and 50 nM, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) close. Divalent cation (e.g. Mg2+) was also tested. The DSA does not the DF does not change with target density (Fig. S6A). This suggests the show high DF in the presence of Mg2+. This is probably because the mutation discrimination capability changes until the negative charge of Mg2+ mediated slow dissociation rate of signal probe prohibits the surface reaches a critical value. Besides, the effect of the magnetic release of bound probe from both targets (Nakano et al., 1999; beads' size was investigated. As shown in Fig. S6B, the DF does not Owczarzy et al., 2008)(Fig. S5). change significantly with differently sized beads. Pushing the size down The high single-nucleotide selectivity or discrimination capability of to nanometer scale may bring changes on the performance of DSA DSA is attributed to the differential thermodynamics of the probe because of the high specific surface area of nanoparticles. The magnetic hybridization of mutant and wild type on beads surface in salt beads in DSA serve as a tool for separation and signal generation, and contained buffer. The thermodynamics and kinetics of DNA hybridiza- might be helpful to enhance single-nucleotide selectivity. tion on particle or slide surface often differ from that in bulk solution (Schmitt and Knotts, 2011). Duplex stability may be greater or weaker 3.3. Multiple-mode downstream analysis of DSA than that in solution. The hybridization thermodynamics of immobi- lized dsDNA have been investigated in the context of planar silicon Flexible design of the signal probe in our method enables multiple- (Wei et al., 2003) and gold nanoparticle (Chen et al., 2009; Xu and mode detection of mutant. G-quadruplex structure can specifically bind Craig, 2005) because of its importance to microarray assay and to hemin, leading to the formation of horseradish peroxidase (HRP)- integrated nanotechnology. The electronic repulsion between oligonu- mimicking DNAzyme. This DNAzyme catalyzes the oxidation of 2,2′- cleotides is commonly the major effect on the hybridization thermo- 2- azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (ABTS in buffer) by dynamics. To gain deep insight into DSA, we first measured the zeta - H2O2 to a colored product ABTS ∙ as signal readout (Li et al., 2009; Wu potential of beads as a function of capture probe density. et al., 2016). A G-rich sequence was concatenated on the 3′ end of the Oligonucleotides are negatively charged under neutral pH. As shown signal probe for colorimetric detection. It is noteworthy that the in Fig. 3A, the zeta potential decreases as a function of the density of addition of the G-rich sequence does not affect the hybridization of capture probe on the bead surface. The performance of these beads in probe and target. As the strategy of DSA, more G-rich probes can be DSA was then further investigated. Note that in all experiments, the released into ion-free buffer in presence of mutant. Upon the addition amount of capture probe is in excess. As shown in Fig. 3B, the signal of 2- of reagents for colorimetric reaction, the oxidation of ABTS is mutant does not show noticeable change with increasing capture probe facilitated by the G-quadruplex structure (top, Fig. 4A). Thus, the density, however, that of wild type decreases significantly. The DF typical green color and strong absorbance at 410 nm of the oxidation therefore increases with the density (Fig. 3C). In addition, we carried product of ABTS ∙- were found in the presence of mutant, however, not out the assay without using fluorescent probes and found that no target found in presence of wild type (bottom, Fig. 4A). The DSA based on G- fi DNA was detected in the supernatant upon the rst separation rich probe offers a dynamic range from 5 to 40 nM and detection limit suggesting the transient binding of signal probe and targets is mainly (S/N 3) of 5 nM (Fig. S7). Colorimetric detection of SNM is therefore ff a ected rather than the stable binding of target and capture probe. The achieved which is attractive for point-of-care diagnosis. electronic repulsion selectively weakens the transient binding of Steady-state fluorescence measurement is not sufficient for detect- mismatched duplex (Fig. 3D). According to above observations, the ing low concentration mutant. Amplification step is therefore required. DF of DSA may depend on the target concentration which weakens the The isothermal exponential amplification reaction (EXPAR) as an concentration robustness. 400 pmol/mg is the saturated concentration attractive amplification technique has been used for nucleic acid of capture probe on the streptavidin coated beads and exhibits high detection due to its high efficiency and robustness (Jia et al., 2010; single-nucleotide resolution. We performed DSA for targets at different Li et al., 2016). To meet the requirement of EXPAR, the signal probe concentrations by using saturated capture probe coated beads where was extended to 20-nt serving as the primer of amplification. The

308 J. Wang et al. Biosensors and Bioelectronics 94 (2017) 305–311

Fig. 3. DSA performance as a function of capture probe density. (A) Zeta potential of magnetic beads as a function of capture probe density. (B) The fluorescence intensity in the second supernatant for mutant and wild type. (C) Discrimination factor comparison of different amounts of capture probe coated beads. (D) The ratio of unbound state of the wild type-probe duplex is gradually increases with the capture probe density, whereas the hybridization thermodynamics of mutant-probe duplex does not show noticeable change. The concentrations of targets and fluorescent signal probe are both 50 nM. length of the complementary sequence of extended signal probe and the BRAF gene were extracted from A375 and HEK-293T cell line, mutant is still 10-nt. As illustrated in Fig. 4B, the EXPAR reaction is respectively. PCR reactions were carried out by using genomic DNA initiated through the hybridization of released signal probe and (varying ratios of mutant/wild type) and 5′-phosphorylated reverse template (green). The formation of a partial double strand can be primer. The single strand DNAs for the DSA were generated by using extended by Vent (exo-) DNA polymerase, resulting in a stable double- lambda exonuclease digesting the extended phosphorylated strand. As stranded DNA containing a recognition site for the nicking enzyme shown in Fig. 5A and B, the mutants can be clearly differentiated from Nt.BstNBI. The cleavage of the upper DNA strand by Nt.BstNBI the corresponding wild types at an abundance as low as 0.1% and 0.5% generated a single-stranded DNA whose sequence is the same as that (S/N 3), respectively. The accuracy of the detection of low abundance of the signal probe. This single-stranded DNA can act as a new primer mutant in PCR products is always limited by genomic diversity within a to trigger the amplification circle. The condition of EXPAR was single cell and replication errors introduced by polymerase. Thus, optimized first by using 1 pM synthetic signal probe as primer (Fig. single cell sequencing and high-fidelity polymerases are useful to S8). By using EXPAR as the downstream assay of DSA, 5 amol mutant enhance the confidence of post-PCR analysis. can be detected within 1 h (bottom, Fig. 4B). As expected, the wild type Table 1 summarizes the performance of some current methods does not show typical amplification curve. The DSA based on G-rich including commercial available approaches (i.e., NGS, allele specific probe offers a broad dynamic range from 0.1 fM to 0.1 pM and PCR and locked nucleic acid probe) for SNM detection. As can be seen, detection limit (S/N 3) of 0.1 fM (Fig. S9). the DSA's sensitivity for low abundance mutation can be compared with other methods and DSA is applicable for cell line samples. More importantly, DSA is low-cost and multiple signaling which are attrac- 3.4. Detection of cancer related SNM at low abundance tive for POC diagnosis.

Detecting mutation at low abundance is crucial for early stage diagnosis because of the low amount of mutant gene in clinical 4. Conclusion samples, particularly in biofluids (Diehl et al., 2008; Ho et al., 2014). To demonstrate the ability of DSA for detecting low abundance In summary, a highly selective method for SNM detection has been mutation, we tested BRAF-D594G (c.1781A > G) and BRAF-V600E developed by constructing dynamic sandwich configuration on mag- (c.1799T > A) mutations which are associated with various types of netic beads. By taking the advantages of the mismatch sensitive cancer (De Roock et al., 2009; Sanchez-Torres et al., 2013)by hybridization thermodynamics of transient DNA binding, high single- fluorescently labeled probes. The fluorescent signal probes for D594G nucleotide discrimination factor was achieved over a broad range of and V600E mutation are 10-nt and 11-nt, respectively, and sodium salt concentration at room temperature. By investigating the DNA concentration for both assays is 400 mM. As shown in Fig. S10, DSA hybridization on particle surface, we found that the stability of single- exhibits high single-nucleotide selectivity for BRAF mutations (syn- nucleotide mismatched duplex can be selectively weakened by the high thetic oligonucleotides). Next, we applied DSA to detect the mutations density negative charge on particle surface. This observation is useful in genomic DNA from cancer cell lines. Mutant and wild-type targets of for nanomaterials based nucleic acid biosensors and holds great

309 J. Wang et al. Biosensors and Bioelectronics 94 (2017) 305–311

Fig. 4. Multiple-mode downstream assays of DSA. (A) Colorimetric detection of mutant. The signal probe in DSA contains G-rich sequence which can trigger the hemin catalyzed oxidation of ABTS. As a result, for mutant, the second supernatant exhibits unique green color. 1 μM capture probe, 150 nM targets, and 200 nM G-rich probe were used. (B) Isothermal amplification. The signal probe in DSA contains the sequence which can be used as the primer of EXPAR. 1 μM capture probe, 0.1 pM targets and 10 pM EXPAR signal probe were used. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) potential for tuning the rate of DNA hybridization based nanoparticle fluorescent materials with better brightness. Due to the merits above, assemble. With this method, BRAF-D594G and BRAF-V600E muta- this method would find broad applications in sandwich assay based tions from A375 cell lines were detected with high-confidence down to biosensors and dynamic assemble of nanomaterials. low abundance of 0.1% and 0.5%, respectively. The sensitivity can be further enhanced by coupling with microfluidic devices or other

Fig. 5. Detection of BRAF mutations. Relationship between the fluorescence intensity and the logarithm of the fractions of mutants, (A) BRAF-D594G, and (B) BRAF-V600E. Insets: linear relationship between fluorescence intensity and the logarithm of the fraction of mutants. The concentrations of capture probe and fluorescent signal probe are 1 μM and 100 nM, respectively. The total concentration of targets is fixed at 100 nM, the fraction of mutants is varied.

310 J. Wang et al. Biosensors and Bioelectronics 94 (2017) 305–311

Table 1 The comparison of some current methods for SNM detection.

Methods Limit of detection of SNM (%) Samples Ref

Allele specific PCR 0.1–0.8 Cell line and clinical samples Paredes et al. (2007) Locked nucleic acid probe 0.1–1 Clinical samples Guha et al. (2013) Enzyme assisted discrimination 0.05–1 Synthetic oligonucleotide Xiao et al. (2012) Simulation-guided probe 1 Cell line Wang and Zhang (2015) NGS 0.1–10 Cell line and clinical samples Forshew et al. (2012) DSA 0.1–0.5 Cell line The present work

Acknowledgements Johnson-Buck, A., Nangreave, J., Kim, D.N., Bathe, M., Yan, H., Walter, N.G., 2013. Nano Lett. 13 (2), 728–733. Johnson-Buck, A., Su, X., Giraldez, M.D., Zhao, M.P., Tewari, M., Walter, N.G., 2015. This work was supported by the National Natural Science Nat. Biotechnol. 33 (7), 730–732. Foundation of China (31600687, 31400915, 21606013, 21405045, Jungmann, R., Steinhauer, C., Scheible, M., Kuzyk, A., Tinnefeld, P., Simmel, F.C., 2010. Nano Lett. 10 (11), 4756–4761. 21522502), Fundamental Research Funds for the Central Universities Jungmann, R., Avendano, M.S., Woehrstein, J.B., Dai, M., Shih, W.M., Yin, P., 2014. Nat. (12060070031, 12060090029 and ys12060026025), Program for New Methods 11 (3), 313–318. Century Excellent Talents in University (China, NCET-13-0789), and Jungmann, R., Avendano, M.S., Dai, M., Woehrstein, J.B., Agasti, S.S., Feiger, Z., Rodal, – Hunan Natural Science Funds for Distinguished Young Scholar A., Yin, P., 2016. Nat. Methods 13 (5), 439 442. Kathiresan, S., Srivastava, D., 2012. Cell 148 (6), 1242–1257. (14JJ1017). Khripin, Y., Myakishev, M.V., Hu, S., Lueders, K., Hamer, D., 2000. Clin. Chem. 46 (11), (1877-1877). – Appendix A. Supplementary material LaFramboise, T., 2009. Nucleic Acids Res. 37 (13), 4181 4193. Li, R.D., Yin, B.C., Ye, B.C., 2016. Biosens. Bioelectron. 86, 1011–1016. Li, T., Wang, E., Dong, S., 2009. J. Am. Chem. Soc. 131 (42), 15082–15083. Supplementary data associated with this article can be found in the Liu, N.N., Jiang, Y.N., Zhou, Y.H., Xia, F., Guo, W., Jiang, L., 2013. Angew. Chem. Int. online version at doi:10.1016/j.bios.2017.03.023. Ed. 52 (7), 2007–2011. Ma, D.L., Guan, J.W., Normandin, F., Denommee, S., Enright, G., Veres, T., Simard, B., 2006. Chem. Mater. 18 (7), 1920–1927. References Nakano, S., Fujimoto, M., Hara, H., Sugimoto, N., 1999. Nucleic Acids Res. 27 (14), 2957–2965. Owczarzy, R., Moreira, B.G., You, Y., Behlke, M.A., Walder, J.A., 2008. Biochemistry 47 – Ang, Y.S., Yung, L.Y.L., 2012. ACS Nano 6 (10), 8815 8823. (19), 5336–5353. – Chen, C.L., Wang, W.J., Ge, J., Zhao, X.S., 2009. Nucleic Acids Res. 37 (11), 3756 3765. Pallela, R., Chandra, P., Noh, H.B., Shim, Y.B., 2016. Biosens. Bioelectron. 85, 883–890. – Chen, X.J., Ying, A., Gao, Z.Q., 2012. Biosens. Bioelectron. 36 (1), 89 94. Paredes, R., Marconi, V.C., Campbell, T.B., Kuritzkes, D.R., 2007. J. Virol. Methods 146, Das, J., Ivanov, I., Montermini, L., Rak, J., Sargent, E.H., Kelley, S.O., 2015. Nat. Chem. 136–146. – 7 (7), 569 575. Rocha-Santos, T.A.P., 2014. TrAC-Trend Anal. Chem. 62, 28–36. De Roock, W., Biesmans, B., De Schutter, J., Tejpar, S., 2009. Mol. Diagn. Ther. 13 (2), Sanchez-Torres, J.M., Viteri, S., Molina, M.A., Rosell, R., 2013. Transl. Lung Cancer Res. – 103 114. 2 (3), 244–250. Diehl, F., Schmidt, K., Choti, M.A., Romans, K., Goodman, S., Li, M., Thornton, K., Schmitt, T.J., Knotts, T.A.T., 2011. J. Chem. Phys. 134 (20), 205105. Agrawal, N., Sokoll, L., Szabo, S.A., Kinzler, K.W., Vogelstein, B., Diaz, L.A., 2008. Schmitt, M.W., Kennedy, S.R., Salk, J.J., Fox, E.J., Hiatt, J.B., Loeb, L.A., 2012. Proc. – Nat. Med. 14 (9), 985 990. Natl. Acad. Sci. USA, 109, 36, pp. 14508–14513. Ding, L.L., Ge, J.P., Zhou, W.Q., Gao, J.P., Zhang, Z.Y., Xiong, Y., 2016. Biosens. Shen, J.W., Li, Y.B., Gu, H.S., Xia, F., Zuo, X.L., 2014. Chem. Rev. 114 (15), 7631–7677. – Bioelectron. 85, 212 219. Sloane, H.S., Kelly, K.A., Landers, J.P., 2015. Anal. Chem. 87 (20), 10275–10282. – Dupuis, N.F., Holmstrom, E.D., Nesbitt, D.J., 2013. Biophys. J. 105 (3), 756 766. Tan, Z.J., Chen, S.J., 2006. Biophys. J. 90 (4), 1175–1190. Forshew, T., et al., 2012. Sci. Transl. Med. 4, 136ra68. Van Cauwenberghe, C., Van Broeckhoven, C., Sleegers, K., 2016. Genet. Med. 18 (5), Guha, M., Castellanos-Rizaldos, E., Makrigiorgos, G.M., 2013. PLoS One 8 (6), e67782. 421–430. Ho, S.L., Chan, H.M., Ha, A.W., Wong, R.N., Li, H.W., 2014. Anal. Chem. 86 (19), Wang, J.S., Zhang, D.Y., 2015. Nat. Chem. 7 (7), 545–553. – 9880 9886. Wei, F., Sun, B., Guo, Y., Zhao, X.S., 2003. Biosens. Bioelectron. 18 (9), 1157–1163. Hwang, M.T., Landon, P.B., Lee, J., Choi, D., Mo, A.H., Glinsky, G., Lal, R., 2016. Proc. Wu, H., Liu, Y.L., Wang, H.Y., Wu, J., Zhu, F.F., Zou, P., 2016. Biosens. Bioelectron. 81, – Natl. Acad. Sci. USA, 113, 26, pp. 7088 7093. 303–308. Jia, H.X., Li, Z.P., Liu, C.H., Cheng, Y.Q., 2010. Angew. Chem. Int. Ed. 49 (32), Xiao, X.J., Zhang, C., Su, X., Song, C., Zhao, M.P., 2012. Chem. Sci. 3 (7), 2257–2261. – 5498 5501. Xu, J., Craig, S.L., 2005. J. Am. Chem. Soc. 127 (38), 13227–13231. Jiang, Y.N., Liu, N.N., Guo, W., Xia, F., Jiang, L., 2012. J. Am. Chem. Soc. 134 (37), Yu, C.J., Wu, S.M., Tseng, W.L., 2013. Anal. Chem. 85 (18), 8559–8565. 15395–15401.

311

Received: 5 July 2017 | Accepted: 5 October 2017 DOI: 10.1002/jcp.26255

ORIGINAL RESEARCH ARTICLE

Fatty acid elongase7 is regulated via SP1 and is involved in lipid accumulation in bovine mammary epithelial cells

Si Chen1 | Zhigang Hu1 | Hua He1,2 | Xiaolin Liu1

1 Shaanxi Key Laboratory of Molecular Biology for Agriculture, College of Animal Science and Fatty Acid Elongase 7 (ELOVL7) is the newly discovered protein on human that Technology, Northwest A&F University, catalyzes the rate-limiting step towards the synthesis of very long-chain fatty acids and Yangling, Shaanxi, P.R. China 2 College of Veterinary Medicine, Northwest exhibits the highest activity toward C18: 3 (n-3) acyl-CoAs, which is the precursor of A&F University, Yangling, Shaanxi, P.R. China eicosapentaenoic acid (EPA, 20: 5n-3). However, in ruminants, an overall

Correspondence understanding of ELOVLs gene family and the transcriptional regulation of ELOVL7 Xiaolin Liu, Email: College of Animal Science remain unknown. The purpose of this study is to investigate the transcriptional and Technology, Northwest A&F University, Yangling, Shaanxi 712100, P.R. China. regulation and the influence of bovine ELOVL7 in bovine mammary epithelial cells Email: [email protected] (bMECs). Quantitative real-time PCR analysis demonstrated that ELOVLs gene family

Funding information had differential expression patterns in bMECs, and bovine ELOVL7 was expressed in a The First-Class General Financial Grant from tissue-specific manner, which was high in kidney, followed by in abdominal fat and in the China Postdoctoral Science Foundation, Grant number: 2015M570856; The National bMECs. Promoter analysis of bovine ELOVL7, including bioinformatics analyzes, dual- 12th“Five-Year”National Science and luciferase reporter assays, protein pull-down assay, Western blot assay, over- Technology Key Project, Grant number: 2011AA100307; The Science and expression and RNA interference assay, have independently and synthetically Technology funds from Northwest A&F demonstrated that transcription factor Sp1 (SP1) specifically interacted with the University, Grant number: A2990215123; − − The First-Class General Financial Grant from GC-box at 143 to 128 base pair on ELOVL7 promoter. Furthermore, the exogenous the Shaanxi Province Postdoctoral Science α-linolenic acid (ALA, 18: 3n-3), strengthened the binding of SP1 to the ELOVL7 Foundation, Grant number: 2016BSHYDZZ44; The Sci-Tech Integrated proximal promoter, resulting in the accumulation of lipid droplets in bMECs. Innovation Engineering Projects of Shaanxi In conclusion, these data suggest that the transcription of bovine ELOVL7 is affected Province, Grant number: 2015KTCL02-11 by the binding of SP1 and the treatment of ALA, moreover, enlightening us the profound role of SP1 in modulating lipid synthesis of the mammary gland in cattle.

KEYWORDS bovine mammary epithelial cells, ELOVL7, lipogenesis, SP1, transcriptional regulation

1 | INTRODUCTION disease (Jump, Depner, & Tripathy, 2012), neurological disorders (Janssen & Kiliaan, 2014), cancers (Schwab et al., 2014), type I diabetes Milk, which is an appropriate nutrition for mammals, is synthesized and mellitus (Lewis et al., 2017), and neovascular eye diseases (Gong et al., secreted in mammary epithelial cells (MECs) via the activation of 2017). Therefore, increasing the amount of LC-PUFAs not only improve hormonal signals (Osorio, Lohakare, & Bionaz, 2016). Fatty acid profiles nutrition value but also upgrade product value of bovine milk. between human breast milk and bovine milk are distinct. Bovine milk In the ruminant mammary gland, the short- and medium-chain contains higher amounts of saturated fatty acids and fewer unsaturated fatty acids (SMFAs) were synthesized via the de novo synthesis fatty acids (Qian et al., 2016). Over the past half-century, the advantages pathway, and long-chain fatty acids (LCFAs) were consecutively of the long-chain polyunsaturated fatty acids (LC-PUFAs) for human catalyzed from C18 precursors through a series of elongation and have been extensively investigated. Increased consumption of desaturation reactions (Palmquist, 2006). The very long-chain fatty LC-PUFAs has health benefits for diseases, including cardiovascular acid elongase (ELOVLs) family is required for the rate-limiting step in

J Cell Physiol. 2017;1–11. wileyonlinelibrary.com/journal/jcp © 2017 Wiley Periodicals, Inc. | 1 2 | CHEN ET AL. the elongation cycle of LCFAs syntheses (Denic & Weissman, 2007; Thus, in this case, for better understanding the mechanism Leonard, Pereira, Spreche, & Huang, 2004). Thus far, seven ELOVL underlying the regulation of ELOVL7 in bovine mammary epithelial cells family members (ELOVL1-7) are identified in mammals and display (bMECs), we have conducted a serious of experiments to illustrate the tissue-specific expression or specific fatty acid substrate selectivity transcriptional regulation of ELOVL7,aswellastheroleofELOVL7 in the (Guillou, Zadrave, Martin, & Jacobsson 2010). ELOVL2, ELOVL4, and process of lipid accumulation. We indicated that ELOVLs showed distinct ELOVL5 contribute to the elongation of PUFAs (Castro, Tocher, & expression levels in bMECs and the expression profile of bovine ELOVL7 Monroig, 2016). However, the in vitro elongase assays indicate that mRNA was found to be tissue-specific pattern, not ubiquitous. We ELOVL7 is an endoplasmic reticulum-bound enzyme with the highest identified that SP1 bound to the proximal promoter of bovine ELOVL7. activity toward C18 acyl-CoAs, especially C18: 3 (n-3) acyl-CoAs and Furthermore, we added the ELOVL7 potential substrate α-linolenic acid C18: 3 (n-6) acyl-CoAs (Naganuma, Sato, Sassa, Ohno, & Kihara, 2011). (ALA, 18: 3n-3) and found that ALA advanced the binding of SP1 to With the development of molecular biology techniques, it became ELOVL7 promoter, indicating that ALA increased ELOVL7 promoter evident that phenotypic characteristics can be affected by the DNA activity and gene expression via SP1 activation in bMECs. Taken transcriptional regulation. In order to regulate the synthesis of LC- together, we provided a wealth of knowledge about the molecular PUFAs in bovine milk, it is essential to investigate the role and the mechanism of the transcriptional regulation and the effect of ELOVL7 in regulation of ELOVL7 in the ruminant mammary gland. bMECs, which provides new insights in modulating the lipid metabolism The transcriptomics analysis on human organs and tissues reveal and some clues for unraveling the transcriptional regulatory network. that ELOVL7 is expressed at high levels in prostate, skin, gall bladder, thyroid, and small intestine, whereas it is expressed low in bone marrow, 2 | MATERIALS AND METHODS heart, liver, ovary, and spleen (Fagerberg et al., 2014). However, the tissue expression profile of ELOLV7 in ruminant remains unknown. Data All animal procedures were carried out in accordance with the from oligodendrocytes studies suggest that elevated expression of Regulations for the Administration of Affairs Concerning Experimental ELOVL7 triggers lipid accumulation in differentiated adipocytes, which Animals (Ministry of Science and Technology, China, 2004) and were results in demyelination and oxidative damage, then inducing inflamma- approved by the Institutional Animal Care and Use Committee at the tion and neuronal degeneration (Shin, Shin McManus, Ptacek, & Fu, Northwest A&F University. Cattles were raised under free food intake 2009). Given the importance of ELOVL7 among varied physical and humanely slaughtered in the Shannxi Kingbull Animal Husbandry conditions, it is essential to clarify the regulation of ELOVL7 in the Company, Ltd (Baoji, Shaanxi, China). process of lipid metabolism in mammary gland. In eukaryotic cells, genes are regulated primarily at the transcriptional level via the recruitment of transcription factors to the DNA sequences. Limited researches have 2.1 | Phylogenetic analysis indicated that ELOVL7 was activating by hormonal status, including insulin and androgen via sterol regulatory element binding transcription ELOVLs coding sequences in eight species were retrieved and factor 1 (SREBF1) (Rodriguez-Cruz, Sanchez Gonazalez, Sanchez Garcia, downloaded from GenBank databases (https://www.ncbi.nlm.nih. & Lopez-Alarcon, 2012; Tamura et al., 2009). However, our previous gov/sites/batchentrez). The detailed information is available in research discovered that transcription factor Sp1 (SP1) bound at the Supplementary Table S1. For the phylogenetic analysis, 56 coding promoter of ELOVL6 to regulate its transcription (Chen, He, & Liu, 2017), sequences were aligned with the ClustalW method. The phylogenetic which enlightened us the similar regulation pattern for the expression of tree was constructed by using the MEGA program (version 6.0) with ELOVL7. the Maximum Likelihood method. Bootstrap values were obtained SP1, the member of Sp/KLF transcription factor family, is viewed as from 1,000 repetitions. The phylogenetic tree was visualized by a basal transcription factor that participates in regulating the expression TreeView software (version 1.6.6). of genes associated with a wide range of cellular processes in mammalian cells (Bajpai & Nagaraju, 2017; Suske, 2017). SP1 has its 2.2 | Quantitative real-time PCR (qPCR) own specific sequence preference, and preferentially binds to GC-rich regions (Jeon et al., 2012; Lu & Archer, 2010). The C-terminal The expression of ELOVLs gene family in bMECs and the tissue multimerization domain of SP1 mediates super-activation of the expression profiles of ELOVL7 in bovine were detected by quantitative promoter containing the consensus site GGGCGG (Beishline & real-time PCR (qPCR). Nine tissues, including heart, liver, spleen, lung, Azizkhan-Clifford, 2015). Interestingly, the majority of the GC and kidney, intestine, stomach, skeletal muscle, and abdominal fat, were CCAAT box-containing promoters lack the TATA box, which indicates collected from three 2-year-old male Qinchuan cattle. Collected that SP1 plays a central role in regulation of the TATA-less genes tissues were stored at −80°C until RNA extraction. Total RNA was (Benner et al., 2013). Study on cancer cells demonstrates that SP1 isolated using TRIzol reagent (Invitrogen, Carlsbad, CA), according to regulates de novo lipogenesis by activating the expressions of fatty the manufacturer's instructions. Subsequently, total RNA (2 μg) was acid synthase (FAS) and sterol regulatory element-binding reverse-transcribed with All-In-One RT MasterMix (ABM, Richmond, protein-1c (SREBP-1c), which are thought to be the critical lipogenic CA). qPCR was performed with a CFX96 Real-Time PCR Detection genes in lipid metabolism (Lu & Archer, 2010). System (Bio-Rad) utilizing the SYBR® Premix Ex Taq II (Takara, Otsu, CHEN ET AL. | 3

Japan). The Cq values were normalized to reference gene (GAPDH) ran bMECs were obtained after three passages. bMECs were grown in on the same plate. Each sample was run in triplicates. The primers 1640 (Hyclone, GE Healthcare) with 10% fetal bovine serum (FBS) selected for qPCR are presented in Supplementary Table S2. (Gibco, Thermo Fisher Scientific, Inc.) in 5% CO2 and 100% humidity at 37°C. Cells were plated at a density of 3 × 105 cells per well in 48-well plates and incubated for 12 hr until they reached 70–80% confluency. 2.3 | Cloning of ELOVL7 promoter and bioinformatics Confluent cells were allowed to grow for an extra 12 hr before analyses harvesting in PBS for nuclear extract preparation. Cells were collected The bovine ELOVL7 promoter (Accession no. AC_000177.1) was by centrifugation at 1,000 g for 5 min and incubated on ice. Nuclear amplified from bovine genomic DNA, which was extracted from a proteins preparation was followed as the protocol on nuclear and bovine blood sample using phenol-chloroform extraction. Primer cytoplasmic protein extraction kit (Beyotime, Beijing, China). The final sequences were the following: ELOVL7 promoter-forward supernatant was stored at −80°C as nuclear proteins. The quantifica- 5′-CAGCCCCTCCTGCTACTTTCC-3′ and ELOVL7 promoter-reverse tion of nuclear protein was performed by BCA method (Beyotime). 5′-GAGGAGCGGAGCCGAAGC-3′. PCR products were inserted into the T Vector pMD-19 (Takara) and sequenced (Invitrogen).The bovine 2.6 | Coupling of biotinylated DNA to magnetic ELOVL7 promoter was analyzed by promoter prediction program at beads https://www.cbs.dtu.dk/ and https://www.genomatix.de/index.html. The analysis of methylation CpG island was conducted at https://www. The 18 base pairs oligonucleotides representing the corn promoter, urogene.org/cgi-bin/methprimer/methprimer.cgi. The putative tran- biotinylated at the 5-terminal of the sense strand. The biotinylated scription factor binding sites were predicted using online prediction DNA probes were synthesized (Invitrogen) and showed in Figure 4c. servers at https://jaspar.genereg.net/ and https://gene-regulation.com. The biotinylated DNA probes were annealed and immobilized onto 2 μM BeaverBeads Streptavidin (Beaver, Suzhou, China) for 1 hr at room temperature in buffer I (pH 7.5; 10 mM Tris-HCl; 1 mM EDTA; 2.4 | Dual-luciferase reporter assays 1 M NaCl; 0.01% (V/V) Tween-20). 6 μg annealed DNA (about 600 Serial 5′-flanking deletion region of the bovine ELOVL7 promoter were pmole biotin) were immobilized onto 300 μg washed BeaverBeads amplified from E7-T vector and subcloned into the MluI and HindIII Streptavidin. The biotinylated DNA coated beads were washed in sites of the pGL3-Basic vector (E1751, Promega, Madison, WI). The buffer I twice and resuspended in 300 μl buffer II (pH 7.4; PBS; 0.05% primers for constructing the vectors are listed in Supplementary (V/V) Tween-20). Table S3. The integrity of each plasmid was confirmed by DNA sequencing (Invitrogen). A total of 0.25 μg constructed plasmids were 2.7 | Pull-down and Western blots then co-transfected with 0.03 μg pRL-TK plasmid into 3T3-L1 for the dual-luciferase reporter assay. Dual-luciferase reporter assay was A total of 0.4 mg nuclear proteins were added in biotinylated DNA carried out by using Varioskan Flash instrument (Life Technologies probes coated beads and the mixture incubated for 1 hr on ice. The Corporation, Carlsbad, CA). The level of firefly luciferase activity was magnetic separation was performed, and the beads were then washed normalized to renilla luciferase activity. 3T3-L1 (CL-173, ATCC, twice with 0.5 ml buffer II and once with buffer II containing 10 times Manassas, VA) cell lines were maintained in Dulbecco's modified excess of salmon sperm DNA (D1626, Sigma, St. Louis, MO). Finally, eagle's medium (DMEM) (Hyclone, GE Healthcare, UT) which was the beads were washed once with PBS and resuspended in 60 μl 0.1% supplemented with 10% fetal bovine serum (FBS) (Gibco, Thermo SDS. The mixture were boiled for 5 min and beads were separated by

Fisher Scientific Inc, Carlsbad, CA) in 5% CO2 and 100% humidity at magnet. Twenty microliters of pull-down proteins were loaded on 12% 37°C and passaged using standard cell culture techniques. Before polyacrylamide gels for SDS-PAGE and then were transferred to transfection, cells were plated at a density of 3 × 105 cells per well in 0.45 μM polyvinylidene fluoride membranes (Beyotime). Western 48-well plates and incubated for 12 hr until they reached 80–90% blots were performed using rabbit polyclonal antibodies directed confluency. Plasmids described above were transfected by Lipofect- against SP1 (21962-1-AP, Proteintech, Rosemont, IL), and HRP- amine 2000 (Invitrogen) according to the manufacturer's instructions. conjugated affinipure goat anti-rabbit IgG (SA00001-2, Proteintech). After 48 hr transfection, the cells were lysed by 1× passive lysis buffer Signals were generated and detected by the enhanced chemilumines- and assayed for firefly luciferase and renilla luciferase activities cence (ECL) system (Beyotime). (Promega). Each sample was performed in triplicates. The experiments were conducted independently three times. 2.8 | Overexpression and RNA interference of SP1

The coding sequence of SP1 was cloned into the pcDNA3.1 over- 2.5 | Cell culture and nuclear extract preparation expression vector (Invitrogen) between EcoRIandNotIsites.Primer Mammary tissues were collected from a lactating Holstein dairy cow. sequences were the following: SP1 CDS-forward 5′-GGGAATTCATGA Bovine Mammary Epithelial Cells were isolated and cultured as GCGACCAAGATCACTCC-3′ and SP1 CDS −reverse 5′- CTCGA described in a previous study (Zhao, Liu, Zhou, & Liu 2010). Purified GCGGCCGCTCAGAAGCCATTGCCACT-3′. All siRNAs were designed 4 | CHEN ET AL. ccording to online program: https://rnaidesigner.thermofisher.com/. transfections in bMECs (overexpression plasmids and siRNAs) were siRNA-Control sequences were as following, sence: 5′-UUCUCCGAACGU performed with Lipofectamine 2000 (Invitrogen) following the manufac- GUCACGUTT-3′, anti-sence: 5′-ACGUGACACGUUCGGAGAATT-3′. turer's protocol. Transfections were performed 12 hr after seeding of siRNA-SP1 were as following, sence: 5′-GCCAAUAGCUACUCAACA bMECs at a density of 3 × 105 cells per well in 48-well plates, and the effects ATT-3′, anti-sence:5′-UUGUUGAGUAGCUAUUGGCTT-3′.Alltransient induced by the treatments were measured after 24 hr. SP1 over-expression

FIGURE 1 The phylogenetic analysis of ELOVLs gene family. The phylogenetic relationship was analyzed by Maximum Likelihood method (MEGA program version 6.0) utilizing ELOVLs mRNA sequences. Numbers above the lines indicate the branch length. The evolutionary distance of 0.1 nucleic acid substitutions per position represented at the scale bars CHEN ET AL. | 5 plasmids were transfected as 0.1 μgmixedwith0.75μl Lipofectamine 2000 3 | RESULTS per well. siRNAs (20 umol/L) were transfected as 5 pmol mixed with 0.25 μl Lipofectamine 2000 per well. The expression changes of relative genes 3.1 | Phylogenetic analysis of ELOVLs in different were measured by qPCR. Each sample was performed in triplicates. The sepcies experiments were conducted independently three times. To understand potential evolutionary relationship of ELOVLs gene family, a phylogenetic tree was constructed by MEGA program 2.9 | Fatty acid treatment, staining, and absorbance (version 6.0). A total of 56 sequences was obtained from eight species. measurement The constructed phylogenetic tree revealed that ELOVLs family could be broadly divided among four groups according to the sequences’ The bMECs were seeded at a density of 8 × 105 cells per well in similarity (Figure 1). Analysis of the phylogenetic relationship showed 48-well plates. These cells were treated with 100 μM α-linolenic acid that ELOVL7 and ELOVL1 had a closer phylogenetic relationship. (U-62-A, Nu-Chek, Elysian, MN) for 24 hr. In order to observe and Among the ELOVL7 genes, cattle had the closer relationship with goat. compare the formation of fatty acids, we conducted the staining The overall phylogenetic tree indicated the timing of ELOVLs gene with oil red O and Nile red. The bMECs were washed with PBS three family expansion. times, and fixed in 10% (v/v) formalin for 30 min at room temperature. After rinsing with 60% isopropanol, cells were stained for 20 min at room temperature and rinsed with PBS three times. 3.2 | ELOVLs expression pattern in bMECs and The EVOS FL Imaging System (Life Technologies) were used to tissue-specific expression of bovine ELOVL7 capture photomicrographs. The imaging of oil red O staining was taken at 10× magnification, and the Nile red staining were taken at The quantitative PCR results in bovine mammary epithelial cells 20× magnification. Following microscopic analysis, 700 μl 60% indicated that bovine ELOVL6 and ELOVL1 were significantly highly isopropanol was added to dissolve the stain. A volume of 200 μl expressed, the expression of ELOVL7, ELOVL4 and ELOVL5 were were added into 96-well plates for measuring the absorbance at averaged (Figure 2a). These findings suggested that the expression 510 nm. The expression changes of SP1 and ELOVL7 were measured of bovine ELOVLs family in bMECs were distinct. To characterize the by qPCR. Each sample was performed in triplicates. The experiments tissue-specific expression pattern of ELOVL7 mRNA, we performed were conducted independently three times. qPCR analyses using bovine cDNAs of nine different tissues and the bMECs. Bovine ELOVL7 exhibited high endogenous expression in a broad range of tissues except liver, stomach and skeletal muscle 2.10 | Statistical analyses (Figure 2b). Expression in kidney was high, which was in agreement All results presented are expressed as mean ± SD. Statistical analyses with the results of a previous study on human (Ohno et al., 2010). were carried out by ANOVA using the software of SPSS 19.0 (IBM, Expression declined in abdominal fat, followed by bMECs, lung, Armonk, NY). To test the effects of treatment, the multipie comparison intestine, heart, and spleen. These results indicated that ELOVL7 were performed by Tukey HSD method. A p value less than 0.05 was exhibited a characteristic tissue-expression pattern, especially high considered statistically significant. in lipid-synthesizing and lipid-secreting organs.

FIGURE 2 The expression profiles of bovine ELOVLs and ELOVL7 mRNA. (a) The mRNA expression patterns of bovine ELOVLs in bMECs. The data were normalized to the expression of ELOVL7 and calculated by using 2−△△Ct. (b) Tissue-specific expression of bovine ELOVL7. The data were normalized to the expression in heart and calculated by using 2−△△Ct. GAPDH was measured as the reference gene. Columns are represented as mean ± SD for three independent experiments performed in triplicate 6 | CHEN ET AL.

3.3 | Bioinformatics analyses and 5′-deletion analysis AC_000177.1). Utilizing the MatInspector program (Genomatix), we of bovine ELOVL7 promoter analyzed the cloned promoter fragment consisting 1,000 bp upstream of the transcription start site (TSS). Promoter analysis by prediction The 5′-flanking region of bovine ELOVL7 promoter was amplified and software identified neither a TATA box nor a CAAT box at 1000 bp was identical with the GenBank database (Accession no. upstream of TSS. However, several GC-rich regions were found on the

FIGURE 3 Bioinformatics analyses and 5′-deletion analysis of bovine ELOVL7 promoter. (a) The predicted CpG islands in the bovine ELOVL7 promoter (+1 to −1000 bp). The red vertical illustrates the GC-rich regions. The blue shading regions indicates the predicated CpG islands. (b) Deletion analysis of bovine ELOVL7 promoter in bMECs. Relative luciferase activity (firefly: renilla) after 48-hr transfection with different constructs. The promoter activities of 3T3-L1 cells (black bars) were shown as the mean ± SD of three independent experiments performed in triplicate. The arrow indicates the transcription initiation site and the direction of transcription. (c) Promoter activities of different deletion constructs. Relative luciferase activities of different deletion constructs were indicated as mean ± SD of three independent experiments performed in triplicate. Left schematic represents various deletion constructs in the core region of bovine ELOVL7 promoter CHEN ET AL. | 7 promoter. The online program MethPrimer revealed two predicted CpG sites between −145 to −128bp promoter region (Figure 4a). The CpG islands within the promoter region of the bovine ELOVL7 results demonstrated that one SP1 binding sites and two CpG sites (Figure 3a). In order to narrow down the core region of bovine ELOVL7 were overlapped in this proximal promoter region. According to the promoter, we constructed a series of pGL3 reporter plasmids relative luciferase activity, the activity of pGL3-P7 was higher with the (designated as P1 to P9), containing various 5′-ends and a common co-transfection of pcDNA3.1-SP1 (Figure 4b). In addition, all the 3′-end (+ 63) of bovine ELOVL7 promoter, and then co-transfected promoter activities were elevated by ALA treatment (Figure 4b), with pRL-TK plasmid into 3T3-L1 cell line for dual-luciferase reporter indicating the potential regulation mechanism of ALA via SP1 in assay. In 3T3-L1 cell line, the dual-luciferase reporter assay showed bMECs. Then we synthesized −145 to −128 bp region as that bovine ELOVL7 promoter activity gave a “down-up-down” pattern 5′-biotinylated sense strands to pull down interacting nuclear proteins along with the piecewise truncation (Figure 3b). The significant from bMECs nuclear extracts (Figure 4c). Western blots analysis decrease of promoter activity in pGL3-P8 construct (−108/+ 63bp) demonstrated that SP1 bound to our biotinylated-DNA probe, which which was compared with pGL3-P7 construct (−143/+ 63bp) may be contains the −145 to −128bp promoter region, suggesting that SP1 attributed to the loss of cis elements contained within the deleted DNA participated in the transcriptional regulation of ELOVL7 through region. To confirm the hypothesis, we constructed three luciferase interacting with its proximal promoter region in bMECs (Figure 4d). reporter plasmids (designated as D1 to D3) to investigate the highest promoter activity region. After measured by dual-luciferase reporter 3.5 | Overexpression and siRNA of SP1 assay, the significant change in promoter activity was found between the deletion construct of D1 and D2, suggesting that this −143 to −128 To independently confirm the above results, the overexpression and may have the positive and dominant transcriptional regulators for knockdown of SP1 assays in bMECs were conducted and the bovine ELOVL7 promoter (Figure 3c). expression of bovine ELOVL7 was detected by relative quantitative PCR. Compared with the control group, the expression level of SP1 was significantly higher (p < 0.05) after 24 hr transfection with the 3.4 | Transcription factor SP1 bind to ELOVL7 pcDNA3.1-SP1 plasmid (Figure 5a). As depicted in Figure 5a, the proximal promoter (−143 to −128) expression level of ELOVL7 also increased as the cells were treated To determine the potential transcription regulation elements within with pcDNA3.1-SP1 plasmid. Additionally, the RNA interference- the corn promoter region, we analyzed the putative binding sites and mediated knockdown of SP1 accompanied with the significantly

FIGURE 4 Western blotting of pull-down proteins. (a) The predicted transcription factors binding sites and CpG sites were underlined. (b) Relative luciferase activity of the ELOVL7 promoter. Gray bar represents the transfection with different constructs. Inner shadow bar represents the 24 hr ALA-treated group. (c) The sequence of biotinylated DNA probe for pull-down assay. (d) Western blot assay verification of SP1 directly bind to biotinylated DNA probe, which is the promoter fragment containing GC box 8 | CHEN ET AL.

FIGURE 5 The effect of SP1 overexpression or inhibition on endogenous expression of ELOVL7 in bMECs. (a) Endogenous SP1 and ELOVL7 expression levels after 24-hours transfection with the pcDNA3.1-SP1 overexpression plasmid. (b) Endogenous SP1 and ELOVL7 expression levels after 24-hr transfection with the siRNA-SP1. GAPDH was measured as the reference gene. Each column represented the mean ± SD of the independent experiment performed in triplicate. *p< 0.05; **p< 0.01

decreased ELOVL7 expression (p < 0.05; Figure 5b). Taken together, erberg, 2006); however, an overall understanding of ELOVLs gene these results independently demonstrated that either overexpression family and the transcriptional regulation in ruminant had not yet been or knockdown of SP1 accompanied by the change of bovine ELOVL7 recognized. In the present study, we first, explore the expression expression. pattern of each ELOVL mRNA in bovine and establish the transcrip- tional regulation of ELOVL7 in bMECs. It is generally accepted that the differences in the transcriptional regulation lead to the change in 3.6 | The treatment of α-linolenic acid increases phenotype. Therefore, we aim to modify the composition of milk, ELOVL7 through activing SP1 especially the content of PUFAs, through the regulation of ELOVL7 The manipulated cells were stained to observe a number of transcription. Extensive experiments demonstrated that mammary cytoplasmic lipid droplets by oil red O and Nile red. The resultes of epithelial cells, which contribute to the synthesis of milk, have well oil red O staning indicated that lipid drops were accumulated after developed fatty acid synthetic, storage and secretion functions transfection with the pcDNA3.1-SP1. Moreover, with the supply of (McManaman, 2012; Oppi-Williams, Suagee, & Corl, 2013). Mammary ALA, these changes were significantly raised (Figure 6a). The results of epithelial cells were viewed as a model to investigate the molecular and Nile red showed that the volume and number of lipid drops were cellular mechanisms of lipid metabolism, so we choose bMECs to study increased in bMECs which were consistent with the oil red O staning the transcriptional regulation of bovine ELOVL7. results (Figure 6a). The absorbance value which represented the According to the qPCR results, ELOVLs gene family shows a amount of fatty acids were in line with the staining results (Figure 6b). distinct expression pattern in bMECs. ELOVL6 and ELOVL1, which The phenotypic consequences demonstrated that lipid droplets were contribute to the elongation of C18:0 and C22:0 respectively (Ohno accumulated in ALA-treated cells compared with non-treated cells. To et al., 2010), exhibited the highest expression levels in bMECs. investigate the underlying molecular mechanism, we performed qPCR However, ELOVL7, which catalyzed the elongation of the PUFAs analyses using cDNAs of manipulated cells. The results indicated that precursor α-linolenic acid (ALA), expressed low mRNA level in bMECs. α-linolenic acid promoted the expression of SP1 and up-regulated the It has already been proven that the characteristic of bovine milk is high mRAN levels of downstream ELOVL7 in bMECs (Figure 6c). Moreover, in SFA and low in MUFA and PUFA (Barreiro, Regal, Diaz-Bao, we conducted qPCR to investigate the expression levels of ELOVLs Vazquez, & Cepeda, 2017), which is consistent with the expression gene family. Interestingly, the expression of other ELVOLs were patterns of bovine ELOVLs in our study. The tissue expression profile of reduced after the treatment of ALA, which revealed other potential ELOVL7 has indicated that bovine ELOVL7 was expressed at the regulation among ELOVLs family via ALA (Figure 6d). Taken together, highest level in kidney, followed by in abdominal fat and in bMECs. The the addition of ELOVL7 substrate ALA induced the accumulation of level of ELOVL7 mRNA was rarely detected in liver, stomach and lipid droplets through activating the lipogenic transcription factors and skeletal muscle. This pattern was in line with the function of ELOVL7, mobilizing downstream lipogenic genes. which mainly involved in modulating the lipid metabolism. In order to explore the transcriptional regulation of ELOVL7,we conducted comprehensive molecular biology experiments in bMECs, 4 | DISCUSSION including dual-luciferase reporter assays, pull-down, Western blots, overexpression and siRNA interference. Concerns about the neurotox- The ELOVL family have been studied extensively on human and mouse icity of ethidium bromide and radioactively labeled molecules, the biotin- (Guillou, Zadravec, Martin, & Jacobsson; 2010; Jakobsson, & West- labelled DNA probes were used to capture the transcription factor and CHEN ET AL. | 9

FIGURE 6 Staining, absorbance and qPCR results after the treatment of 100 uM α-linolenic acid in bMECs. (a) Oil red O and Nile red staining of bMECs treated with 100 uM α-linolenic acid. The stained lipid droplets are shown in red. (b) Quantification of the accumulated lipid in manipulated bMECs. The dye was extracted by 60% isopropanol, and absorbance was measured at 510 nm. Sixty percent of isopropanol was set as control. Each triangle represented the mean ± SD of three independent experiments, which were performed in triplicate. (c) The relative expression levels of SP1 and ELOVL7 in the manipulated bMECs. (d) The relative expression levels of ELOVLs in the manipulated bMECs. Each column represented the mean ± SD of three independent experiments which were performed in triplicate. * p <0.05;**p <0.01 isolated by streptavidin-modified magnetic particles (Ben Aissa, Herrera- to the GC rich region in gene promoter, SP1 inhibits the methylation of Chacon, Pupin, Sotomayor, & Pividori, 2017). All the results clearly CpG islands and activities the gene expression (O'Connor, Gilmour, & demonstrated that SP1 enhanced the expression of bovine ELOVL7 via Bonifer, 2016). Our research also indicated that the SP1 binding site and the interaction with its proximal promoter region in bMECs. the CpG site appeared to overlap in −143 and −128bp promoter region. SP1, which belongs to the SP/KLF family, has the typical DNA- It is evident that PUFAs not only maintain energy production binding domain of three consecutive Cys2His2-type zinc fingers (Suske, throughout the body but also regulate the biological processes in cells. 2017). The SP1-medatied transcription is ubiquitous throughout cell The recently published article demonstrated that eicosapentaenoic development, differentiation and tumorigenesis. SP1 functions by acid (EPA, 20:5n-3) suppressed SELENOP gene expression by interacting with TATA-box binding protein (TBP) and recruiting RNA inactivating its transcriptional regulation (Tajima-Shirasaki et al., polymerase II (Vizcaino, Mansilla, & Portugal, 2015). It has been 2017). In mammal, it has been identified that α-linolenic acid (ALA, increasingly clear that the activity of TBP is required either at TATA- 18:3n-3) is initially converted to stearidonic acid (SDA, 18: 4n-3) by the containing or at TATA-less promoters (Tora & Timmers, 2010). Δ6-desaturase (FADS2), and further is elongated and desaturated into Nowadays, many researches have already established that SP1 played EPA (Burdge & Calder, 2006). Moreover, it is reported that ALA is one an essential role in transcriptional regulation of TATA-less gene of the ELOVL7 potential substrate (Naganuma et al., 2011), which promoters. The analysis of bovine ELOVL7 promoter showed that there enlightens us may be another pathway for the regulation of ELOVL7 was no TATA box in its promoter, which was accordance with previous transcription. In order to illustrate the regulation of bovine ELOVL7,we reports. In addition, SP1 is required for the maintenance of the manipulated the bMECs with the treatment of 100 μm ALA. The methylation free CpG islands in target gene promoters. Through binding results of dual-luciferase reporter assays turned out that ALA 10 | CHEN ET AL. increased the binding of SP1 to the ELOVL7 promoter (Figure 4b). Barreiro, R., Regal, P., Diaz-Bao, M., Vazquez, B I., & Cepeda, A. (2017). Further qPCR assays indicated that SP1 and ELOVL7, which associated Effects of bovine pregnancy on the fatty acid composition of milk: The significance for humans needs. Food Additives & Contaminants with the synthesis of LC-PUFAs, were up-regulated by exogenous ALA Part A, Chemistry, Analysis, Control, Exposure & Risk Assessment, 34(4), in bMECs (Figure 5c). On the contrary, the expression of other ELOVLs 608–616. genes were attenuated with the treatment of ALA (Figure 5d), which Beishline, K., & Azizkhan-Clifford, J. (2015). Sp1 and the ‘hallmarks of needs further work to elucidate the potential mechanism. cancer’. The FEBS Journal, 282(2), 224–258. Ben Aissa, A., Herrera-Chacon, A., Pupin, R R., Sotomayor, M D., & Pividori, With the change of ELOVL7 transcriptional, we examined the M I. (2017). Magnetic molecularly imprinted polymer for the isolation phenotype of bMECs. Utilizing the staining of oil red O and Nile red, and detection of biotin and biotinylated biomolecules. Biosensors & the lipid droplets were accumulated in both the SP1-overexpression Bioelectronics, 88, 101–108. group and the ALA-treated group. The absorbance values were in line Benner, C., Konovalov, S., Mackintosh, C., Hutt, K. R., Stunnenberg, R., & with the staining results. We conjectured that both ALA and SP1 Garcia-Bassets, I. (2013). Decoding a signature-based model of transcription cofactor recruitment dictated by cardinal cis-regulatory increased the expression of ELOVL7, which catalyzes the fatty acid elements in proximal promoter regions. PLoS Genetics, 9(11), e1003906. elongation, resulting in the accumulation of lipid droplets. Burdge, G. C., & Calder, P. C. (2006). Dietary alpha-linolenic acid and health- related outcomes: a metabolic perspective. Nutrition Research Reviews, 19(1), 26–52. 5 | CONCLUSION Castro, LF., Tocher, D R., & Monroig, O. (2016). Long-chain polyunsaturated fatty acid biosynthesis in chordates: Insights into the evolution of Fads and Elovl gene repertoire. Progress in Lipid Research, 62,25–40. In conclusion, SP1 binds to the bovine ELOVL7 promoter and activities Chen, S., He, H., & Liu, X. (2017). Tissue expression profiles and the expression of ELOVL7 in bMECs. Moreover, the supply of ALA transcriptional regulation of elongase of very long chain fatty acid 6 strengthen the binding of SP1 to the ELOVL7 proximal promoter, which in bovine mammary epithelial cells. PLoS ONE, 12(4), e0175777. leading to the accumulation of lipid droplets in bMECs. Our previous Denic, V., & Weissman, JS. (2007). A molecular caliper mechanism for – research discovered that SP1 bound at the promoter of bovine ELOVL6 determining very long-chain fatty acid length. Cell, 130(4), 663 677. Fagerberg, L., Hallstrom, BM., Oksvold, P., Kampf, C., Djureinovic, D., to regulate its transcription (Chen, He, & Liu, 2017), which were proven Odeberg, J., ...Uhlen, M. (2014). Analysis of the human tissue-specific to be the similar regulation pattern for the expression of ELOVL7 in the expression by genome-wide integration of transcriptomics and antibody- present study. These results illustrate that ELOVL7 appear to play an based proteomics. Molecular & Cellular Proteomics, 13(2), 397–406. important role in lipogenesis, and extend our knowledge of bovine Gong, Y., Fu, Z., Liegl, R., Chen, J., Hellstrom, A., & Smith, LE. (2017). Omega- 3 and omega-6 long-chain PUFAs and their enzymatic metabolites in ELOVL7 in controlling lipid synthesis of mammary gland lipid in cattle, neovascular eye diseases. The American Journal of Clinical Nutrition, in addition, enlightening us the profound role of SP1 involving in the 106(1), 16–26. lipid metabolism. Further studies are required to elucidate the Guillou, H., Zadravec, D., Martin, P. G., & Jacobsson, A. (2010). The key roles relationship between SP1 and other ELOVLs family members. of elongases and desaturases in mammalian fatty acid metabolism: Insights from transgenic mice. Progress in Lipid Research, 49(2), 186–199. Jakobsson, A., Westerberg, R., & Jacobsson, A. (2006). Fatty acid elongases CONFLICTS OF INTEREST in mammals: Their regulation and roles in metabolism. Progress in Lipid Research, 45(3), 237–249. The authors declare that they have no conflicts of interest with the Janssen, CI., & Kiliaan, AJ. (2014). Long-chain polyunsaturated fatty acids (LCPUFA) from genesis to senescence: The influence of LCPUFA on contents of this article. neural development, aging, and neurodegeneration. Progress in Lipid Research, 53,1–17. Jeon, BN., Kim, YS., Choi, WI., Koh, DI., Kim, MK., Yoon, JH., ...Hur, MW. AUTHORS’ CONTRIBUTIONS (2012). Kr-pok increases FASN expression by modulating the DNA binding of SREBP-1c and Sp1 at the proximal promoter. Journal of Lipid SC and XL conceived the experiments. SC conducted the experiments, Research, 53(4), 755–766. analyzed the results and wrote the manuscript. ZH and HH contributed Jump, DB., Depner, CM., & Tripathy, S. (2012). Omega-3 fatty acid reagents and materials. All authors have approved the final version of supplementation and cardiovascular disease. Journal of Lipid Research, 53(12), 2525–2545. the manuscript. Leonard, AE., Pereira, SL., Sprecher, H., & Huang, YS. (2004). Elongation of long-chain fatty acids. Progress in Lipid Research, 43(1), 36–54. ORCID Lewis, EJH., Perkins, BA., Lovblom, LE., Bazinet, RP., Wolever, TMW, & Bril, V. (2017). Effect of omega-3 supplementation on neuropathy in type 1 – Xiaolin Liu http://orcid.org/0000-0002-9587-9665 diabetes: A 12-month pilot trial. Neurology, 88(24), 2294 2301. Lu, S., & Archer, MC. (2010). Sp1 coordinately regulates de novo lipogenesis and proliferation in cancer cells. International Journal of Cancer, 126(2), 416–425. McManaman, J L. (2012). Milk lipid secretion: Recent biomolecular aspects. REFERENCES Biomolecular Concepts, 3(6), 581–591. Bajpai, R., & Nagaraju, G P. (2017). Specificity protein 1: Its role in colorectal Naganuma, T., Sato, Y., Sassa, T., Ohno, Y., & Kihara, A. (2011). Biochemical cancer progression and metastasis. Critical Reviews in Oncology/ characterization of the very long-chain fatty acid elongase ELOVL7. – Hematology, 113,1–7. FEBS Letters, 585(20), 3337 3341. CHEN ET AL. | 11

O'Connor, L., Gilmour, J., & Bonifer, C. (2016). The role of the ubiquitously Tajima-Shirasaki, N., Ishii, KA., Takayama, H., Shirasaki, T., Iwama, H., expressed transcription factor sp1 in tissue-specific transcriptional regulation Chikamoto, K., ... Takamura, T. (2017). Eicosapentaenoic acid down- and in disease. The Yale Journal of Biology and Medicine, 89(4), 513–525. regulates expression of the selenoprotein P gene by inhibiting SREBP- Ohno, Y., Suto, S., Yamanaka, M., Mizutani, Y., Mitsutake, S., Igarashi, Y., 1c protein independently of the AMP-activated protein kinase pathway Sassa, T., ...Kihara, A. (2010). ELOVL1 production of C24 acyl-CoAs is in H4IIEC3 hepatocytes. The Journal of Biological Chemistry, 292(26), linked to C24 sphingolipid synthesis. Proceedings of the National Academy 10791–10800. of Sciences of the United States of America, 107(43), 18439–18444. Tamura, K., Makino, A., Hullin-Matsud, F., Kobayashi, T., Furihata, M., Oppi-Williams, C., Suagee, J K., & Corl, B A. (2013). Regulation of lipid Chung, S., ...Nakagawa, H. (2009). Novel lipogenic enzyme ELOVL7 is synthesis by liver X receptor alpha and sterol regulatory element- involved in prostate cancer growth through saturated long-chain fatty binding protein 1 in mammary epithelial cells. Journal of Dairy Science, acid metabolism. Cancer Research, 69(20), 8133–8140. 96(1), 112–121. Tora, L., & Timmers, HT. (2010). The TATA box regulates TATA-binding Osorio, JS., Lohakare, J., & Bionaz, M. (2016). Biosynthesis of milk fat, protein (TBP) dynamics in vivo. Trends in Biochemical Sciences, 35(6), protein, and lactose: Roles of transcriptional and posttranscriptional 309–314. regulation. Physiological Genomics, 48(4), 231–256. Vizcaino, C., Mansilla, S., & Portugal, J. (2015). Sp1 transcription factor: A Palmquist, DL. (2006). Milk fat: Origin of fatty acids and influence of nutritional long-standing target in cancer chemotherapy. Pharmacology & Thera- factors thereon. In P. F. Fox, & P. L. H. McSweeney (Eds.), Advanced dairy peutics, 152, 111–124. chemistry volume 2: Lipids (pp. 43–92). Boston, MA: Springer. Zhao K, Liu HY, Zhou MM, & Liu JX. (2010). Establishment and Qian, L., Zhao, A., Zhang, Y., Chen, T., Zeisel, S, Jia, W., & Cai, W. (2016). characterization of a lactating bovine mammary epithelial cell model Metabolomic approaches to explore chemical diversity of human for the study of milk synthesis. Cell Biology International, 34(7): Breast-Milk, formula milk and bovine milk. International Journal of 717–721. Molecular Sciences, 17(12), 2128. Rodriguez-Cruz, M., Sanchez Gonzalez, R., Sanchez Garcia, AM., & Lopez- Alarcon, M. (2012). Coexisting role of fasting or feeding and dietary SUPPORTING INFORMATION lipids in the control of gene expression of enzymes involved in the synthesis of saturated, monounsaturated and polyunsaturated fatty Additional Supporting Information may be found online in the – acids. Genetics, 496(1), 28 36. supporting information tab for this article. Schwab, U., Lauritzen, L., Tholstrup, T., Haldorssoni, T. I., Riserus, U., Uusitupa, M., & Becker, W. (2014). Effect of the amount and type of dietary fat on cardiometabolic risk factors and risk of developing type 2 diabetes, cardiovascular diseases, and cancer: A systematic review. Food & Nutrition Research, 58(1), 25145. How to cite this article: Chen S, Hu Z, He H, Liu X. Fatty Shin, D., Shin, JY., McManus, MT., Ptacek, LJ., & Fu, YH. (2009). Dicer acid elongase7 is regulated via SP1 and is involved in lipid ablation in oligodendrocytes provokes neuronal impairment in mice. accumulation in bovine mammary epithelial cells. J Cell Annals of Neurology, 66(6), 843–857. Suske, G. (2017). NF-Y and SP transcription factors—New insights in a long- Physiol. 2017;1–11. https://doi.org/10.1002/jcp.26255 standing liaison. Biochimica et Biophysica Acta, 1860(5), 590–597. Article

The Kinase mTORC1 Promotes the Generation and Suppressive Function of Follicular Regulatory T Cells

Graphical Abstract Authors Lifan Xu, Qizhao Huang, Haoqiang Wang, ..., Yuzhang Wu, Xinyuan Zhou, Lilin Ye

Correspondence [email protected]

In Brief To differentiate from Treg cells, follicular regulatory T (Tfr) cells must integrate diverse signals within the germinal center. Xu et al. show that mTORC1 is essential for both the differentiation and functional competency of Tfr cells by activating a transcriptional axis consisting of Stat3-TCF-1-Bcl-6 and thus dictating Tfr cell fate.

Highlights d Tfr cells exhibit elevated mTORC1 signaling compared to Treg cells after stimulation d mTORC1 signaling initiates Treg to Tfr cell differentiation d Differentiated Tfr cells require mTORC1 signaling to perform suppressive effects d mTORC1 regulates Tfr cell differentiation through p-STAT3- TCF-1-Bcl-6 pathway

Xu et al., 2017, Immunity 47, 538–551 September 19, 2017 ª 2017 Elsevier Inc. http://dx.doi.org/10.1016/j.immuni.2017.08.011 Immunity Article

The Kinase mTORC1 Promotes the Generation and Suppressive Function of Follicular Regulatory T Cells

Lifan Xu,1,3 Qizhao Huang,1,3 Haoqiang Wang,1,3 Yaxing Hao,1,3 Qiang Bai,1 Jianjun Hu,1 Yiding Li,1 Pengcheng Wang,1 Xiangyu Chen,1 Ran He,1 Bingshou Li,1 Xia Yang,1 Tingting Zhao,1 Yanyan Zhang,1 Yifei Wang,1 Juanjuan Ou,2 Houjie Liang,2 Yuzhang Wu,1 Xinyuan Zhou,1 and Lilin Ye1,4,* 1Institute of Immunology 2Department of Oncology, Southwest Hospital Third Military Medical University, Chongqing 400038, China 3These authors contributed equally 4Lead Contact *Correspondence: [email protected] http://dx.doi.org/10.1016/j.immuni.2017.08.011

SUMMARY amounts of the transcription factor Bcl-6, the chemokine re- ceptor CXCR5, the co-stimulatory receptor ICOS, and the co- Follicular regulatory T (Tfr) cells differentiate from inhibitory receptor PD-1 (Alexander et al., 2011; Chung et al., conventional regulatory T (Treg) cells and suppress 2011; Linterman et al., 2011; Wollenberg et al., 2011). Resem- excessive germinal center (GC) responses by acting bling Foxp3+ Treg cells, Tfr cells express the transcription factor on both GC B cells and T follicular helper (Tfh) cells. Foxp3 and the receptors associated with Tfr effecter function, Here, we examined the impact of mTOR, a serine/ such as CD25, GITR, and CTLA-4 (Sage and Sharpe, 2015). Tfr threonine protein kinase that senses and integrates cells limit excessive GC reactions and maintain immune homeo- stasis primarily by suppressing Tfh and GC B cells in GC struc- diverse environmental cues, on the differentiation ture (Sage and Sharpe, 2016). and functional competency of Tfr cells in response Inputs from the T cell receptor (TCR), co-stimulatory, and to protein immunization or viral infection. By geneti- inhibitory molecules, including CD28, ICOS, CTLA4, and PD-1, cally deleting Rptor or Rictor, essential components and cytokine receptors, initiate the differentiation of Tfr cells for mTOR complex 1 (mTORC1) and mTOR complex from Treg cells (Chung et al., 2011; Linterman et al., 2011; 2 (mTORC2), respectively, we found that mTORC1 Sage et al., 2013, 2014; Wang et al., 2015). The coordination but not mTORC2 is essential for Tfr differentia- of these early molecular events likely induces the expression tion. Mechanistically, mTORC1-mediated phosphor- of transcription factors required for Tfr differentiation, such as ylation of the transcription factor STAT3 induced Bcl-6 (Chung et al., 2011) and NFAT2 (Vaeth et al., 2014). In addi- the expression of the transcription factor TCF-1 by tion, the expression of SLAM-associated protein (SAP) is promoting STAT3 binding to the Tcf7 50-regulatory increased in Tfr cells and critical to the functional maturation and maintenance of this subset, presumably by mediating the region. Subsequently, TCF-1 bound to the Bcl6 pro- direct interaction of Tfr cells with GC B cells in B cell follicles moter to induce Bcl6 expression, which launched (Linterman et al., 2011). Thus, similar to Tfh differentiation, the Tfr cell differentiation program. Thus, mTORC1 Tfr differentiation is a multistep and multifactor process. The initiates Tfr cell differentiation by activating the mechanisms by which these different signals are integrated to TCF-1-Bcl-6 axis during immunization or infection. dictate the differentiation and function of Tfr cells remain poorly understood. The mechanistic target of rapamycin (mTOR) is an evolution- INTRODUCTION arily conserved serine/threonine kinase that controls various cellular processes, including cell growth, proliferation, and sur- The germinal center (GC) reaction plays a critical role in con- vival, by sensing and integrating environmental cues (Zoncu trolling the invasion of pathogens; however, prolonged and et al., 2011). The mTOR kinase forms two functionally distinct aberrant GC responses are implicated in both autoantibody- complexes: mTOR complex 1 (mTORC1) and mTOR complex mediated autoimmune diseases and B cell malignances (Basso 2 (mTORC2). These complexes share the common catalytic and Dalla-Favera, 2015; Gatto and Brink, 2010). The GC reac- subunit mTOR but have distinct scaffolding subunits, Raptor tion is initiated and amplified by the interaction between GC and Rictor, respectively (Laplante and Sabatini, 2012). Both B cells and follicular helper T (Tfh) cells (Crotty, 2014). A unique mTORC1 and mTORC2 regulate a wide variety of T cell immune subset of Foxp3+CD4+ regulatory T (Treg) cells, follicular regula- responses (Chapman and Chi, 2015; Powell et al., 2012; Xu et al., tory T (Tfr) cells, specializes in repressing GC responses (Sage 2012). Specifically, mTORC1 promotes Th1 and Th17 cell differ- and Sharpe, 2015). Similar to Tfh cells, Tfr cells express large entiation, whereas mTORC2 favors Th2 cell differentiation by

538 Immunity 47, 538–551, September 19, 2017 ª 2017 Elsevier Inc. orchestrating metabolic reprogramming and lineage-specific liferation-indicating dye, followed by transferring labeled cells gene transcription (Delgoffe et al., 2011). The mTORC1 kinase into day 7 OVA/CFA immunized recipients; on day 3 post trans- is reported to potently inhibit the de novo differentiation and fer, we compared the strength of mTORC1 signaling (measured population expansion of Treg cells (Battaglia et al., 2005; by p-S6) between GFP+CXCR5+ Tfr cells and GFP+CXCR5– Treg Delgoffe et al., 2009; Haxhinasto et al., 2008; Liu et al., 2009; cells at different passages of cell division. We noted that, starting Sauer et al., 2008). Paradoxically, however, mTORC1 coordi- from the second passage of cell division when de novo differen- nates cholesterol and lipid biosynthesis to the suppressor func- tiated Tfr cells can be first tracked, mTORC1 activities were tions and homeostasis of activated Treg cells (Zeng et al., 2013). significant higher in Tfr cells than in Treg cells at each passage In addition, the interleukin-2 (IL-2)-mTORC1 signaling axis ap- indicated (Figures S1E and S1F). Moreover, we further divided pears to inhibit Tfh cell differentiation (Ray et al., 2015), whereas donor cells into p-S6high, p-S6intermediate, and p-S6low popula- mTORC2 was reported to be essential for Tfh cell differentiation tions and found that the proportions of CXCR5+ Tfr cells were (Yang et al., 2016; Zeng et al., 2016). Thus, mTOR-mediated highest in p-S6high population while lowest in p-S6low population signaling pathways are closely associated with Treg and Tfh (Figures S1G and S1H). Together, these data revealed that cell differentiation. Tfr cells have dependencies in factors that mTORC1 activity is increased in Tfr cells compared with Treg are necessary for Tfh cells but also for Treg cells. Given that and Tfh cells, and Treg precursors with high amounts of these cell types (Tfh and Treg cells) have apparently differential mTORC1 signaling are more prone to differentiate into Tfr cells requirements regarding mTOR, how the mTOR family would following protein immunization. integrate these signals and coordinate their output to regulate Tfr differentiation is of great interest. A Critical Role of mTORC1 in Early Tfr Differentiation Here, we examined the role of mTOR signaling in the regula- The increase in mTORC1 activity observed in Tfr cells prompted tion of Tfr cell differentiation and effector function. Our findings us to investigate whether mTORC1 plays a functional role in the demonstrate that mTORC1 signaling plays a critical role in the differentiation of Tfr cells from Treg precursors. To this end, we maintenance of humoral immune homeostasis by promoting sorted GITR+CD25+CXCR5– conventional Treg cells (CD45.2+) Tfr cell differentiation and suppressor function. and transiently treated these cells with rapamycin, a pharmaco- logical inhibitor of mTORC1 signaling, for 4 hr and adoptively RESULTS transferred them into WT recipient mice (CD45.1+). At day 8 post OVA/CFA immunization, the number of the donor CD45.2+ Enhanced mTORC1 Signaling in Tfr Cells Foxp3+ Treg cell population, comprised of both CXCR5+ Tfr cells To investigate the relevance of mTOR signaling in the differenti- and CXCR5– Treg cells, was comparable between the rapamy- ation of Tfr cells from Treg precursors, we compared the levels of cin- and vehicle-treated groups; however, both the proportion mTOR activity between Tfr, Treg, and Tfh cells in the draining and number of CD45.2+Foxp3+CXCR5+ Tfr cells were substan- lymph nodes (dLNs) of wild-type (WT) C57BL/6 mice subcutane- tially reduced in the rapamycin-treated Treg group compared ously immunized with ovalbumin (OVA) emulsified in complete with the vehicle-treated Treg group (Figures 2A and 2B). Further- Freund’s adjuvant (CFA). Using flow cytometry analysis, we more, Tfr cells that had differentiated from rapamycin-treated defined Tfr cells as CD4+Foxp3+CXCR5+, conventional Treg precursors exhibited lower expression levels of CXCR5, GITR, cells as CD4+Foxp3+CXCR5–, Tfh cells as CD4+Foxp3–CXCR5+, and CTLA4 than Tfr cells that were derived from the vehicle- and effector CD4+ T cells (Teff) as CD44hiFoxp3–CXCR5– at day 8 treated precursors (Figures 2C and 2D). after immunization (Figure 1A). We observed that Tfr cells The observation of compromised Tfr differentiation by a 4 hr displayed higher amounts of phosphorylated S6 and 4E-BP1, rapamycin treatment on Treg precursors led us to investigate which are key mTORC1 targets (Laplante and Sabatini, 2012), whether mTORC1 signaling plays an important role during the compared with Treg, Tfh, and Teff cells (Figure 1B). Consistent early commitment of Tfr cells from Treg cells. To this end, we with these results, Tfr cells expressed higher amounts of CD71 sorted CXCR5–GFP+ Treg cells from Foxp3-GFP reporter mice (the transferrin receptor) and CD98 (a subunit of the L-amino and labeled these cells with proliferation-indication dye after acid transporter), factors expressed in an active mTORC1- 4 hr treatment with either rapamycin or vehicle. Subsequently, dependent manner (Kelly et al., 2007)(Figure 1C). In contrast, we transferred these labeled Treg cells to recipient mice of the mTORC2 signaling activity in Tfr cells was comparable to 7 days post OVA/CFA immunization. At day 2 and 3 post transfer, that in Treg, Tfh, and Teff cells, as indicated by the similar levels we tracked the conversion of donor CXCR5–GFP+ Treg cells into of Akt phosphorylation at Ser 473 and the phosphorylation of Akt CXCR5+GFP+ Tfr cells in dLNs (Figure S2A). Similar to the results downstream targets FoxO1 and FoxO3a (Powell et al., 2012) of D8 post immunization, at day 2 post transfer, we observed (Figure 1D). Additionally, we compared mTOR activity between both comparable frequency and absolute number of donor Tfr, Treg, Tfh, and Teff cells derived from the spleens of WT GFP+ Treg cells (comprised of both conventional CXCR5–GFP+ mice at day 8 after being acutely infected with lymphocytic Treg cells and converted CXCR5+GFP+ Tfr cells) in both groups choriomeningitis virus Armstrong (LCMV-Arm+). Consistently, (Figures S2B and S2C). We noted that rapamycin-treated donor mTORC1 activity, but not mTORC2 activity, was increased CXCR5–GFP+ Treg cells exhibited much less CXCR5+GFP+ Tfr in Tfr cells compared with Treg, Tfh, and Teff cells (Figures conversion than vehicle-treated counterparts (Figures S2B and S1A–S1D). S2D). The transient rapamycin treatment appeared not to affect To further investigate the association between mTORC1 the overall proliferation of GFP+ Treg cells, as shown by a similar signaling and de novo Tfr differentiation, we sorted Treg cells pattern of proliferation-dye dilution and Ki-67 staining (Figures from Foxp3-GFP reporter mice and labeled these cells with pro- S2E and S2G). We observed that, in contrast to the appearance

Immunity 47, 538–551, September 19, 2017 539 A Treg cells

Immunization Tfr cells

Teff cells Foxp3 Tfh cells CD4 CD44

CXCR5

B Isotype Tfr cells Treg cells Tfh cells Teff cells

Teff cells 500 6 ** ** Tfh cells )

3 Treg cells 450 Tfr cells 4 400 2 350 p-4E-BP-1 MFI Events (% of max) Events (% of max) p-S6 MFI (x10 p-S6 0 p-4E-BP1 300

C

250 10 *

* ) 2 200 8

MFI 150 6 100 4 CD71

50 CD98 MFI (x10 2 Events (% of max) Events (% of max) 0 0 CD71 CD98

D ns ) )

1.2 ns 3 3 5

1.0 4 3 MFI (1x10

0.8 a MFI (x10 2 473 0.6 1 p-Akt Events (% of max) Events (% of max)

0.4 p-FoxO1/3 0 p-AKTs473 p-FoxO1/3a

Figure 1. mTOR Signaling Is Elevated in Tfr Cells following OVA/CFA Immunization (A) Gating strategy of Tfr, Treg, Tfh, and Teff cells in the detection of phosphorylated signaling proteins. (B–D) Comparison of S6 and 4E-BP1 phosphorylation (p) (B), CD71 and CD98 (C), and p-AktS473 and p-FoxO1/3a (D) between Treg cells (CD4+Foxp3+CXCR5–), Tfh cells (CD4+Foxp3–CXCR5+), Teff cells (CD4+Foxp3–CD44+CXCR5–), and Tfr cells (CD4+Foxp3+CXCR5+) 8 days after immunization with OVA/CFA. Blue, green, orange, and red lines in the flow cytometry data represent the gating of Treg, Tfh, Teff, and Tfr cells, respectively, and the solid gray histograms denote the isotype control. The levels of p-S6 and p-4E-BP1 (B), CD71 and CD98 (C), and p-AKTS473 and p-FoxO1/3a (D) summarized beside were calculated by subtracting the mean fluorescence intensities (MFIs) of the isotype controls. Data are representative of two (D) or three (B and C) independent experiments with three mice per group. Center values (B–D) indicate mean. Unpaired t test. *p < 0.05, **p < 0.01; ns, not significant. See also Figure S1.

540 Immunity 47, 538–551, September 19, 2017 ABGated on CD4+Foxp3+ Ctrl Rapa

) ns 2.0 3 12 10 ns * cells 1.5 – 8 )

8 3 6 Tregs (%) Tregs 1.0 + (x10 CXCR5 Tregs (X10 Tregs + + 4 4 0.5 2 Foxp3 Foxp3 0.0 0 Foxp3 0 CD45.2

CD4

50 *** 4 * Ctrl ) 40 3 3 Rapa 30 2 20 cells (x10

Tfr cells (%) 1

10 Tfr

CD44 0 0 CXCR5

C Isotype Ctrl Rapa Events (% of max)

CXCR5 GITR CTLA4

D Ctrl Rapa

) 35 50 2 ) 2

) 15

3 * 30 ** 45 * 10 25 40 R MFI (x10 20 35 5 GIT CTLA4 MFI (x10 CTLA4 CXCR5 MFI (X10 CXCR5 15 30 0

Figure 2. Transient Inhibition of mTORC1 Signaling Impairs the Formation of Tfr Cells without Retardation on Treg Cells (A) Flow cytometry analysis of donor CD45.2+ Treg cells (gated in CD4+Foxp3+, top) and Tfr cells (bottom) in dLNs obtained from recipient mice transferred with rapamycin (Rapa)- or vehicle control (Ctrl)-treated CD45.2+CD4+GITR+CD25+CXCR5– Treg cells, assessed at day 8 after OVA/CFA immunization. The numbers adjacent to the outlined areas indicate the proportion of each population. (B) Summary of the proportion and total cell number of Tfr cells and total Fxop3+ Treg cells and cell number of Fxop3+CXCR5– non-Tfr cells as described in (A). (C and D) Expression (C) and quantification (D) of CXCR5, GITR, and CTLA4 in Tfr cells as described in (A). Blue and red lines represent the gating of control Tfr cells and rapamycin-treated Tfr cells, respectively, and the solid gray histograms denote the isotype control. The data are representative of two independent experiments with four (A–D) mice per group. Center values (B and D) indicate mean. Unpaired t test. *p < 0.05, **p < 0.01, ***p < 0.001; ns, not significant. See also Figure S2. of Tfr conversion as early as the second passage of cell division Treg cells between rapamycin- and vehicle-treated groups, in vehicle-treated donor CXCR5–GFP+ Treg cells, the launch of while a remarkable reduction in Tfr cell conversion was found Tfr cell conversion in rapamycin-treated CXCR5–GFP+ Treg cells in the 4 hr rapamycin-treated group compared to the vehicle- was largely inhibited, and the emergence of CXCR5+GFP+ Tfr treated group (Figures S2I–S2N). Similarly, no obvious differ- cells was barely detected until the fourth passage of cell divi- ences in cell proliferation and survival of donor GFP+ Treg sion (Figure S2F). In addition, we did not notice appreciable were noted between the two groups (Figures S2L, S2O, differences in the survival of transferred donor GFP+ Treg cells and S2P). Taken together, these results supported the notion between vehicle- and rapamycin-treated groups (Figure S2H). that 4 hr rapamycin treatment profoundly impeded the Consistently, at day 3 post transfer, we also noted comparable de novo conversion of CXCR5– Treg cells into CXCR5+ Tfr frequencies and absolute number of transferred donor GFP+ cells at the initiation phase, while exerting minimal effects on

Immunity 47, 538–551, September 19, 2017 541 AB WT Rptorfl/+Foxp3Cre 50 ** 1.6 * WT

) fl/+ Cre 45 5 1.4 Rptor Foxp3 40 1.2 (x10 35 1.0 Tfr cells (%)

30 Tfr cells 0.8

Foxp3 25 0.6 CD4 ns )

5 4 3.0 ns cells 3 – 2.5 ) 5 2.0

Tregs (x10 2 + CXCR5 (x10 + 1 1.5 Foxp3 CD44 0 Foxp3 1.0 CXCR5

C Cre WT Rptorfl/+Foxp3

) 2.4 3 5.5 600 ) 1.8 600 3 ) * * ** 3 * * 2.0 5.0 1.6 400 500 1.4 1.6 4.5 1.2 PD1 MFI 200 CD73 MFI 400 1.2 4.0

ICOS MFI (x10 MFI ICOS 1.0 CTLA4 MFI (x10 MFI CTLA4 CXCR5 MFI (x10 MFI CXCR5 0.8 0 3.5 0.8 300

D Gated on CD4+Foxp3–CD44+ WT Rptorfl/+Foxp3Cre 4 WT 70 ns * ) fl/+ Cre 5 Rptor Foxp3 3 60 2 50

Tfh cells (%) 1

40 Tfh cells (x10 CD44 0 CXCR5

E Gated on B220+CD19+ F Cre fl/+ Cre WT Rptorfl/+Foxp3 WT Rptor Foxp3 ns 4 2.0 * 17 * ) ) 5 3 1.5 16 15 2 1.0 14 GC B (%) 1 5.0 13 NP IgG (Log2) GC B cells (x10 FAS 0 0 12 PNA

Figure 3. The Abolishment of mTORC1 Signaling Affects the Differentiation of Tfr Cells (A) Flow cytometry analysis of CD4+Foxp3+ Treg cells (top) and Tfr cells (bottom) in dLNs obtained from Foxp3Cre control and Rptorfl/+Foxp3Cre mice, assessed at day 8 after NP-OVA/CFA immunization. The numbers adjacent to the outlined areas indicate the proportion of each cell type. (B) Summary of the proportion of Tfr cells and total cell number of Tfr cells, Foxp3+ Treg cells, and Foxp3+CXCR5– cells as described in (A). (C) Quantification of the MFIs of CXCR5, PD-1, CTLA4, ICOS, and CD73 in Tfr cells as described in (A). (legend continued on next page)

542 Immunity 47, 538–551, September 19, 2017 the proliferation or survival of transferred donor CXCR5–GFP+ cells (Figures S3L and S3M). Of note, we observed reduced Treg cells. proportion and absolute number of Foxp3+CXCR5+ Tfr cells To comprehensively investigate the role of cell-autonomous in Rptorfl/+ mice compared to WT mice at day 8 after NP-OVA/ mTORC1 signaling in regulating the differentiation of Treg CFA immunization (Figures 3A and 3B). Moreover, Rptorfl/+ cells into Tfr cells, we crossed Rptorfl/fl mice with Foxp3YFP-Cre Foxp3+CXCR5+ Tfr cells expressed decreased amounts of (Foxp3Cre) mice to generate mice in which the Rptor alleles CXCR5, CTLA4, CD73, ICOS, and PD-1 compared to WT Tfr are conditionally deleted in Foxp3+ Treg cells (hereafter cells (Figure 3C). Consistent with these results, greater numbers referred to as Rptorfl/flFoxp3Cre mice). Foxp3+ Treg cells, but of CD44hiCXCR5hiTfh and PNAhiFAShi GC B cells and higher not Foxp3–CD4+ T cells, exhibited a reduction in the phosphory- titers of NP-specific antibody were seen in Rptorfl/+ mice lation of S6 and 4E-BP1 following TCR-CD28 stimulation (Fig- compared to WT mice (Figures 3D–3F). Given the unaffected ure S3A). We next generated bone marrow (BM) chimeras by proliferation and survival of Treg cells by transient rapamycin mixing BM cells derived from Rptorfl/flFoxp3Cre mice (CD45.2+) treatment or a genetic deletion of a single copy of Rptor alleles, (30%) and WT mice (CD45.1+) (70%) and injecting the mixtures the impaired differentiation of Tfr cells in both settings high- into irradiated WT recipients (CD45.1+). In this setting, the major- lighted a specific and important role of mTORC1 signaling during ity of B and T lymphocytes was of WT origin and would develop Treg to Tfr conversion. relatively normal GC structures that support Tfr differentiation Next, we investigated whether a heightened mTORC1 in response to immunization. At day 8 after OVA/CFA immuniza- signaling was able to enhance the differentiation of Tfr cells. To tion, we observed an approximate 7:3 ratio between WT and this end, we deleted Tsc1, a negative regulator upstream of mutant total CD4+ T cells, while the ratio between WT Foxp3+ mTORC1 signaling (Laplante and Sabatini, 2012), in Foxp3+ Treg cells and Rptor-null Foxp3+ Treg cells became 9:1 (Fig- Treg cells by crossing Foxp3Cre mice with Tsc1fl/fl mice ure S3B), an observation that might reflect defects in the homeo- to generate Tsc1fl/fl Foxp3Cre mice (hereafter referred to as stasis of this population under these conditions (Zeng et al., Tsc1–/–). Following TCR-CD28 stimulation, Foxp3+ Treg cells 2013). Furthermore, we noted that Rptor-null CD45.2+ Treg cells from Tsc1–/– mice exhibited an increased level of S6 phosphory- contributed about 7.3% of total Treg cells, while CD45.2+ Tfr lation compared to counterparts from WT mice (Figure S4A). cells of Rptor-null origin only contributed about 3.3% of total At day 8 post LCMV-Arm+ infection, we observed both higher Tfr cells (Figures S3B and S3C). We also compared the fold proportion and absolute number of Tfr cells in the spleens of differences in total Treg cells as well as in Tfr cells from both Tsc1–/– mice than in those of WT mice (Figures S4B and S4C). origins. It turned out that there was an average 14.3-fold Furthermore, Tsc1-null Tfr cells exhibited higher amounts of change between WT and Rptor-null Treg cells, while there was CXCR5 and GITR expression than WT Tfr cells (Figures S4D). an average 31.3-fold change between WT and Rptor-null Tfr To determine the role of mTORC2 signaling in Tfr cell differ- cells (Figures S3B and S3C). Furthermore, Tfr cells of Rptorfl/fl entiation, we crossed Cd4-Cre transgenic mice with mice Foxp3Cre origin expressed reduced amounts of CXCR5, GITR, harboring loxP-flanked Rictor alleles (Rictorfl/fl) to generate and CTLA4 (Figures S3D and S3E). These results demonstrated Rictorfl/fl Cd4-Cre mice (hereafter referred to as Rictor–/–). that the abrogation of mTORC1 signaling in Treg cells seemed to Following TCR-CD28 stimulation, the mTORC2-mediated phos- result in more severe defects in Tfr differentiation than in Treg phorylation of Akt at Ser 473 and phosphorylation of the Akt expansion. downstream targets FoxO1/3a were substantially reduced in The complete loss of mTORC1 signaling in Treg cells has been Rictor-deficient CD4+ T cells compared to WT CD4+ T cells (Fig- found to profoundly impair the homeostatic proliferation of these ures S5A and S5B). Of note, WT and Rictor–/– Treg cells exhibited cells (Zeng et al., 2013). To clarify a specific role of mTORC1 comparable levels of Tfr cell differentiation following OVA/CFA signaling in regulating Treg to Tfr conversion, we bred Rptorfl/fl immunization (Figure S5C). Furthermore, the expression levels mice with Foxp3YFP-Cre (Foxp3Cre) mice to generate mice in of CXCR5, ICOS, GITR, and CTLA4 of Tfr cells were similar be- which only a single copy of Rptor alleles was conditionally tween the two groups (Figure S5D). Thus, mTORC2 appears to deleted in Foxp3+ Treg cells (hereafter referred to as Rptorfl/+ be dispensable for Tfr cell differentiation. Together, these results mice). We found an approximate half reduction of mTORC1 demonstrated that mTORC1 signaling, but not mTORC2 signaling in Rptorfl/+ Treg cells (Figure S3F), which did not appar- signaling, plays a critical role in regulating Tfr cell differentiation. ently influence Foxp3+ Treg cell population and suppressive function, evidenced by comparable frequency and number of The Role of mTORC1 in Effector Function of Tfr Cells total Foxp3+ Treg cells (Figures 3A and 3B), similar expression As Tfr cells with attenuated mTORC1 signaling exhibited levels of CD73, CTLA4, PD-1, and ICOS (Figure S3G), equivalent decreased amounts of CTLA4, ICOS, and PD-1, we speculated proliferation and survival (Figures S3H–S3K), and similar sup- that mTORC1 signaling in differentiated Tfr cells might modulate pressive effects on effector CD4+ T cells induced colitis in their suppressor function. To test this hypothesis, we sorted Rag-deficient mice between WT Treg cells and Rptorfl/+ Treg already differentiated GITR+CD25+CXCR5+ Tfr cells from WT

(D and E) Flow cytometry analysis of CD4+Foxp3–CXCR5+ Tfh cells (D) and PNAhiFAShi GC-B cells cells (E) in dLNs obtained from Foxp3Cre control and Rptorfl/+Foxp3Cre mice, assessed at day 8 after NP-OVA/CFA immunization. Summary of the frequency and total number of Tfh (D) and GC-B (E) are shown on the right. (F) Titers of NP-specific IgG in sera obtained from Foxp3Cre control and Rptorfl/+Foxp3Cre mice, assessed at day 8 after NP-OVA/CFA immunization. The data are representative of three independent experiments with three (A–F) mice per group. Center values (B–F) indicate mean. Unpaired t test. *p < 0.05, **p < 0.01; ns, not significant. See also Figures S3–S5.

Immunity 47, 538–551, September 19, 2017 543 Figure 4. Rapamycin Treatment Inhibits the A + + (Gated on B220 CD19 ) Suppressive Function of Tfr Cells WT Tfh cells + WT Tfh cells + (A and B) Flow cytometry analysis of PNAhiFAShi WT Tfh cells No transfer Vehicle Tfr cells Rapa Tfr cells GC-B cells (A) gated on the B220+CD19+ population and CD138hiB220lo plasma cells (B) in spleens from recipient mice on day 7 after cell transfer. (C and D) Summary of the proportion and total number of GC-B cells (C) and plasma cells (D) as described above.

FAS The presented data are representative of two inde- PNA pendent experiments with two to four mice per group. Center values (C and D) indicate mean. Unpaired t test. *p < 0.05, **p < 0.01. B (Gated on Live lymphocytes) WT Tfh cells + WT Tfh cells + WT Tfh cells Vehicle Tfr cells Rapa Tfr cells No transfer Rptorfl/flCd4-Cre mice (hereafter called Rptor–/– mice) at day 8 after OVA/CFA immu- nization (Figure S6A) and compared their transcriptional profiles by microarray anal- ysis. We identified 339 genes decreased CD138 and 257 genes increased in Rptor–/– Tfr cells B220 compared with their WT counterparts (Table S1). Then, we selected a set of genes from a C WT Tfh cells + D WT Tfh cells + WT Tfh cells No transfer published dataset (Linterman et al., 2011) Vehicle Tfr cells Rapa Tfr cells that are preferentially expressed in Tfr cells 6

10 )

) 3 10

6 6

6 4 ) ) 4x10 compared with non-Tfr cells for gene set 8 * ** ** 6 3 3x106 * enrichment analysis (GSEA). The genes 6 2 10 6 4 2 2x10 included in the Tfr cell gene set were sub- 1 106 6 GC B cells (% 2 1 1x10 stantially enriched in WT cells, but not in GC B cells (x10

Plasma cells (% –/– Plasma cells (x10 0 0 0 Rptor Tfr cells (Figure 5A), suggesting that mTORC1 signaling played an important role in programming and maintaining Tfr cell mice on day 7 after LCMV-Arm+ infection and transiently treated lineage. Accordingly, these genes were differentially expressed them with rapamycin or vehicle. The sorted cells were combined between WT and Rptor–/– Tfr cells (Figure 5B). We further with WT CD44hiCXCR5+ Tfh cells that had been isolated from WT confirmed these results by quantitative RT-PCR analysis with mice on day 7 after infection and injected into day 1 infected SAP 18 genes that were potentially associated with Tfr differentia- knockout (Sh2d1a–/–) recipients (these mice fail to generate tion and suppressor functions (Figure 5C). Gene profiling data endogenous functional Tfh and Tfr cells upon infection (Linter- revealed that the abolishment of mTORC1 signaling led to the man et al., 2011; Qi et al., 2008; Yusuf et al., 2010). On day 7 after reduction of Tcf7 (encoding transcription factor TCF-1) and cell transfer, the population of PNAhiFAShiCD19+ GC B cells in Bcl6 in differentiated Rptor–/– Tfr cells compared with WT Tfr spleens from the control mice that had not received a cell trans- cells (Figure 5C). This phenotype was confirmed by flow cytom- fer was barely detectable, and a substantial GC B cell population etry analysis comparing TCF-1 and Bcl6 expression between was observed in the spleens of mice that had received exoge- differentiated Rptor-null and WT Tfr cells (Figure 5D). nous Tfh cells alone (Figure 4A), which is consistent with our pre- vious report (Xu et al., 2015). In contrast, the spleens of mice The mTORC1-p-STAT3-TCF-1 Axis in Tfr Cell that had a co-transfer of Tfh cells and vehicle-treated Tfr cells Differentiation exhibited a robust inhibition of the GC response (Figure 4A). Bcl6 is required for Tfh cell differentiation (Johnston et al., 2009; Notably, the spleens of mice that had received WT Tfh and rapa- Nurieva et al., 2009; Yu et al., 2009), and Bcl6 expression is initi- mycin-treated Tfr cells exhibited an increased proportion and ated by the binding of the transcription factor TCF-1 (encoded by total number of GC B cells compared with the spleens of mice Tcf7) to its promoter in Tfh cells (Choi et al., 2015; Xu et al., 2015). that had received WT Tfh and vehicle-treated Tfr cells (Figures Given the similar phenotypic characteristics of Tfr and Tfh cells, 4A and 4C). Similar results were observed in plasma cells (Fig- we hypothesized that the mTORC1 deficiency might impair ures 4B and 4D). These results thus demonstrated that mTORC1 TCF-1 expression, thereby inhibiting Tfr cell differentiation by signaling in differentiated Tfr cells is also important for their sup- reducing the expression of Bcl6 in Tfr precursors. To test this pressive functions. hypothesis, we first investigated whether TCF-1 was capable of binding the putative TCF-1-binding sequence (Bcl6 500) in Transcriptional Profiles of mTORC1-Deficient Tfr Cells the Bcl6 promoter region in sorted Tfr precursors. Using To explore the molecular mechanisms by which mTORC1 chromatin immunoprecipitation (ChIP) followed by quantitative signaling regulates Tfr cell differentiation, we sorted differen- PCR (qPCR), we observed an enrichment of TCF-1 binding tiated CD19–CD4+GITR+CD25+CXCR5+ Tfr cells from WT or to the Bcl6 promoter region (Bcl6 0.5 kb) compared with

544 Immunity 47, 538–551, September 19, 2017 AB Tfr gene signature –/– –/– –/– WT Rptor WT Rptor WT Rptor IL1R2 EGR2 PRKCH High gene expression 0.7 ITGB8 MCM6 ZDHHC2 Low gene expression KIF11 IKZF3 MYO1E 0.6 CCNB2 TTC39C PRDM1 BIRC5 LRBA CKS2 NUSAP1 LAG3 WBP5 0.5 2810417H13RIK PSMA1 ZFP52 GLRX RRM1 HSPA4L 0.4 TNFRSF18 LCLAT1 TANK NRN1 ERGIC1 AW112010 PDCD1 SH2D1A VAMP8 0.3 NCAPG PRC1 CDC14A CDCA8 CCDC104 DYNLT3 0.2 6330403KO7RIK KPNA2 TIMP2 TPX2 PLK4 MALT1 ichment (ES) score 0.1 ICOS SLA S100A9 CCNA2 S100A6 S100A4

Enr MAD2L1 ITGAE SRGN 0.0 HMGB2 GSTA4 NKG7 CENPH ATP6V0D2 ARL3 SPC25 PENK DHRS1 NT5E EZH2 CASP3 TOP2A IRF4 NFIL3 CTLA4 NRP1 GRAMD1B EPHX1 HIF0 SLFN8 CENPK GBP3 BCL6 TNFRSF4 TOX CCNB1 BATF MED7 REPS1 WT –/– TCF7 SAMSN1 ANXA2 2.5 Rptor CENPE PTPLAD1 BTLA STMN1 RGS16 CST7 CDKN2C GLCCI1 IL2RB FIGNL1 IKZF2 S100A10 0.0 HMGN3 VPS54 CCL5 CKS1B RILPL2 ATOX1 RRM2 CCR8 AIM2 IL10 HIF1A EIF2AK2 -2.5 SCCPDH RABGAP1L IGJ CASP4 GBP6 ENTPD1 FOLR4 ARL5A LGALS1 KLRG1 ODC1 DAPP1 0 2,500 5,000 7,500 10,000 12,500 15,000 17,500 20,000 22,500 EEA1 TRIM59 CXCR3 ITGB1 CXCR5 FABP5 Rank in Ordered Dataset TNFRSF9 IL2RA FAM92A TMEM153 Ranked list metric (Signal2Noise) Enrichment profile SYT11 PGLYRP1 FOXP3 CASP1 Hits Ranking metric scores LARP1B PRKCA PLA2G16 ATPIF1 CAPG

C 4 WT 2.5 2.5 3 ) 4 * ** ) * ** * Rptor–/– 3 2.0 2.0 3 2 1.5 1.5 2 2 Icos Irf4 Cxcr5 Il1r2

1.0 Ctla4 1.0 1 1 1 0.5 0.5 expression (fold expression (fold) expression (fold) expression (fold expression (fold) 0 0 0.0 0.0 0

1.5 3 2.0 * ) ** 2.5 * 4 ** * 2.0 1.5 1.0 3 2 1.5 1.0 Gitr 2 Klrg1

1.0 Pdcd1 1 0.5 Prdm1 0.5 1

Il2ra (Cd25) 0.5 expression (fold) expression (fold) expression (fold) expression (fold expression (fold) 0.0 0.0 0.0 0 0

2.5 *** 2.5 4 * 2.5 * 3 * ) * 2.0 2.0 2.0 3 2 1.5 1.5 1.5 2 Il10

1.0 Lclat1 1.0 1.0 Tnfrsf9 1 1

Itgae (Cd103) 0.5 0.5 Nt5e (Cd73)

0.5 expression (fold) expression (fold) expression (fold) expression (fold expression (fold) 0.0 0 0.0 0.0 0 WT Rptor–/– 3 * 10 6 ** ** D 25 4 )

*** ) 2 8 2 *** 2 4 20 6 3 cf7 Bcl6 Batf 4 T 15 1 2 1 MFI (x10 2

2 10 MFI (x10 Bcl-6 TCF- expression (fold) expression (fold) expression (fold) 0 0 0 5 1

Figure 5. mTORC1-Dependent Transcriptional Profiles of Tfr Cells (A) GSEA analysis of gene signatures in WT and Raptor–/– Tfr cells sorted on day 8 after OVA/CFA immunization. (B) Heatmap of the expression of genes described in (A). Red, upregulated gene expression; blue, downregulated gene expression. (C) RT-qPCR of selected genes listed in (A) normalized to their expression level in Rptor–/– Tfr cells. (legend continued on next page)

Immunity 47, 538–551, September 19, 2017 545 isotype-matched IgG-mediated binding and with TCF-1 binding Eight days after LCMV-Arm+ infection, we compared the extent to the first intron of Bcl6 (Bcl6 +2.9 kb, containing no TCF-1- of Tfr cell differentiation between transduced and non-trans- binding motif) (Figure 6A). To further examine TCF-1-mediated duced Rptor–/– Treg cells within the same animal. We observed Tfr cell differentiation, we generated Tcf7fl/fl Cd4-Cre mice (here- that STAT3-CA overexpression to a large extent rescued Tfr after referred to as Tcf7–/–) by crossing Tcf7fl/fl and Cd4-Cre cell differentiation in Rptor–/– Treg cells (Figures 6F and 6G). mice. On day 8 after OVA/CFA immunization, we observed a Furthermore, Rptor-null Tfr cells overexpressing STAT3-CA ex- marked reduction in the proportion and absolute number of Tfr hibited elevated expression levels of both TCF-1 and Bcl6 (Fig- cells in dLNs of Tcf7–/– mice compared with those of WT mice ure 6H). These data suggested that p-STAT3 functions to relay (Figure 6B). As expected, Tcf7–/– Tfr cells expressed reduced upstream of mTORC1 signaling to downstream TCF-1-Bcl-6 amounts of Bcl-6 compared with WT Tfr cells (Figure 6B). Collec- axis for Tfr differentiation from Treg cells. tively, these results revealed that mTORC1-signaling-depen- Next, we speculated that TCF-1 overexpression would also dent TCF-1 expression was critical for Tfr cell differentiation via ‘‘rescue’’ the Tfr cell differentiation defect associated with the regulating Bcl-6 expression. abolishment of mTORC1 signaling. To evaluate this possibility, Next, we sought to determine the mechanisms by which we transduced Treg cells isolated from the spleens of Rptor–/– mTORC1 signaling influences TCF-1 expression. The mTORC1 mice (CD45.2+) with retroviruses carrying a vector encoding signaling has been reported to mediate the phosphorylation TCF-1 and transferred these cells together with splenocytes of STAT3 (Dodd et al., 2015; Laplante and Sabatini, 2013). from WT mice (CD45.1+) into sub-lethally irradiated mice We then investigated whether mTORC1-mediated STAT3 (CD45.1+). At day 8 after LCMV-Arm+ infection, we compared phosphorylation regulated TCF-1 expression during Tfr differ- Tfr cell differentiation between transduced and non-transduced entiation. At day 8 post OVA/CFA immunization, we observed Rptor–/– Treg cells within the same animal. Rptor–/– Treg cells increased amounts of phosphorylated STAT3 (p-STAT3) at site overexpressing TCF-1 exhibited a marked increase in Tfr cell dif- of serine727 in WT Tfr cells compared with WT Treg cells (Figures ferentiation compared to control Rptor–/– Treg cells (Figures 7A 6C and 6D, left panel). Of note, a dramatic reduction in p-STAT3 and 7B). We also examined whether Bcl-6 overexpression could levels was found in Rptor-null Tfr cells compared with WT Tfr restore the defect in Tfr cell differentiation in Rptor–/– Treg cells cells (Figures 6C and 6D, right panel), indicating that mTORC1 using the same strategy. Similar to TCF-1 overexpression, we signaling promotes STAT3 phosphorylation in Tfr cells. Next, observed Bcl-6 overexpression significantly increased Tfr cell we examined whether p-STAT3 directly bound to Tcf7 loci. By differentiation in Rptor–/– Treg cells (Figures 7D and 7E). These analyzing data for chromatin immunoprecipitation and deep data revealed that both TCF-1 and Bcl-6 overexpression could sequencing (ChIP-seq) (Durant et al., 2010), we noted two largely rectify the defects in Tfr cell differentiation of Treg cells conserved p-STAT3-binding peaks enriched at the Tcf7 loci, us- deficient in mTORC1 signaling. Furthermore, both TCF-1 and ing the UCSC browser (http://genome.ucsc.edu/)(Figure S6B). Bcl-6 overexpression in Rptor–/– Treg cells increased the expres- We further identified two conserved consensus p-STAT3-bind- sion levels of CD73, ICOS, and CTLA4, molecules that are ing sequences (TTCnnnGAA) in the 50-regulatory region of Tcf7 associated with Tfr suppressive function (Figures 7C and 7F). (31.8 kb and 17.5 kb, respectively) (Figure S6C). To validate Collectively, these data support the notion that the mTORC1- these putative p-STAT3-binding sites, we performed ChIP- p-STAT3-TCF-1-Bcl-6 axis plays a primary role in the regulation qPCR experiments using sorted Tfr precursors. We observed of Tfr cell differentiation from conventional Treg precursors. an enriched binding of p-STAT3 to both loci, with a greater enrichment observed at the 31.8 kb site, compared with the DISCUSSION binding of isotype-control IgG or the binding of p-STAT3 to the first exon (Tcf7 +0.2 kb, containing no p-STAT3-binding motif) Conventional Treg cells are capable of differentiating into Tfr (Figure 6E). These data suggested the direct regulation of cells that migrate into B cell follicles and inhibit excessive GC TCF-1 expression by phosphorylated STAT3 during Tfr differen- responses induced by infection or immunization. In this study, tiation. Next, we investigated whether the introduction of a we identified the critical role of mTORC1 signaling in this pro- constitutive active form of STAT3 (STAT3-CA) into Rptor-null cess. Treg cells with impaired mTORC1 signaling activity by tran- Treg cells was able to rescue the impaired Tfr differentiation. sient rapamycin treatment or genetic deletion of a single copy We stimulated Treg cells isolated from the spleens of Rptor–/– or double copies of Rptor alleles exhibited profound defects in mice (CD45.2+) with TCR-CD28 stimuli and transduced them Tfr cell differentiation. By contrast, increased mTORC1 signaling with retroviruses carrying a vector encoding STAT3-CA. The in Tsc1/ Treg cells promoted Tfr cell differentiation post viral transduced, STAT3-CA-overexpressing cells were identified as infection. Based on these results, we concluded that mTORC1 CD45.2+GFP+ cells. To increase the uptake of these donor cells, signaling plays a critical role in Tfr cell differentiation. Interest- we depleted splenocytes from WT recipient mice (CD45.1+) us- ingly, mTORC2 seemed not to be required for Tfr cell differenti- ing sub-lethal irradiation and reconstituted their splenocytes ation. Since both mTORC1 and mTORC2 have been reported with a mixture of STAT3-CA-transduced or non-transduced to be essential for Tfh differentiation (Yang et al., 2016; Zeng Rptor–/– Treg cells (CD45.2+) and WT splenocytes (CD45.1+). et al., 2016), while only mTORC1 is important for Treg and Tfr

(D) Quantification of TCF-1 and Bcl-6 in CD4+Foxp3+CXCR5+ Tfr cells of different origins (WT or Rptorfl/fl Foxp3Cre) in dLNs obtained from BM chimeras on day 8 after OVA/CFA immunization. The data presented are representative of two independent experiments (C and D) with three mice (D) or two technical replicates pooled from at least four mice per group (C). Error bars (C) indicate the mean ± SEM. (C) by unpaired t test. (D) by paired t test. *p < 0.05, **p < 0.01, ***p < 0.001. See also Figure S6A.

546 Immunity 47, 538–551, September 19, 2017 E A anti-p-STAT3 4 anti-TCF-1 10 ** IgG *** IgG 3 8 6 2 4 **

Fold Enrichment 1 Fold Enrichment 2

0 0 Bcl6 Bcl6 Tcf7 Tcf7 Tcf7 -0.5K +2.9K -31.8k -17.5k +0.2K

B Gated on CD4+Foxp3+ WT Tcf7 -/- WT Tcf7–/–

40 220 2.5 ** * *** ) 30 5 2.0 200 1.5 180 20 cells (%) 1.0 160 Bcl6 MFI Tfr 10

Tfr cells (x10 0.5 140

CD44 0 0.0 120 CXCR5

C Isotype Isotype WT D Treg cells Tfr cells WT Rptorfl/flFoxp3Cre Treg Tfr cells Rptorfl/flFoxp3Cre 10 **** 10

) **** ) 2 2 8 8

6 6

4 MFI (x10 T3 4

2 2 p-STA p-STAT3 MFI (x10 p-STAT3

Events (% of max) 0 0 p-STAT3

Migr-1 STAT3-CA F WT GFP– GFP+ GFP–（Rptor–/–） GFP+（Rptor–/–） CD44

CXCR5

G H GFP– GFP+ 50 *** 3 * 8 *** ) )

40 2 3 7 30 2 6 20 5 Tfr cells (%) Tfr cells 1

10 Bcl-6 MFI (x10

TCF-1 MFI (x10 4 Events (% of max) 0 Events (% of max) GFP- GFP+ 0 3 GFP- GFP+ Bcl-6 GFP- GFP+ STAT3-CA TCF-1

(legend on next page)

Immunity 47, 538–551, September 19, 2017 547 differentiation, this suggests that Tfr cells are more closely suppresses TCR signaling by the recruiting of phosphatases related to Treg cells than to Tfh cells. SHP-1/2 (Chemnitz et al., 2004; Sheppard et al., 2004). Provided The mTORC1 signaling has been found to be critical for the that TCR signaling is a primary activator of mTORC1 signaling in homeostatic expansion of Foxp3+ Treg cells (Zeng et al., Treg cells (Zeng et al., 2013), it is plausible that PD-1 signaling 2013). However, in the experimental settings of either transient dampens Tfr cell differentiation in part by inhibiting mTORC1 rapamycin treatment or genetic deletion of a single copy of Rptor activation in Tfr precursors. alleles, the expansion, stability, and effector function of Treg The mTORC1 signaling seems to be an intrinsic negative regu- cells were not apparently influenced. Therefore, both models lator of the de novo differentiation and expansion of conventional (transient rapamycin treatment or genetic deletion of a single Treg cells (Battaglia et al., 2005; Delgoffe et al., 2009; Haxhinasto copy of Rptor alleles) provided us a valuable chance to investi- et al., 2008; Liu et al., 2009; Sauer et al., 2008). In sharp contrast, gate the role of mTORC1 signaling specifically in the process mTORC1 functions as an essential rheostat in conventional Treg of Treg to Tfr conversion. The inhibited Tfr differentiation from cell homeostasis and suppressor activity (Zeng et al., 2013). transiently rapamycin-treated or Rptorfl/+ Treg precursors indi- The mTORC1 might mediate these contradictory effects primar- cated that Tfr cells are much more sensitive than Treg cells to ily by regulating distinct cellular metabolic states. The activation mTORC1 activity inhibition. Together, our results provided un- of mTORC1 signaling inhibits the de novo differentiation of Treg ambiguous evidence that the attenuation of mTORC1 signaling cells by fueling glycolytic reactions that are unfavorable for specifically and intrinsically influenced Treg to Tfr commitment, oxidative phosphorylation-mediated Treg cell differentiation (Po- and such effect was likely independent of the role of mTORC1 well et al., 2012); however, mTORC1 facilitates the homeostasis in the expansion of Treg cells. and the suppressor function of naturally occurring Treg cells by The mechanism by which mTOR signaling is amplified in Tfr promoting cholesterol and lipid metabolism (Zeng et al., 2013). precursors to promote Tfr differentiation at early stage of viral Our data demonstrated that mTORC1 regulates Tfr cell differen- infection or immunization remains unclear. The IL-2-STAT5 tiation likely primarily by transcriptionally regulating the TCF-1- axis reportedly enhances mTORC1 activity in virus-specific Bcl-6 axis, as evidenced by the substantial rescue of Tfr cell CD4+ T cells to promote Th1 cell differentiation and inhibit Tfh differentiation by the overexpression of either TCF-1 or Bcl6 in cell differentiation in response to LCMV-Arm+ infection (Ray Rptor–/– Tfr precursors. Thus, mTORC1 appears to differentially et al., 2015). Similarly, IL-2 might also potentiate mTORC1 modulate Treg and Tfr cells via distinct molecular mechanisms. signaling in Tfr precursors to drive the differentiation of these The inhibition of CD8+ T cell-intrinsic mTORC1 signaling by cells into Tfr cells. In addition, virus-specific Th1 cells are the rapamycin treatment profoundly expands the population of major source of IL-2 during acute viral infection, and the IL-2- memory CD8+ T cells (Araki et al., 2009; Pearce et al., 2009). CD25-STAT5 axis has been reported to inhibit Tfh differentia- Thus, rapamycin has the potential as an adjuvant to improve tion (Johnston et al., 2012). Th1-cell-derived IL-2 might limit vaccine efficacy (Araki et al., 2010). The findings of this study GC responses by inhibiting Tfh cell commitment while promoting add an additional rationale for this potential. The administration Tfr cell differentiation through potentiating mTORC1 signaling in of rapamycin might simultaneously improve the memory forma- Tfh or Tfr precursors. In addition, the co-stimulatory receptor tion of CD8+ T cells and boost GC responses by inhibiting Tfr ICOS acts upstream of mTORC1 signaling (Xie et al., 2012). differentiation and suppressive function. However, this hypothe- Loss of ICOS in Treg cells impaired Tfr cell differentiation, indi- sis warrants further investigation. cating that ICOS promotes Tfr cell differentiation potentially by In summary, we have demonstrated that mTORC1 signaling is activating the mTORC1 signaling (Chang et al., 2014). Further- critical for both the differentiation and suppressive activity of Tfr more, signal transduction mediated by the co-inhibitory re- cells due to its role in regulating the TCF-1-Bcl-6 axis in response ceptor PD-1 inhibits Tfr differentiation (Sage et al., 2013). PD-1 to protein immunization or viral infection. Defining approaches

Figure 6. The mTORC1-p-STAT3-TCF-1-Bcl-6 Axis Regulates Tfr Differentiation (A) Binding of TCF-1 to conserved motifs in the Bcl6 promoter region (Bcl6 500) and to a region in Bcl6 without TCF-1-binding motifs (Bcl6 +2.9 kb) (negative control) in CD4+GITR+CD25+ T cells sorted from WT mice at day 8 after OVA/CFA immunization, analyzed by ChIP with antibody to TCF-1 (anti-TCF-1) or isotype- matched control antibody (IgG), followed by quantitative PCR (normalized to their expression level in IgG control). (B) Flow cytometry analysis of CXCR5+ Tfr cells (gated in CD4+Foxp3+ cells) in the dLNs of WT and Tcf7–/– mice on day 8 after OVA/CFA immunization. The proportion and total number of Tfr cells are summarized beside. Quantification of Bcl-6 MFI in Tfr cells is also shown on the right. (C) Expression of p-STAT3 in Treg cells and CD4+Foxp3+CXCR5+ Tfr cells of different origins in dLNs obtained from BM chimeras (WT and Rptorfl/fl Foxp3Cre)on day 8 after OVA/CFA immunization. Blue, green, and red lines on flow cytometry data represent the gating of WT Tfr cells, WT Treg cells, and Rptorfl/fl Foxp3Cre Tfr cells, respectively, and the solid gray histograms denote isotype control (left). (D) Quantification of the MFIs of p-STAT3 in Treg cells and Tfr cells from different origins as described in (C). Lines connect data points for the same mouse. (E) Binding of p-STAT3 to conserved motifs in the Tcf7 50-regulatory region (Tcf7 31.8 kb, 17.5 kb) and to a region in Tcf7 without p-STAT3-binding motifs (Tcf7 +0.2 kb) (negative control) in CD4+GITR+CD25+ T cells sorted from WT mice at day 8 after OVA/CFA immunization, analyzed by ChIP with antibody to p-STAT3 (anti-p-STAT3) or isotype-matched control antibody (IgG), followed by quantitative PCR (normalized to their expression in IgG control). (F and G) Flow cytometry analysis of Tfr population of different origins in spleens obtained from splenic chimeras on day 8 after LCMV-Arm+ infection (gated on CD4+CD25+ cells). The numbers adjacent to the outlined areas indicate the proportion of Tfr cells. The proportion of Tfr cells described in (F) is summarized in (G), and lines connect data points for same mouse. (H) Quantification of the MFIs of TCF-1 and Bcl-6 in Tfr cells of different origins as described in (F). The data are representative of two (A and E–H) or three (B–D) independent experiments with at least three (B–D and F–H) mice or three technical replicates (A and E) per group. Error bars (A and E) indicate the mean ± SEM. Center values (B) indicate mean. (A), (B), and (E) by unpaired t test; (D)–(H) by paired t test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. See also Figures S6B and S6C.

548 Immunity 47, 538–551, September 19, 2017 AB TCF-1-Ov GFP– 60 + ** GFP 40 ns

20 Tfr cells (%) CD4 CD4 0 WT Vector P33 CD45.1 GFP

Migr1 TCF-1-Ov WT GFP– GFP+ GFP– GFP+ CD44 CD44

CXCR5 CXCR5

C WT GFP–(Rptor–/–) GFP+(TCF-1-Ov) ns

) 15 20 *

3 ** * ) 3.5 ) 4 ) ns )

4.0 2 2 3 3 * * ns * 3.5 * 3.0 10 15 3 3.0 2.5 2 5 10 2.5 2.0 1 GITR MFI (x10 ICOS MFI (x10 CD73 MFI (x10 CXCR5 MFI (x10 2.0 1.5 0 5 CTLA4 MFI (x10 0

D E Bcl-6-Ov WT WT GFP– GFP+ 60 GFP–(Rptor–/–) ** GFP+(Bcl6-Ov) 40

20 Tfr cells (%)

0 CD44

CXCR5 F ns 4.0 ns ns 3.5 ) ) 3.5 3 3 * 2.5 ** ) )

) 10 3 3 2 3.5 3.0 * 8 * * 3.0 ** * 2.0 6 3.0 2.5 2.5 4 1.5 2.5 2.0 2.0 GITR MFI (x10

CD73 MFI (x10 2 CTLA4 MFI (x10 ICOS MFI (x10 CXCR5 MFI (x10 2.0 1.5 0 1.0 1.5

Figure 7. TCF-1 and Bcl-6 Overexpression Restores Defective Tfr Differentiation in Rptor–/– Treg Cells Flow cytometry analysis of Tfr population of different origins in spleens obtained from TCF-1 overexpression (A) or Bcl-6 overexpression (D) splenic chimeras on day 8 after LCMV-Arm+ infection (gated on CD4+CD25+ cells). The numbers adjacent to the outlined areas indicate the proportion of Tfr cells, which are sum- marized in (B) and (E). Lines connect data points for the same mouse. Quantification of the MFIs of CXCR5, CD73, ICOS, GITR, and CTLA4 in Tfr cells of different origins are summarized in (C) and (F). The data are representative of three independent experiments with three mice per group. Central values (B, C, E, and F) indicate mean. Paired t test. *p < 0.05, **p < 0.01; ns, not significant.

Immunity 47, 538–551, September 19, 2017 549 toward selectively modulating mTORC1 signaling in Tfr cells may Araki, K., Turner, A.P., Shaffer, V.O., Gangappa, S., Keller, S.A., Bachmann, provide inroads into improving protective immunity or treating M.F., Larsen, C.P., and Ahmed, R. (2009). mTOR regulates memory CD8 autoimmune diseases. T-cell differentiation. Nature 460, 108–112. Araki, K., Youngblood, B., and Ahmed, R. (2010). The role of mTOR in memory STAR+METHODS CD8 T-cell differentiation. Immunol. Rev. 235, 234–243. Basso, K., and Dalla-Favera, R. (2015). Germinal centres and B cell lymphoma- Detailed methods are provided in the online version of this paper genesis. Nat. Rev. Immunol. 15, 172–184. and include the following: Battaglia, M., Stabilini, A., and Roncarolo, M.G. (2005). Rapamycin selectively expands CD4+CD25+FoxP3+ regulatory T cells. Blood 105, 4743–4748. d KEY RESOURCES TABLE Chang, J.H., Hu, H., Jin, J., Puebla-Osorio, N., Xiao, Y., Gilbert, B.E., Brink, R., d CONTACT FOR REAGENT AND RESOURCE SHARING Ullrich, S.E., and Sun, S.C. (2014). TRAF3 regulates the effector function of d EXPERIMENTAL MODEL AND SUBJECT DETAILS regulatory T cells and humoral immune responses. J. Exp. Med. 211, 137–151. B Mice Chapman, N.M., and Chi, H. (2015). mTOR links environmental signals to T cell d METHOD DETAILS fate decisions. Front. Immunol. 5, 686. B Protein immunization and acute viral infection Chemnitz, J.M., Parry, R.V., Nichols, K.E., June, C.H., and Riley, J.L. (2004). B Flow cytometry and antibodies SHP-1 and SHP-2 associate with immunoreceptor tyrosine-based switch B Retroviral transduction and Splenic chimera motif of programmed death 1 upon primary human T cell stimulation, but only receptor ligation prevents T cell activation. J. Immunol. 173, 945–954. B Adoptive cell transfer B Transfer model of colitis Choi, Y.S., Gullicksrud, J.A., Xing, S., Zeng, Z., Shan, Q., Li, F., Love, P.E., B Mixed bone marrow chimeras Peng, W., Xue, H.H., and Crotty, S. (2015). LEF-1 and TCF-1 orchestrate T(FH) differentiation by regulating differentiation circuits upstream of the tran- B Enzyme-linked immunosorbent assay scriptional repressor Bcl6. Nat. Immunol. 16, 980–990. B Microarray and bioinformatic analysis Chung, Y., Tanaka, S., Chu, F., Nurieva, R.I., Martinez, G.J., Rawal, S., Wang, B Quantitative RT-PCR analysis Y.H., Lim, H., Reynolds, J.M., Zhou, X.H., et al. (2011). Follicular regulatory B ( ) Chromatin immunoprecipitation ChIP T cells expressing Foxp3 and Bcl-6 suppress germinal center reactions. Nat. d QUANTIFICATION AND STATISTICAL ANALYSIS Med. 17, 983–988. d DATA AND SOFTWARE AVAILABILITY Crotty, S. (2014). T follicular helper cell differentiation, function, and roles in B Data source disease. Immunity 41, 529–542. Czar, M.J., Kersh, E.N., Mijares, L.A., Lanier, G., Lewis, J., Yap, G., Chen, A., SUPPLEMENTAL INFORMATION Sher, A., Duckett, C.S., Ahmed, R., and Schwartzberg, P.L. (2001). Altered lymphocyte responses and cytokine production in mice deficient in the Supplemental Information includes six figures and two tables and can be X-linked lymphoproliferative disease gene SH2D1A/DSHP/SAP. Proc. Natl. found with this article online at http://dx.doi.org/10.1016/j.immuni.2017. Acad. Sci. USA, 7449–7454. 08.011. Delgoffe, G.M., Kole, T.P., Zheng, Y., Zarek, P.E., Matthews, K.L., Xiao, B., Worley, P.F., Kozma, S.C., and Powell, J.D. (2009). The mTOR kinase differen- AUTHOR CONTRIBUTIONS tially regulates effector and regulatory T cell lineage commitment. Immunity 30, 832–844. L.X., Q.H., H.W., Y.H., Q.B., J.H., Y.L., P.W., X.C., R.H., B.L., X.Y., T.Z., Y.Z., Y.W., and J.O. performed the experiments; Q.B. analyzed the microarray data Delgoffe, G.M., Pollizzi, K.N., Waickman, A.T., Heikamp, E., Meyers, D.J., by GSEA; L.Y. designed the study, analyzed the data, and wrote the paper with Horton, M.R., Xiao, B., Worley, P.F., and Powell, J.D. (2011). The kinase L.X., Q.H., Y.W., X.Z., and H.L.; and L.Y. supervised the study. mTOR regulates the differentiation of helper T cells through the selective acti- vation of signaling by mTORC1 and mTORC2. Nat. Immunol. 12, 295–303. ACKNOWLEDGMENTS Dodd, K.M., Yang, J., Shen, M.H., Sampson, J.R., and Tee, A.R. (2015). mTORC1 drives HIF-1a and VEGF-A signalling via multiple mechanisms We thank the Institute Clinique de la Souris (part of the International Knockout involving 4E-BP1, S6K1 and STAT3. Oncogene 34, 2239–2250. Mouse Consortium) for permission to use the Tcf7 conditional knockout mice. Durant, L., Watford, W.T., Ramos, H.L., Laurence, A., Vahedi, G., Wei, L., We thank CapitalBio Corporation (Beijing, China) for performing microarray Takahashi, H., Sun, H.W., Kanno, Y., Powrie, F., and O’Shea, J.J. (2010). experiments and data analysis, and we thank the core facility center of Diverse targets of the transcription factor STAT3 contribute to T cell pathoge- Third Military Medical University for cell sorting. This study was supported nicity and homeostasis. Immunity 32, 605–615. by grants from the National Natural Science Foundation of China (no. 31500710 to X.Y.), the National Basic Research Program of China (973 Pro- Eri, R., McGuckin, M.A., and Wadley, R. (2012). T cell transfer model of colitis: gram, no. 2013CB531500 to L.Y.), the National Key Research Development a great tool to assess the contribution of T cells in chronic intestinal inflamma- Plan (no. 2016YFA0502202 to L.Y.), and the National Natural Science Founda- tion. Methods Mol. Biol. 844, 261–275. tion of China (no. 81471624 to L.Y.). Gatto, D., and Brink, R. (2010). The germinal center reaction. J. Allergy Clin. Immunol. 126, 898–907, quiz 908–909. Received: December 1, 2016 Haxhinasto, S., Mathis, D., and Benoist, C. (2008). The AKT-mTOR axis regu- Revised: May 12, 2017 lates de novo differentiation of CD4+Foxp3+ cells. J. Exp. Med. 205, 565–574. Accepted: August 18, 2017 He, R., Hou, S., Liu, C., Zhang, A., Bai, Q., Han, M., Yang, Y., Wei, G., Shen, T., Published: September 19, 2017 Yang, X., et al. (2016). Follicular CXCR5- expressing CD8(+) T cells curtail REFERENCES chronic viral infection. Nature 537, 412–428. Johnston, R.J., Poholek, A.C., DiToro, D., Yusuf, I., Eto, D., Barnett, B., Dent, Alexander, C.M., Tygrett, L.T., Boyden, A.W., Wolniak, K.L., Legge, K.L., and A.L., Craft, J., and Crotty, S. (2009). Bcl6 and Blimp-1 are reciprocal and Waldschmidt, T.J. (2011). T regulatory cells participate in the control of antagonistic regulators of T follicular helper cell differentiation. Science 325, germinal centre reactions. Immunology 133, 452–468. 1006–1010.

550 Immunity 47, 538–551, September 19, 2017 Johnston, R.J., Choi, Y.S., Diamond, J.A., Yang, J.A., and Crotty, S. (2012). Steinke, F.C., Yu, S., Zhou, X., He, B., Yang, W., Zhou, B., Kawamoto, H., Zhu, STAT5 is a potent negative regulator of TFH cell differentiation. J. Exp. Med. J., Tan, K., and Xue, H.H. (2014). TCF-1 and LEF-1 act upstream of Th-POK to 209, 243–250. promote the CD4(+) T cell fate and interact with Runx3 to silence Cd4 in CD8(+) Kelly, A.P., Finlay, D.K., Hinton, H.J., Clarke, R.G., Fiorini, E., Radtke, F., and T cells. Nat. Immunol. 15, 646–656. Cantrell, D.A. (2007). Notch-induced T cell development requires phosphoino- Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., sitide-dependent kinase 1. EMBO J. 26, 3441–3450. Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and Laplante, M., and Sabatini, D.M. (2012). mTOR signaling in growth control and Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based disease. Cell 149, 274–293. approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550. Laplante, M., and Sabatini, D.M. (2013). Regulation of mTORC1 and its impact on gene expression at a glance. J. Cell Sci. 126, 1713–1719. Takahashi, K., and Yamanaka, S. (2006). Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell Linterman, M.A., Pierson, W., Lee, S.K., Kallies, A., Kawamoto, S., Rayner, 126, 663–676. T.F., Srivastava, M., Divekar, D.P., Beaton, L., Hogan, J.J., et al. (2011). € Foxp3+ follicular regulatory T cells control the germinal center response. Vaeth, M., Muller, G., Stauss, D., Dietz, L., Klein-Hessling, S., Serfling, E., Lipp, Nat. Med. 17, 975–982. M., Berberich, I., and Berberich-Siebelt, F. (2014). Follicular regulatory T cells control humoral autoimmunity via NFAT2-regulated CXCR5 expression. Liu, G., Burns, S., Huang, G., Boyd, K., Proia, R.L., Flavell, R.A., and Chi, H. J. Exp. Med. 211, 545–561. (2009). The receptor S1P1 overrides regulatory T cell-mediated immune sup- Wang, C.J., Heuts, F., Ovcinnikovs, V., Wardzinski, L., Bowers, C., Schmidt, pression through Akt-mTOR. Nat. Immunol. 10, 769–777. E.M., Kogimtzis, A., Kenefeck, R., Sansom, D.M., and Walker, L.S. (2015). Nurieva, R.I., Chung, Y., Martinez, G.J., Yang, X.O., Tanaka, S., Matskevitch, CTLA-4 controls follicular helper T-cell differentiation by regulating the T.D., Wang, Y.H., and Dong, C. (2009). Bcl6 mediates the development of strength of CD28 engagement. Proc. Natl. Acad. Sci. USA 112, 524–529. T follicular helper cells. Science 325, 1001–1005. Published online December 29, 2014. Pearce, E.L., Walsh, M.C., Cejas, P.J., Harms, G.M., Shen, H., Wang, L.S., Wollenberg, I., Agua-Doce, A., Herna´ ndez, A., Almeida, C., Oliveira, V.G., Faro, Jones, R.G., and Choi, Y. (2009). Enhancing CD8 T-cell memory by modulating J., and Graca, L. (2011). Regulation of the germinal center reaction by Foxp3+ fatty acid metabolism. Nature 460, 103–107. follicular regulatory T cells. J. Immunol. 187, 4553–4560. Powell, J.D., Pollizzi, K.N., Heikamp, E.B., and Horton, M.R. (2012). Regulation Xie, D.L., Wu, J., Lou, Y.L., and Zhong, X.P. (2012). Tumor suppressor TSC1 is of immune responses by mTOR. Annu. Rev. Immunol. 30, 39–68. critical for T-cell anergy. Proc. Natl. Acad. Sci. USA 109, 14152–14157. Qi, H., Cannons, J.L., Klauschen, F., Schwartzberg, P.L., and Germain, R.N. Xu, X., Ye, L., Araki, K., and Ahmed, R. (2012). mTOR, linking metabolism and (2008). SAP-controlled T-B cell interactions underlie germinal centre forma- immunity. Semin. Immunol. 24, 429–435. tion. Nature 455, 764–769. Xu, L., Cao, Y., Xie, Z., Huang, Q., Bai, Q., Yang, X., He, R., Hao, Y., Wang, H., Ray, J.P., Staron, M.M., Shyer, J.A., Ho, P.C., Marshall, H.D., Gray, S.M., Zhao, T., et al. (2015). The transcription factor TCF-1 initiates the differentiation Laidlaw, B.J., Araki, K., Ahmed, R., Kaech, S.M., and Craft, J. (2015). The inter- of T(FH) cells during acute viral infection. Nat. Immunol. 16, 991–999. leukin-2-mTORc1 kinase axis defines the signaling, differentiation, and meta- Yang, J., Lin, X., Pan, Y., Wang, J., Chen, P., Huang, H., Xue, H.H., Gao, J., and bolism of T helper 1 and follicular B helper T cells. Immunity 43, 690–702. Zhong, X.P. (2016). Critical roles of mTOR Complex 1 and 2 for T follicular Sage, P.T., and Sharpe, A.H. (2015). T follicular regulatory cells in the regula- helper cell differentiation and germinal center responses. Elife 5, e17936. tion of B cell responses. Trends Immunol. 36, 410–418. Yu, D., Rao, S., Tsai, L.M., Lee, S.K., He, Y., Sutcliffe, E.L., Srivastava, M., Sage, P.T., and Sharpe, A.H. (2016). T follicular regulatory cells. Immunol. Rev. Linterman, M., Zheng, L., Simpson, N., et al. (2009). The transcriptional 271, 246–259. repressor Bcl-6 directs T follicular helper cell lineage commitment. Immunity Sage, P.T., Francisco, L.M., Carman, C.V., and Sharpe, A.H. (2013). The re- 31, 457–468. ceptor PD-1 controls follicular regulatory T cells in the lymph nodes and blood. Yusuf, I., Kageyama, R., Monticelli, L., Johnston, R.J., Ditoro, D., Hansen, K., Nat. Immunol. 14, 152–161. Barnett, B., and Crotty, S. (2010). Germinal center T follicular helper cell IL-4 Sage, P.T., Paterson, A.M., Lovitch, S.B., and Sharpe, A.H. (2014). The coinhi- production is dependent on signaling lymphocytic activation molecule recep- bitory receptor CTLA-4 controls B cell responses by modulating T follicular tor (CD150). J. Immunol. 185, 190–202. helper, T follicular regulatory, and T regulatory cells. Immunity 41, 1026–1039. Zeng, H., Yang, K., Cloer, C., Neale, G., Vogel, P., and Chi, H. (2013). mTORC1 Sauer, S., Bruno, L., Hertweck, A., Finlay, D., Leleu, M., Spivakov, M., Knight, couples immune signals and metabolic programming to establish T(reg)-cell Z.A., Cobb, B.S., Cantrell, D., O’Connor, E., et al. (2008). T cell receptor function. Nature 499, 485–490. signaling controls Foxp3 expression via PI3K, Akt, and mTOR. Proc. Natl. Zeng, H., Cohen, S., Guy, C., Shrestha, S., Neale, G., Brown, S.A., Cloer, C., Acad. Sci. USA 105, 7797–7802. Kishton, R.J., Gao, X., Youngblood, B., et al. (2016). mTORC1 and mTORC2 Sheppard, K.A., Fitz, L.J., Lee, J.M., Benander, C., George, J.A., Wooters, J., kinase signaling and glucose metabolism drive follicular helper T cell differen- Qiu, Y., Jussif, J.M., Carter, L.L., Wood, C.R., and Chaudhary, D. (2004). PD-1 tiation. Immunity 45, 540–554. inhibits T-cell receptor induced phosphorylation of the ZAP70/CD3zeta signal- Zoncu, R., Efeyan, A., and Sabatini, D.M. (2011). mTOR: from growth signal osome and downstream signaling to PKCtheta. FEBS Lett. 574, 37–41. integration to cancer, diabetes and ageing. Nat. Rev. Mol. Cell Biol. 12, 21–35.

Immunity 47, 538–551, September 19, 2017 551 STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Rat monoclonal PerCP anti-mouse CD4 (clone RM4-5) Biolegend Cat#100538; RRID: AB_893325 Rat monoclonal PerCP Anti-Mouse CD8a (clone 53-6.7) BD Biosciences Cat#553036; RRID: AB_394573 Rat monoclonal Pacific Blue anti-mouse CD19 (clone 6D5) Biolegend Cat#115523; RRID: AB_439718 Rat monoclonal PE anti-mouse CD25 (clone PC61) Biolegend Cat#102008; RRID: AB_312857 Rat monoclonal PE-Cyanine7 anti-mouse CD44 (clone IM7) eBioscience Cat# 25-0441-82; RRID: AB_469623 Armenian Hamster monoclonal PE Anti-Mouse CD95(FAS) BD Biosciences Cat#554258; RRID: AB_395330 (clone Jo2) Armenian Hamster monoclonal APC anti-human/mouse/rat Biolegend Cat#313510; RRID: AB_416334 CD278 (ICOS) (clone C398.4A) Mouse (A.SW) monoclonal APC anti-mouse CD45.1 (clone A20) Biolegend Cat#110714; RRID: AB_313503 Mouse (SJL) monoclonal PE anti-mouse CD45.2 (clone 104) Biolegend Cat#109808;RRID: AB_313445 Rat monoclonal FITC anti-mouse CD279 (PD-1) (clone RMP1-30) eBioscience Cat# 11-9981-82; RRID: AB_465467 Rat monoclonal eFluor 660 anti-mouse FOXP3 (clone FJK-16 s) eBioscience Cat#50-5773-82; RRID: AB_11218868 Rat monoclonal APC anti-mouse CD357 (AITR/GITR) eBioscience Cat#17-5874-81; RRID: AB_469461 (clone DTA-1) Mouse monoclonal PE anti-Bcl-6 (clone K112-91) BD Biosciences Cat#561522; RRID: AB_10717126 Rat monoclonal APC anti-mouse CD71 (Transferrin Receptor) eBioscience Cat#17-0711-82; RRID: AB_1834355 (clone R17217 (RI7 217.1.4)) Rat monoclonal PE anti-mouse CD73 (clone TY/11.8) Biolegend Cat#127206; RRID: AB_2154094 Rat monoclonal PE anti-mouse CD98 (clone RL388) eBioscience Cat#12-0981-83; RRID: AB_465794 Armenian Hamster monoclonal PE anti-mouse CD152 (CTLA-4) Biolegend Cat#106306; RRID: AB_313255 (clone UC10-4B9) Fluorescein labeled Peanut Agglutinin (PNA) (clone FL-1071) Vector Labs Cat#FL-1071; RRID: AB_2315097 Rat monoclonal PE Anti-Mouse CD138 (clone 281-2) BD Biosciences Cat#553714; RRID: AB_395000 Rat monoclonal APC-Cyanine7 anti-mouse CD45R eBioscience Cat#A15399; RRID: AB_2534413 (clone RA3-6B2) Streptavidin PE-Cyanine7 eBioscience Cat#25-4317-82; RRID: AB_10116480 Brilliant Violet 421 Streptavidin Biolegend Cat#405226 LIVE/DEAD Fixable Near-IR Dead Cell Stain Kit, for 633 Life Technologies Cat#L10199 or 635 nm excitation Biotin-SP (long spacer) AffiniPure Goat Anti-Rat IgG (H+L) Jackson Immunoresearch Cat#112-065-143 Purified Rat Anti-Mouse CD16/CD32 (Mouse BD Fc Block) BD Biosciences Cat#553142 (clone 2.4G2) Purified Rat Anti-Mouse CXCR5 (clone 2G8) BD Biosciences Cat#551961 Rabbit monoclonal anti-Human,Mouse,Rat TCF1 (C46C7) Cell Signaling Technology Cat#2206S Rabbit monoclonal anti-Human,Mouse,Rat Phospho-Stat3 Cell Signaling Technology Cat#9134L (Ser727) Rabbit monoclonal anti-Human,Mouse,Rat Phospho-4E-BP1 Cell Signaling Technology Cat#2855S; RRID: AB_560835 (Thr37/46) (236B4) Rabbit monoclonal anti-Human,Mouse,Rat Phospho-S6 Cell Signaling Technology Cat#4858S; RRID: AB_916156 Ribosomal Protein (Ser235/236) (D57.2.2E) Rabbit monoclonal anti-Human,Mouse,Rat Phospho-FoxO1 Cell Signaling Technology Cat#9464S; RRID: AB_329842 (Thr24)/FoxO3a (Thr32) Rabbit monoclonal anti-Human,Mouse,Rat Phospho-Akt Cell Signaling Technology Cat#4060S; RRID: AB_2315049 (Ser473) (D9E) Donkey anti-Rabbit Polyclonal IgG (H+L) Highly Cross- ThermoFisher Scientific Cat#A-21206; RRID: AB_141708 Adsorbed Secondary Antibody, Alexa Fluor 488 (Continued on next page) e1 Immunity 47, 538–551.e1–e5, September 19, 2017 Continued REAGENT or RESOURCE SOURCE IDENTIFIER Cell Proliferation Dye eFluor 450 eBioscience Cat#65-0842 APC BrdU Flow Kit BD Biosciences Cat#552598 Armenian Hamster anti-Mouse LEAF Purified CD3ε(clone Biolegend Cat#100314; RRID: AB_312679 145-2C11) Syrian Hamster anti-Mouse Purified CD28 (clone 37.51) Biolegend Cat#102102; RRID: AB_312867 Biotin Rat anti-Mouse CD8a (clone 53-6.7) Biolegend Cat#100704; RRID: AB_312743 Biotin Rat anti-Mouse CD19 (clone 6D5) Biolegend Cat#115504; RRID: AB_313639 Biotin Rat anti-mouse/human CD45R/B220(clone RA3-6B2) Biolegend Cat#103204; RRID: AB_312989 Biotin Rat anti-mouse CD11c (clone N418) Biolegend Cat#117304; RRID: AB_313773 Biotin Rat anti-mouse Ly-6G (clone 1A8) Biolegend Cat#127604; RRID: AB_1186108 Biotin Rat anti-mouse TER-119/Erythroid Cells (clone TER-119) Biolegend Cat#116204 Biotin Rat anti-mouse NK-1.1 (clone PK136) Biolegend Cat#108704 Mouse PE Anti-Ki-67 Set BD Biosciences Cat# 556027; RRID: AB_2266296 Hamster FITC Anti-Mouse Bcl-2 Set BD Biosciences Cat# 554221; RRID: AB_395312 Rabbit FITC anti-Active Caspase-3 Apoptosis Kit BD Biosciences Cat# 550480; RRID: AB_393697 APC BrdU Flow Kit BD Biosciences Cat# 552598 PE Annexin V Apoptosis Detection Kit eBioscience Cat# 88-8102-72; RRID: AB_2575183 Bacterial and Virus Strains LCMV Armstrong R. Ahmed Emory University Chemicals, Peptides, and Recombinant Proteins Ovalbumin (OVA) Sigma Cat#A5503 NP(16)-OVA Biosearch Technology Cat#N-5051-100 Rapamycin Hua Mai Ke Cat#R081502 Freund’s Adjuvant, Complete, cell suspension Sigma Cat#F5881 Critical Commercial Assays SimpleChIP Enzymatic Chromatin IP Kit (Magnetic Beads) Cell Signaling Technology Cat#9003 BeaverBeads Mag500 Streptavidin Matrix Beaver Cat#22302 Deposited Data Microarray GEO GSE78190 Experimental Models: Cell Lines HEK293T ATCC Cat# CRL-3216, RRID: CVCL_0063 Experimental Models: Organisms/Strains Mouse: C57BL/6J (CD45.2 and CD45.1) Jackson Laboratory RRID: IMSR_JAX:000664 Mouse: B6.Cg-Rptortm1.1Dmsa/J (Rptorfl/fl) Jackson Laboratory RRID: IMSR_JAX:013188 Mouse: STOCK Rictortm1.1Klg/SjmJ (Rictorfl/fl) Jackson Laboratory RRID: IMSR_JAX:020649 Mouse: STOCK Tsc1tm1Djk/JT (Tsc1fl/fl) Jackson Laboratory N/A Mouse: B6.Cg-Tg(Cd4-cre)1Cwi/BfluJ (Cd4-Cre) Jackson Laboratory RRID: IMSR_JAX:022071 Mouse: B6.129(Cg)-Foxp3tm4(YFP/icre)Ayr/J Jackson Laboratory RRID: IMSR_JAX:016959 Mouse: C57BL/6-Tg(Foxp3-GFP)90Pkraj/J Jackson Laboratory RRID: IMSR_JAX:023800 Mouse: B6. Sh2d1a–/– Czar et al., 2001 N/A Mouse: B6. Tcf7fl/fl Steinke et al., 2014 N/A Mouse: Rag1–/–- Jackson Laboratory RRID: IMSR_JAX:002216 Oligonucleotides Primer: Bcl6-BglII-Forward Synthesized by Invitrogen N/A AGTCAGATCTCCACCATGGCCTCCCCGGCTGACA Primer: Bcl6-HpaI-Reverse Synthesized by Invitrogen N/A GTTAACTCAGCAGGCTTTGGGGAGCT Primer: Tcf7-BglII-Forward Synthesized by Invitrogen N/A AGTCAGATCTCCACCATGTACAAAGAGACTGTCTACTCTG (Continued on next page)

Immunity 47, 538–551.e1–e5, September 19, 2017 e2 Continued REAGENT or RESOURCE SOURCE IDENTIFIER Primer: Tcf7-SalI-Reverse Synthesized by Invitrogen N/A AGTCGTCGACCTAGAGCACTGTCATCGGAAG Primer: Stat3-HpaI-Forward Synthesized by Invitrogen N/A GTTAACCACCATGGCTCAGTGGAACCAGCT Primer: Stat3a-EcoRI-Reverse Synthesized by Invitrogen N/A AGTCGAATTCACATGGGGGAGGTAGCAC Primer: Stat3-CA (constitutive-active)-M Forward Synthesized by Invitrogen N/A CATGGATTGCACCTGCATCCTGGTGTCTCCACTTG Primer: Stat3-CA (constitutive-active)-M Reverse Synthesized by Invitrogen N/A GGATGCAGGTGCAATCCATGATCTTATAGCCCATGATG Primers for qPCR and Chip-qPCR, see Table S2 N/A Recombinant DNA Plasmid: MIGR1 (MSCV-IRES-GFP) Xu et al., 2015 N/A Plasmid: MIGR1-Tcf7 overexpressing Xu et al., 2015 N/A Plasmid: MIGR1-Bcl6 overexpressing Xu et al., 2015 N/A Plasmid: STAT3-CA (constitutive-active) This paper N/A Software and Algorithms FlowJo (Tree Star) https://www.flowjo.com/ RRID: SCR_008520 solutions/flowjo Gene Set Enrichment Analysis (GSEA) http://software.broadinstitute. RRID: SCR_003199 org/gsea/index.jsp

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Lilin Ye ([email protected]).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice Rptorfl/fl, Rictorfl/fl, Tsc1fl/fl, Cd4-Cre transgenic, Foxp3YFP-Cre knock-in, Foxp3-GFP reporter, Rag1–/– and C57BL/6J (CD45.2 and CD45.1) mice were obtained from the Jackson Laboratory. Sh2d1a–/– mice were previously described (Czar et al., 2001) and kindly provided by Dr. Hai Qi (Tsinghua University). Tcf7fl/fl mice have been described previously (Steinke et al., 2014; Xu et al., 2015) and were provided by H.H. Xue (University of Iowa) with permission from Institute Clinique de la Souris (part of the International Knockout Mouse Consortium). In brief, Tcf7-targeted mice were crossed with Rosa26-Flippase knock-in mice (Jackson Laboratory) to delete the LacZ-Neo cassette flanked by the Frt sites, converting the targeted allele into Tcf7-floxed allele (Tcf7fl/+). Rptorfl/fl, Rictorfl/fl, Tsc1fl/fl and Tcf7fl/fl mice were bred with Cd4-Cre transgenic or Foxp3YFP-Cre knock-in mice to generate Rptorfl/flCd4-Cre, Rptorfl/fl Foxp3Cre, Rptorfl/+ Foxp3Cre, Rictorfl/fl Cd4-Cre, Tsc1fl/fl Foxp3Cre and Tcf7fl/flCd4-Cre mice. All these strains are C57BL/6 back- ground. All the mice used were analyzed at 6–10 weeks of age, and both genders were included without randomization or ‘‘blinding.’’ BM chimeras were infected after 8–10 weeks of reconstitution. For all the phenotypic analysis, at least 3 animals of each genotype with matched age and gender were analyzed. All mice were housed in specific-pathogen free (SPF) condition. All mouse experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committees of the Third Military Medical University.

METHOD DETAILS

Protein immunization and acute viral infection Ovalbumin (OVA) (A5503; Sigma) or NP(16)-OVA (N-5051-100; Biosearch Technology) was 1:1 emulsified with Freund’s Adjuvants, Complete (CFA) (F5881; Sigma) and injected mice subcutaneously (100 mg of protein/mouse). LCMV-Arm+ virus was provided by R. Ahmed (Emory University), and 2 3 105 plaque-forming units (p.f.u.) were injected intraperitoneal to establish acute infection in mice.

e3 Immunity 47, 538–551.e1–e5, September 19, 2017 Flow cytometry and antibodies Flow cytometry data were acquired by FACSCanto II (BD Biosciences) and analyzed using FlowJo (Tree Star). The antibodies and reagents used for flow cytometry staining were listed in key resources table. Surface staining was performed in PBS containing 2% BSA or fetal bovine serum (wt/vol). For detection of phosphorylated mTOR signaling proteins, spleen or dLN derived lympho- cytes were first stained with surface markers and then were stimulated with anti-CD3 (2 mg /ml, 100302; Biolegend) and anti- CD28 (0.5 mg /ml, 102102; Biolegend) at 37C for 1 hr. After 1 hr stimulation, cells were immediately fixed with Phosflow Lyse/ Fix buffer (558049; BD Biosciences), followed by permeabilization with Phosflow Perm buffer I (557885; BD Biosciences) and staining with primary unconjugated antibodies to p-S6 (Ser 235/236) (D57.2.2E; Cell Signaling Technology), p-4E-BP1 (Thr 37/46) (236B4; Cell Signaling Technology), p-Akt (Ser 473) (#4060S; Cell Signaling Technology), p-FoxO1/3a (#9464S; Cell Signaling Technology). Then, primary unconjugated antibodies were detected by secondary staining with anti-rabbit IgG A488 antibody (A21206; Invitrogen). p-STAT3(49/p-Stat3, BD bioscience) staining were performed after surface staining and stimulated with IL-6 (10ng/ml, 575702; Biolegend) at 37C for 1 hr. CXCR5 staining was described previously (Xu et al., 2015). Briefly, cells were stained with purified anti-CXCR5 (BD PharMingen) for 1 hr in 4C, followed by biotinylated anti-rat IgG (Jackson Immunoresearch), and then APC-labeled streptavidin (Caltag Laboratories) in PBS + 0.5% BSA + 2% FCS + 2% Normal Mouse Serum on ice. Bcl-6, TCF-1 and Foxp3 staining were performed with the Foxp3/Transcription Factor Staining Buffer Set (00-5523; eBioscience). Staining for Ki67, Bcl2 and cleaved Caspase3 were performed with a Cytofix/ Cytoperm Fixation/Permeabilization Kit (554722, BD Biosciences). Annexin V staining were performed with a Annexin V Kit (88-8102-72; eBioscience) according to the manufacturer’s instructions. For in vivo incorporation of the thymidine analog BrdU, mice were given BrdU (1.5 mg BrdU (5-bromodeoxyuridine) in 0.5 mL PBS) intraperitoneally 3 hr before mice were sacrificed. BrdU in T cells was stained with a BrdU Flow Kit (552598; BD Biosciences) according to the manufacturer’s instructions.

Retroviral transduction and Splenic chimera The Tcf7 coding sequences (p33 isoform), Bcl6 coding sequences and Stat3 coding sequence were amplified and cloned into the vectors MIGR1(MSCV-IRES-GFP). The constitutive-active mutant of Stat3 (A662C and N664C, referred as STAT3-CA) was intro- duced by PCR-based site-directed mutagenesis (Takahashi and Yamanaka, 2006). Primers used here were listed in key resources table. Retroviruses were packaged by transfection of 293T cells with the retroviral vectors along with the pCLeco plasmid. CD4+ GTIR+CD25+CXCR5– Treg cells were sorted from Rptor–/– mice (CD45.2+) and activated in vitro by incubating with pre-coated anti-CD3 (2 mg /ml, 100302; Biolegend) and anti-CD28 (0.5 mg /ml, 102102; Biolegend) for 48h at 37C. Subsequently, activated lym- phocytes were spin-infected by centrifugation (800g) with freshly harvested retrovirus supernatants, 8 mg/ml polybrene (H9268; Sigma-Aldrich) and 20 ng/ml of IL-2 (130-098-221; Miltenyi Biotec) at 37C for 120 min. Treg cells were cultured at 37C for 2 days and then were mixed with wide-type (CD45.1+) splenocytes and transferred to sub-lethally irradiated recipient mice (600 rads, CD45.1+), which were infected with LCMV-Arm+ 8-12h later as we previously described (He et al., 2016). A total of 5 3 107 mixed cells (including at least 8 3 105 Rptor–/– Treg cells) were adoptively transferred. Tfr cell differentiation was analyzed in splenic chimeras on day 8 after infection.

Adoptive cell transfer In rapamycin- or vehicle control-treated Treg transfer, for day 8 analysis, 5 3 105 Treg cells (CD4+CD19–GITR+CD25+CXCR5–) from WT mice (CD45.2+) were sorted and treated with rapamycin (5mg/ml, R081502; Hua Mai Ke) or vehicle at 37C for 4h, followed by transferred to recipient mice (CD45.1+) followed by immunization. Otherwise, for proliferation assays, with or without rapamycin treat- ment, 2 3 106 GFP+CXCR5– Treg cells from Foxp3-GFP reporter mice were subsequently labeled with proliferation-indicating dye (Celltrace Violet, 3 mM, 65-0842; eBioscience) according to the manufacturer’s guide. And these labeled cells were transferred into day 7 OVA/CFA immunized recipient mice, followed by analysis on day 2 or 3 after cell transfer. For Tfr-cell function evaluation, 8 3 105 of sorted WT Tfh (CD4+CD19–CD25–GITR–CD44+CXCR5+) cells together with 4x105 Tfr (CD4+CD19–GITR+CD25+CXCR5+) cells (treated with rapamycin or vehicle control) from WT mice (CD45.2+, day 7 after LCMV-Arm+ infection) were adoptively trans- ferred into day 1 infected recipient mice (Sap–/–), followed by analysis on day 7 after cell transfer.

Transfer model of colitis A total of 4 3 105 CD4 Teff cells sorted by FACS (CD4+CD25–GITR–) from CD45.1+ C57BL/6 mice were transferred intraperitoneally into Rag1–/– mice alone or in combination with CD45.2+ Treg cells from Rptorfl/fl Foxp3Cre, Rptorfl/+ Foxp3Cre or WT mice. After T cell reconstitution, mice were weighed and monitored for signs of disease. Mice were euthanized at day 35 after T cell transfer and their colons were dissected and fixed in 4% PFA (vol/vol). Then the fixed tissues were embedded in paraffin. And 5-mm thickness sections were stained with hematoxylin and eosin (H.E staining) as described (Eri et al., 2012; Zeng et al., 2013).

Mixed bone marrow chimeras Bone marrow (BM) cells were harvested from WT (CD45.1+) and Rptorfl/fl Foxp3Cre (CD45.2+) mice. For each chimera, 5 3 106 BM cells (Rptorfl/fl Foxp3Cre and C57BL/6J at a ratio of 3:7) were transferred intravenously into lethally irradiated (two doses of 550 rads each) WT (CD45.1+) recipients. After 8–10 weeks of reconstitution, recipient mice were ready for OVA/CFA immunization.

Immunity 47, 538–551.e1–e5, September 19, 2017 e4 Enzyme-linked immunosorbent assay For analysis of antibody production, serum was collected from mice at day 8 post NP-OVA/CFA immunization. NP-specific IgG was titrated by incubation of serum for 1.5 hr in ELISA plate coated with 500ng NP(18)-CGG (N-5055B-5; Biosearch Technology) per well, followed by incubation with the secondary antibody horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG (1036-05; SouthernBiotech).

Microarray and bioinformatic analysis To isolate Tfr cells, total lymphocytes from WT and Rptor–/– mice on day 8 post immunization were subjected to lineage depletion using biotin-conjugated antibodies (CD8 (53-6.7), B220 (RA3-6B2), CD11b (M1/70), CD11c (N418), Gr-1 (RB6-8C5), TER119 (TER-119) and NK1.1 (PK136), Biolegend or eBioscience) coupled with BeaverBeads Mag500 Streptavidin Matrix (22302; Beaver). The CD4+CD19–GITR+CD25+ CXCR5+ Tfr cells were soretd by a FACSAria II cell sorter (BD Biosciences) at the core facility center of Third Military Medical University. Sorted cells were lysed with TRIzol LS reagent (10296; Life Technologies), and total RNA was ex- tracted according to the TRIzol reagent protocol and submitted to CapitalBio Corporation for microarray analysis. Gene set enrich- ment analysis (GSEA; https://www.broadinstitute.org/gsea/) was performed as previously described (Subramanian et al., 2005).

Quantitative RT-PCR analysis To compare gene expression in Tfr cells from WT and Rptor–/– mice, Total RNA was extracted using TRIzol LS reagent (10296, Life Technologies). RevertAid H Minus First Strand cDNA Synthesis Kit (K1632; Thermo Scientific) were used to generate cDNA. The expression of various genes was examined using AceQ qPCR SYBR Green Master Mix (Q111; Vazyme) on a CFX96 Touch Real-Time System (Bio-Rad). The primers for the test genes are listed in Table S2.

Chromatin immunoprecipitation (ChIP) Sorted CD4+CD19–GITR+CD25+ Treg cells were stimulated with IL-6 (10ng/ml, Biolegend) at 37C for 1 hr. Then 5 3 106 treated cells were conducted ChIP assay with the SimpleChIP Enzymatic Chromatin IP Kit (Magnetic Beads) (#9003; Cell Signaling Technology) according to the manufacturer’s instructions. Chromatin fragments were immunoprecipitated with anti-p-STAT3 (#9134; Cell Signaling Technology), anti-TCF-1 (C46C7; Cell Signaling Technology) or rabbit IgG (#2729; Cell Signaling Technology) coupled with ChIP Grade Protein G Magnetic Beads (#9006; Cell Signaling Technology). After DNA purification, quantitative PCR was per- formed using primers (Table S2) flanking the putative STAT3 or TCF-1-binding sites. The fold enrichment was calculated by normal- izing samples of anti-TCF-1 or p-STAT3 to normal rabbit IgG controls.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis was performed with Prism 6.0 (GraphPad). Unpaired or paired (for retroviral transduction, splenic chimera and BM chimera experiments) two-tailed t test with 95% confidence interval was used to calculate P value. n represents the number of mice used in the experiment, with the number of individual experiments listed in the legend. Graphs show individual samples and center values indicate mean. Significance was defined as p % 0.05.

DATA AND SOFTWARE AVAILABILITY

Data source The accession number for the Rptor/ and WT Tfr Microarray data reported in this paper is GEO: GSE78190 (https://www.ncbi.nlm. nih.gov/geo/query/acc.cgi?token=onmjcqumllkphmt&acc=GSE78190).

e5 Immunity 47, 538–551.e1–e5, September 19, 2017 Article

Cite This: Anal. Chem. 2018, 90, 2355−2361 pubs.acs.org/ac

Aptamer-Based Dual-Functional Probe for Rapid and Specific Counting and Imaging of MCF‑7 Cells Bin Yang, Beibei Chen, Man He, Xiao Yin, Chi Xu, and Bin Hu* Key Laboratory of Analytical Chemistry for Biology and Medicine (Ministry of Education), Department of Chemistry, Wuhan University, Wuhan 430072, China

*S Supporting Information

ABSTRACT: Development of multimodal detection technologies for accurate diagnosis of cancer at early stages is in great demand. In this work, we report a novel approach using an aptamer-based dual-functional probe for rapid, sensitive, and specific counting and visualization of MCF-7 cells by inductively coupled plasma−mass spectrometry (ICP-MS) and fluorescence imaging. The probe consists of a recognition unit of aptamer to catch cancer cells specifically, a fluorescent dye (FAM) moiety for fluorescence resonance energy transfer (FRET)-based “off-on” fluorescence imaging as well as gold nanoparticles (Au NPs) tag for both ICP- MS quantification and fluorescence quenching. Due to the signal amplification effect and low spectral interference of Au NPs in ICP-MS, an excellent linearity and sensitivity were achieved. Accordingly, a limit of detection of 81 MCF-7 cells and a relative standard deviation of 5.6% (800 cells, n = 7) were obtained. The dynamic linear range was 2 × 102 to 1.2 × 104 cells, and the recoveries in human whole blood were in the range of 98−110%. Overall, the established method provides quantitative and visualized information on MCF-7 cells with a simple and rapid process and paves the way for a promising strategy for biomedical research and clinical diagnostics.

ancer is one of the most serious threats to public health based methods have been extended to the determination of C worldwide; it still causes high mortality rate even with the various biomolecules, including peptides,9 proteins,10,11 developed level of clinic medical treatment nowadays.1 DNA,12,13 and RNA,14 as well as cells,15,16 since the pioneering According to the latest cancer statistics released by the National work reported by Zhang et al. in 2001.17 Owing to the superior Central Cancer Registry of China, 4.29 million new cancer performance of ICP-MS for element-specific analysis, ICP-MS- cases and 2.81 million cancer deaths occurred in China in based methods usually exhibit extraordinary advantages such as 2 2015. Specifically, breast cancer is the most common cancer high sensitivity and accuracy, wide dynamic linear range, robust among women, which accounts for 15% of all new cancers in resistance to the matrix, and multiplex detection capability for women. The main cause of cancer-related death is cancer elements and isotopes. Nevertheless, the limitations of ICP- metastasis, which may be largely attribute to the tumor cells MS-based methods should still be noted especially for cell dislodged from tumors into the blood, also known as circulating 3 analysis. As a destructive detection technique, ICP-MS has to tumor cells (CTCs). Early and accurate detection of CTCs face the challenge of providing an in situ noninvasive image that buried in the ocean of numerous normal and benign cells is ff fi would o er unrivaled spatiotemporal resolution and help to therefore signi cant as a general strategy to monitor and exclude false-negative/positive results. Therefore, a combina- prevent the development of cancer and also as a guide to tion of ICP-MS-based quantitative bioanalysis with fluorescence effective therapeutic treatments.4 Toaddressthisgreat imaging is urgently needed to offer more comprehensive and challenge, novel approaches for rapid, sensitive, noninvasive, valuable information over each single assay for more reliable reliable, and accurate detection of extremely rare CTCs in diagnosis of disease and better patient treatment. clinical blood samples are in great demand. To address these issues, we have previously proposed two Up to now, many efforts have been made regarding the kinds of strategies to bridge the barrier between optical imaging detection of CTCs to improve cancer diagnosis and prognosis, 18 fl 5 6 and ICP-MS quantification by introducing quantum dots and including ow cytometry, electrochemical method, quantita- 19 tive real-time reverse transcriptase-polymerase chain reaction upconversion nanoparticles as multifunctional tags, respec- assay,7 inductively coupled plasma-mass spectrometry (ICP- tively. Moreover, Zhang and co-workers reported an integrin- MS)-based method,8 and so on. Among all these methods, ICP- targeted trifunctional probe that comprised a guiding unit, an MS combined with varieties of element-labeling strategies has emerged and become a promising approach for quantitative Received: November 28, 2017 bioanalysis in recent years. By applying suitable elemental tags Accepted: January 8, 2018 as endogenous elemental tags to label target objects, ICP-MS- Published: January 8, 2018

© 2018 American Chemical Society 2355 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361 Analytical Chemistry Article

Figure 1. Schematic illustration of the experimental principle for counting and imaging of cancer cells based on the dual-functional probes. europium chelate tag, and a fluorescent moiety for seeing and this method was not explored by real biological samples counting cancer cells.20 However, these reported methods more analysis. or less suffer from time-consuming operations and complicated Inspired by this idea, we attempt to develop an aptamer dual- fabrication. Besides, the antibody-involved method also has functional probe for rapid and specific counting and imaging of several drawbacks such as high cost and limited analysis targets. cancer cells by fluorescence microscope and ICP-MS. To Therefore, more effort should be taken for the development of demonstrate the feasibility, human breast cancer cell MCF-7 simpler, less time-consuming, less expensive, and more was chosen as a model cell line in this work. A 25-base ′ ′ straightforward approaches that combine fluorescence imaging oligonucleotide (sequence from 5 to 3 : GCA GTT GAT CCT fi TTG GAT ACC CTG G) selected by Ferreira et al. has been and ICP-MS quanti cation. fi fi With the advances in nanotechnology, innumerable nano- widely con rmed as possessing a speci c binding property for mucin 1 protein (MUC1), which is overexpressed on the materials have been synthesized and widely utilized in various 33 fields. Among them, gold nanoparticles (Au NPs) possess surface of MCF-7 cell. In addition, rapid separation of target distinct chemical, physical, and biological properties, and have cells from complex biological matrix under mild conditions was achieved with the introduction of magnetic beads (MBs) in the been demonstrated a good application potential in many fields probes, which not only greatly shortened the analysis time but especially in biomedical analysis.21,22 Au NPs are stable, also extended the application to real clinical samples. biocompatible, slightly toxic, and have a large molar extinction coefficient and broad energy bandwidth. Moreover, Au NPs are also a kind of attractive elemental tag with low background in ■ EXPERIMENTAL SECTION biological sample and satisfactory signal amplification effect.23,24 Apparatus. An Agilent 7500a ICP-MS (Agilent Technol- Compared to those metal chelates tags containing only one ogies, Japan) equipped with a Babington nebulizer was used to ICP-MS detectable atom in each chelate, a 15 nm diameter Au determine Au by monitoring 197Au. The operating parameters NP contains ∼106 Au atoms, which could potentially amplify for ICP-MS detection are summarized in Table S1.AnH- the signal by up to 6 orders of magnitude and would 7000FA transmission electron microscope (TEM, Hitachi, significantly improve the sensitivity of the assay.25 Japan) was used to obtain the TEM images with an acceleration Aptamers are single-stranded oligonucleotide or peptide voltage of 100 kV. Fluorescence spectra were obtained by an LS sequences that can recognize and bind to specific targets 55 spectrophotometer (PerkinElmer, U.S.A.) with the synchro- fi ff Δλ through folding into unique secondary or tertiary structures. nous scan method. The xed wavelength di erence ( )of fl They are generated by an in vitro screening process synchronous scanning uorescence spectroscopy was set at 22 amplification technique known as SELEX (systematic evolution nm, and slit widths of excitation and emission were both set at 10 nm. An AxioObserver Z1 inverted fluorescent microscope of ligands by exponential enrichment) from a large random fl oligonucleotide or peptide sequence pool.26 Aptamers have the (Zeiss, Germany) was utilized for observation and uorescence imaging. ability to bind with a variety of targets such as small organic Materials and Reagents. Streptavidin-modified magnetic molecules, proteins, metal ions, and even entire cells with high μ −1 ffi 27 beads (SA-MBs; 1 m, 10 mg mL ) were purchased from a nity and selectivity. Compared to antibodies, these fi ff Beaverbio Co., Ltd. (Suzhou, China). Streptavidin-modi ed chemically synthesized molecules o er distinct advantages gold nanoparticles (SA-Au NPs; 15 nm, 0.4 mg mL−1) were such as smaller sizes, lower cost, better stability over a wide obtained from Biosynthesis Biotech Co., Ltd. (Beijing, China). range of temperature, solvents, and pH, and a lack of All oligonucleotides used in this study were HPLC-purified and immunogenicity, all of which make aptamers an ideal synthesized by Sangon Biotechnology Co., Ltd. (Shanghai, alternative to protein antibodies.28 Accordingly, several fi 29 China). Their sequences are listed as follows: FAM-modi ed aptamer-based colorimetric biosensors, electrochemical bio- aptamer (Apt-FAM): 5′-biotion-TTT TTG CAG TTG ATC 30 31 sensor and fluorescent biosensor, have been reported for CTT TGG ATA CCC TGG-FAM-3′; Complementary to Apt- the detection of cancer cells. Liu and co-workers demonstrated FAM (c-Apt): 5′-biotion-TTT TTC CAG GGT ATC CA-3′. an approach for fluorescent imaging and mass quantification of The buffers involved in this work are as follows: binding mIgM in live cells by confocal fluorescence microscope and buffer solution contains 10 mM Tris-HCl, 1 mM EDTA, 1 M ICP-MS via a single probe composed of an aptamer and a silver NaCl, and 0.05% Tween-20 (v/v) (pH 7.4); hybridization cluster.32 Unfortunately, the practical application potential of buffer solution contains 10 mM Tris-HCl, 1 mM EDTA, and

2356 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361 Analytical Chemistry Article

Figure 2. TEM images of the as-prepared dual-functional probes (a) and detail with enlarged scale of bare MBs (b) and Apt-FAM-modified MBs (c) after they reacted with c-Apt-modified Au NPs, respectively.

150 mM NaCl (pH 7.4); washing buffer solution is a mixture of Fluorescence Imaging and ICP-MS Determination of 10 mM Tris-HCl, 1 mM EDTA, and 0.1% Tween-20 (v/v; pH MCF-7 Cells. For fluorescence imaging, 7 μL of the above 7.4); phosphate-buffered saline (PBS) is a mixture of 137 mM prepared MB-Apt-FAM-Au NP conjugates were added to μ NaCl, 2.7 mM KCl, 1.9 mM KH2PO4, and 8.1 mM Na2HPO4 MCF-7 cell suspension in 200 L of PBS. Then, the mixture (pH 7.4). All reagents were of analytical grade at least and used was incubated at 37 °C for 180 min with gentle shaking. During without further purification. Milli-Q ultrapure water (18.2 MΩ· the incubation, the target MCF-7 cells would bind with the cm, Millipore, France) was used throughout this work. aptamer modified on MBs specifically, and release Au NPs at Cell Culture. The target MCF-7 cells (human breast same time. The MB-bound MCF-7 cells can be collected with a carcinoma cell line), the control HepG2 cells (human magnet, and the fluorescence image of MCF-7 cells was hepatocellular carcinoma cell line) and SCC-7 cells (squamous observed under blue light excitation. For ICP-MS determi- cell carcinoma cell line) used in this study were purchased from nation, 100 μL of supernatant was collected after magnetic American Type Culture Collection (ATCC, U.S.A.). MCF-7, separation, and followed by introducing into ICP-MS for HepG2, and SCC-7 cells were maintained in Dulbecco’s monitoring the signal of 197Au. As a result, the number of target modified Eagle’s medium (DMEM, Gibco, U.S.A.) supple- cells can be obtained according to the signal intensity of 197Au mented with 10% (v/v) fetal bovine serum (FBS, PAN, indirectly. All of the experiments were performed in triplicate Germany), 100 U mL−1 penicillin, and 100 U mL−1 unless otherwise stated. streptomycin. Cells were cultured at 37 °C in an atmosphere Blood Sample Analysis. Fresh human whole blood of fi of humidi ed 5% CO2. Cells were detached by trypsinization healthy people was obtained from the Zhongnan Hospital of with 0.25% trypsin and 0.02% EDTA in PBS buffer. Wuhan University (Wuhan, China) according to the rules of Preparation of MB-Apt-FAM-Au NP Conjugate. The the local ethical committee. Spiking experiments were design of the dual-functional probe is shown in Figure 1. First, conducted to evaluate the application potential of the proposed Apt-FAM was immobilized on the surface of SA-MBs via a method to real biological samples. A certain number of target specific binding between streptavidin and biotin. Briefly, 100 μL MCF-7 cells in different levels were spiked into lysed human of 10 mg mL−1 SA-MBs were washed three times with 200 μL blood and then subjected to the proposed method followed by of washing buffer solution and then redispersed in 300 μLof ICP-MS determination. binding buffer solution. Subsequently, 200 μLof1μM biotinylated Apt-FAM was added into the SA-MBs dispersion, ■ RESULTS AND DISCUSSION and then the mixture was incubated at 37 °C for 60 min with Design of the Dual-Functional Probe. The detailed gentle shaking to immobilize the Apt-FAM on the surface of principle of this experiment is illustrated in Figure 1. The dual- SA-MBs. After washing with 200 μL of washing buffer solution functional probe is composed of a MB moiety for separation for three times to remove excess aptamers, the Apt-FAM- and collection of target cells, an aptamer unit to recognize modified MBs were resuspended in 500 μL of hybridization target cells via the overexpression of MUC1 on the surface of buffer solution for the following use. In another vial, 100 μLof MCF-7 cells, a fluorescent dye moiety for fluorescence imaging − 5 μM biotinylated c-Apt was mixed with 25 μL of 0.4 mg mL 1 as well as an Au NP tag. The ingenious design of this probe is SA-Au NPs in 400 μL binding buffer solution, and then the the utilization of Au NPs as a quencher and reporter, which mixture was incubated at 37 °C for 60 min with gentle shaking provides both an “off-on” fluorescence signal via fluorescence to immobilize the c-Apt on the surface of SA-Au NPs via resonance energy transfer (FRET) and an elemental tag for biotin−streptavidin specific interaction. The as-prepared c-Apt- ICP-MS determination. The excellent performance of Au NPs modified Au NPs were separated by centrifugation at 12000 g serve as elemental tags in ICP-MS-based quantitative for 15 min and washed with 200 μL of washing buffer solution bioanalysis has been widely confirmed, which may attribute for three times, followed by redispersing in 500 μLof to their sensitive response and low spectral interference in ICP- hybridization buffer solution. Accordingly, 500 μL of the as- MS, as well as low background in biological samples. prepared Apt-FAM-modified MBs were added to the c-Apt- Furthermore, Au NPs can also act as fluorescent quenchers modified Au NPs redispersing solution, and the mixture was in this probe to form an “off-on” system for fluorescence incubated at 37 °C for 60 min to form a MB-Apt-FAM-Au NP imaging, since Au NPs have a wide absorption band and strong conjugate through complementary base pairing. The product fluorescence quenching ability. was separated and washed with 200 μL of washing buffer In the absence of the target cells, the Apt-FAM-modified solution for five times under an external magnetic field. Finally, MBs can bind with the c-Apt-modified Au NPs to form stable the product was redispersed in 200 μL of PBS for further use. MB-Apt-FAM-Au NP conjugates via complementary base

2357 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361 Analytical Chemistry Article pairing. In this case, the fluorescence of FAM is quenched due Au increased along with the increase in density of Apt-FAM to FRET from the FAM to the Au NPs quencher. Whereas when the ratio of MBs to Apt-FAM was lower than 1/200 g when the target cells are added and come into contact with the nmol−1, and the signal intensity of Au started to decrease probe, the target cells would compete with c-Apt-modified Au thereafter when the ratio of MBs to Apt-FAM was higher than − − NPs for binding to the Apt-FAM immobilized on the MBs. It 1/200 g nmol 1. Therefore, a ratio of 1/200 g nmol 1 (MBs/ turns out that the c-Apt-modified Au NPs are released from the Apt-FAM) was employed in the subsequent experiments. conjugates due to a stronger binding force between the target In addition, the amount of MB-Apt-FAM-Au NP conjugates cells and the aptamers. After magnetic separation, the released was one of the key factors in counting and imaging of target Au NPs can be easily separated from the MB-bound target cells cells. It was essential to add excessive of MB-Apt-FAM-Au NP and excess unreacted MB-Apt-FAM-Au NP conjugates. The conjugates to ensure a constant proportion of released Au NPs amount of released Au NPs in the supernatant is in direct to the number of target cells. However, too many MB-Apt- proportion to the amount of target cells, and can be accurately FAM-Au NP conjugates involved may cause a high blank signal determined using Au signal in ICP-MS. Meanwhile, since the and low reproducibility, which may attribute to incomplete Au NPs quencher is released, the fluorescence image of target magnetic separation. Hence, the effect of the amount of MB- cells can be observed with the fluorescence recovery of FAM. In Apt-FAM-Au NP conjugates on Au signal was studied from 2.0 this way, it provides a dual-modal sensing platform for rapid, to 40 μg and the results are shown in Figure S2. As can be seen, sensitive, and specific counting and imaging of MCF-7 cells. the signal intensity of Au increased rapidly when the amount of Characterization of the As-Prepared Dual-Functional MB-Apt-FAM-Au NP conjugates increased from 2.0 to 30 μg, Probe. TEM was used to characterize the morphologies of and then remained constant with further increasing the amount MB-Apt-FAM-Au NP conjugates. As the TEM images show in of MB-Apt-FAM-Au NP conjugates from 30 to 40 μg. Figure 2a, MB-Apt-FAM-Au NP conjugates had a uniform size Therefore, 35 μg of MB-Apt-FAM-Au NP conjugates was around 1 μm with good dispersity. Furthermore, no Au NPs employed in this method. were observed on the surface of the bare MBs (Figure 2b) after Furthermore, the effect of incubation time for the MB-Apt- they reacted with c-Apt-modified Au NPs, whereas several Au FAM-Au NP conjugates and target MCF-7 cells on Au signal in NPs (judging from the size difference, the diameter of Au NPs the range of 20−180 min was also investigated to ensure a was about 15 nm) were clearly observed on the surface of the complete reaction. As the results presented in Figure S3, the Apt-FAM-modified MBs (Figure 2c). In addition, the precise signal intensity of Au increased along with the increase in amount of Au NPs anchored on MBs was further determined incubation time and kept constant after 120 min, indicating that with ICP-MS to confirm the formation of MB-Apt-FAM-Au the reaction could be completed in 120 min. Therefore, an NP conjugates by means of digestion with HNO3. The results incubation time of 150 min was chosen in this work. of ICP-MS analysis indicated that 14.9 ± 0.5 ng of Au NPs was Specificity and Cross-Reactivity. Since the detection of anchored on 1 μg of Apt-FAM-modified MBs. However, as for target cells may be affected by the coexisting numerous of the original MBs without Apt-FAM modification, approx- normal and/or benign cells in complex biological samples, the imately 30-fold less amount of Au NPs was detected, indicating selectivity and specificity of the proposed method need to be that few c-Apt-modified Au NPs could attach to the surface of verified. For this purpose, the human hepatocellular carcinoma unmodified MBs through nonspecific adsorption. All the above cell line HepG2 and squamous cell carcinoma cell line SCC-7 experimental results clearly indicate the successful formation of were chosen as control. The test was subjected to the same MB-Apt-FAM-Au NP conjugates via complementary base analysis procedures as for the MCF-7 cells. As the results pairing. shown in Figure 3, the signal intensity of Au for HepG2 and Optimization of Experimental Conditions. To achieve SCC-7 cells was much lower than that for MCF-7 cells at the the optimal analytical performance, several experimental factors same cell number of 2 × 103 cells. In addition, the cross- involved in this procedure were examined and optimized, including the density of Apt-FAM immobilized on MBs, amount of the dual-functional probe, and incubation time for the dual-functional probe and target MCF-7 cells. In the proposed method, target MCF-7 cells were recognized by the aptamer immobilized on MBs. Therefore, the density of Apt-FAM immobilized on the surface of MBs directly affects recognition capability of the dual-functional probe to target cells, which would reflect in signal response of Au and may further influence the sensitivity of the method. When the density of Apt-FAM immobilized on MBs was too low, the amount of Au NPs anchored would be insufficient, which could lead to low efficiency of Au NPs release as well as low signal response of Au, whereas when the density of Apt-FAM immobilized on MBs was too high, more than one hybrid double-strands could be formed between each Au NP and MB, fi which may increase the resistance for Au NPs release and lower Figure 3. Speci city and cross-reaction tests for MCF-7 cells. Columns A, B, and C present the relative intensity of 197Au in ICP-MS for signal response of Au accordingly. The density of Apt-FAM detecting 2 × 103 of SCC-7 cells, 2 × 103 of HepG2 cells, and 2 × 103 immobilized on MBs was optimized by adjusting the ratio of ff of MCF-7 cells, respectively; Columns E and F represent the relative MBs to Apt-FAM. The e ect of the ratio of MBs to Apt-FAM intensity of 197Au in ICP-MS for detecting 2 × 103 of MCF-7 cells on Au signal was investigated in the range of 1/10−1/500 g mixed with 2 × 104 of SCC-7 cells and 2 × 104 of HepG2 cells, − nmol 1. As the results show in Figure S1, the signal intensity of respectively.

2358 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361 Analytical Chemistry Article reactivity of the method was also verified by mixing 2 × 103 FRET-based “off-on” fluorescence and high selectivity of the MCF-7 cells with 2 × 104 HepG2 or SCC-7 cells, and almost dual-functional probe, and the developed system can be further identical signal intensity of Au was observed for MCF-7 cells applied to cell fluorescence imaging. and other two mixed groups. All the above results demonstrate Fluorescence Imaging. Despite providing quantitative an excellent specificity and selectivity of the proposed method. results of cell numbers, ICP-MS cannot present visualized Moreover, the selectivity and specificity of the method was images due to its inherent limitation. Living cell imaging can further testified by fluorescence imaging. provide visualized images and locations, as well as some other Fluorescence Response of the Dual-Functional Probe detail information, which can also be a supplement to ICP-MS to MCF-7 Cells. To verify that the “off-on” fluorescence of the results for further excluding false positive/negative results. In dual-functional probe could be triggered by target MCF-7 cells, this work, with the help of MB-Apt-FAM-Au NP conjugates, the fluorescence response of the as-prepared dual-functional fluorescence images of target cells were achieved by inverted probe to MCF-7 cells was then investigated, and the results are fluorescent microscope. Fluorescence images of target and displayed in Figure 4. As expected, the fluorescence intensity of control cells are presented in Figure 5a and b, respectively. As can be seen, the noticeable green fluorescence signal was generated from the surface of target cells, and the fluorescence image was in good agreement with the corresponding bright field image. By contrast, the control group HepG2 cells displayed only negligible fluorescence as expected after incubation with the MB-Apt-FAM-Au NP conjugates, indicat- ing that the FRET-based “off-on” fluorescence of the dual- functional probe was hardly activated by nontarget cells. These results demonstrated the specificity of MB-Apt-FAM-Au NP conjugates for target MCF-7 cells, and were in agreement with the results obtained by ICP-MS determination mentioned above. Furthermore, all these results reveal that the barrier between living cell imaging and destructive detector ICP-MS measurement can be bridged with the aptamer-based dual- functional probe, owing to the excellent fluorescence quenching properties and abundant ICP-MS detectable Au atoms composition of Au NPs. Figure 4. Fluorescence spectra of the dual-functional probes in the Analytical Performance. The analytical performance of presence of different numbers of target MCF-7 cells. the developed method of MB-Apt-FAM-Au NP conjugates based assay with ICP-MS measurement for cell counting was evaluated under the optimized conditions. Figure 6 is the the dual-functional probe increased along with increasing dependence of signal intensity of 197Au on MCF-7 cell number. number of MCF-7 cells from 200 to 10000. In the absence of As can be seen, the 197Au signal response in ICP-MS for varying MCF-7 cells, only a very weak fluorescence intensity of the amounts of MCF-7 cells presented a trend similar to the results dual-functional probe was observed with same procedures. The of fluorescence assay. Moreover, a good linear relationship was above results demonstrate the successful establishment of obtained within the cell number ranging from 2 × 102 to 1.2 ×

Figure 5. Fluorescence (a1 and b1), bright field (a2 and b2), and merged images (a3 and b3) of MCF-7 cells and HepG2 cells incubated with the dual-functional probes, respectively.

2359 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361 Analytical Chemistry Article

advantages of each detection technology by integrating fluorescence imaging detection modalities and ICP-MS measurement with single probe, which makes the method simpler and more reliable. It must be stressed that the analysis time is greatly shortened in the developed method compared to traditional cell analysis techniques due to the reduction of incubation times and avoiding of washing or elution operation involved. As is well-known, analytical time is usually one of the restraining factors against the development of rapid and early diagnosis. The success of this study provides a rapid, robust and reliable approach for the early clinical diagnosis of breast cancer. Blood Sample Analysis. In order to evaluate the applicability of the developed method to real biological samples, a spiking experiment was carried out by adding different numbers of target MCF-7 cells into the lysed human blood, and the analytical results are listed in Table 2. As can be Figure 6. Calibration curve of MCF-7 cells obtained using signal intensity of 197Au in ICP-MS. Table 2. Recovery Test for Whole Blood Sample added (×102 cells) found (×102 cells) recovery (%) 4 a 10 cells, and the linear correlation equation was IAu = 0 N.D. 495.7[cell number] + 322810 with a correlation coefficient (R2) 2.0 2.2 ± 0.2 110 of 0.9944. Based on the signal intensity of 197Au, the LOD (3σ) 5.0 5.4 ± 0.2 108 of the proposed method for the MCF-7 cell was calculated to 10 10.9 ± 1.0 109 be 81 cells, and the relative standard deviation (RSD) was 5.6% 20 19.5 ± 1.7 98 for seven replicate determinations of 8 × 102 MCF-7 cells. A 40 41.9 ± 1.7 105 a comparison of the analytical performance of this work with Not detected. those obtained by several other reported approaches for the detection of cancer cells is shown in Table 1. As can be seen, the LOD of the established method is comparable with most of seen, this method has a robust resistibility to the complex the reported ICP-MS based methods, and much lower than that biological matrix with a satisfactory recovery ranging from 98 to obtained by other approaches. Besides, this work takes the 110%. These results clearly indicate that the proposed method can be applied to detect trace amount of target MCF-7 cells in Table 1. Comparison of Different Methods for the Analysis human blood with a good accuracy. of Tumor Cells ■ CONCLUSION linear range LOD In this paper, we demonstrate a simple, rapid, sensitive, and − − detection method target cell (cell mL 1) (cell mL 1) ref specific method using an aptamer-based dual-functional probe QCM CCRF-CEM 1 × 104 to 8 × 103 34 (MB-Apt-FAM-Au NP conjugates) for ICP-MS counting and × 5 1.5 10 fluorescence imaging of cancer cells. Here, Au NPs can not only colorimetric MCF-7 2.5 × 102 to 125 35 × 3 serve as elemental tags for ICP-MS-based quantitative 8 10 ffi fluorescence BGC-823 5 × 102 to 210 36 bioanalysis, but also serve as e cient quenchers to construct 1 × 107 a FRET-based “off-on” fluorescence probe for living cell fluorescence SMMC-7721 2 × 102 to 200 37 imaging. This simple and low-cost design of the dual-functional 2 × 104 probe can take the advantages of each detection technology and photoluminescence MCF-7 5 × 102 to 201 30 offer more comprehensive and valuable information on cancer × 5 1 10 cells. It is greatly expected to be a powerful platform for the atomic force HepG2 1 × 103 to 300 38 microscope 1 × 105 rapid and accurate clinical diagnosis of cancer at early stage. electrochemical MCF-7 1 × 102 to 100 39 Even more importantly, this approach is not limited to the detection 1 × 106 breast cancer cells studied here, it can also extend to detect electrochemical MCF-7 1 × 105 to 1 × 105 40 various types of cancer cells, ssDNA, even nucleases if an 8 detection 1 × 10 appropriate primer DNA (aptamer) is selected. ICP-MS HT-29 4 × 102 to 44 20 × 5 4 10 ■ ASSOCIATED CONTENT ICP-MS SMMC-7721 2 × 102 to 100 8 8 × 103 *S Supporting Information ICP-MS HepG2 2 × 102 to 61 18 The Supporting Information is available free of charge on the 3 × 104 ACS Publications website at DOI: 10.1021/acs.anal- ICP-MS HepG2 8 × 102 to 282 41 × 4 chem.7b04927. 4 10 ff ICP-MS HepG2 4 × 102 to 100 19 Additional information on e ect of the ratio of MBs to 3 × 104 aptamer DNA (Figure S1), effect of the amount of the ICP-MS MCF-7 2 × 102 to 81 this probe (Figure S2), effect of incubation time for the probe 4 1.2 × 10 work and target MCF-7 cells (Figure S3), and operating

2360 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361 Analytical Chemistry Article

parameters of ICP-MS (Table S1) as noted in text (20) Zhang, Z. B.; Luo, Q.; Yan, X. W.; Li, Z. X.; Luo, Y. C.; Yang, L. Anal. Chem. 84 − (PDF). M.; Zhang, B.; Chen, H. F.; Wang, Q. Q. 2012, , 8946 8951. (21) Yeh, Y. C.; Creran, B.; Rotello, V. M. Nanoscale 2012, 4, 1871− ■ AUTHOR INFORMATION 1880. Corresponding Author (22) Peng, H. Y.; Tang, H.; Jiang, J. H. Sci. China: Chem. 2016, 59, * 783−793. Fax: +86-27-68754067. Tel.: +86-27-68752162. E-mail: Mass Spectrom. Rev. [email protected]. (23) Liu, R.; Wu, P.; Yang, L.; Hou, X.; Lv, Y. 2014, 33, 373−393. ORCID (24) Liu, Z.; Li, X.; Xiao, G.; Chen, B.; He, M.; Hu, B. TrAC, Trends Beibei Chen: 0000-0001-7772-5171 Anal. Chem. 2017, 93,78−101. Anal. Bin Hu: 0000-0003-2171-2202 (25) Li, F.; Zhao, Q.; Wang, C.; Lu, X.; Li, X.-F.; Le, X. C. Chem. 2010, 82, 3399−3403. Notes (26) Meng, H. M.; Liu, H.; Kuai, H. L.; Peng, R. Z.; Mo, L. T.; The authors declare no competing financial interest. Zhang, X. B. Chem. Soc. Rev. 2016, 45, 2583−2602. (27) Tan, W.; Donovan, M. J.; Jiang, J. Chem. Rev. 2013, 113, 2842− ■ ACKNOWLEDGMENTS 2862. fi (28) Zhang, H. M.; Zhou, L. J.; Zhu, Z.; Yang, C. Y. Chem. - Eur. J. This work is nancially supported by the National Nature 2016, 22, 9886−9900. Science Foundation of China (Nos. 21575107, 21575108, (29) Zhang, X.; Xiao, K.; Cheng, L.; Chen, H.; Liu, B.; Zhang, S.; 21375097, 21175102), the Science Fund for Creative Research Kong, J. Anal. Chem. 2014, 86, 5567−5572. Groups of NSFC (No. 20921062), the MOE of China, and the (30) Hua, X.; Zhou, Z.; Yuan, L.; Liu, S. Anal. Chim. Acta 2013, 788, Large-Scale Instrument and Equipment Sharing Foundation of 135−140. Wuhan University (LF20170799). (31) Xie, Q.; Tan, Y.; Guo, Q.; Wang, K.; Yuan, B.; Wan, J.; Zhao, X. Anal. Methods 2014, 6, 6809−6814. ■ REFERENCES (32) Liu, R.; Zhai, J.; Liu, L.; Wang, Y. L.; Wei, Y. T.; Jiang, X. L.; Gao, L.; Zhu, H. R.; Zhao, Y. L.; Chai, Z. F.; Gao, X. Y. Chem. (1) Torre, L. A.; Bray, F.; Siegel, R. L.; Ferlay, J.; Lortet-Tieulent, J.; Commun. 50 − Ca-Cancer J. Clin. 65 − 2014, , 3560 3563. Jemal, A. 2015, ,87 108. (33) Ferreira, C. S.; Matthews, C. S.; Missailidis, S. Tumor Biol. 2006, (2) Chen, W.; Zheng, R.; Baade, P. D.; Zhang, S.; Zeng, H.; Bray, F.; 27, 289−301. Jemal, A.; Yu, X. Q.; He, J. Ca-Cancer J. Clin. 2016, 66, 115−132. EMBO Mol. Med. 7 (34) Pan, Y.; Guo, M.; Nie, Z.; Huang, Y.; Pan, C.; Zeng, K.; Zhang, (3) Joosse, S. A.; Gorges, T. M.; Pantel, K. 2015, , Y.; Yao, S. Biosens. Bioelectron. 2010, 25, 1609−1614. 1−11. TrAC, Trends Anal. Chem. (35) Zhang, L. N.; Deng, H. H.; Lin, F. L.; Xu, X. W.; Weng, S. H.; (4) Ho, K. F.; Gouw, N. E.; Gao, Z. Q. Liu, A. L.; Lin, X. H.; Xia, X. H.; Chen, W. Anal. Chem. 2014, 86, 2015, 64, 173−182. − Am. J. Clin. Pathol. 2711 2718. (5) Dunphy, C. H.; Orton, S. O.; Mantell, J. 2004, (36) Han, E.; Ding, L.; Ju, H. Anal. Chem. 2011, 83, 7006−7012. 122, 865−874. J. (37) Xie, Q.; Tan, Y. Y.; Guo, Q. P.; Wang, K. M.; Yuan, B. Y.; Wan, (6) Guo, Y.; Shu, Y.; Li, A.; Li, B.; Pi, J.; Cai, J.; Cai, H.-h.; Gao, Q. J.; Zhao, X. Y. Anal. Methods 2014, 6, 6809−6814. Mater. Chem. B 2017, 5, 5532−5538. ’ (38) Chen, X. J.; Pan, Y. G.; Liu, H. Q.; Bai, X. J.; Wang, N.; Zhang, (7) Koyanagi, K.; O Day, S. J.; Boasberg, P.; Atkins, M. B.; Wang, H. B. L. Biosens. Bioelectron. 2016, 79, 353−358. J.; Gonzalez, R.; Lewis, K.; Thompson, J. A.; Anderson, C. M.; Lutzky, (39) Lv, S.; Guan, Y.; Wang, D.; Du, Y. Anal. Chim. Acta 2013, 772, J.; Amatruda, T. T.; Hersh, E.; Richards, J.; Weber, J. S.; Hoon, D. S. B. − Clin. Cancer Res. 16 − 26 32. 2010, , 2402 2408. (40) Arya, S. K.; Wang, K. Y.; Wong, C. C.; Rahman, A. R. Biosens. (8) Yang, W.; Xi, Z.; Zeng, X.; Fang, L.; Jiang, W.; Wu, Y.; Xu, L.; Fu, Bioelectron. 2013, 41, 446−451. F. J. Anal. At. Spectrom. 2016, 31, 679−685. Talanta Anal. Chem. 88 − (41) Yang, B.; Zhang, Y.; Chen, B.; He, M.; Hu, B. 2017, (9) Tang, N.; Li, Z.; Yang, L.; Wang, Q. 2016, , 9890 167, 499−505. 9896. (10) Liu, R.; Liu, X.; Tang, Y.; Wu, L.; Hou, X.; Lv, Y. Anal. Chem. 2011, 83, 2330−2336. (11) Ko, J.; Lim, H. B. Anal. Chem. 2014, 86, 4140−4144. (12) Luo, Y. C.; Yan, X. W.; Huang, Y. S.; Wen, R. B.; Li, Z. X.; Yang, L. M.; Yang, C. J.; Wang, Q. Q. Anal. Chem. 2013, 85, 9428−9432. (13) Iglesias Gonzalez, T.; Espina, M.; Sierra, L. M.; Bettmer, J.; Blanco-Gonzalez, E.; Montes-Bayon, M.; Sanz-Medel, A. Anal. Chem. 2014, 86, 11028−11032. (14) Zhang, S.; Liu, R.; Xing, Z.; Zhang, S.; Zhang, X. Chem. Commun. 2016, 52, 14310−14313. (15) Bendall, S. C.; Simonds, E. F.; Qiu, P.; Amir, E.-a. D.; Krutzik, P. O.; Finck, R.; Bruggner, R. V.; Melamed, R.; Trejo, A.; Ornatsky, O. I.; Balderas, R. S.; Plevritis, S. K.; Sachs, K.; Pe’er, D.; Tanner, S. D.; Nolan, G. P. Science 2011, 332, 687−696. (16) Giesen, C.; Waentig, L.; Mairinger, T.; Drescher, D.; Kneipp, J.; Roos, P. H.; Panne, U.; Jakubowski, N. J. Anal. At. Spectrom. 2011, 26, 2160−2165. (17) Zhang, C.; Wu, F. B.; Zhang, Y. Y.; Wang, X.; Zhang, X. R. J. Anal. At. Spectrom. 2001, 16, 1393−1396. (18) Yang, B.; Chen, B.; He, M.; Hu, B. Anal. Chem. 2017, 89, 1879− 1886. (19) Yang, B.; Zhang, Y.; Chen, B.; He, M.; Yin, X.; Wang, H.; Li, X.; Hu, B. Biosens. Bioelectron. 2017, 96,77−83.

2361 DOI: 10.1021/acs.analchem.7b04927 Anal. Chem. 2018, 90, 2355−2361