Clonal deletion of thymocytes by circulating dendritic cells homing to the thymus

Transcription

1 26 Nture Pulishing Group Clonl deletion of thymocytes y circulting dendritic cells homing to the thymus Roerto Bonsio 1, M Lucil Scimone 1, Ptrick Scherli 1, Nir Grie 2, Andrew H Lichtmn 2 & Ulrich H von Andrin 1 Dendritic cell (DC) presenttion of self ntigen to thymocytes is essentil to the estlishment of centrl tolernce. We show here tht circulting DCs were recruited to the thymic medull through three-step dhesion cscde involving P-selectin, interctions of the integrin VLA-4 with its lignd VCAM-1, nd pertussis toxin sensitive chemottrctnt signling. Ovluminspecific OT-II thymocytes were selectively deleted fter intrvenous injection of ntigen-loded exogenous DCs. We documented migrtion of endogenous DCs to the thymus in priotic mice nd fter pinting mouse skin with fluorescein isothiocynte. Antiody to VLA-4 locked the ccumultion of peripherl tissue derived DCs in the thymus nd lso inhiited the deletion of OT-II thymocytes in mice expressing memrne-ound ovlumin in crdic myocytes. These findings identify migrtory route y which peripherl DCs my contriute to centrl tolernce. The rndom rerrngement of T cell receptor genes in the thymus continuously genertes utorective T cells tht must e removed from the lymphocyte pool or e rendered innocuous to mintin selftolernce. Most T cells tht encounter their cognte mjor histocomptiility complex (MHC) peptide complex efore completion of thymic development re eliminted y clonl deletion, key process leding to centrl tolernce 1. The presence of n gonist peptide in the thymus cn lso result in the differentition of CD4 + CD25 + regultory T cells 2 tht impose tolernce in the periphery y dmpening the response of conventionl lymphocytes 3. Centrl tolernce is therefore elieved to e restricted to those ntigens tht gin ccess to the thymic environment nd cn e effectively displyed on the surfce of ntigen-presenting cells; tht is, proteins tht re expressed loclly or tht circulte in the loodstrem. Although some tissue-specific proteins my e expressed in the thymus 4,5, no ctive trnsport of self ntigens from the periphery hs een descried so fr. Dendritic cells (DCs) re very efficient ntigen-presenting cells found in ll peripherl orgns nd lymphoid tissues. In the periphery, they re strtegiclly positioned to smple ntigens from tissues nd crry them to drining lymph nodes for processing nd presenttion to recirculting T cells 6. When exposed to pproprite stimuli such s lipopolyscchride, proinflmmtory cytokines or other indictors of cellulr distress, DCs undergo mturtion, resulting in enhnced ntigen presenttion nd incresed expression of costimutlory molecules 6. The mturing DCs gin ccess to lymph nodes through fferent lymphtic vessels nd induce nive T cells to proliferte nd to differentite into effector cells. Immture DCs re lso thought to crry ntigen to lymph nodes nd to interct with nive T cells, ut without previous mturtion stimulus, those interctions result in ortive ctivtion of the T cells, which cn e eliminted, rendered unresponsive or induced to differentite into regultory T cells 7. Tissue-resident DCs cn rech the circultion nd crry ntigen to distl orgns such s the one mrrow vi the lood 8. Thus, we resoned tht in the stedy stte, the DC network would provide suitle mechnism for smpling tissue-specific self ntigens, otherwise invisile to the differentiting thymocytes, nd crrying them to the thymus vi the loodstrem. Here we provide evidence of multistep dhesion pthwy tht recruited circulting DCs to the thymus; we lso show tht peptide-presenting DCs tht homed to the thymus induced clonl deletion of developing ntigen-specific T cells nd, therefore, might mke previously unpprecited contriution to the estlishment of centrl tolernce. RESULTS Immture DCs migrte to the thymus Immture CD11c hi MHC clss II positive DCs constitute rre ut discrete cell popultion in norml mouse nd humn lood 9 (Supplementry Fig. 1 online). This circulting DC popultion is composed of cells with diverse histories; lthough some could derive from newly generted one mrrow emigrnts, there is lso evidence tht DCs in other tissues cn return to the loodstrem fter cquiring ntigenic mteril t peripherl sites 8,9. To model the trfficking properties of circulting DCs, we did doptive trnsfer experiments in which we enriched DC popultions otined from the spleens of donor mice exposed to the lignd for the receptor tyrosine kinse Flt3 (Flt3L), leled the cells with CFSE (croxyfluorescein 1 The CBR Institute for Biomedicl Reserch nd 2 Immunology Reserch Division nd Vsculr Reserch Division, Deprtment of Pthology, Brighm nd Women s Hospitl, Hrvrd Medicl School, Boston, Msschusetts, 2115, USA. Correspondence should e ddressed to U.H.v.A. (uv@cr.med.hrvrd.edu). Received 1 July; ccepted 1 August; pulished online 3 Septemer 26; corrected fter print 29 Septemer 26; doi:1.138/ni1385 NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 1

2 26 Nture Pulishing Group UEA-1 CMTMR C11 c Events Input Thymus Homed cells CD11c CD11c CMTMR C11 77 % 92 % UEA-1 CD11c CFSE CMTMR CMTMR succinimidyl diester) nd dministered them intrvenously to syngeneic recipient mice. Consistent with pulished oservtions 8,we recovered undnt CFSE + CD11c + DCs from spleen, one mrrow, liver nd lung 18 h fter injection. In ddition, we noted tht some of the trnsferred DCs were lso recruited to the thymus (Fig. 1). Compred with the trnsferred input, which contined 2 3% CD11c contminting cells (Fig. 1, left), the popultion of CFSE + cells in the thymus ws enriched for CD11c + cells (Fig. 1, right), in contrst to ll other orgns nlyzed. To determine the locliztion of the trnsferred DCs, we nlyzed thymic cryosections y immunofluorescence t 18 h fter injection. The fluorescence-tgged donor DCs were concentrted in the medull in proximity to the cortico-medullry junction (Fig. 1), in the sme region where endogenous thymic DCs re found 1. In contrst, we found very few homed cells in the cortex. Costining with ntiody to CD11c (nti-cd11c) confirmed tht most homed donor cells were DCs nd loclized together with the ulk of the endogenous thymic CD11c + cell popultion (Fig. 1c). Figure 2 Phenotype of DCs recruited to the thymus. () Stining of thymi nd spleens for B22, CD8, CD11 nd CD11c t 2 h or 24 h fter DC popultions expnded in vivo with Flt3L were leled with CFSE nd dministered intrvenously to wild-type mice (n ¼ 2 mice per group). Dt re presented s percent cells elonging to prticulr suset in the input (gry rs) or gted on the homed CFSE + CD11c + popultion in spleen (filled rs) nd thymus (open rs). () Homing, ssessed y flow cytometry for the presence of CD11c + TRITC + nd CD11c + CFSE + cells. Freshly purified immture DCs (idc) or DCs stimulted overnight with 1 mg/ml of lipopolyscchride (mdc) were differentilly leled with CFSE or TRITC nd were injected intrvenously into wild-type mice (n ¼ 4 or more mice). Spleens nd thymi were collected t vrious times (ove grphs) nd single-cell suspensions were nlyzed. Dt re presented s totl numer of homed CD11c + cells per cells injected., P o.5, nd, P o.1, compred with freshly purified DCs. Dt re pooled from two () or three () experiments (error rs, s.e.m.). CD11c C11 C11 Figure 1 Blood-orne DCs re recruited to the thymus nd loclize to the medull. () Flow cytometry of CD11c expression. After splenic DC popultions were expnded y Flt3L, cells were prtilly purified, leled with CFSE nd injected intrvenously into wild-type mice; 18 h lter, thymi were collected nd single-cell suspensions were nlyzed. Left, CD11c expression in the input popultion; middle, thymocyte suspension showing homed CFSE + cells; right, CD11c expression on gted CFSE + cells. Numers ove rcketed lines indicte the frequency of CD11c + events. Dt re representtive of more thn ten individul experiments. (,c) Confocl microscopy of thymus sections (overly (fr left) nd singlechnnel imges). Mice were injected intrvenously with CMTMR-leled DCs; 18 h lter, thymi were collected nd cryosections 2 mm in thickness were fixed nd were stined with lectin delineting the medull (UEA-1) or with ntiodies to mrkers for the cortex (C-11) or DCs (CD11c). Yellow rrowheds indicte colocliztion of CD11c nd CMTMR signls. Scle rs, 1 mm. Dt re representtive of two experiments. Splenic DCs include t lest three distinct susets: CD8 CD11 + myeloid DCs, CD8 + CD11 DCs (formerly known s lymphoid DCs) nd B22 + CD11c lo plsmcytoid DCs 11. In spleens of mice exposed to Flt3L, we lso found CD8 CD11 popultion tht hs een identified efore in Peyer s ptches nd lymph nodes 12,13. All DCs respond to dnger signls (cteril products, inflmmtory cytokines nd others) y upregulting costimultory molecules nd ntigen-presenttion mchinery nd thus ecome competent to prime nive T cells 6. Mturtion lso influences the expression of chemokine receptors nd dhesion molecules nd, therefore, the trfficking of DCs 14.Thus,wetested the influence of suset memership nd mturtion stte on the migrtion of DCs to the thymus. Comprison of the suset distriution of DCs in the purified input popultion to the frequency of ech suset fter short-term (2-hour) homing showed tht ll DC popultions nlyzed hd similr ccess to the thymus (Fig. 2, left). Only CD8 + DCs were slightly underrepresented in spleen nd thymus compred with the input, ut this Homed CD11c + (per 1 6 injected) Suset frequency (% of CFSE + CD11c + ) 2, 15, 1, 5, Input Spleen Thymus 75 2 h h CD8α + CD11 + CD8α CD11 Spleen Thymus 2 h 24 h 1, 2, B22 + CD11c lo CD8α + 15, 1, 5, CD11 + CD8α CD11 B22 + CD11c lo 1, idc mdc idc mdc idc mdc idc mdc 2 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY

3 26 Nture Pulishing Group c d e ws most likely due to their propensity to ecome trpped in the lungs fter intrvenous injection 8. Suset frequencies remined essentilly unltered fter 24 h (Fig. 2, right), indicting tht ll DCs were retined nd survived eqully well regrdless of their surfce phenotype. In contrst, mturtion y overnight exposure to lipopolyscchride from Escherichi coli reduced the ility of DCs to home to the thymus in short term (2-hour) nd long-term (24-hour) doptive trnsfer experiments y 55% nd 65%, respectively (Fig. 2). This preferentil recruitment of immture DCs ws unique to the thymus, s we found no difference in the numer of homed immture nd mture cells in the spleen (Fig. 2), one mrrow or other lymphoid orgns (dt not shown). We concluded tht ll sutypes of loodorne immture ut not mture DCs cn efficiently gin ccess to the thymus, where they loclize minly in the medull. Moleculr mechnism of DC recruitment Homing of circulting leukocytes to most tissues is medited y multistep dhesion cscdes in which initil selectin-dependent rolling interctions re converted to firm dhesion y chemokine-induced ctivtion of specific integrins 15. To delinete the moleculr components of the multistep cscde responsile for the homing of DCs to the thymus, we did doptive trnsfers in the presence or sence of locking monoclonl ntiodies (mas) to pnel of cndidte trffic molecules. Of the three selectins typiclly involved in rolling interctions 15, only the sence of endothelil P-selectin significntly reduced homing, s demonstrted y ntiody inhiition (Fig. 3; 47% of untreted control) nd y the use of P-selectin-deficient recipients (Fig. 3; 62% of wild-type). However, neither E-selectin nor L-selectin ws involved. Comined lockde of P-selectin nd E-selectin did not result in dditionl inhiition compred with the effect of nti-pselectin lone (dt not shown), ruling out the possiility of n uxiliry function for E-selectin in this setting. We recovered more DCs from the spleen (Fig. 3,) nd peripherl lood (dt not shown) in the sence of P-selectin, indicting tht this pthwy is required for DC trffic to some ut not ll orgns. Leukocyte integrins tht cn medite intrvsculr firm rrest elong to either the 2 or the 4 sufmily 15. Adoptively trnsferred NT L P E NT L P E WT KO WT KO NT α 4 β 2 NT α 4 β 2 WT KO WT KO NT ICAM-1 MAdCAM-1 VCAM-1 NT ICAM-1 MAdCAM-1 VCAM-1 Figure 3 P-selectin nd VLA-4 VCAM-1 interctions medite the recruitment of DCs to the thymus. () Homing of DCs whose popultions were expnded in vivo with Flt3L, then were leled with CFSE nd were injected intrvenously into wild-type mice (n ¼ 6 or more mice per group) either lone (untreted (NT)) or in comintion with 1 mg of ma to L-selectin (L), P-selectin (P) or E-selectin (E); 18 h lter, spleens nd thymi were collected nd single-cell suspensions were nlyzed y flow cytometry for the presence of CD11c + CFSE + cells. Dt re presented s the percentge of homed DCs in untreted recipients. () Homing of CFSE-leled DCs injected intrvenously into wild-type (WT) or P-selectin-deficient (KO) mice (n ¼ 6 mice per group); spleens nd thymi were nlyzed 18 h lter s descried in. (c) Homing of DCs left untreted (NT) or pretreted with 5 mg/ml of locking ma to 4 or 2 integrin, then differentilly leled with CFSE or TRITC nd injected intrvenously into wild-type mice (n ¼ 6 mice per group); spleens nd thymi were nlyzed 2 h lter s descried in. (d) Homing of CFSE-leled DCs injected intrvenously into wild-type mice (n ¼ 4 mice per group) either lone (NT) or in comintion with 1 mg ma to ICAM-1, MAdCAM-1 or VCAM-1; spleens nd thymi were nlyzed 18 h lter s descried in. (e) HomingofCFSEleled DCs injected intrvenously into wild-type (WT) or VCAM-1-knockout (KO) mice (n ¼ 5 mice per group); spleens nd thymi were nlyzed 18 h lter s descried in. Dshed horizontl lines indicte the men homing of untreted DCs in control mice (NT or WT; error rs, s.e.m.)., P o.5, compred with untreted () or wild-type ();, P o.1, compred with wild-type (,e) or untreted(c,d). Dt re pooled from three (,) or two(c e) experiments. DCs pretreted with neutrlizing ma to 2 reched the thymus s efficiently s untreted control cells, wheres lockde of 4 integrins resulted in inhiition of out 8% (Fig. 3c). Homing of DCs to the thymus ws lso reduced y lockde of VCAM-1 (the min endothelil lignd for the integrin VLA-4 ( 4 1 )), ut not y lockde of MAdCAM-1 ( lignd for the integrin 4 7 ), ICAM-1 ( lignd for the integrin L 2 (lso clled LFA-1)) or the integrin M 2 (lso clled Mc-1; Fig. 3d,e). Thus, the only integrin criticl for homing of DCs to the thymus is VLA-4. Consistent with tht conclusion, DCs purified from LFA-1-deficient or integrin 7 deficient donors homed to the thymus s efficiently s wild-type DCs (dt not shown). By nlogy with the moleculr dhesion cscde for homing of DCs to the one mrrow identified y intrvitl microscopy 8, it seems very likely tht VLA-4 VCAM-1 interctions re responsile for firm dhesion, lthough VLA-4 cn lso medite rolling in vitro 16 nd in vivo 17. G protein coupled receptors commonly provide the signl for integrin ctivtion nd consequent rrest of rolling leukocytes NT PTX 2 1 Spleen Thymus NT PTX Spleen Thymus Figure 4 G i -medited signling is required for homing of DCs to the thymus. DC popultions were expnded in vivo with Flt3L nd cells were left untreted (NT) or were incuted for 1 h with 1 ng/ml of PTX, were differentilly leled with CFSE or TRITC nd then were injected intrvenously into wild-type mice (n ¼ 4); 18 h lter, spleens nd thymi were collected nd single-cell suspensions were nlyzed y flow cytometry for the presence of CD11c + CFSE + nd CD11c + TRITC + cells. Dshed horizontl lines indicte the men homing of untreted DCs in control mice (NT; error rs, s.e.m.). Dt re presented s percentge of homed untreted DCs., P o.1, compred with untreted. Dt re pooled from two experiments. NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 3

4 26 Nture Pulishing Group Events d Dy 17 CD45.2 CD4 SP OT-II (% of control) 9.7 % Annexin V CD4 DC DC-OVA OVAp 14.7 % 3.6 % 3 % 13.1 % 11.6 % 1.4 % Annexin V + (%) These receptors recognize specific chemottrctnts (such s chemokines) on the lumenl spect of the endothelium nd signl vi memers of the G i fmily, which re inctivted y Bordetell pertussis toxin (PTX). Indeed, homing of DCs to the thymus ut not the spleen ws sustntilly inhiited y pretretment with PTX (Fig. 4), indicting involvement of one or more G i -dependent chemottrctnt receptors in this pthwy. PTX tretment did not ffect cell viility, s ssessed y nnexin V stining of PTXtreted DCs fter in vivo homing or 18 h of culture in vitro (dt not shown). These results collectively indicted tht immture DCs home to the thymus through multistep dhesion cscde, using P-selectin for rolling, VLA-4 VCAM-1 for firm rrest, nd one (or more) s-yet-unidentified chemottrctnt(s) signling through G i -coupled receptors. Thymus-tropic DCs induce clonl deletion Hving identified nd chrcterized mechnism for the recruitment of circulting DCs to the thymus, we ddressed the functionl consequences of this migrtory event. Given tht resident thymic DCs contriute to negtive selection 7,1 nd tht immture DCs were preferentilly recruited, we hypothesized tht DCs could trnsport to the thymus ntigens cquired in tolerogenic context (in the sence of mturtion stimuli) nd induce negtive selection y presenting them to developing thymocytes. We designed n experimentl protocol to test tht hypothesis (Supplementry Fig. 2 online). We reconstituted lethlly irrdited CD mice with 1:1 mixture of lymphocyte-depleted one mrrow cell smples from CD wildtype mice nd CD OT-II Rg1 / (recomintion-ctivting gene 1 deficient) mice. These one mrrow chimers generted two cohorts of developing T cells: polyclonl CD popultion, nd CD thymocytes exclusively expressing the OT-II T cell receptor, DC CD8 CD4 SP OT-II (% of control) DC-OVA OVAp OT-II WT DC DC + ma DC-OVA DC-OVA + ma OVAp OVAp + ma e c 25 WT WT-OVA I-Aβ KO-OVA OVAp WT OT-II Figure 5 Circulting DCs recruited to the thymus induce poptosis nd clonl deletion of ntigen-specific thymocytes. () Flow cytometry of CD45.2, CD4 nd CD8 expression in single-cell suspensions of thymi from one mrrow chimeric mice (Supplementry Fig. 2), nlyzed on dy 17 fter one mrrow trnsplnttion. Numers ove oxed res indicte percent CD4 SP cells in CD (OT-II) nd CD45.2 (WT) gtes. DC, untreted DCs; DC-OVA, DCs loded with 2 mm OVA( ); OVAp, solule OVA( ) (5 nmol, intrvenously). () Thymi of one mrrow chimers (n ¼ 8 mice per group) nlyzed s descried in. +ma, cells pretreted with 5 mg/ml of nondepleting ma to integrin 4 nd mice treted with 1 mg ma to VCAM-1., P o.1, DC-OVA versus DC- OVA+mA;, P o.1, versus DC. (c) Thymi of one mrrow chimers (n ¼ 7 mice per group), nlyzed nd presented s in. WT-OVA, wild-type DCs loded with 1 mg/ml of OVA; I-A KO-OVA, I-A-knockout DCs loded with OVA., P o.5, versus WT;, P o.1, WT-OVA versus KO-OVA;, P o.1, versus WT. Dt in,c re presented s the frequency of CD4 SP cells mong ll OT-II (CD ) cells in treted mice reltive to the verge frequency of CD4 SP OT-II cells in thymi of control mice tht received only unpulsed DCs. (d,e) Representtive nnexin V stining (d) of OT-II CD4 SP cells in one mrrow chimer treted with DCs lone (dotted line), OVA( )-loded DCs (filled gry histogrm) or free OVA( ) (lck line); nd quntifiction of nnexin V stining (e) in thymi of one mrrow chimers (n ¼ 3 mice per group) for OT-II CD4 SP cells (OT-II) or wild-type CD4 SP cells (WT)., P o.1, versus OT-II DC. Dt re representtive of five () or two(d) experiments or re pooled from three () or two(c,e) experiments. which recognizes peptide of mino cids from chicken ovlumin (OVA( )) in the context of I-A (ref. 18). The time course of our protocol ws dictted y the fct tht intrvenously injected DCs redily ccess the spleen nd other tissues, where they cn ctivte ntigen-specific mture T cells, resulting in the production of inflmmtory cytokines nd steroid hormones tht cuse nonspecific deth of thymocytes 19. To void those confounding side effects, we doptively trnsferred peptide-loded DCs into chimeric recipient mice 2 weeks fter reconstitution; t tht time, intrthymic T cell development ws lredy well under wy (Fig. 5) ut the numer of mture OT-II cells in the periphery ws still negligile (dt not shown). Bone mrrow chimers received intrvenous oluses of DCs untreted or loded with OVA( ) or, s positive control, 5 nmol of solule peptide. Then, 2 d fter the lst injection, we killed the mice nd quntified the frequency nd differentition stte of the polyclonl nd OT-II thymocytes y flow cytometry. Notly, DCs loded with OVA( ) efficiently eliminted CD4 + single-positive (CD4 SP) OT-II thymocytes (Fig. 5). We lso found significnt clonl deletion of OT-II cells in mice tht received loded DCs per injection (P o.5; Supplementry Fig. 3 online). Although tht result ws consistent with the hypothesis tht homed DCs my directly eliminte ntigen-specific T cells in the thymus, our OT-II cells/thymus DC CD4 + CD25 + CD4 + CD25 WT cells/thymus DC-OVA OVAp DC DC-OVA OVAp Figure 6 Presenttion of gonist peptide y lood-orne DCs does not induce intrthymic differentition of OT-II regultory T cells. CD4 SP thymocytes from one mrrow chimer (n ¼ 8 mice per group; Supplementry Fig. 2) were nlyzed for CD25 expression y flow cytometry nd solute numers of CD4 + CD25 nd CD4 + CD25 + regultory T cells were clculted for OT-II popultions () nd wild-type popultions ()., P o.1, nd, P o.1, compred with DCs. Dt re pooled from three experiments (error rs, s.e.m.). 4 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY

5 26 Nture Pulishing Group c CD11c d FITC + DCs (%) Spleen % 8.22% MHC II CD45.2 Spleen Thymus SSC 3.86% 1.36% Prtner-derived cells (%) Control DCs Totl leukocytes Spleen Thymus Time fter priosis (d) Time fter priosis (d) % No A Anti-α 4.57 %.6 % findings could lso hve een explined y the presence of free peptide in the inoculum or y relese of peptide or memrne les from OVA( )-loded DCs, which could hve een cquired y other ntigen-presenting cells in the thymus. To rule out those possiilities, we first sought to determine whether homing of DCs to the thymus ws necessry to induce clonl deletion. The simultneous lockde of 4 integrins on DCs nd of VCAM-1 on the recipients endothelium not only locked migrtion of DCs to the thymus ut lso olished the clonl deletion induced y OVA( )-loded DCs, wheres solule OVA( ) ws eqully effective in the presence or sence of nti-vcam-1 (Fig. 5). Next, we sought to determine whether the homed DCs were directly responsile for ntigen presenttion or cted s ntigen crriers. We incuted wild-type or I-A-knockout DCs with whole OVA nd trnsferred them to one mrrow chimers. Only wild-type DCs nd not I-A-knockout DCs induced deletion of OT-II thymocytes (Fig. 5c), showing tht ntigen presenttion is directly crried out y the doptively trnsferred DCs. Notly, the differentition of wild-type thymocytes proceeded undistured in ll mice (dt not shown), demonstrting tht the deletion of OT-II cells ws ntigen specific. The removl of thymocytes in the process of negtive selection is typiclly ssocited with the induction of poptosis 1, which cn e monitored y the inding of nnexin V to the memrne of dying or FITC + non-dcs (%) FITC No A Anti-α 4. Thymus Spleen Thymus Figure 7 Migrtion of endogenous DCs to the thymus. () Representtive flow cytometry of DCs in the spleen nd thymus of priotic pirs t 5 d (gry lines) or 7 d (lck lines) fter priosis. Filled histogrms, ckground stining in control mice. Numers ove rcketed lines indicte percentge of prtner-derived DCs t 5 d (gry) nd 7 d (lck). Histogrms re gted on CD11c hi MHC-II + events. Dt re representtive of two independent experiments. () Percentge of prtner-derived DCs nd totl leukocytes in priotic mice (n ¼ 4 mice per time point). (c) Detection of DCs in thymi of FITC-pinted mice. Fr left, gting strtegy; remining plots, FITC stining on DCs from thymi of wild-type mice 48 h fter epicutneous ppliction of FITC. Numers in ottom right corners indicte percent FITC + cells. Dt re representtive of four independent experiments. MHC II, MHC clss II; SSC, side sctter; Control, mice not pinted with FITC; No A, PBS lone; Anti- 4, locking ma to 4. (d) Percentge of FITC + DCs (left) or non-dcs (right) in spleen nd thymus fter dministrtion of FITC nd simultneous injection of 2 mg/mouse of locking ma to 4 or PBS. n ¼ 11 mice per group;, P o.5. Dt in,d re pooled from two () or four(d) experiments (error rs, s.e.m.). ded cells. To ddress whether the disppernce of OT-II cells ws due to poptosis, we stined thymocytes of one mrrow chimeric recipients of DC or peptide injections with nnexin V nd compred the stining profiles of CD (OT-II) nd CD45.2 (wild-type) cells. In mice treted with OVA( )-loded DCs, significntly lrger frction of CD4 SP OT-II thymocytes ws poptotic thn in mice treted with unpulsed DCs (P o.1), wheres the percentge of nnexin V positive wild-type cells remined the sme (Fig. 5d,e). These dt demonstrted tht circulting DCs tht home to the thymus present peptides to thymocytes, cusing their removl y poptosis-medited clonl deletion. CD4 + CD25 + OT-II thymocytes resist negtive selection CD4 + CD25 + regultory T cells re criticl in preventing utoimmunity nd keeping in check immune responses 3,utthemechnism underlying their genertion in the thymus is controversil. Although some experiments hve suggested tht gonist peptides in the thymus cn instruct the differentition of regultory T cells 2, other models hve een proposed 2. To ddress tht issue, we nlyzed the expression of CD25 on wild-type (CD45.2 ) nd OT-II (CD ) popultions of CD4 SP thymocytes fter DC injection. Injection of OVA( )-loded DCs or free peptide resulted in mssive deletion of CD4 + CD25 OT-II thymocytes (y 67% nd 9%, respectively; Fig. 6), wheres we detected no increse in the solute numer of CD4 + CD25 + OT-II regultory T cells. However, the selective removl of CD25 OT-II cells resulted in higher frequency of CD25 + OT-II cells in the surviving popultion (from 18% of CD CD4 SP cells in control one mrrow chimers to 35% nd 5% in mice treted with OVA( )-loded DCs nd free peptide, respectively). Figure 8 Clonl deletion of OT-II thymocytes in CMy-mOVA mice depends on 4 integrin function. () PercentV 5 + cells in CMy-mOVA mice nd littermte control mice (Control) irrdited nd reconstituted s descried in Supplementry Fig. 2; 18 d lter, recipient thymi were collected nd stined for CD45.2, CD4, CD8 nd V 5. Results re presented s the percentge of V 5 + cells in the CD CD4 + CD8 (CD45.2) or CD45.2 CD4 + CD8 (CD45.1) popultion; n ¼ 8 mice per group;, P o.5. () Percent V 5 + cells in mice fter intrperitonel dministrtion of 12 mg ma to V β 5 + cells (% of CD4 SP) CD45.2 Control CMy-mOVA integrin 4 on dys 1, 13 nd 16 fter reconstitution of CMy-mOVA mice nd littermte control mice (PBS lone ws injected s control); on dy 18 (48 h fter the lst injection), mice were killed nd nlyzed s descried in. Dt re pooled from two experiments () or were otined in one experiment (; errorrs,s.e.m.;n ¼ 4 mice per group). CD45.1 V β 5 + cells (% of CD4 SP) Control Control + α 4 ma CD45.2 CMy-mOV CMy-mOV + α 4 ma Control Control + α 4 ma CD45.1 CMy-mOV CMy-mOV + α 4 ma NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 5

6 26 Nture Pulishing Group Emphsizing the ntigen specificity of the selection process, the totl numers of CD4 + CD25 nd CD4 + CD25 + wild-type thymocytes remined constnt cross ll experimentl groups (Fig. 6). Endogenous DCs migrte to the thymus DCs re found not only in the lood ut lso in thorcic duct lymph 8,9 (Supplementry Fig. 1), suggesting tht they cn reenter the circultion from the periphery. To investigte whether endogenous circulting DCs could lso rech the thymus, we surgiclly joined CD nd CD mice t the flnks, resulting in the estlishment of shred circultion fter 3 d (priosis). Prtner-derived, fully differentited (CD11c hi MHC clss II positive) DCs migrted to spleen nd thymus s erly s 5 d fter surgery (Fig. 7,). Although the free exchnge of DCs in the spleen lgged ehind the ulk of recirculting leukocytes (Fig. 7, left), this reltionship ws inverted in the thymus, where the frction of prtner-derived DCs ws eight times lrger thn the frction of prtner-derived non-dcs (Fig. 7, right). These results demonstrted tht endogenous, fully differentited DCs hve the ility to migrte to the thymus vi the lood, consistent with pulished studies of priotic mice 21. We then sought to determine whether DCs could trnsport ntigenic mteril from the periphery to the thymus. We pplied fluorescein isothiocynte (FITC), contct sensitizer, to the dorsl skin of wild-type mice. It is known tht fter such tretment, DCs migrte from the skin to the drining lymph nodes 22, ut it is commonly ssumed tht they do not rech ny other orgn. However, t 48 h fter tretment, distinct popultion of FITC + DCs ppered in the thymus (Fig. 7c) nd in the spleen (dt not shown). Notly, mice treted with ma to integrin 4 hd significntly fewer FITC + DCs (P o.5; Fig. 7d, left). This requirement for integrin 4 function ws specific to the thymus, s the numer of FITC + DCs in the spleen did not chnge. Moreover, the frction of non-dcs with right green fluorescent signl ws very smll in the thymus nd ws unffected y tretment with nti- 4 (Fig. 7d, right), suggesting tht skin-resident DCs preferentilly cquired the epicutneously pplied FITC efore trnsporting it to the thymus. Clonl deletion y n extrthymic ntigen To ddress the physiologicl relevnce of the trfficking of DCs to the thymus, we studied the development of OT-II cells in CMy-mOVA trnsgenic mice, in which memrne-ound form of OVA (mova) is expressed exclusively in crdic myocytes (CMy) 23.Wegenerted mixed one mrrow chimers y reconstituting irrdited CMy-mOVA mice (CD ) with CD one mrrow from OT-II Rg1 / donors nd from congenic CD wild-type donors. To detect OT-II thymocytes, we counted CD4 + CD8 CD V 5 + cells 18 d fter reconstitution. We used V 5 s n dditionl OT-II T cell receptor ssocited mrker ecuse despite lethl irrdition, thymi still contined host-derived CD thymocytes, which prohiited the use of CD45.2 lone to identify OT-II cells. The frction of CD4 SP OT-II cells ws significntly smller in the thymi of CMy-mOVA mice thn in littermte control mice (P o.5; Fig. 8). The numer of V 5 + cells ws fourfold higher mong CD cells thn in the CD (non-ot-ii) popultion, ut remined unltered in the ltter whether mova ws expressed in the periphery or not. Thus, OVA-specific thymocytes were selectively deleted in CMy-mOVA hosts. Next we sought to determine how the deleting ntigen cme to e presented in the thymus of CMy-mOVA mice. Although trnsgenic mrna is not detectle y PCR in the thymi of these mice 23, we could not rule out the possiility tht thymic mova mrna ws too scrce to e detectle. Thus, it ws importnt to test whether thymusderived DCs could hve cquired the trnsgenic protein either from n unspecified intrthymic source or y uptke of lood-orne OVA tht might hve een relesed from the hert. Uptke of circulting ntigen y thymic APCs ws most proly instrumentl for the deletion of OT-II thymocytes induced y OVA peptide infusion 24 (Fig. 5). Our results (Fig. 5) showed tht this well-estlished mechnism of clonl deletion y thymus-resident ntigen-presenting cells did not depend on the VLA-4 VCAM-1 pthwy. In contrst, when we treted one mrrow chimeric mice with ma to 4 during the second week fter trnsplnttion, the ma tretment restored the frequency of CD4 SP OT-II cells in CMy-mOVA mice to frequency equivlent to tht in nontrnsgenic control mice (Fig. 8); the frction of V 5 + cells mong CD4 SP CD control cells remined constnt in ll experimentl groups, indicting tht VLA-4 inhiition did not interfere with thymocyte selection itself. Thus, the clonl deletion of OT-II thymocytes in CMy-mOVA mice depended on VLA-4-medited recruitment of cells tht proly cquired the deleting ntigen from crdiomyocytes nd then trnsported it to the thymus. Given our findings descried ove, it seems likely tht t lest some of these peripherl ntigen crriers were hert-derived DCs. DISCUSSION The fct tht DCs set the lnce etween tolernce nd immunity is lredy well estlished. DCs tht rise in the thymus contriute to centrl tolernce y inducing clonl deletion of utorective thymocytes 7,1. Conversely, DCs tht reside in peripherl tissues migrte to secondry lymphoid orgns, where they stimulte T cell responses to pthogens or induce tolernce of utorective T cells tht hve escped thymic selection 6,7.Ourresultsherehvesuggestedtht DCs with extrthymic origin might lso hve role during T cell development. The recruitment pthwy identified here llows tissueresident DCs to collect ntigen in the periphery nd, fter hving returned to the loodstrem, to ccess the thymus. This process my expnd the repertoire of selecting ntigens presented to differentiting thymocytes, llowing peripherl DCs to prticipte in the mintennce of centrl tolernce. How higher orgnisms mintin tolernce towrd the enormous spectrum of peripherl self nd innocuous non-self ntigens hs een the focus of intense study. One mechnism involves ectopic expression of tissue-specific ntigens y medullry thymic epithelil cells 5, which my e directly presented y those cells or my e trnsferred to thymic DCs for cross-presenttion to developing thymocytes 25. However, the scope of ectopic expression seems to e limited in quntity nd qulity: tissue-specific ntigens hve very low expression nd severl re not expressed t ll 5. Indeed, it seems improle tht minute popultion of medullry thymic epithelil cells could fithfully reproduce the overwhelming ntigenic diversity creted y post-trnsltionl modifictions, lterntive mrna splicing nd differentil peptide processing. Moreover, intrthymic gene expression cnnot induce tolernce to the rod rnge of innocuous externl ntigens derived from commensl flor or food. Thymus-tropic DCs tht ptrol the periphery nd then return to the lood might constitute mechnism to prevent misguided T cell responses towrd t lest some of these unpredictle ntigens. Norml lood contins smll numers of circulting differentited DCs 9, which cn cpture lood-orne microorgnisms 26.Thesecells re phenotypiclly distinct from MHC clss II negtive DC precursors 27,28 or CD11c lo B22 + plsmcytoid DCs 29. Although some of the circulting DCs re proly recent emigrnts from the one mrrow, there is mounting evidence tht others hve entered the lood fter 6 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY

7 26 Nture Pulishing Group migrting (or originting) in other tissues 8,9. In fct, DCs re found in thorcic duct lymph, indicting tht there is sutle ut continuous flux of tissue-derived DCs tht re-enter the lood 8,22. Incresed numers of circulting DCs my lso e encountered in clinicl settings; in cncer vccine trils, utologous DCs modified ex vivo hve een infused into tumor ptients 3,31. It is likely tht the infused DCs exert similr iologicl ctivities, presumly t the sme ntomicl loctions s endogenous circulting DCs. In our experiments, we used splenic DCs whose popultions were expnded in vivo to otin sufficient numer of cells tht could e trcked fter doptive trnsfer. Control experiments showed no sustntil differences in the migrtory ctivity of splenic versus lood-derived DCs or DC popultions expnded with Flt3L versus nonexpnded DC popultions (dt not shown). Moreover, the fct tht ll DC susets injected were recruited eqully to the thymus mkes it unlikely tht our findings reflect the nomlous ctivity of nonrepresenttive supopultion of CD11c + cells in donor spleens. Of note, it hs een reported tht thymic ut not splenic DCs home to the thymus fter doptive trnsfer 32. Those findings differ from our oservtions in which splenic DCs were recruited to the thymus with efficiency similr to or etter thn tht of thymic DCs (dt not shown). Tht discrepncy might e due to the different methods used to purify splenic DCs: extensive mnipultion to isolte DCs from the spleen cn result in DC mturtion, which, s we hve shown here, ttenutes migrtion of DCs to the thymus. Our results hve shown tht DCs use clssicl multistep dhesion cscde to home to the thymus through microvessels in the corticomedullry junction: P-selectin is required for optiml homing nd most likely medites tethering nd rolling interctions; G i - medited, PTX-sensitive chemottrctnt signl is solutely required, s is true for the homing of lymphocytes to most lymphoid orgns 15 ; nd, finlly, VLA-4 VCAM-1 interctions re essentil to medite firm rrest nd my lso contriute to rolling. The dhesion molecules prticipting in this cscde re the sme s those tht medite the recruitment of common lymphoid progenitors to the thymus 33. However, the chemottrctnt signl seems to e distinct from tht used y lymphoid progenitors. Those lst cells depend on CCR9- CCL25 to home to the thymus 33, wheres DC homing ws unltered in mice treted with nti-ccl25 or severl other locking mas to thymic chemokines, including CCL2, CCL5, CCL2 nd CXCL12 (dt not shown). It lso seems unlikely tht thymus-expressed CCR7 lignds re involved, s CCR7 is preferentilly expressed on mture DCs, which migrted poorly to the thymus. Wht is the fte nd function of homed DCs in the thymus? Our dt hve indicted tht they cn induce ntigen-specific clonl deletion. DCs must home to the thymus to exert this effect, nd they present the selecting ntigen utonomously, without medition y other ntigen-presenting cells. Tht is consistent with pulished findings showing tht ntigen-loded DCs cn eliminte ntigenspecific thymocytes in vitro 34 nd tht resident thymic DCs induce negtive selection in vivo 7,1. It ws initilly ssumed tht the responsile DCs rise exclusively in the thymus, possily from common lymphoid progenitor 35 ; however, susequent studies with priotic nimls 21 hve suggested the existence of n immigrting DC popultion tht might prticipte in negtive selection nd intrthymic genertion of regultory T cells 36. Studies of the lifespn nd turnover of thymic DCs hve shown tht there re two kineticlly distinct popultions of pproximtely equl size: the first is seprted y only 2 d from its lst proliferting precursor nd therefore is most likely generted loclly; the second is composed of DCs tht hve not divided for t lest 1 week 37 nd whose entry into the thymus is independent of lymphoid progenitors 21. Consistent with those oservtions, we hve found tht prtnerderived DCs in priotic mice quickly ppered in the thymus s erly s 2 d fter shred circultion ws estlished. Tht result suggested tht fully differentited DCs ccess the thymus from the circultion, s lood-orne DC precursors re exceedingly rre nd require out 5 d to cquire full-fledged DC phenotype 28. At dys 5 nd 7 fter priosis, the frction of prtner-derived DCs in the thymus reched out 2%, wheres their frequency mong DCs in the lood ws out 3% (dt not shown). Assuming tht the prtner-derived DCs fithfully report the trfficking ctivity of ll circulting DCs, we cn estimte tht t lest 7% of thymic DCs were lood derived. Thus, if we ssume totl numer of out CD11c hi MHC clss II positive DCs in the thymus of young dult mice 1, t lest would e lood-derived immigrnts. Intrvitl microscopy of T cell DC interctions in lymph nodes hs shown tht single DCs cn contct s mny s T cells per hour 38. Although such mesurements hve not yet een done in the thymic medull, similr contct frequency in this environment would llow the seemingly smll lood-derived frction of thymic DCs to engge thymocytes in over interctions per dy. An importnt issue is the origin of the circulting DCs tht entered the thymus. Using FITC pinting, we were le to identify in the thymus DCs tht hd cquired FITC, presumly in the skin. The ppernce of FITC + DCs ws dependent on 4 integrins, suggesting tht FITC + DCs trfficked to the thymus y undergoing specific dhesive interctions. Bsed on pulished findings 8 nd dditionl dt presented here, we estimte the numer of recirculting DCs tht pss through the thorcic duct to rnge from out to cells per dy. Given tht the thorcic duct conducts only lymph fluid from elow the diphrgm, the totl flux of DCs tht reenter the lood of mouse through efferent lymphtic conduits could rech out cells per dy. Our homing experiments showed tht the frction of doptively trnsferred immture DCs tht could e recovered from the thymus ws out.8%. Tht proly underestimtes the true homing efficiency of circulting DCs, ecuse mny doptively trnsferred cells die soon fter intrvenous injection. Nonetheless, even tht conservtive estimte suggests constnt migrtory strem of t lest 8 DCs per dy tht deprt from peripherl tissues nd ccess the thymic medull in norml mouse. Given the high efficiency of DCs in mking contcts with T cells in lymphoid tissues nd the reltively long time spent y thymocytes in the medull (5 d on verge 5 ), it is possile tht most medullry thymocytes encounter peripherl DCs efore leving the thymus. Consistent with our hypothesis, utorective OT-II thymocytes developing in CMy-mOVA mice were selectively deleted y n 4 integrin dependent process. The most plusile interprettion of tht oservtion is tht crdic phgocytic cells, most proly DCs, cquired mova from crdiomyocytes nd migrted to the thymus through VLA-4 VCAM-1 interctions. Nonetheless, in other experimentl settings, tissue-restricted ntigens expressed in the pncres 39 or the centrl nervous system 4 hve filed to induce efficient negtive selection in the thymus. However, nother study hs descried one mrrow dependent deletion of myelin sic protein specific thymocytes recognizing n epitope of myelin sic protein (mino cids ) not expressed in the thymus 41. Tht rises the possiility tht the contriution of peripherl DCs to centrl tolernce might depend on the tissue origin nd/or the nture of the selecting ntigen. Further studies re needed to explore tht hypothesis. Finlly, the finding tht mture DCs hve sustntilly reduced cpcity to home to the thymus suggests mechnism to sfegurd NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 7

8 26 Nture Pulishing Group ginst indvertent deletion of T cells tht respond to pthogenssocited ntigens. DCs mture when they encounter dnger signls, prticulrly in the context of tissue dmge nd microil infections 6. The loss of thymic tropism should llow mture DCs in the circultion to void sequestrtion in the thymus nd thus to ccumulte more efficiently t other sites, such s the one mrrow nd spleen, where their crgo of pthogen-derived ntigens cn elicit productive immune responses y fully differentited T cells 8,9. In summry, we hve shown here tht peripherl lood-orne DCs cn crry ntigens to the thymic medull, where they re recruited from the circultion through specific multistep dhesion cscde involving P-selectin nd G protein coupled stimulus tht presumly triggers the high-vidity inding of VLA-4 to VCAM-1. This previously unpprecited migrtory pthwy cn result in ntigen-specific clonl deletion of thymocytes nd suggests possile function for peripherl DCs in estlishing nd mintining centrl tolernce. METHODS Mice. C57BL/6 (CD45.2 Ly5.2) mice were purchsed from The Jckson Lortory, Tconic or Chrles River Lortories. Congenic CD45.1 Ly5.1 (B6.SJL-Ptprc Pep3 /BoyJ) nd P-selectin-deficient mice on C57BL/6 ckground 42 (B6.129S7-Selp tm1by /J) were otined from The Jckson Lortory. I-A-knockout (B6.129-H2-A1 tm1gru N12) mice were purchsed from Tconic. P-selectin-deficient mice on mixed C57BL/6 129 ckground nd control mice were provided y D. Wgner (The CBR Institute for Biomedicl Reserch, Hrvrd Medicl School, Boston, Msschusetts) 43. Mice with conditionl deficiency in VCAM-1 were otined from P. Koni (Medicl College of Georgi, August, Georgi) nd were red s descried 44. T cell receptor trnsgenic OT- II Rg1 / mice (C57BL/6-TgN(OT-II.2)-RAG1 tm1mom ) were purchsed from Tconic through the Ntionl Institute of Allergy nd Infectious Diseses exchnge progrm 18,45. CMy-mOVA mice were otined from A. Lichtmn (Brighm & Women s Hospitl, Boston, Msschusetts) nd hve een descried 23. Mice were housed in n specific pthogen free, virl ntiody free fcility nd were used in ccordnce with guidelines of the niml committees of the CBR Institute nd Hrvrd Medicl School (Boston, Msschusetts). Regents. Fluorochrome-conjugted mas to mouse B22 (RA3-6B2), CD4 (RM4-5), CD8 (53-6.7), CD11 (M1/7), CD11c (HL3), CD25 (PC61), CD45.2 (14) nd V 5 (MR9-4), nd purified neutrlizing mas to mouse integrin 2 (GAME-46), ICAM-1 (3E2) nd P-selectin (RB4.34) were purchsed from BD Biosciences. Neutrlizing mas to mouse integrin 4 (PS/2), L-selectin (Mel-14), E-selectin (9A9), MAdCAM-1 (MECA-367) nd VCAM-1 (MK2.7) were purified from the superntnts of cultured hyridoms otined from Americn Type Culture Collection. Chimeric, nondepleting, nti- 4 (CRL19.11) ws provided y R. Plfrmn (Celltech, London, UK). For immunofluorescence, FITC-conjugted ntiody to pn cytokertin (C-11) ws otined from Sigm-Aldrich; iotinylted Ulex Europeus gglutinin I (UEA-1) ws purchsed from Vector Lortories. OVA( ) (H 2 N-ISQAVHAAHAEINEAGR-OH) ws purchsed from New Englnd Peptide. Endotoxin-free OVA in crude egg white extrct ws gift from M. Boes (Hrvrd Medicl School, Boston, Msschusetts) 46. PTX ws from Cliochem. DC isoltion nd culture. C57BL/6 mice were injected sucutneously with to B16 melnom cells secreting Flt3 lignd s descried 47. After 1 14 d, mice were killed nd DCs were purified y density-grdient centrifugtion over Optiprep (Sigm-Aldrich) y collection of the low-density frction 8. These preprtions routinely contined 75 85% CD11c + DCs. In some experiments, DC mturtion ws induced y culture for h in complete medium in the presence of 1 mg/ml of lipopolyscchride (E. coli.26:b6; Sigm-Aldrich). Mture DC cultures typiclly resulted in enrichment in CD11c + cells (9 95%) nd in the upregultion of clssicl mturtion mrkers (CD86 nd MHC clss II) for ll CD11c + cells. Homing ssy. Immture or mture DCs were leled for 15 min t 37 1C with 3 mm CFSE, 2 mm TRITC (tetrmethylrhodmine-5-(nd-6)-isothiocynte) or 1 mm CMTMR (5-(nd-6)-(((4-chloromethyl)enzoyl) mino)tetrmethylrhodmine; ll from Moleculr Proes). Ded cells nd excess lel were removed y centrifugtion over FCS, nd to leled DCs were then injected in the til veins of syngeneic recipient mice. In some experiments, DCs were pretreted with 5 mg/ml of locking ma nd were wshed efore injection to void the inhiition of dhesion molecules on simultneously injected control popultions; for inhiition of endothelil dhesion molecules, 1 mg ma ws injected long with the leled DCs. Mice were killed fter 18 h in most experiments nd fter 2 h in experiments in which homing receptors were locked on the DCs to minimize new expression fter ma tretment. Then, single-cell suspensions were generted from spleens nd thymi nd cell smples were incuted with nti-cd11c nd nlyzed on FACScliur (BD Biosciences). The totl numer of homed DCs ws clculted y multipliction of the frction of CFSE + (or TRITC + )CD11c + events y the totl cellulrity of the trget orgn. Immunofluorescence. Thymi were emedded in optimum cutting temperture compound nd were snp-frozen. Sections 2 mm in thickness were prepred, were ir-dried nd were fixed for 1 min with ice-cold ethnol. After eing locked with 1% (volume/volume) mouse serum, sections were incuted with vrious regents. Cortex nd medull were identified y stining with ma C-11 to pn cytokertin 48 nd the lectin UEA-1 (ref. 49), respectively. For secondry detection of UEA-1 iotin nd CD11ciotin, sections were incuted with Alex Fluor 633 conjugted streptvidin (Moleculr Proes). Clonl deletion ssy. C57BL/6 congenic (CD ) recipients were lethlly irrdited on dy y the dministrtion of two doses of 65 rds. Bone mrrow cells were isolted from donor OT-II Rg1 / (CD ) nd wild-type (CD ) mice. Mture lymphocytes were removed y incution with iotinylted ntiodies to NK1.1 (PK136), CD3 (145-2C11) nd CD19 (1D3), followed y immunomgnetic depletion with streptvidin-conjugted dyneds (DynlBiotech); to of the remining cells were injected intrvenously into ech irrdited recipient mouse. Then, 2 weeks lter, fresh DCs were isolted from C57BL/6 mice injected with B16 cells secreting Flt3 lignd, were purified s descried ove, were loded for 45 min t 37 1C with 2 mm OVA( ) with or without 5 mg/ml of nondepleting, 4 -locking ntiody CRL19.11 nd were dministered intrvenously to the one mrrow chimers. In some experiments, wild-type or I-A-knockout DCs were loded for 2 h t 37 1C with 1 mg/ml of whole OVA. Mice were treted twice, on dys 14 nd 15 fter irrdition, with loded or unloded DCs or with 5 nmol of solule peptide nd were killed on dy 17 unless otherwise indicted. In some experiments, regultory T cells were counted y gting on CD25 hi events (men fluorescent intensity of more thn 5) mong CD4 SP cells. Priosis. For priosis, 6- to 8-week-old femle C57BL/6 mice were nesthetized to full muscle relxtion nd were joined s descried 5. The lterl spects of ech mouse were shved, mtching skin incisions were mde from the olecrnon to the knee joint of ech mouse nd the sucutneous fsci ws luntly dissected to expose out.5 cm of skin. Olecrnons nd knees were joined with single 2- silk suture, nd the dorsl nd ventrl skins were pproximted y stples or continuous suture. FITC pinting. Mice were nesthetized nd hir ws removed from their dorsl skin. After skin ws stripped three times with trnsprent tpe to weken the strtum corneum, 5 ml of.5% solution of FITC in cetone nd olive oil (4:1, volume/volume) ws pplied to the skin. The use of diutyl phthlte ws voided to minimize skin sensitiztion nd ensuing DC mturtion. Mice were killed nd orgns were collected for nlysis 48 h lter. For detection of the rre FITC + CD11c hi MHC clss II positive cells, nd totl events were collected for spleen nd thymus, respectively. Sttisticl nlysis. Dt re presented s men ± s.e.m. unless otherwise noted. Sttisticl significnce ws ssessed y the two-tiled unpired Student s t-test for comprison of two groups or y one-wy nlysis of vrince followed y the Student Newmn Keuls post-test for more thn two groups. Differences with P vlues of less thn.5 were considered significnt. 8 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY