cd8+ recent thymic emigrants home to and efficiently repopulate the small intestine epithelium

8
CD8 + recent thymic emigrants home to and efficiently repopulate the small intestine epithelium Tracy L Staton 1 , Aida Habtezion 2 , Monte M Winslow 1 , Tohru Sato 2 , Paul E Love 4 & Eugene C Butcher 1–3 Prevailing knowledge dictates that naive ab T cells require activation in lymphoid tissues before differentiating into effector or memory T cells capable of trafficking to nonlymphoid tissues. Here we demonstrate that CD8 + recent thymic emigrants (RTEs) migrated directly into the small intestine. CCR9, CCL25 and a 4 b 7 integrin were required for gut entry of CD8 + RTEs. After T cell receptor stimulation, intestinal CD8 + RTEs proliferated and acquired a surface phenotype resembling that of intraepithelial lymphocytes. CD8 + RTEs efficiently populated the gut of lymphotoxin-a-deficient mice, which lack lymphoid organs. These studies challenge the present understanding of naive ab T cell trafficking and suggest that RTEs may be involved in maintaining a diverse immune repertoire at mucosal surfaces. The immune system is compartmentalized into primary, secon- dary and tertiary organs 1 . Mature lymphocytes are generated in primary lymphoid organs. After export, these naive lymphocytes are thought to recirculate in the secondary lymphoid organs, where they can be activated by cognate antigen. Once activated, lymphocytes can enter tertiary nonlymphoid sites, such as the skin and intestine, where they can function in clearing infection. Present knowledge suggests that naive ab T cell circulation is restricted to secondary lymphoid tissues 1,2 . The small intestine tertiary site is important in host defense. The gut epithelium forms a border between the outside environment and the body interior and is a potential entry site for antigens 3 . Of the immune cells that take up residence in the epithelium (called ‘intraepithelial lymphocytes’ (IELs)), more than 90% are CD8 + T cells. Because of the large surface area of the small intestine, IELs comprise a considerable fraction of the body’s T cells, and approxi- mately 1 10 9 IELs can be found in the human small intestine 4–6 . IELs have an activated or memory phenotype 7 and, once associated with the epithelium, do not recirculate throughout the body 8 . CD8 + IELs have high expression of a E integrin, which interacts with E-cadherin on epithelial cells and allows retention of IELs in the gut epithelium 9 . The CD8 + IEL compartment contains CD8aa + gd and ab T cells as well as conventional CD8ab + ab T cells 10–12 . Studies suggest that CD8ab + ab T cell receptor (TCR)–positive IELs originate from the thymus 13,14 , specifically via naive T cells that are activated in the Peyer’s patch that then migrate to the gut epithelium 15–17 . Recent thymic emigrants (RTEs), a distinct subpopulation of naive ab T cells, have ‘preferential’ access to survival factors and niches and are ‘preferentially’ incorporated into the peripheral naive ab T cell pool 18 . Splenic RTEs are less responsive than other naive ab T cells to TCR stimulation 19 and may be especially susceptible to tolerization 20 . CD8 + RTEs have unique adhesive and chemotactic properties. Unlike other naive ab T cells, RTEs express a E integrin and CCR9, which is involved in recruiting memory T cells to the intestines 21,22 . CCR9 is a chemoattractant receptor expressed on small intestine–specific T cells, and its ligand, CCL25, is expressed by small intestine epithelial cells 23 . CD8 + RTEs also express a 4 b 7 integrin, which is the receptor for the intestinal vascular addressin MAdCAM-1. Those features support the hypothesis that CD8 + RTEs might home to the mucosal surface of the small intestine without prior activation and ‘reprogramming’ of homing properties in peripheral lymphoid tissues. Moreover, because RTEs establish and contribute to the TCR diversity of the peripheral ab T cell pool, we reasoned that RTEs might also maintain the TCR diversity of the IEL population 24 . Here we demonstrate that CD8 + RTEs homed to the gut and show that this migration required a 4 b 7 integrin, CCR9 and CCL25. RTE gut homing occurred independently of prior antigen recognition in peripheral lymphoid tissues. After entry into the gut wall, CD8 + RTEs proliferated and differentiated, ultimately acquiring the pheno- type of resident IELs. CD8 + RTEs populated the IEL compartment of lymphopenic mice more efficiently than naive non-RTE CD8 + T cells. In the steady state, CD8ab + TCRab + RTEs were readily recovered from the resident IEL population and expressed a diverse TCR repertoire. Our data indicate that trafficking of naive CD8 + ab T cells, particularly RTEs, is not limited to secondary lymphoid tissues. Instead, CD8 + RTEs can home directly into the intestinal wall, where they might help to maintain the TCR diversity of the mucosal CD8ab + ab T cell pool. Received 21 November 2005; accepted 14 February 2006; published online 2 April 2006; corrected after print 2 May 2006 (details online); doi:10.1038/ni1319 1 Program in Immunology and 2 Laboratory of Immunology and Vascular Biology, Department of Pathology, Stanford University School of Medicine, Stanford, California 94305, USA. 3 Center for Molecular Biology and Medicine, Veterans Affairs Palo Alto Health Care System, Palo Alto, California 94304, USA. 4 Laboratory of Mammalian Genes and Development, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland 20892, USA. Correspondence should be addressed to E.C.B. ([email protected]). NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 1 ARTICLES © 2006 Nature Publishing Group http://www.nature.com/natureimmunology

Upload: eugene-c

Post on 29-Jul-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

CD8+ recent thymic emigrants home to and efficientlyrepopulate the small intestine epithelium

Tracy L Staton1, Aida Habtezion2, Monte M Winslow1, Tohru Sato2, Paul E Love4 & Eugene C Butcher1–3

Prevailing knowledge dictates that naive ab T cells require activation in lymphoid tissues before differentiating into effector

or memory T cells capable of trafficking to nonlymphoid tissues. Here we demonstrate that CD8+ recent thymic emigrants

(RTEs) migrated directly into the small intestine. CCR9, CCL25 and a4b7 integrin were required for gut entry of CD8+ RTEs.

After T cell receptor stimulation, intestinal CD8+ RTEs proliferated and acquired a surface phenotype resembling that of

intraepithelial lymphocytes. CD8+ RTEs efficiently populated the gut of lymphotoxin-a-deficient mice, which lack lymphoid

organs. These studies challenge the present understanding of naive ab T cell trafficking and suggest that RTEs may be

involved in maintaining a diverse immune repertoire at mucosal surfaces.

The immune system is compartmentalized into primary, secon-dary and tertiary organs1. Mature lymphocytes are generated inprimary lymphoid organs. After export, these naive lymphocytes arethought to recirculate in the secondary lymphoid organs, where theycan be activated by cognate antigen. Once activated, lymphocytescan enter tertiary nonlymphoid sites, such as the skin and intestine,where they can function in clearing infection. Present knowledgesuggests that naive ab T cell circulation is restricted to secondarylymphoid tissues1,2.

The small intestine tertiary site is important in host defense. Thegut epithelium forms a border between the outside environment andthe body interior and is a potential entry site for antigens3. Of theimmune cells that take up residence in the epithelium (called‘intraepithelial lymphocytes’ (IELs)), more than 90% are CD8+

T cells. Because of the large surface area of the small intestine, IELscomprise a considerable fraction of the body’s T cells, and approxi-mately 1 � 109 IELs can be found in the human small intestine4–6.IELs have an activated or memory phenotype7 and, once associatedwith the epithelium, do not recirculate throughout the body8. CD8+

IELs have high expression of aE integrin, which interacts withE-cadherin on epithelial cells and allows retention of IELs in the gutepithelium9. The CD8+ IEL compartment contains CD8aa+ gd andab T cells as well as conventional CD8ab+ ab T cells10–12. Studiessuggest that CD8ab+ ab T cell receptor (TCR)–positive IELs originatefrom the thymus13,14, specifically via naive T cells that are activated inthe Peyer’s patch that then migrate to the gut epithelium15–17.

Recent thymic emigrants (RTEs), a distinct subpopulation of naiveab T cells, have ‘preferential’ access to survival factors and niches andare ‘preferentially’ incorporated into the peripheral naive ab T cell

pool18. Splenic RTEs are less responsive than other naive ab T cells toTCR stimulation19 and may be especially susceptible to tolerization20.CD8+ RTEs have unique adhesive and chemotactic properties. Unlikeother naive ab T cells, RTEs express aE integrin and CCR9, which isinvolved in recruiting memory T cells to the intestines21,22. CCR9 is achemoattractant receptor expressed on small intestine–specific T cells,and its ligand, CCL25, is expressed by small intestine epithelial cells23.CD8+ RTEs also express a4b7 integrin, which is the receptor for theintestinal vascular addressin MAdCAM-1. Those features support thehypothesis that CD8+ RTEs might home to the mucosal surface ofthe small intestine without prior activation and ‘reprogramming’ ofhoming properties in peripheral lymphoid tissues. Moreover, becauseRTEs establish and contribute to the TCR diversity of the peripheralab T cell pool, we reasoned that RTEs might also maintain the TCRdiversity of the IEL population24.

Here we demonstrate that CD8+ RTEs homed to the gut and showthat this migration required a4b7 integrin, CCR9 and CCL25. RTE guthoming occurred independently of prior antigen recognition inperipheral lymphoid tissues. After entry into the gut wall, CD8+

RTEs proliferated and differentiated, ultimately acquiring the pheno-type of resident IELs. CD8+ RTEs populated the IEL compartment oflymphopenic mice more efficiently than naive non-RTE CD8+ T cells.In the steady state, CD8ab+ TCRab+ RTEs were readily recoveredfrom the resident IEL population and expressed a diverse TCRrepertoire. Our data indicate that trafficking of naive CD8+ abT cells, particularly RTEs, is not limited to secondary lymphoidtissues. Instead, CD8+ RTEs can home directly into the intestinalwall, where they might help to maintain the TCR diversity of themucosal CD8ab+ ab T cell pool.

Received 21 November 2005; accepted 14 February 2006; published online 2 April 2006; corrected after print 2 May 2006 (details online); doi:10.1038/ni1319

1Program in Immunology and 2Laboratory of Immunology and Vascular Biology, Department of Pathology, Stanford University School of Medicine, Stanford, California94305, USA. 3Center for Molecular Biology and Medicine, Veterans Affairs Palo Alto Health Care System, Palo Alto, California 94304, USA. 4Laboratory of MammalianGenes and Development, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland 20892, USA. Correspondenceshould be addressed to E.C.B. ([email protected]).

NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 1

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

RESULTS

Homing of CD8+ RTEs to nonlymphoid tissue

To analyze the peripheral homing patterns and tissue distribution ofCD8+ RTEs in adult mice, we labeled emigrating thymocytes byinjection of intrathymic fluorescein isothiocyanate (FITC). At 24 hafter FITC injection, all peripheral FITC+ cells are RTEs25,26. Inagreement with published data, we detected FITC+ CD8+ RTEs inthe spleen, peripheral lymph nodes, mesenteric lymph nodes, Peyer’spatches and blood26 (Fig. 1a). To determine if CD8+ RTEs could alsolocalize to the small intestine, we isolated IELs and lamina proprialymphocytes. FITC+ cells expressing the RTE markers aE integrin andCCR9 were present in these gut T cell populations and displayed anaive CD62LhiCD44lo surface phenotype (Fig. 1a and data notshown). These results suggest that CD8+ RTEs migrate both to thegut and to secondary lymphoid organs.

Naive CD8+ RTEs among IELs

We next sought to determine if naive T cells could be found in thesteady-state resident IEL population. There was a small (comparedwith secondary lymphoid organs) but detectable fraction ofCD44loCD62LhiCD8ab+ ab T cells (2–10% in 4- to 12-week-oldmice) in the IEL population (Fig. 1a, left). A large percentage ofCD62LhiCD44lo IELs also expressed CCR9 and aE integrin, suggestingthat RTEs were enriched for this population (Fig. 1a and data notshown). In addition, there were similar proportions of FITC+ cells inthe CD44loCD62Lhi CD8ab+ ab T cell populations in the secondarylymphoid organs and the small intestine (Fig. 1a, right). These resultsfurther support the idea that RTEs can migrate to the gut.

Thymic origin of gut CD8+ RTEs

To rigorously examine the mechanisms regulating RTE trafficking, wedeveloped a short-term homing assay using enriched CD8+ single-positive (SP) thymocytes as a source of RTEs. CD8+ SP thymocytes arethe immediate precursors of RTEs, and adoptive transfer of CD8+ SPthymocytes results in their forced emigration27,28. We isolated CD8+

SP thymocytes by depleting samples of CD4+CD8+ double-positiveand CD4+CD8– SP thymocytes, followed by staining for CD8 (to

exclude CD4–CD8– double-negative thymocytes). CD8+ SP thymo-cytes had high expression of aE integrin and CD62L and intermediateexpression of a4b7 integrin (in contrast to high expression on memorycells) and CCR9 (Fig. 1b and data not shown). The amounts of aE

integrin, CD62L, CCR9 and a4b7 integrin on CD8+ SP thymocyteswere similar to those on FITC+ CD8+ RTEs in the spleens of micegiven intrathymic injection of FITC (data not shown). FITC+ CD8+

RTEs and adoptively transferred FITC-labeled CD8+ SP thymocyteshad similar tissue distribution (Fig. 1a,c), suggesting that the twopopulations had similar migratory capabilities. Unlike methods usedbefore to identify RTEs (such as TCR rearrangement excision circleanalysis or the use of recombination-activating gene–green fluorescentprotein (GFP) mice), detection of RTEs in mice receiving FITC-labeled CD8+ SP thymocytes is not influenced by cell division29,30.

CD8+ RTE localization in the small intestine

To identify the specific intestinal region to which CD8+ SP thymocytesmigrate, we transferred carboxyfluoroscein succinimidyl ester (CFSE)–labeled Thy-1.2+ CD8+ SP thymocytes into Thy-1.1+ congenic reci-pients. Staining of small intestine tissue sections isolated 1d (Fig. 2a)or 8 d (data not shown) after transfer demonstrated the presence ofThy-1.2+ cells in the lamina propria as well as in the region adjacent tothe epithelium. Counterstaining for the CD31 endothelial markerconfirmed that few if any Thy-1.2+ cells were present in vessels ateither time point (data not shown).

Secondary lymphoid tissue–independent gut localization

After antigen stimulation in secondary lymphoid tissues, naive abT cells are ‘reprogrammed’ to home to nonlymphoid sites31.

a b

c

CD8αβ+

Spleen

Naive89 2.0

88 1.9

90 1.8

IEL Spleen

PLN Blood

78 1.9

1.5 1.5

0.9 1.24 2.0

RTE

PLN

MLN

CD

62L

CD62L CCR9

Enriched CD8 RTE(gated on CD8α+CD4–

αE integrin α4β7 integrinC

ell n

umbe

r

Cel

l num

ber

Cel

l num

ber

FIT

C

CD8α

Cel

l num

ber

)

CD44 CD8β

PP

IEL

CD8αβ+

CD44loCD62Lhi

CD8αβ+ CD62Lhi

FIT

C

Figure 1 Homing of CD8+ RTEs to the gut. (a) CD44loCD62Lhi (naive) cells

of the CD8ab+ fraction (left) and FITC+ (RTE) cells of the CD44loCD62Lhi

CD8ab+ fraction (right) in various tissues (left margin). Numbers beside

boxed areas indicate the percentage of cells in each fraction. PLN,

peripheral lymph node; MLN, mesenteric lymph node; PP, Peyer’s patch.

(b) Expression of aE integrin, CD62L, CCR9 and a4b7 on CD8+ SP

thymocytes. Shaded histograms indicate isotype control staining.

(c) Analysis of FITC-labeled CD8+ SP thymocytes 24 h after transfer. Plotsare gated on CD8ab+ CD62Lhi cells; numbers beside boxed areas indicate

the percent of CD44loCD62Lhi CD8ab+ cells that are FITC+. All data are

representative of at least three independent experiments.

Uninjected Thy-1.1 Thy-1.2 RTE into Thy-1.1

FIT

C

CD8α

Thy

-1.2

Hem

atox

ylin

IEL

9 5

5

Spleen

Blood

CD8αβ+ CD62Lhia b

PLNabsent

from Lta–/–

Figure 2 Localization of CD8+ SP thymocytes in the gut. (a) Staining of

small intestine sections from Thy-1.1+ congenic BALB/c recipients 1 d after

transfer of CFSE-labeled Thy-1.2+ CD8+ SP thymocytes (right) or no cells

(left). Sections were stained with an anti-Thy-1.2 and were developed with

NovaRED. Original magnification, �20 (top row) and �40 (bottom row).

(b) Flow cytometry of FITC-labeled CD8+ SP thymocytes after transfer into Lta–/–

recipients. Plots are gated on CD8ab+CD62Lhi cells; numbers beside boxed

areas indicate percent FITC+. All data are representative of three experiments.

2 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

Therefore, although unlikely because of the short 24-hour time frameof our experiment, it was possible that before gut localization,adoptively transferred CD8+ SP thymocytes could have passed throughPeyer’s patches or mesenteric lymph nodes and could have been‘reprogrammed’ to allow gut homing. To rule out a requirement for‘reprogramming’ in lymphoid tissues, we analyzed lymphotoxin-a-deficient (Lta–/–) mice, which lack organized secondary lymphoidtissues and lymphoid structures in the small intestine and splenicwhite pulp32,33. Even in the absence of secondary lymphoid structures,naive ab T cells were present in the CD8ab+ IEL compartment inthe small intestines of Lta–/– mice (data not shown), and adopt-ively transferred FITC-labeled CD8+ SP thymocytes migrated to thegut epithelium of Lta–/– mice (Fig. 2b). These results support the hypo-thesis that CD8+ RTEs can transit directly from the thymus to the gut.

Homing mechanisms regulating gut localization

To investigate the mechanisms regulating gut localization of CD8+ SPthymocytes, we analyzed the involvement of several candidate surfacemolecules. We first focused on CCR9 because it is expressed by CD8+

RTEs but not by other naive cells and is required for intestinallocalization of activated and memory ab T cells21,22,34. We mixedwild-type and Ccr9–/– CD8+ SP thymocytes labeled with the redfluorescent cell linker PKH26 and FITC, respectively, and transferredthe cells into wild-type recipient mice. Although wild-type and CCR9-deficient CD8+ SP thymocytes migrated with equal efficiency tolymph nodes, Peyer’s patches, spleen and blood, CCR9-deficientCD8+ SP thymocytes migrated much less efficiently than wild-typeCD8+ SP thymocytes to the small intestine epithelium (Fig. 3a,b).We obtained similar results in experiments using reciprocal labeling(FITC-labeled wild-type and PKH26-labeled Ccr9–/– CD8+ SP

thymocytes), and FITC- and PKH26-labeled wild-type CD8+ SPthymocytes transferred together migrated with equal efficiency tothe gut, confirming that the dyes did not have differential effects onhoming. In addition, we isolated CD4+ SP thymocytes, which do notexpress CCR9, from wild-type and Ccr9–/– mice, labeled the cells withFITC or PKH26 and transferred them together into wild-type reci-pients. Wild-type and Ccr9–/– CD4+ SP thymocytes had similarhoming patterns (Fig. 3b).

Developmental defects intrinsic to Ccr9–/– thymocytes might haveindirectly affected their migration patterns. To confirm that CCR9 wasrequired for trafficking of CD8+ SP thymocytes to the gut, we assessedthe function of the CCR9 ligand CCL25. Gut homing of CD8+ SP

a b c dCD8α+

KO (FITC) CD8α

PLN

/SI I

EL

WT→WTCD8

KO→WTCD4

KO→WTCD8

Spleen PLN MLN PP IEL

0.10

0.05

0Blood

WT

(P

KH

26)

FIT

C

FIT

C+ C

D8+

(%

)

LN

IEL

LN

Isotype

Isotype *

Anti-CCL25

Anti-CCL25

IEL

1.6

1.3 1.2

0.2 0.02

32

16

8

4

2

1

0.5 0.0

0.8

1.6

2.4

0.07

Figure 3 CCR9 and CCL25 in gut homing of CD8+ SP thymocytes. (a) Flow cytometry of PKH26-labeled wild-type (WT) and FITC-labeled Ccr9–/– (KO) CD8+

SP thymocytes transferred together into wild-type recipients. Plots are gated on CD8a+ cells; numbers in top right corners indicate the Ccr9–/–/wild-type ratio

from the boxed areas. (b) Ratio of transferred SP thymocytes in the peripheral lymph node (PLN) versus small intestine (SI IEL). SP thymocytes isolated from

wild-type or Ccr9–/– mice were differentially labeled and transferred together into wild-type mice (horizontal axis). Each dot represents an individual recipient

mouse. (c,d) Flow cytometry of FITC-labeled CD8+ SP thymocytes transferred into mice treated with isotype control antibody or anti-CCL25. Numbers beside

boxed areas in plots (c) and graphed data (d) indicate the percent of cells in the CD8a+ population that are FITC+. In d, data represent mean ± s.e.m. of

values obtained with various tissues (horizontal axes) normalized to values obtained for peripheral lymph nodes of four individual mice per group. *, P o 0.08.

CD8α

FIT

C

LN

Isotype Anti-CD62L

IsotypeCD62L

IEL

CD8α

FIT

C

FIT

C+

CD

8+ (

%)

FIT

C+C

D8+

(%

)

LN0.9 0.9

0.1

1.0

0.8 0.8

0.4 0.5

0.01

0.03 0.03

<0.1

2.0

1.0

0.0

1.0

0.5

0

0.08

0.04

0

0.02

0.01

0

0.050

0.025

0

Isotypea b

FIT

C+C

D8+

(%

)

1.0

0.5

0

dc

e f

Anti-α4 integrin

Isotype *

*

*****

α4 integrin

IEL

Spleen

PLNM

LNBloo

d PP

Spleen

PLNM

LNBloo

d PP

Spleen

PLNM

LNBloo

d PP

IEL

IEL

IEL

CD8α

FIT

C

LN

Isotype Anti-αE integrin

Isotype NSαE integrin

IEL

Figure 4 The a4 integrin in gut homing of CD8+ SP thymocytes. FITC-

labeled CD8+ SP thymocytes were pretreated with isotype control or

antibodies specific for a4 integrin (a,b; n ¼ 4), CD62L (c,d; n ¼ 3) or aE

integrin (e,f; n ¼ 3) and were adoptively transferred into wild-type recipient

mice, followed by flow cytometry. Representative plots (a,c,e) are gated on

CD8a+ cells; numbers beside or above boxed areas indicate the percent ofCD8a+ cells that are FITC+. Graphed data (b,d,f) represent mean ± s.e.m.

of the percentage of CD8a+ cells that are FITC+ in each tissue, normalized

to that of peripheral lymph nodes (b) or spleen (d,f) *, P o 0.02; **,

P o 0.003; ***, P o 0.001; NS, not significant.

NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 3

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

thymocytes was less efficient in mice treated with a neutralizingantibody specific for CCL25 than in mice treated with an isotypecontrol antibody (Fig. 3c,d). These results suggest that CCR9 isrequired for gut homing of CD8+ SP thymocytes.

Next we characterized the function of adhesion molecules. Wetransferred FITC-labeled CD8+ SP thymocytes into mice treatedwith either an isotype control antibody or a neutralizing antibodyspecific for a4b7 integrin, a4 integrin, CD62L or aE integrin. Wemaintained saturating quantities of blocking antibodies throughoutthe experiments, as confirmed by staining of cells from treated micewith a labeled version of the blocking antibody (SupplementaryFigure 1 online).

Blockade of a4 integrin inhibited homing of CD8+ SP thymocytesto the gut and Peyer’s patches (Fig. 4a,b). The a4 integrin can pairwith the b1 or b7 chain to form a4b1 or a4b7 integrin. Blockade withan antibody specific for the a4b7 integrin also inhibited gut homing ofCD8+ SP thymocytes (data not shown). CD62L is essential for homingof naive cells to the lymph nodes. CD62L blockade inhibited homingof CD8+ SP thymocytes to the peripheral lymph nodes but did notalter homing of CD8+ SP thymocytes to the small intestine (Fig. 4c,d).The aE integrin distinguishes CD8+ RTEs from other naive CD8+

cells34. However, antibody blockade of aE integrin did not inhibittrafficking of CD8+ SP thymocytes to the small intestine (Fig. 4e,f).These results indicate that a4b7 integrin is required for the homing ofCD8+ SP thymocytes to the small intestine.

Gut CD8+ RTE proliferation

To determine whether CD8+ SP thymocytes proliferate in the periph-ery, we analyzed CFSE dilution of adoptively transferred CFSE-labeled

CD8+ SP thymocytes. By 8 d after transfer,CD8+ SP thymocytes in the small intestinehad undergone multiple rounds of cell divi-sion (Fig. 5a). CD8+ SP thymocytes at othersites, including peripheral lymph nodes,mesenteric lymph nodes, Peyer’s patches,spleen and blood, did not undergo suchextensive proliferation (Fig. 5b and data notshown). We used Thy-1-mismatched recipi-ents in similar experiments, which indicatedthat CFSE labeling alone was adequatefor tracking transferred cells accurately (datanot shown).

The proliferation of gut CD8+ SP thymo-cytes could be driven by TCR signals or bysignals received specifically in the intestinalenvironment (such as growth factors or cyto-kines). To determine whether proliferation ofCD8+ SP thymocytes was antigen induced,we used OT-I TCR-transgenic recombina-tion-activating gene 1–deficient (Rag1–/–)mice35,36. These mice express a single TCRand therefore cannot respond to as diverse apanel of antigens as wild-type mice can. Welabeled OT-I+Rag1–/– and C57BL/6 CD8+ SPthymocytes with CFSE and transferred thecells into separate C57BL/6 recipient mice. By5 d after transfer, gut OT-I+Rag1–/– CD8+ SPthymocytes had undergone fewer divisionsthan gut C57BL/6 CD8+ SP thymocytes(Fig. 5a). In contrast, there was no differencein the proliferation profiles of OT-I+Rag1–/–

and C57BL/6 CD8+ SP thymocytes at other peripheral sites (Fig. 5band data not shown). These results indicate that CD8+ SP thymocytesproliferated in response to an antigen present exclusively in the gut.OT-I+Rag1–/– and C57BL/6 CD8+ SP thymocytes migrated with equalefficiency to the gut (Supplementary Figure 2 online), suggesting thatTCR specificity does not affect homing.

Intraintestinal differentiation of CD8+ RTEs

Resident IELs have an aE integrin–high, CD62Llo surface phenotype,whereas CD8+ SP thymocytes have an aE integrin–intermediate,CD62Lhi phenotype. At 2 d after adoptive transfer, gut OT-I+Rag1–/–

and C57BL/6 CD8+ SP thymocytes retained their aE integrin–inter-mediate, CD62Lhi surface phenotype. However, by 8 d after transfer,gut OT-I+Rag1–/– and C57BL/6 CD8+ SP thymocytes downregulatedCD62L and upregulated aE integrin to amounts seen on resident IELs(Fig. 5c). CD8+ SP thymocytes at other peripheral sites retained theiroriginal aE integrin–intermediate, CD62Lhi surface phenotype (Fig. 5dand data not shown). These data indicate that proliferation is notmandatory for differentiation from naive CD8+ SP thymocytes intoresident IELs and that differentiation can occur in the absence ofspecific TCR stimulation.

Efficiency of CD8+ RTE gut homing

As RTEs are distinguished from non-RTE naive CD8+ T cells by theirunique pattern of CCR9 and aE integrin expression, we sought todetermine whether CD8+ RTEs home more efficiently than non-RTEnaive CD8+ T cells to the gut. We mixed purified RTE and non-RTEnaive CD8+ T cells labeled with FITC and PKH26, respectively, andinjected the cells into wild-type mice. CD8+ SP thymocytes consistently

0

B6 B6 OT-I B6

Cel

ls in

CF

SE

divi

sion

pea

k (%

)

Day 2 Day 8Day 5

25

50

75

100

Time after transfer

IEL PLNa b

c d

0

25

50

75

100

CD62L

Day 2 Day 5 Day 8

CD62L

α E in

tegr

in

α E in

tegr

in

Day 2 Day 5 Day 8

B6

B6

OT-I

B6

B6

B6

OT-I

B6

123 0 123 0 123 0 123 0 123 0 123 0

IEL PLN

Division

Day 2 Day 8Day 5

Time after transfer

123 0 123 0 123 0 123 0 123 0 123 0Division

Cel

ls in

CF

SE

divi

sion

pea

k (%

)

*

*

B6 B6OT-I B6

Figure 5 Intraintestinal proliferation and differentiation of CD8+ SP thymocytes. (a,b) CFSE dilution

analysis of cell division. C57BL/6 (B6) or OT-I+Rag1–/– (OT-I) CD8+ SP thymocytes were isolated,

labeled with CFSE and transferred into C57BL/6 recipient mice, which were killed on day 2, 5 or

8 after cell transfer. Based on CFSE dilution, cells were classified as having undergone zero to threecell divisions (horizontal axes). Data represent the mean ± s.e.m. of three individual recipients in each

group. *, P o 0.001. (c,d) Representative CD62L and aE integrin profiles of CD8ab+CFSE+ cells (from

a,b) on day 2, 5 or 8 after transfer (n ¼ 3).

4 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

homed to the gut more efficiently than non-RTE naive CD8+ T cells(Fig. 6a). However, both populations homed equally efficiently toother peripheral tissues. In contrast, CD8+ SP thymocytes and per-ipheral naive CD8+ RTEs (aE integrin–positive, CD62L+) homed withequal efficiency to all peripheral tissues (Fig. 6b). These data furtherconfirm the validity of our CD8+ SP thymocyte adoptive transfermodel, as CD8+ SP thymocytes homed comparably to splenic andperipheral CD8+ T cells with an RTE surface phenotype.

We compared the ability of CD8+ SP thymocytes and non-RTEnaive CD8+ cells to reconstitute the lymphopenic intestine of micewith severe combined immunodeficiency (SCID). The intestinalepithelium of SCID mice is devoid of CD8ab+ ab T cells12. However,we noted the presence of a small population of CD8aa+ gd T cells ingut of SCID mice. As CD8+ RTEs express CD8ab+ and TCRab, theseCD8aa+ gd T cells were ‘gated out’ of our analyses. We sorted CD8+

SP thymocytes from ‘actin-GFP’ mice and non-RTE naive CD8+

T cells from wild-type mice, mixed the cells and transferred themtogether into SCID recipient mice (Fig. 6c). We analyzed reconstitu-tion at various time points from 2 d to 4 weeks after transfer. By 7 dafter transfer, the RTE/non-RTE ratio was higher in the gut than inthe blood of each mouse (Fig. 6d). These results indicate that RTEshome more efficiently to the gut than do their non-RTE naive CD8+

T cell counterparts (Supplementary Fig. 3 online).

DISCUSSION

Our results have demonstrated that CD8+ RTEs have a unique surfacephenotype that allows them to migrate directly from the blood tothe small intestine. Using gene-targeted mice and in vivo antibodyblockade, we have shown that this gut entry required a4b7

integrin, CCR9 and CCL25. CD8+ SP thy-mocytes were more efficient than non-RTECD8+ T cells in entering and repopulatingthe small intestine. After entry into the gut,CD8+ SP thymocytes underwent post-thymicdifferentiation involving TCR-dependentproliferation and TCR-independent, micro-environment-dependent acquisition of aresident IEL surface phenotype.

It is generally believed that naive T cellhoming is restricted to organized secondarylymphoid tissues, particularly the spleen,lymph nodes and Peyer’s patches. Therefore,our substantial recovery of naive CD8+ T cellsfrom the small intestine epithelium was unex-pected. We did several rigorous experimentsto show that naive CD8+ T cells, particularlyRTEs, were present in the small intestine andwere not due to contamination from Peyer’spatches or blood. There were naive CD8+

T cells and RTEs in IEL populations fromwild-type and from Lta–/– mice, suggestingthat the presence of these cells in the gut wasnot due to contamination from Peyer’spatches or small lymphoid aggregates in thesmall intestine. Histological localization ofadoptively transferred CD8+ SP thymocytesthat could be identified by fluorescence andgenetics demonstrated that these cells werelocalized specifically in the lamina propriaand IEL compartments. Recruitment ofRTEs to the gut was specifically blocked

by inhibition of CCR9-CCL25 interactions or by blockade of a4

integrin, ruling out the possibility of blood contamination.The finding that naive CD8+ T cells comprise a substantial

percentage of small intestine IELs suggests that the trafficking ofnaive ab T cells is not limited to secondary lymphoid organs. It will benecessary to evaluate the extent of homing and the functionalrelevance of naive ab T cells in other nonlymphoid peripheral tissues.In particular, our results raise the possibility that ab T cell populationsresident in the skin, brain, liver and lungs may also be derived, at leastin part, from naive ab T cells recruited directly from the blood.

After entry into the small intestine (but not after entry intosecondary lymphoid tissues), CD8+ SP thymocytes underwent multi-ple rounds of cell division and adopted a resident IEL phenotype.Specifically, aE integrin, which tethers IELs to neighboring epithelialcells, was induced9. Our studies with OT-I+Rag1–/– RTEs suggestedthat specific TCR stimulation was necessary for maximal gut prolif-eration of CD8+ SP thymocytes. Those results do not rule out thepossibility that growth factors cooperate with TCR signals in regulat-ing proliferation, but they indicate that the local population expansionof RTEs is driven by their response to antigens in the small intestine.In contrast, acquisition of a resident IEL phenotype (aE integrin–high,CD62Llo) occurred independently of TCR signals. That change in cellsurface phenotype may have been due to transforming growth factor-b receptor signaling, as there is high expression of transforminggrowth factor-b in the intestinal environment37 and it is involved inaE integrin induction in vitro38,39.

The mechanisms regulating antigen presentation in the small intes-tine differ from those that regulate antigen presentation in lymphnodes. Specifically, epithelial cells are the main antigen-presenting cells

a c

b d

4

3

2

1

0PLN MLN

FIT

C/P

KH

26

CD

8

CD

8

CD

8

47 52

α E in

tegr

in

CD8 RTE (GFP+) Injected cellsNaive CD8 cells (GFP–)

CD4 CD4 CD62L GFPBlood PP IEL

*

NS

4

3

2

1

0PLN MLN

FIT

C/P

KH

26

GF

P+/G

FP

(RT

E/n

on-R

TE

)

Blood PP IEL 10

1

2

3

4

5

6 BloodIEL

2Day 7 Day 21

Time after SCID reconstitution

3 4 5 6

Figure 6 ‘Preferential’ homing of CD8+ RTEs to the gut. (a,b) FITC-labeled CD8+ SP thymocytes were

transferred together with equal numbers of PKH26-labeled CD8+, aE integrin–negative, CD62L+ cells (a)

or CD8+, aE integrin–positive, CD62L+ cells (b) into wild-type recipient mice. At 24 h after transfer, the

FITC/PKH26 ratios were calculated for various tissues (horizontal axes) and were normalized to those for

peripheral lymph nodes. A ratio of more than 1 indicates a competitive homing advantage for RTEs.

Data represent mean ± s.e.m. of four individual mice per group. *, P o 0.01; NS, not significant.

(c) Flow cytometry of CD8+ SP thymocytes from actin-GFP mice and CD8+, aE integrin–negative,

CD62L+ cells from wild-type mice, mixed and transferred together into SCID recipient mice. In the

first three plots, boxed areas indicate sort gates; dot plots show the purity of sorted populations afterreanalysis. Numbers beside boxed areas (far right plot) indicate percent GFP– cells (left) or GFP+

cells (right). Plots are representative of three independent experiments. (d) GFP+/GFP– ratio among

CD8ab+TCRb+TCRgd– cells in the blood and small intestines of SCID mice analyzed at various times

after transfer (horizontal axis). Data from six mice are presented individually (1–6, below bars).

NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 5

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

in the gut40. In the uninflamed state, intestinal epithelial cells arethought to present antigen in a suppressive way because they havelower expression of costimulatory molecules than do ‘professional’antigen-presenting cells41. In addition, transforming growth factor-bpresent in the gut epithelium contributes to the generation of animmunosuppressive microenvironment37. That may explain why RTEsand other IEL subsets proliferate and have an effector or memoryphenotype but do not normally cause overt inflammatory symptoms.

Published work suggests a thymic origin for CD8ab+TCRab+

IELs13. IELs were thought to originate from naive ab T cells activatedin Peyer’s patches, which subsequently migrated into the smallintestinal epithelium15,16. Here we have shown that at least someresident IELs trafficked in a naive state directly from the thymusthrough the blood into the gut. Moreover, although the lymphopenicgut can be reconstituted by total lymph node cells42, our resultshave shown that the gut could be reconstituted by a naive CD8+ SPthymocyte population ‘preprogrammed’ in the thymus to migrate tothis site.

What can RTEs contribute to the IEL population? In contrast to thepreviously activated cells ‘programmed’ in the Peyer’s patches to enterthe gut, RTEs have a diverse repertoire of TCRs. The TCRb repertoireof naive CD8+ IELs is similar in diversity to that of splenic and lymphnode RTEs and CD8+ SP thymocytes but is less diverse than that ofmemory CD8+ ab T cells (data not shown)43,44. It is expected that theTCR repertoire of IELs becomes more restricted over time because ofgut-specific antigen recognition and clonal amplification. Thus, con-tinuous RTE recruitment may be an important mechanism formaintaining TCR diversity at mucosal surfaces.

METHODSMice. BALB/c, C57BL/6, SCID and Lta–/– mice were purchased from Jackson

Laboratories. OT-I+Rag1–/– and C57BL/6 control mice were purchased from

Taconic Farms35,36. Thy-1.1+ congenic BALB/c mice were a gift from

I. Weissman (Stanford University School of Medicine, Stanford, California).

Ccr9–/– mice45 and actin-GFP mice (transgenic mice expressing enhanced GFP

under the control of the promoter of the gene encoding chicken b-actin)46 have

been described. Unless stated otherwise, mice used were between 4 and 12

weeks of age. Animals were housed in a specific pathogen–free facility. All

animal protocols were in accordance with institutional, federal and state

guidelines and were approved by the Veterans Affairs Palo Alto Health Care

System Institutional Animal Care and Use Committee (Palo Alto, California).

Flow cytometry. Antibodies to CD4, CD8a, CD44, CD49d, CD62L, CD103–aE

integrin, a4b7, TCRb and TCRgd and rat IgG2ak and IgG2bk isotype controls

were from BD Pharmingen. Antibody to CD8b (anti-CD8b) was from

eBioscience. Cells were stained in Hank’s balanced-salt solution containing

1% bovine serum and were analyzed on an LSRII (BD Biosciences) or

were sorted with a FACSVantage (BD Biosciences). CCR9 staining (clone

242503) was done according to the manufacturer’s protocol (R&D Systems).

Data were analyzed with CellQuest software (BD Biosciences) and FlowJo

software (Treestar).

IEL preparation. IELs were isolated as described47 with some modifications.

After removal of Peyer’s patches, small intestines were rinsed twice with Hank’s

balanced-salt solution containing 2% bovine calf serum, were cut into pieces

5 mm in length and were incubated at 37 1C while being stirred for two

10-minute sessions, with a change of media between sessions. IELs were

collected from the interface of a 40–70% Percoll gradient.

CD8+ RTE enrichment and labeling. Total thymocyte suspensions from 4- to

8-week-old mice were depleted of CD4+ thymocytes with anti–mouse CD4

MACS beads and LD columns according to the manufacturer’s protocol

(Miltenyi Biotech). After depletion, 50–70% of the population (CD8+ SP

thymocytes) was CD8+CD4–TCRhi and the remaining cells were CD4–CD8–.

Cells were labeled with FITC (Sigma), CFSE (Molecular Probes) or PKH26

(Sigma) before transfer, as described25,48.

In vivo antibody blockade. FITC-labeled CD8+ SP thymocytes were preincu-

bated with blocking antibody in vitro for 20 min at 25 1C, after which both cells

and antibody were transferred intravenously into recipient mice. Recipient mice

were analyzed 24 h after transfer. For confirmation of complete blockade of

these cell surface receptors, cells from treated mice were stained with a labeled

antibody of the same clone as the blocking antibody and were analyzed by flow

cytometry. Antibodies specific for a4 integrin (clone PS/2; 200 mg), aE integrin

(clone M290; 250 mg), CD62L (clone MEL14; 100 mg), a4b7 integrin (clone

DATK32; 200 mg) and the isotype control anti–human CD44 (clone 9B5) were

generated ‘in house’ (Supplementary Fig. 1). For CCL25 blockade, mice were

pretreated 2 h before transfer of CD8+ SP thymocytes with 100 mg of antibody

specific for CCL25 (clone 89818) or an isotype control antibody (clone

141945). Both antibodies were purchased from R&D Systems.

Intrathymic FITC injection. These injections were done as described49. A

ventral midline incision was made one third down the sternum to expose the

thymus. A Hamilton syringe was used to inject 10 ml of a 1-mg/ml solution of

FITC (Sigma). The skin incision was closed with Vetbond (3M) and mice were

allowed to recover under a heat source.

Adoptive transfer. Each wild-type or Lta–/– recipient was injected intravenously

with approximately 1 � 107 cells after depletion CD8+ SP thymocytes. SCID

mice were reconstituted with a 1:1 mix of approximately 5 � 105 sorted RTEs

and naive non-RTEs.

Histology. Tissues were embedded in optimum cutting temperature com-

pound (Tissue Tek; Sakura) and were frozen at –80 1C. Frozen tissue sections

6 mm in thickness were fixed in acetone, were blocked with PBS containing 2%

BSA and 2% goat serum and were stained with anti-Thy-1.2 or isotype control

antibody. Slides were washed with PBS containing 2% BSA before and after the

incubation steps, and images were visualized with a confocal microscope

(Nikon Eclipse TE300) equipped with Lasersharp software (Bio-Rad Labora-

tories). Immunoperoxidase staining was done on similarly sectioned tissues

with biotin–anti-Thy-1.2 and a streptavidin-peroxidase kit (Zymed Labora-

tories) and the substrate NovaRED, which was used according to the manu-

facturer’s guidelines.

Statistics. All data are presented as mean ± s.e.m. For antibody blockade

experiments in Figures 3 and 4, Student’s t-test was used for comparisons of

two groups. In Figure 5, the data set was collapsed and the w2 test was used. In

Figure 6, a paired Student’s t-test was used for comparison of IELs and blood.

Note: Supplementary information is available on the Nature Immunology website.

ACKNOWLEDGMENTSWe thank E. Resurrecion for histology, and L. Rott and C. Crumpton for cellsorting. Supported by the National Institutes of Health (AI47822 and GM37734to E.C.B.; DK07056 to A.H.; and DK060000 to T.S.), the Department ofVeterans Affairs (E.C.B.), the Howard Hughes Medical Institute (M.M.W.)and the Stanford Digestive Disease Center (DK56339).

COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.

Published online at http://www.nature.com/natureimmunology/

Reprints and permissions information is available online at http://npg.nature.com/

reprintsandpermissions/

1. Picker, L.J. & Butcher, E.C. Physiological and molecular mechanisms of lymphocytehoming. Annu. Rev. Immunol. 10, 561–591 (1992).

2. Weninger, W., Manjunath, N. & von Andrian, U.H. Migration and differentiation of CD8+

T cells. Immunol. Rev. 186, 221–233 (2002).3. Cheroutre, H. IELs: enforcing law and order in the court of the intestinal epithelium.

Immunol. Rev. 206, 114–131 (2005).4. Hayday, A., Theodoridis, E., Ramsburg, E. & Shires, J. Intraepithelial lymphocytes:

exploring the Third Way in immunology. Nat. Immunol. 2, 997–1003 (2001).5. Ferguson, A. Intraepithelial lymphocytes of the small intestine. Gut 18, 921–937

(1977).

6 ADVANCE ONLINE PUBLICATION NATURE IMMUNOLOGY

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

6. Holmes, G.K., Asquith, P., Stokes, P.L. & Cooke, W.T. Cellular infiltrate of jejunalbiopsies in adult coeliac disease in relation to gluten withdrawal. Gut 15, 278–283(1974).

7. Kim, S.K. et al. Induction and visualization of mucosal memory CD8 T cellsfollowing systemic virus infection. The role of b7 integrins in CD8 T cell traffickingduring an antiviral immune response. J. Immunol. 163, 4125–4132 (1999).

8. Klonowski, K.D. et al. Dynamics of blood-borne CD8 memory T cell migration in vivo.Immunity 20, 551–562 (2004).

9. Cepek, K.L. et al. Adhesion between epithelial cells and T lymphocytes mediated byE-cadherin and the aE b7 integrin. Nature 372, 190–193 (1994).

10. Goodman, T. & Lefrancois, L. Expression of the gd T-cell receptor on intestinal CD8+

intraepithelial lymphocytes. Nature 333, 855–858 (1988).11. Bonneville, M. et al. Intestinal intraepithelial lymphocytes are a distinct set of gd

T cells. Nature 336, 479–481 (1988).12. Guy-Grand, D. et al. Two gut intraepithelial CD8+ lymphocyte populations with different

T cell receptors: a role for the gut epithelium in T cell differentiation. J. Exp. Med. 173,471–481 (1991).

13. Eberl, G. & Littman, D.R. Thymic origin of intestinal ab T cells revealed by fatemapping of RORgt+ cells. Science 305, 248–251 (2004).

14. Guy-Grand, D. et al. Extrathymic T cell lymphopoiesis: ontogeny and contribution togut intraepithelial lymphocytes in athymic and euthymic mice. J. Exp. Med. 197,333–341 (2003).

15. Arstila, T. et al. Identical T cell clones are located within the mouse gut epithelium andlamina propia and circulate in the thoracic duct lymph. J. Exp. Med. 191, 823–834(2000).

16. Guy-Grand, D., Griscelli, C. & Vassalli, P. The mouse gut T lymphocyte, a novel type of Tcell. Nature, origin, and traffic in mice in normal and graft-versus-host conditions.J. Exp. Med. 148, 1661–1677 (1978).

17. Lambolez, F. et al. The thymus exports long-lived fully committed T cell precursors thatcan colonize primary lymphoid organs. Nat. Immunol. 7, 76–82 (2006).

18. Berzins, S.P., Boyd, R.L. & Miller, J.F. The role of the thymus and recent thymicmigrants in the maintenance of the adult peripheral lymphocyte pool. J. Exp. Med.187, 1839–1848 (1998).

19. Boursalian, T.E., Golob, J., Soper, D.M., Cooper, C.J. & Fink, P.J. Continued maturationof thymic emigrants in the periphery. Nat. Immunol. 5, 418–425 (2004).

20. Alferink, J. et al. Control of neonatal tolerance to tissue antigens by peripheral T celltrafficking. Science 282, 1338–1341 (1998).

21. Svensson, M. et al. CCL25 mediates the localization of recently activated CD8ab+

lymphocytes to the small-intestinal mucosa. J. Clin. Invest. 110, 1113–1121(2002).

22. Johansson-Lindbom, B. et al. Selective generation of gut tropic Tcells in gut-associatedlymphoid tissue (GALT): requirement for GALT dendritic cells and adjuvant. J. Exp.Med. 198, 963–969 (2003).

23. Wurbel, M.A. et al. The chemokine TECK is expressed by thymic and intestinalepithelial cells and attracts double- and single-positive thymocytes expressing theTECK receptor CCR9. Eur. J. Immunol. 30, 262–271 (2000).

24. Berzins, S.P., Godfrey, D.I., Miller, J.F. & Boyd, R.L. A central role for thymic emigrantsin peripheral T cell homeostasis. Proc. Natl. Acad. Sci. USA 96, 9787–9791 (1999).

25. Butcher, E.C., Weissman, I.L. & Scollay, R.G. Direct fluorescent labeling of cells withfluorescein or rhodamine isothiocyanate. I. Technical aspects Direct fluorescent label-ing of cells with fluorescein or rhodamine isothiocyanate. II. Potential application tostudies of lymphocyte migration and maturation. J. Immunol. Methods 37, 97–108(1980).

26. Scollay, R.G., Butcher, E.C. & Weissman, I.L. Thymus cell migration. Quantitativeaspects of cellular traffic from the thymus to the periphery in mice. Eur. J. Immunol.10, 210–218 (1980).

27. Gabor, M.J., Godfrey, D.I. & Scollay, R. Recent thymic emigrants are distinct from mostmedullary thymocytes. Eur. J. Immunol. 27, 2010–2015 (1997).

28. Dyall, R. & Nikolic-Zugic, J. The majority of postselection CD4+ single-positivethymocytes requires the thymus to produce long-lived, functional T cells. J. Exp.Med. 181, 235–245 (1995).

29. Ye, P. & Kirschner, D.E. Reevaluation of T cell receptor excision circles as a measure ofhuman recent thymic emigrants. J. Immunol. 168, 4968–4979 (2002).

30. Hazenberg, M.D., Verschuren, M.C., Hamann, D., Miedema, F. & van Dongen, J.J. T cellreceptor excision circles as markers for recent thymic emigrants: basic aspects,technical approach, and guidelines for interpretation. J. Mol. Med. 79, 631–640(2001).

31. von Andrian, U.H. & Mackay, C.R. T-cell function and migration. Two sides of the samecoin. N. Engl. J. Med. 343, 1020–1034 (2000).

32. Pabst, O. et al. Cryptopatches and isolated lymphoid follicles: dynamic lymphoidtissues dispensable for the generation of intraepithelial lymphocytes. Eur. J. Immunol.35, 98–107 (2005).

33. De Togni, P. et al. Abnormal development of peripheral lymphoid organs in micedeficient in lymphotoxin. Science 264, 703–707 (1994).

34. Staton, T.L., Johnston, B., Butcher, E.C. & Campbell, D.J. Murine CD8+ recent thymicemigrants are aE integrin-positive and CC chemokine ligand 25 responsive. J. Immunol.172, 7282–7288 (2004).

35. Mombaerts, P. et al. RAG-1-deficient mice have no mature B and T lymphocytes. Cell68, 869–877 (1992).

36. Hogquist, K.A. et al. T cell receptor antagonist peptides induce positive selection. Cell76, 17–27 (1994).

37. Barnard, J.A., Warwick, G.J. & Gold, L.I. Localization of transforming growth factor bisoforms in the normal murine small intestine and colon. Gastroenterology 105, 67–73(1993).

38. Parker, C.M. et al. A family of b7 integrins on human mucosal lymphocytes. Proc. Natl.Acad. Sci. USA 89, 1924–1928 (1992).

39. Kilshaw, P.J. & Murant, S.J. A new surface antigen on intraepithelial lymphocytes in theintestine. Eur. J. Immunol. 20, 2201–2207 (1990).

40. Mayer, L. & Shlien, R. Evidence for function of Ia molecules on gut epithelial cells inman. J. Exp. Med. 166, 1471–1483 (1987).

41. Hershberg, R.M. & Mayer, L.F. Antigen processing and presentation by intestinalepithelial cells - polarity and complexity. Immunol. Today 21, 123–128 (2000).

42. Camerini, V. et al. Generation of intestinal mucosal lymphocytes in SCID micereconstituted with mature, thymus-derived T cells. J. Immunol. 160, 2608–2618(1998).

43. Blumberg, R.S., Yockey, C.E., Gross, G.G., Ebert, E.C. & Balk, S.P. Human intestinalintraepithelial lymphocytes are derived from a limited number of T cell clones thatutilize multiple Vb T cell receptor genes. J. Immunol. 150, 5144–5153 (1993).

44. Regnault, A., Cumano, A., Vassalli, P., Guy-Grand, D. & Kourilsky, P. Oligoclonalrepertoire of the CD8aa and the CD8ab TCR-ab murine intestinal intraepithelialT lymphocytes: evidence for the random emergence of T cells. J. Exp. Med. 180,1345–1358 (1994).

45. Uehara, S., Grinberg, A., Farber, J.M. & Love, P.E. A role for CCR9 in T lymphocytedevelopment and migration. J. Immunol. 168, 2811–2819 (2002).

46. Okabe, M., Ikawa, M., Kominami, K., Nakanishi, T. & Nishimune, Y. ‘Green mice’ as asource of ubiquitous green cells. FEBS Lett. 407, 313–319 (1997).

47. Lefrancois, L. Carbohydrate differentiation antigens of murine T cells: expression onintestinal lymphocytes and intestinal epithelium. J. Immunol. 138, 3375–3384(1987).

48. Valdez, Y. et al. Major histocompatibility complex class II presentation of cell-associated antigen is mediated by CD8a+ dendritic cells in vivo. J. Exp. Med. 195,683–694 (2002).

49. Butcher, E.C., Scollay, R.G. & Weissman, I.L. Direct fluorescent labeling of cellswith fluorescein or rhodamine isothiocyanate. II. Potential application to studiesof lymphocyte migration and maturation. J. Immunol. Methods 37, 109–121(1980).

NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 7

A R T I C L E S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y

CD8+ recent thymic emigrants home to and efficiently repopulate the small intestine epitheliumTracy L Staton, Aida Habtezion, Monte M Winslow, Tohru Sato, Paul E Love & Eugene C ButcherNature Immunology 7Nature Immunology 7Nature Immunology , 482–488 (2005); published online 2 April 2005; corrected after print 2 May 2006

In the version of this article initially published, the vertical axis label ‘FITC’ is missing from the right column in Figure 1a. The correct figure is presented here.The error has been corrected in the PDF version of the article.

a b

c

CD8αβ+

Spleen

Naive89 2.0

88 1.9

90 1.8

IEL Spleen

PLN Blood

78 1.9

1.5 1.5

0.9 1.24 2.0

RTE

PLN

MLN

CD

62L

CD62L

Enriched CD8 RTE(gated on CD8α+CD4–

αE integrin α4β7 integrin

Cel

l num

ber

Cel

l num

ber

Cel

l num

ber

FIT

C

CD8α

Cel

l num

ber

CCR9

CD44 CD8β

PP

IEL

CD8αβ+

CD44loCD62Lhi

CD8αβ+ CD62Lhi

FIT

C

NATURE IMMUNOLOGY VOLUME 7 NUMBER 5 MAY 2006

ERRATA©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

atur

eim

mun

olog

y